Introduction

The generation of excited triplet carbonyls and of singlet molecular oxygen, O2 (1Δg), has long been reported to occur in various biological processes, based on the observation of low-level (also called ultraweak) chemiluminescence (CL)1,2,3,4,5,6,7,8,9,10,11.

Triplet-excited carbonyl species can be generated by photoexcitation of carbonyl compounds. Importantly, electronically excited carbonyls can also be generated by chemiexcitation and undergo further typical photochemical processes, i.e. without photoexcitation, which consequently was independently called by G. Cilento (University of São Paulo)10 and by E. H. White (Johns Hopkins University) as “photochemistry in the dark”11. Some examples of such “dark” reactions are the dismutation of alkoxyl radicals12, thermal decomposition of 1,2-dioxetanes13,14, thermolysis of oxetanes (reverse [2+2] Paternò-Büchi reaction)15 and dismutation of alkyl peroxyl radicals, known as the Russell reaction16,17. The quantum yield of excited triplet carbonyl generation may vary from 0.1% up to 60% in these reactions18. Of potential biological interest are triplet carbonyls arising from the annihilation of oxyradical intermediates during lipid peroxidation6,18,19,20,21.

Enzyme-catalyzed peroxidation can also yield excited triplet carbonyls, as in the case of aerobic oxidation of 2-methylpropanal (isobutyraldehyde or isobutanal, IBAL) catalyzed by horseradish peroxidase (HRP), which gives rise to formic acid and triplet acetone22. This reaction is thought to occur by HRP-catalyzed addition of molecular oxygen to the α-carbon of IBAL, yielding a 1,2-dioxetane intermediate whose homolysis renders acetone in the triplet state22,23,24. Accordingly, the chemiluminescence spectrum matches the phosphorescence spectrum of triplet acetone (λmax ~ 430 nm). In addition, iso-propanol and pinacol (2,3-dihydroxypropane) ultimately formed by hydrogen abstraction from the carbohydrate portion of HRP by triplet acetone were found in the spent reaction mixtures, thus a process that can be here classified as a source of “photo” chemical products, although formed in the dark.

The fact that the excitation energy of acetone to its triplet state is about 335 kJ.mol−112 whereas that of O2 (1Δg) is 94.2 kJ.mol−125,26 makes the triplet-triplet energy transfer process thermodynamically viable. Briviba et al.27 detected monomol light emission of O2 (1Δg) at 1,270 nm in CCl4 during the thermal decomposition of 3-hydroxymethyl-3,4,4-trimethyl-1,2-dioxetane.

Singlet molecular oxygen exhibits a pair of electrons whose opposite spins in the highest occupied molecular orbital gives O2 (1Δg) dienophilic properties, which explains its significant reactivity toward electron-rich organic molecules, particularly with those exhibiting conjugated double bonds28, leading to the formation of allylic hydroperoxides, dioxetanes or endoperoxides2,5,17,29,30. Singlet molecular oxygen has been shown to be generated in biological systems. As possible biological sources of O2 (1Δg), one can cite (i) enzymatic processes catalyzed by peroxidases or oxygenases; (ii) several reactions that take place in cells, such as annihilation of lipid peroxyl radicals (Russell reaction)16,17,30,31; (iii) ozone oxidation of amino acids, peptides and proteins32; (iv) reactions of hydrogen peroxide with hypochlorite or peroxynitrite33,34,35; (v) thermolysis of endoperoxides36,37,38,39,40,41,42,43,44,45; (vi) in vitro photodynamic processes involving type II photosensitization reactions by suitable dyes46,47,48,49; (vii) UV irradiation of aromatic amino acids in proteins and immunoglobulins5,50; and (viii) metal-induced decomposition of a thymine hydroperoxide51. Production of O2 (1Δg) during phagocytosis in polymorphonuclear leukocytes has also been described52,53,54 and observed in photodynamic therapy, where the production of this reactive oxygen species (ROS) has been demonstrated using different photosensitizers, including methylene blue, eosin and rose bengal46 or dye-containing nanoparticles47,48. Some endogenous photosensitizers may also lead to the generation of O2 (1Δg) upon exposure to UVA radiation5,55. Photodynamic therapy has been applied successfully in both antimicrobial and antitumor treatments46,47,48,49, including inactivation of viruses in human plasma56.

Thus, there is a potential mechanistic crosstalk between O2 (1Δg) and triplet carbonyl in biological environments where both excited species can be produced, either by alkoxyl and alkylperoxyl radical dismutation or by triplet-triplet energy transfer from excited carbonyls. Hence, several hypotheses, such as the production of triplet carbonyls from O2 (1Δg)-driven peroxidation of polyunsaturated fatty acids, have been proposed and demonstrated experimentally, although triplet carbonyl products have been detected in only a few systems21.

This investigation addresses the question whether electronically excited O2 (1Δg) can unequivocally be produced by energy transfer from excited triplet acetone to triplet molecular oxygen O2 (3Σg−) dissolved in aqueous solution. We used the thermolysis of 3,3,4,4-tetramethyl-1,2-dioxetane (TMD)57,58 and the HRP/IBAL/O2 system22 as chemical and enzymatic sources of triplet acetone, respectively12. The generation of O2 (1Δg) was monitored by direct spectroscopic detection and characterization of O2 (1Δg) monomol light emission in the near-infrared region at 1,270 nm. Singlet molecular oxygen was also detected indirectly by electron paramagnetic resonance spectroscopy (EPR) of 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) formed by the reaction of the spin trap 2,2,6,6-tetramethylpiperidine (TEMP) with O2 (1Δg). Further, the reaction mechanism was investigated by tracing the energy transfer from triplet excited ketone species to [18O]-labeled triplet molecular oxygen [18O2 (3Σg−)] through the detection of [18O]-labeled O2 (1Δg) [18O2 (1Δg)]. Chemical trapping experiments of 16O2 (1Δg) and 18O2 (1Δg) were performed using the anthracene-9,10-diyldiethane-2,1-diyl disulfate disodium salt (EAS) trap by monitoring the corresponding endoperoxide (EASxOxO, x = 16 or 18) with high-performance liquid chromatography coupled to electrospray ionization tandem mass spectrometry (HPLC-ESI-MS/MS).

Results

Characterization of singlet molecular oxygen generated by energy transfer from triplet acetone to triplet molecular oxygen by CL measurements

Chemiluminescence produced by a chemical reaction provides useful information about the excited species being generated. Here, the production of O2 (1Δg) in response to the collision of excited triplet acetone with ground state molecular oxygen was investigated by monitoring the near infrared (NIR) light emission at 1,270 nm, which corresponds to the singlet delta state monomolecular light emission decay of oxygen (1Δg → 3Σg−) (Equation 1)2,59,60. The measurement of ultra-weak light emission or low level CL originating from this radioactive transition is an important method for the detection and characterization of O2 (1Δg).

The CL arising from the thermal decomposition of 10 mM TMD at 70°C in air-equilibrated CCl4 or acetonitrile (Fig. 1A(b) and 1A(a), respectively) was recorded in the UV-visible region. The CL spectrum of 10 mM TMD in CCl4 shows a peak at 430 nm (Fig. 1B), which was assigned to the triplet excited acetone14. Fig. 1C and 1D depict the time course of monomol light emission of O2 (1Δg) at λ = 1,270 nm and the NIR spectrum of O2 (1Δg), respectively. Since the lifetime of O2 (1Δg) in acetonitrile is much lower than in CCl4 (5.0–8.0 × 10−5 s and 0.02–0.08 s, respectively), the TMD/O2 (3Σg−) NIR light emission in acetonitrile was very low under similar experimental conditions61,62. For comparison, the time course and spectrum of NIR light emission were recorded during the thermolysis of 1,4-dimethylnaphthalene-1,4-endoperoxide (DMNO2)60 in methanol (Fig. 1E and 1F).

Figure 1
figure 1

Chemiluminescence studies of TMD in organic solvents.

TMD (10 mM) was incubated in air-equilibrated organic solvents at 70°C. (A) Time course of total UV-visible light emission of TMD in acetonitrile (line a) and in CCl4 (line b); (B) The chemiluminescence spectrum matches the phosphorescence spectrum of TMD-generated triplet excited acetone20 in CCl4; (C) NIR light emission of O2 (1Δg) at 1,270 nm during the thermolysis of TMD in CCl4 (line b) and in acetonitrile (line a); (D) O2 (1Δg) spectrum, corresponding to the monomol light emission recorded during incubation of TMD in CCl4; and (E and F) thermodissociation of 10 mM DMNO2 in methanol, as a control, which generated O2 (1Δg) monomol light emission at 1,270 nm and the NIR spectrum of released O2 (1Δg), respectively.

Full size image

The rate of triplet ketone produced by TMD concentrations ranging from 2 to 10 mM in CCl4 was estimated to be 4.89 ± 0.98 nM min−1. The molecular oxygen concentration available in the solvent induces a saturation effect of O2 (1Δg) steady-state concentration. Briviba et al.27 estimated the yield of O2 (1Δg) produced by an analogue of TMD, 3,3,4,-tetramethyl-4-hydroxy-1,2-dioxetane to be 0.2%.

Since the concentration of O2 in solution can limit the generation of O2 (1Δg) by TMD thermolysis, additional luminescence experiments were performed using CCl4. Ten minutes after starting the reaction, pure O2 was purged inside the cuvette in an attempt to enhance O2 (1Δg) generation (Fig. 2). As expected, the influx of molecular O2 into the system decreased the intensity of UV-visible light (Fig. 2A) due to energy transfer of the generated triplet acetone to molecular oxygen, although a slight decrease in NIR monomol light emission of O2 (1Δg) was observed (Fig. 2B). In this respect, we note that, although triplet molecular oxygen is known to be a triplet carbonyl suppressor18, McGarvey et al.61 reported an inverse correlation between molecular oxygen quenching of different triplet naphthalenes in benzene and the generation of O2 (1Δg). This finding was then correlated to structural differences in naphthalene and not to changes in O2 concentration.

Figure 2
figure 2

Effect of pure O2 purging on the chemiluminescence intensity elicited by TMD.

(A) UV-visible light emission time course of triplet excited acetone during the thermolysis of 5 mM TMD in CCl4 at 70°C and (B) Monomol light emission of O2 (1Δg) recorded during the decomposition of 5 mM TMD in CCl4 at 70°C. The arrow in both graphs indicates the time elapsed in O2 purging.

Full size image

Since the sorbate anion was proposed as a probe for testing the presence or intermediacy and roles of triplet species in biological systems18, the quenching effect of sorbate on TMD-generated triplet acetone luminescence was also examined (Supplementary Fig. 1).

Although 0.5 mM sorbate was able to quench ~25% of the triplet acetone chemiluminescence in CCl4 (Supplementary Fig. 1A), the NIR light emission generated by O2 (1Δg) did not change significantly (Supplementary Fig. 1B).

Triplet acetone is also produced by O2-mediated oxidation of IBAL by molecular oxygen, catalyzed by HRP (Fig. 3). The total chemiluminescence was recorded in D2O at pD 7.4 in the presence of 5 µM HRP and 10 mM IBAL (Fig. 3A). Low-level O2 (1Δg) NIR light emission was also detected at 1,270 nm under similar experimental conditions (Fig. 3B).

Figure 3
figure 3

Chemiluminescence studies of O2-mediated oxidation of IBAL catalyzed by HRP.

(A) Total UV-visible light emission of triplet excited acetone in deuterated phosphate buffer (pD 7.4) and (B) NIR light emission of O2 (1Δg) at 1,270 nm after injection of 10 mM IBAL in a solution of 5 µM HRP in D2O, pD 7.4 at 37°C.

Full size image

Singlet Molecular Oxygen Spectrum in the Near-Infrared Region

The generation of O2 (1Δg) by the thermal cleavage of TMD was also confirmed by recording the spectrum of the light emitted in the near-infrared (NIR) region (Fig. 1D). For comparison, the spectrum of O2 (1Δg) generated by thermolysis of DMNO260 was also recorded (Fig. 1F). Both spectra showed an emission band with maximum intensity at 1,270 nm, characteristic of the monomolecular decay of singlet oxygen delta state. Additional proof that the light emitted in the TMD reaction corresponds to O2 (1Δg) was obtained by testing the effect of solvents. The intensity of light emitted in the reaction performed in CCl4 was higher than in acetonitrile, which is consistent with the longer lifetime of O2 (1Δg) in CCl462. The quenching effect of lycopene63 on the NIR chemiluminescent reaction of TMD thermolysis was also observed (data not shown).

Detection of O2 (1Δg) by EPR

Indirect analysis of O2 (1Δg) in D2O by electron paramagnetic resonance was performed using 2,2,6,6-tetramethylpiperidine (TEMP) as the spin trap (Supplementary Fig. 2). The lifetime of O2 (1Δg) in D2O is similar to that observed in acetonitrile (5.0–6.5 × 10−5 s)62. The EPR spectrum depicted in Supplementary Fig. 2A (line a) shows a triplet signal (aN = 1.60 mT, g-shift = −0.5) obtained upon incubation of 30 mM TEMP with 4 mM TMD in normally aerated D2O. The pre-addition of 0.4 µM commercial standard 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) to the reaction mixture intensified the EPR signal significantly, thus suggesting the generation of O2 (1Δg)64 (line b).

EPR experiments using TEMP were also conducted with the HRP/IBAL system, as depicted in Supplementary Fig. 2B. The EPR spin-trapping signal obtained also overlaps the TEMPO signal, showing the same coupling constants. This finding provides further evidence of the generation of O2 (1Δg) by the HRP-treated aldehyde65.

When the reaction of TMD was conducted in the presence of 30 mM TEMP and 32 mM sorbate, no significant decrease in TEMPO was observed (data not shown). Although sorbate can reportedly suppress triplet acetone generated from TMD18, diene quenching was unable to compete actively with the excitation of oxygen in the presence of TEMP.

When 20 µM HRP and 50 mM IBAL were incubated with 8 mM sorbate, the EPR signal of TEMPO was suppressed (Supplementary Fig. 2B, line e).

Detection of [18O]-Labeled Singlet Molecular Oxygen in the Chemical and Enzymatic Reactions

To better characterize the mechanism involved in the generation of O2 (1Δg) by the thermolysis of TMD or HRP-catalyzed aerobic oxidation of IBAL, [18O]-labeled O2 (3Σg−) was used as a triplet energy acceptor. The generated [18O]-labeled O2 (1Δg) was trapped with the anthracene derivative, EAS (Fig. 4)5,30. The corresponding endoperoxides (EASxOxO, x = 16 or 18) were detected by HPLC-ESI-MS/MS.

Figure 4
figure 4

Chemical trapping of [18O]-labeled O2 (1Δg) [18O2 (1Δg)] with disodium salt of anthracene-9,10-diyldiethane-2,1-diyl disulfate (EAS) yielding the corresponding double-[18O]-labeled 9,10-endoperoxide (EAS18O2).

Full size image

In the dark, energy transfer from triplet ketone to 16O2 or [18O]-labeled molecular oxygen led to a mixture of mainly two anthracene endoperoxide derivatives, namely, the fully labeled 9,10-endoperoxide (EAS 18O18O) and the related unlabeled endoperoxide (EAS 16O16O), plus a small amount of partially labeled endoperoxide (EAS18O16O).

Figures 5 and 6 and Supplementary Fig. 3 to 7 show the typical chromatograms for EASxOxO analysis with UV and MS/MS detection. Analysis of the products by UV absorption at 210 nm showed two peaks corresponding to the endoperoxides EAS18O18O and EAS16O16O and to EAS at the time windows 7.2 to 7.9 min and 9.2 to 12.2 min, respectively, for the TMD (Fig. 5A and Supplementary Fig. 3A) and HRP/IBAL systems (Fig. 6A and Supplementary Fig. 4A and 5A). The tandem mass spectrometry detection of EAS18O18O (m/z 230→212) and EAS16O16O (m/z 228→212) was performed by the Selected Reaction Monitoring (SRM) mode. SRM detection based on the fragmentation of precursor ions at m/z 230 (Fig. 5B and 6B) and 228 (Fig. 5C and 6C), which generated the product ion at m/z 212, shows the presence of EAS18O18O and EAS16O16O, respectively. The identity of the precursor ions was confirmed based on an analysis of the mass spectra of product ions derived from each of the endoperoxides (Fig. 5E and 5F,and Fig. 6E and 6F).

Figure 5
figure 5

EAS chemical quenching studies of O2 (1Δg) produced during thermal cleavage of TMD in the presence of [18O]-labeled molecular oxygen.

HPLC-ESI-MS/MS analysis of 8 mM EAS incubated with 8 mM TMD for 2 h at 70°C in deuterated phosphate buffer (pD 7.4). (A) UV chromatogram at 210 nm. EASO2 endoperoxides containing 16O or 18O eluted at 7.8 min. (B) SRM chromatogram of EAS18O18O (m/z 230→212) with a determined area integration of 118,754 (A.U.). (C) SRM chromatogram of EAS16O16O (m/z 228→212) with area integration of 11,882 (A.U.). (D) Full mass spectrum obtained from peak at 7.8 min within the mass range of 200-255 m/z. (E) Product ion spectrum from precursor ion at m/z 230 and (F) Product ion spectrum from precursor ion at m/z 228.

Full size image
Figure 6
figure 6

EAS chemical trapping of O2 (1Δg) generated by the HRP-catalyzed oxidation of IBAL in the presence of [18O]-labeled molecular oxygen.

HPLC-ESI-MS/MS analysis of 8 mM EAS upon incubation for 24 h with 5 µM HRP and 50 mM IBAL at 37°C in deuterated phosphate buffer (pD 7.4). (A) UV chromatogram at 210 nm. Endoperoxides EAS16O16O and EAS18O18O eluted at 7.8 min. (B) SRM chromatogram of EAS18O18O (m/z 230→212). (C) SRM chromatogram of EAS16O16O (m/z 228→212). (D) Full mass spectrum obtained from peak at 7.8 min within mass range of 100–280 m/z. (E) Product ion spectrum from precursor ion at m/z 230. (F) Product ion spectrum from precursor ion at m/z 228.

Full size image

Energy transfer from excited triplet acetone generated by thermal cleavage of TMD

The thermolysis of 10 mM TMD in deuterated phosphate buffer (pD 7.4) performed in an 16O2 or 18O2 atmosphere resulted in the generation of the corresponding EAS 9,10-endoperoxides containing the 18O or 16O isotope (EASxOxO) (Fig. 5 and Supplementary Fig. 3).

Formation of endoperoxides, which was confirmed by HPLC-ESI-MS/MS analysis, occurred through the mass transition of m/z 230 →212 to EAS18O18O and m/z 228 →212 to EAS16O16O (Fig. 5B and 5C and Supplementary Fig. 3B and 3C). In the presence of [18O]-labeled O2, the amount of EAS18O18O (Fig. 5B) was ten-fold greater than that of EAS16O16O (Fig. 5C). The EAS18O2 endoperoxide formed in the presence of the triplet acetone chemical generator system shows an intense [M-2H]2− ion at m/z 230 corresponding to a molecular weight of 462 (Fig. 5D). This strongly attests to the incorporation of two [18O]-labeled oxygen atoms into the anthracene derivative molecule. This finding also confirms that O2 (1Δg) is produced by energy transfer from TMD-generated triplet acetone and not through direct oxygen atom transfer from the 1,2-dioxetane, which lacks 18O in its molecular structure. Important to note is the fact that the amount of EAS18O2 formed in the experiment reached a level of 90%38.

Energy transfer from enzymatically generated excited triplet acetone

The generation of O2 (1Δg) by energy transfer from HRP-catalyzed production of excited triplet acetone from IBAL oxidation was monitored using water-soluble EAS, which can react with O2 (1Δg), yielding EASO2 as the specific oxidation product (Fig. 4). To this end, EAS was incubated at 37°C with HRP and IBAL in an 16O2 or 18O2 atmosphere. The resulting 9,10-endoperoxides EAS16O16O and EAS18O18O were analyzed by HPLC-ESI-MS/MS (Fig. 6 and Supplementary Fig. 4 and 5). As expected in 16O2 atmosphere, the endoperoxide EAS16O2 (MW 458) produced in the ezymatic reaction exhibits a [M-2H]2− ion at m/z 228 (Supplementary Fig. 5D). Only the SRM chromatogram of EAS16O2 with the mass transition m/z 228 to 212 can be detected at 7.9 min (Supplementary Fig. 5C). The SRM chromatogram showed no peaks for the EAS18O2 mass transition (m/z 230 to 212) (Supplementary Fig. 5B).

Conversely, when the HRP-catalyzed reaction was conducted under [18O]-labeled dioxygen (18O2) enriched atmosphere, the fully labeled endoperoxide (EAS18O18O) appeared as the most abundant ion at m/z 230 (Fig. 6D). The ion corresponding to unlabeled endoperoxide (EAS16O16O) at m/z 228 was also detected with a relative abundance of about 50% compared to the fully labeled endoperoxide. Trace amounts of partially labeled endoperoxide (EAS18O16O) was also observed (Supplementary Fig. 4C). The detection of unlabeled and partially labeled endoperoxides can be attributed to residual oxygen (16O2) present in the reaction media after the freeze-thawing cycles to replace the dissolved 16O2 with 18O2.

Subsequently, the experiments were conducted in the presence of [18O]-labeled O2 (3Σg−). The EASxO2 endoperoxides formed in the presence of the triplet ketone enzymatic generator systems show two intense [M-2H]2− ions at m/z 228 and 230 (Fig. 6D), corresponding to the molecular weights of 458 and 462 for the endoperoxides EAS16O16O and EAS18O18O, respectively. This is indicative of the incorporation of two 16- or 18- oxygen atoms into the anthracene endoperoxide molecules. The signal of the ion corresponding to the unlabeled anthracene endoperoxide at m/z 228 was detected with a relative abundance of 50% compared to that of the [18O]-labeled oxygen anthracene endoperoxide molecule at m/z 230. The generation of traces of EAS18O16O (Supplementary Fig. 4C) and minor amounts of EAS16O2 was also observed in the EASO2 MS spectrum, which can be attributed to residual 16O2 contaminant after the freeze-thawing cycles to replace the dissolved 16O2 with 18O2 and to subdue the incidence of natural light during sample handling. Because the initial step of HRP-catalyzed IBAL oxidation involves the generation of an IBAL resonant α-hydroperoxyl/enolyl radical, which ultimately yields the 3-hydroxy-4,4-dimethyldioxetane intermediate – the putative precursor of triplet acetone and formic acid65 by thermolysis, O2 (1Δg) may have arisen from the radical, according to the Russell mechanism16. This route can be safely disregarded because TMD alone would not have been able to yield a consistent amount of EAS18O16O (Fig. 5) and the radical does not bear a geminal hydrogen, a necessary condition for singlet molecular oxygen generation by the Russell reaction16. Sulfur stable isotope distribution in EAS16O2 and EAS18O2 molecules was also observed by Ultra High Resolution MS, providing further confirmation of the EAS endoperoxide structures (Supplementary Fig. 6 and 7).

When the IBAL/HRP system was investigated under aerated condition (Supplementary Fig. 7), an analysis of the peak corresponding to m/z transition 228 to 212 indicated that 2.0 ± 0.4 µM O2 (1Δg) is formed. A previous report stated that the HRP-catalyzed oxidation of IBAL generates at least 20% O2 (1Δg)65. Much less optimistic, the yield of O2 (1Δg) measured in our enzymatic experiments points to approximately 0.1%. Nevertheless, one must consider that the HRP-catalyzed reaction consumes the dissolved oxygen65, thus gradually suppressing the generation of both triplet carbonyls and O2 (1Δg).

Moreover, when the HRP/IBAL enzymatic reaction was conducted in the presence of 5 mM sorbate ion, a decrease in EASO2 was observed (Supplementary Fig. 8). The EASO2 EAS transition peak is 3-fold lower than the control peak (Supplementary Fig. 8B). Compared to the TMD chemiluminescence experiment in the presence of 0.5 mM sorbate (Supplementary Fig. 1A), the quenching efficiency of 5 mM sorbate is lower in the generation of O2 (1Δg) that accompanies the HRP-catalyzed oxidation of IBAL. This is predicted by the fact that the enzymatic system probably produces fewer triplet carbonyls than TMD and that the enzyme structure offers a collisional barrier for triplet acetone quenching produced in the active site22. Excited triplet acetone was estimated to be produced at a rate of 0.19 µM.min−1 by 10 mM IBAL in the presence of 5 µM HRP (Fig. 3). A noteworthy fact is that the decay of light emission parallels the oxygen consumption by the enzymatic reaction66.

Discussion

It is well established that electronically excited triplet carbonyl products are produced chemically or enzymatically via the thermolysis of dioxetane intermediates1,3,4,8.

Carbonyls in the triplet excited state are known to undergo unimolecular reactions (e.g., isomerization, α- and β-cleavage) and bimolecular processes (e.g., hydrogen abstraction, (2+2) cycloadditions), or to act as an electronic energy donor to a wide spectrum of biomolecules, thus triggering typically photochemical reactions. This inspired Cilento10,67 and White11, in the mid-1970s, to postulate independently that chemically or enzymatically generated triplet species in cells may drive physiological and/or pathological processes in the dark, a phenomenon they coined as “photochemistry and photobiology without light,” or “photochemistry in the dark.” The isomerization of natural products (e.g., colchicine, santonin), initiation of polyunsaturated fatty acid peroxidation, generation of the plant hormone ethylene, formation of cyclobutane thymine dimers and several other biological processes, have been predicted and some of them have been shown to occur in the dark via triplet carbonyl intermediates18.

Our results show for the first time that singlet molecular oxygen is produced enzymatically. This paper described the generation of O2 (1Δg) via energy transfer from excited triplet acetone from both the thermolysis of TMD and the aerobic oxidation of HRP-catalyzed IBAL.

The chemiluminescent catalytic activity of hemeproteins such as cytochrome c acting on the peroxidation of fatty acids24,29 and soybean lipoxygenase68 or myeloperoxidase69 inducing the oxidation of IBAL, were also accounted for by enzymatic sources of triplet excited species. The generation of methylglyoxal and diacetyl, putatively in the triplet state, by the oxidation of myoglobin-catalyzed aerobic oxidation of acetoacetate and 2-methylacetoacetate, respectively, was reported more recently70.

From the biological viewpoint, it is worth mentioning that the generation of electronically excited triplet carbonyls in biological systems has been shown to cause oxidative injury to biologically important molecules such as DNA71 and proteins, to trigger lipid peroxidation72 and to induce phosphate-mediated permeabilization of isolated rat liver mitochondria73.

In this work the formation of O2 (1Δg) in chemical and enzymatic reactions was clearly demonstrated by direct detection of the O2 (1Δg) monomol light emission at 1,270 nm using a photomultiplier coupled to a monochromator (Fig. 1C, 2B and 3B); and the observation of the effect of D2O on the acquisition of the spectrum of the light emitted in the near infrared region showing an emission with maximum intensity at 1,270 nm (Fig. 1D).

Another evidence supporting the involvement of this mechanism was obtained by the direct detection of radicals TEMPO in the incubation reaction of TMD or HRP/IBAL with TEMP (Supplementary Fig. 2). The observed EPR spectrum suggests the presence of O2 (1Δg) in the reaction mixture due to a mechanism involving energy transfer from the excited triplet acetone generated to molecular oxygen.

Finally the transfer mechanism involved in the generation of O2 (1Δg) was studied using [18O]-labeled molecular oxygen. Experiments conducted with 18O2 in the presence of EAS (Fig. 4), showed that TMD thermolysis and the enzymatic HRP/IBAL generation of excited triplet acetone yields a mixture of endoperoxides containing 18O and/or 16O atoms namely EAS16O16O, EAS18O18O (Fig. 5 and 6 and Supplementary Fig. 3 to 7), EAS16O18O (Supplementary Fig. 4). Comparison of the relative amounts of EAS16O16O:EAS16O18O:EAS18O18O detected before and after removal of molecular oxygen showed a significant increase in the amount of EAS18O18O and a decrease in the amount of both EAS16O16O (Fig. 5 and 6). These results indicate that the reactions yield mainly 18O2 (1Δg). The differences observed with and without [18O]-labeled molecular oxygen shows that the 16O-oxygen molecule present in the reaction mixture decreases the amount of detected 18O2 (1Δg).

The decrease in the amount of 18O2 (1Δg) detected in the presence of oxygen may be explained by an energy transfer mechanism between 18O2 (1Δg) and 16O2 (3Σg−), yielding 16O2 (1Δg) and 18O2 (3Σg−) as recently demonstrated for aqueous system by Martinez et al.44.

The chemiluminescence, EPR and chemical trapping of 18O2 (1Δg) experiments were also performed in the presence of sorbate, showing a triplet carbonyl quenching effect.

Quenching of triplet carbonyls by the addition of conjugated dienes such as hexa-2,4-dienoates (sorbates)18 or even by the presence of ground state, triplet molecular oxygen can abate the level of chemical damage promoted by triplets to studied targets, either biomolecules or cell organelles74,75.

Considering the enzymatic reactions that give rise to 18O2 (1Δg) through a dioxetane intermediate involving a peroxyl radical. An alternative mechanism by which the formation of 18O2 (1Δg) could be explained is the Russell mechanism16. This requires the generation of [18O]-labeled IBAL peroxyl radicals that recombine to form a hypothetical tetraoxide intermediate, which then decomposes to generate O2 (1Δg). However, this mechanism can be disregarded because it requires the presence of a geminal hydrogen in the IBAL-derived hydroperoxyl radical for the formation of O2 (1Δg) and the detection of O2 (1Δg) containing a mixture of 16O and 18O atoms.

Conclusion

The present study unequivocally demonstrates that singlet molecular oxygen is generated by energy transfer from chemically and enzymatically produced excited triplet acetone to ground state triplet molecular oxygen in aqueous solution (Fig. 7).

Figure 7
figure 7

Singlet molecular oxygen generated enzymatically (A) and chemically (B).

Full size image

This was substantiated by ultraweak CL studies in the near IR region at 1,270 nm with both chemical and enzymatic sources of triplet acetone, which is characteristic of the singlet delta state monomolecular decay of excited molecular oxygen. Indirect analysis based on mass spectrometry and EPR measurements strongly supports the formation of O2(1Δg). Moreover, the use of [18O]-labeled molecular oxygen in association with HPLC-ESI-MS/MS analysis is a highly suitable way to gain relevant mechanistic insights into the formation of singlet molecular oxygen and the decomposition pathways of initially generated peroxide compounds such as dioxetanes and subsequently excited ketones.

The quantum yield of singlet molecular oxygen was found to be higher in aqueous medium than previously demonstrated in organic solvents. This work proposes that enzymatically generated triplet carbonyl may be a contributing source of O2 (1Δg) in non-illuminated biological systems such as root and liver tissues, as earlier proposed independently by Cilento10 and White11.

Biological implication - Taking into consideration that (i) molecular oxygen is c.a. ten times more soluble in membranes that in aqueous medium, (ii) membrane peroxidation involves the intermediacy of alkoxyl and alkylperoxyl radicals derived from polyunsaturated fatty acids, whose dismutation affords triplet carbonyls18 and (iii) phosphate-induced and sorbate-inhibited deleterious permeabilization of mitochondrial membranes via amplification of triplet products73, it is of utmost interest to investigate the participation of singlet molecular oxygen in membrane damage induced by pro-oxidants. In addition, membrane cholesterol and proteins could also be victimized by singlet molecular oxygen formed from triplet carbonyls leading to loss or gain of biological functions. In this regard, noteworthy are the findings by several groups76,77,78 that cholesterol secoaldehyde formed by addition of ozone or singlet molecular oxygen to cholesterol may be implicated in atherosclerosis, Alzheimer disease and apoptosis involving signaling pathways.

Methods

Materials Used

Peroxidase from horseradish (HRP) type VI, K2HPO4, KH2PO4, NH4HCO3, 2,2,6,6-tetramethylpiperidine (TEMP), 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) and hexa-2,4-dienoic acid (sorbic acid) were purchased from Sigma (St. Louis, MO). 2-Methylpropanal (isobutyraldehyde or isobutanal, IBAL), D2O and CCl4 were purchased from Aldrich (Steinheim, Germany). HPLC grade solvents were acquired from Merck (Darmstadt, Germany). IBAL was distilled before use. Deuterated phosphate buffer at pD 7.4 (equivalent to pH 7.0) was prepared by mixing D2O stock solutions of KH2PO4 and K2HPO4. 3,3,4,4-Tetramethyl-1,2-dioxetane (TMD) was prepared as previously described by Kopecky et al.57,58. Standard anthracene-9,10-diyldiethane-2,1-diyl disulfate disodium salt (EAS) endoperoxide (EASO2) was prepared by methylene blue photosensitization in aerated deuterium water containing 8 mM EAS and was subsequently quantified spectrophotometrically30,34,35,44,45. 1,4-Dimethylnaphthalene (DMN) endoperoxide (DMNO2) was also prepared by UVA irradiation of DMN/methylene blue and then quantified spectrophotometrically60.

Low level luminescence emission of excited triplet acetone produced by thermal cleavage of TMD or oxidation of IBAL by HRP/H2O2 and NIR detection of the monomol light emission of O2 (1Δg)

TMD dissolved in CCl4 at concentrations ranging from 2 to 10 mM was transferred from ice to a cuvette holder set at a temperature of 70°C. The light emission was immediately recorded by a FLSP 920 photon counter (Edinburgh Instruments, Edinburgh, UK) consisting of two UV-Visible Hamamatsu detectors R9110, maintained at −20°C by a CO1 thermoelectric cooler also purchased from Edinburgh Instruments. The detector used to measure the steady-state light emission from TMD thermal cleavage was not preceded by any monochromator; therefore, light was recorded directly from the cuvette source. To trace the TMD-elicited chemiluminescence, a second detector was used and its wavelength was determined using a monochromator35. The chemical yield of 10 mM TMD-generated triplet acetone was confirmed in acetonitrile, reportedly evaluated as approximately 30%58. During each experiment, the monomol light emission of O2 (1Δg) at 1,270 nm was monitored using the third detector coupled to the device, a Hamamatsu H10330A-45 apparatus (Hamamatsu city, Japan), also preceded by a monochromator. To determine the O2 (1Δg) generation rate, 2 mM DMNO2 in CCl4 was used as the standard29,33,34,35,60. The same procedure was applied in the quenching studies of triplet acetone with different concentrations of sodium sorbate18. CL arising from the thermal decomposition of 10 mM TMD at 70°C was performed in acetonitrile or air-equilibrated CCl4. The CL spectrum of the triplet excited acetone was obtained from 10 mM TMD in CCl4 at 70°C. The same equipment and procedure were used to observe the generation of triplet excited acetone produced during 5 µM HRP-catalyzed oxidation of 10 mM IBAL and 0.10 mM H2O2 in deuterated aqueous 50 mM phosphate buffer (pD 7.4) at 37°C.

The NIR spectrum of O2 (1Δg) at 1,270 nm was produced by thermal decomposition of 1,4-dimethylnaphthalene endoperoxide (DMNO2) at 50°C60.

EPR spin-trapping studies with TEMP

Samples containing 50 mM TEMP64 and 4 mM TMD were prepared in phosphate buffer (pH 7.4) and incubated for 4 min at 60°C. The reacting solutions were then transferred to an appropriate cuvette and the EPR spectra recorded in an EMX spectrometer (Bruker, Silberstreifen, Germany), using the following parameters: frequency: 100.0 kHz; amplitude: 0.5 mT; time constant 81.920 ms; time conversion: 40.960 ms; and gain: 2.52 × 104.

The experiment with enzymatically generated triplet acetone were conducted in 50 mM phosphate buffer (pH 7.4) that contained 20 µM HRP, 100 mM IBAL and 50 mM TEMP, incubated for 6 min at 37°C and immediately transferred to the cuvette. The EPR parameters used were the same as for TMD, both in the absence and presence of 8 mM sorbate, a triplet acetone quencher. In order to confirm the attribution of the EPR signal to the reaction product of TEMP treated with O2 (1Δg), the spectrum was spiked by addition of 0.4 µM TEMPO to the solution. Computational simulations of the EPR signals were performed using the Winsin program79.

Chemical trapping of O2 (1Δg) by EAS and [18O]-labeled experiments

To unequivocally attest singlet molecular oxygen 9,10-cycloaddition to the EAS probe yielding EASO2, a sample containing 8 mM TMD and 8 mM EAS was prepared in 18O2-purged solutions as follows. TMD/EAS samples were transferred to a closed system and degassed by three freeze-thaw cycles using a vacuum pump. The degassed solution was saturated with 18O2 for 2 h and heated at 70°C. The same procedure was employed as above but without dearating the solution. These samples were kept at 70°C using a Termomixer (Eppendorf, City, Germany) for 24 h.

The degassing-saturation procedure using argon gas was applied to prepare the reaction mixture containing 8 mM EAS, 5 µM HRP, 0.1 mM H2O2 and 50 mM IBAL, except that it was kept in the dark under room temperature for 72 h. Quantification of EAS18O2 using the HRP/IBAL system was carried out under similar concentration conditions, but the reacting solutions were kept under continuous stirring using a Termomixer apparatus (Hamburg, Germany) for 24 h at 37°C.

HPLC-ESI-MS/MS detection of EASO2

HPLC-ESI-MS/MS analyses of the anthracene endoperoxide EASO2 were conducted by injecting 25 µL of the sample in a Shimadzu HPLC system (Tokyo, Japan) coupled to a mass spectrometer Quattro II triple quadrupole (Micromass, Manchester, UK). Endoperoxide EASO2 was separated using a Luna C18 reverse phase column, 250 × 4.6 mm, 5 µM particle size (Phenomenex, Torance, CA,) that was kept at 25°C. The liquid phase consisted of 25 mM ammonium formate (solvent A) and acetonitrile:methanol 7:3, v/v (solvent B) with linear gradient of 25% B during 15 min, 25 to 70% B for 1 min, 70% B until 25 min, 70 to 25% B during 1 min and 25% B until 30 min. The eluent was monitored at 210 nm with a flow rate of 0.8 mL.min−1. First 5 min of run gradient was discarded and 10% of flow rate was directed to the mass spectrometer. Ionization of the sample was obtained by electrospray ion source (ESI) in the negative ion mode using the following parameters: source temperature, 120°C; desolvation temperature, 200°C; cone voltage, 15 V; collision energy, 10 eV. The endoperoxides EASxOxO were detected by the loss of the oxygen molecule, in the Selected Reaction Monitoring mode (SRM). The transitions recorded were m/z 228→212 for EAS16O16O, m/z 229→212 for EAS18O16O and m/z 230→212 for EAS18O18O.

UHR-ESI-Q-TOF detection of EASO2

High resolution mass spectrometry analysis of EASxOxO endoperoxides were performed in an UHPLC Agilent coupled to an UHR-ESI-Q-TOF Bruker Daltonics MaXis 3G mass spectrometer with CaptiveSpray source in the negative mode. The UHPLC mobile phase consisted of ammonium formate (solvent A) and acetonitrile:methanol 7:3, v/v (solvent B) with the following linear gradient: 25% B during 15 min, 25 to 70% B for 1 min, 70% B until 25 min, 70 to 25% B during 1 min and 25% B until 30 min. Endoperoxides was separated on a Luna C18 reverse phase column, 250 × 4.6 mm, 5 µM particle size (Phenomenex, Torance, CA) and monitored at 210 nm. The flow rate was 0.8 mL.min−1. Reverse phase column was kept at 30°C. The ESI conditions were: capillary, 4.0 kV; dry heater, 180 °C; dry gas, 8.0 l/min; end plate, −450 V. Nitrogen was used as collision gas and the CID (collision-induced dissociation) energy was 10 eV. The instrument was externally calibrated using an ESI low concentration tuning mix over the m/z range of 100 to 2000. The Bruker Data Analysis software (version 4.0) was employed for data acquisition and processing.