University of Illinois
Name: Neil Kelleher
Position: Professor, chemistry department, University of Illinois, 2006 to present
Background: Postdoctoral associate, Harvard Medical School, 1997 to 1999; Research assistant, Cornell University with Tadhg Begley and Fred McLafferty, 1993 to 1997; PhD in bioanalytical chemistry, Cornell University, 1997
Neil Kelleher and his colleagues have developed a platform to analyze histone H3, allowing for comprehensive characterization of histone codes. Their method, drawing on top-down proteomics, uses hydrophilic interaction chromatography and Fourier transform mass spectrometry. Using the platform, Kelleher and his co-researchers found 170 distinct codes of histone H3. Their research appears in the June 1 edition of Nature Methods.
Below is an edited version of a conversation ProteoMonitorhad with Kelleher last week on his work.
Why was the top-down approach so important to your research?
Because it detects modifications in combination with others, which is the hypothesis around the histone code.
[In a bottom-up approach] you disconnect the modifications from each other by cleaving the peptide bond, so you lose the information that will tell you the relationships between modifications.
Describe what you did.
[Our work] is focused on histone H3, the most complicated of all the histones. And we detected 170 distinct codes or molecules of histone H3 that float around inside cells. So for the first time we have a very clear estimate of the molecular diversity created through combinatorial post-translational modification.
Some of the estimates were in the thousand or tens of thousands of forms. And we found 170, so it clarifies a very complex measurement problem.
What techniques did you use?
It’s a combination of chromatography and mass spectrometry [involving] four-dimensional separation. The first stage is reverse phase liquid chromatography.
The second dimension is something called HILIP chromatography [for hydrophilic interaction chromatography]. That separates based on post-translational modifications. The third dimension is high-resolution Fourier transform mass spectrometry. And the fourth dimension is high-resolution tandem mass spectrometry.
And that was using an FTMS or some other type of mass spec?
FT; we’re an FTMS lab.
When you first used the top-down approach the mass spectra was still too complex for complete characterization of the modifications, right?
On H3, that’s right. We chopped the proteins from a 15-kilodalton protein to a 5-kilodalton peptide, the first 50 amino acids where most of the action is. And we analyzed that piece almost exhaustively from about 1 million human cancer cells.
Where did you get your tissue?
It was a cell line made from a cancer; it wasn’t a primary cancer tumor.
Top-down proteomics have been used by other people. How was what you did different in terms of your approach or technique?
I’d say it’s a combination of lots of little innovations. We have software that interprets the data quickly. We have electron capture dissociation, which was invented in 1997. There are a lot of little things that aren’t especially novel, but they were all combined to realize this advance.
There are only three modification types — well, there are more than three — but there’s methylation, phosphorylation and acetylation. And [we found] 170 distinct forms, [or] combinations of those different modification types.
How many of these were novel?
The answer is none and all. None in the sense that all the modifications we found were known in isolation. What we did was tell the world what the combinations of the modifications are, how they populate histone H3 in human cells. It’s just establishing the basis set of expressed human H3, what’s there. It’s actually quite simple in that regard.
What kind of conclusions can you draw from your work?
You can start to look at the combinations of the modifications and what links one modification to another. For example, methylation of lysine 4 leads to acetylation on the same histone tail. So there’s a linkage between methylation at lysine-4 and acetylation.
And this has been guessed at through lots of work over years in chromatin. This is sort of a protypical example, and now we can quantify and show the direct relationship between these two types of modifications.
The histone code is a hypothesis. We don’t know if there really is such a thing.
Yes and no, again. Yes in the sense that there is clearly a link between some modifications and others. So the question is the extent to which it is true, and the extent to which it is like a computer code and very prescriptive. Or is it sort of a loose suggestion?
Clearly modifications are important, then you get into the argument about how rigorous of a code it is.
In terms of verifying the existence of such a code, what is the significance of your work?
It doesn’t resolve the question, but it does prove some relationships like the one I just discussed about methylation on lysine-4 and aceylation, and it directly supports that linkage. So it puts more eggs into the basket of a more rigorous interpretation of a code.
And what does it mean in terms of cracking the code, if it exists?
It’s a method [that] needs to be applied now to lots of different systems, stem cells, progression of cancer, and other differentiation models. It needs to be applied to lots of different areas in biology, and together with complementary, existing approaches, will answer the big question in time.
Who was the manufacturer of the FTMS you used?
We used a custom instrument to do that work. It was in collaboration with Thermo Fisher Scientific.
How many Tesla on the magnet?
Would it make any difference it you increase that to 15 Tesla?
We have two 12 Tesla [instruments]. … It would make a little difference, but it wouldn’t change the outcome of the study.
What’s next? Are continuing this method to identify additional combinations of modifications?
Right, and then look at the biology of the modifications, the function. We’re applying it in stem cell biology, and cancer, epigenetics.