NEW YORK (GenomeWeb) – Researchers from the Max Planck Institute for Brain Research have developed a synthetic biology-based method for selecting particular cell types for proteomic analysis.
Described in a paper published last week in Nature Biotechnology, the approach uses a mutant methionyl-tRNA synthetase (MetRS) controlled by cell-specific promoters to label proteins with a non-canonical amino acid, azidonorleucine (ANL), that can then be pulled down using conventional click chemistry.
Using the approach, the researchers labeled the proteomes of mouse excitatory principal neurons, identifying more than 200 proteins that are differentially regulated in those cells when the subject mice were exposed to increased sensory cues.
The method is generally applicable to a wide variety of cell types, said MPI Professor Erin Schuman, the paper's senior author, and will allow researchers to generate cell-specific protein expression data that has traditionally been challenging to come by.
"This is something people have tried to address with transcriptomics," Schuman said. However, she noted, past research has found poor correlation between proteomic and transcriptomic measurements, making it difficult to use one as a proxy for another. Some recent studies, though, have found that transcripts can serve as effective proxies for protein expression in a gene-specific manner.
The method was developed in Schuman's lab over the course of some ten years. It relies on a mutated form of MetRS the lab created, MetRS L274G, that allows methionine tRNA to substitute ANL for methionine. The researchers developed a mouse line in which expression of MetRS L274G is controlled by the Cre recombinase, and by crossing that mouse line with other lines in which Cre expression is controlled by a cell-type-specific promoter, they are able to label the proteomes of those specific cell types with ANL.
The ANL-labeled proteins can then be pulled down using biotin-streptavidin chemistry and analyzed using mass spec or another technology.
In the Nature Biotechnology paper, Schuman and her colleagues applied the method to several different brain cell types, looking at the proteomes of hippocampus, cerebellum, and glial cells, finding sets of unique and enriched proteins characteristic of each. Using a principal components analysis of the proteomic datasets obtained from these three cell types, they found that they could be clearly separated by their protein profiles, indicating the method's ability to isolate specific cells for proteomic analysis.
They also compared the hippocampal proteomes of mice housed in conventional cages to those of mice kept in an enriched sensory environment. While the majority of their proteomes overlapped, 225 proteins differed significantly between the two groups, among them a number of proteins involved in neuron and synapse function and signaling molecules involved in protein translation and degradation.
The researchers also looked to see if insertion of ANL changed the behavior of the proteins in the labeled cells, Schuman said, noting that metabolic pulse experiments found that the half-life of the labeled proteins didn't differ from that of their unlabeled versions. She added that her lab explored this question more extensively in previous work where they used live-imaging experiments to follow the behavior of labeled proteins. In those experiments, they found, for instance, that ANL-labeled proteins localized to appropriate cellular destinations and otherwise showed normal behavior.
"So we think that at least the majority of the proteins that we've identified appear to function normally," she said.
Schuman said she and her colleagues are now exploring the technique for use in a number of research questions. For instance, they are currently developing strains of mice where they can label all dopaminergic cells, those that produce the neurotransmitter dopamine.
"Dopamine product is important in a lot of diseases and in Parkinson's disease in particular," she said. "We will be able to cross that mouse to any disease model mouse that we want. So we'll be able to, for instance, look at the changes in dopaminergic neurons in Parkinson's [mouse models]. We could even do that in a developmental time course. You could look at, say, how this population of cells changes over time. And this isn't something you could do before."
"It can be used in any kind of system where people are interested in what their specific cell type looks like in a certain disease state versus a normal state," Schuman added. "All of these [cell-specific] proteomes are now accessible, and I think looking at them at the interface of disease will be very interesting."
Her group is also continuing to refine the labeling technique, which Schuman said remains fairly inefficient. While the mutant MetRS L274G is able to use ANL instead of methionine, it still prefers to incorporate methionine into peptides if it is available.
With this in mind, she and her collaborators are exploring whether additional mutations to MetRS could produce a synthetase that would have a stronger preference for ANL than the L274G mutant.
Another challenge is to optimize expression levels of the mutant MetRS L274G molecule, Schuman said. "You want a high enough level of expression … but in some experiments when you overexpress something, the cell has mechanisms for turning down expression. So finding that sort of sweet spot where you maximize expression without invoking the cellular defense mechanisms is one challenge."
Countering the inefficiency of the labeling process is the fact that, in theory, the MetRS L274G needs to replace just one of a protein's methionines with ANL for the researchers to then pull it down.
"In, say, a 200-amino acid protein, we only need one of the methionines to be labeled with ANL in order to get that protein," Schuman said. "We can be sensitive enough to detect changes of around a 30 percent difference, which is, I would say, pretty sensitive, considering that most pathological states often involve doublings of things. So I think [the method] can be quite sensitive."