Page 1 of 1
HCD or with ETD
Posted: Mon Nov 21, 2011 3:44 pm
We are looking for Orbitrap Velos. But I wonder if we need to attach ETD or not for PTM analysis.
So I want to know your opinion for PTM discovery with HCD fragmentation.
(ETD addition is pricy....)
Can HCD fragmentation work well for PTM discovery?
Posted: Tue Nov 22, 2011 7:23 am
In my experience, HCD works quite well for phosphopeptide analysis, a lot of high profile global phospho-peptide analyses use HCD exclusively. ETD of course will add unique sites and peptides.
Glycopeptides are a bit of a different story. If you are interested, Alma Burlingame's group has done some really nice work on these peptides and a good starting point would be Chalkley, PNAS 2009, "Identification of protein O-GlcNAcylation sites using electron transfer dissociation mass spectrometry on native peptides.".
Everything I've read of his says that HCD can be useful for sequencing the glycopeptides, but that to localize the site ETD is absolutely necessary.
Posted: Tue Nov 22, 2011 8:25 am
Thank you for valuable response and introducing good reference.
I agree that ETD works well for determing the site.
I saw many papers relating with phosphorylation analysis by HCD.
If possible, can you give me a reference of HCD analysis?
I am interested in which kinds of modification HCD can work.
Posted: Tue Nov 22, 2011 8:37 am
Here are two recent papers that discuss collision-based methods for phosphoproteomics. The first is in JPR by the Mann group, where they claim HCD is better (http://pubs.acs.org/doi/abs/10.1021/pr100637q
). The second is in MCP by the Gygi group, where they claim CID is better (http://www.mcponline.org/content/early/2011/09/13/mcp.M111.009910.short
). The dispute is discussed on this thread
. However, it is clear that both can be effective for analyzing phosphopeptides.
Posted: Tue Nov 22, 2011 10:37 am
This is a really good question. I think ETD has received both unfair praise and unfair criticism and it is probably difficult to discern the reality based just upon the literature. The same can be said about HCD but to a lesser degree. I think it comes down to two questions (1) what do you want out of your mass spectrometer and (2) how willing are you to work for it?
If you just want lots of IDs (unmodified or phosphorylated peptides) than you should probably just use CAD or HCD. I have always had better experience with HCD (and OMSSA)
but others have shown good results with CAD (and Sequest)
. You can probably add a few more IDs by also using ETD (either rerunning the sample or using a decision tree method
). But the increase in IDs you get will be fairly minor. And to ensure that you get a benefit you need to be very careful about how you run your experiment. Your anion flux needs to be tuned up very well and your reaction time needs to be properly optimized. In addition ETD runs should exclude precursors < +3 z and > ~800 m/z.
ETD does have some areas where it is very beneficial. For instance, analysis of intact histones or histone tails pretty much necessitates ETD or ECD (ETD paper
, ECD paper
). Also, if you are interested in specific proteins or regions of specific proteins, trypsin may not generate good peptides and you mayb be forced to use a different enzyme. In these cases ETD can be really helpful for characterization. And I have never analyzed glycopeptides but according to the post above that may be a good time to use ETD as well.
Another thing to keep in mind with both ETD and HCD (ETD more so) is that they require vigilant instrument maintenance and tuning. This is one reason why I think many people have been disappointed by both methods. You cannot just walk up to a mass spectrometer and use ETD or HCD as you would with CAD. If ion transfer calibrations are not perfect HCD performance can really suffer. The 'first mass' setting is also crucial for HCD. Setting it below 110 m/z can cause major problems. Setting it to at least 125 is optimal. For ETD to work well you need large populations on anions and quick fill times. Then you need to infuse a peptide and optimize the reaction time. Then once running the your sample you need to check injection times every day to make sure things don't change. If you are dilgent about these things you can get some nice results. But these methods are not as robust as CAD.
Finally, I just want to qualify this post by mentioning that it's been a little while since I have used these methods. If you want the best advice about tuning instruments and selecting settings I would encourage you to post here about it (we have some members that are quite knowledgable in these areas) or email groups specifically about how they set up their instrument.
I hope that helps.
Posted: Tue Nov 22, 2011 11:57 am
Doug and Craig,
Thank you for the comment.
Our lab is service lab and looking for new one.
One issue is if we need to add ETD on Orbitrap Velos or not.
ETD is good but it costs extra $$$ & extra maintenance (we need to do).
As Doug said, I also heard ETD needs to be attention for keep running.
Many university lab do not attach ETD on Orbitrap (not because of cost).
So if we do not add ETD, I want to clarify what we can do for customer and not for customer.
ETD: Good for glycopeptide and most PTM analysis
HCD can do most of stable modification analysis. ...can we say that?
(YES, Phospho, Acetylation, ubiquitination, SUMO, ...)
Thank you for taking your time.
Posted: Mon Nov 28, 2011 12:21 pm
Doug asked me to comment as I have worked in a MS service facility. I have limited ETD experience with the LTQ-ETD system.
It is questionable on whether you will need ETD or not. Doug and Craig have pretty well covered the pros and cons and types of analysis you can do. Remember that ETD is rather touchy and not as robust as HCD/CID. ETD makes a good supplement to HCD/CID rather than a replacement. You will also need much more starting signal for ETD as the ETD cleavage is a rare event and the fragments are very low in abundance. The HCD data on the Orbi Velos that I have seen look quite good and I expect it to be used more often in the future. Glycopeptides are difficult as you mostly get sugar cleavage, but often you don't see enough fragmentation on larger N-linked glycans to reach the core GlcNAc residues with affect the total mass adjustment. Peptide backbone fragmentation is rare with larger glycans and thus ETD fills this gap by assisting with site determination. ETD also helps when working with larger, higher-charged, peptides as you can get more fragmentation information than HCD/CID.
In regards to what you can tell your customer base...
HCD/CID Yes: Acetylation, Methylation, Ubiquitination, SUMO, Phosphorylation, Glycosylation (typically structure only, up to a point, often no peptide ID)
ETD Yes: Everything in HCD/CID, Glycosylation (site localization only, peptide ID), Phosphorylation
Also keep in mind the added cost for ETD (both upfront and maintenance) and the added time needed to tune everything as you likely will not use it every day. If user fees are your main facility income, I think you likely will not recoup the added cost for the ETD option over its lifespan as the number of projects will be quite limited compared to your normal protein ID experiments. From a purely scientific perspective, ETD (even with its flaws) can be your only option to solve some of your investigator's questions and an important tool in the toolbox.
Hope this helps.
Posted: Wed Dec 07, 2011 11:59 am
The discussion on the thread has been excellent. I thought I would add to it by giving the perspective from someone who just acquired an instrument with ETD.
I just purchased an OrbiTrap Elite with ETD for a shared resource facility at a non-profit research institution. ETD is an expensive option so I gave significant thought into why I would need this capability. There were three reasons for including ETD with the Elite. First, to primarily aid with peptide mapping experiments. There are numerous instances where I need to map a protein (e.g. in vitro kinase experiments to locates sites of phosphorylation) and potential proteolytic peptides could be quite large. When performing our LC-MS experiments, these larger peptides go undetected. I am hoping that ETD will provide sequence information on the larger peptides resulting in more complete maps. Our instrument has only been in operation for two weeks, but I envision using ETD on every ion that is >+3 while performing mapping experiments. Second, modification site localization. There are occurrences when analyzing phosphopeptides that I know a phosphopeptide has been detected but the site of modification within the peptide is dubious. I hope that ETD will assist with better phosphopeptide site localization. Third, large-scale peptide identification. With proteomics experiments, I plan to use ETD in a decision tree format to identify larger peptides.
Jason's assessment of the role of ETD within a core facility was very good. I see this capability as a tool in my toolbox to be used on occasion; not something that will be used all the time. If the added cost of ETD is too much, then you might think of identifying other core facilities that could provide this capability on a fee for service basis.
Unfortunately, my post does not get to the heart of the original topic of this thread; namely, will HCD be sufficient for PTM discovery or is ETD necessary. I don't have enough experience in that area yet, but I fully expect ETD to complement CID and HCD. I think the biggest question will be the level of complementarity it provides. Good luck with your big decision.