Hi Jan,

The way I have approached experiments like this are to complete seperate experiments for each biological replicate (ie 5 different sets of 6-plexes). Three biological replicates is generally accepted as sufficient for analysis. Is there a particular reason that you are runing five? If you were to pool samples from each group, I feel that you would negate the reason for completing biological replicates to begin with. It is also important to randomize the tag assignment in an experiment of this magnitude so that there is no tag-related bias. I use www.random.org to for this. 

Depending on the instrument that you are using, Thermo has a 10-plex kit available as well. I don't have any personal experience with this, but the small amount of data that I have seen with it seems to be convincing: http://www.piercenet.com/product/amine-reactive-10-plex-tandem-mass-tag-reagents.

For software, we have used ProteoIQ in the past, but are currently evaluating Proteome Discoverer, Scaffold, and PEAKS as means to quantitate the data as well. 

Please let me know if you have any other questions that I might be able to help with. I have spent the last two years working completely with TMT proteomic related projects. 

Best,

Kristin

Depending on the instrument you plan to use, you have to be very careful with the known interference problem using iTRAQ/TMT tags. Essentially, what you get is a mixture of quantitative results of the precursor +-0.XDa window. Whatever is present in that window will add to your quantification.

We have done simple tests using BSA treated with TMT in mixture 1:1, 1:10 and 10:1. The quantitative results are all over the place.

For more than 3 years now, there is an increased awareness of this problem in the proteomics community.

This is addressed with the latest instrument generation where you can do synchoronous precursor selection MS3. In previous generation instruments the MS3 resulted in so little sensitivity that it was not widely adopted.

In other words, you have more things to consider than only how to pool your samples.

Very interesting point Sonja, never realized interference might play such an important role for this.  We use LTQ Velos Pro.. Proteome Discoverer calculates % of isolation interference, is there any value considered to be "dangerous"? I know that the lower the better, but do you use any treshold value for peptides you don't count with?

Thanks Kristin for your reply, I was wondering whether it would be better to use pooled control sample for each experiment (5 different experiments of 6-plexes (5 biological replicates), each replicate separately, but related to the same, control, pooled sample). Because if you do it separately for each replicate, one set of 6-plex might be completely different from the others but only due to disparity of one control animal. 

Thank you once again

Jan

We currently do not use it, because the results of that said BSA sample led us to believe that we cannot accurately quantify our data. If we cannot determine it for BSA, we did not expect this to work reliably with something more complicated. 

In extreme cases, we have seen a peptide that was not labeled at all, yet still gave outstanding quantitation results. Initially puzzled, we analyzed our data more carefully and realized that this all came from interfering ions. If you do not look for this, you might be entirely blind to it, but it has been reported in the literature. Gygi proposed the MS3, Coon proposed gas-phase purification. So, it is a problem that is known in the field. 

You can look at your MS1 and you will almost never find an isolation of a single precursor. The bigger problem though is that these very small ions that are co-isolated can throw off your quantitative results entirely, regardless of the percentage isolation that is reported. As said, in extreme cases, it could be that this 1% or less is contributing to 100% of your quantification results. And it should be a concern, if you go through all that trouble preparing your samples and all you get is uncertainty.

We are using the MS3 approach proposed by the Gygi group on our Orbitrap Velos Pro with TMT-labeled samples. On our instrument, comparison of a complex mixture spiked-in with a set of proteins at known concentrations showed increased accuracy for the MS3 approach in contrast to the MS2-HCD. As Sonja already mentioned, a multi-notch algorithm increasing the sensitivity of reporter ions in MS3 described by McAllister et al. (http://pubs.acs.org/doi/abs/10.1021/ac502040v) is implemented in the new generation Orbitrap Fusion.

For older instruments which are not equipped with the multi-notch algorithm, Thermo suggested different protocols for selection of MS2 product ions (and increasing the sensitivity) for further quantification of reporter ions ins MS3. You should check which of these methods work best for your instrument and sample.

If you are going for a TMT10-plex (or a TMT8-plex) sample as suggested by Kristin you need to use a resolution of 30k at the OT for your MS3 spectra to achieve baseline separation of the reporter ions (see Werner T et al., Anal Chem. 2012 Aug 21;84(16):7188-94. http://pubs.acs.org/doi/abs/10.1021/ac301553x)!

I have two sipmle, maybe naive questions about that. What if I narrow the isolation window? I presume it's not sufficient, cause you did not mention it..

And after CID fragmentation in MS2, will be those tags on fragments still available for MS3 in HCD?

The last thing, as I understood this issue is not so new, I have never seen it in any paper (that is interested in biology more that the method itself)...

Ting et al. (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3205343/) show the impact of smaller isolation windows for quantitation accurancy (fig. 2B and 2D. They lowered the isolation window from m/z of 2 to 0.5 but interference was not completely eliminated. Additionally, your number of peptide IDs is decreased by roughly 20% using this setting (fig S4).

You mentioned using Proteome Discoverers isolation interference for eliminating interference. Ting et al. also investigated the effect of isolation specificity thresholds which is summarized in fig. S3. Their conclusion was that using isolation specificity is insufficient in predicting or eliminating the interference problem.

For the MS3 workflow: We are using an isolation window of m/z 1.2 for CID-MS2 which is a tradeoff between sensitivity and specificity as we want to include at least the +1 isotope peak of our precursor for higher signal intensity. For the consecutive HCD-MS3 of the MS2 we use a m/z window of 2.5 as the background of our MS2 spectra are pretty low. The majority of the product ions (in MS2) still contain the TMT tags and thus you will be able to monitor the reporter ions in MS3 (shown in figure 3A in the Ting paper)

Thank you all for your suggestions. I'll try the MS3 method, I hope I will not lose most information in contrast to MS2 method..

Good luck!