By Tony Fong
Name: Victor Tapia
Position: PhD student and scientific assistant, Molecular Recognition Laboratory, Institute for Medical Immunology of the University Hospital Charité, Berlin, 2006 to present
Background: Scientific assistant, Research Institute for Molecular Pharmacology and the research group Systems Immunology, Institute for Theoretical Biology of the HU-Berlin, 2004 to 2006
If protein arrays are the poor cousins to mass spectrometry in proteomics research, peptide arrays could be described as the stepchild to protein arrays.
In a keyword search of the PubMed database, the number of hits for peptide arrays — 14,535 — was nearly half of the number of hits for protein arrays.
The September edition of Methods in Molecular Biology, which focuses on peptide microarrays, could raise the profile for the technology. In an introductory chapter, Victor Tapia and his colleagues at the Institute for Medical Immunology of the University Hospital Charité in Berlin write that while proteins are "highly informative systems and well suited to investigate protein interactions occurring over extensive complementary surfaces," and protein arrays are "predestined to support proteomic research," factors such as stability, native folding, "or activity of the immobilized proteins" create limitations for protein arrays.
"Peptides, on the contrary, are chemically resistant and have more modest dynamics to achieve their active conformations," the authors write.
The chapter also lays out a brief developmental history of peptide arrays and points out some recent applications of the technology. ProteoMonitor spoke this week with Tapia. Below is an edited transcript of the conversation.
Describe for me the work you're doing with peptide arrays.
My work [focuses on] the production of peptide arrays [that are] macro-sized. That means, for example, on cellulose, in situ. … We [also] make peptides in situ on the same platform, cellulose, and then we keep them and put them on glass slides, which allows much more miniaturization. Everybody calls them microarrays.
I [have] dedicated much work to the description of the precision and accuracy of measurement [that is] achieved from such methods, and my goal is to implement an array of peptides that [would] allow me to make diagnostics of immunologically relevant diseases.
So your work is mainly in the development of these peptides?
Well, that's the goal. … I am not at that stage now. I am just trying to achieve this.
The paper is really a broad history of peptide arrays.
It's two things. … The first part looks at the history of peptide arrays and the focus of the development of peptide synthesis … but I also try to give an overview of what has been done with it and take some technical details.
In the beginning of the paper, there's a suggestion that peptide arrays can be more useful than protein arrays for proteomics research. Can you explain this?
Well, I could not [put it the way] you've stated it. I was much more careful stating [it] because I am not advising people to forget proteins, to immobilize proteins. I was just trying to get people aware of the difficulties of working with proteins in this fashion, because it is not clear that protein structure is so robust [that they can] survive the generation chemistry.
What is the advantage of peptide arrays then? You're suggesting that the structures of the peptides are more robust than that of the proteins.
Yes, exactly. That's the point. [Working with peptides] is much more reductionist than working with proteins and you certainly cannot embrace the whole information capacity that a protein gives you, so working with the whole protein is really something else.
But it's more reliable to work with proteins in a reductionist [manner]. …It is certainly demonstrated that peptides do partially conserve the functions of proteins.
[ pagebreak ]
Are these peptide arrays more suitable for profiling work or for characterizing?
I would say they're not so suited for heavy-duty analysis of protein function, but they're indeed very good at providing and exploring protein function. Let's say, exploring binding with a proteomic set of peptide motifs. That is, you search for a consensus in a [dataset] and see all targets represented in the human proteome, let's say, and print as many of them [as you can] on a chip.
Why are they not suitable for heavy-duty analysis of protein functions?
Because of the gap in information between peptides and whole proteins. …You are working in a reductionist way and you're studying the binding between peptides and protein modules. There isn't much work done in which the modules are printed and generated in situ or freshly prepared and then immobilized on solid-phase platforms.
So this is very interesting but technologically demanding, but you still don't know, for example, if protein context affects the behavior of this protein module.
Why are peptide arrays not widely used, especially in comparison to mass spectrometers, or even in comparison to protein arrays?
I would say that the instrument that you need in the lab to carry out this [work] is very expensive and knowledge demanding. You really have to have a couple of experts on this to produce your peptides and things like this.
It's not very easy to cross the line between biology and chemistry to establish this method. The instruments are very expensive especially if you're working on glass supports. You need a very dedicated lab for this.
Is there a lack of biological knowledge that's holding back this technology?
I think mass spectrometry, for example, and peptide arrays are really not directly comparable. They're really [complementary] methods for studying protein function, so they're not competing at all.
You can study the binding on mass spectrometry but you have to do serious interventions on your probes like cross-linking, which really are not able to, let's say, capture the state-to-state situations of binding. And this is possible on microarrays.
I myself have [shown that it] was possible for us to measure state-to-state binding on a glass peptide microarray.
How difficult was it to do that?
It's not difficult at all. The difficulty is producing the peptide arrays and things like this. When you work with fluorescence detection, the signals are reliable enough to get from your signal intensity your data, pass them through a mathematical model … and then with the value of statistics, you can make a model of your signal intensity and predict affinities.
Or if you know the affinity, you can predict the analyte concentrations. It requires more than one experiment. It requires a few experiments so that you can build an analyte concentration series.
That is, you incubate each replica of your peptide arrays on the different analyte concentrations, and then you can use a very handy mathematical model to predict affinity.
Can you do high-throughput experiments on a peptide array?
Yes, exactly. That's the point. … On glass when you use … resin-based or cellulose-based peptide synthesis, you cleave them off and you can put on a slide … up to I think 6,000 different peptides.
When I spoke with a protein array researcher, he said that one of the barriers to greater adoption of that technology is a kind of herd mentality. Mass spec is the technological flavor of the month kind of thing, so everyone is jumping on that bandwagon. Do you see that with peptide arrays, as well?
[ pagebreak ]
I would not deny this [happens], but I think protein arrays have their own problems, which are not easy to hide. And that is really what I think is the problem — of keeping the sensible fold, the very dynamic fold of proteins, intact after this harsh chemical process for immobilization and things like this.
You know how much care you have to take to maintain your proteins for months … so you can imagine when you have them immobilized with heavy chemicals partially and you leave them dry.
Can you describe any kind of scientific breakthroughs that have been achieved with peptide array technology?
Yes, in the review [I pointed out three instances] and one is the work of [Richard]Jones in 2006. … They made a quantitative protein interaction network for growth factor receptors. And what they did is what I always dreamt of doing. They did it quicker.
They proved that you can really predict affinities … and with this quantitative information derived from microarrays, they were able to draw interaction maps and to make a hierarchy of these interactions and an affinity hierarchy.
So they have another dimension of information with these interactions, weighted interaction networks, you would say.
Another dimension of information that you can't get with protein arrays or a mass spec?
You could do this also with protein arrays. Now, the problem would be that you still have to demonstrate with protein arrays that all your features are biologically active, and that, I think, for protein arrays is a real problem.
What are some of the significant bottlenecks with peptide arrays?
For me the greatest [obstacle] is that you have certain lengths of the peptide that you cannot produce with reliability. I would say more than 40 [amino acids], I would not take seriously unless you have analytical data behind this, which is in confrontation with a high-throughput approach.
You also have some problems with sequences, which are difficult to produce. They are difficult to immobilize … in an oriented fashion. Let's say the enzymes of the peptide only are immobilized and the rest are free. Depending on the sequence, it could be that the immobilization procedure is not so specific, and that certainly would compromise the biological activity.
And how would you overcome this?
You overcome this during the peptide synthesis. … The peptide synthesis is a step-wise synthesis, so you are using building blocks … so when you're done with your sequence, you add some non-native components, or building blocks, which allow you chemoselective immobilization.
Are there any new technologies from other groups that you find particularly interesting?
I find very interesting the possibility to use some affinity capturing methods other than antibodies, for example, functionalizing your platforms with streptavidin. You know this interaction between streptavidin and biotin is very high. It's one of the [strongest] in nature, so using this interaction to immobilize peptides is very interesting.
You immobilize streptavidin molecules on the slide, and you label your peptides with biotin, and then you have a kind of sandwich set where you have glass streptavidin bound to biotin, which is labeled to your peptide.
And then you can incubate with analytes specific for the peptide.
There are several people who have been working on this and I am happy to have them here in Berlin where I am working, so one can learn much from them.