Combined contributions of carotenoids and chlorophylls in two-photon spectra of photosynthetic pigment–protein complexes—A new way to quantify carotenoid dark state to chlorophyll energy transfer?

Light is absorbed by green plants through a complex network of multiple antenna complexes and funneled to the reaction center. There, the initial charge separation takes place that drives all biochemical reactions following in photosynthesis.³¹ Plants have the ability to react to widely varying light conditions.^8,9 This enables them to use the available light effectively for photosynthesis and, on the other hand, to react if there is more light than can be utilized by the reaction center. Otherwise, deleterious photoproducts may be formed. A protective mechanism, which is also known as non-photochemical quenching (NPQ), is triggered under saturating light conditions.^22,28 The precise molecular mechanisms of this process are still not entirely clear. In the past, several possible mechanisms have been suggested, including charge transfer from the first excited Car state Car S₁ to the Chl Qy state,^1,18 charge transfer only between Chls,^25,26 energy transfer from Chl Qy to the Car S₁ state,^11,30 or quenching excitonic states.^7,17 Despite several decades of research, many aspects are still poorly understood.^{1–4,7,10,17,18,23,25,26,29,30,33} Nevertheless, most researchers agree that electronic interaction between Chls and Cars seem to be crucial. In addition, the carotenoid based xanthophyll cycle, which is driven by varying light conditions, does certainly play an important role.^11,27

When exploring the role of carotenoids in light-harvesting as well as its regulation, one obstacle is the fact that transitions between the Car electronic ground state and their first excited state, Car S₁, are optically forbidden in conventional one-photon spectroscopy. However, it has been shown that TPE is able to excite this Car state as the optical selection rules are different in TPE vs OPE.^{6,15,20,32,36,37,39}

In contrast to one-photon absorption spectroscopy, the measurement of two-photon spectra is complicated by the fact that the decrease in light-intensity after a sample due to two-photon absorption (TPA) is so low that it is very hard to detect, in particular, for biological samples. Therefore, in most cases, two-photon excitation (TPE) spectra are measured in which a fluorescence emission intensity is detected as a function of the two-photon excitation wavelength rather than directly measuring a two-photon absorption cross section. All spectra shown in this Communication are based on such two-photon excitation spectra. To compare the extent by which distinct samples differ in the magnitude of two-photon absorption is then only possible by directly comparing the two-photon excitation spectra at exactly identical or defined concentrations. This was done by us in the past very carefully in order to determine relative contributions of different pigments to spectra of entire pigment–protein complexes. To derive directly absolute TPA cross sections from TPE spectra would require direct comparisons of TPE spectra to TPE spectra of samples with exactly known TPA cross section and fluorescence quantum yields. Since the latter is often hard to know with precision, we usually hesitate to provide absolute TPA cross section estimates based on such comparisons. However, when the fluorescence originates in certain samples from the exact same emitter, then the relative TPE magnitudes largely reflect relative TPA cross sections, even if it does not give the absolute numerical TPA-cross section values. The situation is more complicated for the detection of TPE spectra originating from the optical forbidden Car S₁ state of carotenoids. Since carotenoids do not fluorescence themselves, their Car S₁ TPE spectrum is typically recorded using the chlorophyll or tetrapyrrole fluorescence observed after Car S₁ → Chl or Car S₁ → tetrapyrrole energy transfer.

However, this possibility was used by Bode et al. to introduce a coupling parameter $ΦCouplingCar S1−Chl$ that basically allows us to quantify differences of Car S₁ → Chl energy transfer of different pigment–protein complexes and even during the light-harvesting regulation in living plants. The idea is simple: If the Car S₁ → Chl energy transfer changes or is different in distinct complexes, then the amount of chlorophyll fluorescence observed after Car S₁ TPE varies accordingly. To correct for changes in the chlorophyll fluorescence quantum efficiency itself (which occurs, for example, during regulation of light-harvesting) as well as contributions of direct chlorophyll two-photon excitation, the parameter is normalized by dividing the chlorophyll fluorescence observed after TPE, F^TPE, by the fluorescence observed after direct chlorophyll OPE, F^OPE,

This parameter is unitless and does not provide absolute Car S₁ → Chl energy transfer quantum efficiencies, $ΦETCar S1→Chl$⁠, but allows direct comparisons of the amount in Car S₁ → Chl energy transfer of different samples measured under identical conditions.

For example, if an LHC II complex with a certain carotenoid composition has higher Car S₁ → Chl energy transfer efficiency than another LHC II complex with a different carotenoid composition, then also the experimentally observed $ΦCouplingCar S1−Chl$ will be higher for the former. Likewise, if the Car S₁ → Chl energy transfer efficiency increases or decreases in a plant, for example, due to regulation, then also $ΦCouplingCar S1−Chl$ will increase or decrease accordingly. Due to the normalization by F^OPE, this is still the case, even if the fluorescence quantum efficiency changes or if there is significant contribution from direct Chl TPE. More details can be found in a brief explanation in the supplementary material of Ref. 14.

However, this approach clearly requires the ability for two-photon as well as one-photon experiments of the same samples under close to identical conditions. Tedious calibration procedures and controls are necessary for this. For example, similar OPE and TPE excitation probabilities as well as transition energies should be used, and of course, the very same spots of the samples should be observed at the same time. If the OPE and TPE excitation probabilities—i.e., the fraction of molecules that are in the excited state—are very different using both excitation methods, then biases due to, for example, different saturations cannot be excluded. This, of course, is not easy to control as the excitation conditions of OPE and TPE are quite different.

Even though $ΦCouplingCar S1−Chl$ intrinsically corrects for direct chlorophyll TPE contributions, there had been discussions in the past about the extent of direct chlorophyll contributions in TPE spectra.

To address this question quantitatively, Liao et al.²⁴ and Gacek et al.¹⁴ explored 1:1 carotenoid–tetrapyrrole dyads for which it was known that there is a significant Car S₁ → tetrapyrrole energy transfer. The comparison of TPE spectra observed from these 1:1 dyads with TPE spectra of the tetrapyrroles only as well as Chl a and Chl b at identical concentrations allowed to determine the contributions of a single carotenoid in these spectra in direct comparison to a single Chl a or Chl b molecule. These data demonstrated that the carotenoids investigated in this study had a larger TPE contribution than Chl a, TPE_{Chl a}(λ), in a very wide spectral range [compare Fig. 1(a) with Fig. 1(b)]. It is important to note that this experimental observation only provides a lower limit of a single carotenoids TPA cross section compared to that of a single chlorophyll molecule, as the carotenoid contribution in TPE spectra depends on the amount of Car S₁ → Chl or Car S₁ → tetrapyrrole energy transfer. Remember that carotenoid TPE can only be detected via Chl or tetrapyrrole fluorescence observed after Car S₁ → Chl or Car S₁ → tetrapyrrole energy transfer. Assuming, for example, that the Car S₁ → tetrapyrrole energy transfer in the dyads was 30% would mean that the actual TPA cross section of a single carotenoid is by a factor of 1/0.3 larger compared to the chlorophylls or tetrapyrrole two-photon cross sections than observed in the TPE spectra. This is indicated by the upper, opaque red data in Fig. 1(a) that represent the carotenoids two-photon cross sections compared to that of Chl a [Fig. 1(b)] when assuming that the Car S₁ → tetrapyrrole energy transfer in the dyads was 30%.

In summary, the spectra TPE_Car(λ), TPE_Chl a(λ), and TPE_Chl b(λ) shown in Figs. 1(a)–1(c) reflect the relative contributions of a single carotenoid, Chl a or Chl b molecule in two-photon excitation spectra. In two-photon excitation spectra, carotenoid contributions are diminished when there is less than 100% Car S₁ → Chl or Car S₁ → tetrapyrrole energy transfer. When assuming 30% Car S₁ → tetrapyrrole energy transfer in the 1:1 dyads that were used to determine these relative contributions, then the TPA cross sections of a single carotenoid are reflected by the opaque red curves in Fig. 1(a) compared to that of Chl a and b in Figs. 1(b) and 1(c). It also needs to be noted that Car S₁ → tetrapyrrole energy transfer from higher vibronic Car S₁ states might be more effective than from lower vibronic Car S₁ states. This could explain the increasing Car S₁ contributions in experimentally TPE spectra observed at shorter wavelengths—the actual TPA cross sections of carotenoids at longer wavelengths might be potentially higher than reflected in the experimental TPE spectra.

The assignment of spectral features in the two-photon spectra to distinct vibronic transitions is not trivial. Recent results seem to indicate that the Q_x bands or higher vibronic bands of chlorophylls and other tetrapyrroles have higher TPA cross sections than in the corresponding OPE spectra. In addition, the interpretation of carotenoid TPE spectra is more complicated than previously anticipated by most researchers and the mentioned hot state energy transfer might play a role in comparison of carotenoid TPA and TPE spectra. Further clarification should be subject to future studies with additional experiments.

Similar looking data have also been observed by the z-scan technique³⁵ that actually provides TPA spectra instead of TPE spectra. There seem to be shifts between TPA and TPE data, and this might indeed be due to differently energy transfer efficiencies from hot carotenoid states, but as mentioned above, this requires future studies.

As all data presented here are based only on the detection of fluorescence from the lowest Chl Qy band, it is useful to briefly consider if a part of the excitation could be missed by emission from higher states. However, it is known for Chl a and Chl b that Kashas rule applies pretty strictly as there are no reports so far on chlorophyll fluorescence from states higher than Qy. In addition, for the carotenoids involved in the present study, no violation of the Kashas rule is known, as they virtually show no fluorescence at all. However, as mentioned above, energy transfer from higher states could potentially contribute to the shape of the observed TPE spectra.

In a recent publication,¹³ the apparent discrepancy of different groups claiming that TPE spectra of pigment–protein complexes are either dominated by carotenoid or chlorophyll contributions could be resolved by a very simple consideration: Even though the TPA cross section of a single carotenoid is higher than that of a single Chl a or Chl b molecule [Figs. 1(a)–1(c)], most pigment–protein complexes contain far more chlorophylls than carotenoids and not all carotenoids transfer energy from Car S₁. It turned out that both is correct: TPE spectra of pigment–protein complexes can be dominated by either carotenoid or chlorophyll contribution, depending on the actual TPE wavelength used. For example, in TPE wavelength ranges around 1100/550 nm, there is very little Chl TPE but high Car TPE, whereas TPE spectra have large Chl TPE contributions in wavelength ranges around 1300/650 nm because of the large number of Chls in these complexes.¹³ Even if only one out of the four carotenoids in LHC II would transfer Car S₁ → chlorophyll energy with a high efficiency, changes in the direction of this pairwise energy transfer could be physiological very important and still be detected by $ΦCouplingCar S1−Chl$ albeit the contribution to the TPE spectrum would be rather small in certain wavelength ranges.

For example, LHC II contains 4 Car, 6 Chl b, and 8 Chl a molecules (Fig. 1). Summing the TPE spectra of 4 Cars [Fig. 1(d)],

and comparing that to the sum of the TPE spectra of 6 Chl b and 8 Chl a [Fig. 1(e)],

demonstrate that the high number of Chls results in a significant Chl contribution even though a single Chl a molecule has a significant smaller contribution than a single Car molecule [compare Figs. 1(a) and 1(b)].

Fortunately, the finding that—depending on the wavelength range—TPE spectra contain information about carotenoid and chlorophyll contributions opens now a new, much easier way for the determination of $ΦCouplingCar S1−Chl$ than the previously more demanding way to derive information about carotenoid and chlorophyll contributions from separate TPE and OPE experiments.

This is illustrated in Fig. 1(f). The black data are an experimentally observed TPE spectrum of LHC II. If there are higher contributions from Car S₁ → Chl energy transfer in a certain complex, then the corresponding contribution of the 4 Cars [Fig. 1(d)], f_car, should be higher in the LHC II TPE spectrum [Fig. 1(f)]; if it is lower, the 4 Cars would contribute correspondingly less. Thus, when fitting both the contribution of the 4 Car spectrum [Fig. 1(d)], f_car, and the contribution of the 6 Chl b + 8 Chl a spectrum [Fig. 1(e)], f_Chl, to the experimental LHC II spectrum,

the resulting Car contribution, f_Car, compared to the Chl contribution, f_Chl, also reflects the amount of Car S₁ → Chl energy transfer,

In Fig. 1(f), for example, a fit with f_Car = 57% of the 4 Car spectrum in Fig. 1(d) and f_Chl = 100% of the 6 Chl b + 8 Chl a spectrum (brown) has the best agreement with the experimentally observed LHC II spectrum (black). This means that in LHC II, the Car S₁ → Chl energy transfer quantum efficiency is 57% compared to that observed in 1:1 Car–tetrapyrrole dyads from which the Car spectra in Figs. 1(a) and 1(d) were derived. Similarly, as $ΦCouplingCar S1−Chl$ values obtained from a comparison of TPE and OPE [Eq. (1)], also the $ΦCouplingCar S1−Chl$ values determined by analyzing contributions of Car TPE, f_Car, and Chl TPE f_Chl, to experimental TPE of pigment–protein complexes [Eq. (2)] do only allow relative comparisons of the Car S₁ → Chl or Car S₁ → tetrapyrrole energy transfer efficiencies between different samples (in this case, between the 1:1 Car–tetrapyrrole dyad and LHC II). However, it will also allow us to monitor relative changes in the Car S₁ → Chl energy transfer in a single sample, for example, during light-harvesting regulation in plants. However, experimentally it is much easier to measure single TPE spectra under identical conditions than comparing TPE and OPE data of the same sample.

In Fig. 1(f), the analysis was done with a Car spectrum derived from a Car–tetrapyrrole 1:1 dyad that contained a carotenoid with 8 conjugated double bonds (8DB). However, LHC II contains various different carotenoids with different numbers in conjugated double bonds, partly depending on a plants current adaption status (xanthophyll cycle). We, therefore, repeated the analysis also with a Car TPE spectra derived from a 1:1 Car–tetrapyrrole dyad that contained a carotenoid with 10 conjugated double bonds (10DB). In this case, it was reported earlier²¹ that a 10DB carotenoid was still capable of significant Car S₁ → tetrapyrrole energy transfer while using carotenoids with 11 DB did not show sufficient Car S₁ → tetrapyrrole energy transfer anymore. In addition, we tested the analysis using an LHC II spectrum reported in Refs. 5, 37, and 13 as well as different Chl a and Chl b TPE spectra reported in Refs. 5, 13, and 14 (Fig. 2). The Chl TPE spectra reported in Ref. 14 were observed using acetone solutions for Chl a and Chl b, whereas the Chl TPE spectra reported in Ref. 5 were observed using acetonitrile/ethyl acetate/water as solvent. Figure 2 demonstrates that independent from different experimental data, the agreement of the spectra fitted analogously to the procedure shown in Fig. 1 is quite well and that the values for $ΦCouplingCar S1−Chl∝fCarfChl$ observed with different experimental data differ only in a range of 54% ± 7% (Table I). Be reminded again that these values are values relative to the energy transfer quantum efficiencies in 1:1 Car–tetrapyrrole dyads from which the Car contributions to TPE spectra [Fig. 1(a)] were derived from. These energy transfer quantum efficiencies in 1:1 Car–tetrapyrrole dyads are not known with high precision but are likely not 100%. The slight blue shift of the fitted spectra is likely due to different environment of the pigments in solution compared to the protein environment in LHC II.

Determining the contributions of Car TPE, f_Car, and of Chl TPE, f_Chl, from experimental TPE of pigment–protein complexes principally requires even only the measurement at two different TPE wavelengths, for example, in spectral regions that are either dominated by Cars or Chls. When the TPE spectrum of a sample is only measured at two wavelengths, λ₁ and λ₂, Eq. (4) becomes

From this, the contributions, f_Car and f_Chl, can be directly calculated as

For example, when using λ₁ = 1050 nm and λ₂ = 1200 nm, the corresponding values are in a range of $ΦCouplingCar S1−Chl$ = 58 ± 13, which is in good agreement with the value observed fitting the whole spectra (Table I). Clearly, using only two TPE wavelengths provides less accurate results for $ΦCouplingCar S1−Chl$⁠, as also indicated by the larger error margin, but obviously this simpler approach still provides reasonable results when choosing appropriate wavelengths. When there are shifts of the spectra caused, for example, by coupling or environment, then this approach will be also less accurate than using the entire spectra for the analysis. Ideally, the two wavelengths are chosen to reflect spectral ranges that are either dominated by Car or Chl signals. In addition, when using the average of smaller wavelength ranges, such as λ₁ = 1070 ± 20 nm and λ₂ = 1320 ± 20 nm, reasonable agreement with the values observed from fitting entire spectra [Eq. (4)] is observed (Table I). The measurement with only two TPE wavelengths, λ₁ and λ₂, opens the possibility of experiments in which fast changes in $ΦCouplingCar S1−Chl$ could be monitored. This is of particular interest for measurements of $ΦCouplingCar S1−Chl$ during the regulation of photosynthetic organisms and will be subject of future studies.

To compare the $ΦCouplingCar S1−Chl$ values from LHC II with a different complex, we did an analogous analysis also using known experimental TPE data from PS 1. Advantageously, for PS 1, even an absolute Car S₁ → Chl energy transfer quantum efficiency of $ΦETCar S1→Chl$ ∼ 47% could be derived from kinetic two-photon pump–probe signals that were observed in the spectral region of the strong Car S₁–Car S_n transient absorption and that were directly compared with that of β-carotene alone (Fig. 5 from Ref. 38). As detailed below, this can be used to estimate also a corresponding absolute value for LHC II considering the relative $ΦCouplingCar S1−Chl$ values of both complexes.

PS 1 contains 22 β-carotene molecules and 96 Chl a molecules.¹² Thus, fitting the fractions of Car and Chl contributions to experimental PS 1 TPE spectra analogously to the procedure described by Fig. 1 and Eqs. (2)–(4) needs to be done using the following equations:

When using only two TPE wavelengths, λ₁ and λ₂, the analogous equations are

Figure 3 and Table II show the corresponding fits and values for PS 1. The fact that the experimental PS 1 TPE spectrum contains less structural features than the fitted ones can be explained by the fact that the spectrum used for fitting the Chl contribution is based on Chl a spectra in solution only [Fig. 1(b)] that are not subject to any heterogeneous broadening present in the very different protein environments of the Chl molecules in the complex. Fitting the whole spectra resulted in a slightly higher average value of $ΦCouplingCar S1−Chl=$ 64 ± 13 than the average value of $ΦCouplingCar S1−Chl=$ 54 ± 7 observed for LHC II. Likely, this is because PS 1 only contains Chl a, which has a lower Qy energy than Chl b. This is potentially advantageous for Car S₁ → Chl energy transfer. For PS 1, it has to be considered that no data were available below 1100 nm, which is a spectral range dominated by Car TPE. However, at 1100 nm, there are still rather large Car and rather little Chl a TPE contributions. Therefore, the values observed with fitting the whole spectra should be still similarly accurate as the ones observed for LHC II. Only when using values as large as λ₂ = 1300 nm in the two-wavelength approach [Eqs. (13) and (14)], the error margins become quite large for the PS 1 values, which is likely due to the low signal in that spectral region. Once more, it is important to note that these $ΦCouplingCar S1−Chl$ values are values relative to the energy transfer quantum efficiencies in 1:1 Car–tetrapyrrole dyads from which the Car contributions to TPE spectra [Fig. 1(a)] were derived from. These are not known with high certainty. However, as mentioned above, the absolute Car S₁ → Chl energy transfer quantum efficiency of PS 1 could be estimated from two-photon pump–probe amplitudes to be $ΦETCar S1→Chl$ ∼ 47%.³⁸ 

The obtained values of $ΦCouplingCar S1−Chl=$ 54 ± 7 for LHC II and of $ΦCouplingCar S1−Chl=$ 64 ± 13 for PS 1 indicate that the absolute Car S₁ → Chl energy transfer quantum efficiency in LHCII is about 16% smaller than that of PS 1. Given the reported efficiency value of $ΦETCar S1→Chl$ ∼ 47% for PS 1, this indicates an absolute value for the Car S₁ → Chl energy transfer quantum efficiency in LHC II of 40%, which is in reasonable agreement with previous estimates. However, further studies are necessary because the $ΦCouplingCar S1−Chl$ value for PS 1 is derived from a TPE spectrum that misses important parts in the blue range.

In summary, the new proposed approach allows determining the extent in Car S₁ → Chl energy transfer only from varying Car and Chl contributions in TPE spectra without any comparison between different OPE and TPE experiments. Thus, the careful calibration procedures necessary for comparing results from these different types of TPE/OPE experiments are not necessary anymore, and the approach likely provides a much easier and quicker access to $ΦCouplingCar S1−Chl$ in photosynthetic samples than previous methods. However, future studies are necessary to further assess the accuracy of the proposed fitting procedure, for example, by comparison with the results of the previous TPE/OPE method. If successful, the suggested approach will be much more flexible and will allow measuring also very dynamic changes in the Car S₁ → Chl energy transfer quantum efficiency, such as that observed in the regulation of living photosynthetic complexes. This would enable many more studies of various conditions, mutants, and organisms in a routine fashion. It would also allow to better estimate absolute Car S₁ → Chl energy transfer quantum efficiencies by comparison of the fitted $ΦCouplingCar S1−Chl$ with that of complexes, for which the absolute Car S₁ → Chl energy transfer quantum efficiency has been reported by other methods, such as the $ΦETCar S1→Chl$ value derived for PS 1 from kinetic two-photon pump–probe data and direct comparison to β-carotene data (Fig. 5 from Ref. 38).

There are no new experimental data in this Communication. The main topic of this Communication is the proposal of a new and simpler way to analyze Car S₁ → Chl energy transfer based on recent findings that signals from carotenoids as well as chlorophylls can be extracted from two-photon spectra of photosynthetic pigment–protein complexes and organisms. Therefore, the following experimental procedures describe how the results shown here to illustrate this new proposal have been obtained in previous work.

Two laser systems were used for the TPE spectra of Gacek et al. For the range of 950–1060 nm, a Chameleon Ultra II was directly fed into the confocal microscope. From 1050 to 1400 nm, an optical parametric oscillator (IR OPO) driven by the Chameleon Ultra II, 80 MHz was used. An electron multiplying charge coupled device camera (EMCCD) was used for detection of the fluorescence (iXonEm+ 897 black-illuminated, Andor Technology).¹⁴ 

Measurements by Hilbert et al. were done with a Femtolite-A-10-fiberlaser (IMRA, 48 MHz, 16.1 mW, 783 nm pulses) and a home-build Ti:Sa amplifier pumped by a ND:YLF laser (Quandronix model 527 DP-H, 527 nm pulsed of 6.5 mJ, repetition rate 1 kHz). To vary the wavelengths a traveling-wave optical parametric superflourescence (TOPAS) was pumped by the amplifier (1 kHz, 700 mW, 780 nm). The fluorescence was detected by an AT200 CCD camera Photometrics.¹⁶ 

TPE measurements by Betke et al. were performed with a Ti:Sapphire regenerative femtosecond amplifier (Coherent Legend Elite) pumped by a Coherent Mira 900 fs oscillator (800 nm, 1 kHz). 1.2 W optical pump to power a wavelength tunable optical parametric amplifier (OPA) (Coherent OperA Solo) was provided by the regenerative laser amplifier. The emission was detected with a photo-multiplier tube.^5,13

LHC II by Walla et al. was measured with a Coherent 9450 OPA seeded by a Coherent RegA900 and Mira Seed Ti:sapphire oscillator with model 9150 stretcher/compressor (pulse width ∼85 fs). The spectrum was detected with a photomultiplier tube (Hamamatsu R928 red-extended).³⁷ 

This work was supported by generous grants from the German science foundation [Deutsche Forschungsgemeinschaft (DFG) (Grant Nos. INST 188/334-1 FUGG, GRK2223, and INST 188/334-1 FUGG)].

The authors have no conflicts to disclose.

The data that support the findings of this study are available from the corresponding author upon reasonable request.