A new bright green-emitting fluorescent protein – engineered monomeric and dimeric forms

Authors


  • Note
    The nucleotide sequence data are available in the DDBJ/EMBL/GenBank databases under the accession number FN597286 and the protein sequence data are in UniProtKB/TrEMBL with the accession number D1J6P8.

L. Regan Department of Molecular Biophysics and Biochemistry, Yale University, New Haven, CT 06520, USA
Fax: (203) 432 5175
Tel: (203) 432 9843
E-mail: lynne.regan@yale.edu

Abstract

Fluorescent proteins have become essential tools in molecular and biological applications. Here, we present a novel fluorescent protein isolated from warm water coral, Cyphastrea microphthalma. The protein, which we named vivid Verde fluorescent protein (VFP), matures readily at 37 °C and emits bright green light. Further characterizations revealed that VFP has a tendency to form dimers. By creating a homology model of VFP, based on the structure of the red fluorescent protein, DsRed, we were able to make mutations that alter the protein’s oligomerization state. We present two proteins, mVFP and mVFP1, that are both exclusively monomeric, and one protein, dVFP, which is dimeric. We characterized the spectroscopic properties of VFP and its variants in comparison with enhanced green fluorescent protein (EGFP), a widely used variant of GFP. All the VFP variants are at least twice as bright as EGFP. Finally, we demonstrated the effectiveness of the VFP variants in both in vitro and in vivo detection applications.

Structured digital abstract

Abbreviations
dVFP

dimeric VFP

EC

extinction coefficient

EGFP

enhanced GFP

FCS

fluorescence correlation spectroscopy

FMRP

fragile X mental retardation protein

FP

fluorescent protein

GFP

green fluorescent protein

GST

glutathione S-transferase

hpf

hours post-fertilization

mVFP

monomeric VFP

QY

quantum yield

tD

diffusion time

t½

half time

T-Mod

TPR-based recognition module

TPR

tetratricopeptide repeats

VFP

vivid Verde fluorescent protein.

Introduction

Fluorescent proteins (FPs) have become ubiquitous tools in biological and biomedical research. Since the cloning and exogenous expression of green fluorescent protein (GFP) from the jellyfish Aequorea victoria, researchers have sought new variants of this protein, as well as of other FPs, with properties that are well-suited for a particular application [1–3]. Extensive mutagenesis has been performed on FPs to better tailor their properties to the needs of biologists [1,2,4,5]. Of special interest are FPs with new excitation and emission wavelengths, FPs with increased brightness, FPs that are monomeric and FPs that mature rapidly at 37 °C.

GFP is a 238-amino-acid protein, whose chromophore is formed by the post-translational re-arrangement of an internal Ser-Tyr-Gly sequence to a 4-(p-hydroxybenzylidene)-imidazolidine-5-one structure [6]. The crystal structure of GFP revealed that the chromophore is buried in the center of a β-barrel structure [7,8]. Amino acid mutagenesis and protein engineering were carried out on GFP to improve its spectral characteristics, oligomeric state and chromophore-maturation at 37 °C [6,9–12]. A broad range of GFP variants with fluorescence emission ranging from blue to yellow regions of the visible spectrum was created [1,2]. Enhanced green fluorescent protein (EGFP) is a widely used variant of GFP, which has mutations at two positions: F64L and S65T [9,10]. EGFP is brighter and matures more rapidly at 37 °C than wild-type GFP [1,9]. Protein engineering of EGFP has yielded several green variants with improved characteristics, such as Emerald FP. This Emerald FP has improved photostability and brightness compared with EGFP [11]. Another GFP variant is the ‘superfolder’ GFP that is designed to fold faster at 37 °C. This ‘superfolder’ GFP is also brighter and more acid resistant than either EGFP or Emerald FP [12]. A weak tendency of GFP and its variants to dimerize was completely eliminated using point mutations at F223K, L221K, or A206K [13,14].

Another FP, DsRed, from the sea anemone Discosoma striata, is also of great interest to researchers because its intrinsic fluorescence is red rather than green [15,16]. The chromophore of DsRed is closely related to that of GFP, being formed by the re-arrangement of an internal Gln-Tyr-Gly tripeptide [15]. The extended conjugation in the chromophore causes the red-shift observed in DsRed and other red FPs [4]. DsRed forms a strong tetramer both in solution and in crystal and its chromophore maturation is very slow [17–19]. As a result of these limitations, DsRed has been a target of protein engineering and mutations to improve its chromophore maturation rate and to reduce oligomerization [20–22]. A directed evolution approach was performed on DsRed to make a monomeric version, mRFP1, which has a total of 33 amino acid mutations [21]. In addition to DsRed, there are many other FPs, ranging from blue-, cyan-, green- and yellow- to red-emitting, which have different spectral properties, brightness, and stabilities, that have been isolated from reef corals and other Anthozoa species [1,2]. Most of these FPs display a higher degree of oligomerization, which is detrimental for cellular labeling [17,18,23]. To overcome FP oligomerization, mutations must be made at the monomer–monomer interface. The exact nature of such interfaces varies depending on the nature and origin of the FP [2].

Many FPs, either isolated from natural sources or engineered from GFP or DsRed, are known and available [1,2]. However, only a few of the current FPs are widely used in various cell-imaging applications and most of them have certain limitations [1,2,24]. A continuing effort must be made to improve the spectral characteristics and stabilities of the FPs, or alternatively, to search for new FPs with optimal properties, for maximum utility in cellular imaging.

The natural habitat of A. victoria is the cool waters off the northwest coast of Washington State. One might expect organisms that inhabit warmer waters to have evolved FPs that mature more rapidly at higher temperatures. Here we describe the characterization and modification of a novel FP that was isolated from Cyphastrea microphthalma, a scleractinian coral found in the warmer waters of the Australian Great Barrier Reef (Fig. 1). Several new fluorescent organisms were identified by diving at night with UV illumination, and the FPs were cloned from these organisms and expressed in Escherichia coli [25,26]. We found that the vast majority of proteins characterized indeed mature robustly and rapidly at 37 °C. Here we report the properties of one of the novel green-emitting FPs, vivid Verde FP (VFP), which exhibits useful properties. VFP is very bright, matures rapidly at 37 °C and we have engineered exclusively monomeric or dimeric variants of it. These properties are particularly well suited to a variety of molecular and biological applications.

Figure 1.

 (A) Amino acid sequence alignment of VFP with DsRed and EGFP. The chromophore-forming amino acid residues are shown in bold and are underlined. The amino acid residues (N158 and T160) of VFP, where the mutations were made, are indicated by a bold letter. The conserved Arg and Glu (corresponding to Arg96 and Glu222 of GFP) residues are shown on a gray background. (B) A scleractinian coral, Cyphastrea microphthalma, collected in 1.2 m of water off Lizard Island on the Australian Great Barrier Reef. (C) Overlay of the absorption (abs), fluorescence-excitation (ex) and fluorescence-emission (em) spectra of VFP. The samples were excited at 450 nm and the emission spectra were measured from 465 to 650 nm. The fluorescence excitation spectra were obtained from 250 to 515 nm by monitoring the emission at 530 nm. The spectra were normalized at the maximum peak.

Results

Sequence of the new FP and relation to other known FPs

A new FP, VFP, was isolated and cloned from the C. microphthalma coral, collected in 1.2 m of water off Lizard Island on the Australian Great Barrier Reef [25,26]. The alignment of the amino acid sequences of VFP, DsRed and EGFP is shown in Fig. 1A. The amino acid residues that form the chromophore are in bold and underlined. The chromophore residues at positions 66, 67 and 68, following the amino acid residues numbering in DsRed, are QYG in VFP, QYG in DsRed and TYG in EGFP. VFP shows greater sequence identity overall to DsRed than to EGFP, with 53% sequence identity to DsRed and only 20% sequence identity to EGFP. Sequence alignment demonstrates the conservation of many positions in VFP, which are presumably structurally and/or functionally important. Arg96 and Glu222 of GFP, which were proposed to participate in chromophore maturation [27], are also conserved in DsRed and VFP. The VFP coding sequence was deposited in the EMBL nucleotide sequence database under the accession number FN597286. Using the Swiss Institute of Bioinformatics BLAST Network Service, the VFP sequence was found to have the highest sequence identity, of 83%, to a GFP isolated from coral Montastraea cavernosa [28]. Sequence alignment also showed that there are several cyan, green, or red FPs and chromoproteins from coral in which the chromophore is formed by amino acids QYG, the same as in VFP.

VFP exhibits maximum excitation and emission peaks at 491 and 503 nm, respectively, as shown in Fig. 1C. These spectral properties are more similar to those of EGFP rather than to those of DsRed, despite the fact that the sequence of VFP is more closely related to DsRed than to EGFP. The chromophore formation in GFP involves cyclization, oxidation and dehydration, and in DsRed and other coral FPs, an additional oxidation step occurs [4,29–31]. Previously, DsRed chromophore maturation has been shown to proceed through a green-emitting anionic GFP-like intermediate, which has excitation and emission peaks at 475 and 499 nm, respectively [17]. However, it has also been proposed that the red-emitting chromophore of DsRed and of related chromoproteins is produced from a blue-emitting neutral form of a GFP-like chromophore, the green anionic species being the dead-end product [32]. The GFP-like chromophore of VFP is stable and further conversion into the red-emitting chromophore was not observed.

Two tryptophan residues at positions 93 and 143 of DsRed, located in the immediate vicinity of the chromophore, are conserved in VFP (corresponding to positions 89 and 139). Thus, the absorption spectrum of VFP showed a peak at 280 nm (Fig. 1C) as a result of the presence of these Trp residues, and excitation at 280 nm gave an emission peak at 503 nm.

Oligomeric state of VFP

For many applications, it is essential that the FP used to ‘tag’ another protein is monomeric [24]. If an FP is not monomeric, then its oligomerization may influence the behavior of the tagged protein, thus perturbing the system under study. We used gel-filtration chromatography to assess the oligomeric state of VFP. To allow direct comparison with a known protein, we also purified EGFP, which is monomeric at concentrations of < 1 mg·mL−1 [33]. A gel-filtration chromatogram of VFP showed a major peak and a shoulder, indicating a mixture of dimer and monomer species (Fig. S1). We therefore sought to design mutations to shift the equilibrium to a fully monomeric state. With this goal in mind, we aligned the sequences of DsRed and VFP and created a homology model for VFP.

It is known that DsRed forms a strong tetramer, both in solution and in the crystal structure [17,18]. An examination of the crystal structure of the DsRed tetramer shows that the monomers are arranged as a dimer of dimers, with AB (or CD) and AC (or BD) interfaces, as illustrated in Fig. 2A. The AB interface is dominated by hydrophobic interactions, whereas the AC interface is comprised predominantly of salt bridges and hydrogen bonds [19]. Thus, the formation of VFP dimer could be caused by the interaction of either AB or AC. Several point mutations (such as I125R, H162K, A164R and I180T) on the surface of DsRed are documented in the literature, which convert the DsRed tight tetramer into a monomer [1,21]. We compared the residues at these positions in VFP with those in DsRed to identify mutations in VFP that might shift the monomer–dimer equilibrium towards monomer. The corresponding amino acid residues in VFP are H121, N158, T160 and T176, allowing us to identify possible mutations in VFP as H121R, N158K and T160R. We focused on examining the N158K and T160R mutations. The locations of these mutations in the AC interface are indicated in Fig. 2B. The rationale for the N158K mutation is that it replaces a polar uncharged Asn with a positively charged Lys and this mutation should disrupt the AC dimerization interface. In DsRed, His162 of the A monomer is involved in a stacking interaction with His162 of the adjacent C monomer, whilst simultaneously making an electrostatic interaction with Glu176 of the C monomer, forming what appears to be an important part of the AC interface [19]. In VFP, residue 158 (corresponding to residue 162 in DsRed) is Asn and residue 172 (corresponding to residue 176 in DsRed) is Asp. By contrast, in T160R mutations, the polar uncharged Thr was replaced with the positively charged Arg. In DsRed, position 164 is occupied by Ala, which creates small hydrophobic patches in the AC interface and, by replacing it with Arg, the AC interaction is disrupted. Also, previous studies showed that substituting hydrophilic or charged amino acids for hydrophobic and neutral residues of the FP tetrameric interfaces could generate the monomer form of the protein [13,21].

Figure 2.

 (A) A cartoon illustration of DsRed tetramer arranged as a dimer of dimers with AB (=CD) and AC (=BD) interfaces. (B) Structure of two of the four subunits of the tetrameric DsRed consisting of the AC polar interface. The positions of the amino acid residues 158 and 160 (corresponding to amino acid residues 162 and 164, respectively, in DsRed) where mutations were made, are indicated by lines. The chromophore at the center of the β-barrel structure is shown in black sticks. Protein Data Bank (PDB) code: 1GGX. [55]

The mutations were made individually, with the intention of combining any of them if an individual mutation was insufficient to cause the VFP to monomerize. We expressed and purified each VFP mutant (N158K or T160R) and assessed its oligomeric states using gel-filtration chromatography. We found that either the N158K mutation or the T160R mutation is sufficient to convert VFP into an exclusively monomeric species (Fig. S1). We named these monomeric N158K and T160R mutants as mVFP1 and mVFP, respectively. In the course of the cloning, we also serendipitously isolated the T160A mutant of VFP. Gel-filtration chromatography revealed that the T160A mutant is fully dimeric, with no evidence of the monomer–dimer equilibrium that we observed for VFP (Fig. S1). Presumably, the introduction of small hydrophobic patches on the surface of the protein promotes strong dimer formation. We named this dimeric variant of VFP as dVFP.

Spectral properties of VFP and its variants

We proceeded with further characterizations of all four proteins, namely VFP and its variants mVFP1 (N158K), mVFP (T160R) and dVFP (T160A). The excitation and emission spectra for all four proteins were identical, with an excitation maximum of 491 nm, an emission maximum of 503 nm and a Stokes shift of 12 nm (Table 1, Fig. 1C and Fig. S2). The measured extinction coefficient (EC) of VFP was 83 700 M−1·cm−1, which was higher than that of EGFP (54 400 m−1·cm−1). The ECs calculated for mVFP1 and mVFP were 80 400 and 85 000 m−1·cm−1, respectively. However, a higher EC value of 107, 000 m−1·cm−1 was observed for dVFP. The increase in the EC value of dVFP compared with VFP might be caused by its tight dimer formation. Table 1 summarizes these data, alongside the measured results for EGFP and Venus for comparison. Venus is a variant of yellow FP with a fast maturation and high brightness [34]. The results obtained for EGFP and Venus are consistent with the values reported in the literature [34,35]. For comparison, Table 1 includes a list of selected green-emitting FPs that have spectral properties relevant to VFP variants. We reported the fluorescence excitation and emission wavelength peaks, molar EC, quantum yield (QY), oligomeric states, relative brightness and photostability of these selected FPs.

Table 1.   Spectral properties of VFP and its variants in comparison to EGFP, Venus and selected FPs. The excitation (Ex) and emission (Em) wavelengths, the molar extinction coefficients (EC), the quantum yield (QY), the oligomeric states, the relative brightness and the photostability are listed. The relative brightness was calculated from the product of EC and QY. Photostability was calculated based on 100% EGFP measured at the same time. ND, not determined.
ProteinEx (nm)Em (nm)EC × 10−3 (m−1·cm−1)QYOligomeric statesRelative brightnessPhotostability (% EGFP)Reference
VFP49150383.71.0Weak dimer8433This work
mVFP49150385.00.86Monomer7316This work
mVFP149150380.40.84Monomer6811This work
dVFP491503107.01.0Dimer10739This work
EGFP48850954.40.60Monomer33100This work
Venus51552793.30.75Monomer7027This work
GFPs – Anthozoa
 AzamiGreen49250555.00.74Monomer41ND44
 mWasabi49350970.00.80Monomer565345
 ZsGreen49350543.00.91Tetramer39ND15
 copGFP48250270.00.60Tetramer42ND46
 cmFP51250351258.80.66Tetramer399247
 aacuGFP250251393.90.71ND67ND48
 aeurGFP504515145.70.67ND98ND48
 eechGFP1497510124.20.75ND93ND48
 efasGFP496507125.80.80ND101ND48
 gfasGFP492506102.50.73ND75ND48
 plamGFP50251498.60.96ND95ND48
 sarcGFP48350076.70.96ND74ND48
GFP – Aequorea derivatives
 Superfolder48551083.30.65Monomer549012
 mEmerald48750957.50.68Monomer395811
Cyan FPs – Anthozoa
 mTFP1462492640.85Monomer546349
 Photoconvertable FPs
 Kaede50851898.80.88Tetramer872650
 dEos (G)506516840.66Dimer552451
 mEos2 (G)506519560.84Monomer472152
Photoswitchable FPs
 Dronpa503517950.85Monomer81ND53
 mKikGR505515490.69Monomer34754
 cerFP505494505540.55Tetramer305447

The EC and QY for each FP were determined and the product of these two parameters (EC x QY) provides the relative brightness (Table 1). We used the reported EGFP QY of 0.60 [35] as a reference for calculating the QY of VFP and its variants. The relative QY values of VFP and its variants ranged from 0.84 to 1.0, which was higher than those of both EGFP and Venus. Thus, as a result of having a high EC and a high QY, VFP and its variants produced high relative brightness. It is evident that the dimeric form, VFP or dVFP variant, is brighter than the monomeric form of VFP. Either the mVFP1 or the mVFP variant is at least twice as bright as EGFP, and the dVFP variant is much brighter than Venus. To our knowledge, there is no monomeric green-emitting FP available to date that is at least twofold brighter than EGFP, except for the photoswitchable Dronpa FP (Table 1).

We also investigated the pH dependence of VFP and its variants' fluorescence emission at 503 nm upon excitation at 491 nm, as shown in Fig. S3. VFP and its variants were found to have pH stability profiles similar to those of EGFP between pH 6 and pH 10.

Fluorescence correlation spectroscopy measurements of size and photostability

We used fluorescence correlation spectroscopy (FCS) to investigate the photobleaching, molecular brightness and oligomeric states of the FPs in more detail. The traces of FCS autocorrelation curves obtained for EGFP and mVFP are shown in Fig. 3A. No shifts in the autocorrelation curves were observed for EGFP as a function of laser power intensity. However, the diffusion curves shifted to the left for VFP and its variants as the laser power intensity was increased (Fig. 3A and Fig. S4). This shift in autocorrelation curves to the left, noted by shorter apparent diffusion times (tD), is indicative of photobleaching.

Figure 3.

 (A) Representative FCS autocorrelation curves of EGFP and mVFP taken at increasing laser power intensities from 0.25 to 5 μW. A shift in the autocorrelation curve to the left, to apparently shorter tD values, as a function of laser power intensity, was observed for VFP and its variants. The autocorrelation curves are normalized to the number of molecules obtained from the fitting autocorrelation function [G(t)]. (B) Photobleaching curves for the EGFP, Venus, mVFP and dVFP under mercury arc lamp illumination using a wide-field microscope. The relative photostability of VFP and its variants are reported in Table 1.

The autocorrelation curves for each sample were fitted using single-diffusion or two-diffusion component equation. The best-fit curve was assessed based on the residual of the fitting. A detailed analysis of the other photophysical dynamics (e.g. triplet blinking), occurring at the submillisecond timescale, is beyond the scope of this paper and will be presented elsewhere. The tD value and the average fluorescence intensity were determined from the fitting of the autocorrelation curves taken at 0.25 μW laser power, as reported in Table 2. At this low laser power intensity, the effects of other photophysical processes were minimized. The relative molecular brightness of the FPs was calculated by dividing the average fluorescence intensity by the number of molecules within the illuminated region. The results obtained support our earlier findings that VFP and its variants are nearly twofold brighter than EGFP, based on the counts per molecule (kHz/molecule) in Table 2 and Table S1.

Table 2.   Summary of FCS analysis. The autocorrelation curves of each FP obtained at 0.25 μW laser power intensity were fitted using a single-diffusion component equation. The brightness, expressed as counts per molecule, was calculated by dividing the intensity by the number of molecules.
FPsDiffusion time (ms)Intensity (Hz) 1 × 104Counts per molecule (kHz·molecule−1)
EGFP0.486 ± 0.0122.49 ± 0.050.262 ± 0.005
Venus0.543 ± 0.0193.25 ± 0.020.251 ± 0.002
VFP0.646 ± 0.0044.07 ± 0.141.76 ± 0.06
mVFP10.460 ± 0.0073.40 ± 0.190.485 ± 0.028
mVFP0.472 ± 0.0073.70 ± 0.140.476 ± 0.018
dVFP0.763 ± 0.0043.00 ± 0.111.55 ± 0.06

At 0.25 μW laser power intensity, the measured relative tD values of either mVFP1 or mVFP were comparable to that of the EGFP, indicating that both variants are monomeric. Furthermore, VFP and dVFP have tD values greater than that of EGFP, indicating higher oligomeric states (Table 2). These results supported our findings on the oligomeric states of the VFP variants using gel-filtration chromatography, as described earlier.

Based on our FCS results, we noticed that photobleaching occurs in VFP and its variants. This observation prompted us to investigate, in greater detail, the rate of photobleaching of VFP and its variants in comparison to that of EGFP and Venus using wide-field microscopy, as described in the Materials and methods. Figure 3B depicts the relative photobleaching curves of EGFP, Venus, mVFP and dVFP from 0 to 500 s. We determined the relative half time (t½) to photobleach the VFP samples, EGFP and Venus (Fig. S5). Based on the t½ values, we calculated the percentage of photostability of the VFP and its variants relative to 100% EGFP. We also included, in Table 1, the reported photostability of some FPs relative to 100% EGFP measured at the same time. The photostability data of other green-emitting FPs have not yet been reported or determined. The mVFP1 and mVFP variants, which have 11% and 16% photostability, respectively, were less photostable than VFP and dVFP. However, the dVFP variant has 39% photostability, and thus exhibits greater photostability than Venus and other photoconvertable or photoswitchable FPs. For other imaging applications [24], the difference in photostability has no relevance. Even with this photostability, our VFP variants can be useful for numerous in vitro and in vivo detection applications.

Application of VFP variants as detection markers

It has been shown previously in our laboratory that a protein recognition domain, tetratricopeptide repeats (TPR), fused to EGFP can be used to detect the protein–peptide interaction in a single step, completely eliminating the use of primary and secondary antibodies in western blot analysis [36]. The TPR-based recognition module (T-Mod) was demonstrated to bind specifically to MEEVF peptide fused to glutathione S-transferase (GST) [36]. The fusion of FP to T-Mod can completely eliminate the need for any antibodies or developing procedures, which makes western blotting faster, simpler and less costly. We adapted this experiment to show the usefulness of mVFP and dVFP brightness in comparison to EGFP. We expressed and purified the T-Mod fused to EGFP, mVFP or dVFP. Following the SDS/PAGE of E. coli-expressing GST–MEEVF lysate, gels were transferred to poly(vinylidene difluoride) membrane and processed as for western blotting. After blocking the membrane, we incubated the blots separately with different T-Mod–FPs for 1 h at room temperature. The membrane was then visualized using a UV transilluminator at 302 nm, as shown in Fig. 4A. The visible band indicated by an arrow is the GST–MEEVF protein detected by the binding of T-Mod–FP. The bands from T-Mod–mVFP or T-Mod–dVFP were at least two-fold brighter than that of the EGFP. Additional bands were visible in the membrane incubated with T-Mod–dVFP as a result of the intense brightness of the dVFP protein. This result illustrates the benefit of having high brightness, in terms of sensitivity, in a practical detection application.

Figure 4.

 (A) Comparison of the T-Mod fused to EGFP, mVFP, or dVFP as a replacement for antibodies in western blot analysis. A duplicate SDS-polyacrylamide gel used in western blotting was stained with Coomassie Brilliant Blue. Lanes 1, precision plus protein standard (BioRad); lane 2, lysate; lane 3, lysate supplemented with 1 mg·mL−1 of purified GST–MEEVF; lane 4, purified GST–MEEVF. The arrow indicates the GST–MEEVF protein band. (B) Microinjection of KH–mVFP fusion mRNA into zebrafish embryos. The expression of KH–mVFP protein was monitored in the embryos at 6 and 14 hpf using fluorescence microscopy. Zebrafish embryos without RNA injections were used as a control.

Application of mVFP as an in vivo marker

To demonstrate that our VFP can be used for in vivo labeling, we chose the monomeric form, mVFP, and fused it to the KH domains of fragile X mental retardation protein (FMRP). We injected mRNA encoding the KH–mVFP fusion protein into zebrafish embryos at the one-cell stage. Live embryos at 6-h post-fertilization (hpf) and at 14 hpf (10-somite) stages were mounted on glass slides and visualized using a fluorescence microscope, as shown in Fig. 4B. The fluorescence signals from zebrafish embryos with KH–mVFP were more intense than those of the control, which showed a faint cellular autofluorescence.

Discussion

We have described a detailed characterization of a new FP from the warm water coral, C. microphthalma, collected off Lizard Island on the Australian Great Barrier Reef. The protein, which we named VFP, matures rapidly at 37 °C and emits bright green fluorescence. VFP, as isolated, showed a propensity to form fairly weakly associating dimers. By creating a homology model of VFP, we were able to create surface mutations that convert VFP into either an exclusively monomeric species (N158K or T160R) – which we named mVFP1 and mVFP, respectively, or into an exclusively dimeric species (T160A) – which we named dVFP. This rational approach to creating monomeric variants can be used as a guide for re-engineering other coral FPs that have higher oligomeric forms.

These novel proteins have features that will be useful for a variety of applications. The mVFP1 and mVFP variants are both monomeric and fluoresce at least twice as brightly as EGFP. The dimeric dVFP is even brighter, being at least 1.5 times as bright as Venus. For applications where oligomerization is not critical, the use of the dVFP variant would be advantageous because of its high brightness. When a bright, monomeric protein is desired, mVFP1 or mVFP would be the proteins of choice. Based on the list of reported FPs (either wild-type or engineered) (Table 1), none is both monomeric and at least two-fold brighter than EGFP, except for photoswitchable Dronpa. The data we presented should allow investigators to choose which VFP variant is the most appropriate for their specific research application.

With regards to photostability, VFP and its variants photobleached at a faster rate than EGFP. The vast majority of reports in the literature describing green-emitting FPs isolated from corals do not include photostability measurements, which makes it difficult to assess the level of photostability of VFP variants in relation to other coral FPs [2,24]. However, for many imaging applications, this photobleaching property will not be influential [24,34]. In conclusion, the monomeric and the dimeric forms of VFP represent viable alternatives to the widely used EGFP and Venus.

Materials and methods

Plasmid constructions and mutations

The plasmids encoding VFP, EGFP and Venus with polyhistidine tags were constructed as previously described [26,37]. The VFP coding sequence was deposited in the EMBL nucleotide sequence database under the accession number FN597286 and in the UniProtKBT/TrEMBL protein sequence database under the accession number D1J6P8. Site-directed mutagenesis (QuikChange Site-Directed Mutagenesis Kit; Stratagene, Cedar Creek, TX, USA) was used to introduce the N158K and T160R mutations into VFP. Mutations were verified by DNA sequencing (W. M. Keck, Foundation Facility, Yale University, CT, USA).

Sequence alignment and homology modeling

Sequence alignment of VFP with EGFP and DsRed was performed using clustalw2 (EMBL-EBI). Homology modeling was carried out using swiss-model [38].

Recombinant protein expression

The proteins were expressed in E. coli DH10β cells grown in Luria–Bertani (LB) liquid medium for 24 h at 37 °C. The cells were harvested by centrifugation and the pellets were resuspended in lysis buffer (50 mm Tris/HCl, pH 7.4, 300 mm NaCl) supplemented with a tablet of complete EDTA-free protease inhibitor cocktail (Roche) and 5 mmβ-mercaptoethanol. The lysate was sonicated, then centrifuged. The supernatant solution was loaded into Ni-nitrilotriacetic acid agarose (Qiagen, Valencia, CA, USA), and the pure protein was eluted with 50 mm Tris/HCl, pH 7.4, 150 mm NaCl, 200 mm imidazole. The fractions containing the protein were pooled and dialyzed into 50 mm Tris/HCl, pH 7.4, 150 mm NaCl. The purity of the samples was determined by SDS/PAGE. The proteins were concentrated by centriprep YM-10 with 10 000 MWCO (Amicon, Billerica, MA, USA) to about 100–200 mm then stored in aliquots at −20 °C. The buffer used in all spectroscopic analyses was 50 mm Tris/HCl, pH 7.4, 150 mm NaCl, unless otherwise noted.

Analytical gel-filtration chromatography

The molecular sizes of the purified FPs were analyzed using a Superdex S200 10/30 gel-filtration column (Amersham Pharmacia) by FPLC at room temperature. A 100 mL sample of < 0.01 mg·mL−1 of each FP was injected into the column at a flow rate of 0.5 mL·min−1 and the absorbance was monitored at 280 nm. The oligomeric states of the VFP and its variants were determined based on the EGFP elution time and protein standards (Bio-Rad, Hercules, CA, USA).

Absorption spectroscopy

The absorbance spectra of the FPs were recorded on a Hewlett Packard 845X UV-visible Chemstation. The ECs of the FPs were calculated based on the absorbance of the native and acid-denatured or alkali-denatured proteins. The ECs of the GFP-like chromophores used in the calculation are 44 000 m−1·cm−1 at 447 nm in 1 m NaOH [33] and 28 500 m−1·cm−1 at 382 nm in 1 m HCl [39]. For yellow FP, Venus, the EC of the chromophore was back-calculated using 22 000 m−1·cm−1 at 280 nm in 10 mm Tris/HCl.

Fluorescence spectroscopy

Fluorescence excitation and emission measurements were performed using a PTI Quantamaster C-61 two-channel fluorescence spectrophotometer. The samples were excited at 450 nm and emission spectra were measured from 465 to 650 nm with a 2 nm slit-width. Fluorescence excitation spectra were obtained from 250 to 515 nm by monitoring the emission at 530 nm with a 2 nm slit-width. The QY values of the VFP and its variants were determined relative to EGFP (QY = 0.60 [35]). The pH dependence of VFP and its variants' fluorescence emission at 503 nm were measured upon excitation at 491 nm at room temperature. pH titrations were performed using a series of 100–200 mm citrate-phosphate buffer (pH 2.0–11.0) containing 150 mm NaCl.

FCS

FCS measurements were made on a laboratory built instrument, based around an inverted microscope with a 488 nm DPSS laser for excitation, as previously described [40,41]. All measurements were carried out on FP samples of approximately 100 nm using varying laser power intensities from 5 to 0.25 μW measured on the table before entering the microscope. The output of the detection channels was autocorrelated in a digital correlator (Correlator.com). Control measurements were performed using Alexa 488 solutions to ensure the proper alignment of the confocal optics and the absence of artifacts in the FCS. The autocorrelation curves were fitted using a single- or two-component equation, as previously described [41]. The parameters extracted from the fittings were relative tD number of molecules, and fluorescence intensities.

Photobleaching

Photobleaching measurements of purified FP samples were performed using a inverted wide-field microscope equipped with a 100 W mercury arc lamp similar to those described in the literature [42]. The FP samples were mixed with mineral oil, and about 5 μL of the mixture was sandwiched between a glass slide and a cover slip. A neutral density filter was used initially for sample alignment, which was removed when the actual measurements were being made. The FP samples were imaged using a 50 ms exposure time and a frame rate of one image per second. The measurement was taken in a 600 s time span under constant illumination.

Western blot assays

The T-Mod–FP and GST–MEEVF constructs were prepared as previously described [36]. The FP fused to T-Mod was EGFP, mVFP, or dVFP. Each construct was transformed into E. coli BL21(DE3) cells and the protein was purified following the protocol previously described [36]. The GST–MEEVF lysate was obtained from 6-mL overnight culture cell pellet by adding 1 mL of B-Per (Pierce, Rockford, IL, USA) and shaking, with occasional vortexing, for 10 min. The lysate was supplemented either with or without 1 mg·mL−1 of purified GST–MEEVF protein. The samples mixed with a reducing loading buffer were loaded precisely into 4–12% gradient SDS-polyacrylamide gels together with an equivalent amount of purified GST–MEEVF protein. The gels were run at room temperature for 1 h at a voltage of 120 V using NuPAGE buffer (Invitrogen, Carlsbad, CA, USA). One gel was stained with Coomassie Brilliant Blue while the other gels were transferred onto a poly(vinylidene difluoride) membrane (Millipore, Billerica, MA, USA). Membrane transfer was carried out in a cold room for 3 h at a constant current of 380 mAmp. The transfer buffer used contained 24 mm Tris-base, 192 mm glycine, 10% methanol and 0.01% SDS. The membranes were blocked in 5% non-fat milk in TBS-T (20 mm Tris-base, pH 8.0, 150 mm NaCl, 0.1% Tween-20) overnight at 4 °C with shaking. The membranes were then incubated individually with each 5 μm T-Mod-FP fusion construct in TBS-T containing 0.1% nonfat milk for 1 h at room temperature with shaking. The membranes were washed three times with TBS-T, for 10 min each wash, visualized using a UV transilluminator at 302 nm and the images captured using a digital camera (Kodak, Rochester, NY, USA).

mRNA microinjection assay

To assemble the KH–mVFP fusion construct, the deleted KH domain of human FMRP – hFMRP(KH1-KH2Δ) – was fused with the N-terminus of mVFP and cloned into the mammalian PCS2 + vector. The construct was sequenced (W. M. Keck Foundation Facility, Yale University) and named KH–mVFP for simplicity. The in vitro synthesis of large amounts of capped RNA was carried out using the mMESSAGE mMACHINE kit (Applied Biosystems/Ambion, Austin, TX, USA) following the manufacturer’s protocols. The capped transcription reaction was prepared at room temperature and then incubated at 37 °C for 2 h. TURBODNase (Ambion) was added to the reaction and incubated at 37 °C for another 15 min to remove the template DNA. The RNA was purified using the RNeasy Mini kit (Qiagen). The concentration of the RNA was determined using a UV-vis spectrometer and then the RNA was stored at −80 °C until use.

The RNA microinjections were performed at the one-cell stage using standard protocols [43]. The injection solution consisted of 200 ng·μL−1 of KH–mVFP and 0.15% Phenol Red in Danieau’s solution. Live embryos at 6 and 14 hpf stages were manually dechorionated and mounted in methylcellulose. In parallel, we also mounted embryos without RNA injections as a control. Fluorescent images were acquired on a Zeiss Axioskop microscope using a 20 × objective and an FITC filter. Color adjustment of the fluorescent images was made equally for both KH–mVFP-injected and control zebrafish using ImageJ software.

Acknowledgements

We thank Dr. Joseph Wolenski of the MCDB Imaging Facilities at Yale University for helping us with photobleaching experiments; and Dr. Scott Holley and Jamie Schwendinger-Schreck of the MCDB at Yale University for performing the RNA microinjection assay in zebrafish. We also thank the members of the Regan laboratory for comments and suggestions on the manuscript. This work is funded by HFSP (RGP44/2207 to L.R.), Leslie H. Warner Postdoctoral Fellowship (to R.P.I.), NIH (GM070348 to H-T.K. and Earthwatch Institute (Grant: ‘Luminous Life in the Great Barrier Reef’ to V.A.P.)

Ancillary

Advertisement