2.1. Specific Bioluminescence Activity
The coelenterazine analogues (
Figure 1) were tested with aequorin from
Aequorea victoria [
17,
18,
19] and obelin from
Obelia longissima [
20,
21]. These Ca
2+-regulated photoproteins were selected because they are the best studied photoproteins [
12,
22,
23] and because they frequently display different properties and activities upon being activated by the same coelenterazine analogue [
24]. Among the semi-synthetic photoproteins examined, only Aq_A6, Aq_B14, Aq_B15, Ol_A1, Ol_A6, Ol_B1, Ol_B3, Ol_B4, and Ol_B14 have preserved sufficient bioluminescence activities (7–80%) as compared to those of the corresponding photoproteins with native coelenterazine (CTZ) (
Table 1). The activation of obelin and aequorin by other CTZ derivatives produced semi-synthetic photoproteins with very poor bioluminescence activities amounting to less than 1% of those with native CTZ.
The highest specific bioluminescence activities among the semi-synthetic photoproteins studied were obtained for those activated by analogues A6 and B14 (
Figure 1); both obelin and aequorin retained 50–80% of bioluminescence activity (
Table 1). Of note is that both CTZ derivatives contain OH-group at the 6-phenyl ring (
Figure 1), which seems to be extremely important for the photoprotein bioluminescence. In contrast, the presence of a triple bond (A6) or the absence of the hydroxy group in the 2-benzyl ring of CTZ (B14) (
Figure 1) is apparently not crucial for the bioluminescence reaction of photoproteins.
The use of CTZ analogue B15 comprising 5-methylfuryl-2 groups as C-6 and C-8 substituents of imidazopyrazinone core (
Figure 1) produced semi-synthetic aequorin and obelin that retain 30.0% and 2.5% of bioluminescence activities of the corresponding photoproteins activated by native CTZ, respectively. However, with CTZ analogues B1 and B3 with 4-fluorophenyl and 4-hydroxymethylphenyl substituents at 6-position of the imidazopyrazinone core, respectively (
Figure 1), the semi-synthetic obelins displayed higher activities (48.0% and 22.7% of that with native CTZ) than the corresponding semi-synthetic aequorins, whose bioluminescence activities were less than 1% compared to that of aequorin activated with native CTZ (
Table 1).
Although the degree of identity between the amino acid sequences of obelin from
O. longissima and aequorin from
A. victoria is only 63.8% [
23], the substrate-binding cavity of these photoproteins is formed by strictly conserved residues [
12,
22]. There are several amino acid residues that differ, but only one of them significantly affects the bioluminescent properties of obelin and aequorin. According to the photoprotein crystal structures, the 6-(p-hydroxy)-phenyl group of CTZ is in a close proximity from Phe88 in obelin and the corresponding Tyr82 in aequorin (residue is numbered according to aequorin spatial structure) [
6,
7].
The role of these residues was revealed by substitution of Phe to Tyr in obelin and Tyr to Phe in aequorin [
25]. The F88Y mutation in obelin resulted in a shift of its bioluminescence maximum to shorter wavelengths (λ
max = 453 nm) and disappearance of the shoulder at 400 nm, thus making its light emission spectrum similar to that of the wild-type aequorin (λ
max = 465 nm). On the contrary, the Y82F aequorin displayed bioluminescence maximum at 500 nm with an additional shoulder at 400 nm, like the emission spectrum of the wild-type obelin. It was proposed that it is a hydrogen bond between OH-groups of Tyr in aequorin and the 6-(p-hydroxy)-phenyl substituent of coelenterazine that determines the differences in light emission spectra of aequorin and obelin [
25]. Later, the determination of spatial structure of F88Y obelin mutant in two conformational states (before and after bioluminescent reaction) confirmed this hypothesis [
26]. This hypothesis is also supported by the computational studies, which showed that the spectral properties of aequorin are significantly affected by the hydrogen bonds between His16, Tyr82, Trp86, and coelenteramide [
27].
For it is quite possible that the hydrogen-bond network around the substituent in position 6 of the imidazopyrazinone core is the reason of differences in bioluminescence activities of aequorin and obelin with certain coelenterazine analogues, the bioluminescence activities of F88Y obelin and Y82F aequorin mutants activated by B1 and B15 CTZ analogues were tested. Indeed, the bioluminescence activity of F88Y obelin mutant with B15 analogue appeared to be higher when compared to wild-type obelin (
Table 2). In contrast, the activation of this obelin mutant by B1 analogue caused the decreased activity. The same was observed for Y82F aequorin—the mutant activated with B1 analogue demonstrated higher bioluminescence, whereas its activity with B15 compound was reduced, i.e., aequorin mutant behaves similarly to wild-type obelin with regard to CTZ analogues (
Table 2). These results clearly show that the hydrogen-bond network and other possible interactions of the residues surrounding the substituent at position 6 of the CTZ imidazopyrazinone core may influence not only the light emission color of photoproteins [
25,
26] but also the efficiency of bioluminescence reaction of the photoprotein in the case of its activation by CTZ analogues with modifications of this substituent.
Coelenterazine-
f (or CTZ-
f) is a well-known CTZ analogue with a 4-fluorobenzyl group as C-2 substituent, commonly used as a substrate because photoproteins activated by this compound display higher sensitivity to Ca
2+ ions [
28,
29] and the bioluminescence activity comparable to that of photoproteins with native CTZ [
29]. The B1 analogue from the present study contains 4-fluorophenyl group as the C-6 substituent, however, the bioluminescence activity of Aq_B1 is poor and Ol_B1 retains only half of the obelin activity with native CTZ (
Table 1). The observed difference between CTZ-
f and B1 analogues suggests the substituent at coelenterazine C-6 position to be of considerable importance for the photoprotein bioluminescence.
Recently, a number of novel coelenterazine analogues with various modifications of C-2, C-6, and C-8 substituents were tested with aequorin and obelin. Among those there were several analogues with 4-fluorobenzyl group as the C-2 substituent combined with another substitution at C-6 or C-8 position of coelenterazine [
24]. Specific bioluminescence activities of the corresponding semi-synthetic obelins and aequorins were substantially lower compared to photoproteins activated with CTZ. There were few exceptions such as analogue with additional fluoro group on the 6-(4-hydroxyphenyl) substituent of coelenterazine displaying the same bioluminescence activity as native coelenterazine with both aequorin and obelin (97–100%) or aequorin activated with naphtyl analogue (30%) [
24]. It is important to note, that in the cited study obelin and aequorin activated by the same CTZ analogue often displayed different bioluminescence activities as well.
As is mentioned above, most of the studied analogues were tested as the substrates of Renilla and Gaussia luciferases. The bioluminescence activity of RLuc with all CTZ analogues from A- and T-series was shown to be significantly lower than that with native CTZ [
15]. The same was observed for most of the modified CTZs from the B-series, with the exception of B2 and B9 analogues with which RLuc exhibited 90% and 25% of native CTZ bioluminescence activity, respectively [
16]. Interestingly, both aequorin and obelin showed poor bioluminescence activity with B2 and B9 analogues as compared to the Aq_WT and Ol_WT (
Table 1). It is evident that while being the coelenterazine-dependent bioluminescence proteins and consequently sharing the same chemical mechanism of bioluminescence, Ca
2+-regulated photoproteins and Renilla luciferase utilize coelenterazine and its analogues differently due to the diverse environment of the active sites and other distinguishing features of these proteins.
In addition, most of the studied coelenterazine compounds had a poor performance with GLuc, especially the analogues from A- and T-series, which displayed insignificant bioluminescence signals [
15,
16]. The best performance with GLuc was demonstrated by B2 analogue, however, it was only 5% of the GLuc bioluminescence with native CTZ. The strikingly different performance of the CTZ analogues with two coelenterazine-dependent luciferases, RLuc and GLuc, once again indicates the great influence that protein microenvironment has on the interaction with luciferin.
2.2. Bioluminescence and Fluorescence Spectra
Most of the semi-synthetic aequorins displayed bioluminescence spectra similar to that of Aq_WT with a maximum around 460–470 nm (
Table 1). There were only two exceptions—Aq_A2 and Aq_B2, which revealed light emission maxima shifted to shorter wavelengths (λ
max at 446 and 452 nm, respectively). It should, however, be mentioned that the bioluminescent activities of these semi-synthetic aequorins were very low (
Table 1). In contrast, when CTZ analogues were used as obelin substrates, bioluminescence spectra of the corresponding semi-synthetic photoproteins were significantly influenced. The emission at 390 nm corresponding to that from the neutral coelenteramide [
1], clearly visible in the bioluminescence spectrum of Ol_WT, disappeared in the spectra of Ol_A2-A5, Ol_B2, Ol_B14, and Ol_B16 (
Table 1,
Figure 2B). At the same time bioluminescence maxima of Ol_A2-A5 and Ol_B2 were shifted toward shorter wavelengths with λ
max around 430–445 nm. The most significant changes in the light emission spectrum were detected for obelin activated with B11 analogue, which has a naphthyl substituent in the C-6 position of imidazopyrazinone core. Ol_B11 emitted light with λ
max at 390 nm and a shoulder at 482 nm (
Figure 2B), i.e., in fact, its bioluminescence spectrum was the “inverse copy” of that of Ol_WT. Light emission spectra of other semi-synthetic obelins were very similar to that of Ol_WT (
Table 1). Of note is that semi-synthetic photoproteins with the highest bioluminescence activities did not show any promising spectral shifts (
Table 1,
Figure S1). According to a recent study performed on semi-synthetic aequorins and obelins, only the analogues containing electron-donating groups (m-OCH3 and m-OH) on the C-6 phenol moiety or an extended resonance system at the C-8 position (1-naphthyl and α-styryl analogues) provided a significant red shift of light emission, up to 44 nm in the case of obelin and α-styryl analogue [
24].
The Ca
2+-regulated photoproteins with peroxy adducts in the active sites are non-fluorescent, but after the bioluminescence reaction ceases they become able to fluoresce at visible wavelengths for as long as the reaction product, coelenteramide, remains bound within a substrate-binding cavity of the photoprotein. Aequorin and obelin variants displayed various types of fluorescence spectra, with maxima ranging from 405 nm to 510 nm, both monomodal and bimodal with different ratios suggesting the presence of different coelenteramide forms (
Table 1,
Figure 2C and
Figure S2).
It was proposed that coelenteramide, the product of the bioluminescence reaction, can exist in different ionic forms [
30]. These are a neutral species with a fluorescence emission maximum around 400 nm, an amide anion with a maximum around 450 nm, a phenolate anion with a maximum around 480–490 nm, and a pyrazine-N(4) anion resonance form with a maximum in the 535–550 nm range (
Figure 2A). In the photoprotein bioluminescence reaction, neutral coelenteramide is believed to be the primary excited product, while the light at longer wavelengths (λ
max = 460–495 nm) originates from the excited phenolate anion arising from the proton dissociation from the hydroxyl group of the 6-(p-hydroxy)-phenyl substituent of coelenteramide in the direction to His22, which is located nearby at a hydrogen-bond distance [
12,
26]. The hybrid quantum mechanics and molecular mechanics methods combined with the molecular dynamics method confirmed experimental conclusions that the neutral form of coelenteramide is the primary excited state product and that the excited phenolate anion coelenteramide is the main aequorin emitter [
31,
32].
It is well known that the p
K* of a phenolic group is several units below its p
K in the ground state [
33]. If this p
K* falls well below 6.5, which is the expected p
K of His, rapid transient proton dissociation and its “transient displacement” toward the N atom of His will occur, with simultaneous generation of the excited phenolate anion. Since its fluorescence lifetime is 5–6 ns [
34], there is more than enough time for the proton to dissociate before radiation. However, in addition to the His residue, the OH-group of the phenol attached at C-6 of coelenterazine is also surrounded by Trp and Tyr in aequorin or Trp and Phe in obelin within hydrogen-bonding distance which also affect the spectral properties of photoproteins. It is believed that a reason for the lack of 400-nm shoulder in the aequorin bioluminescence spectrum is a more effective proton dissociation due to the additional H-bond between the OH-group of Tyr82 (Phe in this position in obelin) and
p-OH of the phenol attached at C-6 of the 2-hydroperoxy adduct of coelenterazine.
Thus, the variety of fluorescence spectra of the Ca
2+-discharged semi-synthetic photoproteins could be a result of improper orientation of the corresponding peroxy adducts in the active site that might disturb the environment of the substituent at C-6 position or increase the distance between this substituent and His22, thus tampering the proton dissociation. It is also interesting to note that most of the studied coelenterazine analogues, especially those of A- and T-series, tend to produce a neutral coelenteramide as a fluorescence emitter with a maximum around 400–420 nm despite the fact that in most cases the main bioluminescence emitter was detected to be a phenolate anion with a maximum around 470–480 nm (
Table 1).
2.3. Absorbance Spectra of Coelenterazine Analogues in Ethanol, and Active and Ca2+-Discharged Semi-Synthetic Photoproteins
To address the matter of whether semi-synthetic photoproteins with low bioluminescence activity can even form an active photoprotein complex we determined the absorbance spectra of CTZ analogues in ethanol, as well as active and Ca2+-discharged photoproteins.
Native coelenterazine in ethanol (or methanol) has the absorption spectrum with a maximum at 434 nm with a small shoulder at 350 nm (
Table 3,
Figure S3). Photoproteins activated by CTZ display red-shifted broad absorption maximum with λ
max around 460 nm conditioned by 2-hydroperoxy adduct of CTZ bound within the active site. After bioluminescence reaction ceases, the absorption at 460 nm disappears and the maximum at 335 nm corresponding to the bound reaction product, coelenteramide, becomes apparent (
Figure S3).
Similar to native CTZ, all the analogues in ethanol have absorbance spectra with maxima/shoulders in the visible and near-UV regions (
Table 3). However, for more than half of them the absorption at near-UV region was higher than that in the visible region. Since the higher absorption in the near-UV region was probably conditioned by the product arising as a result of spontaneous oxidative decarboxylation of CTZ analogues, we can reasonably assume that many of these analogues were unstable in solution in the presence of oxygen and consequently this could be one of the reasons accounting for a very low activity of the corresponding semi-synthetic photoproteins. The other reasons could be an improper spatial orientation of the peroxy adduct in the coelenterazine-binding site or the disruption of the interactions between the key amino acid residues and analogues of coelenterazine due to the chemical modifications of the latter.
The absorbance spectra of active semi-synthetic photoproteins are diverse and depend on CTZ analogue used for activation (
Table 3). For example, both Ol_A6 and Aq_A6 display red-shifted absorbance maxima at 460 nm, which match exactly those of Ol_WT and Aq_WT. A similar red shift of the absorbance maxima was observed for Ol_A1 and Ol_B1. However, the absorbance spectrum of Ol_B3 was shifted to a shorter wavelength (λ
max = 420 nm) as compared to that of B3 analogue in ethanol (
Table 3). Both Ol_B14 and Aq_B14 revealed the same absorbance spectra as the B14 analogue has in ethanol with one distinction—whereas semi-synthetic photoproteins had a higher absorption in the near-UV region with a shoulder at 434 nm, analogues in ethanol displayed the absorbance spectrum with λ
max at 434 nm and a shoulder at 344 nm (
Table 3). Also noteworthy is that a similar case is observed for ctenophore photoproteins, for which absorption spectrum corresponds to that of native CTZ in methanol [
35]. The most unusual absorption spectrum appeared to be shown by aequorin with analogue B15 (
Figure 1) comprising the two methylfuryl groups with the absorption maximum at 400 nm (
Table 3). Of note is that the absorbance spectra of the above-mentioned semi-synthetic photoproteins were undoubtedly conditioned by the bound 2-hydroperoxy adduct of the corresponding CTZ analogue since all of them retained 10–80% of bioluminescence activity compared to photoproteins activated by native CTZ (
Table 1).
The absorbance spectra of semi-synthetic photoproteins with low bioluminescence activities (
Table 1) can be divided into two groups. Both aequorin and obelin with A2-A5, B4, B6, B8-B10, B16, or T1-T3 analogues displayed spectra matching those of the Ca
2+-discharged photoproteins, i.e., bound with the reaction product (
Table 3). It implies that many of these analogues could be unstable in solution or fail to form a stable peroxy adduct within the substrate-binding cavity of the photoprotein. Another group includes the analogues that produce semi-synthetic photoproteins with different absorption spectra for the active and discharged forms in the case of aequorin or obelin. For example, after the bioluminescence reaction the absorption spectrum of Aq_B7 significantly changed, while Ol_B7 displayed the absorption spectrum matching that of its Ca
2+-discharged variant (
Table 3). The same pattern was observed for B11 and B13 analogues. It is possible that in the absorption spectra of these semi-synthetic photoproteins the band specific for a bound 2-hydroperoxy adduct of coelenterazine analogue is located at different wavelengths as compared to the photoproteins with native coelenterazine. The differences in the absorption spectra of obelin and aequorin with these analogues may be due to the hydrogen-bond network and the residues surrounding the substituent at position 6 of the CTZ imidazopyrazinone core of a substrate in these photoproteins, i.e., similar to what was demonstrated for B1 and B15 analogues (
Table 2).
Thus, the low activity of certain semi-synthetic obelins and aequorins can be brought about by several key reasons, among which are instability of some analogues in solution and environment of substituent at C-6 position of the CTZ imidazopyrazinone core influencing the efficiency of 2-hydroperoxy adduct formation.
2.4. Sensitivity of Semi-Synthetic Obelins and Aequorins to Calcium
The main application of Ca
2+-regulated photoproteins is based on their ability to emit light on Ca
2+ binding. Five decades ago, aequorin was used as an intracellular Ca
2+ indicator for the first time and since then Ca
2+-regulated photoproteins have been widely applied to detect calcium ions in biological systems [
36]. Successful cloning of the corresponding cDNAs allowed intracellular expression of the recombinant apophotoproteins. After external addition, coelenterazine diffuses into the cells and forms an active photoprotein. Such cells with a “built-in” calcium indicator can be used to measure intracellular Ca
2+ concentration with high efficiency, and this approach has many advantages over fluorescent Ca
2+ probes [
37].
It is well known that the use of CTZ analogues instead of native CTZ might change sensitivity of photoproteins to Ca
2+ [
24,
38]. Hence, we determined calcium sensitivity of Aq_A6, Aq_B14, Ol_A6, Ol_B1, and Ol_B14 (
Figure 3) since only these semi-synthetic photoproteins have shown bioluminescence activities high enough for reliable measurements of light signals over the explored [Ca
2+] range (
Table 1). The vertical ranges of the Ca
2+ concentration-effect curves for semi-synthetic obelins and aequorins were very similar to those of photoproteins activated by native CTZ (the vertical range of the curves spans approximately 7 log units) with small distinctions—the vertical ranges of Ol_B14 and Aq_B14 as well as of Aq_A6 were slightly reduced for 0.3 and 0.45 log units, respectively, on account of the increase of Ca
2+-independent bioluminescence (EGTA condition, left part of the curves). Of note is that all the Ca
2+ concentration-effect curves had the same maximum slope of approximately 2.5, meaning that three calcium ions are needed to bind with photoprotein to initiate its light emission [
13,
39].
The Ca
2+ concentration-effect curves for Ol_WT, Ol_A6, and Ol_B1 were almost superimposed (
Figure 3). Only for Ol_B14, the curve showed a clear difference, since it was shifted to the region of lower Ca
2+ concentration. The same effect was observed for both semi-synthetic aequorins, however, the curve for Aq_A6 was shifted to the region of lower Ca
2+ concentration more than the corresponding curve for Ol_A6 (
Figure 3). Thus, among the semi-synthetic photoproteins tested only Aq_B14 and Ol_B14 clearly displayed higher Ca
2+ sensitivity compared to that of the corresponding photoproteins activated by native CTZ. It is worth noting that while bioluminescence activities of semi-synthetic obelins with CTZ-
f and B1 analogues were different, the effect of these substrate modifications on calcium sensitivity was quite similar [
29].
In cells, the concentration of free Mg
2+ exceeds that of Ca
2+ by several orders of magnitude [
40] and, being a competing ion, Mg
2+ may decrease Ca
2+ affinity of the Ca
2+-binding sites of many EF-hand Ca
2+-binding proteins. Therefore, the effect of physiological concentration of Mg
2+ ions on the Ca
2+ affinity of semi-synthetic photoproteins was studied (
Table 4). Mg
2+ ions noticeably affected only aequorin and its variants; the presence of magnesium ions shifts the Ca
2+ concentration-effect curves to the right relative to those without Mg
2+. Moreover, Mg
2+ ions also decreased Ca
2+-independent luminescence of aequorins (
Figure A1) and increased the [Ca
2+]
limit and
Kd approximately twofold, even more in the case of Aq_B14 (
Table 4). The effect of 1 mM Mg
2+ on the Ca
2+ concentration-effect relations of semi-synthetic obelins was expectedly less pronounced [
24,
41].
2.5. Rapid-Mixing Kinetics
The speed with which the luminescence responds to rapid changes in Ca
2+ concentration is an important characteristic of Ca
2+-regulated photoproteins. It defines the ability of photoproteins to properly track the intracellular calcium transients sometimes occurring within the millisecond timescale [
42,
43]. Only a few semi-synthetic photoproteins possessing bioluminescence activities high enough for stopped-flow measurements were tested (
Table 5). When A6, B1, or B14 analogue was used as a substrate of obelin, the rate constants characterizing rise and decay of the light signal were not significantly influenced. In contrast, bioluminescence kinetics of Aq_A6 strikingly differed from that of Aq_WT (
Table 5). The rise constant of this semi-synthetic aequorin turned out to be 2.5 times faster than that of the aequorin with native CTZ. It is interesting to note that a similar 2.3-fold increase of rise constant was found earlier for the aequorin activated by CTZ-
i analogue [
29]. In addition, the decay kinetics of this aequorin variant was satisfactorily described by a two-exponential decay function and consequently by two rate constants—“fast” (
k1) and “slow” (
k2), i.e., similar to all studied obelins (
Table 5). According to the recently proposed unanimous kinetic model for photoprotein bioluminescence, which incorporates the “positive cooperativity” between the Ca
2+-binding sites II and III [
44], the fast decay component of bioluminescence was attributed to the intermediate that arises after calcium binding with the unpaired N-terminal Ca
2+-binding site I. Thus, appearance of the “fast” component in the decay kinetics of light signal of Aq_A6 might be attributed to significant structural changes in the N-terminal domain, owing to the coelenterazine analogue used as a substrate.