Recent opinions on the "classical discussion points" in proteomics

If it doesn't fit into any other category post it here.
m/z
Angiotensin Member
Angiotensin Member
Posts: 47
Joined: Thu Oct 17, 2013 5:06 am

Recent opinions on the "classical discussion points" in proteomics

Postby m/z » Mon Apr 28, 2014 11:23 pm

I was hoping for some opinions on these "classical discussion points" in proteomics:

1)
To label or not in quantitative proteomics?
Label-free proteomics seems to become more and more popular, also among top proteomics researchers in the field? Label-free is easy and cheap - of cause you have to validate your results by for example western blotting, but I guess that also goes for iTRAQ, ICAT and SILAC? So in many proteomics studies (perhaps most published studies?) label-free is actually sufficient?

2)
Online 2DLC vs. gel-LCMS
Isn't online 2DLC becoming more and more "old school"? I have tried 2DLC and I found it to be fairly challenging. The problem is that everything is so integrated that it can be difficult to trouble-shoot the method (i.e. identify sources of contamination). Here gel-LCMS is more easy since it consists of individual steps and you have time to stop and validate things. Also, from what I understand some expert groups use offline SCX (for example SCX tips), but specifically do not use online 2DLC perhaps for the same reason? And with the emerging and very fast MS instruments perhaps 1DLC will in most cases be sufficient in the future?

Maybe I am missing something? Please feel free to share your opinion.
Thanks!

tsbatth
Ubiquitin Member
Ubiquitin Member
Posts: 58
Joined: Thu Jul 12, 2012 7:26 pm

Postby tsbatth » Tue Apr 29, 2014 4:13 am

Im not sure what you mean by 2DLC, do you mean all in an online manner like MudPit ? I don't think that's popular much anymore but certainly offline fractionation by SCX, ERLIC, high pH reverse phase are fairly common and superior to gel based fractions in most cases.

m/z
Angiotensin Member
Angiotensin Member
Posts: 47
Joined: Thu Oct 17, 2013 5:06 am

Postby m/z » Tue Apr 29, 2014 4:44 am

Yes, by 2DLC I meant mudpit (or online SCX-RP LC).

Christopher
Angiotensin Member
Angiotensin Member
Posts: 42
Joined: Thu Dec 27, 2012 12:26 pm

Postby Christopher » Tue Apr 29, 2014 5:28 am

1. I think this really depends on your application. In label free analysis you really live and die by the quality and reproducibility of your system, whereas this is less of a problem in a labeled sample (although obviously you want it to be as good as possible). I think if you can get by with doing an extended gradient single-shot style analyses of your sample, then label free is very nice. You can incorporate some level of internal standard in the form of synthetic peptides to help calibrate between runs and get really nice data. The obvious advantage to label free analysis is that you decrease the complexity of the MS1 data, which can lead to cleaner quantification, depending on your sample type. I think the biggest limitation, in both label-free and labeled quant data is the software, and the methods used for transforming and normalizing the data now that we are starting to get up to large numbers of samples. I think this is improving with some recent developments in R packages though.

There are some other analyses types, such as turnover measurements that are going to require labeling.

2. Again I think this depends on your application. I wouldn't say 2DLC is old school, if anything is old school it is gel-LCMS. I do think it is far more common to just do offline 2D LC, as this gives you much more flexibility for your UPLC platform. It doesnt always have to be plumbed to do a 2D LC separation. Also, you have the flexibility of using a wider bore column, meaning you can hit it with more material. I think 2DLC will see a resurgence once lab-on-a-chip platforms start to really come into their own as people start to push the amount of sample required for good proteome depth lower and lower.

Zhang
Angiotensin Member
Angiotensin Member
Posts: 27
Joined: Fri Aug 24, 2012 8:08 am

Postby Zhang » Tue Apr 29, 2014 6:33 am

Label-free with single-shot is simple and economic. Using high resolution instruments, you can reach very deep as well. But labelling with 2D-LC is even deeper. We observed lower CV in protein quantification comparing to label-free.

gabe
Angiotensin Member
Angiotensin Member
Posts: 25
Joined: Sat Nov 19, 2011 5:56 pm

Postby gabe » Tue Apr 29, 2014 6:37 am

I think in both cases it depends on what you're used to. All proteomic techniques take a certain amount of practice and infrastructure to perform well. If you've never done 2DLC or label-free analyses before, they'll be hard. I've been doing 2DLC (MudPIT) for years and its much easier for me than other techniques (e.g. offline SCX) because I have everything in-place to perform it and I know how to troubleshoot. Having said that:

(1) Label-free analysis requires good, reproducible chromatography and appropriate software for analysis, but once you have those in-place it's a great method. That said, techniques like SILAC, reductive dimethyl labeling, and iTRAQ/TMT are superior and make me wonder why anyone attempts to use label-free quantitation anymore...

(2) I don't think 2-D fractionation will ever be replaced by 1D. You'll always be able to get much more depth-of-coverage with 2D separations, whether it's geLCMS, online 2D-LC, offline 2D-LC, IEF/LC, etc. As I said above, the choice of separation techniques in the first dimension depends on what is easiest for you and your lab. I prefer MudPIT but all of these techniques can be made to perform well and will remain popular.

m/z
Angiotensin Member
Angiotensin Member
Posts: 47
Joined: Thu Oct 17, 2013 5:06 am

Postby m/z » Tue Apr 29, 2014 11:57 am

I have primarily been working with gel-LCMS and label-free and I just now realize how biased my opening post is ;-) Maybe I am the one who is old school...:-/

However, now that my cover is blown in this matter: there are papers that demonstrate that label-free is as good as labelling-strategies. I haven't seen any paper showing the opposite? I think that there are even some papers that conclude that label-free can be more sensitive than labelling strategies?

I admit that if labelling strategies can overcome issues with reproducibility of chromatography and are actually easier to implement than label-free then that is some very valid arguments in favour of using labelling for the conventional proteomics studies: by the way, in the opening post I was thinking of the "good but not exceptional" proteomics paper that constitutes the majority of papers - not expert studies or method-development studies by masters in the field. Label-free just seemed the most obvious choice to me based on my overall impression of the field. But yes, I am biased...

I agree that offline SCX and LCMS is more sensitive that Gel-LCMS. I don't know of any direct comparisons of Gel-LCMS vs Mudpit?

PhilG
Phosphoserine Member
Phosphoserine Member
Posts: 11
Joined: Thu Aug 04, 2011 1:15 pm

Postby PhilG » Tue Apr 29, 2014 12:47 pm

For a direct comparison of Gel-LCMS vs MudPit (and other fractionation methods), check out:

Proteomics 5 (2005) 2018-2028.

Assuming I got the details from the paper correct, a 13-cycle MudPit analysis of a 65 ug yeast lysate identified 524 proteins while 38 2-mm regions cut from an SDS-PAGE gel of an identical sample identified 898 proteins. Fairly labor intensive, but if the goal is to identify the most proteins and get the highest sequence coverage per protein, then the Gel-LCMS method is considerably better. This paper is somewhat dated and used older ion trap instruments. It would be interesting to know what the results would be with current instruments.

After reading this paper a number of years ago I moved away from MudPit and relied more on Gel-LCMS fractionation. I think having approximate protein mass information from Gel-LCMS can be quite useful in some experiments (e.g. why is my 20 kDa protein identified in a 75 kDa region?). The protein mass info is lost when digesting to peptides and separating by SCX or MudPit.

m/z
Angiotensin Member
Angiotensin Member
Posts: 47
Joined: Thu Oct 17, 2013 5:06 am

Postby m/z » Tue Apr 29, 2014 11:19 pm

@ PhilG: Interesting paper:
- 38 gel slices for one sample is a lot! In most published gel-LCMS studies I believe each sample is sliced into <10 fractions. Of cause the optimal depends on the sample and the LCMS platform.

- For someone experienced in Mudpit Gel-LCMS may not seem like a tempting alternative and it may not be. It depends on many things, I agree. However, I would still argue that Mudpit requires really good LCMS skills whereas the sample preparation of Gel-LCMS can be performed by basically anyone and a conventional 1D LCMS analysis with a modern LCMS platform is now becoming very robust. But then again, some may find that Mudpit is equally robust?

To me the unique advantages with Gel-LCMS are:
- If you load ~60ug on the gel each piece should on average contain ~6-10ug protein. But if you slice in a smart way you can isolate troublemakers such as albumin and big structural proteins to primarily a few fractions and that may improve the overall number of ID's. If you digest your sample first, as in most 2D LC workflows, then albumin peptides will be in all fractions.
- After slicing you can store your gel pieces for a long time in the fridge. The proteins are not going anywhere.
- With Gel-LCMS you can asses if there is equal total amount of protein loaded from each sample by looking at the gel lanes. Measuring the protein concentration can be misleading if one sample contains a few high abundant proteins - when comparing tissue samples this can be a real challenge due to the varying amount of plasma protein and extracellular matrix protein present in individual tissue specimens. I don't know that any other method than SDS page can provide this information?
- For subcellular fractionation Gel-LCMS is helpful since you can assess if your biochemical fractionation work has had any substantial effect on the overall protein profile. If the lanes look similar you may not have accomplished any real enrichment of a desired subproteome.

- There is probably more labor/time with Gel-LCMS as compared to Mudpit:
I think the typical Gel-LCMS workflow of a single sample can be summarized by:
Day1: SDS PAGE (3 hours) & comassie staining (1hour) & destaining (over night),
Day2: slicing into 10 peices and dicing (1hour) & washing, reduction, alkylation (3 hours) & trypsin digestion (over night),
Day 3: Peptid extraction and speed vac. concentration (3-4 hours) & 10 x LCMS (~20 hours) (note: It is not necessary to do zip tipping since SDS PAGE is also cleaning the proteins).
Day 4: Data analysis

Biomarker
Albumin Member
Posts: 88
Joined: Thu Aug 09, 2012 3:28 am

Postby Biomarker » Tue Apr 29, 2014 11:47 pm

@ m/z

You can actually do day 1 & 2 days work in one day. This has been made possible with PAGE blue stain from thermo. It just take one hour to stain and the same time to destain. This is just technical input..

cheers..
m/z wrote:@ PhilG: Interesting paper:
- 38 gel slices for one sample is a lot! In most published gel-LCMS studies I believe each sample is sliced into <10 fractions. Of cause the optimal depends on the sample and the LCMS platform.

- For someone experienced in Mudpit Gel-LCMS may not seem like a tempting alternative and it may not be. It depends on many things, I agree. However, I would still argue that Mudpit requires really good LCMS skills whereas the sample preparation of Gel-LCMS can be performed by basically anyone and a conventional 1D LCMS analysis with a modern LCMS platform is now becoming very robust. But then again, some may find that Mudpit is equally robust?

To me the unique advantages with Gel-LCMS are:
- If you load ~60ug on the gel each piece should on average contain ~6-10ug protein. But if you slice in a smart way you can isolate troublemakers such as albumin and big structural proteins to primarily a few fractions and that may improve the overall number of ID's. If you digest your sample first, as in Mudpit, then albumin peptides will be in all fractions.
- After slicing you can store your gel pieces for a long time in the fridge. The proteins are not going anywhere.
- With Gel-LCMS you can asses if there is equal total amount of protein loaded from each sample by looking at the gel lanes. Measuring the protein concentration can be misleading if one sample contains a few high abundant proteins - when comparing tissue samples this can be a real challenge due to the varying amount of plasma protein and extracellular matrix protein present in individual tissue specimens. I don't know that any other method than SDS page can provide this information?
- For subcellular fractionation Gel-LCMS is helpful since you can assess if your biochemical fractionation work has had any substantial effect on the overall protein profile. If the lanes look similar you may not have accomplished any real enrichment of a desired subproteome.

- There is probably more labor/time with Gel-LCMS as compared to Mudpit:
I think the typical Gel-LCMS workflow of a single sample can be summarized by:
Day1: SDS PAGE (3 hours) & comassie staining (1hour) & destaining (over night),
Day2: slicing into 10 peices and dicing (1hour) & washing, reduction, alkylation (3 hours) & trypsin digestion (over night),
Day 3: Peptid extraction and speed vac. concentration (3-4 hours) & 10 x LCMS (~20 hours) (note: It is not necessary to do zip tipping since SDS PAGE is also cleaning the proteins).
Day 4: Data analysis

tsbatth
Ubiquitin Member
Ubiquitin Member
Posts: 58
Joined: Thu Jul 12, 2012 7:26 pm

Postby tsbatth » Wed Apr 30, 2014 1:41 am

PhilG wrote:For a direct comparison of Gel-LCMS vs MudPit (and other fractionation methods), check out:

Proteomics 5 (2005) 2018-2028.

Assuming I got the details from the paper correct, a 13-cycle MudPit analysis of a 65 ug yeast lysate identified 524 proteins while 38 2-mm regions cut from an SDS-PAGE gel of an identical sample identified 898 proteins. Fairly labor intensive, but if the goal is to identify the most proteins and get the highest sequence coverage per protein, then the Gel-LCMS method is considerably better. This paper is somewhat dated and used older ion trap instruments. It would be interesting to know what the results would be with current instruments.

After reading this paper a number of years ago I moved away from MudPit and relied more on Gel-LCMS fractionation. I think having approximate protein mass information from Gel-LCMS can be quite useful in some experiments (e.g. why is my 20 kDa protein identified in a 75 kDa region?). The protein mass info is lost when digesting to peptides and separating by SCX or MudPit.


But 2005 is a long time ago, there have been many improvements in sample prep, instrumentation since then. Just in single shot analysis of human proteome we can get 4-5 thousand proteins in 1-2 gradient runs from 1 ug. You can get more depth from gel-lcms but from my experience high pH reverse phase gives better coverage/depth in half the time of Gel-LCMS. Of course as you said the protein mass information is lost. I personally don't have much experience with MudPit but I guess it wont be the cleanest method for your instrument since you'll be putting salt through your samples, column etc...

m/z
Angiotensin Member
Angiotensin Member
Posts: 47
Joined: Thu Oct 17, 2013 5:06 am

Postby m/z » Wed Apr 30, 2014 2:15 am

tsbatth wrote:But 2005 is a long time ago, there have been many improvements in sample prep, instrumentation since then. Just in single shot analysis of human proteome we can get 4-5 thousand proteins in 1-2 gradient runs from 1 ug. You can get more depth from gel-lcms but from my experience high pH reverse phase gives better coverage/depth in half the time of Gel-LCMS. Of course as you said the protein mass information is lost. I personally don't have much experience with MudPit but I guess it wont be the cleanest method for your instrument since you'll be putting salt through your samples, column etc...



4-5000 proteins is clearly better than what we can do with any workflow, including Gel-LCMS - (we dont have a state-of-the-art MS instrument).
Do you do RP with high pH of intact proteins on a normal uHPLC? Or do you digest first?

Christopher
Angiotensin Member
Angiotensin Member
Posts: 42
Joined: Thu Dec 27, 2012 12:26 pm

Postby Christopher » Wed Apr 30, 2014 8:29 am

Some newer studies have been published that demonstrate the coverage you can obtain with current-gen instrumentation and methods.

http://www.ncbi.nlm.nih.gov/pubmed/23540446

Or using 3-d online chromatography setups with long run times.

http://www.ncbi.nlm.nih.gov/pubmed/23863870

There was a paper in 2012 I believe in MCP where they looked at the yeast proteome, I want to say between stress conditions, but I cannot remember. But they used an in-gel method and achieved almost complete coverage of the yeast proteome. They divided the gel into 28 slices I believe. I wish I could find the paper, but I cant.

There was also a nice paper recent in Nature Methods demonstrating the HiREF LC-MS method that you could use to obtain great coverage of the proteome while potentially discovering novel gene products.

In summary, I think you can find a paper that demonstrates really nice data from any fractionation and analysis method. It really comes down to what works well for what you are trying to investigate and what you have available to you in your lab.

We often do high-pH reversed phase in our lab, almost always on peptides, on a regular HPLC.

m/z
Angiotensin Member
Angiotensin Member
Posts: 47
Joined: Thu Oct 17, 2013 5:06 am

Postby m/z » Wed Apr 30, 2014 11:54 pm

Sounds interesting. I hope to be able to test pre-fractionation of peptides/proteins by RP at high pH followed by LCMS, as an alternative to our usual Gel-LCMS workflow.

In summary, perhaps there is agreement that with high-end LCMS platforms any of the established proteomics workflows can now deliver thousands of protein identifications from biological samples - whether it be Gel-LCMS, online 2D, offline 2D or even direct 1D LCMS (with a 50cm column) and, in addition, both label-free and labeling strategies are becoming more and more user-friendly. Sorry if I seem to be nitpicking here, but I also believe that it is also NO LONGER true that labeling is superior to label-free.

tsbatth
Ubiquitin Member
Ubiquitin Member
Posts: 58
Joined: Thu Jul 12, 2012 7:26 pm

Postby tsbatth » Thu May 01, 2014 4:58 am

m/z wrote:4-5000 proteins is clearly better than what we can do with any workflow, including Gel-LCMS - (we dont have a state-of-the-art MS instrument).
Do you do RP with high pH of intact proteins on a normal uHPLC? Or do you digest first?


Digest first. I can show you our setup if you like, we're just across the street remember :P .

m/z
Angiotensin Member
Angiotensin Member
Posts: 47
Joined: Thu Oct 17, 2013 5:06 am

Postby m/z » Thu May 01, 2014 9:12 am

tsbatth wrote:Digest first. I can show you our setup if you like, we're just across the street remember :P .


Sounds good! :-)


Return to “Other”

Who is online

Users browsing this forum: No registered users and 1 guest