Cytochrome b6f complex inhibition by antimycin-A requires Stt7 kinase activation but not PGR5

Ferredoxin-plastoquinone reductase (FQR) activity during cyclic electron flow (CEF) was first ascribed to the cytochrome b6f complex (b6f). However, this was later dismissed since b6f inhibition by antimycin-A (AA) could not be reproduced. AA presumably fails to ligate with haem bh, at variance with cytochrome bc1 complex, owing to a specific Qi-site occupation in b6f. Currently, PROTON GRADIENT REGULATION5 (PGR5) and the associated PGR5-Like1 are considered as FQR in the AA-sensitive CEF pathway. Here, we show that the b6f is conditionally inhibited by AA in a PGR5-independent manner when CEF is promoted. AA inhibition, demonstrated by single b6f turnover and electron transfer measurements, coincided with an altered Qi-site function which required Stt7 kinase activation by a strongly reduced plastoquinone pool. Thus, PGR5 and Stt7 were necessary for b6f activity and AA-sensitive electron transfer in CEF-favouring conditions. Extending previous findings, a new FQR activity model of the b6f is discussed.


Introduction 1
Light is captured by two photosystems (PSI and PSII) and their associated light 2 harvesting complexes (LHCI and LHCII) which results in the splitting of water by PSII and 3 the reduction of ferredoxin (Fd) by PSI. Reduced Fd carries the electrons to Fd-NADP(H) 4 oxidoreductase (FNR) which generates NADPH in this linear electron flow (LEF) 5 process, thus providing the reducing equivalents for CO2 fixation in the Calvin Benson 6 cycle. Light-driven charge separation and water splitting generate a membrane potential 7 () and a proton gradient (pH), respectively, and this proton motive force (pmf) drives 8 ATP synthesis. Both photosystems are functionally connected by the cytochrome b6f 9 complex (b6f) which sophisticatedly transfers electrons from plastoquinol (PQH2) to 10 plastocyanin 1, 2 . These steps involve an electron bifurcation within the stromal PQH2 11 binding pocket, the Qo-site. The first donated electron from PQH2 enters the high-12 potential chain and the second one enters the low-potential chain. The proton 13 release/binding that is associated with the interconversion of PQH2 to plastoquinone 14 (PQ) generates a pH, and charge separation within the low-potential chain produces 15 . Besides minor subunits (PetG, L, M and N), the b6f core subunits are cytochrome f 16 (cyt.f), the Rieske iron sulphur protein (ISP), subunit-IV and cytochrome b6. According to 17 their redox midpoint potential (Em), which is between 300 mV (Rieske ISP) and 380 18 mV (cyt.f), the high-potential chain differs from the low-potential chain. The latter is 19 formed by three redox cofactors: haems bl (Em  130 mV), bh (Em  35 mV), and ci. 20 Haem ci is near haem bh in the stromal Qi-site and is linked to cytochrome b6 via a single 21 thioether bond, lacking amino acid axial ligands. Its Em ranges from 100 mV to 22 approximately 150 mV, when ligated to semiquinone analogues 3 . Moreover 4 , the 23 presence of a  may modulate the Em of haem ci, resulting in the shared electron to 24 reside as bh red /ci ox . The structural properties of haem ci were suggested by some 25 authors 5, 6 to be relevant for cyclic electron flow (CEF) where Fd reduces haem ci and 26 thereby equips the b6f with Fd-PQ reductase (FQR) activity. CEF between PSI and the 27 b6f is important in ATP-depleted conditions to drive the Calvin Benson cycle by 28 compensating for the excess of NADPH, and by protecting PSI from photodamage. The 29 latter is realized by various processes such as photosynthetic control of the b6f that slows 30 down PQH2 oxidation in the Qo-site at low lumen pH. 31 The light capturing efficiency may be unevenly distributed within the photosynthetic 32 machinery since light quality is dynamically changing. Therefore, lateral movements of 33 mobile LHCII transiently serve as compensatory adjustments between PSII and PSI, 34 which is termed state transitions 7 . A crucial sensor/responder function for state 35 transitions is the PQ pool redox state and a thylakoid-associated Ser/Thr protein kinase. 36 In Chlamydomonas reinhardtii, the kinase is termed Stt7 (STN7 in Arabidopsis thaliana) 37 and it is activated by a reduced PQ pool. Various phosphatases were identified as kinase 38 antagonists and, in order to phosphorylate LHCII and favour its movement towards PSI, 39 Stt7 activation requires a functional b6f Qo-site 8 . The kinase interacts with the b6f 9 and it 40 is not known whether Stt7 is required for the b6f function although it phosphorylates 41 various residues in subunit-IV and in the loosely attached subunit PETO 10, 11, 12 . The 42 latter is an algal CEF effector protein that was identified in a PSI-b6f supercomplex 13,14,15,43 16 . FNR, which is considered as a b6f subunit 17,18,19 , was routinely detected in these 44 enriched fractions together with proteins of the CEF pathway that depends on PROTON 45 GRADIENT REGULATION5 (PGR5). We recently found that the algal b6f failed to 46 operate only in CEF-promoting conditions when PGR5 was absent, which we attributed 47 to a dysfunctional FQR activity of the b6f 20 . This implied that the b6f operates via a 48 canonical Q cycle in LEF conditions and as FQR during CEF for which PGR5 ensures Fd 1 supply by recruiting FNR to the membrane 20, 21 . Our findings challenged an 2 acknowledged model that ascribes the FQR to be formed by PGR5/PGR5-like1 22, 23 . The 3 PGR5 interactome was mainly discovered owing to the observation 24 that the b6f was not 4 inhibited by antimycin-A (AA), unlike PGR5-dependent slow chlorophyll fluorescence 5 kinetics in vitro 22 . AA is a well-known Qi-site inhibitor in the respiratory cytochrome bc1 6 complex 25 . Initial studies demonstrated inhibition of the b6f by AA 26 which, at the time, 7 supported the view that the b6f acts as FQR during Fd-dependent cyclic 8 photophosphorylation 27,28,29 . 9 Here, we show in vivo that the b6f is exclusively inhibited by AA in CEF-favouring 10 conditions as a function of the PQ pool redox state. The inhibition was independent of 11 PGR5 but relied on Stt7 kinase activity. Efficient electron transfer under CEF-promoting 12 conditions relied on the AA sensitive step of photosynthesis, i.e. on the FQR activity of 13 the b6f. A refined model of our previous findings is discussed where Stt7 primes the Qi-14 site and PGR5 provides Fd. 15

Results 1
Conditional antimycin-A sensitivity of the cytochrome b6f complex. Based on the 2 structure of the algal b6f with the unusual Qi-site occupation 5 and the ligation of AA to 3 haem-bh in the respiratory cytochrome bc1 complex 25 , it has been proposed that AA fails 4 to act as Qi-site inhibitor in b6f 2 .The proposal has been reinforced by various conflicting 5 reports on the b6f inhibitor potency of AA in isolated chloroplasts from vascular plants 24, 6 27, 28, 29 . To test AA efficacy in vivo, we used the green alga Chlamydomonas reinhardtii 7 grown at moderate light. Indeed, we did not observe an effect of AA on the b6f kinetics in 8 light-adapted cells that were mixed thoroughly throughout the measurement, i.e. kept in 9 an oxic state (Fig. 1a). In both control and AA sample, the majority of photo-oxidized cyt.f 10 was re-reduced within 10-ms, and the b-haems net reduction and oxidation was finished 11 ~2-ms and ~30-ms after the flash, respectively. The redox amplitudes of the b-haems 12 signals were slightly increased in the presence of AA and the reactions ceased at pre-13 flash levels, like in the control. When monitoring the electrochromic shift (ECS) of the 14 photosynthetic pigments in these samples, the electrogenic contribution of the b6f during 15 the first 10-ms after the flash (b-phase) was hardly visible in the control of Fig. 1b. As 16 expected in light-adapted cells, the b-phase was masked by the coinciding and very 17 efficient  consumption via ATP synthase (in the thioredoxin-mediated reduced state).

18
The b-phase was more apparent when AA was present, owing to a slowdown of ATP 19 synthesis. Again, this was expected since the inhibitory AA impact on respiration 20 interfered with metabolic coupling between chloroplasts and mitochondria 30, 31 . Thus, the 21 exchange of reducing equivalents from the plastid for mitochondrial ATP resulted in a 22 more reduced, slightly ATP-depleted chloroplast which is known to slowdown algal ATP 23 synthesis 32 . The addition of PSII inhibitors to oxic AA samples revealed a pronounced b-24 phase that coincided with a further slowdown of the ATP synthase activity (grey symbols 25 in Fig. 1b), pointing to an even more ATP-depleted chloroplast. Next, we monitored 26 electron transfer rates during a 10-s illumination period (Fig. 1c). Since the samples were 27 light-adapted, the electron sink capacity of the Calvin Benson cycle was still active after 28 30-s darkness at the onset of re-illumination. Accordingly, initial rates of ~175 charge 29 separations/s/PSI levelled off in the controls after ~500-ms of illumination which yielded 30 quasi steady state rates of ~75 charge separations/s/PSI. A transient slowdown which 31 was followed by a re-acceleration was observed between 10-ms and 500-ms of 32 illumination. In AA samples after the 30-s dark period, the initial rates of ~200 charge 33 separations/s/PSI were slightly higher. Compared to controls, elevated rates were 34 observed up to 50-ms of illumination. It is of note that a slight, AA concentration-35 dependent stimulation of non-cyclic photophosphorylation has been observed in earlier 36 reports as well 29 . Following the AA-stimulated phase up to 50-ms, a period of diminished 37 electron transfer was observed and a virtual steady state of ~75 charge 38 separations/s/PSI was established no later than 2.5-s of illumination, owing to a slower 39 re-acceleration phase. Compared to AA samples, only slightly lower initial electron 40 transfer transients up to 25-ms of re-illumination were measured when PSII was inhibited 41 via HA/DCMU in the presence of AA. In these strongly ATP-depleted samples after 30-s 42 darkness, the light-induced electron transfer ceased exponentially within ~200-ms and 43 yielded steady state rates of ~5 charge separations/s/PSI after ~500-ms light. previous report 20 , the b6f kinetics of the controls in Fig. 1d differed from oxic samples.

46
Cyt.f re-reduction was slightly delayed, finishing shortly after 10-ms darkness. The initial 47 b-haems signals in anoxic controls showed a similar 1-ms net reduction after the flash. 48 However, kinetics differed profoundly thereafter by displaying a large net oxidation phase 1 up to 60-ms, followed by a b-haems re-reduction after ~500-ms darkness. In contrast to 2 oxic samples, the presence of AA inhibited the b6f under anoxic conditions, both on the 3 levels of cyt.f re-reduction and b-haems redox kinetics (cf. open symbols in Figs. 1a and 4 1d). The impaired b6f contribution to the light-dependent  generation was also visible 5 on the level of ECS kinetics after the flash, as shown by the absence of the b-phase 6 upon AA treatment (cf. open circles in Figs. 1b and 1e). Since ATP synthase is slowed 7 down in reducing conditions 32 , the b-phase became apparent in the controls of Fig. 1e. 8 When AA-treated anoxic cells were further illumination upon PSII inhibition via 9 HA/DCMU, the b-phase was partially restored (grey symbols in Fig. 1e). This suggests 10 mitigation of the AA effect in absence of PSII photochemistry. We also checked electron 11 transfer rates (Fig. 1f), clearly demonstrating the inhibitory effect of AA in anoxic cells.

12
The controls displayed sustained initial transients with up to ~250 charge 13 separations/s/PSI in the first 10-ms of re-illumination (at least). This highly active initial 14 phase was followed by an exponential decrease in activity to reach the apparent steady 15 state of ~75 charge separations/s/PSI after ~300-ms in the light. AA treatment 16 diminished both the initial phase and the steady state. The rates in AA samples, starting 17 from ~165 charge separations/s/PSI, levelled off at ~40 charge separations/s/PSI after 18 50-ms of light. Unexpectedly, slightly higher transients were observed between 10-ms 19 and 200-ms of light when PSII was subsequently inhibited. 20 21 1 Fig. 1 The conditional sensitivity of the cytochrome b6f complex to antimycin-A depends 2 on a cellular redox poise. Here, light-adapted WT cells are shown that experienced 30-s 3 darkness before the measurements. Representative kinetics and, where indicated, 4 means of three biological replicates (± SD) are shown. a Redox signals for cytochrome f 5 (cyt.f, squares) and b-haems (circles) are shown in oxic WT controls (-AA, closed 6 symbols) and after treatment with antimycin (+AA, open symbols). The AA treatment did 7 not substantially alter cyt.f re-reduction upon a flash, shown by a signal increase in the 8 positive direction that finished no later than 50-ms. AA did also not affect the b-haems 9 reduction (positive signals) and re-oxidation to pre-flash levels, finishing at ~100-ms. b 10 The electrochromic shift (ECS) signals varied slightly since the b6f-dependent 11 electrogenic 10-ms phase upon a laser flash was obscured in light-adapted controls 12 (black) by a highly active ATP synthase that resulted in a fast ECS decay phase. 13 Slowdown of the latter revealed the b6f-phase in AA samples (white), also upon PSII 14 inhibition by addition of HA and DCMU (grey, see Material and Methods). c An ECS-15 based electron transfer rate (ETR) measurement during a 10-s re-illumination shows the 16 gradual decrease of high initial ETRs to quasi steady state levels, established no later 17 than 2.5-s. Unlike AA treatment, PSII inhibition lowered steady state ETR. See text for 18 details. d Redox signals in the b6f, as in panel a, are shown for anoxic cells. While cyt.f in 19 the controls was similar as in oxic samples, the b-haems oxidation phase reached 20 signals below the pre-flash levels, followed a by a re-reduction at ~500-ms. AA treatment 21 inhibited cyt.f reduction and b-haems redox kinetics. e ECS kinetics revealed the b6f-22 phase in the control due to slow ATP synthesis. AA treatment abolished the b6f-phase 23 and light adaptation in the absence of PSII activity partially restored it (cf. white and grey 24 10-ms phases). f Re-illumination of anoxic controls sustained an elevated ETR up to 25-25 ms. The steady state ETR and the induction phase depended on AA-sensitive 26 processes. The induction phase could be partially recovered when cells were illuminated 27 in absence of PSII photochemistry. 28 Mitigated antimycin-A sensitivity of the cytochrome b6f complex in the stt7-1 1 kinase mutant. The observations in Fig. 1 that show different AA effects in oxic and 2 anoxic WT raised the question whether the PQ pool redox state governs the (partially 3 reversible) AA sensitivity of the b6f, and whether active modulation via the Stt7 kinase is 4 involved. To test this, we examined the stt7-1 strain under anoxic conditions in Fig. 2  5 since the mutant is fully devoid of the kinase 10 . Interestingly in anoxic stt7-1, AA 6 treatment had a milder inhibitory effect on b6f redox reactions after the flash (Fig. 2a). 7 Accordingly, the b-phase in the ECS kinetics was less affected (Fig. 2b). When 8 monitoring photosynthetic electron transfer during the 10-s illumination in the stt7-1 9 control (Fig. 2c), the initial rate of ~240 charge separations/s/PSI resembled WT control 10 (cf. Fig. 1f). The mutant, however, did not sustain high initial rates up to (at least) 10-ms 11 light where rates have dropped to ~200 charge separations/s/PSI already. Thereby, stt7-12 1 levelled off immediately to establish a quasi steady state of ~40 charge 13 separations/s/PSI after ~200-ms in the light, which was lower than WT. The initial rates 14 and the steady state in AA-treated stt7-1 were very similar as in the control, with 15 exception of slightly lower rates during the transition between 10-ms to 300-ms. To 16 summarize the conditional AA effect in WT b6f and the data of stt7-1, the rate constants 17 for cyt.f reduction (kf-red, Fig. 2d) and b-haems oxidation (kb-ox, Fig. 2e) are shown. This 18 data served as an estimate as to whether the b6f was limiting during the induction phase 19 of photosynthesis shown in Figs. 1c, 1f and 2c. Apart from the above-mentioned slight 20 delay of re-reduction, kf-red was not significantly altered in anoxic WT (~320 s -1 ) compared 21 to oxic cells (~350 s -1 ). The presence of AA in oxic WT samples produced an insignificant 22 accelerating effect on kf-red. Moreover, kf-red in anoxic WT was faster than in the absence 23 of Stt7 kinase (~190 s -1 ) and was severely slowed down in the presence of AA (~40 s -1 ). 24 The AA-induced slowdown of kf-red was insignificant in stt7-1 (~150 s -1 ). As reported 25 previously 20 , anoxic WT in Fig. 2e showed slightly faster kb-ox (~160 s -1 ) compared to oxic 26 samples (~100 s -1 ). Anoxic stt7-1 (~90 s -1 ) was significantly slower than WT. AA did not 27 produce a significant inhibitory effect on kb-ox in stt7-1 (~80 s -1 ) which stood in contrast 28 with the AA inhibition in WT (~40 s -1 ). 29 1 Fig. 2 The antimycin-A sensitivity of the cytochrome b6f complex was less pronounced in 2 anoxic stt7-1 cells. The light-adapted samples were kept in darkness for 30-s before the 3 measurements. Representative kinetics and, where indicated, means of three biological 4 replicates (± SD) are shown. a The b6f redox kinetics are shown and resemble anoxic 5 WT signals of Fig. 1d. At variance with WT, AA treatment produced only a slight 6 slowdown of cyt.f reduction and b-haems oxidation in stt7-1. b Accordingly, the 7 electrogenic b6f-phase of the ECS signals was less affected by AA, c as well as the 8 electron transfer rate (ETR) during re-illumination in the presence of AA. The stt7-1 was 9 less capable to sustain WT-like ETR after several ms of light and produced a lower 10 steady state, compared to Fig. 1f. The b6f kinetics analysis of Figs. 1 and 2 is 11 summarized in panels d and e (see Materials and Methods). Unless marked, P is 12 indicated in italics (two-tailed Student's t-test **P < 0.005 and ***P < 0.0005). d The cyt.f 13 reduction rate constants (kf-red) showed that only anoxic WT was inhibited by AA and kf-red 14 in anoxic stt7-1 controls was slower than WT. e The apparent oxidation rate constants for 15 the b-haems (kb-ox) showed that anoxic condition had an accelerating effect in WT 16 controls which coincided with AA susceptibility. In stt7-1 anoxic controls, kb-ox was slower 17 than in WT and not significantly affected by AA. 18

19
The cytochrome b6f complex in pgr5 is sensitive to antimycin-A. After demonstrating 1 AA inhibition in the anoxic WT b6f, and linking the effect to the Stt7 kinase and the PQ 2 pool redox state, we followed our initial study 20 and questioned whether the AA sensitivity 3 of WT under CEF-favouring conditions is linked to PGR5 function. In other words, is the 4 PGR5 polypeptide involved in forming the AA-sensitive FQR machinery, as suggested in 5 a widely accepted model for this type of CEF 22, 23 ? To test this, cells were grown in weak 6 light to prevent PSI photodamage in pgr5 21, 33 . As expected, the low-light anoxic WT 7 showed an AA effect on the level of b6f kinetics (Fig. 3a) as well as the electrogenic b-8 phase after the flash ( Supplementary Fig. 1a). The same applied for electron transfer 9 transients during photosynthetic induction (Fig. 3b), except that slightly higher apparent 10 steady state rates of ~100 charge separations/s/PSI were obtained after ~300-ms in the 11 WT control (cf. Fig. 1f for WT grown in moderate light). When measuring anoxic pgr5, we 12 also observed a slowdown of b6f kinetics in the presence of AA (Fig. 3c). However, the b-13 haems oxidation amplitude was almost not diminished in pgr5 AA samples, unlike the 14 electrogenic b-phase amplitude ( Supplementary Fig. 1b). The control pgr5 failed to 15 sustain efficient electron transfer rates which was evidenced here by the low virtual 16 steady state of ~35 charge separations/s/PSI (Fig. 3d). Nevertheless, during the initial 17 phase up to (at least) 10-ms of illumination, the mutant resembled WT and high rates of 18 ~235 charge separations/s/PSI were observed in pgr5 controls. As expected from the AA 19 effect on the b6f (Fig. 3c), electron flow was severely affected by the inhibitor in pgr5 (Fig.  20 3d), which resembled WT during the first 100-ms of light and levelled off at only ~25 21 charge separations/s/PSI. 22 The 30-s dark period served to dark-relax the system before measurements, i.e. restore 23 the availability of electron acceptors/donors, diminish the pH-dependent electron flow 24 downregulation, etc. However, various dark processes influence the PQ pool redox state 25 such as the type II NAD(P)H dehydrogenase which has an activity of ~2 electrons/s 34, 35 .

26
In the absence of oxygen, i.e. the substrate for plastid terminal oxidase, this should 27 favour PQH2 production in the dark which may have influenced our measured initial 28 electron transfer rates. To exclude this potential source of additional electrons in the 29 chain before the measurements, the b6f and photosynthetic induction were analysed in 30 the same WT and pgr5 samples after a short dark period of 700-ms as well. Compared  31 to the 30-s dark ECS (i.e. ), the values at 700-ms were similar in the controls 32 ( Supplementary Fig. 1c). After 700-ms, pH was likely not completely collapsed. 33 Likewise, re-equilibration of the redox carriers was still ongoing. Therefore, only the first 34 100-ms after the flash are shown for anoxic WT in Fig. 3e. The cyt.f re-reduction to pre-35 flash levels was finished no later than 20-ms during which the b-haems displayed a net 36 oxidation, despite electrogenic charge transfer from haem bl red to haem bh ox (see inset in 37 Supplementary Fig. 1c). AA treatment slowed down both kinetics (Fig. 3e). Re-38 illumination after the short dark period yielded high initial rates of ~225 charge 39 separations/s/PSI in WT controls after 10-ms light, which levelled off at ~85 charge 40 separations/s/PSI (Fig. 3f). The AA samples were identical, compared to 30-s dark (cf. 41 Fig. 3b). After 700-ms darkness, the flash-induced b6f turnover in pgr5 controls differed 42 from WT by showing a slightly slower cyt.f re-reduction and absence of large net redox 43 changes in the b-haems (Fig. 3g) when electrogenic events took place in the b6f (see 44 inset in Supplementary Fig. 1c). As shown in Fig. 3h, the initial rates after 10-ms re-45 illumination were not as high when pgr5 controls experienced the short dark period, 46 yielding ~155 charge separations/s/PSI (cf. Fig. 3d). Again, AA treatment in pgr5 47 inhibited the induction and maintenance of photosynthetic electron transfer during the 10-48 s re-illumination. 49 1 Fig. 3 The conditional sensitivity of the cytochrome b6f complex to antimycin-A is 2 displayed in pgr5. Light-adapted WT and mutant cells experienced both 30-s (panels 3 ad) and 700-ms darkness before the measurements (panels eh). Representative 4 kinetics and, where indicated, means of three biological replicates (± SD) are shown. a 5 The b6f kinetics in anoxic WT and the AA effects resemble Fig. 1d despite different 6 growth conditions. b This also applied for the light-induced electron transfer rate (ETR) 7 although they developed a slightly higher steady state compared to Fig. 1f. c The b6f 8 kinetics in anoxic pgr5 resembled WT from panel a. Both, the cyt.f re-reduction and b-9 haems oxidation were slowed down by AA. At variance with WT, the b-haems amplitude 10 was not altered in the presence of AA. d The mutant managed to produce WT-like ETR, 11 at least during the first 10-ms of light, followed by a rapid ETR drop to a steady state 12 lower than in WT. AA treatment had an effect on the induction phase, less on the steady 13 state. e Compared to panel a, the short dark-relaxation of 700-ms resulted in slightly 14 modified b6f kinetics in anoxic WT which mainly involved a smaller b-haems oxidation 15 amplitude. f ETR after 10-ms light was almost as high as in panel b but it declined more 16 rapidly with ongoing illumination. g The 700-ms b6f signals in pgr5 differed from WT by 17 showing slightly slower cyt.f reduction and almost no net redox changes of b-haems (for 18 electrogenic events see Supplementary Fig. 1). h Compared to panel d, a significantly 19 diminished ETR was seen in pgr5 already with the first record at 10-ms light. 20 To summarize the b6f characterization of low-light cultures in Fig. 3, Fig. 4a shows that 1 the AA-induced slowdown of kf-red after 30-s dark was similar in WT and pgr5. The 2 corresponding kb-ox are shown in Fig. 4b. In line with a previous report 20 , kb-ox in the pgr5 3 control was slower than in WT (Fig. 4b). However, AA treatment yielded similar low kb-ox 4 in both strains. When cells experienced only 700-ms darkness in Fig. 4c, kf-red was similar 5 in WT controls (cf. Fig. 4a for 30-s dark). This was unexpected and suggests weak 6 photosynthetic control in low-light WT cultures, which could explain the above-mentioned 7 high initial rates of ~225 charge separations/s/PSI after 700-ms dark (Fig. 3f). However, 8 kf-red in pgr5 controls was slower than in WT after 700-ms darkness (Fig. 4c). The pgr5 kb-9 ox parameter could not be obtained after 700-ms darkness (cf. Fig. 3g) since b-haems 10 reduction and oxidation kinetics (see b-phase in Supplementary Fig. 1c) nullified net 11 redox changes independently of AA. 12 Independent of PGR5 but related to the above-mentioned weak photosynthetic control in 13 low-light WT cultures, we noticed that WT cells responded to some extent differently after 14 a 700-ms dark period when they were grown in moderate light, i.e. 40 µmol 15 photons/m 2 /s. Therefore, the moderate light cultures from Figs. 1 and 2 were also 16 monitored after 700-ms darkness ( Supplementary Fig. 2). As expected 20 , there was no 17 net oxidation of b-haems in relation to the pre-flash level in oxic WT cells (Supplementary 18 Fig. 2a). The transient electron transfer re-acceleration before 500-ms of illumination was 19 missing after the short dark period (Supplementary Fig. 2b; cf. Fig. 1c). The major 20 difference between anoxic cultures from the two light growth conditions was a less 21 pronounced net b-haems oxidation phase in WT from moderate light (Supplementary 22 Fig. 2c; cf. Fig. 3e) that coincided with a significantly slower onset of electron transfer in 23 the control samples ( Supplementary Fig. 2d; cf. Fig. 3f). Apart from the missing AA 24 effect, stt7-1 controls resembled WT on the levels of b6f kinetics and electron transfer 25 rates ( Supplementary Figs. 2e and 2f). This stood in contrast to the above-mentioned 26 stt7-1 phenotype after 30-s darkness. Fig. 4d shows the kf-red after 700-ms dark in 27 moderate light cultures and, when compared with Fig. 2d (showing kf-red after 30-s dark in 28 those cultures), a strong photosynthetic control resulted in slowdown of kf-red for all 29 control samples. For instance, kf-red in oxic WT was slowed down from ~350 s -1 to ~140 s -30 1 , i.e. by a factor of 2.5 when cells dark-relaxed for 700-ms. Accordingly, the higher kf-red 31 in anoxic WT after 30-s resembled the invariably low anoxic stt7-1 when the cells dark-32 relaxed for 700-ms only. This could explain the disappearing differences between WT 33 and stt7-1 on the electron transfer rates after 700-ms darkness (Supplementary Figs. 2d 34 and 2f). 35 1 Fig. 4 The b6f kinetics analysis of Fig. 3 is summarized and cyt.f reduction from samples 2 in Figs. 1 and 2 are characterized after 700-ms darkness. Means of three biological 3 replicates are shown (± SD) and P is indicated in italics (two-tailed Student's t-test). a 4 The cyt.f reduction rate constants (kf-red, see Materials and Methods) show that both 5 strains from Fig. 3 were inhibited by AA after 30-s darkness to similar extents. b 6 Accordingly, the apparent oxidation rate constants for the b-haems (kb-ox) showed the 7 same trend. In pgr5 controls, kb-ox was slower than in WT. c The kf-red after 700-ms 8 darkness were slower in pgr5 controls. A significant AA effect was seen in WT only. The 9 kf-red in WT controls was comparable to panel a. d The 700-ms kf-red parameters of 10 moderate light cultures are shown. The short dark period slowed down kf-red in all 11 controls, except the stt7-1 which was constantly slower unless AA was present (cf. Fig.  12 2d). 13

Discussion 1
The b6f has been initially suggested as the AA-sensitive FQR component 26,27,29,36 which 2 was later revised 24 . The important discovery of the PGR5 interactome 22, 37 recognized 3 PGR5/PGR5-Like1 as the AA-sensitive FQR for PGR5-dependent CEF, suggested by in 4 vitro measurements. Here, we presented in vivo evidence that the cellular redox poise 5 may equally dismiss and qualify the b6f as AA-sensitive FQR component in the same 6 biological sample (Fig. 1). Although we also observed mild AA effects in the chloroplast 7 that were associated with respiration inhibition in oxic samples (see also Supplementary  8 Text 1), the contrasting AA effect on the b6f in anoxic conditions could explain the above-9 mentioned conflicting results when hunting for the AA-sensitive FQR in isolated 10 chloroplasts. More specifically, the strongly reduced PQ pool activated the Stt7 kinase 11 which itself was required for the AA-sensitive b6f feature (Fig. 2). Our data in Figs. 3 and 12 4 ruled out that PGR5 was responsible for both AA sensitivity of the b6f, and efficient 13 electron transfer at the very onset of photosynthesis. 14 This challenges the above-mentioned PGR5/PGR5-Like1 model 23 . Instead, a recent 15 proposal is substantiated 20 which links the PGR5 polypeptide to the b-haems redox 16 reactions in CEF-favouring conditions. It was suggested that PGR5 facilitates access to 17 stromal electrons for the b6f Qi-site. In this scenario, supported by biochemical data 21 , 18 PGR5 could be involved in efficient FNR recruitment to the thylakoid membrane (on the 19 levels of PSI and b6f). Moreover, the b6f was proposed to operate in two types of Q 20 cycles 20 : a canonical type during LEF and a Fd-assisted Q cycle during CEF. The latter 21 Q cycle concept has been introduced elsewhere and early reports tied it to AA inhibition 22 of Fd-dependent cyclic photophosphorylation 29,36,38,39 . The dysfunctionality of the pgr5 23 b-haems redox reactions became more apparent in Figs. 3 and 4 when only a short dark 24 period was given for equilibration before the flash-induced b6f turnover. Only WT showed 25 a fast b-haems net oxidation (Fig. 3e) and we interpret the flat pgr5 b-haems signals (Fig.  26 3g) as an inefficient replenishment of reduced Fd to assist in the modified Q cycle. 27 Accordingly, a pre-reduced haems bh red /ci red couple in anoxic WT may produce two PQH2 28 at the Qi-site per Qo-site turnover after a flash by utilizing Fd 20 . Only a fraction of WT b6f 29 might be engaged in this reaction after 700-ms darkness, considering the small 30 amplitude compared to 30-s dark (cf. Figs. 3a and 3e). The 700-ms b6f kinetics in pgr5 31 (Fig. 3g) are unlikely to be linked to enhanced photosynthetic control (lowering kf-red, Fig.  32 4c) but rather to inefficient b-haems oxidation in anoxic conditions. The latter eventually 33 hampers the FeS domain movement of the Rieske ISP to prevent intrinsic short-circuits 34 in the high-potential chain 40, 41 . Moreover, the diminished pmf in the mutant under these 35 conditions 20 , evidenced by the lower steady state electron transfer rates (Fig. 3), was not 36 in favour of enhanced photosynthetic control in pgr5. Another observation that argued 37 against this was the WT-like ATP synthase activity upon a flash ( Supplementary Fig. 2). 38 Fd shortage in pgr5 may be less severe after prolonged dark-relaxation and, on average, 39 the b6f population produced only a small, yet significant kb-ox difference after 30-s 40 darkness (Fig. 4b). Under these conditions, the observed high initial electron transfer in 41 anoxic pgr5 resembled WT at least up to 10-ms of illumination (Figs. 3b and 3d). The 42 comparable initial rates might be distantly related to findings in Arabidopsis pgr5 showing 43 WT-like CEF during the induction phase, despite a different CEF regulation capacity in 44 pgr5 42 . It was demonstrated in Chlamydomonas that Fd-binding proteins PETO and 45 ANR1 interact with the b6f, and at least PETO has access to the stromal b6f region 12, 43, 46 44 . If these interactions are PGR5-independent, a pool of Fd may be retained in proximity 47 of the algal Qi-site by these auxiliary CEF proteins. Thus, despite modified FNR 48 recruitment in pgr5 21 , the auxiliary Fd-binding proteins could accommodate CEF for a 1 limited number of turnovers which resulted in WT-like initial electron transfer rates after a 2 longer dark equilibration period. 3 It is important to note that our study only indirectly addressed CEF efficiency, judged 4 from the AA inhibition of total electron transfer. We did not discriminate between LEF and 5 CEF with intent (see also below). Although AA binding/unbinding may occur during 6 continuous illumination, as suggested for other b6f Qi-site inhibitors 45 , electron transfer 7 transients in the presence of AA should be mainly a function of LEF efficiency. The 8 transients depended on the number of electrons accumulated in the chain during the 9 700-ms/30-s dark interval and the number of the active (AA-insensitive) b6f in the 10 sample. A weakness of CEF estimations by using AA is that LEF efficiency may vary in 11 controls and AA samples on the level of PSII. In theory, PSII acceptor side limitation is 12 modulated upon disproportionation of PQ reduction by PSII and PQH2 oxidation by the 13 diminished functional b6f population (in the presence of AA). The functional b6f population 14 (in the absence of AA) is lower in pgr5 under CEF-favouring conditions so that this 15 disproportionation may be an intrinsic mutant feature, resulting in low LEF efficiency (Fig.  16 3) and a higher proportion of closed PSII centres 20 . 17 Here, we report for the first time that the b6f represented a bottleneck in anoxic stt7-1 as 18 well, which was possibly linked to a modified AA affinity in the Qi-site. When comparing 19 with WT, the lower b6f activity in stt7-1 could result from an inefficient Qi-site turnover (in 20 analogy to pgr5 above). Since the stt7-1 steady state electron transfer rate was lower 21 than in WT (independently of the dark adaptation period), the prevailing pH was 22 presumably higher in WT after 700-ms darkness. Considering the significant 23 photosynthetic control in anoxic WT (Figs. 2d and 4d), the 700-ms behaviour of WT was 24 disregarded to interpret stt7-1 since the majority of the pH was expectedly consumed 25 only after 30-s darkness. Accordingly, the diminished electron transfer at the very onset 26 of light was more pronounced when stt7-1 cells dark-relaxed for 30-s (cf. Figs. 1f and 2c) 27 instead of 700-ms (cf. Supplementary Figs. 2d and 2f). In contrast to pgr5 with 28 insufficient access to Fd at the Qi-site, we propose that the effect in anoxic stt7-1 was 29 due to incomplete switching to the Fd-assisted Q cycle. Stt7 interacts with the b6f 9 and 30 phosphorylates the complex directly 10 . Compared to pgr5, absence of the kinase may 31 have more severe consequences on the b6f function in the dark-relaxed state. Therefore, 32 a significant slowdown of kf-red in addition to kb-ox was observed in anoxic conditions after 33 30-s darkness (cf. Figs. 2d and 4a), as well as a significant drop of electron transfer in 34 the first 10-ms of light (cf. Figs. 2c and 3d). This distinguished stt7-1 from pgr5 in addition 35 to the AA effect. Since stt7-1 was not fully AA resistant, it could be that a certain AA 36 susceptibility was imposed by our conditions or that additional reactions were 37 contributing to the b6f modulation, which could also be a function of the PQ pool redox 38 state. 39 Nevertheless, we link these Stt7-dependent b6f adjustments to more efficient (total) 40 electron flow which is AA-sensitive and the major contributing process during the first 41 100-ms of light (cf. Figs. 1f and 2c). This initial phase is presumably dominated by CEF 42 although a recent analysis in the Chlamydomonas stt7-9 mutant revealed that CEF was 43 not affected in the mutant 15 . It should be noted that stt7-9 is a leaky strain, unlike stt7-1 44 used here 10 . Thus, stt7-9 still contained the phosphorylation(s) in the b6f that might be 45 responsible for the AA sensitivity 10 . We showed that candidate phosphorylation sites are 46 found in the stromal region of the interacting subunit-IV and PETO, the loosely attached 47 CEF effector 10,11,12,16 . Furthermore, the CEF rates in stt7-9 were obtained in the 48 presence of DCMU 15 . The major methodical bottleneck of CEF measurements is that 1 CEF shares several electron carriers with LEF. One disentanglement strategy is to 2 abolish CEF by inhibitors such as AA. Potential drawbacks on LEF efficiency have been 3 outlined above. Another strategy to measure CEF is to abolish LEF, using PSII inhibitors 4 such as DCMU. In our study, it was intended to benefit from the PQ reductase activity of 5 PSII to ensure maximal PQ pool reduction throughout the illumination regime in anoxic 6 cells. Accordingly, we did not make further attempts to disentangle LEF and CEF since 7 we also noticed in our conditions that PSII activity contributed to AA sensitivity of the b6f 8 (b-phase in Fig. 1e) and suppression of electron flow in the presence of AA (Fig. 1f). This 9 may reflect the PSII-dependent redox poise during CEF which has been recognized and 10 discussed elsewhere 46,47 . Therefore, AA-sensitive CEF measured in the absence of PSII 11 photochemistry (or in PSI-specific far-red light) bears the risk of underestimating true 12 maximal (PSII-poised) rates. In the absence of PSII activity, an increased AA-insensitive 13 b6f population will eventually feed on the FQR activity of the proportionally diminished 14 AA-sensitive b6f population, which should be in favour of non-cyclic processes to 15 compete for the electron downstream of PSI.  1 and  6 2). The growth light for WT and pgr5 was set to 10 µmol photons/m 2 /s at otherwise 7 identical conditions (Fig. 3). Grown cultures were diluted ~6-fold at least once after 8 inoculation and grown to a density of ~2 × 10 5 cells/ml before harvesting (5000 rpm, 5 9 min, 25°C). Cells were resuspended at 20 µg chlorophyll/ml in TP supplemented with 10 20% (w/v) Ficoll. Oxic samples were resuspended throughout the measurements in 2-ml 11 open cuvettes. For oxygen-deprived conditions, cells were supplemented with 50 mM 12 glucose, 10 U glucose oxidase and 30 U catalase in the cuvette, and then overlaid with 13 mineral oil for 20 min in the dark. The enzyme stock solutions to obtain anoxic samples 14 were freshly prepared. 15 In Vivo Spectroscopy. Both oxic and anoxic samples were light-adapted in the cuvette 16 for 20 min before the measurements of the electrochromic shift (ECS) and the 17 cytochrome b6f redox kinetics. The measurements were obtained using a JTS-10 18 (Biologic) and are described in detail elsewhere 20 . In brief, all absorption changes were 19 normalized to the ECS I/I signals that were produced (<300 µs) after a saturating laser 20 flash in the presence of 1 mM hydroxylamine (HA, from 1 M aqueous stock) and 10 µM 21 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU, from 10 mM ethanolic stock). Thus, HA 22 and DCMU were used as PSII inhibitors to obtain the density of active PSI centres in the 23 sample (measured as 1 charge separation/PSI). For b6f measurements, the flash 24 intensity was lowered (to avoid double turnovers) which can be seen in Fig. 1b. There, 25 both photosystems (with a ratio of ~1:1) were actively trapping photons and, when 26 referring to the light-induced ECS amplitude (first records at 0.3-ms), the flash produced 27 an accumulated ~0.7 charge separations/PSI. Once one photon trap (PSII) was inhibited 28 in the presence of AA, about half the PSI population was excited by the flash (grey 29 symbols in Fig. 1b). LEDs emitting ~150 µmol photons/m 2 /s of 630-nm actinic light (AL) 30 were used for light adaptation in the cuvette. The AL was interrupted for 34-s during 31 which two sub-saturating laser flashes were fired at 700-ms and 30-s. After these 32 flashes, cytochrome f (554-nm -0.27×(563-nm -546-nm)) and the haems bl/bh (563-nm) 33 were measured, using a baseline drawn between 573-nm and 546-nm 20 . The 34 interference filters were altered in a fixed order and the first set of records was dismissed 35 after continuous AL adaptation to account for the altered light/dark regime. AL was on for 36 ~1-min in between each dark period-containing measurement sequence. For 37 calculations, means of six technical replicates were used for each wavelength. The dark 38 phases were fit with a single-exponential function after the flash to obtain rate constants 39 for cytochrome f reduction (kf-red, fit from 0.3-ms to 100-ms in oxic samples and 1-ms to 40 100-ms in anoxic samples) and b-haems oxidation (kb-ox, from 1-ms to 500-ms which had 41 to be shortened in some anoxic samples owing to an established dark re-reduction 42 phase). The standard errors of kf-red and kb-ox were comparable for each biological 43 replicate whereas inhibition reduced the fit quality occasionally (Supplementary Fig. 3).

44
Antimycin-A (AA, Sigma, Lot#061M4063V from 2011 and Lot#079M4102V from 2020 45 were used) was prepared freshly as 40 mM ethanolic stock and incubation at 40 µM in 46 AL was for 30-min. Where indicated, these cells were illuminated for another 20-min in 47 the absence of PSII photochemistry. The ECS (520-nm -546-nm) was also 48 deconvoluted in the dark pulse method where AL was shuttered briefly during a 10-s AL 1 induction phase 39, 51, 52 . As for the b6f dark periods above, AL-adapted cells were 2 measured after 700-ms and 30-s darkness (four technical replicates with 1-min AL in 3 between). Two separate measuring sequences were used to avoid accumulating effects 4 of shortly spaced dark intervals at the beginning of the induction phase 51 . AL was 5 switched off at t = 0-ms in the sequences so that the linear ECS slopes before (SL from -6 4-ms to -1-ms) and after darkness (SD from 1-ms to 5-ms) gave electron transfer rates 7 before darkness (SL -SD, expressed in separations/PSI/s). Four detecting pulses, 8 spaced by 1-ms, recorded the ECS for slope calculations. The shown rates at 1-ms 9 illumination were calculated from separate slopes obtained during 2-ms AL in light-10 adapted cells after 30-s dark (15 replicates spaced by 5-s