Name: Jonathan Sweedler
Position: James R. Eiszner Family Chair in Chemistry, University of Illinois
Background: PhD, University of Arizona; NSF Postdoctoral Fellow, Stanford University
Jonathan Sweedler is the James R. Eiszner family chair in chemistry at the University of Illinois and a leading researcher in the field of peptidomics, with a particular interest in neuropeptidomics.
While proteomics typically uses trypsin digestion to break proteins down into peptides for mass spec analysis, peptidomics skips this step, instead analyzing the endogenous peptides found in an organism. This approach offers insights into the organism's native peptide population and these peptides' roles across a wide range of biological processes.
Although similar in a number of respects to proteomics, peptidomics comes with its own set of issues, including unique sample prep, dynamic range, sensitivity, and informatics concerns.
This week, ProteoMonitor spoke to Sweedler about the field's applications, potential, and challenges.
Below is an edited version of the interview.
What distinguishes peptidomics from proteomics?
Peptidomics is the measuring of the global peptides in a tissue as opposed to the proteins. So if you have a section of brain, you'll end up with the protein content from proteomics, and if you're doing a metabolomic experiment you'll be measuring the metabolites. Peptidomics is trying to capture the information on the peptides that were endogenously present while the tissue was living.
To measure proteins, one of the common ways is to digest the proteins into peptides and measure them. In peptidomics we're not doing that [digestion] step, so you're seeing ideally the peptides that were present in the tissue. So, in one sense, it's a strange way to say it, but because of the limitation of the ability to characterize full-length proteins – and obviously people with top-down proteomics are trying to alleviate that [limitation] – the process for measuring proteins involves making a bunch of peptides. In peptidomics, we're looking for the [endogenous] peptides.
Does peptidomics bleed into top-down proteomics, then, given the latter's concern with looking at proteins with their post-translational modifications and processing intact?
It would get people mad at me for saying this, but peptidomics is like top-down proteomics [in that] you're looking at the endogenous forms of the molecules. Except the difference is that top-down proteomics tends to be looking at larger molecules and has a separate set of technical issues. Peptidomics happens to be looking at the class of molecules that are back into the sweet spot of mass spectrometry, and that's the peptide range.
What challenges, technical and otherwise, are involved in peptidomics?
Well, In the case of neuropeptides – and this is the area where I work – you take a large protein, which is called a neuropeptide pro-hormone, and endogenous enzymes cleave this into a number – say, 10 peptides – and then those peptides sometimes have post-translational modifications on the C- or N-termini, and so you could have, as an example, acetylation, or pyro-glu, or amidation of peptides in the middle of the protein because it cleaved them and modified them. You wouldn't get that with a proteomics experiment.
In neuropeptidomics, very often, some of the peptides are transported from one part of the brain to another, and others aren't. I'm often asked this by reviewers who understand proteomics. They say, well, you only detected some of the peptides, where are the others? And the answer is they aren't even there.
Peptidomics has as an issue that the dynamic range of neuropeptides and other peptides can be ten orders of magnitude. And oftentimes peptides from different pro-hormones can be at vastly different levels, and the physiological response depends on the form and the ratio of levels. From a neuropeptide point of view, these are the molecules that regulate the nervous system, and one of the hard things is that adjacent neurons can have different neuropeptide complements. So if you want to understand how a particular neuronal network works, you should really know what particular neural peptides are present. So you need high sensitivity and ideally in many cases you really want to go to select cells, and ideally you would like some dynamic and temporal information.
If you move away from peptides, and you think about how people measure other cell-signaling molecules in the brain, they use electrochemistry and they measure dynamics and spatial information with an electrode. That's hard to do with mass spectrometry. So a lot of what my group has worked on is how to make peptidomics or how to make the global characterization of neuropeptides and hormones work on very small samples down to single cells. And that becomes completely a dynamic range and detectability issue.
What techniques are peptidomics researchers using to tackle these challenges?
One thing you can do is that if most of the neuropeptides are contained in a cell … you can take the neuron, and leave it relatively intact and you can get it onto, for example, a MALDI mass spectrometry stage.
Is MALDI the primary mass spec platform currently used in peptidomics?
It depends on what you're trying to do. You can do electrospray. We've actually stuck cells into a capillary recently, and we can measure the contents using mass spectrometry – you put the whole cell into a capillary and do capillary electrophoresis mass spectrometry. So there's a range of approaches depending on what you're trying to do. MALDI imaging has become very useful. If you're doing imaging, and you want spatial information, something like MALDI is compatible.
What are the complications of the informatics given that you're not just looking at trypsin cleavages as in proteomics but at cleavages from a wide variety of enzymes?
Depending on the cell type there are a lot of stereotypical cleavages. Where a signal peptide cleaves off [for instance], there's a lot of informatics to predict that. For cleavages of pro-hormones there are di-basic, tetra-basic, and even some mono-basic sites, and those cleavages, just like trypsin, are fairly predictable. What's not predictable is that, depending on the state of the cell, it may express different complements of these processing enzymes. There are some unusual cleavages, and those are unusual because they can't be predicted, and some of those create very important molecules. You wouldn't necessarily without knowledge predict that angiotensinogen is converted to angiotensin-1 and how that goes to [angiotensin]-2 and -3.
How well developed are informatics resources like databases and search algorithms for peptidomics?
There are some informatics resources. We have something called Neuropred, which is for neuropeptide prediction. We have this for a number of animal models. There's a group in Belgium that has a peptide-processing [resource], and then there's SwePep, by [Uppsala University researcher] Per Andren's group. We're actually trying to combine ours with SwePep.
So there are peptide-prediction algorithms and peptide databases and obviously a lot of the global proteomic databases list peptide forms as well. Some of the typical processing approaches have to be changed a little bit, and you really have to validate it. It's not like trypsin where you expect all your peptides to be present at the same level and maybe you're not going to detect them all but they should all be there. Oftentimes [in proteomics], you see people take a very large protein, digest it, and say, 'Look, this is how well my approach works.' And for a peptidomics approach, you really don't expect all the peptides from a pro-hormone to be present, and so in that sense some of the statistical tests have to be done differently.
What are some potential applications of peptidomics?
In the 1970s, to discover a new neuropeptide … it was work that took literally tons of brain material to get a structure. And now there are many cases … where you can look at a single cell and discover a new neuropeptide. Because of that type of technological advance, the rate of neuropeptide discovery is accelerating, so there are more being reported, and because of that there's a lot of interest in what they do.
Neuropeptide receptors are very wonderful targets of pharmaceutical intervention. If a cell releases a peptide to control something, then understanding that gives us potential ways of controlling, alleviating, modifying these types of factors. So they're biologically relevant for a couple reasons. One is that it tells us something about how you can modulate neuronal function, which may be related to cell survivability, behavioral implications, or alleviating mental disorders. We certainly know that … classical transmitters [like serotonin or dopamine] are important, but it's becoming more and more obvious that neuropeptides are absolutely critical players.
My group works on circadian rhythm … and we see [that] at different times of day under different stimulation conditions, a small number of neurons – only 10,000, in this one brain region – can release amazingly complex and distinct sets of neuropeptides. So we're trying to understand what peptides are present, when they're present, and what they do.
What level of support and interest has peptidomics received from mass spec vendors compared to proteomics?
The good news from a mass spectrometry point of view is that to be able to separate and detect peptides, it doesn't really matter – it's the front end sampling that matters. If I take a section of brain that has a bunch of pro-hormones that were digested by endogenous enzymes, or I take the proteins out and digest them [using trypsin as in a proteomics experiment], the next step may be very similar. The first steps of sampling are very different, but the mass analyzers aren't different, so in that sense it's well supported because it's peptide identification.
How about with regard to the informatics on the backend?
That takes more work. We're still using [vendor] software, and some of it needs to be modified. Sometimes we do it manually, but certainly a lot of [commercial proteomics software] can be used.
Is there any commercial software that is specifically for peptidomics?
No, not really.
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.