Protocol Online logo
Top : New Forum Archives (2009-): : ChIP and Next Generation Sequencing

Chip problem: antibody or cell number?? - (Apr/15/2009 )

Hi,

I am really in a fix with Chip assay.

I am doing ChiP with an antibody for a transcription factor (TF) using Upstate's Chip assay kit. The antibody I am using is the one used for Chip successfully and published from another lab. The difference between them and us is the cell used for the assay. They used a cell line (MEF) which expresses the TF uniformly, while we are using a mouse tissue, in which only ~3% cells of the tissue expresses the TF. So the population of the target cells used is very very small in our case.

I used about 20 million cells (~0.6 million target cells) per assay. To increase the chance to meet the antibody with the target TF_Chromatin complex, I incubated the antibody (2ug) with sheared chromatin solution in 500ul (100ul chromatin solution + 400ul Chip Dilution Buffer), which is twice as high concentration as original protocol, at 4C overnight. In the beginning, I sheared chromatin in 1% SDS so that the concentration of SDS in the solution was 0.2%. The results were very variable: sometimes I saw 4-8 fold difference between target sample (qPCR Ct value: ~31) and negative control samples (IgG and No antibody, Ct value: ~34), but sometimes there was no difference. Later, thanks to this web, I noticed bad effects of SDS for the Chip and used 0.5% SDS for the chromatin shearing (the size of sheared DNA was as good as before (~500bp)) to lower the concentration of SDS in the solution to 0.1%. This time, results became worse than before. All signals become lower: qPCR Ct values of the sample and negative control were around 34 or not detected. No significant signal/niose fold difference was seen.

As a positive control, I am doing Chip using an antibody for a different TF, which also expresses in the ~3% cells of the tissue. In this assay, I have positive results consistently in both 0.2% and 0.1% SDS containing Chip. So the cell number I applied to the Chip should be OK.

In my understanding, the latter results (no signal!) could be the real results unfortunatelly. Maybe because of the lower affinity of the antibody. Should I give up using this tissue? If there are some suggestions or advice to improve the assay, please let me know.

Thanks in advance for reading this long letter and cooperation to fix the problem.

-ChipChimp-

Since you are doing tissue ChIP which is quite different from cell lines especially only 3% of the cells are expected to be positive, the results could be the real situation. You have set up your experiment nicely with proper positive and negative controls. Is it possible to increase the amount of tissue in the assay? Have you check the performance of your primers?

-pcrman-

pcrman on Apr 15 2009, 09:30 PM said:

Since you are doing tissue ChIP which is quite different from cell lines especially only 3% of the cells are expected to be positive, the results could be the real situation. You have set up your experiment nicely with proper positive and negative controls. Is it possible to increase the amount of tissue in the assay? Have you check the performance of your primers?


Thank you for your advice. Although it is impossible to increase the ratio of positive cells/negative cells, it is possible to increase the whole cell number twice or maybe three times more than before. I am wondering that the increase of the cell number may increase the background, but it is worth trying in my case since I have no positive signals. About the peformance of the primer, I made a series of DNA diluted samples and did qPCR to check the PCR efficiency. The result was ~95%. So, primers seems to be OK.

If you find a good antibody, you get positive results easily, but if not, it is very tough and time consuming to reach the goal...That is my impression about Chip. I will push myself more to get positive results.

Thanks.

-ChipChimp-

Wow you are in quite a fix. Here are my observations from what you've said. I also use the Upstate kit and so have plenty of experience with it.

Is there a way to increase the population of target cells in your sample, e.g. FACS sorting?

I like your idea for trying to increase the chance of an antibody-TF interaction by reducing the amount of ChIP Dilution buffer you add to the sample. Unfortunately the assay is designed for a 1:10 dilution and the concentration of SDS in your final sample is probably too high (though it sounds like you've been addressing that as I've read on). The concentrations of the other components in the dilution buffer are also probably too high. Salt, SDS and possibly EDTA will all interfere. Happily upstate provide the recipes for their buffers, you could try making your own variant that changes the concentrations of the various components so that when you add 400ul to 100ul sample the final concentration is the same as if it were diluted 1:10 normally.

Your positive control sounds good, you've certainly ruled out the cell number possibility but different antibodies can behave differently in the same protocol. How confident are you that the TF you're interested in binds to the site you are PCRing? Is there another site you can PCR as a control for this antibody to which it's known the TF binds?

You could try a couple of different antibodies if you are worried that it might be the problem. Like others though, your result sounds as though it could be the correct one and your TF doesn't bind the site you are looking at under the conditions you're testing. Does your transcription factor need activating in some way before it will bind to DNA, is there a way to artificially activate it to give yourself the best possible chance of detecting its presence on chromatin?

-JPchip-

JPchip on Apr 29 2009, 04:58 PM said:

Is there a way to increase the population of target cells in your sample, e.g. FACS sorting?


First, sorry for late reply. I was away from chip recently.

I could say that it is impossible. Alternatively, I can transfect a cell line with expression vector of the target TF and promoter containing plasmid and then do chip using the transfected cells, although there is no biological meanings in it even though I get positive results. I am doing luciferase assay now and as a preliminary result, I found that my target TF activates luciferase activity, meaning that the expressed TF binds to a promoter region I hooked up to luciferase. So, as a plan"B", I am thinking to use these cells for the chip.

JPchip on Apr 29 2009, 04:58 PM said:

I like your idea for trying to increase the chance of an antibody-TF interaction by reducing the amount of ChIP Dilution buffer you add to the sample. Unfortunately the assay is designed for a 1:10 dilution and the concentration of SDS in your final sample is probably too high (though it sounds like you've been addressing that as I've read on). The concentrations of the other components in the dilution buffer are also probably too high. Salt, SDS and possibly EDTA will all interfere. Happily upstate provide the recipes for their buffers, you could try making your own variant that changes the concentrations of the various components so that when you add 400ul to 100ul sample the final concentration is the same as if it were diluted 1:10 normally.


I am thinking to use no-SDS lysis buffer used in Nat Protocol 2006. Previously, I used this buffer once for fun under the DNA shearing condition I used to use with SDS containing lysis buffer. I got no DNA :wacko: . The condition was too weak to break nuclear membrane. So now, I am more seriously optimizing DNA shearing condition with more tissues than that I used before.

JPchip on Apr 29 2009, 04:58 PM said:

Your positive control sounds good, you've certainly ruled out the cell number possibility but different antibodies can behave differently in the same protocol. How confident are you that the TF you're interested in binds to the site you are PCRing? Is there another site you can PCR as a control for this antibody to which it's known the TF binds?


Yes! That is a good point. Since I am PCRing for the site I have used as a bait for yeast-one-hybrid screening, the binding site might be different in vivo. Unfortunately, I do not know other PCR sites as a control, instead, I designed primers for all possible binding sites which I can expect from target TF's binding domain along the promoter sequence. I hope it might work.

JPchip on Apr 29 2009, 04:58 PM said:

You could try a couple of different antibodies if you are worried that it might be the problem. Like others though, your result sounds as though it could be the correct one and your TF doesn't bind the site you are looking at under the conditions you're testing. Does your transcription factor need activating in some way before it will bind to DNA, is there a way to artificially activate it to give yourself the best possible chance of detecting its presence on chromatin?

In the beginning, I used all 5 commercially available antibodies for the target TF. But in the end, I just focused on two antibodies used/published from another lab, because 1) for the several trial, there was no difference in the results between antibodies and 2) if I use all antibodies every time, I have to use many mice for the chip but there is no space for the breeding in our animal facility so that mice will run out in several assays.

As far as I know, there is no way to activate the TF to have the best results. But I will keep it in mind.

Thanks!!

-ChipChimp-

Hi there,

Just a quick question - are you cross-linking in serum-free media?
Clare

-Clare-

Clare on May 6 2009, 06:21 AM said:

Hi there,

Just a quick question - are you cross-linking in serum-free media?
Clare


Hello,

Yes. I use neurobasal media or PBS and either way, I added nothing.

Thanks

-ChipChimp-