Crystal structure of semisynthetic obelin‐v

Abstract

Coelenterazine‐v (CTZ‐v), a synthetic derivative with an additional benzyl ring, yields a bright bioluminescence of Renilla luciferase and its “yellow” mutant with a significant shift in the emission spectrum toward longer wavelengths, which makes it the substrate of choice for deep tissue imaging. Although Ca²⁺‐regulated photoproteins activated with CTZ‐v also display red‐shifted light emission, in contrast to Renilla luciferase their bioluminescence activities are very low, which makes photoproteins activated by CTZ‐v unusable for calcium imaging. Here, we report the crystal structure of Ca²⁺‐regulated photoprotein obelin with 2‐hydroperoxycoelenterazine‐v (obelin‐v) at 1.80 Å resolution. The structures of obelin‐v and obelin bound with native CTZ revealed almost no difference; only the minor rearrangement in hydrogen‐bond pattern and slightly increased distances between key active site residues and some atoms of 2‐hydroperoxycoelenterazine‐v were found. The fluorescence quantum yield (Φ _FL) of obelin bound with coelenteramide‐v (0.24) turned out to be even higher than that of obelin with native coelenteramide (0.19). Since both obelins are in effect the enzyme‐substrate complexes containing the 2‐hydroperoxy adduct of CTZ‐v or CTZ, we reasonably assume the chemical reaction mechanisms and the yields of the reaction products (Φ _R) to be similar for both obelins. Based on these findings we suggest that low bioluminescence activity of obelin‐v is caused by the low efficiency of generating an electronic excited state (Φ _S). In turn, the low Φ _S value as compared to that of native CTZ might be the result of small changes in the substrate microenvironment in the obelin‐v active site.

Short abstract

PDB Code(s): 7O3U; 1QV0, 1QV1, 1EJ3, 4NQG, 3KPX, 1UHK, 1UHJ, 1UHI, 1UHH, 5ZAB, 1EL4, 4MRX.

1. INTRODUCTION

Coelenterazine (CTZ), an imidazopyrazinone compound formed by several heterocycles, is one of the most widespread and well‐studied luciferins. It serves as a substrate of bioluminescent reaction in at least nine phyla of marine luminous organisms including jellyfishes, hydroids, ctenophores, copepods, soft corals, worms, crustaceans, mollusks, and vertebrates. ¹ Many of these organisms cannot synthesize CTZ or its derivatives by themselves and thus obtain it from their diet as reported for the hydrozoan jellyfish Aequorea. ² The extensive presence of CTZ in the ocean food chains explains well its abundance as a substrate of bioluminescence reactions in a variety of marine organisms. The deep‐sea luminous copepod Metridia, which is a typical planktonic organism at the bottom of food pyramid, was shown to endogenously synthesize CTZ from two molecules of l‐tyrosine and one molecule of l‐phenylalanine. ³ Thus, this and most likely other copepods can serve as a source of CTZ substrate for many other luminous organisms.

There are various marine CTZ‐dependent bioluminescence systems which use classic luciferases such as Renilla, Gaussia, or Metridia luciferases ⁴ , ⁵ , ⁶ or Ca²⁺‐regulated photoproteins such as aequorin, obelin, or berovin ⁷ , ⁸ , ⁹ , ¹⁰ as CTZ‐oxidizing enzymes. All known Ca²⁺‐regulated photoproteins consist of single polypeptide chain apoprotein (~22 kDa) to which the oxygenated CTZ, 2‐hydroperoxycoelenterazine (2‐hydroperoxyCTZ), is tightly bound. Unlike classic luciferases that “turn over” as an enzyme does, photoproteins can emit light only once because they require new pre‐oxygenated luciferin molecule to be incorporated in the protein active center to produce bioluminescence. The process of photoprotein activation, that is, the formation of 2‐hydroperoxyCTZ, is slow and involve incubation of apoprotein and CTZ in the presence of oxygen and reducing agents such as dithiothreitol or β‐mercaptoethanol under Ca²⁺‐free conditions. ¹¹ , ¹² Since the CTZ molecule becomes oxygenated in the process of activation, later on the photoprotein light emission is completely independent of the presence of molecular oxygen in the reaction mixture, and this is the other feature that distinguishes photoproteins from classic luciferases.

Photoprotein bioluminescence is initiated upon binding of calcium ions to the EF‐hand Ca²⁺‐binding sites of a protein. This event results in small structural changes within the substrate‐binding site leading to the fast decarboxylation of 2‐hydroperoxyCTZ with the elimination of CO₂ and generation of the protein‐bound product, coelenteramide (CTD), in an excited state. ¹³ Relaxation of excited CTD to the ground state is accompanied by light emission with λ_max in the range of 465–495 nm depending on the photoprotein type. These bioluminescent proteins attract great interest due to their broad analytical potential. The main analytical application of the Ca²⁺‐regulated photoproteins comes from their ability to emit light on binding calcium ions at physiological range of [Ca²⁺] within cells. ¹⁴ In fact, photoprotein aequorin is the first calcium indicator applied to studying the role of calcium ions in the regulation of cellular function. ¹⁵

Following the discovery of CTZ, a large number of its analogs have been synthesized in order to acquire the substrates with altered properties such as enhanced bioluminescence intensity, higher or lower decay rate, modified calcium sensitivity, or different emission color. ¹ , ¹⁶ Of note is that the light emission shifted to the long wavelength region is an important feature for the reporter molecule to be efficiently applied for in vivo imaging. Thus, the demand for the red‐shifted bioluminescence reporters is quite high. Numerous CTZ derivatives with various modifications of imidazopyrazinone core or its C2, C6, and C8 substituents are synthesized and tested as the substrates of CTZ‐dependent bioluminescent proteins and some of them are currently used as the substrates of Ca²⁺‐regulated photoproteins and luciferases in different applications including in vivo imaging. ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰

Coelenterazine‐v (CTZ‐v) is a vinylene‐bridged π‐extended coelenterazine derivative (Figure 1). ²¹ The introduction of an additional conjugated double bond (—CH═CH—) leads to a planar conformation of the 6‐(p‐hydroxy)‐phenyl group relative to pyrazine ring and results in an increase of rigidity of the overall structure of the molecule. Due to this steric difference, when used as a substrate of different CTZ‐dependent proteins, CTZ‐v substantially affects their light emission characteristics as compared to those with CTZ. In case of Renilla luciferase, for example, the use of CTZ‐v significantly increases the peak light intensity but provides the total light output comparable to that of the native CTZ (73%). ²² In contrast, the light outputs of Ca²⁺‐regulated photoproteins and Oplophorus luciferase with CTZ‐v are drastically reduced. As for the photoproteins, the luminescence intensity of natural aequorin activated by CTZ‐v (aequorin‐v) comprises only 3.5% of that of aequorin with native CTZ. ²² A similar result is obtained for the recombinant obelin activated with CTZ‐v (obelin‐v) which retains only 1.7% of bioluminescence light output of obelin activated with unmodified CTZ. ²³

FIGURE 1.

Coelenterazine (a) and coelenterazine‐v (b)

The most remarkable feature of CTZ‐v as a bioluminescent substrate is its ability to shift light emission of several CTZ‐dependent bioluminescence proteins toward longer wavelengths when switching substrate from CTZ to the CTZ‐v analog. For instance, bioluminescence spectrum maxima of native Renilla and Oplophorus luciferases with CTZ‐v are shifted from 475 to 512 nm and from 454 to 480 nm, respectively. ²² Moreover, certain luciferase mutants display even greater spectral shifts with CTZ‐v. In the past decade, the mutants of Renilla luciferase with bioluminescence spectra red‐shifted up to λ _max = 547 nm have been described. With CTZ‐v as a substrate, the spectral shift appears to be even more impressive (λ _max = 588 nm). ²⁴ Thus, with the construction of red‐shifted luciferase mutants, CTZ‐v has significantly strengthened its position as a powerful substrate for in vivo imaging.

CTZ‐v was also shown to significantly change bioluminescence spectrum of Ca²⁺‐regulated photoproteins. When activated with CTZ‐v, obelin displays red‐shifted bioluminescence with a peak at 532 and a shoulder at 415 nm while the obelin with CTZ emits light with a maximum at 480 nm and a shoulder at 400 nm. ²³ However, low bioluminescence activities of both aequorin‐v and obelin‐v yet prevent the usage of CTZ‐v analog as a substrate of the photoproteins.

In recent years, tremendous efforts have been directed toward shifting the blue photoprotein bioluminescence to the longer wavelengths through the mutagenesis or by fusing photoprotein with fluorescent proteins. ²⁵ , ²⁶ , ²⁷ The construction of red light‐emitting Ca²⁺‐regulated photoprotein variants would promote the development of a potent set of reporters for monitoring calcium signaling in cells and tissues in various applications. The employment of long‐wavelength CTZ analogs is definitely an attractive tool to achieve the goal of obtaining Ca²⁺‐regulated photoproteins with red‐shifted light emission suitable for the in vivo assays. However, so far when being used as a substrate of the photoproteins, all promising CTZ analogs that could ensure the long‐wavelength shift of the light emission including CTZ‐v yield a low bioluminescence output. The lack of structural data on photoproteins with CTZ‐v bound within the substrate‐binding cavity hinders the productive speculations on the reasons behind the low efficiency of photoprotein bioluminescence with CTZ‐v and consequently limits the experimental approaches to overcome this drawback.

In the present study, we report the crystal structure of Ca²⁺‐regulated photoprotein obelin bound with 2‐hydroperoxy adduct of CTZ‐v (obelin‐v) at 1.80 Å resolution as well as some photophysical properties of obelin‐v.

2. RESULTS AND DISCUSSION

2.1. Overall structure of obelin‐v

Similar to the obelin bound with 2‐hydroperoxy adduct of native CTZ (obelin), ²⁸ , ²⁹ obelin‐v is a globular molecule with the same radius of ~25 Å, formed by two sets of four helices designated as A–D and E–H in the N‐ and C‐terminal domains, respectively (Figure 2a,d). RMSD values of the main chain/side chain atomic positions of obelin‐v versus obelin are 0.814/1.457 Å (PDB: 1QV0) and 0.815/1.540 Å (PDB: 1QV1) (Table 1 and Figure 2b). The comparison of the spatial structures clearly shows that the presence of the 2‐hydroperoxyCTZ‐v molecule instead of 2‐hydroperoxyCTZ has no significant effect on the overall obelin structure. The structures of EF‐hand motives in obelin‐v are also very similar to those of obelin, thereby indicating the absence of local changes provoked by the replacement of CTZ to CTZ‐v (Table 1 and Figure 2b). Of note is that the differences between obelin‐v and obelin structures are slightly higher than those between cognate Ca²⁺‐regulated photoproteins with 2‐hydroperoxyCTZ; RMSD values of the main chain atoms for obelin (PDB: 1QV0) versus aequorin (PDB: 1EJ3), mitrocomin (PDB: 4NQG), and clytin (PDB: 3KPX) are within the range of 0.563–0.785 Å.

FIGURE 2.

Crystal structure of obelin‐v. (a) Overall structure of obelin‐v. The helices are marked by capital letters A–H. The loops are designated I–IV. (b) Superimposition of obelin‐v (pink, PDB: 7O3U), obelin bound with one Ca²⁺ (cyan, PDB: 1QV1), and obelin with no Ca²⁺ (green, PDB: 1QV0). The 2‐hydroperoxyCTZ and 2‐hydroperoxyCTZ‐v molecules are displayed as stick models in the center of the protein; calcium ions are shown as balls. The 2‐hydroperoxyCTZs and calcium ions are colored according to the structure color. (c) Electron‐density map of 2‐hydroperoxyCTZ‐v molecule in the substrate binding cavity of obelin‐v. (d) Amino acid sequence of obelin from O. longissima. ⁸ The helices are shown as pink sticks; the loops involved in the binding of Ca²⁺ are shown as blue sticks

TABLE 1.

Comparison of crystal structures of obelin‐v (PDB: 7O3U), obelin bound with one Ca²⁺ (PDB: 1QV1), and obelin with no Ca²⁺ (PDB: 1QV0)

Structural parts of photoprotein	7O3U vs. 1QV1	7O3U vs. 1QV0
	RMSD values (Å) a
Overall b	0.815/1.540 c	0.814/1.457
N‐terminal domain d	0.666/1.491	0.656/1.418
C‐terminal domain	0.676/1.393	0.663/1.272
EF‐hand I e	0.499/0.843	0.468/0.791
EF‐hand II	0.380/1.437	0.383/1.300
EF‐hand III	0.561/1.040	0.492/1.014
EF‐hand IV	0.408/1.498	0.434/1.349

^a RMSD was calculated using the program Superpose (LSQKAB) (CCP4 Program Suite).

^b Residues 7–195 for all structures.

^c RMSD values of main/side chain atoms.

^d N‐terminal and C‐terminal domains are defined as residues 7–106 and 107–195.

^e EF‐hands I, II, III, and IV are defined as residues 17–54, 58–104, 110–141, and 149–180 for obelins.

Since obelin‐v crystal structure is the first solved structure of semisynthetic obelin, the effect of other CTZ analogs on spatial organization of this photoprotein is unknown. However, by now several crystal structures of semisynthetic aequorins have been determined. ³⁰ , ³¹ These are aequorin bound with 2‐hydroperoxy adducts of CTZ‐n, CTZ‐br, CTZ‐i, CTZ‐cp, and CTZ‐cf3. Comparison of aequorin crystal structure with 2‐hydroperoxyCTZ with those bound with the listed CTZ analogs shows that the replacement of CTZ to the mentioned analogs does not disturb the overall aequorin structure. RMSD values of the main chain atomic positions for aequorin (PDB: 1EJ3) versus aequorin‐n (PDB: 1UHK), aequorin‐br (PDB: 1UHJ), aequorin‐i (PDB: 1UHI), aequorin‐cp (PDB: 1UHH), and aequorin‐cf3 (PDB: 5ZAB) are within 0.359–0.408 Å range. Noteworthy is that the effect these CTZ analogs have on the overall aequorin structure is twice lower in contrast to the influence of CTZ‐v on the overall obelin structure (Table 1) even in the case of CTZ‐n containing bulky β‐naphthylmethyl substituent at C2 position.

The 2‐hydroperoxyCTZ‐v resides in hydrophobic internal cavity in the same position and orientation as 2‐hydroperoxyCTZ (Figure 2b). The electron density distribution near the C2 position of the CTZ‐v molecule clearly displays two oxygen atoms, which corresponds to the peroxide group of the 2‐hydroperoxyCTZ‐v molecule (Figure 2c). It is not always the case with previous photoprotein crystal structures; the electron density next to the C2 position of obelin (PDB: 1EL4) or clytin (PDB: 3KPX), for example, could only be fitted by a single oxygen atom (Figure S1). ²⁸ In the case of another obelin crystal structure (PDB: 1QV0), the electron density of the terminal oxygen atom is very weak but becomes significantly more intensive after soaking the crystals with trace [Ca²⁺] (PDB: 1QV1). ²⁹ The soaking procedure results in the appearance of calcium ion in the first Ca²⁺‐binding site of obelin. Although Ca²⁺ turns out to be of incomplete pentagonal bipyramidal coordination characteristic of EF‐hand Ca²⁺‐binding proteins, ³² , ³³ it is suggested that even incomplete attachment of Ca²⁺ increases a rigidity of the residues composing the active site and decreases flexibility of the terminal oxygen atom, thereby enhancing its electron density. This conclusion agrees with the results of studies on stability of different conformational states of obelin. ³⁴ It is shown that conformational states of obelin with calcium (either calcium‐loaded apo‐obelin or Ca²⁺‐discharged obelin) are more stable than those without calcium.

However, there is another plausible explanation of the weak electron density for the terminal oxygen atom of peroxide group at C2 position. With the use of aequorin crystals, it has been demonstrated that electron density corresponding to this oxygen atom decreases with the increase of exposure time to X‐ray irradiation that, in fact, is accompanied by the damage of peroxide group. ³⁰ This statement looks reasonable because it was shown that the damage to peroxide group in aequorin can already be induced by near‐UV irradiation. ³⁵ Moreover, in the case of ctenophore Ca²⁺‐regulated photoproteins even the exposure to visible light is sufficient to damage peroxide group at C2 position of CTZ. ¹ , ³⁶

Similar to obelin (PDB: 1QV1), ²⁹ the first Ca²⁺‐binding loop in obelin‐v structure is also occupied by calcium ion (Figure 3). However, in contrast to obelin structure in which the appearance of calcium results from crystals soaking with trace [Ca²⁺], obelin‐v most likely binds calcium ion from the crystallization solution. In obelin‐v, the coordination geometry of Ca²⁺ coincides with that of obelin (PDB: 1QV1); three oxygen ligands are provided by the side chains of Asp30, Asn32, and Asn34, another one is given by the main chain of Lys36, and two oxygen ligands are supplied by water molecules (Figure 3). Glu41 in position 12 of the Ca²⁺‐binding loop which always provides two oxygen ligands to coordinate calcium ions in photoproteins ³⁷ , ³⁸ does not contribute to the coordination of this calcium ion, thereby indicating incomplete binding of Ca²⁺ as it takes place in the case of obelin (PDB: 1QV1).

FIGURE 3.

Stereoview of the first Ca²⁺‐binding loop of obelin‐v. Calcium ion (green) and water molecules (red) are shown as balls. Oxygen and nitrogen atoms of amino acid residues are colored in red and blue, respectively. Hydrogen bond distances are shown with dashed lines. Distances are indicated in Å

Noteworthy is that calcium is manifested only in the first Ca²⁺‐binding loop of both obelin‐v and obelin. This can occur only if the affinity of the first Ca²⁺‐binding loop to calcium significantly exceeds the affinity of the Ca²⁺‐binding loops resided in the C‐terminal domain. In contrast, for aequorin it is suggested that the Ca²⁺‐binding loop III located in the C‐terminal domain has the highest affinity to Ca²⁺ as compared to the affinities of other Ca²⁺‐binding loops. This conclusion is based on the determination of K _D for Ca²⁺ of synthetic peptide fragments of 20–22 amino acid residues that mimic the aequorin EF‐hand motifs ³⁹ as well as NMR studies on binding of magnesium ions with Ca²⁺‐binding loops of aequorin. ⁴⁰ If this is the case then Ca²⁺‐binding sites with the highest affinity to Ca²⁺ can have different location in different Ca²⁺‐regulated photoproteins. Since Ca²⁺‐binding sites of highest affinity bind calcium ion in the first instance, it brings about dissimilar order of calcium binding in the different photoproteins and consequently could mainly account for the revealed distinctions in light emission kinetics and sensitivity to calcium of photoproteins originated from different organisms. ¹⁴ , ⁴¹

Of note is that in all Ca²⁺‐regulated photoproteins the C‐terminus caps the substrate‐binding cavity, ensuring a solvent‐inaccessible and nonpolar environment of the CTZ derivative. ²⁹ , ⁴² , ⁴³ The inaccessibility of the internal substrate‐binding cavity of photoproteins to solvent is additionally secured by the hydrogen bond interactions between helix A residues and residues from helix H and the C‐terminus. ⁴⁴ , ⁴⁵ In obelin‐v, like in obelin, the inaccessibility of the internal substrate‐binding cavity to solvent is also ensured by the hydrogen bonds formed by Arg21 located in helix A and C‐terminal Pro as well as Asp187 and Phe178 (Figure 4).

FIGURE 4.

Hydrogen bond network formed by Arg21 located in helix A and C‐terminal Pro 195 together with Asp187 and Phe178, which ensures solvent inaccessibility of the substrate‐binding cavity in obelin‐v (a) and obelin (b)

2.2. Hydrogen bond network in the substrate‐binding cavity of obelin‐v

The residues surrounding the 2‐hydroperoxyCTZ‐v molecule within internal photoprotein cavity at the distance of 4 Å originate from helix A (His22, Met25, and Leu29), helix B (Ile42, Lys45, and Ile50), helix C (Phe72), helix D (Phe88 and Trp92), helix E (Ile111, Trp114, Gly115, and Phe119), helix F (Tyr138), the loop linking helices F and G (Ile144), helix H (Met171, His175, and Trp179), and the C‐terminus of the protein (Tyr190). This amino acid environment is practically identical to that forming the substrate‐binding cavity of obelin (PDB: 1QV1). Only one difference is found—in obelin‐v Trp135 from the helix F is excluded from the 4 Å range that is due to both the slight shift of this residue and the small change in orientation of the second oxygen atom of 2‐hydroperoxy group of CTZ‐v (Figure S2).

In obelin‐v, the amino acid residues and water molecule proposed to affect the decarboxylation reaction and emitter formation ¹³ , ⁴⁶ reside in the same positions of the substrate‐binding cavity as in obelin (PDB: 1QV1) (Figure 5a). There are only a few spatial differences to be mentioned, namely in obelin‐v the Trp92 and Trp179 and to a lesser extent His22 and Phe88 are slightly shifted as compared to their positions in obelin. It is worth noting that the small shifts in the region of His22‐Phe88‐Trp92 triad and 2‐hydroperoxyCTZ‐v observed in the obelin‐v crystal structure are overall consistent with the results of optimization of S₁‐state structures in both obelin and obelin‐v, ²³ where changes in spatial arrangement of the key residues and hydrogen bond distances due to the substrate modification are clearly visible suggesting good prognostication ability of the theoretical calculations.

FIGURE 5.

Substrate‐binding cavities of obelin‐v and obelin. (a) Stereoview of superimposition of the corresponding 2‐hydroperoxyCTZ molecules with the key residues facing into the substrate‐binding cavities of obelin (cyan) and obelin‐v (pink). The 2‐hydroperoxyCTZs and water molecules are colored according to the structure color, water molecules are shown as balls. Two‐dimensional drawing of the hydrogen bond network in obelin‐v (b) and obelin (c). Hydrogen bonds are shown as dashed lines, possible hydrogen bonds are shown as arrows. Distances are given in Å

The hydrogen bond networks formed by the atoms of 2‐hydroperoxy adducts of CTZs and the surrounding key amino acid residues in obelin‐v and obelin are shown in Figure 5b,c. In both obelins, His22 is hydrogen bonded with the oxygen of the 6‐(p‐hydroxy)‐phenyl group of CTZ, and the hydroxyl group of Tyr190 forms a strong hydrogen bond to the peroxide group at C2 position of CTZ. Although the Nε atom of Trp92 is at a hydrogen bond distance from the oxygen of the 6‐(p‐hydroxy)‐phenyl group of both CTZ and CTZ‐v, the corresponding hydrogen bonds are not revealed by PyMOL program in both obelins. His175 residues form strong hydrogen bonds with Tyr190 residues and are found at hydrogen bond distances from carbonyl oxygens at C3 position (Figure 5b,c). In addition, in both obelins the oxygen of Tyr138 hydroxyl group forms hydrogen bonds with N1 of both 2‐hydroperoxyCTZs.

The substrate‐binding cavities comprise two water molecules (W₁ and W₂) which are located in the same positions in both obelins (Figure 5b,c). The W₁ is stabilized by hydrogen bonds with the surrounding residues and the OH group of the 2‐(p‐hydroxy)‐benzyl substituent of both CTZs. Another water molecule (W₂) which is proposed to be involved in decarboxylation reaction serving as a bridge for proton transfer to dioxetanone anion ¹³ , ³⁸ is positioned by hydrogen bonds with Tyr138 and His64 (Figure 5b,c). In addition, in obelin‐v W₂ might form a weak hydrogen bond with Nε atom of Trp114 whereas in obelin this hydrogen bond is not determined by PyMOL probably due to a slightly higher distance (Figure 5c). The third water molecule (W₃) which is possibly involved in the formation of a “proton channel” for proton transfer from solvent to dioxetanone anion ⁴⁶ is found close to the surface of the photoprotein molecule and is stabilized by hydrogen bonds with His64, Ser47, and Thr61 in both obelins (Figure 5b,c).

In obelin, there is a hydrogen bond revealed between Nε atom of Trp179 and oxygen of the carbonyl group at C3 position of 2‐hydroperoxyCTZ while in obelin‐v the corresponding bond is absent (Figure 5b,c). This difference is undoubtedly caused by the shift of Trp179 due to the additional ring in CTZ‐v leading to an increase in distance by ~0.6 Å between Nε atom of Trp179 and carbonyl group.

2.3. Spectral properties of obelin‐v

Absorption spectra of free CTZ and CTZ‐v, active obelin and obelin‐v, and the corresponding Ca²⁺‐discharged photoproteins are shown in Figure 6. Owing to the presence of additional ring the absorption spectrum of free CTZ‐v is red‐shifted as compared to native CTZ. In a visible region CTZ‐v in solution displays a spectrum with λ _max at 458 nm and two shoulders at ~430 and 490 nm (Figure 6a). Conversion of CTZ‐v into 2‐hydroperoxyCTZ‐v within obelin substrate‐binding cavity shifts its absorption toward longer wavelengths—obelin‐v reveals a spectrum with λ _max at 473 nm and two shoulders at ~450 and 510 nm. The Ca²⁺‐discharged obelin‐v, bearing the reaction product CTD‐v, displays a prominent absorption peak at 380 nm with a shoulder at ~360 nm (Figure 6a). For comparison, Figure 6b shows absorption spectra of the corresponding states of native CTZ—free CTZ in solution, 2‐hydroperoxyCTZ within obelin, and CTD within Ca²⁺‐discharged obelin have absorption spectral maxima at 435, 460, and 344 nm, respectively.

FIGURE 6.

Absorption spectra. (a) Free CTZ‐v (black), obelin‐v (red) and Ca²⁺‐discharged obelin‐v (blue). (b) Free CTZ (black), obelin (red) and Ca²⁺‐discharged obelin (blue). Left ordinate axes are for absorbance in the range of 250–400 nm, right ordinate axes—for the range of 400–550 nm. CTZs are in ethanol; obelins—in 1 mM EDTA, 20 mM Tris–HCl pH 7.2; Ca²⁺‐discharged obelins—in 1 mM CaCl₂, 20 mM Tris–HCl pH 7.2. All samples are in concentration of 8.93 μM

The bioluminescence spectra of obelin and obelin‐v and fluorescence spectra of their Ca²⁺‐discharged conformational states are summarized in Table 2 and Figure 7. The obelin emits blue light with a spectral maximum at 480 nm and a shoulder at 390 nm (Figure 7a). The use of CTZ‐v as a substrate for the reaction gives bimodal emission with maxima at 415 and 535 nm. The main maximum of obelin‐v is red‐shifted by 55 nm while a maximum at shorter wavelengths is shifted only by 25 nm as compared to those of obelin (λ _max = 480 nm, shoulder at 390 nm) (Table 2 and Figure 7a). The bimodal bioluminescence spectrum of obelin‐v is not unique. The mutants of obelin ⁴⁶ and aequorin ¹ with substitution of Trp92 and Trp86 to Phe, respectively, and activated by native CTZ also display bimodal emission spectra with λ _max at 410 and 470 nm for obelin mutant and with λ _max at 400 and 455 nm for aequorin mutant. The light emission spectrum of aequorin activated by CTZ‐e, ¹ , ²¹ , ²² CTZ analog similarly to CTZ‐v containing an additional but different linker (—CH₂—CH₂—) between C5 atom of imidazopyrazinone and Cα atom of 6‐(p‐hydroxy)‐phenyl group of CTZ is also bimodal but with peaks at 405 and 472 nm. Of note is that the activity of this semisynthetic aequorin is ~50% of that of aequorin with native CTZ. ¹

TABLE 2.

Spectral characteristics of obelin and obelin‐v and their Ca²⁺‐discharged conformational states

	Obelin	Obelin‐v
Bioluminescence activity (%)	100.0	1.7
Bioluminescence maximum a (nm)	480/390	535/415
Fluorescence maximum b (nm)	515/415	530/435
Fluorescence quantum yield c	0.19 ± 0.02	0.24 ± 0.02

^a For bimodal spectra λ _max is shown in bold.

^b Excitation wavelength is 373 nm.

^c Fluorescence quantum yield was determined at 25°C.

FIGURE 7.

Spectral properties of obelin and obelin‐v. (a) Bioluminescence spectra of obelin (left ordinate, black line) and obelin‐v (right ordinate, red line) normalized to protein concentration. Insertion shows normalized bioluminescence spectra of obelin (black line) and obelin‐v (red line) (b) Fluorescence spectra of corresponding Ca²⁺‐discharged photoproteins divided by the absorption at the wavelength of excitation (373 nm). Vertical lines indicate the wavelength of maxima of the components determined by fitting with the sum of two Gaussian functions

The fluorescence spectrum of Ca²⁺‐discharged obelin‐v is also bimodal with peaks at 435 and 530 nm, while fluorescence of the corresponding obelin conformational state reveals one prominent maximum at 515 nm with almost indistinguishable shoulder at 415 nm (Table 2 and Figure 7b). Thus, the excited state energies of two obelin‐v emitters are lower than that of obelin, which is in good agreement with the observed red shifts in the absorption spectra (Figure 6a). Noteworthy is that the red shift of fluorescence maximum of Ca²⁺‐discharged obelin‐v relative to that of Ca²⁺‐discharged obelin is only 15 nm while the bioluminescence maximum of obelin‐v is shifted by 55 nm (Table 2 and Figure 7b). The shift of fluorescence peak at shorter wavelengths is also less (20 nm) than the one between the corresponding bioluminescence peaks (25 nm), but the difference between bioluminescence and fluorescence maxima for the short wavelength peaks is not significant as in the case of emission maxima at longer wavelengths. The observed long wavelength shifts in absorption and emission spectra of obelin‐v could be a result of an enhanced π‐electron system of CTD‐v (i.e., increased degree of conjugation) owing to addition of a vinylene bridge between the pyrazine and phenol rings of CTZ, which is in agreement with a general rule of photophysics of polyaromatic compounds. ⁴⁷ Of particular note is that the experimentally revealed long wavelength shifts of the emission maxima of obelin‐v were successfully predicted through combined QM/MM calculations as well as molecular dynamics simulations and were calculated as 25 and 84 nm for the emission at shorter and longer wavelengths, respectively. ²³

The quantum yield (Φ _BL) of bioluminescent reaction is an important characteristic that, in fact, shows the efficiency of conversion of substrate molecules into photons during reaction. The quantum yields of reactions catalyzed by even cognate bioluminescent proteins utilizing the same substrate can vary significantly. For instance, quantum yields of reactions catalyzed by cognate luciferases from the Brazilian click beetle, Pyrearinus termitilluminans, and firefly Luciola cruciata which use the same substrate, d‐luciferin, are 0.61 and 0.43, respectively, differing almost 1.5 times. ⁴⁸ , ⁴⁹ Whereas the quantum yields of reactions catalyzed by beetle or firefly luciferases are high the Φ _BL of reaction of coelenterazine oxidation catalyzed by luciferase from the sea pansy Renilla reniformis is only 0.053. ⁵⁰ At the same time, the quantum yield of bioluminescent reaction of Ca²⁺‐regulated photoproteins which also use coelenterazine as a substrate is significantly higher varying in the range of 0.19–0.25. ⁵¹

The quantum yield of bioluminescent reaction is the result of multiplication of three efficiencies: the yield of the ground state (S ₀) reaction (Φ _R), the singlet chemiexcitation yield (Φ _S), and the fluorescence quantum yield of the chromophore (Φ _FL). Different values of Φ _BL can be attributed to changes in either Φ _R, Φ _S, or Φ _FL, or to a combination of changes thereof. We determined the fluorescence quantum yields of Ca²⁺‐discharged obelin and obelin‐v by using a comparative method with quinine sulfate as a reference (Table 2). ⁵² It turns out that the fluorescence quantum yield of Ca²⁺‐discharged obelin‐v exceeds that of Ca²⁺‐discharged obelin ~1.3‐fold. The Φ _R value characterizes the ground state yield which in our case is, in fact, the yields of CTD‐v and CTD after bioluminescent reactions. Since both obelin and obelin‐v are ready‐to‐go enzyme‐substrate complexes containing 2‐hydroperoxy adducts of CTZ and CTZ‐v, the Φ _R values for both reactions should be very similar and, in effect, quite high. For instance, only CTD with a little contamination by coelenteramine, which is the product of CTD hydrolysis, could be isolated after bioluminescent reaction of aequorin. ⁵³ Thus, the ineffective generation of a singlet electronic excited state (Φ _S) is most likely the reason of the low bioluminescence activity of photoproteins activated by CTZ‐v.

The low yield of the singlet electronic excited state at the use of CTZ‐v as a substrate of photoproteins is not conditioned by its molecular structure since in the case of Renilla luciferase CTZ‐v provides a light output comparable to that of native CTZ. ²² It means that low Φ _S value is rather caused by different amino acid environment of CTZ‐v in active sites of Ca²⁺‐regulated photoproteins and Renilla luciferase and probably by different residues involved in both catalytic oxidation of the substrate and emitter formation. At the same time, we cannot exclude the small changes in positions of some residues in the substrate‐binding cavity revealed in obelin‐v structure to affect the Φ _S value.

In obelin‐v, the distance between His22, which is proposed to be involved in the emitter formation, and OH group of 6‐(p‐hydroxy)‐phenyl substituent of the substrate is increased for 0.2 Å as compared to that in obelin (Figure 5b,c). According to the suggested proton‐relay mechanism, ¹³ , ³⁸ , ⁵⁴ the neutral CTD is the primary excited product in bioluminescence of Ca²⁺‐regulated photoproteins emitting light at the shorter wavelengths. Light emission at longer wavelengths is proposed to originate from the excited phenolate anion that arises as a result of transient proton dissociation of the OH group of the 6‐(p‐hydroxy)‐phenyl substituent of CTZ in the direction to His22. The elongation of hydrogen bond between His22 and OH group might decrease the efficiency of the excited‐state proton dissociation ¹³ , ⁵⁵ yielding an increased short wavelength emission from the neutral CTD* (Figure 7).

The other residues that are known to control the light emission color in photoproteins and which slightly change their positions in obelin‐v are Trp92 and Phe88 (Figure 5). The substitution of those in obelin and the corresponding residues in aequorin bring about the changes in bioluminescence spectra of photoproteins, thereby indicating their significant influence on the excited state of CTD. ⁴⁶ , ⁵⁶ As this takes place, none of these mutations exerts an effect on the overall structure and the dimensions of the active sites compared to the wild type photoprotein. ⁵⁷ Based on these findings, one attributes the changes of light emission color to the alteration of hydrogen bond network around OH group of 6‐(p‐hydroxy)‐phenyl substituent of CTZ. Thus, even the small changes in hydrogen bond arrangement similar to those observed in obelin‐v could affect the excited state generation and consequently the light emission spectrum of photoproteins.

Trp179 along with His175 and Tyr190 form another triad in photoprotein active site. ¹³ In contrast to the above mentioned His—Trp—Tyr (Phe in obelin) triad surrounding the 6‐(p‐hydroxy)‐phenyl substituent of CTZ, the substitution of these residues does not influence light emission color but significantly affects the bioluminescence activity as well as stability and efficient formation of 2‐hydroperoxy adduct of CTZ. ⁵⁸ , ⁵⁹ The Trp179 is another residue which position is slightly changed in obelin‐v (Figure 5). While in obelin the nitrogen atom of indole ring of this Trp is hydrogen bonded with C3 carbonyl oxygen of 2‐hydroperoxyCTZ, in obelin‐v this hydrogen bond is absent due to the increase of a distance by 0.6 Å. It is worth attention that at the replacement of Trp179 in obelin (Trp173 in aequorin) to amino acids with different donor–acceptor properties of side chains, only obelin mutants W179Y and W179F correspondingly retain 23% and 67% of bioluminescence activity. ⁵⁸ This clearly indicates that this Trp and its proper orientation relative to 2‐hydroperoxyCTZ is important for the effective light emission.

These His—Trp—Tyr triads are also proposed to perform another function. According to the theoretical studies, the hydrogen bond network formed by these residues can severely affect the pathways of dioxetanone decomposition, and consequently the efficiency of light emission reaction of photoproteins. ⁶⁰ In order to provide a complete electron transfer for the formation of a radical pair, the combination of two processes is believed to be important, which are the deprotonation of the phenolic OH group at the C6 position of CTZ facilitated by the Tyr82—His16—Trp86 triad and intramolecular charge transfer achieved within the neutral peroxide by a proton delivery conciliated by another triad motif, Tyr184—His169—Trp173. If this is indeed the case, then even the small changes in positions of these residues and alterations in hydrogen bond distances revealed in substrate‐binding cavity of obelin‐v can substantially influence the light output efficiency.

3. CONCLUSIONS

In this study, we report the crystal structure of the semisynthetic obelin bound with 2‐hydroperoxy adduct of CTZ‐v determined at 1.80 Å resolution. Comparison of the crystal structure of obelin‐v with that of obelin bound with native 2‐hydroperoxyCTZ reveals no significant structural changes in either overall structure or in CTZ‐binding pocket. Only small changes in positions of several residues leading to the minor alterations in the hydrogen bond network and the hydrogen bond distances between some active site residues and atoms of 2‐hydroperoxyCTZ‐v molecule are found. We also determined the fluorescence quantum yields of Ca²⁺‐discharged obelin‐v and Ca²⁺‐discharged obelin. Since Φ _FL value of Ca²⁺‐discharged obelin‐v turns out to be even higher than that of Ca²⁺‐discharged obelin, we propose that a reason for the low bioluminescence activity of obelin‐v is ineffective generation of a singlet electronic excited state (Φ _S) that, in turn, could be a result of small changes in hydrogen bond network around 2‐hydroperoxyCTZ‐v molecule.

4. MATERIALS AND METHODS

4.1. Materials

Native CTZ and CTZ‐v were obtained from NanoLight Technology, a division of Prolume Ltd. (Pinetop, AZ, USA). Other chemicals, unless otherwise stated, were from Sigma–Aldrich and the purest grade available.

4.2. Protein preparation

For apophotoprotein production, the transformed Escherichia coli BL21‐Gold (DE3) Codon Plus (RIPL) cells were cultivated with vigorous shaking at 37°C in LB medium containing ampicillin (200 μg/ml). Protein expression was induced with 0.5 mM IPTG at OD₅₉₀ 0.6–0.8 and the cultivation was continued for another 3 h. Recombinant apo‐obelin was extracted from E. coli inclusion bodies by 6 M urea as previously reported. ⁶¹ Apophotoprotein was purified on a DEAE‐Sepharose Fast Flow column (GE Healthcare, Chicago, IL, USA), and then concentrated with the use of 10 kDa Amicon Ultra Centrifugal Filters (Merck Millipore). The concentration of apophotoprotein was determined using the corresponding molar extinction coefficient at 280 nm calculated with the ProtParam tool. ⁶²

The active photoproteins were produced as described elsewhere. ⁶¹ Apophotoproteins were activated overnight with a two‐fold molar excess of CTZ or CTZ‐v in a buffer 5 mM EDTA, 10 mM DTT, 20 mM Tris–HCl pH 7.2 at 4°C. Active photoproteins were separated from apophotoproteins, unbound CTZs, and DTT by ion‐exchange chromatography on HiTrap Q HP column or Mono Q™ 5/50 (GE Healthcare, Chicago, IL, USA). Active photoproteins and apophotoproteins were eluted as separate peaks with the linear salt gradient of 1 M NaCl in 5 mM EDTA, 20 mM Tris–HCl pH 7.2. Protein concentration was determined with the Dc Bio‐Rad protein assay kit.

4.3. Crystallization, data collection, structure solution, and crystallographic refinement

For crystallization, obelin‐v sample after ion‐exchange chromatography was concentrated to 9.0 mg/ml using 10 kDa Amicon Ultra Centrifugal Filters (Merck Millipore) with simultaneous replacement of chromatography buffer to a buffer containing 2 mM EDTA, 10 mM Bis‐Tris pH 6.5 (Hampton Research). The crystals of obelin‐v were grown using the hanging‐drop vapor‐diffusion technique. The best condition for obelin‐v crystallization was a solution of 2.1 M dl‐Malic acid pH 7.0 (Wizard IV, Emerald BioSystems). Similar to the obelin crystals, ⁵⁴ the crystals of obelin‐v grew as light‐yellow rod‐shaped crystals after 5–7 days at 4°C. Of note is that the crystals grew only when the drop comprised 2 μl of protein solution and 1 μl 2.1 M dl‐Malic acid. For X‐ray diffraction analysis, the crystals were picked up from the crystallization drop using fiber loops and were flash‐cooled in liquid nitrogen. The solution of 30% glycerol was used as a cryoprotectant.

Data from the obelin‐v crystals were collected at the Spring‐8 beam line 41XU, Hyogo, Japan, using a Pilatus3 6 M detector (X‐ray wavelength 1.0000 Å). Native diffraction data were indexed, integrated and scaled to 1.80 Å resolution using the XDS software. Phases were determined by molecular replacement with Phaser ⁶³ using the structure of obelin from Obelia longissima (PDB: 1QV0) as a search model. The final model of the obelin‐v was refined with PHENIX. ⁶⁴ Manual adjustments to the model were performed using Coot. ⁶⁵ The final refinement statistics are shown in Table 3. For the purposes of visualization and superposition of the molecular structures the PyMOL 2.5.0 (Schrödinger, LLC) was applied. RMSD was calculated using the program Superpose (LSQKAB) of CCP4 Program Suite. ⁶⁶ , ⁶⁷ Parameters to detect hydrogen bonds were 3.6 Å for an ideal geometry and 3.2 Å for minimally acceptable geometry, 180° for a hydrogen bond cone, and 63° for the maximal hydrogen bond angle. ⁶⁸

TABLE 3.

Data collection and structure refinement statistics for obelin‐v

Data collection
Space group	P6222
Cell dimensions
a, b, c (Å)	42.88, 42.88, 375.44
α, β, γ (°)	90.0, 90.0, 120.0
Resolution (Å)	50.00–1.80 (1.91–1.80)
R merge (%)	10.9 (62.9)
I/σI	10.69 (1.73)
Completeness (%)	99.4 (96.5)
CC (1/2)	99.5 (88.3)
Redundancy	8.26 (5.68)
Refinement
Resolution (Å)	37.14–1.80
No. of reflections	20,430
R work/R free (%)	24.0/26.4
Model composition
Nonhydrogen atoms	1,663
Protein residues	189
B‐factors (Å2)
Protein	57.016
Ligand/ion	29.760
Water	53.685
RMSDs
Bond lengths (Å)	0.012
Bond angles (°)	1.488
Validation
MolProbity score	2.14
Clash score	11.88
Poor rotamers (%)	3.66
Ramachandran plot statistics (%)
Favored regions	97.37
Allowed regions	2.63
Disallowed regions	0.00

Note: Values in parentheses are for the highest resolution shell.

The spatial structure of obelin‐v was obtained by X‐ray diffraction of protein crystals with final resolution of 1.80 Å and deposited at PDB under ID 7O3U (Table 3). The final model of obelin‐v includes 188 of the 195 amino acids, the 2‐hydroperoxy CTZ‐v molecule, the ion of Ca²⁺, and 86 solvent molecules. Residues 1–6 are not visible in the electron‐density maps, as is commonly observed for the N‐terminal residues in the structures of other Ca²⁺‐regulated photoproteins. The chemical structure of the 2‐hydroperoxyCTZ‐v was deposited at PubChem data base under SID 440975074.

4.4. Determination of the specific bioluminescence activities

Specific bioluminescence activities of obelin and obelin‐v were measured with luminometer BLM‐003 (Oberon‐K, Krasnoyarsk, Russia) by rapid injection of 10 μl of photoprotein solution in 1 mM EDTA, 20 mM Tris–HCl pH 7.2 into a luminometer cell containing 490 μl of 2 mM CaCl₂ in 50 mM Tris–HCl pH 8.8 at room temperature. In order to calculate specific bioluminescence activities, the bioluminescence signal was integrated over 10 s.

4.5. Spectral measurements

Absorption spectra were obtained with an UV‐2600 double‐beam spectrophotometer (Shimadzu, Kyoto, Japan) or Cary 5000 spectrophotometer (Agilent Technologies, Australia).

Bioluminescence spectra were measured with a Cary Eclipse spectrofluorometer (Agilent, Santa Clara, CA, USA) in 1 mM EDTA, 20 mM Tris–HCl pH 7.2. The slit was set to a bandpass of 5 nm. Bioluminescence was initiated by injecting CaCl₂ in 20 mM Tris–HCl pH 7.2 into the protein solution. The concentration of free calcium was around 0.5 μM to provide a constant light level during the spectral scans.

Ca²⁺‐discharged photoprotein samples for the fluorescence measurements were prepared after the bioluminescence reaction ceased. All photoprotein samples were desalted on the HiTrap Desalting column with Sephadex G‐25 resin (GE Healthcare, Chicago, IL, USA) equilibrated with 1 mM CaCl₂, 20 mM Tris–HCl pH 7.2. The fluorescence spectra of Ca²⁺‐discharged photoproteins were recorded with Fluorolog‐3 spectrofluorometer (Horiba Jobin Yvon, France). The fluorescence was measured within the range of 385–740 nm after excitation at 373 nm, excitation and emission slits were set to a bandpass of 1.8 nm.

All bioluminescence and fluorescence spectra were corrected for the detector spectral sensitivity with an algorithm supplied with the instruments.

To determine the maxima of the spectral components the emission spectra were converted in the wavenumber scale and fitted with a sum of Gaussian functions ⁶⁹ using Origin Lab software. The λ _max of the emission bands were obtained from the approximating curves after inverse conversion to the wavelength scale.

4.6. Determination of the fluorescence quantum yield

The fluorescence quantum yield of Ca²⁺‐discharged obelin and obelin‐v was obtained by the comparative method using quinine sulfate in 0.05 M H₂SO₄ (Φ _FL = 0.6 ± 0.02) as the reference ⁵² and calculated according to the following equation ⁷⁰ :

$$\mathrm{\Phi FL},\mathrm{x}=\mathrm{\Phi FL},\mathrm{st}\frac{m_x}{m_{\mathrm{st}}}\frac{n_x^2}{n_{\mathrm{st}}^2}$$

where x and st indicate the sample and standard solution, respectively, Φ _FL is the fluorescence quantum yield, m is the gradient derived from the linear regression analysis when plotting integrated fluorescence intensity against absorbance, and n is the refraction index of the solutions (the value of each was 1.334). The excitation wavelength was 373 nm for all samples. All samples were diluted in order that the absorption at 373 nm was less than 0.05 to ensure that the fluorescent quantum yield would be independent of concentration effects.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS

Marina D. Larionova: Investigation (equal); visualization (equal). Lijie Wu: Formal analysis (equal). Elena V. Eremeeva: Funding acquisition (equal); investigation (equal); project administration (equal); writing – original draft preparation (equal); writing – review and editing (equal). Pavel Natashin: Formal analysis (equal); visualization (equal). Dmitry V. Gulnov: Investigation (equal). Elena V. Nemtseva: Formal analysis (equal); methodology (equal); validation (equal), writing – review and editing (equal). Dongsheng Liu: Conceptualization (equal); funding acquisition (equal). Zhi‐Jie Liu: supervision (equal); methodology (equal); writing – review and editing (equal). Eugene S. Vysotski: Conceptualization (equal); funding acquisition (equal); project administration (equal); writing – review and editing (equal); supervision (equal).

Supporting information

Figure S1 Electron‐density map of 2‐hydroperoxyCTZ molecule in the substrate binding cavity exemplifying visibility of two oxygen atoms (obelin mutant OL138F, PDB: 4MRX) (a) or one oxygen atom (clytin, PDB: 3KPX) (b).

Figure S2. Stereoview of superimposition of the corresponding 2‐hydroperoxyCTZ molecules with the amino acid environment at the distance of 4 Å in obelin‐v (pink) and obelin (cyan). The 2‐hydroperoxyCTZs are colored according to the structure color.

Larionova MD, Wu L, Eremeeva EV, Natashin PV, Gulnov DV, Nemtseva EV, et al. Crystal structure of semisynthetic obelin‐v . Protein Science. 2022;31:454–469. 10.1002/pro.4244

DATA AVAILABILITY STATEMENT

The diffraction and structural data have been submitted to protein data bank with PDB‐entry 7O3U and authors agree to make all protocols and data available for the readers.