Excess manganese increases photosynthetic activity via enhanced reducing center and antenna plasticity in Chlorella vulgaris

Photosynthesis relies on many easily oxidizable/reducible transition metals found in the metalloenzymes that make up much of the photosynthetic electron transport chain (ETC). One of these is manganese, an essential cofactor of photosystem II (PSII) and a component of the oxygen-evolving complex, the only biological entity capable of oxidizing water. Additionally, manganese is a cofactor in enzymatic antioxidants, notably the superoxide dismutases—which are localized to the chloroplastic membrane. However, unlike other metals found in the photosynthetic ETC, previous research has shown exposure to excess manganese enhances photosynthetic activity rather than diminishing it. In this study, the impact of PSII heterogeneity on overall performance was investigated using chlorophyll fluorescence, a rapid, non-invasive technique that probed for overall photosynthetic efficiency, reducing site activity, and antenna size and distribution. These measurements unveiled an enhanced plasticity of PSII following excess manganese exposure, in which overall performance and reducing center activity increased while antenna size and proportion of PSIIβ centers decreased. This enhanced activity suggests manganese may hold the key to improving photosynthetic efficiency beyond that which is observed in nature.


Results
Photosynthetic performance and flux. Previous work revealed that C. vulgaris accumulated manganese beyond equilibrium concentrations and that exposure to elevated levels of manganese resulted in an increased photosynthetic capacity over time as measured using F V /F M , the quantum yield of primary photochemistry 24 . This enhanced photosynthetic performance accompanied a significant increase in intracellular and adsorbed manganese, despite little to no effect on the overall turbidity. In order to determine the mechanisms responsible for this enhanced performance, cells were grown to mid-exponential phase-the point at which cells have the highest chlorophyll content and are most photosynthetically efficient-and measured at a concentration of 2.5 µg mL −1 total pigments to ensure the differing pigment content did not affect results ( Fig. 1) 36,37 . It is essential to control for total chlorophyll prior to taking fluorescent measurements, as chlorophyll provides a proxy for total photosystem number, where lower amounts of chlorophyll is associated with lower numbers of PSII (Supplemental Fig. 1). Normalizing to chlorophyll analyzed ensures measurements assess the efficiency as opposed to the quantity of photosystems. F V /F M was significantly increased and decreased in comparison to the control at 750 and 1000× manganese, respectively (Fig. 2a). More interesting, however, was the change in performance index on a per absorption basis (PIabs). As this parameter quantifies PSII activity normalized to total photon absorption, trapping of excitation energy, and conversion of energy to electron transport, it is a more accurate representation of photosynthetic activity 38,39 . The PIabs of the 500 and 750× manganese cultures were 2.1 and 2.4× higher than that of the control cultures (Fig. 2b). However, photosynthetic performance decreased dramatically in the 1000× manganese cultures, indicating that while increased manganese improves the photosynthetic performance to an extent, there is still a threshold at which overall activity is decreased.
In addition to an increasing performance index, exposure to increased manganese decreased the overall Chl fluorescence output. Both F 0 and F M decreased as manganese concentration increased, with F M having an intensity of 1.755 ± 0.120 in the control cultures and an intensity of 1.128 ± 0.057 in the 1000× Mn cultures for a 36% depletion. These decreases could occur from either an increase in NPQ or an increased amount of active reducing reaction centers (or a combination of the two). However, the JIP test parameters indicate a more complex story (Fig. 3). All of the specific fluxes per reaction center decrease compared to the control, meaning the trapped energy (TR 0 /RC), electron transport flux (Et 0 /RC), dissipated energy flux (DI 0 /RC), and reduction of the acceptor side of PSI (RE 0 /RC) have all diminished. However, that is combined with an overall decrease in Abs/ RC, the average absorbed photons per reaction center and an indicator of relative antenna size. Fewer photons would equate to less overall flux, but may increase the efficiency; in previously published work, Chlorella sp.
Reducing centers. Several fluorescence protocols enable the comparison of reaction center activity. In order to determine the relative amount of active Q A sites, the initial point of electron transport in PSII, the cells were exposed to DCMU, an herbicide that selectively binds to the Q B site of PSII and inhibits electron transfer [45][46][47] . The resulting fluorescent transient reaches its maximum output once all Q A sites are reduced and does not allow for PSII turnover, generating a plot with an initial slope that inversely correlates to the relative amount of active Q A sites within a sample (Fig. 4). Manganese appears to increase the activity of Q A reducing centers, peaking at the 1000× manganese concentration with a 25% increase over the control. The relative number of non-reducing QB sites were determined using the double hit method for collecting the OJIP fluorescence transient. Similar to the Q A reducing activity, the B 0 , or relative amount of non-reducing Q B sites, decreases as the concentration of manganese increases, with peak activity occurring at 750× manganese (Fig. 5a). The 1000× manganese cultures are also less active compared to the 500× and 750× manganese cultures, but, unlike the Q A sites, do not surpass the control cultures, demonstrating an increased activity of Q B even at 70 mM manganese.
In order to understand the full effect of Q A and Q B activity, it is necessary to comprehensively measure the activity of PSII reaction centers. Previous studies have accomplished this through measurements of S-state populations, wherein the fluorescence decay after the fourth flash is controlled almost entirely by PSII X 48-50 . When measured using the difference between S 4 and S 0 of the S-state cycle, it would appear that PSII X (relative inactive PSII) were increasing, despite noted increases to reducing center activity (Fig. 5b). However, these differences did not show statistical significance, a problem described in a previous study of C. vulgaris exposed to excess lead 32 . When using the OJIP method to determine the percent of non-silent reaction centers (Eq. 5), all concentrations 500× and over increased in overall activity, with the 750× manganese showing a peak activity 68% higher than  To determine the relative proportion of PSIIα to PSIIβ, we used fast fluorescence induction (FFI), which implements a single turnover saturating flash to induce rapid Q A reduction 51 . Interestingly, FFI revealed an overall decrease in PSIIβ centers (Fig. 6b). This indicates that although Chl content and ABS/RC are both diminished in cultures exposed to excess manganese, they are still primarily relying on PSIIα supercomplexes, therefore suggesting that the decreased antenna size is from a decreased number of antennas and associated reactions centers rather than interconversion to smaller complexes.
To further analyze the manganese-induced changes to the antenna complexes, low temperature (77 K) fluorescence spectroscopy was used to assess the distribution between LHCI and LHCII and its dependence on manganese concentration. Low temperature fluorescence results in a dual peaked trace with maxima at 685 nm and 730 nm, and has been used to distinguish the presence of state 1 and state 2 transitions in vivo, through which state 1 is shown through a shift toward lower wavelengths 52 . These transitions enable a more finely tuned balance of light absorption by the photosystems and allow rapid acclimation to changes in light intensity, making state transitions a key component of photoprotection 53,54 . Previous research has isolated the component parts of the photosynthetic cores and LHCs in order to distinguish their contributions to the overall spectra [55][56][57][58][59] . Thus, by assuming a non-linear Gaussian distribution and aligning to the best fit, the component peaks representing LHCII (681 nm), CP47/CP43 (686.5 nm), CP47 (695.5 nm), PSI core (721.5 nm), and LHCI (735 nm) could be individually assessed (Table 1, Supplemental Fig. 2). This revealed a decrease in PSII-associated LHCII at the 250×, 500×, and 750× manganese concentrations, mirroring that which was described from the decreasing ABS/ RC and correlating manganese with decreased antenna size. By taking the ratio of the LHCII:LHCI peaks, it is possible to delineate relative number of antennas associated with PSII compared to PSI, which can suggest the prominence of PSII-initiated linear electron flow relative to PSI-initiated cyclic electron transfer. As the concentration of Mn increases, the LHCII:LHCI ratio increases, with significantly increased ratios at 250× (p = 0.02) and 500× (p = 0.01) manganese. The rising LHCII:LHCI ratio demonstrates that manganese either promotes an ETC that is increasingly reliant on PSI-initiated light harvesting, or that that C. vulgaris grown in excess manganese exhibits fewer LHCII antenna complexes. The latter is more likely as the cultures with increased manganese had lower levels of pigment accumulation ( Fig. 1) and the aforementioned decreased ABS/RC (Fig. 6a). This is complicated by the decrease in PSIIβ centers (Fig. 6b); however, since multiple photosystems can receive photons from one supercomplex, it is possible that fewer overall LHCs are needed in the higher Mn cultures.

Discussion
While photosynthesis and light harvesting relies on several important metal cofactors, excess exposure to many of them-including copper, iron, and magnesium-leads to disruption of the photosynthetic ETC 60,61 . Our results indicate that this is not the case with manganese: in fact, overall performance increases with increasing manganese concentrations (Fig. 2b). Based on fluorescence measurements, this is facilitated by an increase in the activity of reaction centers paired with a decrease in PSIIβ complexes and decrease in average antenna size (Figs. 5,6). This increased activity could result from an increase in manganese concentration within the chloroplast itself. While determining subcellular localization was outside the scope of this study, previous studies in algae have shown a tendency for metals to accumulate both in cell walls as well as in plastids [62][63][64] . Furthermore, www.nature.com/scientificreports/ our previous work determined that C. vulgaris accumulates intracellular manganese up to 55× the surrounding medium, suggesting that these manganese stores may be compartmentalized into an organelle or complexed in order to vastly exceed equilibrium concentrations 24 . Chloroplastic storage could be facilitated by plastidal manganese transporters. Chlamydomonas reinhardtii, a model green alga, encodes two distinct chloroplastic manganese transporters via PAM71-HL and CGLD1 [65][66][67] . However, these proteins are not found in the Chlorella genome; a BLAST search resulted in no matches over a 50% similarity threshold. Unfortunately, the limited genetic annotation of C. vulgaris prohibits thorough understanding of the genetic and proteomic regulation that could increase manganese stores outside of equilibrium. OEC assembly is not facilitated by protein chaperones, but requires a redox potential similar to that of Chl a, enabling the initial Mn 2+ to Mn 3+ oxidation to occur on the surface of PSII 68 . Thus enhanced manganese concentration in the chloroplast could increase the accessibility of manganese for PSII binding sites, and push the reaction kinetics forward toward more efficient de novo OEC synthesis, thus increasing the rate of photoassembly. These increases in photoassembly efficiency may then contribute to the observed performance increases by enabling more efficient rate of photorepair following the turnover of PSII D1 dimers 69,70 . Photoinhibition and the subsequent photorepair cycle is inevitable, and the overall efficiency of the photosynthetic apparatus is dependent on the difference between the rate of photoinhibited and photorepaired PSII 71 . Increasing the efficiency of this photorepair would therefore generate a net increase in active PSII units compared to basal conditions, potentially increasing the overall performance and resulting in the increased activity observed herein. This is supported by studies that have shown manganese-deficient plants to decrease in photosynthetic efficiency due to PSII structural instability, with the magnitude of instability related to manganese binding affinity to PSII and suggesting that photorepair is, in part, concentration dependent 72 . However, since PSII repair requires the translation of D1 subunits directly into the membrane, it is likely that the synthesis and insertion of D1 is rate-limiting 68 . Therefore,   The relative amount of NPQ in cells exposed to increasing levels of manganese, measured using pulse amplitude modulated fluorometry. (f) The oxygen production of cells exposed to increasing concentration of manganese, measured using a Clark-type electrode. www.nature.com/scientificreports/ it is likely that increased manganese concentration has a threshold for increasing repair efficiency, as shown here through the decreased photosynthetic flux at the higher concentrations of manganese. The likelihood that photorepair efficiency has improved is further supported by the decrease in PSIIβ complexes. Previous research has determined that PSIIβ complexes are localized to the stroma lamellae of the chloroplast, in which damaged PSII monomers are repaired and new PSII units are synthesized [73][74][75] . Thus a decreasing amount of PSIIβ could indicate a faster flux out of the stroma lamellae. The decreased PSII-bound LHCII complexes in the higher Mn cultures as revealed by the low temperature fluorescence supports this further, as complexes could more easily share LHCII super complexes if there is less flux toward PSII repair. However, this switch to a predominantly LHCI-based system could also be an artifact of the enhanced PSII activity. If the flux of electrons through PSII increases due to increasing reducing center activity, the plastoquinone pool will then become oxidized faster. This would then increase the flow of electrons toward the downstream proteins, potentially enhancing the turnover rate of PSI. Ergo the switch to PSI-bound LHCI may be generated from the increased rate of oxidation at the acceptor side of PSII as the entire photosynthetic electron transport chain moves electrons at a faster overall flux.
Another possibility is that the manganese is helping protect the photosynthetic ETC from sustaining damage in the first place, which may explain why the level of YII increases while the NPQ remains unchanged across concentrations (Fig. 5d, e). While unexplored in algae, manganese has been shown to increase oxidative stress resistance in organisms with active superoxide dismutase enzymes in a manner dependent on high manganese/ iron ratios within the cell, allowing such a boost in antioxidant capacity that it helped facilitate resistance to gamma-radiation [76][77][78] . Furthermore, it was shown that this manganese-facilitated resistance protected proteins, not DNA; thus, manganese may have a role in protecting proteinaceous components of the photosynthetic   79 . Additionally, adding excess manganese to the growth media of yeast with lethal superoxide dismutase knockouts rescued the phenotype, allowing growth under aerobic conditions and increasing scavenging activity with no change in enzymatic catalase or peroxidase activities 80 . Manganese is also capable of catalase-like activity and can disproportionate H 2 O 2 in the presence of bicarbonate, independent of enzymatic processes 81 . Furthermore, manganese can form non-proteinaceous complexes through mixtures of phosphate, peptides, carbohydrates, and nucleoside bases, with which it can actively scavenge superoxide radicals [82][83][84] . Manganous phosphate salts are particularly effective, as they rapidly catalyze the scavenging of superoxide radicals using a mechanism completely distinct from that of superoxide dismutases, in which Mn 2+ reacts to form a short-lived MnO 2+ before disproportionating to form manganese phosphate, dioxygen, and hydrogen peroxide 85 . Further studies are needed to determine if increased reducing activity is due to an increase in photorepair efficiency, a decrease in photoinhibition, or a combination of the two.

Materials and methods
Strain and culture growth conditions. Cultures of C. vulgaris (Carolina Biological Supply) were maintained on lysogeny broth agar plates and for batch cultures, inoculated into 25 mL of modified Chlorella medium supplemented with 20 g L −1 dextrose. All cultures were grown in 50-mL sterile Erlenmeyer flasks capped with aluminum foil 37 . Cultures were grown in triplicate, using a 1-mL inoculum from a stationary phase culture and kept under constant white light at 30 µmol photons m −2 s −1 via an LED light at 25 °C with an orbital rotational speed of 100 rpm (verified using a tachometer).
Experimental design. Prior to culture inoculation, manganese-deprived modified Chlorella medium (containing 20 g L −1 dextrose) was dosed with an autoclaved stock of MnCl 2 and ultrapure H 2 O, ensuring that all cultures had equal concentrations of all other medium components. The control cultures' media was dosed to generate a working concentration of 0.070 mM MnCl 2 , the standard concentration for modified Chlorella medium 37 . Experimental cultures' media were 250× (17.5 mM) to 1000× (70.0 mM) the control with an n = 6. Cultures were maintained under constant light and grown until they reached mid-exponential phase, shown through a spectroscopic cell density of between 8 and 10 AU and peak pigment concentration for that particular experimental group. Mid-exponential phase was chosen to assay each culture at its peak photosynthetic productivity.

Pigment extraction.
Pigments were extracted using DMSO as previously described and measured from 470 to 700 nm 37 . Chl a, chlorophyll b (Chl b), and total carotenoids were calculated using the following equations 86 : Photosynthetic electron transfer fluxes. Photosynthetic electron transfer fluxes were inferred from in vivo Chl a fluorescence using a Photon Systems Instruments FL 3500 fluorometer following a 5 min dark adaptation as previously described 36 . To ensure reproducibility between samples, algal cultures were diluted to 2.5 µg mL −1 total pigments using medium of the same manganese concentration before dark adaptation and subsequent measurements. The OJIP protocol included a 1-s actinic illumination using a 630-nm light at an intensity of 2400 µmol photons m −2 s −1 . Fluorometry parameters (JIP test) were calculated as previously outlined (Stirbet et al. 87 ).

Determination of reducing center activity.
To further increase understanding of the mechanisms at work and to estimate the relative amount of Q B non-reducing centers, the double hit method was used, collecting Chl a fluorescence data at two subsequent 1-s pulses 32,34,88 . This method generates sequential OJIP transients that can be separated and normalized to t = 0. While the first pulse was conducted following dark adaptation, meaning that all the reaction centers were open, the second pulse will only excite so-called 'fast-opening' reaction centers, allowing for the calculation of non-reducing centers (centers which are unable to open in time for the second pulse) using the equation: where F V /F M is derived from the first pulse and F V */F M * is derived from the second pulse. The redox state of the OEC can be determined using short, actinic light flashes to sequentially advance from S0 to S4, where each S-state represents a progressively more oxidized state of the OEC 89,90 . The contribution of inactive PSII (PSII X ) centers can be estimated by the difference between the S4 S0 fluorescence decay and the initial F 0 , since the fluorescence decay following the fourth flash is primarily controlled by inactive centers 48,91 . S-states were measured through 4 saturating flashes at 80,000 µmol photons m −2 s −1 voltage that were 100 µs in duration and 300 ms apart, each causing a single turnover of PSII 32 . Cells were dark adapted for 5 min prior to measurements. www.nature.com/scientificreports/ In addition to quantifying inactive Q B centers, active Q A reducing centers were quantified through the inhibition of PSII with 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU), a selective inhibitor that blocks the Q B site [92][93][94] . When DCMU-inhibited cells are measured for Chl a fluorescence, the normal polyphasic transient is replaced by a single-phase plateau, indicating the single turnover of Q A to Q A -. By comparing the slope of the normalized curve when t = 300 μs, it is possible to infer the relative number of active reducing centers, as a decreased slope would indicate a longer time needed to reduce all Q A , therefore suggesting an overall increase in activity 39 . Dark adapted samples were exposed to 50 mM DCMU prepared in DMSO for 10 min before measuring using the same protocol used for the JIP test. DCMU was prepared as a 1.5 M stock and combined with cell media to form a 1 mL solution to maintain DMSO concentrations below 5% and prevent damage to photosynthetic proteins. GraphPad Prism v.7.01 was used to perform the integrations following normalization 32 .
It is also possible to calculate active Q A centers using JIP test parameters, in which the absorption flux per reaction center (ABS/RC) is used in the following equation 30,32,95,96 : For control (0.07 mM manganese) cultures, the average ABS/RC of all cultures was used as the control, in order to determine the overall standard error in activity. In order to report increases in activity, the RC si value was multiplied by − 1 and added to 100, resulting in a measurement of percent activity as compared to the control condition.
Last, the size of the reducing pool can be estimated by calculating the normalized complimentary area over the OJIP curve; since this area is assumed to be proportional to the number of reductions and oxidations of one Q A center, it enables relative comparison of the number of electron carriers along the ETC 39,97 . To calculate the normalized area, the OJIP was double normalized where F O = 0 and F P = 1 before using GraphPad Prism v.7.01 to calculate the complementary area above the curve. To allow comparisons between experimental groups, the area was divided by the variable fluorescence.
Non-photochemical quenching analysis. Quenching analysis was used on dark-adapted cells with an actinic intensity of 300 µmol photons m −2 s −1 , a saturating pulse intensity of 64,000 µmol photons m −2 s −1 , and a measuring flash voltage of 80%. There was a dark relaxation duration of 20 s between pulses 98 . Photochemical coefficients were calculated as previously reported 39 .
Antenna size and heterogeneity. Antenna size was estimated by the calculation of Abs/RC by the following equation 33,39 : While this parameter gives a relative size difference of antenna complexes attached to active centers, it does not show the influence of different antenna types. To determine the partitioning between PSIIα and PSIIβ centers, flash fluorescence induction (FFI) analysis was performed after applying a 50 µs single-turnover flash using the 100% relative flash energy on the instrument with the detector sensitivity set at 3%, as outlined in previously published research 32,35,51 . The area over the double normalized curves was integrated using OriginPro (Origin-Lab, Northampton, MA) and used to generate a semi-log plot consisting of a fast, sigmoidal phase attributed to PSIIα and a slow, linear phase attributed to PSIIβ. The intercept of these two phases reflects the proportion of PSIIβ reaction centers 35,99-101 . Analysis of state transitions. State transitions were analyzed using a home-built tabletop 77 K fluorescence spectrometer as previously published, using the FLAME-S spectrophotometer in place of the JAZ-EL200 (Ocean Optics, Largo, FL) 102 . Cell cultures were diluted to 3 µg mL −1 Chl a with ultrapure water and immediately frozen in liquid N 2 . Chlorophyll were excited using a wavelength of 435 nm and spectra were collected with an integration time of 350 ms and averaged over 16 scans. Peaks for the 77-K fluorescence spectra were determined in OriginPro 2019b using the Multiple Peak Fit tool. Briefly, data were subjected to multiple Gauss fittings in NLFit with Auto Parameter Initialization, six peaks were necessary to account for the area under the spectral envelope.
Oxygen-evolving complex activity. Photosynthetic oxygen production was measured using a Clarktype oxygen electrode (Oxygraph; Hansatech, UK) at 25 °C as previously described 36 . Oxygen measurements were conducted by exposing 1:1 media diluted cell suspension in modified Chlorella media to a 1700-µmol pho-tons· m −2 s −1 white light after a 5-min dark adaptation phase. High light intensity was chosen in order to ensure light saturation of PSII-the initial rate of oxygen production was measured, normalized to Chl a, and used for comparison across culture conditions. Statistical analysis. Statistical analyses were performed with GraphPad Prism 7.01 (GraphPad Software, Inc.). Data were analyzed through a one-way Kruskal-Wallis test, comparing each experimental group to that of the control. The family-wise error rate for each figure was maintained at 0.05 using the Holm-Bonferroni method. Statistical significance is indicated numerically through increasing asterisks, where * indicates p ≤ 0.05, www.nature.com/scientificreports/ ** indicates p ≤ 0.01, *** indicates p ≤ 0.005, and **** indicates p ≤ 0.001. Figures show the means of data (n = 6) and the error bars denote the standard error of the measurement, unless otherwise stated.

Data availability
All data generated or analyzed during this study are included in this published article and its supplementary information files.