How electron tunneling and uphill excitation energy transfer support photochemistry in Halomicronema hongdechloris

Halomicronema hongdechloris, the first cyanobacterium reported to produce the red-shifted chlorophyll f (Chl f ) upon acclimation to far-red light, demonstrates remarkable adaptability to diverse light conditions. The photosystem II (PS II) of this organism undergoes reversible changes in its Chl f content, with levels ranging from practically zero under white-light culture conditions to a Chl f :Chl a ratio of up to 1:8 when exposed to far-red light (FRL) in the 720-730 nm range for several days. Our ps time-and wavelength-resolved ﬂ uorescence data obtained after excitation of living H. hongdechloris cells indicate that the Soret band of a far-red (FR) chlorophyll involved in charge separation absorbs at 470 nm. At 10 K, the fluorescence decay at 715-720 nm is still fast with a time constant of 165 ps indicating an efficient electron tunneling process. However, additionally, there is efficient excitation energy transfer (EET) from 715-720 nm to 745 nm with the latter resulting from FR Chl f , which mainly functions as light-harvesting pigment upon adaptation to FRL. From there, excitation energy is efficiently transferred towards the primary donor in the reaction center of PS II with an energetic uphill EET mechanism inducing charge transfer. The fluorescence data is well explained with a secondary donor P D1 represented by a red-shifted Chl a molecule with characteristic fluorescence around 715 nm and a more red-shifted FR Chl f with fluorescence around 725 nm as primary donor at the Chl D1 or P D2 position


Introduction
Chlorophyll f (Chl f) is the most red-shifted chlorophyll molecule discovered in nature to date.
It was initially discovered in 2010 in the cyanobacterium Halomicronema hongdechloris (Chen et al. 2010;Chen et al. 2012).Chl f differs from Chl a by a formyl group present at the C-2 position, making it [2-formyl]-Chl a (Chen et al. 2010;Willows et al. 2013).Similar to other species like Acaryochloris marina, H. hongdechloris employs a chemically unique pigment to facilitate adaptation to far-red light conditions (Chen et al. 2010;Chen et al. 2012;Willows et al. 2013;Allakhverdiev et al. 2016;Schmitt et al. 2019;Friedrich and Schmitt 2021;Schmitt et al. 2021).Employment of red-shifted Chl species most likely reflects evolutionary adaption to the light conditions in the local environment (Ueno et al., 2019).This is well known for A. marina, which is adapted to an ecological niche that is abundant in far-red (FR) and nearinfrared light, accumulating up to 99 % of its total chlorophyll content as Chl d.This substitution of Chl a with Chl d in A. marina's photosystems extends the spectrum of light that can be harnessed for oxygenic photosynthesis and carbon fixation by the organism to the near infrared region (Mielke et al. 2011).Chl d mainly functions as a light-harvesting pigment in A. marina with an absorption maximum at 714 -718 nm for PS II in living cells (Miyashita et al. 1996, Marquardt et al. 1997, Marquardt et al. 2000, Mimuro et al. 1999, Mimuro et al. 2000, Mimuro et al. 2004, Chen et al. 2005, Petrasek et al. 2005, Schmitt et al. 2006, Schmitt 2011, Theiss et al. 2011).
However, to this day there is still a controversial debate about the nature of the primary and secondary donor in the PS II reaction center (RC) of H. hongdechloris.This accounts not only for the chemical nature, but also for the structural position of the involved chlorins, since recent studies discuss either Chl a, Chl d or Chl f as possible candidates for the four central donor side chlorins in the PS II RC, which are PD1, PD2 ChlD1 and ChlD2 (Nurnberg et al. 2018;Zamzam et al. 2020;Judd et al. 2020).It is proposed that the composition of PS II in H. hongdechloris is similar to Chroococcidiopsis thermalis which contains four Chl f, one Chl d (Judd et al. 2020) and about 30 Chl a molecules, while it seems to be clear that the primary donor is formed by a red-absorbing pigment with a maximum of the QY absorbance band at 726 nm (Zamzam et al. 2020;Judd et al. 2020).Zamzam et al. (2020) studied two different purified, isolated PS II core complexes.For the white light (WL) measurements, these authors used Thermosynechococcus elongatus, whereas Chroococcidiopsis thermalis was used for the FRL measurements.Based on their results, Zamzam and coworkers established a model for the central reaction center, in which the pigments of the four Chl molecules (PD1, PD2, ChlD1, and ChlD2) and the two pheophytin molecules (PheoD1 and PheoD2) were varied to understand the changes induced by exposing the cells to FRL.Zamzam et al. arrived at the conclusion that the actually available data cannot finally settle the discussion about the composition of the PS II in Chroococcidiopsis thermalis.Therefore, it is still not decided, which molecule at which position exactly is the primary electron donor but it is proposed that it is either ChlD1 or PD2 (Zamzam et al. 2020;Judd et al. 2020) (For the nomenclature of the pigments see also Fig. 7).
In the PS I reaction center, the primary donor is formed by FR Chl d in A. marina (Itoh et al. 2001;Itoh et al. 2007;Tomo et al. 2008;Schenderlein et al. 2008).The nature of the primary donor of PS II in A. marina is most probably a Chl a/Chl d heterodimer as proven by the electrochromic shift introduced by a charged Chl d cation (Schlodder et al. 2007) or possibly a Chl d homodimer (Tomo et al. 2007).Therefore, the idea of functionally highly distinct and specialized heterodimers is appealing to understand the photochemistry together with a possible uphill excitation energy transfer (EET) in far-red light-absorbing species.
Based on observations of cyanobacterial core complexes and eukaryotic PS I/LHC systems, it appears that chlorophylls with a red-shifted absorbance spectrum, which are generated within the antenna system, retard the flow of energy to the reaction centers by acting as energy traps that transiently store energy on the nanoscale and over nanoseconds in time.The energy transfer process is kinetically adjusted by involving energetically uphill-directed processes, until the excitation energy ultimately reaches the special pair, where charge separation is induced.(Jennings et al. 2003;Croce et al. 2000;Gobets and van Grondelle 2001;Trissl 1993;Allakhverdiev et al. 2016).Thermally activated energy transfer from red spectral states to bulk Chls in the PS I/LHC system from maize thylakoids was indeed observed as a possible ratelimiting step for charge separation (Croce et al. 2000), with activation energies between 7 and 13 kJ•mol -1 complying with the Arrhenius-Eyring theory (Jennings et al. 2003).In an algal Ostreobium species, which can perform assimilation under light above 700 nm and is known for its high content of red Chl a molecules, direct evidence for uphill energy transfer from red Chls to the special pair was observed in the PS II (Wilhelm and Jakob 2006).In chloroplasts, uphill energy transfer can be detected through blue-shifted anti-Stokes fluorescence, even when the shift is greater than 100 nm, as demonstrated in a study utilizing 800 nm excitation (Hasegawa et al. 2011).
The strongly temperature-dependent anti-Stokes fluorescence following the Arrhenius-Eyring theory can qualitatively explain uphill EET in the PS II of H. hongdechloris (Schmitt at al. 2019).As "thermal" uphill energy transfer is strongly temperature-dependent, it can even be used for the determination of the local temperature (Hasegawa et al. 2011;Schmitt et al. 2019;Friedrich and Schmitt 2021).
Previous investigations showed that EET occurs from red-shifted Chl f with an absorption maximum of up to 800 nm to the primary donor in the PS I of H. hongdechloris, which is most probably composed of a Chl a/Chl a' heterodimer with an absorption maximum at 704 nm in FRL-adapted cells (Kurashov et al. 2019).We had shown that the uphill energy transfer, which can bridge an energy gap of up to 13 kJ•mol -1 in H. hongdechloris, is additionally supported by an entropy gain, since few Chl f molecules (about 12.5 % of total Chl content under FRL) are energetically coupled to a large pool of Chl a molecules, which might explain the efficient harvesting of red-shifted light at 745 nm.745 nm light is lower in energy as compared to the FR light absorbed by A.marina, which fills its PS II complexes by more than 90 % with Chl d.
For H. hongdechloris, the limited amount of Chl f is a direct prerequisite for the efficiency of the process of uphill energy transfer promoted by an entropy gain.The entropy effect in H. hongdechloris reduces the free energy gap between Chl f and Chl a during the endothermal EET process (Schmitt at al. 2019;Friedrich and Schmitt 2021).Recent studies on the molecular composition of PS I have confirmed the function of selected Chl f molecules as FRL absorbers in the periphery of PS I, which efficiently transfer their energy to the Chl a/Chl a' heterodimer (Kato et al. 2020).It is reasonable to postulate that both photosystems should be able to function in a balanced manner under far-red light (FRL) conditions to effectively drive carbon fixation (Friedrich and Schmitt 2021).
Following-up on our former studies, we further investigated H. hongdechloris and found that the species shows an adaptation mechanism of the phycobilisomes (PBS) under changing light conditions in the form of fast state transitions of PBS between different PS II complexes with different pigment composition (Ho et al. 2020;Schmitt et al. 2021).After FRL acclimation, the PBS of H. hongdechloris are found to be localized in discrete clusters or functionally associated primarily with photosystem II (PS II) containing Chl a. Upon brief exposure to white light, blue light (405 nm), or red light (630 nm), the PBS undergo a rapid mobilization, decouple from Chl a-containing PS II for a few seconds and recouple with Chl f-containing PS II, thereby enabling an effective energy transfer directly from the PBS to Chl f (schmitt et al., 2021).The presence of such a heterogeneity regarding PS II complexes containing either Chl a and Chl f or Chl a only strongly suggests that the structure of at least one or two molecules of the secondary and possibly also of the primary donor at the PD1 PD2 and ChlD1 position, respectively, is based on Chl a after FRL adaptionat least in a fraction of PS II reaction centers and only one or two molecules are substituted by Chl f just in a fraction of the reaction centers (Blankenship et al., 2019).Zamzam et al. (2020) recently proposed that the secondary donor in the reaction center of PS II in Chroococcidiopsis thermalis might consist of red-shifted Chl a at the PD1 position while the primary donor is a FR chlorophyll at the ChlD1 position, which is either a Chl d or a Chl f.This was concluded after observing an electrochromic shift of 21 ± 4 cm -1 due to formation of the semiquinone anion, QA •-.From the wavelength (726 nm) it was concluded that the primary donor (PD) is a FR chlorophyll; however, the true chemical nature of the PD is still elusive.Judd et al. (2020), observed the same electrochromic and added that the PD might also be located at the PD2 position.We assume that the PS II of Chroococcidiopsis thermalis is possibly similar to H. hongdechloris but not necessarily identical as no evidence for Chl d in the PS II of H. hongdechloris was found so far.
In our studies using UV-vis absorption spectroscopy, evidence of the formation of Chl f was confirmed in H. hongdechloris, and time-integrated and time-resolved fluorescence spectroscopy were used to calculate decay-associated spectra (DAS) to monitor EET processes between photosynthetic pigments in intact H. hongdechloris cells (Schmitt et al. 2019;Friedrich and Schmitt 2021;Schmitt et al. 2021).
According to Zamzam et al. (2020), the primary electron donor is named P720 and claimed to be a FRL Chl f (in case of H.hongdechloris we exclude Chl d) attached to the ChlD1 position in the D1 subunit in the heterodimer of the reaction center of PS II.The secondary electron donor is assumed to be a red shifted Chl a in the PD1 position with slightly shorter emission wavelength (see also Fig. 7).Electron transfer between the Chl a and the FRL-absorbing chlorophyll takes place within a few picoseconds; therefore, both pigments are strongly coupled and EET occurs efficiently from PD1 to ChlD1 at room temperature.For our simulations we set this value to 50 ps which is the resolution limit of our measurement technique.According to Zamzam et al. (2020) the excited state of FRL-absorbing Chl d/Chl f (P720) at ChlD1 is energetically lower than Chl a in Chroococcidiopsis thermalis, and there is no backward electron transfer assummed to be possible to the Chl a at PD1.The excited state is, therefore, localized at P720.However, the authors showed that FRL-absorbing Chl d/Chl f not only occurs as a pigment in the antenna, but is directly involved in primary charge separation.
We propose that the localization of the excited state to P720 represents an optimum between the high-energy state of many Chl a molecules and the low-energy state of a few Chl f molecules.The picture is similar to an "uphill" EET supported by entropy, but it does not involve direct EET.More likely, it represents an energetic distribution of the electron wave function that gives rise to a charge separation from an energetically lower state, which in turn induces electron transfer (ET) into an energetically higher state.We follow this picture assuming that it is essential to bear in mind the fact that wave functions are extended, can overlap, and can give rise to electron tunneling processes to "overcome" or "penetrate" energetic barriers.
In such a sense, ET can occur via a Chl a in the PD1 position, where Chl a is practically involved as a supporting molecule with an energetic state between P680 and P720.The spectral position of the Chl a in the PD1 position is adjusted at the energetic optimum for efficient charge separation, ET to Pheo and subsequent stabilization by ET to QA.
Our low-temperature measurements indicate that an electron tunneling process with a time constant of about 165 ps supports ET to Pheo.This is inferred from the temperatureindependence of the corresponding rate constant, which is generally not accounted for by the Arrhenius-Ayring theory of reaction rate constants.Therefore, we reach beyond this theory and our previous considerations by including an electron tunneling mechanism into the process of charge separation, which might help to understand the efficient functional design of PS II in far-red light-adapted organisms.

Materials and Methods
Cell culture.Halomicronema hongdechloris cultures were kept at room temperature (25 °C) in 50 mL Erlenmeyer flasks containing KES seawater medium under constant shaking, as described (Tomo et al. 2014;Chen et al. 2012) and were grown under far-red light (FRL) by illuminating culture flasks by a circular array of 720 nm LEDs.Far-red light intensity was adjusted to 10 µE•m -2 •s -1 , since at intensities exceeding this value, the cultures bleached rapidly, as described (Li et al. 2014).For reference, cells were grown under white light (WL) with an intensity of 30 µE•m -2 •s -1 .White light was obtained from "warm white natural neon tube" resembling a temperature spectrum of 3000 K (Osram, Munich, Germany).
Absorption spectra were recorded on methanolic total pigment extracts prepared from intact WL-or FRL-adapted H. hongdechloris filament bundles in a 1-cm quartz cuvette between 300 nm and 800 nm with a UV-1800 spectrophotometer (Shimadzu, Berlin, Germany) at room temperature (298 K).
Fluorescence emission a spectra were recorded on intact H. hongdechloris filament bundles at room temperature in KES medium.For this purpose, cell colonies were placed directly in the middle of regular 1-cm quartz cuvettes with 1 mm optical path length.The cells were placed inside the excitation light beam and adjusted to maximize the fluorescence signal.The fluorescence and excitation spectra were measured with a Fluoromax-2 spectrofluorometer (Horiba Jobin Yvon, Bensheim, Germany).The cell filaments were excited at different wavelengths using a Xenon lamp (Osram XBO 100, Carl Zeiss, Germany) as light source and wavelength selection with a monochromator, while emission spectra were recorded between 600-800 nm during excitation with 430 nm and 470 nm at room temperature.Slits were set to 4 nm for excitation and 4 nm for emission, while the integration time was 1 s and the increment 1 nm.Excitation spectra were recorded observing the fluorescence at 710 and 730 nm from 400 nm up to 10 nm below the corresponding observation wavelength.

Time-and wavelength-correlated single photon counting (TWCSPC) Spectrally resolved
fluorescence decay curves were recorded on intact H. hongdechloris filament bundles at room temperature in KES medium.For this purpose, cell colonies were transferred into the ca. 1 mm diameter tip of a Pasteur pipette, which served as spectroscopic cuvette placed in the laser focus with a multi-axis translation stage.Measurements were performed employing a Hamamatsu R5900 16-channel multi-anode photomultiplier tube (PMT) with 16 separate output (anode) elements and a common cathode and dynode system (PML-16C, Becker&Hickl, Berlin, Germany) as described in (Schmitt et al. 2019).A 470 nm pulsed laser diode (LHD-470, Picoquant, Berlin) delivering 80 ps full-width at half-maximum (FWHM) pulses at a repetition rate of 20 MHz was used for excitation.The fluorescence was observed via a 488 nm longpass filter (AHF Analysentechnik, Tübingen, Germany).

Calculation of the decay-associated spectra (DAS) and decay associated yield spectra (DAYS).
Decay associated spectra (DAS) were determined by fitting a theoretical model for the time resolved fluorescence decay F(t) for each measured wavelength section i (i=1...16).This model function is a multiexponential decay (up to four components) with individual amplitude prefactors Aj(i) (j=1,..,4) for each measured wavelength section but common (global) fluorescence decay times j (j=1,..,4) for the whole spectral range: As the measured data is a convolution of the system´s instrumental response function, IRF, with the ideal data model the fit function FIT denotes to The wavelength dependent Aj(i) denote the amplitudes for the decay times j (j=1,..,4).A Levenberg-Marquardt algorithm was used to minimize the difference between the model and the data, i.e., an error functional based on the reduced Chi square, χ r 2 (λ), for each wavelength channel.The IRF was obtained by measuring the response of the detection system to scattered excitation laser pulses.The modelled exponential decay (two exponential components) was convolved with the IRF averaged over all time channels t´ as described in form of a continuous time axis in eq. ( 2) before the comparison with the measured data.The calculated χ r 2 (λ) was minimized by varying the model input parameters until convergence was reached.For this calculation, the software of Globals Unlimited® (University of Illinois, Urbana, USA) was used.The fluorescence lifetimes, j, were considered global variables while the amplitude of each decay component, Aj(i), was considered to be independent and a function of the emission wavelength.The result of this analysis is visualized by plotting Aj(i) for each lifetime j which represents the DAS that reveal the spectral distribution of individual decay components.DAS can therefore provide information on different dynamic processes such as direct fluorescence decay of spectrally overlapping states and contributions of reabsorbed light energy.
The determination of the DAYS follows the DAS as described in detail in (Schmitt 2011) by multiplying the spectral shape of the fit amplitudes of all decay components in the DAS with the corresponding time constant of the decay component.In contrast to the DAS the amplitudes represented in the DAYS are therefore proportional to the overall signal amplitude (registered number of photons) that contributes to that component: ( )  -720 nm appears, which was absent in all WL-grown samples, and is attributed to the formation of Chl f (see green arrow 3 in Fig. 1B) (Chen et al. 2010;Chen et al. 2012;Schmitt et al. 2019;Friedrich andSchmitt 2021, Schmitt et al. 2021).The relative content of the Chl f emission (determined as area of the Chl f absorption at 700-720 nm compared with a single Gaussian fit of the Qy absorption band of Chl a at 665 nm) reaches up to 10 % after 12 days of FRL illumination (Schmitt et al. 2019), but does not seem to grow further (data not shown).The corresponding difference spectrum (Fig. 1B) shows a small positive maximum that peaks at 710 nm in the diference spectrum (Fig. 1B) and is attributed to Chl f.There is a small absorption peak at 470 nm in both, WL and FRL adapted cells (Fig. 1A  Gaussian bands between 700 -800 nm, with the following emission maxima obtained from an optimal fit: 723 ± 26 nm, 739 ± 19 nm, 745 ± 25 nm. Figure 2A shows the absorbance of total methanolic pigment extracts from H. hongdechloris cells, adapted to WL (dashed black line) and adapted to FRL for five days (red line).Besides the maximum at 665 nm, which is attributed to the Qy band of Chl a and was used for normalization, the Chl a Soret bands are localized at 400 -450 nm, but strongly overlap with carotenoid absorption between 400 and 500 nm.A distinct increase in absorption can be observed for the FRL-adapted samples in the 400-500 nm wavelength range due to the formation of Chl f and carotenoids (in particular β-carotene) as noted previously (Chen et al. 2012;Schmitt et al. 2019).
Most probably, the Soret band of FR absorbing Chls in the PS II is found at 470 nm (green arrow 1 in Fig. 2A).
Transient changes in carotenoid content may be a general response to changes in light conditions, not only during FRL acclimation but also during the first days of re-acclimation to WL, and possibly play a protective role, since carotenoids are efficient quenchers of ROS species and wellknown non-photochemical quenchers of excited singlet and triplet states in phycobilisome pigments (Schmitt at al. 2014;Schmitt and Allakhverdiev 2017;Maksimov et al. 2014;Maksimov et al. 2015;Maksimov et al. 2016).
Figure 2A shows a clearly resolved shoulder in the absorbance spectrum around 710 nm indicating Chl f production (green arrow 2 in Fig. 2A).The shoulder arises after two days of cultivation under FRL and continuously grows up to the mentioned value.Room temperature fluorescence emission spectra upon excitation of Chl a and Chl f with 430 nm light measured from WL-adapted H. hongdechloris cells and from cell samples adapted to FRL for five days are shown in Figure 2B (dashed black and red curve, respectively).The FRL-illuminated samples represent a superposition of Gaussian bands centered at 650 nm, 686 nm, 715 nm, 739 nm, and 745 nm according to our former studies (Schmitt et al. 2019).The fluorescence spectrum obtained after excitation with 470 nm (Fig. 2C) indicates the presence of an additional Gaussian band at 723 ± 26 nm (see green arrow 2 in Fig. 2C and red curve in Fig. 2D) upon fitting the long wavelength emission with three Gaussian bands (see Fig. 2D).The arrows in Fig. 2C indicate the reduction in fluorescence occurring at 650 nm (see green arrow 1) for FRL-adapted cells, while fluorescence at 745 nm increases (see green arrow 3).The fluorescence spectra are normalized to the amplitude at 686 nm, because this represents the Chl a emission peak, and Chl a is the most stable ("invariant") pigment over time under all cultivation conditions.
The peak at 650 nm is attributed to PBS emission, the one at 686 nm to Chl a emission, and in the long-wavelength regime at 715 nm, 723 nm, 739 nm and 745 nm, the signals are dominated by low-energy Chl a species, red-shifted PBS emission, and Chl f, respectively (Majumder et al. 2017;Tomo et al. 2014;Schmitt et al. 2019).These contributions have not been disentangeled yet.Most probably, the bands at 723, 739 and 745 nm are all resulting from FR chlorophylls that have a redshifted Soret absorption band with strong absorption at 470 nm, or they are functionally coupled to such molecules.In addition, direct reduction of the PBS content was observed in FRL-adapted cells indicating that PBSs degrade during FRL adaption down to the size of red light-absorbing APC cores.This even leads to changes of the thylakoid-membrane distance because the smaller PBSs allow for a closer approach of the thylakoids (Li et al. 2014, Majumder et al. 2017, Li et al. 2018).This explains the reduced fluorescence emission attributed to PBSs around 650 nm as indicated in Figure 2B (green arrow "1") which is not even visible after excitation with 470 nm (Fig 2C ), where the PBS absorption is low.The question arises, which emitter is responsible for the fluorescence around 723 nm (region "2" in Fig. 2B).
Thus, at least three distinguishable long-wavelength emitters with remarkably different emission wavelengths of 715 nm (mainly at 430 nm excitation) or 723 nm (mainly at 470 nm excitation), 739 nm and 745 nm are involved in the far-red fluorescence spectra of FRL-adapted H.Only a very minor fraction of far-red fluorescence at 710 nm and 730 nm is excited via the Soret band of Chl a or Chl f in WL adapted samples (see green arrow "1" and "3", respectively, in Fig. 3A and Fig. 3B for the dashed curve).In marked contrast, in FRL-adapted cells (red curves in Fig. 3), the 710 nm fluorescence (Fig. 3A), and especially the 730 nm fluorescence (Fig. 3B) is preferentially excited between 400 -470 nm via the Chl Soret bands (see green arrow "1" and green arrow "3" in Fig. 3 B): There is a small but characteristic peak at 470 nm, at which the Soret bands of red-shifted Chl a and Chl f show stronger absorption.This indicates that the 730 nm fluorescence mainly results from FR chlorophylls, which might be red-shifted Chl a or Chl f .However, the 730 nm fluorescence is also directly excited via the Soret band of Chl a at 430 nm (indicating that it is strongly coupled to the FR Chl).In Figure 3, the difference between the dashed black curve and the red curve is more pronounced for the fluorescence emission at 730 nm (see green arrows in Fig. 3B) as compared to 710 nm (green arrows in Fig. 3A) indicating that mainly the 730 nm emission is stimulated due to the FRL adaptation process.In addition, for both, 710 nm emission and 730 nm emission, the excitation shifts to the red in FRL adapted samples as compared to WL adapted samples when excited via PBSs (550 -650 nm, green arrows "2" in Fig. 3), which is in line with the finding that PBS degrade down to the long wavelength absorbing APC cores in FRL-adapted samples, which are responsible for the peak at 650 nm.
Overall, it is less likely to induce the red fluorescence at 710 nm by direct excitation within the Soret band of Chls.Rather, the 710 nm fluorescence is mainly excited via PBS absorption with an additional peak at 650 nm for APC, while fluorescence at 730 nm is more likely excited via the   4A,C) and 710 ps in FRL-adapted samples (blue triangles in Fig. 4B,D) and a slow one with 1.1 ns in WL-(blue triangles in Fig. 4A,C) and 1.8 ns in FRLadapted samples (green triangles in Fig. 4B,D).Especially the decay associated yield spectra (DAYS) show the components at 650 nm (PBS emission), 686 nm (Chl a) emission, and in the long-wavelength regime at 715-725 nm and 740 nm (Fig. 4C,D).In WL-adapted (Fig. 4A,C) the medium component peaks at 650-680 nm and at 740 nm, while the fast 78 ps component peaks at 715-720 nm (see especially the DAYS, Fig. 4C).This finding indicates that the fastest EET/ET processes occur from a molecule showing fluorescence at 715 nm.In addition, PS I fluorescence around 720-730 nm contributes in, both, WL-and FRL-adapted samples.
In FRL-adapted cells (see DAS in instead of 500 ps in WL-adapted samples (red circles in Fig. 4A,C) .Additionally, in FRL-adapted samples, the fast 170 ps component (red circles in Fig. 4B,D) shows peaks at 685 nm, 720 nm and 745 nm indicating strong coupling between the corresponding states.The subsequent ET from the primary donor and the charge stabilization upon electron transfer to the primary plastoquinone QA are prolongated in FRL-adapted as compared to WL-adapted cells (Zamzam et al., 2020).The longest decay time of 1.1 ns results from PBPs in WL-adapted samples at 650 -680 nm (blue triangles, Fig. 4 A,C), while a FR excitation energy trap is formed in FRL-adapted samples at 740 nm (green triangles, Fig. 4 B,D).
The DAS (Fig. 4 A,B) and DAYS (Fig. 4 C,D) show broad emission bands of the time constants because (i) the fast component (black curves) subsumes fast EET and ET processes, which are difficult to resolve because of the strong spectral overlap of all molecules, and (ii) due to the limited spectral resolution of the DAS of 12.5 nm.Therefore, we conducted low-temperature measurements in order to benefit from reduced spectral overlap of the molecules under these conditions.
At low temperatures in WL-adapted cells (see DAS in Fig. 5A,C,E), four spectral regions of fluorescence emission can be discriminated.This is mainly visible in the DAYS (Fig. 6A,C,E): The first is observed around 650 nm, which comprises mainly of PBS fluorescence, the second is located around 685 nm, which can predominantly be attributed to Chl a (in PS II), the third is located around 720 nm, and, finally, the most red-shifted is found around 740 nm with very small amplitude.As discussed above, the band at 720 nm most probably results from FR Chl which can be Chl a or f.
In WL-adapted samples, we observed energetic decoupling of pigments upon cooling to very low temperatures, whereas in FRL-adapted cells, the emissions at 720 nm and 740 nm remain coupled (see blue, black and red curve in Fig. 6E,F).For the WL-adapted cells, three components are necessary to obtain satisfactory fit results.The fastest component (black squares) increases from 85 ps at 250 K (Fig. 5A) to 170 ps at 10 K (Fig. 5E), and it dominates the emission at 720 nm.The intermediate component (red circles) rises fom 490 ps at 250 K (Fig. 5A) to 610 ps at 10 K (Fig. 5E).Both components dominate DAS (Fig. 5) and DAYS (Fig. 6).In addition, a third component with a lifetime between 1.1 ns (250 K) and 1.6 ns (10 K) is found in WL adapted samples (blue triangles in Fig. 5A,C,E), which contributes to both, the PBS regime (650 nm) and the Chl a regime (685 nm).In the DAS, the contribution of these components at 650 nm and 685 nm is hardly visible, since the excitation of PBSs and Chl a is low at 470 nm.In the DAYS, it can be seen that both, Chl a and PBSs, slightly emit with both time constants (470 -600 ps, red circles and 1.1 ns -1.6 ns, blue triangles).The long fluorescence lifetimes and the isolated peaks indicate that the PBSs and Chl a, which are partially excited at 470 nm, are not functionally coupled to the reaction center, and decay with slow time constants typical for distorted PBSs and isolated PBSs (Li et al. 2001;Schmitt et al.. 2006;Schmitt 2010).The reason for the decoupling at low temperatures (10 K) might be due to slow freezing with a cooling rate of 1 K/s, which typically leads to PBS decoupling (Schmitt et al. 2006;Schmitt 2010;Schmitt et al. 2011;Schmitt 2020;Maksimov et al. 2013).Interestingly, the red states at 720 nm and 740 nm decay fast even at the lowest temperature.The time constants at 10 K are determined as 160 ps (black squares, 720 nm) and 600 ps (red circles, 740 nm).
The fast time constants of 90 ps (black squares) and 240 ps (red circles) in FRL-adapted samples at 10 K indicate that there is still functional EET between the primary donor and the red antenna Chl f, and that charge separation still occurs even at the lowest temperature investigated.Basically, two mechanisms contribute to the electron transfer (ET) processes in course of charge separation: The first is thermally activated ET with a probability given by the Boltzmann factor  − ∆  , and the second is caused by electron tunnelling (ETUN), which is temperature-independent, and remains as the only ET mechanism at the lowest temperatures (10 K).
To further insights into the mechanism of EET, thermally activated ET, and ETUN processes at low temperatures, the logarithm of the fastest rate constant found in WL-adapted samples and of the second fastest rate constant in FRL-adapted samples was plotted against the reciprocal temperature in the form of Arrhenius graphs, which indicate thermally activated temperature regimes (linear rise of   with 1/T) and the contribution of electron tunnelling as a constant, i.e. temperature-independent, rate constant.For the analysis of the temperature dependence of the rate constants according to the Arrhenius-Eyring theory, the fastest rate constant was chosen for WL-adapted cells, since these samples exhibit no significant EET between the excited species at deep temperatures.Therefore, it is assumed that this rate constant represents the ET from the primary donor.In FRL-adapted samples, one further decay component was necessary to fit the data satisfactorily, and the fastest component (see black curve in Fig. 5B,D,E and Fig. 6B,D,E) exhibits a negative amplitude at 740 nm, which indicates fast EET to Chl f.Therefore, it is assumed that the second fast time constant (red curves) represents ET from the primary donor.However, it can be seen, that the qualitative temperature dependence is similar for both rate constants (Fig. 8).
As can be inferred from Fig. 8A, the Arrhenius plot is a fairly linear function just for sufficiently high temperatures between 290 K (1/T = 0.0034) and 130 K (1/T = 0.077), while for lower temperatures, the rate constant stays essentially constant.In this low-temperature regime, a temperature-independent electron tunneling process is apparently observed.
For a quantitative analysis, an extended function was derived that includes both, thermally activated ET as well as electron tunneling (ETUN).In the most simple assumption, we anticipate that the overall rate constant for the electron transfer is composed of the sum of a temperature-dependent term () according the Arrhenius-Ayring theory, and a temperatureindependent term 0 k , which represents electron tunnelling as described by eq. ( 4): ( ) Therefore, the natural logarithm of the overall rate constant for the electron transfer is a transcendental function for the temperature-dependent ET part that contributes to overcome the free energy DG according to ( ) (5) For temperatures approaching 10 K, the temperature-dependent term from the Arrhenius-Eyring theory can be neglected, and the low-temperature approximation holds, since ( ) , it follows: And with the defintion ( ) This can be simplified so that the slope of ( ) , which comprises the entropy difference and an (unknown) normalisation factor A0 that has the character of an effective entropy involving different numbers of electron donor and electron acceptor molecules (similar to a degeneration factor).
The following results are obtained for WL-adapted samples: and for FRL adapted samples:

Simulation of the DAS
In order to interpret the DAS of FRL-adapted H. hongdechloris cells at low temperatures (10 K), the DAS measured at 10 K were simulated by setting up a minimal kinetic model in the form of a rate equation system.This model assumes that the secondary donor (PD) in the reaction center of PS II in H. hongdechloris might consist of red-shifted Chl a at the PD1 position, while the primary donor is a FR Chl at the ChlD1 position, which most probably is a Chl f as depicted in Fig. 9.The corresponding simulations of WL-and FRL-adapted samples at room temperature can be found in Schmitt et al. (2019) and Schmitt et al. (2021).
Since the DAS were all recorded upon excitation with 470 nm into the Soret band of the primary donor in PS II, these models mainly encompass the primary charge separation and the production of a radical pair consisting of the chlorophyll special pair and a pheophytin (denoted as Chl + /Pheo), from which the electron is then transferred further and charge separation is stabilized by transfer to the primary plastoquinone (PQ) QA.In addition, the antenna complexes contain phycobilisomes (PBS) and domains of Chl a and Chl f, which are coupled by EET (see Fig. 9).Fluorescence emission bands were modeled as Gaussian bands with a full-width at half-maximum (FWHM) of 20 nm.With the choice of the time constants as depicted in Table 1, the simulated DAS as shown in Fig. 10  The results show that even at 10 K, charge separation occurs within 200 ps (kDPhe) with just about 10 % recombination probability (2 ns), which is only slightly smaller than the one of open and active PS II RCs at room temperature.The charge stabilization by ET to plastoquinone QA (kPheQ) is assumed to occur with 670 ps to fit the experimental data.Thus, for FRL-adapted cells, EET from Chl a to the far-red Chl as primary donor and the subsequent charge separation is even possible at 10 K.
If the system is simulated without an electron tunneling process, the lowest electronically excited state at 740 nm would be quickly populated at 10 K and would decay slowly with the intrinsic lifetime (there 1.7 ns) visible as long fluorescence decay time (see Fig. 10, left side).However, such a long-lived energy trap is not observed in the cells at 10 K. Therefore, the fluorescence at 740 nm can not beling to an isolated long-lived trap even at low temperatures (10 K).

Conclusion
To summarize, we identified a small absorption band at 470 nm in FRL-adapted H. hongdechloris cells, which we attribute to the primary donor of PS II.Most probably, this molecule is a Chl f either at at the ChlD1 or PD2 position (Zamzam et al. 2020;Judd et al. 2020).
Time-resolved fluorescence dynamics measured upon excitation with 470 nm light at room temperature and down to cryogenic temperatures of 10 K suggest that the molecule, which absorbs at 470 nm facilitates or undergoes a charge transfer, which is intact with a time constant of 160 ps even at 10 K in WL-adapted H. hongdechloris cells.Charge transfer at such a low temperature is supported by an electron tunneling mechanism.Evidence for this was provided from Arrhenius analyses of the fast fluorescence component, which exhibited a temperature-dependent and an additional temperature-independent contribution for the overal ET rate constant.The latter, temperature-independent component is characterized by a time constant of 160 ps.The situation is similar in WL-and FRL-adapted cells, with the main difference being that in FRL-adapted samples, a further EET process to 740 nm can be found indicating the existence of a long-to QA, charge stabilisation, membrane dynamics and plastoquinone mobility, which ultimately separates the charges and enables an efficient and directional electron transfer.

3 )
Figure 1.(A) Room temperature absorption spectra of methanolic extracts from H. hongdechloris cells adapted to WL (black line) and illuminated with FRL for three days (red line).(B) Difference absorbance spectra of H. hongdechloris extracts from cells grown for three days under FRL and the corresponding control sample grown under WL as shown in panel (A).

Figure 1
Figure1shows distinct changes in pigment composition when H. hongdechloris cells were grown under FRL in comparison to the WL control sample (Fig.1A, red vs. black curve).After three days of FRL illumination, a distinct additional absorbance feature with maximum at 700 , Fig.2A, green arrow "1"), which might be caused by the Soret band of FR absorbing Chl.The review of the absorption spectra of isolated PS I(Kurashov et al. 2019) and isolated PS II complexes(Judd et al. 2020;Zamzam et al. 2020) indicates, that this peak at 470 nm is mainly caused by a pigment or pigments localized in photosystem II.To further analyze the pigment composition absorbance and fluorescence spectra of WL-and FRL-adapted samples are compared in Figure2.After five days of FRL adaption, more profound changes are observed between 400 and 500 nm due to increased contents of βcarotene, as noted previously(Chen et al. 2012).When put back to WL of an intensity of 30 µE•m -2 •s -1 , the content of Chl f started to decline (data not shown) indicating that Chl f formation in response to illumination conditions is a reversible process.In addition, prominent changes due to redistribution and remodeling of PBS can be observed(Chen et al. 2012;Chen et al. 2010;Schmitt et al. 2021).However, since the bilin pigments are covalently attached to the PBPs, they cannot be extracted by organic solvents, and, therefore, do not contribute to the spectral changes shown in Fig.1A und B).

Figure 2 .
Figure 2. (A) Room temperature absorbance spectra of methanolic extracts of H. hongdechloris cells adapted to WL (dashed) as the initial control sample and after illumination with FRL for five days (red).The absorbance spectra are adjusted to the same absolute value at 665 nm.(B) Room temperature fluorescence emission spectra of intact WL-adapted H. hongdechloris cells upon excitation with 430 nm (dashed black curve), and for samples adapted to FRL (red curve) for five days.The fluorescence spectra are adjusted to the same absolute value at 680 nm.(reprinted with permission from Friedrich and Schmitt 2021).(C) Room temperature fluorescence emission spectra of intact WL-adapted H. hongdechloris cells upon excitation with 470 nm (dashed black curve) and samples adapted to FRL (red curve) for five days.(D) Fit of the fluorescence emission band from FRL-adapted cells as shown in Fig. 2C with three hongdechloris cells.The most red-shifted peaks at 739 nm and, possibly, 745 nm are attributed to Chl f fluorescence, in line with previous observations(Chen et al. 2012, Schmitt et al. 2019).The presence of fluorescence in the 715-740 nm range in WL-adapted cells, in which no Chl f was detected(Chen et al. 2012), suggests that the emission strongly overlaps with far-red-shifted PBSs and red Chl a molecules already present in WL-adapted cells.Since the Soret band of Chl a is located at 430 nm, but shifts to the red in red-shifted Chl a, the transition from 715 nm upon 430 nm excitation to 725 nm upon 470 nm excitation indicates several subspecies of Chl a or Chl a vs.Chl f, respectively.

Figure 3 .
Figure 3. Fluorescence excitation spectra of H. hongdechloris cells adapted to WL (dashed black curves) and illuminated for five days with FRL (red curves), monitored at observation wavelengths 710 nm (A) and 730 nm (B).(Reprinted with permission from Friedrich and Schmitt 2021).

Figure 3
Figure 3 compares fluorescence excitation spectra of WL-and FRL-adapted H. hongdechloris cells.Of note, far-red emission occurs also in WL-adapted cells (Fig. 3A,B dashed black curves) at 710 nm and 730 nm.
Soret bands of both, Chl a and FR Chl like Chl f in FRL adapted samples.Time-resolved fluorescence spectra: DAS and DAYS at room temperature.Spectrally-resolved fluorescence decays were recorded with 470 nm excitation (illumination intensity 100 W/m²) on intact filament bundles of H. hongdechloris cells, either WL-adapted or adapted to FRL for five days, using multichannel TCSPC.The resulting DAS and DAYS of WLand FRL-adapted samples are shown in Figure4A,B.

Figure 4 .
Figure 4. Decay-associated spectra (DAS) upon 470 nm excitation for WL-(A) or FRL-adapted cells (B) after global approximation by a sum of three exponential decay components for WLadapted cells (78 ps, black squares, 500 ps, red circles and 2.1 ns, blue triangles), and four components for FRL-adapted cells (40 ps, black squares, 170 ps, red circles, 710 ps, blue triangles and 1.8 ns, green triangles), respectively.The decay-associated yield spectra (DAYS) were calculated according to eq. 3 (see Materials and Methods) and are presented in (C) for WL-adapted and in (D) for FRL-adapted samples.
Fig. 4B and DAYS in Fig. 4D) the fastest decay component of 40 ps (black squares) shows a peak at 700-710 nm and a distinct rise kinetics at 745 nm indicating a fast EET between FR Chl (most probable a far red Chl a) and the most red Chl molecules, most probable far-red Chl f at 745 nm.In addition, a long-wavelength component of 1.8 ns (green triangles in Fig 4 D) shows peaks at 720 nm and 745 nm indicating the development of a longwavelength trap in the FRL-adapted samples, which refers to FR Chl.The medium time component is prolongated in FRL-adapted samples with 710 ps (blue triangles in Fig. 4B,D)

Figure 7 .
Figure 7. Excitation energy transfer scheme (reprinted with modification with permission from Mamedov et al. 2015)

Figure 8 .
Figure 8. (A) Arrhenius plots of the rate constant of fastest fluorescence component in WLadapted samples (170 ps decay time at 10 K), and of the second fastest decay component in FRLadapted samples (240 ps decay time at 10 K). (B) Expanded view of the thermally-activated temperature range from (A), fits of the negative slopes of the   values with 1/T are shown with resulting activation energies determined from the fitting results.
, and the total kinetics follows the Arrhenus-Eyring approximation: the intersection with the y-axis is determined by the were obtained.While Fig. 10 (left side) shows the simulation result without electron tunneling (ETUN), the right side of Fig. 10 includes a tunneling rate constant of 4 ns -1 (250 ps time constant), which is included as the optional electron tunneling pathway in Fig. 9 (green arrow with ).Remarkably, the agreement between experimental and simulated DAS (compare Fig 10, right side with Fig. 5F) is satisfactory.

Figure 9 .
Figure 9. Kinetic scheme for the kinetic constants in the PS II reaction center of H. hongdechloris.

Figure 10 .
Figure 10.Simulation of the DAS without (kETUN=0) (left side) and with electron tunneling (kETUN =4 ns -1 )(right side) for the EET and ET processes in FRL-adapted samples as depicted in Fig. 9 at

Table 1
Rate constants used in the simulation shown in Fig.10