Isobaric labeling of immunoprecipitation

[niki...@gmail.com]

Hi there,

I was wondering if anyone has ever done an immunoprecipitation followed by a 10-plex isobaric tagging (TMT) to quantify the proteins that are bound to the bait relatively to 3 conditions. How do you handle the data when you need to:

1. Subtract the proteins that are non-specific to the bait. I use the IgG as a negative control bait and I was thinking of excluding anything bound to IgG.

2. How do you normalize data from an IP? Can you do log2 transformation followed by median-centre normalization?

Additionally, how complex could a 10-plex immunoprecipitation experiment could be? Is it worth doing high pH fractionation prior to LC-MS/MS?

I am not sure how tricky such an experiment and its analysis would be.

Thank you very much for your help.

Niki

[anneg1...@gmail.com]

Hi Niki,

I'm not sure to understand exactly the experimental design: you have 3 conditions for the IPs and 3 biological replicates, and an IgG controls for each IP, so 18 samples. And you performed TMT labeling for the 9 direct IPs analyzed in one run, is that correct? And the controls were run independently in an other run (with TMT or not)?

You probably have many proteins identified in the IPs using an antibody against an endogenous proteins. I wonder wether simply substracting the list of proteins identified in the IgG controls will be sufficient to eliminate your background and end-up with a "clean" interactome, before doing the statistical analysis between your 3 conditions. Is it? If not, then your stats may give you many proteins that could simply be regulated in the cell betwen your 3 conditions (identified as background), not only those that are differentially binding to your bait between these 3 conditions...

Here we run the IP experiments mostly in label-free mode, so I don't know if it can help, but we do it in this way:

1/ We normalize the intensities (adjust the median of each column, before or after log transformation) between ALL samples (IPs AND controls for each experimental condition and biological replicates, e.g. in your case the 18 samples). In doing that you assume that most of the proteins are not changing between IPs and controls, i.e that most proteins identified in your IPs are contaminants (which is true most of the time) and that your control properly reflects this background (this is more tricky, and IgG may not be a control as relevant as for example an IP using the same antibody performed in cells which are KO for your bait, but at least you should check that you get equivalent numbers of proteins identified in assays and controls...)

2/ For each experimental condition, we would do first the differential analysis of IPs against the corresponding control, using the biological replicates and a simple t-test + a fold change cut-off for example, or a t-test + FDR correction with BH or with the permutation-based method in Perseus. This would give the confident binding partners of your bait.

3/ For this interactome, we then compare the different conditions: we normalize the intensities to the bait (to correct for different IPs efficiencies in the experiments) and look which partners are differentially binding to the bait between these conditions.

I'm not sure how I would organize that in a 10-plex TMT experiment....

Anne

[niki...@gmail.com]

Hi Anne,

Thank you very much for replying and really sorry for my late. Somehow I lost the notification. The experimental design of the experiment is a bit complicated. So, I will explain better:

I have 3 conditions A, B, C where I did an IP of estrogen receptor (ER-my bait). I ran this IP 3 times per each condition. Then I labelled the samples with TMT in a 10-plex. I had 9 ER IP plus one IgG IP (IP of a pool of samples) to make sure I have one IgG incorporated in the 10-plex. Additionally to this, I did a 6-plex where I had 3 IgG IPs pools of the biological experiments for the conditions and 3 ER IPs. I wanted to use this as a reference to subtract the contaminants from the TMT 10-plex in addition to having the one IgG control in the 10-plex. I wanted to be extra careful, because I was doing it first time.

Yes, exactly: I was also thinking the same of simply subtracting the list of proteins identified in the IgG controls. I will try it. Sometimes I have seen that a protein that interacts with ER let's say with 6 peptides identified or even with 40 peptides identified, a few times I saw it with 1 peptide in the IgG or 5-6 respectively. What do you do then? How do you set the cut off of contaminant or not contaminant?

Secondly, I think I got most of the part that you suggested in steps 1 and 2. However, I got a bit stuck with the 3rd step. So, you end up with an interactome done from step 1 and 2 and then you go back to the original, median normalized intensities and you subtract from each condition the intensity of the bait? Do you do that bait normalization after the median normalization stage done in step 1, correct?

Just to clarify: In step 1, the adjustment of each column to the median is done separately for each column/sample.

Thank you very much again. Even though, you do label-free, your post is very helpful. I will see how far I get.

Best wishes,

Niki