A novel fluorescent protein purified from eel muscle
We discovered that some isolated eel skeletal muscle cells exhibited green fluorescence under a fluorescence stereomicroscope, and we successfully isolated a novel fluorescent protein from the eel muscle homogenate. The protein was a monomer with a molecular mass of 16.5–17 kDa and showed minor and major peaks at 280 and 493 nm, respectively, in the absorption spectrum. The molar extinction coefficient at 493 nm was 41,300 M−1 cm−1 and A280/A493 was 0.083. Excitation and emission spectra of the protein showed maxima at 493 and 527 nm, respectively. Heat treatment at 95°C for 10 min or 5% trichloroacetic acid treatment of the protein caused aggregation of the protein but did not release any fluorescent components such as FAD into the supernatant after centrifugation. Fluorescence of the protein remained after native PAGE, but not after SDS-PAGE. These results indicate that the purified fluorescent protein is not a flavoprotein, and that its fluorescent chromophore is a covalently bound one, such as green fluorescent protein (GFP) from jellyfish Aequoria victoria, but that its fluorescence requires its native conformation within the protein. Based on these results, we can conclude that the fluorescent protein obtained from eel skeletal muscle is a novel GFP-like protein.
KeywordsEelFluorescent proteinGFPGFP-like proteinIsolated muscle cellsSkeletal muscle
We begin this report by describing how a novel fluorescent protein was discovered from eel skeletal muscle. It is known that lipoprotein concentrations in fish sera, including eel sera, are several times higher than those in human sera . We have already shown that primary cultured eel hepatocytes actively synthesize and secrete very-low-density lipoprotein (VLDL)-like lipoprotein . On the other hand, eel skeletal muscle contains lipids at levels of as much as ~20% of the wet weight of the muscle . Then we investigated the lipid transportation mechanism from liver to muscle through lipoproteins. To do this, we studied whether the VLDL-like lipoprotein labeled with fluorescein isothiocyanate (FITC) binds to isolated eel muscle cells. However, some of the isolated eel muscle cells showed green fluorescence without the addition of the FITC-labeled lipoprotein under a fluorescence stereomicroscope. This indicated the occurrence of some fluorescent materials in eel muscle cells. Thus, we attempted to purify these fluorescent materials from the eel muscle.
The purified fluorescent material was a monomeric protein with a molecular mass of 16.5–17 kDa, and the excitation and emission maxima of the protein were 493 and 527 nm, respectively. The absorption spectrum of the purified protein showed two peaks at 280 and 493 nm; the latter was the main peak. Although flavoproteins also show green fluorescence [4–7], their absorption spectra are different from that of the purified fluorescent protein from eel skeletal muscle.
This report describes the purification, some physicochemical properties, and some internal amino-acid sequences of the purified fluorescent protein.
Materials and methods
Aquacultured eels (Anguilla japonica) weighing about 250 g were obtained from Sueyoshi Co. at Kagoshima, Japan.
Isolation of muscle cells
Muscle cells were isolated by collagenase digestion, principally following the approaches of Rosenblatt et al.  and Alam et al. . An eel anesthetized in 0.25% 2-phenoxyethanol was placed on sheets of paper and wiped with 70% ethanol. A round slice approximately 3 cm in width was cut away 5–10 cm below the anus. The skin and bone were removed from the round slice, and the skeletal muscle was minced into small pieces with scissors in a clean bench. Small pieces of the skeletal muscle were washed with 40–50 ml of sterilized phosphate-buffered saline three times and incubated in 15 ml of Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Grand Island, NY, USA) containing 30 mg collagenase S-I from Streptomyces parulus (Nitta-Zeratin, Osaka, Japan) in a 100 ml beaker at 30°C for 2 h, rotating at 100 rpm. After collagenase digestion, isolated muscle cells were filtered and washed on a single layer of nylon gauze with DMEM containing 10% horse serum (Gibco) and 0.5% fetal bovine serum (Gibco). Muscle cells on the gauze were transferred to a 6-cm plastic culture dish and suspended gently by a Pasteur pipette in 4 ml of the same medium.
Extraction of green fluorescent protein from eel muscle
Eels were anesthetized in 0.25% 2-phenoxyethanol, and the skins of the eels were removed. Four fillets obtained from two eels were minced by a mincer, and 100 g of minced muscle was homogenized in 600 ml of cold 20 mM Na phosphate buffer (pH 7.5)–1 mM EDTA by a mixer. The homogenized muscle was centrifuged at 6,000×g for 30 min at 5°C, and the supernatant was carefully transferred to a 2 l beaker so as not to contaminate the solid lipid. After centrifugation, the precipitate was rehomogenized in 400 ml of cold 20 mM Na phosphate buffer (pH 7.5)–1 mM EDTA, and the rehomogenized precipitate was centrifuged. This supernatant was added to the previous supernatant and they were filtered through a Buchner funnel containing filter paper (No. 2) in a cold room (5°C). The filtrate was used as the homogenate.
Purification procedures for the eel fluorescent protein
The purification steps used were as follows: salting out by ammonium sulfate; gel filtration by Sephadex G-75 (GE Healthcare Bio-Science, Piscataway, NJ, USA); absorption chromatography by hydroxyapatite (Bio-Rad Laboratory, Hercules, CA, USA); gel filtration by Sephadex G-50 (GE Healthcare Bio-Science); absorption chromatography by hydroxyapatite; and anion-exchange chromatography by Source 15Q (GE Healthcare Bio-Science). All of these procedures were carried out in a cold room at 5°C, except for the last two procedures, which were carried out at room temperature. Further details are described in “Results.”
The purified fluorescent protein was digested with trypsin (sequence grade 16,800 U/mg) (Promega, Madison, WI, USA), principally according to the in-gel digestion described by Rosenfeld et al. . After SDS-PAGE (17% polyacrylamide gel) of the purified fluorescent protein in the presence of reduced reagent of dithiothreitol (DTT), the fluorescent protein band stained with Coomassie Brilliant Blue R250 was cut from the gel and minced into 1–2 mm cubes by a sterilized knife. The cubes were destained and dehydrated, and the protein in the cubes was reduced by DTT and alkylated by iodoacetoamide. Trypsin digestion at 37°C for 1 h and the extraction of peptides were carried out according to Rosenfeld et al. .
Peptides extracted from the gel were lyophilized, and 110 μl of 0.1% trifluoroacetic acid (TFA) was added. The 110 μl 0.1% TFA solution was centrifuged at 11,000×g for 5 min, and the supernatant was used as the peptides obtained by trypsin digestion of the purified fluorescent protein.
Purification of peptides
The peptides obtained by trypsin digestion were purified by a Sephasil Peptide C18 (5 μm, ST 4.6/100) column (GE Healthcare Bio-Science). The column was set on an Äkta Purifier System and washed adequately with 0.1% TFA before use. Then 100 μl of the peptides were applied to the column and eluted with a 50 ml linear gradient containing 0–60% acetonitrile in 0.1% TFA, measuring the absorbance at 215 nm. Each peptide fraction was lyophilized, dissolved in 20 μl of 0.1% TFA, and the dissolved peptide was centrifuged at 11,000×g for 1 min. The supernatant was used for amino-acid sequence analysis with a protein sequencer (PE Biosystems Procise 492, Applied Biosystems, Foster City, CA, USA).
SDS-PAGE and protein assay
SDS-PAGE was performed according to the method of Laemmli . Low molecular weight markers of phosphorylase b (97 kDa), albumin (66 kDa), ovalbumin (45 kDa), carbonic anhydrase (30 kDa), trypsin inhibitor (20.1 kD), and α-lactalbumin (14.4 kDa) (Bio-Rad Laboratory) were used for SDS-PAGE. Markers of transferrin (75 kDa), ovalbumin (45 kDa), myoglobin (17.6 kDa), and ribonuclease A (13.7 kDa) (GE Healthcare Bio-Science) were used for gel filtration by a Superdex 75HR column (GE Healthcare Bio-Science). Protein was determined by the method of Smith et al.  using bovine albumin as a standard protein.
Observation of eel skeletal muscle and isolated muscle cells using a fluorescence stereomicroscope
Purification of fluorescent protein from eel skeletal muscle
The eel fluorescent protein was purified as follows. The muscle homogenate was subjected to ammonium sulfate fractionation, and the precipitates that formed between 60 and 90% saturation with ammonium sulfate were collected by centrifugation. The precipitates were dissolved in and dialyzed against 10% glycerol–20 mM Na phosphate buffer (pH 7.5)–1 mM EDTA–0.15 M NaCl (glycerol buffer). Glycerol buffer was used to avoid aggregation of proteins during dialysis and gel filtration through the Sephadex G-75 column (26 × 400 mm). After the gel filtration, the fluorescent protein did not form any aggregates. The flow rate and the fraction size of the gel filtration were 0.26 ml/min and 3 ml, respectively. Pooled fractions with high fluorescence intensities were dialyzed against 20 mM Na phosphate buffer (pH 7.0) and used as the G-75 fraction.
The G-75 fraction containing 52.8 mg protein was applied to a hydroxyapatite column (36 × 48 mm) equilibrated with phosphate buffer. Adsorbed proteins were eluted by a 600 ml linear gradient from 20 mM Na phosphate buffer (pH 7.0) to 0.25 M K phosphate buffer (pH 7.0), and the flow rate and the fraction size were 1.89 ml/min and 7 ml, respectively. Pooled fractions with high fluorescence intensities were dialyzed against 20 mM Tris–HCl buffer (pH 7.5)–1 mM EDTA–0.15 M NaCl (Tris buffer) and concentrated by a PM10 membrane (Millipore, Billerica, MA, USA).
When the proteins were lyophilized for concentration, the recovery of the fluorescent protein decreased sharply because most of the proteins became insoluble after lyophilization. The concentrated sample was used as the hydroxyapatite fraction.
Five milliliters of the hydroxyapatite fraction were applied to a Sephadex G-50 column (26 × 900 mm) equilibrated with Tris buffer, and the flow rate and the fraction size were 0.20 ml/min and 3 ml, respectively. Pooled fractions with high fluorescence intensities were dialyzed against phosphate buffer and concentrated by a PM membrane. The concentrated sample was used as the G-50 fraction.
About 4 ml of the G-50 fraction was applied to a hydroxyapatite column (10 × 125 mm) set on an Äkta Purifier System (GE Healthcare Bio-science) and equilibrated with phosphate buffer. Adsorbed proteins were eluted with a 24 ml linear gradient from 20 mM Na phosphate buffer (pH 7.0) to 0.25 M K phosphate buffer (pH 7.0), and the flow rate and the fraction size were 1 ml/min and 2 ml, respectively. The pooled fractions with high fluorescence intensities were dialyzed against 20 mM Tris–HCl buffer (pH 7.5) and concentrated by a YM10 membrane (Millipore). The concentrated sample was used as the hydroxyapatite fraction.
Summary of purification of the eel fluorescent protein
Total protein (mg)
8.41 × 105
4.64 × 105
Sephadex G-75 26 × 400 mm
4.19 × 105
Hydroxyapatite 36 × 48 mm
4.04 × 105
Sephadex G-50 26 × 970 mm
3.33 × 105
Hydroxyapatite 10 × 125 mm
1.47 × 105
2.43 × 105
Source 15Q 4.6 × 100 mm
2.28 × 106
1.64 × 105
Molecular weight determination by SDS-PAGE and Superdex 75 gel filtration
Absorption spectrum of the fluorescent protein
Excitation and emission spectra of the fluorescent protein
Effects of heating, acetone and trichloroacetic acid on the fluorescent protein
Comparison of the fluorescence of partially purified proteins after native PAGE and SDS-PAGE
After the proteins purified by a Sephadex G-50 column had been applied to native PAGE with a 17% polyacrylamide gel, the fluorescence in the gel was measured with an image analyzer (FLA 2000). As shown in Fig. 7a, only one fluorescent band was observed among a few proteins. However, after SDS-PAGE of the same proteins on a 17% polyacrylamide gel in the presence of DTT, no fluorescent band was observed (Fig. 7b).
Amino-acid sequences of peptides obtained by trypsin digestion
The purified protein (20 μg) was denatured by heating at 95°C for 5 min in the presence of SDS and DTT, and applied to SDS-PAGE with a 17% polyacrylamide gel. After staining the protein, the protein band was cut from the gel, and in-gel digestion by trypsin (according to Rosenfeld et al. ) was carried out.
Amino-acid sequences of the peptides obtained by trypsin digestion of the purified fluorescent protein
Amino acid sequence similarities between the obtained peptides and the proteins of other bony fishes were investigated using the Basic Local Alignment Search Tool (BLAST) website (http://blast.ncbi.nlm.nih.gov/Blast.cgi). The sequence of -LVYVQK-WDGKETTVR-ELSDGGDATTPTL- (the sequence for the peptides in fractions 5, 6 and 7) showed 66% sequence identity with the sequence of amino acid residues from 91 to 118 of fatty acid-binding proteins from muscle of Atlantic salmon Salmo salar (accession number NP_001117050) and heart of rainbow trout Onchorhynchus mykiss (accession number NP_001118185). These fatty acid-binding proteins consist of 133 amino acids. Furthermore, the peptide from fraction 8 showed 70% sequence identity with the sequence from amino acid residues 13 to 22 of the same proteins. However, the peptide from fraction 9 showed very low sequence identity with the same proteins.
Absorption spectrum of the eel fluorescent protein
A green fluorescent protein (GFP) was discovered and purified for the first time from jellyfish Aequorea victoria . Flavoproteins show fluorescence like GFP, and their fluorescence is due to the flavin adenine dinucleotide (FAD) within their proteins. However, absorption spectra of GFPs and flavoproteins are different. We then tried to compare the absorption spectra of the eel fluorescent protein and some flavoproteins [4–7]. As shown in Fig. 5, the eel fluorescent protein shows an absorption spectrum with maxima at 280 and 493 nm, and the main absorption maximum was 493 nm. The molar absorption coefficient at 493 nm and the A280/A493 ratio were 41,300 M−1 cm−1 and 0.083, respectively. On the other hand, flavoproteins such as glycine oxidase , d-amino acid oxidase , and amine oxidase  show absorption spectra with maxima at 273–280, 372–380, and 454–456 nm, respectively. The main absorption maxima of these flavoproteins are 273–280 nm, and their ratios of A273–280/A454–456 are 5.54–8.7. Their molar absorption coefficients at 454–456 nm are 11,810–12,600 M−1 cm−1. These characteristics of flavoproteins are different from those of the eel fluorescent protein in the following three ways. The first is the lack of an absorption maximum at 372–380 nm in the eel fluorescent protein. The second is that main absorption maximum was 493 nm in the eel fluorescent protein but is 273–280 nm in flavoproteins. Therefore, the ratios of A273–280/A454–456 for flavoproteins are rather higher than the A280/A493 ratio of the eel fluorescent protein (about 67–105 times higher). The third is the lower molar absorption coefficients at 454–456 nm in flavoproteins compared to that at 493 nm in the eel fluorescent protein. These results suggest that the eel fluorescent protein is not flavoprotein.
All GFP-like proteins found so far contain a chromophore induced from the amino-acid sequence of -X-Tyr-Gly- within the proteins by cyclization, dehydration, and aerial oxidation [14, 15]. This chromophore within the native protein shows a characteristic absorption spectrum with maxima near 400 and 500 nm. If the phenolic hydroxyl derived from the Tyr of the chromophore is protonized, the absorption peak near 400 nm is the main one, but if it is ionized, the peak near 500 nm is the main one . The absorption spectrum of the eel fluorescent protein suggests that the latter form of GFP is present, as found in EGFP, one of theGFP variants, and Azami Green obtained from Galaxeidae coral [14, 16]. The value of the molar absorption coefficient at 493 nm for the eel fluorescent protein was 41,300 M−1 cm−1, and this value was within the range of those found for GFP-like proteins [16, 17].
Characteristics of the fluorescence of the eel fluorescent protein
As shown in Fig. 6, the excitation/emission maxima of the the eel fluorescent protein were 493/527 nm and those of flavoprotein  and GFP-like proteins such as Topaz  and Azami Green  were 454–456/530, 514/527, and 492/507 nm, respectively. The fluorescence of flavoprotein is known to be about 1/20 of that of resolved flavin from the same sample. This is suggested to be due to the strong quenching of the fluorescence of the coenzyme on flavin binding to the protein moiety [4, 5]. The low molar absorption coefficients at 454–456 nm of flavoproteins seem to reflect their low fluorescence. The fluorescence of the eel fluorescent protein was yellowish green, like Topaz fluorescent protein, but the excitation maximum at 493 nm of the eel fluorescent protein was lower than that at 514 nm of Topaz , and it was almost the same as that at 492 nm of Azami Green .
The chromophore within GFP is covalently bound to the protein of GFP, but its fluorescence is lost by denaturing the protein through methods such as heating at 90°C for 5 min or acid and alkaline treatment at pH 2.0 and pH 12.0 [19, 20]. These results show that the fluorescence of GFP requires that the molecule adopts its native conformation. However, the chromophore of GFP remains intact when GFP is denatured, and the chromophore can then be prepared after papain digestion of the denatured GFP in order to elucidate the structure of the chromophore [21, 22]. As described in “Results,” when the purified eel fluorescent protein was treated at 95°C for 10 min, the supernatant showed no fluorescence after the heat treatment. Similarly, the supernatant showed no fluorescence after treatment with 90% acetone or 5% TCA. Flavoproteins containing noncovalently bound FAD show fluorescence of FAD in the supernatant after denaturation by heating or acid [4, 5]. If the eel fluorescent protein is like a flavoprotein and contains a covalently bound FAD, such as peroxisomal l-pipecolic acid oxidase from monkey liver , its fluorescence should be detectable after SDS-PAGE of the denatured protein by an image analyzer. As shown in Fig. 7a and b, no fluorescence was detected after SDS-PAGE of the eel fluorescent protein, but fluorescence was detected after native PAGE. These results indicate that the fluorescent chromophore of the eel fluorescent protein is a covalently bound one, such as GFP, but that its fluorescence requires its native conformation within the protein.
Other characteristics of the eel fluorescent protein
After SDS-PAGE of the purified eel fluorescent protein, the band stained with Coomassie Brilliant Blue R-250 was cut and used for in-gel digestion by trypsin according to Rosenfeld et al. . The peptides recovered after trypsin digestion at 37°C for 1 h were separated using a Sephasil peptide column. As shown in Table 2, amino-acid sequences of nine peptides were determined. As reported by Shimomura  or Cody et al. , the fluorescent chromophore of Aequorea GFP was prepared from papain-digested GFP, and its structure was elucidated. We did not try to detect the fluorescent chromophore in the trypsin-digested sample due to the very small quantity available. Among the nine peptides, the sequence of -X-Tyr-Gly-, which forms the fluorescent chromophore of 4-(p-hydroxybenzylidene)-5-imidazoline within Aequorea GFP [14, 15], was not found. If we had found this sequence, it would suggest that it is used for chromophore formation within the eel fluorescent protein. The sequence of the peptide in fraction 6 contained that of the peptide in fraction 4, and the other peptides did not overlap. The total number of amino acid residues in the peptide sequences determined was 69, which was about 45% of the amino acids in the eel fluorescent protein calculated from the molecular weight of the protein (16.5 kDa), as shown in Figs. 3 and 4. The sequences of the peptides from fraction 5 to 9 were compared with the protein sequences of other bony fishes using BLAST. The results showed that all of the peptide sequences except for the peptide from fraction 9 had sequence similarities with fatty acid-binding proteins from Atlantic salmon and rainbow trout, as described in the “Results.”
The molecular weight of the eel fluorescent protein was smaller than the GFPs from jellyfish and GFP-like proteins from stony coral animals or flavoproteins reported so far. The molecular weight of GFP from Aequorea victoria and GFP mutants is 28 kDa. Most GFP-like proteins from stony coral animals consist of four identical subunits, and the molecular weight of each subunit is about 28 kDa . Flavoproteins such as glycine oxidase from Bacillus subtilis  and d-amino-acid oxidase from Rhodotorula gracilis  have molecular weights of 47 and 39 kDa, respectively. Furthermore, the eel fluorescent protein was a monomer. The low molecular weight and monomeric nature of the eel fluorescent protein would be very useful features if it was used as a marker for gene expression in the future  if the eel fluorescent protein is a group of GFPs.
We observed fluorescence in a slice of eel muscle, as shown in Fig. 1a, and the eels used for the experiments were aquacultured eels. We also observed fluorescence in muscle of natural eels—not only yellow eels but also silver eels—captured in the river (data not shown). Yellow eels are immature eels in the river and silver eels are eels that migrate from the river to the sea for spawning. Though the physiological functions of the eel fluorescent protein are uncertain, the results obtained using BLAST suggest that the eel fluorescent protein functions as a fatty acid-binding protein. However, the fatty acid-binding proteins from Atlantic salmon or rainbow trout do not contain the -X-Tyr-Gly- sequence necessary for fluorescent chromophore formation in GFPs.
We conclude that the eel fluorescent protein is a GFP-like protein based on its absorption spectrum, denaturation experiments, and a comparison between native PAGE and SDS-PAGE. If the cDNA of this protein is cloned (which is currently under investigation), it can be expected to be a good marker for gene expression.
We thank Dr. Kobayashi, Takara Bio Inc., for helpful suggestions. This work was supported in part by a Grant-in-Aid for Basic Research B (10460095) from the Ministry of Education, Culture, Sports, Science and Technology, Japan, and was also supported by Takara Bio Inc. We also thank Messrs Adachi and Honda for technical support.