Abstract The green fluorescent protein (GFP) from the jellyfish Aequorea victoria has proven to be a powerful tool in plant genetic transformation studies. This paper reviews the history and the progression of the expression of GFP variants in transgenic plants. The distinguishing features of the most useful GFPs, such as those including the S65T chromophore mutation and those with dual ex- citation peaks, are discussed. The review also focuses on the utility of GFP as a visual selectable marker in aiding the plant transformation process GFP has been more im- portant in monocot transformation compared with dicot transformation. Finally, the potential utility of new fluo- rescent proteins is speculated upon. Keywords Fluorescent proteins �� GFP �� Gene expression �� Marker genes �� Plant transformation Introduction The jellyfish green fluorescent protein (GFP) has be- come a very effective marker for use in plant genetic transformation research. For the first time in plant biolo- gy, researchers had at their disposal a universal, in vivo, and real-time transgenic visible marker in GFP. Several review papers have been written about the uses of GFP in plant biology (Haseloff and Amos 1995 Leffel et al. 1997 Haseloff and Siemering1998), however the present review focuses on the expression of different Aequorea victoria GFP variants in transgenic plants with the spe- cific purpose of assessing their utility in plant transfor- mation research. The historical development of GFP variants will be reviewed. In addition, the trends towards mutational variant development of Aequorea GFPs will be extrapolated to speculate on the usefulness of recently cloned non-Aequorea GFP genes in plant transformation. Early variants of green fluorescent protein The utility of GFP in plant transformation and expres- sion was first demonstrated in plant cells, not intact transgenic plant tissues. Niedz et al. (1995) were the first to show that wild-type Aequorea GFP could be visual- ized in plant cells ��� in this case sweet orange (Citrus sin- ensis) protoplasts. There were two other published stud- ies using wild-type GFP. Hu and Cheng (1995) demon- strated that GFP could be synthesized in corn protop- lasts. However, they failed to observe GFP in trans- formed Arabidopsis thaliana or tobacco cells, presum- ably the result of low expression of the wild-type gene. Using a stronger promoter, another group was able to vi- sualize wild-type GFP in corn and arabidopsis cells (Sheen et al. 1995). Both the latter two groups used heat shock promoters to attempt to drive GFP with inducible expression as well Sheen et al. (1995) were successful, while Hu and Cheng (1995) were not. The difference in the promoters used for the experiments likely do not ex- plain the disparate results ��� these are more likely due to the excitation source: laser (Sheen et al. 1995) versus in- candescent lamp with excitation filters (Hu and Cheng 1995). These studies show that successful GFP detection is highly dependent on the strength and source of the ex- citation sources. Nonetheless, the experience with a low expression of wild-type GFP encouraged researchers to modify it to forms that could be more effectively synthe- sized in plants. Haseloff et al. (1997) reported that a cryptic intron existed in the wild-type Aequorea GFP that caused a ab- errant splicing in plant cells between nucleotides 380 and 463, thereby creating an 84-nucleotide intron. When the cryptic splice sites were altered with silent mutations, a variant called mGFP4 was produced (Haseloff et al. 1997) that had essentially wild-type spectral characteris- tics: maximal excitation at 395 nm and maximal emis- Communicated by A. Komamine C.N. Stewart Jr (���) Department of Biology, University of North Carolina-Greensboro, NC 27402-6174, USA e-mail: nstewart@uncg.edu Tel.: +1-336-3344980, Fax: +1-336-3345839 Plant Cell Rep (2001) 20:376���382 DOI 10.1007/s002990100346 REVIEW C.N. Stewart Jr The utility of green fluorescent protein in transgenic plants Received: 17 November 2000 / Revision received: 16 March 2001 / Accepted: 9 April 2001 / Published online: 7 June 2001 �� Springer-Verlag 2001

377 sion at 509 nm. mGFP4 was successfully expressed in soybean suspension cultured cells (Plautz et al. 1996), arabidopis (Haseloff et al. 1997), tobacco (Stewart 1996) and other plants. However, several researchers reported that mGFP4 was not very stable in its fluorescence, espe- cially under field conditions, even though it was ex- pressed in the plant at levels that should have yielded visible green fluorescence (Stewart 1996 Leffel et al. 1997 Harper et al. 1999). A similar synthetic human co- don-optimized GFP with a wild-type chromophore was created by Haas et al. (1996) that also eliminated the cryptic intron. Since humans and corn have very similar codon usage, the gene proved to be well expressed in plants. When it was expressed in plants it yielded 20 times more fluorescence than the wild-type gene (Chiu et al. 1996). Common GFP variants The various GFP mutants that are most commonly used for plant transformation experiments are shown in Ta- ble 1. Haseloff's group also introduced mutations that con- ferred increased GFP heat stability and altered spectral qualities. The V163A and S175G mutations proved to provide better folding and hence green fluorescence at 37��C (Siemering et al. 1996). Coupled with the I167T mutation (Heim et al. 1994), the variant, mGFP5, has du- al excitation peaks at 395 nm and 475 nm and an emis- sion peak at 509 nm (Siemering et al. 1996). The variants mentioned above thus far retained a wild-type chromophore of SYG (peptides 65���67) and, therefore, an unaltered emission spectrum from wild- type GFP. The most important chromophore alteration to plant biology has been the S65T change that created a single blue excitation peak (489 nm optimum) and red- shifted the excitation optimum to 511 nm (still green) and also the less often-used S65C mutation (Heim et al. 1995). When in an essentially mGFP4 background (few codons changed, cryptic intron removed), the S65T and S65C mutations increased detection limits by up to 19- fold (Reichel et al. 1996). Chiu et al. (1996) demonstrat- ed that the S65T mutation provided a fluorescence gain of up to 100-fold in plant cells after human codon opti- mization was performed. The synthetic S65T gene with the cryptic intron removed was called sGFP (S65T) (Haas et al. 1996). Harper et al. (1999) demonstrated in the field that the sGFP-S65T gene was expressed up to 0.2% and had strong fluorescence characteristics. Not all mutants have yielded increased fluorescence. A sub-opti- mal change is the Y66H, which makes a blue fluorescent protein (BFP) that does not fluoresce well in plants (Reichel et al. 1996). Several other important modifications have been made to improve GFP expression in plants. Pang et al. (1996) produced a synthetic (pgfp) S65T and S65C vari- ants, with versions with and without the substitution of a potato ST-LS1 intron 2 in place of the cryptic intron. The effect of adding the intron boosted the fluorescence of the synthetic S65T/S65C versions 150-fold compared with that of the wild-type GFP. Other GFP variants have been expressed in plants. The commercially available EGFP (Clontech) has the S65T as well as the F64L and Y145F mutations and is human codon-optimized (Yang et al. 1996). Gene shuf- fling was used to produce a mutant that has greater solu- bility and fluorescence (Crameri et al. 1996). The mut3GFP has the V164A (V163A) mutation that putative- ly improves folding at higher temperatures (Siemering et al. 1996) as well as the F100S (F99S) and M154T (m153T) mutations. Davis and Vierstra (1998) further modified mGFP4 to include the mut3 mutations and called it smGFP (sm = soluble, modified). In addition, they added the S65T mutation, in yet another variant, Y66H. They called these smRS-GFP (RS = red-shifted) and smBFP, respectively. Surprisingly, when expressed in arabidopsis, there was no difference found in fluores- cence between smGFP and smRS-GFP, but both were improvements over mGFP4. smBFP, like all BFPs de- scribed to date, had little fluorescence. smGFP, like mut3, showed primarily UV excitation (with a minor blue peak) and smRS-GFP had an excitation maximum of 495 nm and emission maximum at 507 nm. Table 1 The various GFP mutants that are most commonly used for plant transformation experiments. The mutations listed are amino acid substitutions in standard format (for example: S65T = serine to threonine at the 65th amino acid) Mutant (source) Excitation/emittance (nm) Mutationsa Wild-type Aequorea victoria 395b, 475/507 None mGFP4 (Haseloff) 395b, 475/509 CI, SDM mGFP5 (Haseloff) 395, 475/509 CI, V163A, S167T, S175G, SDM SGFP S65T (Chiu) 489/511 CI, S65T, synth PGFP (S65T) (Pang) 489/511 CI, S65T, synth, intron added EGFP(Clontech, Yang) 488/507 CI F64L, S65T, Y145F, synth smGFP (Davis and Vierstra) 395b, 475/509 CI F99S, M153T, V163A, SDM smRS-GFP (Davis and Vierstra) 490/510 CI S65T, F99S, M153T, V163A, SDM a CI The cryptic intron in the wild-type gene has been altered SDM site-directed mutagenesis has been performed synth the gene is codon-optimized (humanized) b Major excitation peak where more than one excitation peak is present