Passive fluorescent protein markers are indispensable for dynamic cellular imaging; however, they are unselective, introduce constant background fluorescence, and require continuous observation. Photoactivatable fluorescent proteins have now been developed whose fluorescence can be switched on and off by illumination, allowing selective and direct tracking of tagged objects without the need for continuous imaging. The “kindling fluorescent protein” is a photoactivatable marker with a novel twist: it turns itself off after a selectable period.
Although the green fluorescent protein (GFP) revolution began over a decade ago with the molecular cloning of the Aequorea victoria GFP (40), there is no slackening in the pace of discoveries and novel applications. The popularity and ubiquitous use of GFP stems from spontaneous chromophore formation without the need for accessory enzymes (9, 24), high fluorescence quantum yield (ϕF), stability toward photobleaching, proteolysis and extremes of pH, and ease of protein fusion due to its monomeric fold and accessible NH2 and COOH termini. On the other hand, the emission characteristics of GFP are not well suited for many applications. The 510-nm emission is difficult to separate from blue cellular autofluorescence (37) and penetrates tissue poorly. The required violet (380 nm) or blue (475 nm) illumination tends to be cytotoxic. The apparent steady-state emission of GFP is more complicated than was initially apparent and depends on illumination history. In extremely dilute solution, GFP molecules are observed to blink or have long-lived on and off states, with time scales ranging from microseconds to hours (15).
In efforts to overcome the cellular autofluorescence problem, mutagenesis of GFP led to variants that emit in the range of blue to yellow; however, many of these variants suffer from low emission or high sensitivity to environmental conditions (14, 22, 49, 54). Thus it was the discovery of yellow-and red-shifted GFP homologs in coral reef organisms of the class Anthozoa (33) that provided for the creation of the next generation of fluorescent protein (FP) markers. The diversity of colors available from Anthozoa FPs creates new opportunities for multicol-or labeling and fluorescence resonance energy transfer (FRET) applications (52). As expected, red emission reduces the problems associated with background autofluorescence, and longer wavelengths penetrate cellular tissues much more effectively.
Also as expected, the molecular architecture of the Anthozoa proteins is identical to that of GFP; however, the autocatalytic pathway of chromophore formation introduces several new steps, beginning with a GFP-like intermediate (18), as shown in FIGURE 1A⇓. The intermediate chromophore is further oxidized by molecular oxygen (to an acylimine) such that the protein backbone extends the conjugation of the chromophore. The acylimine is an unstable species in aqueous solution, and apparently much of the diversity of emission color results from subsequent chemical reactions within the protein to produce novel chromophore variations, including cleavage of the polypeptide backbone (see below) or additional cyclization reactions.
Despite their promising attributes, the application of Anthozoa FPs as labels is not without technical problems. These proteins form obligate oligomers (usually tetramers), which is a serious obstacle for labeling individual molecules or cellular structures. Oligomerization of the fluorescent tag very often disrupts the function and/or localization of fusion partners and may even result in cell death. Furthermore, since the GFP-like chromophore seems to be the default in the maturation pathway, many red-shifted Anthozoa FPs exhibit slow and incomplete maturation, resulting in mixed red and green emission (5, 43, 51). However, in one case, slow maturation has been effectively exploited to create a time-dependent multicol-or marker (the so-called “fluorescent timer” protein; Ref. 47). More recently, combinations of directed evolution and rational and random mutagenesis have been used to eliminate many of these undesirable properties. The effort has resulted in rapidly maturing monomeric red FPs (mRFPs) with high brightness (the product of the extinction coefficient and quantum yield) and an amazing variety of emission colors (8, 17, 45, 56).
The primary subject of this review is a member of a closely related but distinct class of Anthozoa proteins (7, 19, 30) that strongly absorb visible light but are nonfluorescent and hence referred to as chromoproteins (CPs). Of particular interest to us is the deep purple-colored CP asFP595, which was cloned from the sea anemone Anemonia sulcata and is responsible for the purple coloration found in the tips of the animal’s tentacles (30). Wild-type asFP595 is essentially non-fluorescent. Remarkably, on illumination with green light in the range of 540–560 nm, the protein becomes transiently fluorescent, a process termed “kindling” (12). When green light illumination ceases, the fluorescent (kindled) state relaxes thermally back to the nonfluorescent state or it can be instantly quenched by illumination with blue light in the range of 450 nm (12), allowing full control over the fluorescent signal. asFP595 is commonly known as the “kindling fluroescent protein” or KFP (11) and is now recognized to be but one example of a new and growing class of photoactivatable FPs (PAFPs), i.e., proteins whose fluorescence properties can be controlled by illumination with light of appropriate wavelength and intensity.
Photoactivatable FPs as Research Tools
Although passive FP markers are invaluable for applications such as determining subcellular protein localization, their continual turnover (synthesis and degradation) within the cell limits their utility in measuring temporal expression patterns or motility of fusion partners. With conventional fluorescent labels, this problem has been addressed by kinetic microscopy techniques such as fluorescence recovery after photobleaching (FRAP) or fluorescence loss in photobleaching (FLIP), which perturb the steady-state fluorescence signal in a defined region of the cell (25, 28).
PAFPs offer significant advantages over photobleaching techniques. For example, photobleaching suffers from errors introduced by the continual turnover and synthesis of new (unbleached) proteins. With photoactivation, one can use a confocal microscope to selectively mark a subpopulation of PAFP molecules with illumination at a specific wavelength and then directly follow the signal of the photoactivated population, thus providing higher contrast between signal and background. Furthermore, photoactivation typically has much higher quantum efficiency than photobleaching and often requires shorter periods of activating illumination at longer (and hence less cytotoxic) wavelengths. Thus PAFPs offer improved temporal resolution over photobleaching techniques.
PAFPs can be grouped into three general classes based on their mechanism of activation. In one class, photoactivation with ultraviolet (UV) to violet light gives rise to a large increase (~100- to 300-fold) in green fluorescence (~510–520 nm) on excitation with blue light (~460–480 nm). Examples include the photoactivatable T203H mutant of GFP (PA-GFP; Ref. 38) and the photoswitchable cyan FP (PS-CFP) derived from the colorless, monomeric Aequorea coerulescens protein (13). Before photoactivation, PA-GFP fluorescence is dim when illuminated with blue light; however, once activated, the fluorescence signal is increased 100-fold. PS-CFP switches emission from predominantly cyan (468 nm) to predominantly green (511 nm) on photoactivation, yielding a 300-fold increase in the green emission and a 5-fold decrease in cyan emission for a 1,500-fold change in the optical contrast. Activation of these PAFPs is irreversible and widely suspected to involve decarboxylation of a conserved glutamic acid residue, a process that is known to occur in wild-type GFP (6, 38, 50).
The high optical contrast of these markers on photoactivation can be exploited to follow movements of tagged proteins, organelles, and cells over broad time scales in what amount to optically labeled pulse-chase experiments. For example, photoactivation of discrete cellular locations (e.g., the cytoplasm or nucleus) in cells expressing PA-GFP has been used to measure the rate of passive diffusion across the nuclear envelope (38, 48). As an alternative methodology to previous FLIP studies (16), photoactivation of PA-GFP has demonstrated dynamic nucleoplasmic shuttling of the catalytic α-subunits and regulatory β-subunits of the CK2 kinase (48). The essential role of this kinase in the regulation of signal transduction pathways and the inferred existence of mixed populations of CK2 subunits in different cellular compartments suggest an important balance in its temporal expression, localization, and assembly. By photoactivating a cytosolic subpopulation of PA-GFP-CK2α transfected into NIH 3T3 cells, it was shown that this chimeric protein equilibrates with the nucleus in ~15 min compared to 5 min for PA-GFP alone (48), arguing for active transport of this subunit. In contrast, PA-GFP-CK2β active transport is a much slower process and suggests independent nuclear import mechanisms of the CK2 subunits.
In another example, exchange of membrane protein components has been investigated using PA-GFP-lgp120 (a lysosomal membrane protein) and PS-CFP-hDAT (human dopamine transporter) chimeras. These experiments demonstrated microtubule-dependent exchange of membrane components between lysosomes (38) and direct exchange of cargo proteins between endosomes (13), respectively. Although the monomeric nature of PA-GFP and PS-CFP make them particular well suited as fusion partners, their high contrast and irreversible activation also present opportunities to study cellular migratory behavior and lineage in developing organisms (27, 46).
A second class of PAFPs cloned from reef organisms irreversibly switches from green to red fluorescence on illumination with UV to violet light, a process termed photoconversion. Examples include the so-called “Kaede” and “EosFP” proteins (1, 34, 36, 57). Both are tetrameric and are thus not well suited for applications as fusion partner. However, the green fluorescence, which is excited by blue light far removed from the UV required for photoconversion, allows for identification of labeled structures before photoconversion. Subsequent to photoconversion, the resulting ~2,000-fold contrast in red fluorescence is advantageous for studying organelle and cell dynamics (1, 4, 35). In one interesting application, Kaede-containing mitochondrial import sequences have been used to demonstrate fusion of mitochondria in higher plants (converted red fluorescent and green fluorescent mitochondria combine to become yellow; Ref. 4).
The third class of PAFPs, of which KFP is a member, includes the recently developed monomeric green PAFP, “Dronpa” (2, 20). Like KFP, Dronpa is a photoswitchable label that can be turned on or off using appropriate illumination wavelengths. Again, the activated state is long-lived and is completely reversible. In the fluorescent state, the absorbance maximum of Dronpa is 503 nm (indicative of an anionic chromophore) with emission at 518 nm (ϕF = 0.85). Activated Dronpa is readily visualized using the standard 488-nm line of an argon-ion laser. Moderate illumination of Dronpa causes photobleaching (quantum efficiency = 0.00032) and switches the absorbance maximum to ~400 nm, indicative of a protonated chromophore. Relatively weak illumination at 400 nm (quantum efficiency = 0.37) efficiently reverses the process. In an ingenious application, a Dronpa fusion was used to study nuclear import and export of the extracellular signal-regulated kinase (ERK), which is inefficient in the absence of extracellular factors. In the experimental protocol, overall cellular fluorescence was first erased by using 488-nm illumination, then specific regions such as the cytosol or nucleus are reactivated with a short 405-nm flash (2). In the presence of epidermal growth factor, repeated cycles of Dronpa de- and reactivation demonstrated that the rate of nuclear exchange of ERK increases in a time-dependent manner (2).
Finally, novel photoactivatable mRFPs (PA-mRFPs) derived from mRFP (53) are under development, and progress in this area can to some extent be attributed to earlier efforts to understand the basis of KFP photoactivation (7, 12, 19). The rapid progress in PAFP and imaging technologies have paved the way for exciting new possibilities for monitoring cellular dynamics on a broad range of spatiotemporal levels. The PAFPs described above differ in many aspects of their photophysical behavior and suitability for specific applications. These topics were recently discussed in two excellent reviews (10, 29), to which the interested reader is referred.
The KFP as a Marker
Currently, KFP has yet to find widespread use for two reasons: obligate tetramer formation and low brightness. However, as mentioned above, oligomerization problems were overcome during the development of monomeric mRFPs, which have in turn been used to develop PA-mRFPs (53), so there is good reason to believe that the same approach will be successful with KFP. Drawbacks aside, KFP has some potentially useful properties that the currently available reversible PAFPs, such as Dronpa and PA-mRFPs, do not offer. One such property is photoactivation by green light (540–560 nm), which is less toxic to cells than the activation wavelengths used for Dronpa and PA-mRFPs (400 and 360 nm, respectively). A second useful property is the spontaneous but slow reversion to the nonfluorescent state, the lifetime of which can be controlled by mutagenesis or variations in temperature. Variants of KFP at a single amino acid position exist with fluorescence half-lives in the convenient range from 7 to 200 s (3, 11, 12). Furthermore, one of these variants, A143G (known as KFP-1), is reported to become irreversibly fluorescent on prolonged illumination with intense green light (11), suggesting that an extremely wide range of fluorescent state half-lives is achievable. The tetrameric nature of KFP makes it best suited for organelle or cellular tracking, whereas red fluorescence and the possibility of irreversible activation will be useful in tissues where depth of light penetration is a limiting factor. For example, irreversible activation of KFP has been successfully used to study embryo development in Xenopus laevis and to track the movements of mitochondria in mammalian cells (11).
In the case of reversible photoactivation, optimized KFPs could be designed to measure diffusion rates of cellular objects and to monitor exchange of factors between cellular compartments, as has been demonstrated with Dronpa fusions to the ERK (2). One could in principle tailor the half-life of the spontaneous inactivation for a particular experiment such that fluorescence quenching by illumination is not required. A detailed understanding of the molecular switching process will lead to designer PAFPs whose photophysical properties can be tailored to fit specific applications.
Biophysical Properties of KFP
KFP is an obligate tetramer with protomer molecular mass of ~28 kDa. The protein is distantly related to GFP (23% identity in amino acid sequence) but is much more similar (~50% identity) to representative Anthozoa-derived FPs. Each protomer is synthesized as a single polypeptide. However, the mature form contains a chain break between residues Cys62 and Met63, leading to two polypeptides of ~8 and ~20 kDa that can be separated only under strongly denaturing conditions (31, 60). The chain break evidently results from the process of chromophore formation, but at present the exact nature of neither the chemical reaction nor the chromophore itself has been established.
Zagranichny et al. (60) studied chromopeptides isolated from KFP and, on the basis of NMR and spectroscopic studies, proposed that the cleavage reaction results in a N-unsubstituted ketimine instead of a carbonyl oxygen on the chromophore (FIGURE 1A⇑). However, this suggestion was challenged by Yampolsky et al. (59), who pointed out that N-unsubstituted ketimines are highly unstable in aqueous solution. The latter authors synthesized and studied a carbonyl-containing model chromophore, whose spectral properties closely match those of intact KFP. Interestingly, in aqueous solution, the model chromophore is ~10-fold more fluorescent than native KFP, suggesting that within KFP the protein fold suppresses fluorescence (59).
KFP strongly absorbs green light and displays one absorbance peak in the visible region centered at 570 nm, characteristic of the anionic form of the chromophore. Initially, fluorescence from KFP at 595 nm is extremely weak (ϕF < 0.001; Ref. 11); however, on illumination of sufficient intensity near the absorbance maximum, the protein is activated into a transiently fluorescent state that can be rapidly quenched by illumination with blue light (~450 nm). Recovery of the dark state is a relatively slow process with a half-life of 7 s (wild-type) at room temperature. When activated, a new absorbance peak appears at 450 nm, characteristic of the protonated chromophore, with a concomitant decrease in absorbance at 570 nm. This observation suggests that light activation is accompanied by a change in the chromophore environment so as to favor the protonated neutral state. Furthermore, the action spectrum for quenching of the kindled state corresponds closely to the 450-nm peak (12), suggesting that illumination of this species causes quenching. Curiously, fluorescence from the 450-nm chromophore species has not been reported.
Chudakov et al. (12) initially studied the basis for the kindling phenomenon using a combination of homology modeling and mutagenesis. They proposed that the activation process involved trans-cis isomerization of the chromophore as illustrated in FIGURE 1⇑. By amino acid sequence comparisons, three key internal amino acids adjacent to the chromophore (positions 143, 158, and 197) were identified and were proposed to control the chromophore configuration. In GFP, large side chains at the first two positions (His148 and Phe165 in GFP numbering) act to stabilize the cis configuration, precluding the trans configuration. In KFP, the corresponding positions are occupied by small side chains (Ala143 and Ser158) not found in any known FP, suggesting that either chromophore isomer might be accommodated.
Interestingly, substitution of serine 158 in KFP with the larger valine (12), as found in some FPs, converts KFP into a FP. Also, substitution of Ala143 with serine, typically found in Anthozoa FPs, results in increased fluorescence from both the nonkindled and kindled states and extends the half-life of the fluorescent state to ~200 s (3) while retaining the blue-light quenching property. Finally, on substitution of Ala143 with glycine, a variant referred to as KFP-1, the half-life of the kindled state is ~50 s at room temperature. KFP-1 is capable of irreversible activation under intense illumination (11).
Quillin et al. (42) used Arrhenius plot analysis to demonstrate that in KFP-1 the spontaneous deactivation is accurately modeled as a thermally driven process and estimated the energy barrier toward relaxation to the dark state to be ~71 kJ/mol. This value is somewhat higher than the ~55 kJ/mol estimated for cis-trans isomerization of model compounds in solution (21), but if cis-trans isomerization does take place within the protein interior, the higher value could reflect the additional energy required for rearrangements of the chromophore cavity.
Structural Studies of KFP
When evaluating the results of structural studies of any protein molecule, one must keep in mind that crystallographic studies give a time- and space-averaged view of the structure and thus may not reveal the key species in a dynamic process. A potential pitfall with KFP is that the quantum yield of the activated state is low (ϕF 3 0.12). One possible explanation for this is that the majority of the molecules in an activated sample remain in the nonfluorescent state. Electron-density maps will reveal the average structure, and in this case the active species might be obscured.
The extremely high optical density of the crystals and time dependence of the kindling phenomenon present significant challenges for structural studies of the activated state of KFP. Thus first to appear were reports of the putative dark state of KFP-1 (A143G). Two groups independently determined structures at 2.1 and 1.38 Å resolution [Protein Data Bank codes 1XQM (58) and 1XMZ (42)], based on data collected from crystals that were flash-frozen at 100 K. The molecular models are very similar and reveal the overall fold of KFP to be very similar to that of GFP, consisting of an 11-stranded β-barrel surrounding a central α-helix from which the chromophore is formed (FIGURE 2⇓). The electron-density maps verified the existence of the chain break immediately preceding the chromophore.
Due to prior reports describing the crystal structure analysis of the closely related nonfluorescent blue chromoprotein Rtms5 (41), there were few surprises in the KFP-1 structures. The salient features are 1) the chromophore is nonplanar and strongly bent or twisted in both KFP-1 and Rtms5 (FIGURE 3⇓, A AND B) and 2) the configuration of the KFP-1 chromophore is trans about the double bond (FIGURE 3⇓, A AND E). The trans configuration is stabilized by hydrogen bonds between the chromophore phenolate oxygen and Ser158 (FIGURE 2⇑).
The trans-noncoplanar configuration of the KFP chromophore is in strong contrast to the cis-planar configuration found (with one exception) in strongly FPs of known structure (e.g., DsRed; FIGURE 3C⇑). However, the red FP eqFP611 (39) provides an important exception because the chromophore is trans-planar (FIGURE 3D⇑). Therefore, trans-cis isomerization per se cannot completely account for the activation of KFP.
The models presented by Wilmann et al. (58) and Quillin et al. (42) differ somewhat with respect to side-chain conformational variability, which is partly a consequence of the difference in diffraction data resolution. In the structure reported by Quillin et al. (42), the highly conserved His197 was found to be statistically distributed between two alternate conformations. In the major conformation, the imidazole ring is “up” and perpendicular to the plane of the chromophore, whereas in the minor conformation the imidazole ring is “down” and stacks against the chromophore. The stacking interaction is seen in other Anthozoa FPs eqFP611, amFP486, EosFP, and zFP538 (23, 36, 39, 44). In the structure described by Wilmann et al. (58), His197 is modeled only in the stacking configuration, leaving a void corresponding to the major configuration observed by Quillin et al. (42).
Both groups agree, on the basis of modeling studies, that the chromophore cavity can easily accommodate either the cis or the trans form of the chromophore, but His197 must be in the stacked configuration for the cis chromophore to be accommodated. This led Quillin et al. (42) to propose that His197 might act as part of a “binary switch” to determine the fluorescent state. To highlight the potential importance of this histidine, we note that the PAFP Dronpa contains the corresponding histidine residue and also that the conversion of GFP into its photoactivatable form, PA-GFP, requires only the substitution Thr203 → His, again at the corresponding position (38).
Activated state structures
Recently, Andresen et al. (3) presented crystal structures of asFP595 variants in putative “activated” states, as well as the corresponding dark-state structures. For activation studies, these authors chose to work with the variant asFP595-A143S because of the long lifetime of the activated state. Crystals were irradiated using a high-intensity filtered light source, then flash frozen for data collection. Electron-density maps were obtained at various resolutions between 1.3 and 1.9 Å. The key conclusions of this study were that indeed, after irradiation, the chromophore could be visualized in the cis configuration (FIGURE 3F⇑). Furthermore, after short periods of illumination, the chromophore was observed to be in a statistically distributed mixture of cis and trans configurations. These results are consistent with the notion that light activation drives formation of the cis and presumably fluorescent isomer. Thus the hypothesis of Chudakov et al. (12) seems to have been confirmed. It is very interesting that the cis configuration appears to be stabilized by a new hydrogen bond between the chromophore hydroxyl group and the introduced serine 143. This could account for the longer lifetime of the activated state compared to KFP-1 (A143G) or wild-type, since both lack the serine and, hence, cannot form this hydrogen bond.
On the other hand, Andresen et al. (3) did not offer an explanation for why the cis configuration, as opposed to the trans configuration, should be fluorescent. Moreover, in the putative activated structures, the chromophore is non-coplanar (FIGURE 3F⇑). In fact, the twist and tilt angles (FIGURE 4⇓) are comparable to if not more distorted from planarity than those from the dark-state structures described by Wilman et al. (58) and Quillin et al. (42) (Table 1⇓).
In parallel, we carried out structural studies of KFP-1 in the light-activated state at the Argonne National Laboratory (Henderson JN and Remington SJ, unpublished observations). Crystal activation was achieved at several wavelengths with a high-power laser, and the crystals were flash frozen at 100 K. A 1.62-Å electron-density map calculated for KFP-1, activated at 540 nm, surprisingly reveals the chromophore to be in the trans configuration, with His197 in the stacking conformation. The activated state thus observed is very similar to dark-state described by Wilmann et al. (58), except for chromophore planarity. In the Wilmann et al. structure, the twist angle is 8.8° and the tilt 21°, whereas in the putative KFP-1 light-activated structure the twist and tilt angles are 1.2 and 13°, respectively (Table 1⇑). Multiple structural studies of KFP-1 obtained with and without laser illumination at various temperatures in the range of 100—298 K tend to support the general observation that activation involves movement of His197 into the stacked configuration with a concomitant increase in chromophore planarity.
We are convinced that trans to cis isomerization is an initial step in the process of activation; however, which state (cis or trans) actually corresponds to the fluorescent species remains unclear. It is worth noting that, in the A143S variant studied by Andresen et al. (3), the presence of serine at position 143 seems to permanently lock His197 into the stacked state so that its configuration remains unchanged throughout the photoactivation process. This leads to the possibility that the details of the photoactivation process are different for KFP-1 (A143G) and the A143S variant. The possibility also exists that the structure observed by Andresen et al. (3) represents a trapped intermediate in the pathway of KFP photoactivation.
Structures of permanently fluorescent KFP variants
Andresen et al. (3) also described the crystal structure analysis of flash-frozen crystals of a permanently fluorescent variant, asFP595-S158V, at 1.7-Å resolution. We have independently determined the crystal structure of a different fluorescent variation, asFP595-A143S/S158V, at 1.7-Å resolution (Henderson JN and Remington SJ, unpublished observations). In these fluorescent variants, the chromophore configuration is cis (FIGURES 2⇑ AND 3, G AND H⇑) and most importantly coplanar (especially the double mutant; Table 1⇑). We also observe the stabilizing hydrogen bond between serine 143 and the chromophore as described by Andresen et al. (3) for the activated state of asFP595-A143S.
However, close comparison of the atomic models reveals large differences in chromophore placement (up 3 Å) between these two fluorescent variants and all of the other structures [see Figure 4B⇑ of Andresen et al. (3) for an independent comparison]. This shift is likely due in part to the unfavorable interaction between the introduced valine 158 and the chromophore (FIGURE 2⇑). Although the structures of the permanently fluorescent variants provide valuable information, one may question whether these studies give definitive insight into the photoactivation process.
For a large class of compounds of biological relevance, cis-trans isomerization is a natural consequence of rearrangements in bonding electrons subsequent to the absorption of a photon. In the excited state, electron density is promoted into anti-bonding orbitals, favoring 90° rotations of one or more double bonds, which can then relax into a different isomeric configuration. Such rearrangements form, for example, the basis for human vision and the light-driven proton pumps found in unicellular organisms (26, 32, 55). Thus it is certainly attractive to hypothesize that light-induced trans-cis isomerization is necessary for the photoactivation of KFP or other photoactivatable proteins, but is it sufficient?
While pondering this question, it is important to keep in mind that visible fluorescence is one of many competing processes that can dissipate the energy of chromophores in the excited state. Light emission is usually less efficient than vibrational or other modes of energy dissipation, which is why most compounds that absorb visible light are merely colorful. Thus the key features of a transiently fluorescent species might not be observable by crystallographic techniques.
Yet comparative structural studies of both strongly FPs and nonfluorescent chromoproteins provide strong hints about the nature of the active configuration, and these in turn suggest that the factors that determine fluorescence efficiency probably cannot be reduced to a simple question of cis or trans. Indeed, planarity analyses of FP and chromoprotein structures (Table 1⇑) suggest that coplanarity is more important than isomer configuration in determining the efficiency of fluorescence.
It is clear that, in all these structures, the protein shell exerts profound influence on the chromophore configuration, yet this is very poorly understood. The cis form of the chromophore is more stable than the trans form in solution (21), because the trans form suffers from an unfavorable clash between the phenol moiety and the imidazolinone oxygen (FIGURE 1⇑). Nevertheless, the protein environments of Rtms5, KFP, and eqFP611 overcome this energetic hurdle and favor the trans form. In nonfluorescent Rtms5 and KFP, the trans form is accommodated by tilting or twisting the planes of the chromophore to minimize this steric clash, whereas in fluorescent eqFP611, bond angles within the trans-planar chromophore “open up” to relieve the tension.
On the other hand, in all strongly fluorescent proteins, the chromophore is maintained in a strictly planar configuration. The reasons for this have not been thoroughly investigated and remain unclear. Likewise, it is not entirely clear why chromophore planarity is disrupted in the chromoproteins (42). Although His197 is clearly important in KFP and for efficient fluorescence in other Anthozoa FPs (23), it is not highly conserved in the chromoproteins (12). In Rtms5, the equivalent Arg197 seems to enforce the nonplanar conformation (41). Yet strongly fluorescent proteins such as GFP and DsRed lack the histidine in this position. Therefore, depending on context, chromophore planarity is likely dictated by multiple elements in the protein environment.
Thus a number of important questions on structure-function relationships in FPs and chromoproteins remain to be investigated. Once these questions are resolved, the factors that influence the lifetime of the photoactivated state will need to be determined. We are confident that further structural analyses of these and other photoactivatable proteins will reveal the guiding principles. In turn, such understanding will lead to important new tools and advances in the methodology for cellular and molecular biologists.
This work was supported by National Science Foundation Grant MCB-0417290 to S. J. Remington and National Institutes of Health Training Grant GM-07759 to the Institute of Molecular Biology.
- © 2006 Int. Union Physiol. Sci./Am. Physiol. Soc.