Introduction

Photosynthetic organisms often work under strong light stress when the intensity of incident light exceeds the ability of the photosynthetic cells to utilize light energy for photosynthesis. Protection from the photodynamic damage caused by excess light energy is a prerequisite of survival and normal functioning of the photosynthetic apparatus. After the pioneering work by Griffiths et al. (1955) and subsequent papers of other groups, it is generally adopted that carotenoids with nine and more conjugated double bonds (CDB) play an important role in protection of living tissues against photooxidative stress (Griffiths et al. 1955; Mathews-Roth et al. 1974; Foote 1976; Cogdell et al. 2000; Britton 2008 and refs therein). Contrary to this opinion, A. A. Moskalenko’s Lab (Institute of Basic Biological Problems, Russia) obtained the data, which suggested that carotenoids photosensitize the 1O2-mediated destruction of bacteriochlorophyll (BChl) in the LH2 complexes of purple bacteria (Makhneva et al. 2007, 2009, 2020, 2021). As shown quite recently, the action spectrum of this process corresponds to the absorption spectra of rhodopin and lycopene. However, the quantum yield of the BChl photodestruction is very low, equal to 0.0003 (Klenina et al. 2022). Based on these observations, one could assume that carotenoids are capable of low efficient generation of singlet oxygen under photoexcitation in the bacterial photosynthetic apparatus. Perhaps, this process because of its low efficiency is not very important for photodamage of bacterial cells. However, we feel that it is of great interest from the viewpoint of photophysics of carotenoid molecules. It is noteworthy, that before our studies, photogeneration of 1O2 by isolated molecules of bacterial carotenoids had never been investigated, although it is known that 1O2 formation is rather efficiently photosensitized by the visual pigment—retinal, a polyene with 6 CDB (Delmelle 1979; Krasnovsky and Kagan 1979; Edge and Truscott 2018).

In this connection, couple of years ago, we have started a systematic study of 1O2 photosensitization by bacterial carotenoids with different lengths of CDB using direct detection of carotenoid-photosensitized 1O2 phosphorescence. It should be noted that photosensitized phosphorescence of singlet oxygen at 1270 nm in aerated solutions of different pigments, including chlorophylls, BChls and retinals, was firstly observed by one of us in nineteen seventies (Krasnovsky 1976, Krasnovskiĭ 1977; Krasnovsky 1979; Krasnovsky and Kagan 1979). Presently, detection of this phosphorescence is used worldwide as the most accurate method of singlet oxygen detection and investigation. The most common phosphorescence mechanism corresponds to the following scheme:

$$\begin{aligned} & ^{1} {\text{P}}_{0} + hv \to^{1}\!\!{\text{P}}^{*} \to ^{3}\!\!{\text{P}}^{*} +^{3}\!{\text{O}}_{2} \to^{1}\!{\text{O}}_{2}\\& ^{1}{\text{O}}_{2} \to ^{3}\!{\text{O}}_{2 } + hv\;\left( {1270\;{\text{nm}}} \right), \end{aligned} $$

where 1P0, 1P*, and 3P* are molecules of a pigment-photosensitizer in the ground and excited singlet and triplet states, 1O2 is the 1Δg singlet state of molecular oxygen (Krasnovsky 1979, 2008; Schweitzer and Schmidt 2003 and refs therein).

For our work, new 1O2 phosphorescence spectrometers assembled recently by A.A. Krasnovsky’s group (Bach Institute of Biochemistry, Russia) (Krasnovsky et al. 2019) were applied. In the previous papers, polyenes with a short chain of CDB were studied: phytoene (3 CDB), phytofluene (5 CDB), and ζ-carotene (7 CDB) (Ashikhmin et al. 2020, 2022). It was found that photoexcitation of these polyenes causes generation of 1O2 in aerated solutions. The quantum yields of 1O2 generation by phytoene and ζ-carotene appeared to be low: ≈ 0.02 and 0.014, respectively. Unexpectedly, phytofluene efficiently generated 1O2. First estimation of the quantum yield (ΦΔ) of this process based on the steady-state phosphorescence measurements led us to a value of 0.85 ± 0.05 (Ashikhmin et al. 2020). In the next paper, using more accurate time-resolved measurements, we arrived to a slightly lower value 0.39 ± 0.06, which is still very high (Ashikhmin et al. 2022). The rate constants of 1O2 quenching by these carotenoids were also obtained (Ashikhmin et al. 2020, 2022). In the present work, the methods developed in the prior papers are applied to natural bacterial carotenoids with long chain of CDB: neurosporene (9 CDB), lycopene (11 CDB), rhodopin (11 CDB), and spirilloxanthin (13 CDB) (Fig. 1).

Fig. 1
figure 1

The structural formulae of the studied carotenoids

Full size image

Materials and methods

Carotenoid isolation

Carotenoids were obtained from cells of purple photosynthetic bacteria. Neurosporene was isolated from Rhodobacter sphaeroides strain G1C (received from Prof. Takaichi S., Nippon Medical School, Japan). Lycopene was purified from Ectothiorhodospira haloalkaliphila. Rhodopin and spirilloxanthin were extracted from Allochromatium vinosum and Rhodospirillum rubrum, respectively. The bacterial cells used for carotenoid isolation were grown at 28 ± 2 °C in anaerobic conditions under illumination by white light of an incandescent lamp 90 W/m2 on the Hutner's, Pfennig's, Larsen's and Ormerod's media, respectively (see Ashikhmin et al. 2017). Under these conditions, the carotenoids with the short CDB chain were not accumulated in bacterial cells (Ashikhmin et al. 2017). For accumulation of the carotenoids with short CDB chains, inhibitors of carotenoid biosynthesis should be added as was done in our prior papers (Ashikhmin et al. 2020, 2022).

Carotenoid isolation was carried in two stages as described earlier (Ashikhmin et al. 2017, 2022). At the first stage, cells were destroyed by ultrasonic disintegration. A mixture of photosynthetic pigments (carotenoids and bacteriochlorophyll) was extracted from the membranes, and then, carotenoids were extracted from the mixture. In the second stage, individual carotenoids were released by HPLC on the Agilent 1200 (Agilent Technologies, USA) with a preparation column (10 × 250 mm) with a reverse phase of Waters Spherisorb ODS2 (Waters, USA), using the solvent gradient described earlier (Ashikhmin et al. 2017). The purity of isolated carotenoids was additionally checked on an analytical HPLC device Shimadzu (Shimadzu, Japan) using an Agilent Zorbax SB-C18 reverse phase column (4.6 × 250 mm) (Agilent Technologies, USA) at 22 °C. Carotenoids were identified by their absorption spectra and retention time.

A typical HPLC chromatograms of isolated carotenoids are indicated in Supplementary materials. The absorption coefficients of carotenoids were taken from Britton (1995): neurosporene—156; lycopene—185; rhodopin—166 and spirilloxanthin—147 mM−1 cm−1 (light fraction of petroleum ether).

Singlet oxygen detection

Singlet oxygen was studied in aerated pigment solutions by measuring photosensitized phosphorescence at 1270 nm corresponding to the radiative deactivation of the singlet (1Δg) state of oxygen molecules. Measurements were carried out using the spectrometers described elsewhere (Krasnovsky et al. 2019). The spectrometers allow detection of singlet oxygen phosphorescence under stationary and pulsed LED excitation of photosensitizers. It also makes possible to analyze the spectral parameters (emission and excitation spectra) of phosphorescence and record its kinetics traces after short exciting flashes. To measure excitation spectra, we had a set of five LEDs with the emission maxima at 347, 367, 400, 421, and 460 nm, obtained from the firms Polironic (Russia) and Artleds (Russia). When measuring phosphorescence parameters, the emission power of LEDs on the surface of the cuvette did not exceed 10 mW/cm2. The half-width of the LED radiation bands was 11–26 nm. For time-resolved measurements, pulsed LEDs equipped with a modulating electronic chips and emission maxima at 365 nm (half bandwidth—18 nm), 395 nm, and 450 nm (half bandwidth—11 nm) (Polironic, Moscow) were used.

LED emission was focused in a spot with a diameter of 5 mm on the surface of a 1 cm quartz cuvette with a tested solution. The average excitation intensity was controlled by the ThorLabs PM-100D power meter with S120VC sensor photovoltaic head (ThorLabs, USA) and Ophir ORION-TH power meter with a 20C-SH thermal sensor head (Ophir Optronics, Israel). Phosphorescence of 1O2 was detected at an angle of 90° relative to the excitation beam using a cooled photomultiplier FEU-112 with the S-1 spectral response (Ekran Optical Systems, Russia) through the cut-off filters that transmit IR emission at the wavelength ≥ 1000 nm, and one of three replaceable interference filters with maximum transmittance at 1230, 1270, and 1310 nm and half-band width of 10 nm. The stationary phosphorescence intensity was recorded using a digital millivoltmeter (Econix-Expert, Russia).

When time-resolved phosphorescence measurements were carried out, the repetition rate and duration of LED pulses were regulated by an external pulse generator. The pulse duration in most measurements was 10 μs, and the pulse repetition frequency was changed in the range of 5–100 Hz. The PMT signal was sent through a broadband (400 MHz) preamplifier to an electronic board operating in a time-resolved photon counting mode (Parsek, Russia). The counting board, which was launched from an additional pulse of the generator synchronized with the LED pulse, divided the time interval between flashes into 256, 512, or 1024 channels. The signal of the board came through the USB port to a personal computer that processed the signal. Kinetic curves were obtained by accumulating PMT impulses in each channel during 2–30 min. The quantum yield of 1O2 generation (ΦΔ) was measured using as a reference an organic photosensitizer phenalenone (perinaphthenone, 1H_phenalen-1-one) (Merck, USA), which has a broad absorption band with a maximum at 350 nm. It is known that the quantum yield of the 1O2 photogeneration by phenalenone is close to 1(0.95 ± 0.05) (Oliveros et al. 1991; Schmidt et al. 1994). The experiments were carried out using hexafluorobenzene (C6F6) as a solvent, which was specially purified for our experiments (Pim-Invest, Russia). This solvent was employed because there are no hydrogen atoms in its molecule. Therefore, the lifetime of singlet oxygen in hexafluorobenzene is very long (16 ms) that facilitates the detection and studies of 1O2. In addition, C6F6 does not enter chemical interaction with carotenoids and singlet oxygen (Krasnovsky et al. 2019; Ashikhmin et al. 2020, 2022).

Results

Neurosporene

The absorption spectrum of neurosporene in hexafluorobenzene has three major peaks at 417, 441, and 470 nm (Fig. 2A). Under laser irradiation of neurosporene solutions (2–8 μM), we observed weak phosphorescence at 1270 nm. Excitation spectrum of phosphorescence was estimated using three continuous LEDs with the emission wavelengths of 399, 461, and 530 nm. Details of the measurement procedure were described in previous papers of our group (Ashikhmin et al. 2020, 2022). With each LED, luminescence was detected through three interference filters transmitting light at 1270, 1230, and 1310 nm. Then, at each wavelength, the background luminescence of a quartz cuvette with acetone or 1:1 (v/v) mixture of C6F6 and acetone having no carotenoids was recorded. The background signals were subtracted from the signals obtained from carotenoid solutions. The spectral maximum of the resulting difference signals was always observed at 1270 nm that corresponds to the spectral maximum of singlet oxygen phosphorescence.

Fig. 2
figure 2

Neurosporene in C6F6. A The absorption spectrum of neurosporene (4 μM) (solid line) and the excitation spectrum (vertical lines) of 1O2 phosphorescence at 1270 nm estimated using a set of LEDs (100 mW) with fixed radiation wavelengths, the spectra are normalized at 461 nm (Iph is the phosphorescence intensity in photons per second, n is a number of incident photons), B decays of singlet oxygen phosphorescence at 1270 nm photosensitized by phenalenone after LED pulses (2 μs, 405 nm, 1.2 μJ, 10 kHz repetition rate, 10 min accumulation time) in Cartesian or semi-logarithmic coordinates in solutions containing phenalenone (3 μM) and 2.6 (1) and 8.7 (2) μM neurosporene, and C the Stern–Volmer plot for the dependence of the phosphorescence decay rates in phenalenone solutions from the concentration of added neurosporene

Full size image

The main maximum of the phosphorescence excitation spectrum was always observed at 460 nm that corresponds to the main maximum of the neurosporene absorption spectrum (Fig. 3A). Note that the carotenoids with short CDB chains (phytoene, phytofluene, and ζ-carotene) studied in our prior papers have zero absorbance at this wavelength. The long wavelength maxima of these short-chain carotenoids were at 293, 369, and 420 nm, respectively (Ashikhmin et al. 2020, 2022). Therefore, it is excluded that admixture of these carotenoids contributed to the phosphorescence excitation in solutions of neurosporene and other carotenoids investigated in the present work.

Fig. 3
figure 3

Determination of the quantum yields of 1O2 photogeneration by carotenoids using lycopene in aerated hexafluorobenzene. A Decays of.1O2 phosphorescence after 10 μs LED pulses (3.6 μJ, 405 nm, 10 kHz repetition rate) in the solution of lycopene (1.8 μM with absorbance of 0.13 at 405 nm) (1, 3) and mixed solution of phenalenone (with absorbance of 0.13 at 405 nm) and lycopene (2, 4) in the Cartesian (1, 2) or semi-logarithmic coordinates (3, 4). The decay curves (2, 4) were reduced 40-fold compared to the original records in order to be seen on the scale of the present figure. The decays were recorded with the 10 μs delay after the start of LED pulses and 40 min accumulation time. B Spectra of the phosphorescence emission in solution of lycopene alone (black bars) and in the mixed solution of phenalenone and lycopene (gray bars)

Full size image

The phosphorescence decay time in neurosporene solutions was much shorter than the lifetime of singlet oxygen (τΔ) in neat C6F6. For detailed analysis, 1O2 phosphorescence was studied in mixed solutions containing both phenalenone and neurosporene. The excitation power was selected so that the degree of the carotenoid destruction during the measurement time did not exceed 5% of the initial absorbance value. Figure 3B indicates that decays of 1O2 phosphorescence arising after excitation of phenalenone by short excitation pulses are exponential. The phosphorescence decay time declined after the addition of neurosporene and the increase of the neurosporene concentration. At the same time, the initial phosphorescence intensity obtained by extrapolating the exponentials to the zero time after the flash [Iph(0)] remained unchanged. The latter implies that within the range of the carotenoid concentrations used in our experiments the carotenoid did not influence the rate of the 1O2 generation by phenalenone but they strongly increase the decay rate of singlet oxygen (Fig. 2 B, C).

Quenching of 1O2 by neurosporene obeys the Stern–Volmer Eq. (1) (Fig. 1C):

$$ k_{{{\text{obs}}}} = k_{{\text{o}}} + k_{{\text{q}}} \left[ {{\text{Car}}} \right], $$
(1)

where kobs is the observed rate constant of phosphorescence decay (in s−1) after LED pulses in solutions containing both a photosensitizer and Car, k0 is the rate constant of the phosphorescence decay in phenalenone solutions without Car, and kq is the bimolecular rate constant of 1O2 quenching by Car. The value of kq was calculated from the slope of the linear Stern–Volmer plot. The neurosporene concentrations were calculated from the absorbance measurements using the molar absorption coefficient indicated by Britton 1995.

As a result, it was obtained that the neurosporene kq is rather close to the rate constant of the diffusion- limited reactions (Table 1), being many-fold greater than kq obtained previously for ς-carotene and phytofluene (Ashikhmin et al. 2022). It is noteworthy that this constant corresponds to a physical quenching process, because under conditions of our experiments, the bleaching of neurosporene was negligible or was not observed at all. This conclusion agrees with the rates of chemical quenching of singlet oxygen obtained previously in solutions of β-carotene and fucoxanthin (Krasnovsky and Paramonova 1983).

Table 1 Quantum yields of 1O2 generation (ФΔ) and rate constants for 1O2 quenching (kq) in solutions of carotenoids in hexafluorobenzene obtained in the present and prior (Ashikhmin et al. 2020, 2022) works of our group
Full size table

The quantum yields of singlet oxygen were determined using both steady-state and time-resolved phosphorescence measurements. For steady-state measurements, the following basic equation was applied:

$$ I_{{{\text{ph}}}} = I_{{{\text{ex}}}} \left( {1 - 10^{ - A} } \right)\Phi_{\Delta } k_{{\text{r}}} \tau_{\Delta } , $$
(2)

where Iph and Iex are the steady-state intensities (Einstein per second) of phosphorescence and exciting light, respectively. A is absorbance of a photosensitizer at the wavelength of excitation, kr is the radiative rate constant for 1O2 phosphorescence emission, and τΔ is the real lifetime of singlet oxygen in solutions.

As Iph/(Iex(1 − 10−A)) is equal to the phosphorescence quantum yield (Φph), Eq. 2 can be simplified:

$$ \Phi_{{{\text{ph}}}} = \Phi_{\Delta } k_{{\text{r}}} \tau_{\Delta } . $$
(3)

The values of ΦΔ for carotenoids (ΦΔ)car can be calculated from the ratios of the ΦΔ values in solutions of carotenoids and phenalenone (Eq. 4). In this case, constants kr are reduced:

$$ \left( {\varvec\Phi_{\Delta } } \right)_{{{\text{car}}}} = \left(\varvec\Phi_{{{\text{ph}}}} / \tau_{\Delta } \right) {_{{{\text{car}}}} /} \left(\varvec\Phi_{{{\text{ph}}}}/ \tau_{\Delta }\right)_{{{\text{phen}}}} . $$
(4)

Since carotenoids are very strong quenchers of 1O2, the τΔ in carotenoid solutions was about 150-fold smaller than in solutions of phenalenone (16 ms, Krasnovsky et al. 2019).

For time-resolved measurements, another equation was used.

$$ I_{{{\text{ph}}}} \left( t \right) = I_{{{\text{ex}}}} \left( {1 - 10^{ - A} } \right)\Phi_{\Delta } k_{{\text{r}}} (\exp \, ( - t/t_{\Delta } )), $$
(5)

where Iph(t) is the instant phosphorescence radiative rates in Einstein per second and Iex is the intensity of short exciting LED pulse in Einstein per pulse. As phosphorescence decays were exponential, extrapolation of the exponential functions to the zero time after exciting pulse is equal to

$$ I_{{{\text{ph}}}} \left( 0 \right) = I_{{{\text{ex}}}} \left( {1 - 10^{ - A} } \right)\Phi_{\Delta } k_{{\text{r}}} . $$
(6)

This function allows for calculation of (ΦΔ)car:

$$ \left( {\Phi_{\Delta } } \right)_{{{\text{car}}}} = \left[ {I_{{{\text{ph}}}} \left( 0 \right)/\left( {I_{{{\text{ex}}}} \left( {1 - 10^{ - A} } \right)} \right)} \right]_{{{\text{car}}}} /\left[ {I_{{{\text{ph}}}} \left( 0 \right)/\left( {I_{{{\text{ex}}}} \left( {1 - 10^{ - A} } \right)} \right)} \right]_{{{\text{phen}}}} . $$
(7)

Figure 3A illustrates the procedure of the quantum yield measurement using time-resolved detection of singlet oxygen phosphorescence. This procedure was similar for all investigated carotenoids. For illustration, the lycopene solution was chosen.

In the experiment with neurosporene, excitation was produced by 2 μs pulses of LED at 405 nm. The concentration of neurosporene was 2.0 μM. The absorbance of neurosporene at 405 nm was 0.2. In addition, a mixed solution containing both phenalenone and the carotenoid was prepared. The phenalenone concentration was 3 μM. Absorbance at 405 nm was 0.13 in a 1 cm cell. Comparison of the Iph(0) (the zero time phosphorescence intensity) in solutions of phenalenone and neurosporene led us to the value of the quantum yield of 1O2 generation by neurosporene equal to 0.02. Using steady-state measurements, we arrived to the value of 0.04. Average value is indicated in Table 1. Surprisingly, it was slightly higher than that obtained previously for ζ-carotene (Ashikhmin et al. 2022). Nevertheless, due to much stronger quenching of 1O2 by neurosporene, the phosphorescence signal was rather weak. The 3–5-fold smaller signals would be near the noise level and could not be reliably analyzed.

Lycopene

The absorption spectrum of lycopene has three main peaks at 442, 469, and 500 nm (Fig. 4A). The studies were carried out according to the same plan as the studies of neurosporene. Under laser irradiation of lycopene solutions, weak phosphorescence at 1270 nm was observed. Excitation spectrum of phosphorescence was estimated using three continuous LEDs. Figure 4A indicates that the action spectrum correlates with the main maxima of the lycopene absorption spectrum. The phosphorescence decay time in the lycopene-containing solutions decreased with the increase of the lycopene concentration (Fig. 4B). The rate constant of physical quenching of 1O2 by lycopene was calculated from the Stern–Volmer plot of the phosphorescence decay rates versus lycopene concentrations. The obtained kq was about 2-fold greater than that for neurosporene (Table 1). It is known that lycopene is probably the strongest quencher of singlet oxygen among the C40 carotenoids with 11 CDB (Di Mascio et al. 1989; Conn et al. 1991). Our data are consistent with this statement, although the difference from other carotenoids is less than that obtained by Di Mascio et al. and resembles the data by Conn et al. (1991).

Fig. 4
figure 4

Lycopene in C6F6. A The absorption spectrum (solid line) and the excitation spectrum of 1O2 phosphorescence at 1270 nm estimated using a set of LEDs with fixed radiation wavelengths, normalized at 461 nm (vertical lines) (Iph is the phosphorescence intensity in photons per second, n is a number of incident photons), B decays of singlet oxygen phosphorescence after LED pulses (10 μs, 405 nm, 6 μJ, 10 Hz repetition rate, 15 min accumulation time) in Cartesian or semi-logarithmic coordinates in solutions containing phenalenone alone (1) and after addition of 0.03 (2) and 0.07 (3) μM rhodopin, and C the Stern–Volmer plot for the dependence of the phosphorescence decay rates in phenalenone solutions from the concentration of added lycopene

Full size image

Figure 3A illustrates the procedure of the quantum yield measurement using time-resolved detection of singlet oxygen phosphorescence in lycopene-containing solutions. Both steady-state and time-resolved methods showed that the quantum yield of 1O2 generation by lycopene is equal to that for ζ-carotene and about 2-fold smaller than in neurosporene solutions, although this difference between the ΦΔ values for neurosporene and lycopene is close to the measurement error range (Table 1).

Rhodopin

The absorption spectrum of rhodopin has three main peaks at 442, 469, and 496 nm, which are close to the main absorption peaks of lycopene (Fig. 5A). The action spectrum of weak phosphorescence at 1270 nm observed under laser irradiation of rhodopin solutions correlated with the rhodopin absorption spectrum (Fig. 5A). The phosphorescence decay time in rhodopin-containing solutions decreased with the increase of rhodopin concentrations (Fig. 5B). The rate constant of the physical quenching of 1O2 by rhodopin was calculated from the Stern–Volmer plot of the phosphorescence decay rates versus rhodopin concentrations. The kq obtained was about 10% smaller than that for lycopene (Table 1). The quantum yield of 1O2 photogeneration by rhodopin was estimated from comparison of the 1O2 phosphorescence intensities in the mixed solutions of phenalenone and rhodopin with the same parameters of the solutions of rhodopin alone. For excitation of phenalenone, the LED with the radiation wavelength of 367 nm was used. For excitation of rhodopin, the 460 nm LED was applied. Phenalenone and rhodopin had equal absorbance at the excitation wavelengths. Other experimental conditions were similar to those indicated in Fig. 4, and therefore, the data are not shown. The obtained quantum yield is presented in Table 1.

Fig. 5
figure 5

Rhodopin in C6F6. A The absorption spectrum (solid line) and the excitation spectrum (vertical lines) of 1O2 phosphorescence estimated using a set of continuous LEDs (100–150 mW) with fixed radiation wavelengths, the spectra are normalized at 461 nm. B Decays of singlet oxygen phosphorescence after LED pulses (10 μs, 405 nm, 6 μJ, 10 Hz repetition rate, 15 min accumulation time) in Cartesian or semi-logarithmic coordinates in solutions containing phenalenone alone (1) and after addition of 0.04 (2) and 0.09 (3) μM rhodopin, and C the Stern–Volmer plot for the dependence of the phosphorescence decay rates in phenalenone solutions from the concentration of added rhodopin

Full size image

Spirilloxanthin

The absorption peaks of spirilloxanthin are markedly shifted to longer wavelengths compared to rhodopin. The main absorption maxima are at 462, 490, and 523 nm (Fig. 6). We could not obtain reliable signal of 1O2 phosphorescence in solutions of this pigment. Therefore, it can be proposed that the phosphorescence intensity in solutions of spirilloxanthin is smaller than the sensitivity limit of our spectrometers. It was estimated that the phosphorescence quantum yield is less than 0.5. The rate constants of 1O2 quenching by rhodopin and spirilloxanthin appeared to be very similar. The quenching is physical with the kq values near the rate constants for the diffusion-limited reactions.

Fig. 6
figure 6

Spirilloxanthin in C6F6. A Absorption spectrum of spirilloxanthin, B decays of singlet oxygen phosphorescence after LED pulses (10 μs, 405 nm, 6 μJ, 10 Hz repetition rate, 15 min accumulation time) in Cartesian or semi-logarithmic coordinates in solutions containing phenalenone alone (1) and after addition of 0.03 (2) and 0.08 (3) μM spirilloxanthin, and C the Stern–Volmer plot for the dependence of the phosphorescence decay rates in phenalenone solutions from the concentrations of added spirilloxanthin

Full size image

Discussion

Thus, it has been shown for the first time that isolated carotenoid molecules with long chains of CDB (9–11): neurosporene, lycopene, and rhodopin are capable of singlet oxygen generation upon photoexcitation at the wavelength region of their main absorption maxima. The quantum yields were estimated to be (1.5–3.0) × 10−2 in solutions of all investigated carotenoids. Photogeneration of singlet oxygen was not observed in spirilloxanthin solutions (Table 1). It was estimated that ФΔ in the spirilloxanthin solutions is at least 3-fold smaller than in solutions of lycopene. Comparison with the data of our prior papers (Ashikhmin et al. 2020, 2022) indicates that the carotenoids with long chains of CDB generate 1O2 much less efficiently than phytofluene (5 CDB) and slightly more efficiently than ζ-carotene (7 CDB) (Table 1).

It is known that carotenoids emit fluorescence under photoexcitation (see references in Table 2). A simplified energy diagram of the main electronic transitions in carotenoid molecules is shown in Fig. 7. Major absorption bands correspond to vibronic So(1Ag−) → S2 transitions (Frank and Christensen 2008). According to Fujii et al. (2001a, b), the S2 level is in fact a sum of two levels 1Bu− and 1Bu+ (S2 and S3). Anyway, for simplicity, sum of these levels is denoted in this paper by one symbol S2. At longer wavelengths, much weaker symmetry forbidden So(1Ag−) → S1(2Ag−) transition exists, which is normally not seen in the absorption spectra. Fluorescence is emitted due to radiative deactivation of the S2 and S1 states. As indicated in Table 2, an apparent correlation is seen between the ФΔ in solutions of phytoene, phytofluene, and ζ-carotene, and the literature values of the quantum yields (Фf1) and lifetime (τf1) of the fluorescence corresponding to the forbidden S1 → S0 transitions of these carotenoids (see also Ashikhmin et al. 2020, 2022). At the same time, no correlation was observed between ФΔ for these carotenoids and fluorescence parameters of the S2 → S0 transitions Фf2 and τf2. The τf2 are very short (of 0.1–0.2 ps) and almost independent of the structure of the investigated carotenoids. The Фf2 values are at least 100-fold smaller than the values of Фf1 (Table 2).

Table 2 Relative quantum yields of 1O2 generation by carotenoids in hexafluorobenzene calculated from the data of Table 1
Full size table
Fig. 7
figure 7

Tentative energy diagram of the electronic transitions in carotenoid molecules resulting in photosensitization of the 1O2 formation

Full size image

These observations allowed us to propose that the carotenoids with a short chain of CDB generate 1O2 due to energy transfer to oxygen from the T1 triplet states, which are populated owing to interconversion of the forbidden S1 state, which in its turn, is formed due to spontaneous deactivation of the S2 state (Fig. 7) (Ashikhmin et al. 2022). The proposed mechanism agrees with the reported energy and temporal parameters of the triplet states of phytoene, phytofluene, and ζ-carotene (Bensasson et al. 1976).

However, for the carotenoids with 9–13 CDB, there is no correlation between ФΔ and literature values of Фf1 and τf1 (Table 2). For instance, ФΔ for neurosporene is slightly greater than that for ζ-carotene, whereas the Фf1 and τf1 for neurosporene are much smaller than those for ζ-carotene. Moreover, the yield of neurosporene fluorescence corresponding to the S2 → S0 transition (Фf2) is 2-fold higher than Фf1, despite the very short lifetime (0.1–0.2 ps) of the S2 → S0 fluorescence (Andersson et al. 1995). When number of CDB increases from 5 to 11, Фf2 increases by more than three orders of magnitude, while the Фf1 decreases by more than two orders (Table 2 and refs therein). The both values of ФΔ and Фf2 are almost the same in the range of carotenoids molecules with 9–11 CDB. It is reasonable to propose based on these observations that singlet oxygen is generated due to the triplets (T2) formed owing to direct interconversion of the S2 state without intermediate S1 population (Fig. 7). To the best of our knowledge, dual triplet state formation has never been observed experimentally (see for instance, Bensasson 1976, Burke et al. 2000). Therefore, it might be suggested that T2 triplets transfer their energy to oxygen directly with 1O2 formation or populate T1 due to internal energy transfer from T2. Further experiments are needed for verification of these assumptions.

Returning to the bacterial complexes of purple bacteria, it is known that rhodopin takes about 90% of the total carotenoid content in the LH2 bacterial cells of Allochromatium minutissimum (Britton 2008; Klenina et al. 2022 and refs therein). As Table 1 indicates, the ФΔ value in rhodopin solutions is about 100-fold greater than the quantum yield of the 1O2-mediated destruction of BChl in the LH2 complexes of purple bacteria (Klenina et al. 2022).

However, it should be taken into account that BChl strongly quenches singlet oxygen (Krasnovsky 1977, 1979; Krasnovsky et al. 1985; Borland et al. 1987; Egorov et al. 1990). This process is mainly due to a physical mechanism of energy dissipation:

$$^{1} {\text{O}}_{2} + {\text{BChl}} \to^{3}\!\!{\text{O}}_{2} + {\text{BChl}} + {\text{heat}}. $$

Contribution of BChl oxygenation is only 3% (Krasnovsky et al. 1985). If it is also true for BChl in bacterial LH2 complexes, we can expect that the quantum yield of rhodopin-photosensitized BChl photodestruction in LH2 complexes should be about 9 × 10−4. This value reasonably correlates with the data of Klenina et al. (2022). Further studies of this problem are in progress.

The rates constants for 1O2 quenching by carotenoids with 9–13 CDB were found to be close to the rate constants of the diffusion-limited reactions. To our knowledge, the exact values of kq for phytoene, phytofluene, ζ-carotene, rhodopin, and spirilloxanthin were obtained in our present and prior works (Ashikhmin et al. 2020, 2022) for the first time. Our data support the results of many researchers (Di Mascio et al. 1989; Conn et al. 1991; Foote 1976; Frank and Christensen 2008 and refs therein) that carotenoids with long chain of CDB efficiently deactivate 1O2 due to a physical energy dissipation process. It might be that the carotenoids quench 1O2 due to the reverse transfer of energy from singlet oxygen to low-lying triplet levels of carotenoids. Our data do not contradict this hypothesis. However, they are not sufficient for detailed mechanistic analysis of this process.

Thus, our data suggest that the carotenoids with long chain of CDB combine the ability to efficiently deactivate 1O2 with the ability to serve as weak photosensitizers of singlet oxygen generation. Hence, under certain conditions they might play a role of low efficient photosensitizers of photodynamic damage. At the same time, they have a very high potential as protectors of the photosynthetic structures against singlet oxygen attack. However, in the bacterial photosynthetic systems, another potent quencher of singlet oxygen BChl is also present (Krasnovsky 1977; Krasnovsky 1979; Krasnovsky et. al. 1985; Borland et al. 1987; Egorov et al. 1990), due to which the protective function of carotenoids is not so obvious. Similar processes occur also in chloroplasts where both carotenoids and chlorophyll are potent 1O2 quenchers what was discussed many times in the papers of one of us (Krasnovsky 1994 and refs therein).