In August 2016, the scientific community was rocked by the announcement of the untimely passing of Roger Tsien (below), recipient of the 2008 Nobel prize in chemistry, along with Osamu Shimomura and Martin Chalfie, for their discovery and work on the green fluorescent protein*, GFP. It is no exaggeration to say that this one biomolecule has revolutionised imaging techniques within the field of molecular biology. Indeed, a significant majority of graduate students in the biological sciences surely encounter GFP or derivatives thereof on a day-to-day basis in the lab, probably with looks of weariness and resignation.
As someone working in the field of chemical biology, or more specifically on protein modifications, it is probably something of a surprise that I’ve never had cause to use GFP myself (other than a chance encounter with a few badly labelled “GFP-X” tubes stuffed in an old lab freezer drawer…).I have however had the good fortune to encounter Tsien himself, in the context of a lecture at my department (Oxford). It was a high-class talk, and, on a totally unrelated note, had absolutely nothing whatsoever to do with fluorescent proteins. That goes in stark contrast firstly to a large chunk of his career, and secondly to this short article, in which I briefly discuss the history of the discovery, and some of the key impacts it has had.
For this, it is first necessary to journey back – mentally, at least – to 1962, where Osamu Shimomura, a researcher at Princeton University, was studying the jellyfish Aequorea victoria (below, upper). His work culminated in the successful isolation of two proteins from this organism: aequorin, and what we now know as GFP (below, lower). The remarkable feature of the latter is that, as later outlined by Martin Chalfie and Roger Tsien, its fluorescence depends only on the presence of molecular oxygen (which is of course rather ubiquitous); aside from this, it is an entirely intrinsic feature of the protein. In more specific details (but avoiding delving into chemical structures and terminology and so forth) a certain region of the protein is forced into a very tight “turn” structure in its natural form; this in turn causes a reaction which, after subsequent oxidation, forms the requisite chromophore.
Given the lack of requirement of any auxiliary factors for fluorescence that might, for example, be specific to the jellyfish, it would seem that this protein will, if generated, fluoresce automatically, wherever it is produced. Indeed, Chalfie and coworkers subsequently showed that was entirely possible to generate functional GFP in other organisms such as e. coli and c. elegans (which are rather simpler and thus far more useful for genetic engineering/manipulation, protein production and so on than a relatively obscure jellyfish). It was this transition to such model organisms of molecular biology that truly opened up the potential of GFP to this rapidly growing research area.
GFP on its own was not necessarily so useful, albeit it was one of the more aesthetically pleasing proteins around. However, it can be readily fused to another protein of interest through manipulation of the DNA sequence that encodes it, whilst losing none of its own fluorescence properties – the target protein is also typically largely unaffected. This enabled, for the first time, specific proteins to be visualised and tracked dynamically inside live cells under the microscope. Of course, prior to this, generic stains specific for e.g. sub-cellular structures the nucleus existed, but these were only applicable to dead cells.
Thus, GFP allows real-time studies into the localisation, movement (cellular trafficking) and interactions of any protein to which it can be fused. Cellular moieties with which the protein interacts are, by association, visible (see above). These previous sentences should likely be flagged up in bright red, bold typeset – the importance of this development to our understanding of biological systems can hardly be underestimated, for reasons I hope are relatively obvious. It is important to appreciate that if the protein of interest is not being produced in the organism at a given time – that is, the DNA encoding that protein is not being transcribed to RNA and translated to protein – then GFP is also not produced, and as such fluorescence is absent. This enables the use of GFP as a so-called “reporter” for not only where, but also when, a given protein is being produced; it allows for temporal monitoring of gene expression (i.e. protein production).
One could almost certainly write a substantial book on the applications of GFP – indeed, a quick search reveals a number exist – but that’s not the purpose of this article. If Shimomura was responsible for isolation of the protein itself, and Chalfie for its subsequent transferral to “model organisms” such as e. coli, Tsien was largely focussed on improvement of the natural protein. Indeed, his work led to the design of a range of mutant GFP variants exhibiting, for example, improved brightness, or altered spectral properties. This area of research was undoubtedly greatly aided by the GFP crystal structure, solved in 1996 by a group of researchers including Tsien himself. He ventured beyond GFP itself to other fluorescent proteins, in particular designing a wide number based on the dsRED protein from a coral species (discosoma, for those of you interested).
Combined with the GFP mutants, this led to a range of “standard” fluorescent proteins exhibiting a wide variety of colours. This not only looked rather attractive, and indeed has allowed for “bacterial artwork” (see below), but enabled simultaneous tracking and following of more than one protein species in vivo, by fusing each one to differently-coloured fluorescent protein species. In turn, this led to the ability to follow protein interactions through the phenomenon known as FRET (fluorescence resonance energy transfer). Without going into the details (largely due to a lack of complete understanding on the part of this author), this essentially refers to the ability of one fluorophore to transfer energy to an adjacent one, the level of which is highly sensitive to the inverse of the distance between them.
Tsien applied this concept to develop a fluorescence-based calcium sensor. 2 fluorescent protein mutants were linked to a calcium-binding protein, which, upon binding calcium, exhibited a change in conformation that forced the fluorescent proteins closer together, thus increasing FRET efficiency between them. As such, FRET could be correlated to intracellular calcium levels.
Even from this somewhat whistle-stop tour of GFP, I hope its impact has been made at least partially clear. Of course, it benefited from concomitant developments in microscopy itself that allowed its full potential to be realised. The only question mark against it is whether the fusion of a target protein to another fairly large protein (GFP) might itself affect its function and localisation inside cells, i.e. whether what we see is truly representative. To this end, alternative strategies for protein visualisation, such as labelling with small-molecule fluorophores, or incorporation of unnatural fluorescent amino acids, may continue to be developed. However, there is little doubt we will continue to use GFP as a tool in molecular biology, to follow specific biomolecules in live systems and further understand them, for some time, and for this we remain indebted to the work of Tsien, and of Chalfie and Shimomura.
- *For anyone not familiar – proteins represent one of the fundamental biological macromolecules in our cells. They carry out a huge range of functions, from structural to performing chemical reactions. They are encoded by DNA sequences known as genes.