GREEN FLUORESCENT PROTEIN

The green revolutionGreen fluorescent protein allows gene expression and

protein localization to be observed in living cells.

From Leuwenhoek's light microscope to the develop-ment of the electron microscope, immunofluorescenceand video microscopy, advances in cell biology haveoften come about through the development of new tech-nologies that allow us to see better what is happeninginside cells. I will give two examples. First, gene expres-sion has progressed from something that could only beassayed for a population of cells in a test tube, as studiedby Jacob and Monod, to something that can be assayed atthe single cell level through the use of in situ hybrid-ization and immunofluorescence techniques. This hasallowed the study of the developmental biology of multi-cellular organisms at a new level of detail. Second, theability to observe the behavior of molecules within liv-ing cells by video microscopy has changed the way thatwe think of cells. For example, the cytoskeleton wasuntil recently thought to be a static structure, a 'skel-eton' seen in fixed samples. Video microscopy hasdemonstrated that the cytoskeleton is actually a highlydynamic structure, changing during the cell cycle, cellu-lar morphogenesis and development, and in response toenvironmental cues.

In part, this revolution in understanding has come aboutthrough the ability to isolate the major structural pro-teins of the cytoskeleton, to covalently label these pro-teins with fluorescent molecules, and to introduce thelabeled proteins into cells. This has allowed the directobservation of cytoskeletal dynamics by fluorescencemicroscopy. Although this technique could, in theory, beapplied to any protein of interest, it has several limita-tions. It requires that the protein be purified in quantity,that the protein maintain its function after covalentmodification with the fluorescent molecule, and that the

Fig. 1. Expression of GFP under the control of the mec-7promoter illuminates two neurons in the worm C. elegans.

labeled protein be introduced into cells - typically bymicroinjection - potentially perturbing the cells andmaking any interpretation of observed effects difficult.

Clearly, what is needed to overcome these problems is afluorescent genetic tag that can be added to a codingsequence, or placed behind a promoter of interest, mak-ing a fluorescent product when expressed. Just such amolecule, green fluorescent protein (GFP), has recentlyappeared on the scene. GFP comes from the jellyfishAequorea victoria and is naturally fluorescent; whenexcited by light of one wavelength, it emits light ofanother wavelength. In fact, many jellyfish emit greenlight when disturbed (often seen when riding in a boat atnight). In Aequorea, the green fluorescence is due to theaction of two proteins, aequorin and GFP.

Aequorin is a Ca2 +-activated photoprotein that excitesGFP by an unknown mechanism, making GFP emitgreen light; aequorin itself has been used as a sensor ofintracellular calcium [1]. But aequorin is not an essentialpart of the green-light-producing reaction: purified GFPcan be made to emit green light by exciting it directlywith blue light. How does GFP manage this feat of fluor-escence? The gene for GFP was cloned recently [2]; theencoded protein consists of 238 amino-acid residues(molecular weight 27 kD) and it is present as a monomerin solution. Unlike some other fluorescent proteins, thechromophore responsible for GFP fluorescence is com-pletely contained within the coding sequence of the pro-tein. The chromophore is a cyclic derivative of thetripeptide serine-dehydrotyrosine-glycine [3]. Remark-ably, GFP expressed in Escherichia coli has the same spec-tral properties as the natural protein [4], demonstratingthat the GFP chromophore can form without the assis-tance of any Aequorea enyzmatic activities. GFP absorbsblue light, maximally at a wavelength of 395 nm withweaker absorbance at 470 nm, and emits green light at509 nm. Excitation with 395 nm light results in rapidphotobleaching of GFP fluorescence, whereas excitationwith 470 nm light results in relatively stable fluorescence.

GFP thus has all of the right properties to make it auseful fluorescent tag: its fluorescence is apparently notspecies-specific, it does not require any unusual co-factors, the protein is relatively small, and it does notform multimers that might interfere with the function ofthe protein to which it is fused. The first use of GFP as acell biological tool was as a marker for gene expression inthe nematode Caenorhabditis elegans [4]. The GFP codingsequence was placed under the control of a promoter for

© Current Biology 1995, Vol 5 No 3262

Tim STEARNS

DISPATCH 263

Fig. 2. Expression of GFP andGFP-tubulin in yeast cells. Yeast cellsexpressing either (a,b) GFP or (c,d)a-tubulin-GFP fusion protein, viewedby (a,c) fluorescence microscopy or(b,d) Nomarski microscopy. GFP is dis-tributed throughout the cytoplasm,whereas the ca-tubulin-GFP is localizedto microtubule structures.

the mec-7 tubulin gene, which is expressed only in asubset of neurons. In worms expressing this regulatedGFP, the pattern of fluorescence in vivo was similar tothat previously seen with antibodies against Mec-7 pro-tein. GFP filled the cytoplasm of the expressing cells,making it possible to see the cell bodies and the processesof growing neurons (Fig. 1). The GFP-expressing wormswent on to become viable adults, indicating that GFP isnot toxic.

The experiments with C. elegans established that GFP isuseful as a marker of gene expression, and it is is now inwide use for this purpose in the community of C. elegansresearchers. The requirements for a fluorescent tag that isuseful for following proteins inside cells are more strin-gent however; it must be able to function when fused toother proteins, and it must not interfere with the func-tions of the proteins to which it is fused. Wang andHazelrigg [5] demonstrated that GFP can fulfil bothrequirements. They fused the GFP coding sequence tothat of the Drosophila exuperantia (exu) gene. The Exuprotein is required for the localization of specific messen-ger RNAs in the developing oocyte. GFP fusions toeither the amino terminus or the carboxyl terminus ofExu resulted in a fluorescent protein that could comple-ment the phenotypic defect of a null exu allele. Thelocalization of the fluorescent Exu-GFP fusion proteinwas similar to that of the native Exu protein, as deter-mined by immunofluoresence. But the sensitivity ofdetection afforded by the GFP tagging was greater thanthat of immunfluorescence and allowed the authors toidentify new sites of Exu localization.

These experiments demonstrate that GFP can be usedfor both transcriptional and translational fusions - its

expression can be driven by a promoter of interest, or itcan be fused directly to the protein product. The realpotential of GFP, though, is in studying the dynamics ofevents in living cells. Although not yet published, thereare many anecdotal accounts of GFP translational fusionsbeing used to study the distribution of proteins andstructures in cells over time. In my laboratory, for exam-ple, we have fused GFP to a yeast tubulin gene. Thetubulin-GFP fusion protein is fluorescent and is incor-porated into microtubules, allowing visualization of themicrotubules in living yeast cells for the first time (Fig.2). Because the GFP chromophore is relatively resistantto photobleaching under the appropriate illuminationconditions, it has been possible to use video microscopyto observe microtubules over extended periods and todiscover new aspects of their function. Conversely,because the GFP chromophore can be made to bleachrapidly under different illumination conditions, it is alsopossible to perform marking experiments, whereby thefluorescence of a subset of GFP molecules is inactivated.As an example, one can photobleach a portion of amicrotubule, distinguishing it from the total population,and observe specifically the fate of that portion.Although they present some difficulties, such photo-bleaching experiments can provide unique informationabout the relative movement of elements within the cell.

If there is a result that could temper the enthusiasm forGFP, it would be that GFP activity is not completelyindependent of the cell type in which it is beingexpressed. Although strong fluorescence has been ob-tained in bacteria, yeast, worms and flies, there are manyreports of failure in mammalian cells. It is not yet clearwhether the defect lies in the expression of theprotein, its stability, or the formation of the chromophore.

264 Current Biology 1995, Vol 5 No 3

The generation of antibodies against GFP, or the creationof epitope-tagged versions of it, should make it possible todistinguish between these possibilities. It may be necessaryto develop mutant versions of GFP in order for it tofunction efficiently in mammalian cells.

As is the case for many technological advances, it takesonly short reflection to come up with many possibleapplications for GFP in molecular biology. Some of thepotential uses for GFP that are currently being pursuedinvolve modifications of existing techniques that rely onexpression of, for example, the bacterial enzyme lacZ anddetection of its expression in fixed cells treated with achromogenic substrate. For example, GFP is being usedas an expression reporter in the two-hybrid 'interactiontrap' gene expression system; a fluorescence-activated cellsorter (FACS) can be used to sort out expressing cellsfrom a non-expressing population. GFP can be used as afluorescence tag in a random cDNA library; GFP fusionproteins that localize to intracellular places of interest canbe identified by fluorescence microscopy. Other uses forGFP will be developed as we learn more about its fluo-rescent properties. For example, it might be possible totake advantage of photochemical changes in the proteinthat occur during excitation to make photoactivatableversions of GFP. These would become fluorescent only

after activation with light of the appropriate wavelength.An exciting recent paper demonstrates that the absorp-tion and emission spectra of GFP can be altered in usefulways by mutagenesis, resulting in mutant proteins thatemit different coloured light, or are brighter than thewild-type protein under certain illumination conditions[6]. Clearly the greening of cell biology is just beginning.

References1. Speksnijder J, Corson DW, Sardet C, Jaffe LF: Free calcium pulses

following fertilization in the ascidian egg. Dev Biol 1989,135:182-190.

2. Prasher DC, Eckenrode VK, Ward WW, Prendergast FG, Cormier M:Primary structure of the Aequorea victoria green-fluorescent pro-tein. Gene 1992, 111:229-233.

3. Cody CW, Prasher DC, Westler WM, Prendergast FG, Ward WW:Chemical structure of the hexapeptide chromophore of theAequorea green-fluorescent protein. Biochemistry 1993 32:1212-1218.

4. Chalfie M, Tu Y, Euskirchen G, Ward WW, Prasher DC: Greenfluorescent protein as a marker for gene expression. Science 1994,263:802-805.

5. Wang S, Hazelrigg T: Implications for bcd mRNA localization fromspatial distribution of exu protein in Drosophila oogenesis. Nature1994, 369:400-403.

6. Heim R, Prasher DC, Tsien RY: Wavelength mutations and post-translational autoxidation of green fluorescent protein. Proc NatlAcad Sci USA 1994, 91:1250112504.

Tim Stearns, Department of Biological Sciences, StanfordUniversity, Stanford, California 94305-5020, USA.