Cytochrome cM downscales photosynthesis under photomixotrophy in Synechocystis sp. PCC 6803

Photomixotrophy is a metabolic state, which enables photosynthetic microorganisms to simultaneously perform photosynthesis and metabolism of imported organic carbon substrates. This process is complicated in cyanobacteria, since many, including Synechocystis sp. PCC 6803, conduct photosynthesis and respiration in an interlinked thylakoid membrane electron transport chain. Under photomixotrophy, the cell must therefore tightly regulate electron fluxes from photosynthetic and respiratory complexes. In this study, we show via characterization of photosynthetic apparatus and the proteome, that photomixotrophic growth results in a gradual reduction of the plastoquinone pool in wild-type Synechocystis, which fully downscales photosynthesis over three days of growth. This process is circumvented by deleting the gene encoding cytochrome cM (CytM), a cryptic c-type heme protein widespread in cyanobacteria. ΔCytM maintained active photosynthesis over the three day period, demonstrated by high photosynthetic O2 and CO2 fluxes and effective yields of Photosystem II and Photosystem I. Overall, this resulted in a higher growth rate than wild-type, which was maintained by accumulation of proteins involved in phosphate and metal uptake, and cofactor biosynthetic enzymes. While the exact role of CytM has not been determined, a mutant deficient in the thylakoid-localised respiratory terminal oxidases and CytM (ΔCox/Cyd/CytM) displayed a similar phenotype under photomixotrophy to ΔCytM, demonstrating that CytM is not transferring electrons to these complexes, which has previously been suggested. In summary, the obtained data suggests that CytM may have a regulatory role in photomixotrophy by reducing the photosynthetic capacity of cells. One sentence summary The cryptic, highly conserved cytochrome cM completely blocks photosynthesis in Synechocystis under three days of photomixotrophy, possibly by suppressing CO2 assimilation.


Introduction
Switching between different trophic modes is an advantageous feature, which provides great metabolic flexibility for cyanobacteria. For a long time, these photosynthetic prokaryotes were considered as a group of predominantly photoautotrophic organisms (Smith 1983, Stal andMoezelaar 1997). Lately, accumulating evidence marks the physiological and ecological importance of trophic modes involving organic carbon assimilation, e.g. photomixotrophy (Zubkov andTarran 2008, Moore et al 2013). Dissolved organic carbon, most notably monosaccharides, glucose and fructose, accumulates in the environment, mainly during phytoplankton blooms (Teeling et al 2012, Ittekot et al 1981. During photomixotrophy, photosynthetic organisms must balance the consumption of organic carbon sources with photosynthesis and carbon fixation.
In the model cyanobacterium, Synechocystis sp. PCC 6803 (hereafter referred to as Synechocystis), photomixotrophy is further complicated by the operation of anabolic and catabolic processes occurring in the same cellular compartment and by the presence of an interlinked thylakoid membrane-localised electron transport pathway involved in both photosynthesis and respiration (Vermaas et al. 2001, Mullineaux 2014. In Synechocystis, photosynthetic linear electron flow is similar to other oxygenic photoautotrophs. In photosystem (PS) II and PSI, the energy of the harvested photons induces charge separation. Electron from the PSII primary donor P680 passes via pheophytin and the primary quinone Q A , to the secondary quinone, Q B . Oxidized P680 + is the strongest biological oxidizing molecule, which drives water splitting on the lumenal side of PSII. When Q B is doubly reduced, it binds two protons from the cytosol, converting plastoquinone (PQ) to plastoquinol (PQH 2 ), then diffuses into the membrane PQ pool. Cytochrome (Cyt) b 6 f receives two electrons and transfers an electron to the mobile small protein, plastocyanin (Pc) or cytochrome c 6 (Cyt c 6 ). An electron is subsequently transferred to PSI, replacing a newly excited electron that is transferred from the PSI reaction center P700 + via several co-factors to ferredoxin (Fed). Lastly, electrons are transferred from Fed to NADP + by ferredoxin-NADP + reductase (FNR) to generate NADPH. In the respiratory electron transfer pathway, PQ is reduced by NAD(P)H dehydrogenase-like complex I (NDH-1) and succinate dehydrogenase (SDH), using electrons ultimately derived from Fed (Shuller et al 2019) and succinate, respectively. Electrons from PQ-pool can be transferred to a thylakoid-localized respiratory terminal oxidase (RTO), cytochrome bd-quinol oxidase (Cyd), or via Cyt b 6 f then Pc/Cyt c 6 , to a second RTO, an aa 3 -type cytochrome-c oxidase complex (Cox). How Synechocystis regulates electron input from PSII and the NDH-1 and SDH complexes into the photosynthetic electron transport chain and to respiratory terminal acceptors under photomixotrophic conditions, is not fully understood. Moreover, Synechocystis encodes four isoforms of flavodiiron proteins (FDPs), Flv1-4, which likely utilize NAD(P)H (Vicente et al. 2002, Brown et al 2019 or reduced Fed (Santana-Sanchez et al. 2019). These proteins function in light-induced O 2 reduction as hetero-oligomers consisting of either Flv1/Flv3 or Flv2/Flv4 (Helman et al. 2003, Mustila et al. 2016, Allahverdiyeva et al. 2015, Santana-Sanchez et al. 2019. In Synechocystis, a c-type Cyt, the water-soluble Cyt c 6 (formerly referred to as Cyt c 553 ), can substitute for Pc under conditions of copper deprivation (Durán et al 2004). Cyt c 6 belongs to the Cyt c family, whose members are characterized by a covalently bound c-type heme cofactor. C-type Cyts are further classified into groups such as the Cyt c 6 -like proteins, Cyt c 555 , Cyt c 550 , and CytM (Bialek et al. 2008). Apart from the well-established role of Cyt c 6 in electron transfer (Kerfeld and Krogman 1998) and the role of Cyt c 550 (PsbV) in stabilizing the PS II water splitting complex (Shen and Inoue 1993), most of the Cyt c proteins remain enigmatic.
CytM is conserved in nearly every sequenced cyanobacteria with the exception of the obligate symbionts, Candidatus acetocyanobacterium thalassa and Candidatus Synechococcus spongiarum (Supplemental Fig. S1, Bialek et al 2016). In Synechocystis, CytM is encoded by sll1245 (Malakhov et al 1994). Nevertheless, its subcellular location is ambiguous. An early study localised CytM to the thylakoid and plasma membranes in 'purified' membrane fractions (Bernroitner et al 2009). However, cross contamination between membranes was not determined, which has been an issue in studies using similar separation techniques (Sonoda et al 1997, Schultze et al 2009. In later proteomics studies, CytM has not been detected or localised using membranes purified by either two-phase aqueous polymer partitioning or subcellular fractionation (Baers et al 2019). However, the structure of the hydrophobic N-terminus resembles a signal peptide, which suggests that CytM is targeted to a membrane. Sequence similarity to the N-terminus cleavage site of Synechocystis Cyt c 6 suggests that the N-terminus is processed and the mature 8.3 kDa protein is inserted into the lumen (Malakhov et al 1994). However, cleavage does not seem to occur in vivo, as the protein extracted from various cyanobacterial species, including Synechocystis, Synechococcus elongatus PCC 6301 and Anabaena sp. PCC 7120, was found to be around 12 kDa (Cho et al 2000, Bernroitner et al 2009, implying that the hydrophobic N-terminus remains on the protein and serves as a membrane anchor. It has been suggested that CytM may play a role in respiratory or photosynthetic electron transfer (Vermaas et al 2001, Bernroitner et al 2009. In Synechocystis, CytM was shown to reduce the Cu A center of Cox in vitro with similar efficiency as Cyt c 6 (Bernroitner et al 2009). However, given the midpoint potential of CytM (+150 mV), electron transfer from Cyt b 6 f (+320 mV) to CytM would be energetically uphill (Cho et al 2000). Notably, CytM is unable to reduce PSI in vitro (Molina-Heredia et al 2002). Thus, it is difficult to see how the protein would substitute for Cyt c 6 or Pc. Importantly, CytM is not detected under photoautotrophic conditions (Baers et al 2019) and deletion of the gene does not affect net photosynthesis or dark respiratory rates (Malakhov et al 1994) under these conditions. Cold, high light and salt stress, however, induce gene expression and the stress-induced co-transcriptional regulation between cytM (CytM), petJ (Cyt c 6 ) and petE (Pc) suggests a stress-related role in electron transfer (Malakhov et al 1999).
Besides environmental stresses, CytM has been linked to organic carbon-assimilating trophic modes. A strain of Leptolyngbya boryana was found to grow faster than wild-type (WT) in dark heterotrophy. Genome re-sequencing revealed that the fast-growing strain harbours a disrupted cytM (Hiraide et al 2015). In line with this, the cytM deletion mutant of Synechocystis demonstrated a growth advantage over the WT under dark and light-activated heterotrophic conditions, and under photomixotrophic conditions (Hiraide et al 2015). Under dark heterotrophic conditions, Δ CytM had higher respiratory and photosynthetic activity. However, the physiological mechanism and the functional role of CytM remains entirely unknown.
In this study, we sought to uncover the physiological background behind the growth advantage of photomixotrophically grown Synechocystis Δ CytM by characterizing its photosynthetic machinery and the proteomic landscape. We demonstrate that a mutant lacking CytM circumvents over-reduction of the PQ-pool during photomixotrophic growth, enabling higher rates of net photosynthesis. In order to meet the substrate demand for enhanced growth, the mutant accumulates transporter proteins, cofactor biosynthetic enzymes and slightly adjusts central carbon metabolism. Although the function of CytM was previously associated with Cox, both thylakoid respiratory terminal oxidases, Cox and Cyd, were found to be dispensable for the metabolic advantage conferred by deletion of CytM in photomixotrophy. We conclude that when cells are exposed to high glucose conditions, CytM reduces the photosynthetic capacity and contributes to regulating the redox state of the intertwined photosynthetic and respiratory electron transport chain, in order to accommodate this new energy source. light. In line with previous studies (Malakhov et al 1994, Hiraide et al 2015, no growth difference was observed between Δ CytM and WT under photoautotrophic conditions ( Fig 1A).

Deletion of
Next, we characterized growth under photomixotrophic conditions. To determine how different starting glucose concentrations affected photomixotrophic growth (Fig. 1A, B), we supplemented the medium with 5 mM and 10 mM glucose and cultivated the strains under constant 50 µmol photons m light. Based on optical density measurements (OD 750 ), all cultures with added glucose grew substantially faster than those cultured photoautotrophically (Fig 1A, B). Deletion of cytM had no effect on cells grown at 5 mM glucose. However, when cultured with 10 mM glucose, Δ CytM demonstrated 1.9±0.4 (p=6E-6) higher OD 750 than WT and Δ Cox/Cyd/CytM demonstrated 1.9±0.6 (p=0.002) higher OD 750 compared to Δ Cox/Cyd, after three days. In line with this, Δ CytM consumed more glucose than WT ( Fig. 2A), as quantified by measuring the glucose concentration of the cell-free spent media on the third day of photomixotrophic growth.
The effect of deleting cytM was tested then by culturing strains under photomixotrophic conditions but with different light regimes (  constant light), when the highest difference in OD 750 was observed. The cell size, cell number per OD 750 and chlorophyll (chl) concentration per cell was determined. No difference was observed in cell size between Δ CytM and WT (Supplemental Fig. S3), and the cell number per OD 750 was similar in both strains (Fig. 2B), confirming that the difference in OD 750 reflects higher growth. However, the chl a content per cell increased in Δ CytM (Fig. 2C), suggesting that the photosystem content or PSII/PSI ratio has been altered in this strain. illumination supplemented with 10 mM glucose. Samples were taken on the third day.
Overall, the most pronounced growth advantage of Δ CytM over the WT was observed when cells were exposed to a light intensity of 50 µmol photons m − 2 s − 1 and glucose concentration of 10 mM. Therefore, these conditions were used for all subsequent phenotyping experiments examining cells cultured photomixotrophically. The same phenotype manifested in the triple Δ Cox/Cyd/CytM mutant, showing that Cox and Cyd are not required for the growth advantage. Moreover, we demonstrate that deletion of cytM leads to a higher cellular chl a content, which implies an altered photosynthetic machinery when cells are cultured photomixotrophically.

Deletion of CytM circumvents the gradual over-reduction of the PQ-pool in photomixotrophy
To determine how deletion of CytM may alter the photosynthetic machinery, we first assessed the activity of PSII by probing chl fluorescence (  illumination for three days, with or without 10 mM glucose, respectively. Prior to measurements, the cell suspension was adjusted to 15 µg chl ml To determine how over-reduction of the PQ-pool gradually increases in WT over three days of photomixotrophic growth, we monitored the redox kinetics of the PSII primary electron acceptor Q A (Fig. 4) by firing a single-turnover saturating flash on dark-adapted cells then relaxation of the chl fluorescence yield was recorded in the period of subsequent darkness.
No difference was observed between WT and Δ CytM cells cultured photoautotrophically (Supplemental Fig. S9) and on the first day of photomixotrophy, both WT and Δ CytM cells demonstrated typical flash-fluorescence relaxation in the darkness. On the second day, WT cells demonstrated a substantial slow-down in Q A -to-Q B electron transfer reflected by slow decay kinetics (Fig. 4B), while on the third day, there was almost a complete loss of Q A -to-Q B electron transfer (Fig. 4C). The kinetics from the third day resembled a curve recorded on photoautotrophically cultured WT supplemented with DCMU prior to the measurement (Supplemental Fig. S9). This supports the conclusion that Q A -to-Q B electron transfer was almost completely inhibited in WT on the third day of photomixotrophy.
Interestingly, Δ Cox/Cyd and Δ Cox/Cyd/CytM displayed pronounced waving in the fluorescence yield relaxation kinetics ( Fig. 4 A-C). The wave phenomenon is an unusual pattern in the decay of flash-induced chl fluorescence yield in the dark. The feature is characterized by a dip, corresponding to transient oxidation of Q A − and a subsequent rise, reflecting re-reduction of the PQ-pool by NDH-1 (Deák et al 2014). During growth over the three-day period, the wave phenomenon in Δ Cox/Cyd became less evident due to stronger inhibition of Q A -to-Q B electron transfer. In contrast, Δ Cox/Cyd/CytM displayed prominent waving during all three days of photomixotrophic growth, demonstrating that Q A − re-oxidation was being sustained. Slight waving was reported previously (Ermakova et al 2016), although to our knowledge, this is the first study demonstrating that glucose induces a strong wave phenomenon in Δ Cox/Cyd. Next, we analysed net photosynthesis by probing the O 2 production capacity of cells (Fig.  4D). When WT cells were grown photomixotrophically, only marginal net O 2 production was observed on the third day. Strikingly, in the presence of the artificial electron acceptor, 2,6dichloro-p-benzoquinone (DCBQ), the O 2 evolving capacity of WT was restored. DCBQ accepts electrons from Q A and/or Q B , virtually disconnecting PSII from the PQ-pool (Srivastava et al 1995). This suggests that PSII is fully functional in WT and that overreduction of the PQ-pool hindered PSII electron transfer to PQ-pool. On the contrary, DCBQ did not increase O 2 production in Δ CytM, implying that the PQ-pool is well oxidized. Immunoblotting performed on total protein extracts from WT and Δ CytM demonstrated a higher accumulation of PSII reaction center protein D1 in Δ CytM compared to WT (Fig 4E), suggesting that PSII is preserved in Δ CytM during the photomixotrophic growth.
These results demonstrate that during photomixotrophic growth, the PQ-pool gradually becomes over-reduced in WT, leading to drastically slower and eventually fully inhibited electron transfer from PSII to the PQ-pool on the third day. Deletion of CytM circumvents over-reduction of the PQ-pool and maintains PSII reaction center protein D1 amounts. , with DCBQ-treated WT considered as 100%. Values are means ± SD, n = three biological repeats. For numerically expressed rates of control samples, see Fig. 6. Immunoblot analysis with D1-N antibody (E) was performed on samples taken on the third day. 15 µg total protein extract was loaded per 100% lane, 50% and 200% corresponds to 7.5 µg and 30 µg, respectively.

CytM has a larger pool of oxidizable PSI than WT in photomixotrophy
Next, we determined activity of PSI by monitoring the redox kinetics of P700, the primary electron donor of PSI (Fig. 5), which was performed simultaneously with chl fluorescence measurements (Fig. 3). First, the maximal amount of P700, P m , was determined (Fig. 5A). Compared to cells cultured under photoautotrophic conditions, WT cells grown photomixotrophically had 45.2±0.03% lower P m . Interestingly, the difference between Δ CytM cultured under photomixotrophic and photoautotrophic conditions was negligible (17.2±19.3%). Thus, under photomixotrophic conditions, Δ CytM had 132±18.7% higher maximum amounts of oxidizable P700 than WT (Fig. 5A). In line with this, immunoblotting revealed higher levels of a reaction center subunit of PSI, PsaB, in Δ CytM compared to WT under photomixotrophic growth (Fig. 5B). To determine the PSI:PSII ratio, samples were analysed at 77K by measuring chl fluorescence emission. No statistical difference was observed between WT and Δ CytM (Supplemental Fig. S10), demonstrating that the PSII:PSI ratio is the same in both strains.
The PSI effective yield Y(I), was also quantified, and was three times lower in photomixotrophically cultured WT cells compared to those grown photoautotrophically (Supplemental Fig S5B). This is due to a strong donor side limitation of PSI Y(ND), which demonstrates electron shortage to P700 + (Supplemental Fig S5C). In contrast, photomixotrophically cultured Δ CytM demonstrated similar Y(I) and only slightly increased Y(ND) compared to photoautotrophically cultured WT and Δ CytM (Supplemental Fig. S5B, C). As a result, Δ CytM had more than three times higher Y(I) than WT in photomixotrophy (Supplemental Fig. S5B).
Next, pulse-induced P700 fast kinetics were compared between photoautotrophically and photomixotrophically cultured WT (Fig. 5E) and Δ CytM (Fig 5F). These fast kinetics reveal the dynamics of P700 oxidoreduction during saturating pulses on millisecond scale. Saturating pulses are flashed in darkness (P m D ) under far-red light (P m ) and actinic red light (P m '). Typically, photoautotrophically cultured WT (Supplemental Fig. S11A) demonstrates transient P700 + re-reduction during light pulses. However, photomixotrophically grown WT did not exhibit the typical transient re-reduction (see P m D on Fig. 5E). Importantly, P700 + relaxation after the pulse (Fig. 5E) was slower compared to that observed in photoautotrophically cultured cells (Supplemental Fig. S11A). Collectively, these results confirm that fewer electrons were transferred to P700 + , leading to higher Y(ND) in photomixotrophically grown WT. Photomixotrophically cultured Here, we have shown that the effective yield of PSI in photomixotrophically cultured WT cells was considerably lower compared to photoautotrophically cultured cells, due to an electron shortage at P700 + . This phenotype is eliminated by deleting cytM, as increased Y(I), higher amounts of oxidizable P700 (P m ) and PsaB was observed in Δ CytM compared to WT on the third day of photomixotrophy.

Figure 5. Characterization of PSI in cells cultured photomixotrophically.
The maximal amount of oxidizable P700, P m (A), and immunoblotting of PSI reaction center protein, PsaB (B), in cells cultured photomixotrophically. P700 oxidoreduction slow (C, D) and fast kinetics (E, F) was measured in parallel with fluorescence (Fig. 3). Fast kinetics curves (E, F) are normalized to P m and referenced against their respective minimum P700 signal detected after the pulse. Cultivation, sample preparation and experimental parameters are similar to those detailed in Fig. 3. P 0 , initial P700; P m D , maximum P700 in darkness; P m , maximum P700 under far-red light; P m ', maximum P700 under red actinic light. white light, O 2 consumption and gross production rates were similar, resulting in nearly zero net photosynthetic O 2 production. This is in line with the data obtained by O 2 electrode (Fig. 4D). Corresponding to the minor net photosynthetic O 2 production, the rate of CO 2 consumption was negligible (Fig. 6A,  Supplemental Fig. S12A). Importantly, no light-induced O 2 consumption was observed in WT ( Fig. 6A, C), although a substantial amount of Flv3 was detected by immunoblotting ( Fig 6B). Although the thylakoid-localized RTOs, Cox and Cyd, were shown to be active in light (Ermakova et

Δ
CytM exhibited a positive net O 2 production rate and active CO 2 consumption (Fig. 6E). Strikingly, gross O 2 production was approximately 10 times higher compared to WT and 18 O 2 consumption in light followed a triphasic pattern, a characteristic trend reflecting the contribution of Flv1/3 and Flv2/4 to O 2 consumption in light (Santana-Sanchez et al. 2019). The triphasic pattern in Δ CytM was observed as an initial burst of O 2 consumption following the dark-to-light transition, which faded after 1-1.5 min and continued at a relatively constant rate (Fig. 6C). Accordingly, immunoblotting confirmed higher accumulation of the Flv3 proteins in  Similar to WT, Δ Cox/Cyd (Fig 6C) showed barely any photosynthetic activity on the third day of photomixotrophic growth. During illumination, net O 2 production remained negative, and CO 2 consumption was found to be negligible (Fig. 6A, Supplemental Fig. S12B). Only residual gross O 2 production was observed and O 2 consumption was not stimulated by light (Fig 6A, D). Flv3 protein abundance in Δ Cox/Cyd was comparable to WT (Fig. 6B) Δ Cox/Cyd/CytM displayed a triphasic O 2 consumption pattern under illumination (Fig. 6F) and the light-induced O 2 consumption was comparable to that of Δ CytM in steady state (Fig. 6F). Compared to Δ Cox/Cyd, Δ Cox/Cyd/CytM had higher levels of Flv3 ( Fig 6B). Notably, deleting cytM in the Δ Cox/Cyd mutant did not enhance dark respiration, whereas Δ CytM had higher rates compared to WT.
To conclude, mutants lacking CytM sustained a steady electron flux towards O 2 and CO 2 in photomixotrophy, reflected by substantial net O 2 production and active CO 2 consumption under illumination both in Δ CytM and Δ Cox/Cyd/CytM.

CytM cells accumulate transport proteins and cofactor biosynthetic enzymes
In order to understand the metabolism of photomixotrophically grown WT and Δ CytM and how this affects the faster growth observed in the mutants, we analysed the total proteome by nLC-ESI-MS/MS via the data-dependent acquisition (DDA) method. Samples for analysis were collected on the second day, when both WT and Δ CytM cells were in late exponential phase and a significant growth difference was observed between the strains (Fig. 7A).
In total, 2415 proteins were identified (Supplemental Table S3), despite the fact that the dataset was slightly biased against basic (Fig. 7D) and hydrophobic proteins (Fig. 7E), which is a known issue with this technique (Chandramouli and Qian 2009). Out of 2415 proteins, 634 were quantified, with 162 displaying a statistically different abundance in Δ CytM compared to WT (FC >1.5 and FC < -1.5 (ANOVA p<0.05) (Supplemental Table S4). The functional classification of differentially regulated proteins (Fig. 7C) revealed that apart from unknown or hypothetical proteins, mainly transporters and biosynthetic enzymes were altered in photomixotrophically cultured Δ CytM cells. Table I shows a selection of proteins whose abundance was different in Δ CytM compared to WT. The highest fold change was observed in transport proteins. Among these, the constitutive low-affinity ABC-type phosphate transporters (PstA1, PstB1, PstB1', PstC), periplasmic P i -binding proteins (SphX, PstS1) and extracellular lytic enzymes (PhoA, NucH) are more abundant in Δ CytM. Among proteins related to C i uptake, a thylakoid β type carbonic anhydrase, EcaB, was 2.32 times (P=7.50E-03) more abundant in Δ CytM. EcaB is a CupA/B-associated protein, proposed to regulate the activity of NDH-1 3 (NDH-1 MS) and NDH-1 4 (NDH-1 MS') (Sun et al 2018). NDH-1 3 facilitates inducible CO 2 -uptake, whereas NDH-1 4 drives constitutive CO 2 -uptake (Ogawa et al 1991). CupB is exclusively found in the NDH-1 4 complex and converts CO 2 into HCO 3 − . Interestingly, no significant change was observed in the level of the glucose transporter GlcP, although the growth advantage of Δ CytM was glucose concentration-dependent.
Chl a biosynthetic enzymes were found to accumulate in the mutant (Table I). ChlL, a subunit of the light-independent protochlorophyllide reductase (Wu and Vermaas 1995), and ChlP (4.61E-03), a geranylgeranyl reductase (Shpilyov et al. 2005), were 9.28 fold (P=5.32E-03) and 1.52 fold (P=4.61E-03) upregulated in Δ CytM, respectively. The incorporation of chl into photosystems likely increases due to the elevated level of Pitt, a protein contributing to the formation of photosynthetic pigments/proteins at the early stages of biogenesis (Schottkowski et al 2009). The ligand of the tetrapyrrole ring of chl is Mg 2+ and accordingly, the magnesium uptake protein MgtE accumulated in Δ CytM along with a periplasmic iron-binding protein, FutA2, part of the complementary uptake-system of iron, a vital element of the photosynthetic machinery (Kranzler et al 2014). Among pigment biosynthetic enzymes, Δ CytM showed increased levels of the heme oxygenase Ho1, catalyzing the final step in the production of biliverdin (Willows et al 2000). Biliverdin is the precursor of phycocyanobilin, which is incorporated into phycobilisomes, the light-harvesting complexes of Synechocystis.
Among the photosynthetic proteins, the PSI reaction center subunit PsaB was found in equal amounts in WT and Δ CytM. However, immunoblotting with an anti-PsaB antibody demonstrated that Δ CytM contained higher amounts of PsaB than WT (Fig. 5B). This discrepancy may be due to the fact that despite the robustness of the MS-based DDA method, hydrophobic membrane proteins are prone to misquantification. Via MS analysis, quantification of psbA encoded D1 was not successful. Therefore, its abundance was only determined by immunoblotting (Fig. 4E), which revealed higher levels of D1 proteins in Δ CytM compared to WT. Interestingly and somewhat contradictorily, the amount of PSII assembly proteins encoded by the PAP-operon (Wegener et al 2008) decreased in the mutant. We also note that lower levels of NorB, a quinol-oxidizing nitric oxide reductase located in the plasma membrane (Büsch et al 2002), were observed in Δ CytM.
Since the growth advantage of Δ CytM is glucose-dependent, alterations are expected in the abundance of the intermediary carbon metabolic enzymes. In Synechocystis, roughly 100 enzymes participate in this metabolic network. In our study, 40 were quantified and surprisingly, only a few proteins were differentially regulated in Δ CytM. One notable example is phosphofructokinase PfkA, the key regulatory enzyme of the glycolytic Embden-Meyerhof-Parnas pathway, which was 1.86 times (P=1.96E-05) less abundant in Δ CytM, suggesting that carbon flux might be redirected into the Entner-Doudoroff or oxidative pentose phosphate pathways. Phosphoglycerate kinase Pgk, which is involved in each glycolytic pathway, was 2.06 times (P=1.27E-05) as abundant in Δ CytM. Phosphoenolpyruvate synthetase PpsA, a protein that catalyses the first step of gluconeogenesis, is 2.21 times (P=3.10E-04) less abundant in Δ CytM.
To conclude, global proteomic analysis revealed that photomixotrophically cultured Δ CytM accumulates transporter and chl biosynthetic proteins, while slight changes in the glycolytic and photosynthetic proteins were also observed. , with the ellipsis marking the sampling day. Cells were cultured similarly to those used in the biophysics analysis, except that the cells for proteomics were pre-cultivated under atmospheric CO 2 in order to fully adapt the cells to these conditions. Importantly, extra preculturing step did not affect the growth of the experimental cultures. Relaxation of the flashinduced fluorescence yield in the dark (A) was measured in the absence (closed symbols) and in the presence of 20 µM DCMU (open symbols). Differentially regulated proteins in Δ CytM were grouped by functional classification (C). In total, 2415 proteins were identified, out of which 634 proteins were quantified and 162 were differentially regulated. The practical significance of differentially regulated proteins was set to FC >1.5 and FC <-1.5 (ANOVA p<0.05). Effect of isoelectric point (pI) (D) and hydrophobicity (GRAVY) (E) of the proteins on the identification rate was determined. Black squares mark all the 3507 predicted proteins in Synechocystis, lilac circles mark each proteins identified in WT and in Δ CytM.

Discussion
The effect of importing and metabolising organic carbon on the bioenergetics properties of cyanobacteria over a long-term period is not fully understood. Previous studies have focused on the cellular changes following relatively short-term (from 10 min to 24 h) exposure to organic carbon (Lee et al 2007, Takahashi et al. 2008, Haimovich-Dayan et al 2011, Zilliges and Dau 2016. The majority of these reports suggest partial inhibition of photosynthetic activity, whereas some studies demonstrate increased net photosynthesis under air-level CO 2 after 2 h exposure to 10 mM glucose (Haimovich-Dayan et al 2011). However, longterm changes to bioenergetics processes, particularly photosynthesis, remain to be elucidated. In this study, we investigated the effect of long-term photomixotrophic growth on WT and Δ CytM cells, most notably on the photosynthetic machinery, by analysing chlorophyll fluorescence, the redox kinetics of P700, real time O 2 and CO 2 fluxes and changes within the proteome.

Gradual over-reduction of the PQ-pool limits photosynthesis in photomixotrophically cultured WT
By characterizing WT cells shifted from photoautotrophic to photomixotrophic conditions, we show that photosynthesis was markedly downscaled over three days of cultivation. This is deduced from the low PSII and PSI yield (Fig. 4C, D; Supplemental Fig. S5A, B) and most importantly, the negligible net O 2 production (Fig. 6A, C) and CO 2 fixation rates (Fig. 6A) on the third day. Since addition of an artificial PSII electron acceptor, DCBQ, restores O 2 evolving activity of PSII (Fig. 5D), we can conclude that the highly reduced state of the PQpool limits electron flow from PSII and consequently, water splitting. However, residual PSII activity is ensured by circulating electrons in a water-water cycle. This was demonstrated by residual PSII gross O 2 production ( Fig. 6 A, C), nearly equalling O 2 consumption in the light, resulting in practically zero net O 2 production. Interestingly, PSI and FDPs fail to oxidize the over-reduced PQ-pool. High donor side limitation at PSI (Supplemental Fig. S5C) shows that the photosynthetic electrons accumulate in the PQ-pool and are not transferred to PSI. These results suggest that photosynthetic electron flow is restricted downstream to PSII. Importantly, over-reduction of the PQ-pool increases gradually during photomixotrophic growth (Fig. 4). Based on thermoluminescence profiles, binding of quinones to PSII is delayed by glucose, although after two hours, the yield of PSII is not affected (Haimovich-Dayan et al 2011). We show here that after 24 hours growth (Fig. 4A), Q A − reoxidation kinetics are comparable to photoautotrophically grown cells (Fig. S9), indicating that the PQpool is well oxidized. Over-reduction of the PQ pool occurs over the next two days when Q Ato-Q B electron transfer becomes nearly completely blocked ( Fig. 4B and 5C).
The gradual downscaling of photosynthesis could be due to a number of factors including: (i) spatial isolation of PSII via rearrangement in the thylakoid to another location or (ii) posttranslational modifications of PSII. Rearrangement of thylakoid-localised complexes, specifically NDH-1 and SDH, has been observed in response to redox-regulated changes in the electron transport chain (Liu et al 2012). Applying the same analogy to PSII, the highly reduced state of the PQ-pool might trigger the complexes to arrange into a more sparse distribution during photomixotropic growth. Although cyanobacterial thylakoids are densely packed membranes (Kaňa et al 2013), lateral heterogeneity (Agarwal et al 2010) and mobility of PSII (Casella et al 2017) has been previously demonstrated. In the (ii) second scenario, post-translational modifications near the Q B -binding pocket hinders the effective oxidation of Q A − . D1 is phosphorylated at the C-terminus under certain conditions (Angeleri et al 2016), although the Q B -binding site resides at the N-terminus.

Δ
CytM Surprisingly, deletion of CytM reverses downscaling of photosynthesis in photomixotrophy, resulting in a profile similar to WT and Δ CytM cells grown under photoautotrophic conditions. Importantly, Δ CytM demonstrated unrestricted electron flow between PSII and PSI. The well oxidized PQ-pool allowed unrestricted Q A -to-Q B electron transfer (Fig. 5C) and the rate of gross O 2 production (Fig 6AE) was ten times higher in Δ CytM than it was in WT cells cultured under photomixotrophic conditions. Contrary to WT, Δ CytM did not demonstrate PSI donor-side limitation (Supplemental Fig. S5C). Finally, the abundance of D1 (Fig. 5E), PsaB (Fig.6B), and PetA and PetB (Table I), the core subunits of PSII, PSI and Cyt b 6 f, respectively, was higher in Δ CytM than in WT, although the PSI:PSII ratio was unaltered (Supplemental Fig. S10). As a consequence, the rate of net O 2 production and CO 2 consumption ( Fig. 6-7) was significantly higher in Δ CytM, demonstrating that deletion of CytM conserves photosynthetic activity and circumvents over-reduction of the PQ-pool in photomixotrophy.
The exact mechanism by which Δ CytM alleviates blockage of the electron transport pathway was not elucidated in this work, nor has an exact role for this protein been determined in previous studies. CytM has been suggested to play a role in transferring electrons from Cyt b 6 f to Flv1/3, limiting productivity but providing a possible alternative route for safely transferring electrons to O 2 (Hiraide et al 2015). However, given the low midpoint potential of CytM, a large energy barrier would have to be overcome in order for electron transfer downstream of Cyt b 6 f to occur (Cho et al 2000). Moreover, we demonstrated that the absence of CytM does not decrease O 2 photoreduction driven by FDPs in Δ Cox/Cyd/CytM (Fig. 6F), thus excluding this possibility. Recently, a cyanobacterial ferredoxin, Fed2, was shown to play a role in iron sensing and regulation of the IsiA antenna protein, a protein which is typically expressed when cells are exposed to low-iron conditions (Schorsch et al 2018). Similar to Fed2, it is possible that CytM plays a regulatory role in the cell, rather than being directly involved in electron transport.
Under conditions when cells are exposed to high concentrations of glucose or other sugars, CytM may regulate photosynthetic electron transport and carbon fixation by limiting CO 2 uptake and decreasing the total amount of photosynthetic proteins, which in turn reduces photosynthesis. In line with this, we observed accumulation of EcaB in Δ CytM (Table I; Fig.  8). Enhanced EcaB levels likely results in greater inorganic carbon assimilation, higher carbon fixation, and increased turnover of NADPH, the terminal electron acceptor in linear photosynthetic electron transport. This in turn likely limits over-reduction of the photosynthetic electron transport chain.
Regardless of the exact role of CytM, it is clear that deletion of this protein significantly increases growth of Synechocystis in photomixotrophy (Fig. 1), in line with previous studies (Hiraide et al 2015). This is possibly due to an increase in photosynthetic capacity combined with efficient assimilation of glucose into central metabolism, resulting in greater biomass accumulation. This clearly results in increased production of proteins required for enhanced growth, including those involved in phosphate uptake (PstA1, PstB1, PstB1', PstC) ( Table I), import of Mg 2+ (MgtE) Zn 2+ (ZiaA) and Fe 2+ (FutA2) and production of chl (ChlP, ChlL) ( Table  I; Fig. 8).
In conclusion, deletion of CytM allows Synechocystis to maintain efficient photosynthesis and enhanced growth under photomixotrophy. While we have not determined the exact function of CytM, we propose that it plays a role in reducing photosynthesis under natural conditions when both light intensity and glucose concentration fluctuates (Hieronymi andMacke et al 2010, Ittekkot et al 1985), and the redox state of the intertwined photosynthetic and respiratory electron transfer rapidly changes.

Plasmid construction
The genome sequence of Synechocystis released 11.05.2004 was consulted via Cyanobase (http://genome.kazusa.or.jp/cyanobase) for primer design. Primers are listed in Supplemental Table S1. The cytM (sll1245) gene was deleted by amplifying a 906 bp fragment upstream of cytM using primers CytMleftfor and CytMleftrev and a 932 bp fragment downstream of cytM using primers CytMrightfor and CytMrightrev, followed by insertion of the respective fragments into the SacI/EcoR1 and XbaI/BamH1 sites of pUC19 to generate pCytM-1. The BamH1 digested npt1/sacRB cassette from pUM24Cm (Ried and Collmer, 1987) was inserted into the BamH1 site between the upstream and downstream fragments in pCytM-1 to generate pCytM-2.

Construction of cytM deletion mutants
Unmarked mutants of Synechocystis lacking cytM were constructed via a two-step homologous recombination protocol according to . To generate marked mutants approximately 1 µg of plasmid pCytM-2 was mixed with Synechocystis cells for 6 hours in liquid media, followed by incubation on BG-11 agar plates for approximately 24 hours. An additional 3 mL of agar containing kanamycin was added to the surface of the plate followed by further incubation for approximately 1-2 weeks. Transformants were subcultured to allow segregation of mutant alleles. Segregation was confirmed by PCR using primers CytMf and CytMr, which flank the deleted region. To remove the npt1/sacRB cassette to generate unmarked mutants, mutant lines were transformed with 1 µg of the markerless CytM-1 construct. Following incubation in BG-11 liquid media for 4 days and agar plates containing sucrose for a further 1-2 weeks, transformants were patched on kanamycin and sucrose plates. Sucrose resistant, kanamycin sensitive strains containing the unmarked deletion were confirmed by PCR using primers flanking the deleted region (Supplemental Fig. S2B). The ∆ Cox/Cyd/CytM unmarked strain was generated via the same method in the background of the unmarked ∆ Cox/Cyd strain (Lea-Smith et al., 2013).

Cultivation
Pre-experimental cultures were grown in 30 ml BG-11 medium buffered with 10 mM TES-KOH (pH 8.2) in 100 ml Erlenmeyer flasks. Cultures were shaken at 120 rpm at 30°C and exposed to constant white fluorescent light of 50 µmol photons m − 2 s − 1 intensity in a Sanyo Environmental Test Chamber (Sanyo Co, Japan) which was saturated with 3% CO 2 . Starter cultures were inoculated at 0.1 OD 750 and cultivated for three days with density typically reaching 2.5±0.5 OD 750 .
Experimental cultures for growth and photophysiological experiments were inoculated in 30 ml fresh BG-11 media at 0.1 OD 750 from harvested pre-experimental cultures. The media was buffered with 10 mM TES-KOH (pH 8.2), the CO 2 concentration was atmospheric, and cultures were agitated in 100 ml Erlenmeyer flasks at 120 rpm in AlgaeTRON AG130 cool- every 24 h. For photophysiological studies, cells were cultivated under condition (c) for three days. For proteomics analysis, cells were cultivated similarly to (c), with the exception of an extra three day long pre-cultivation step at atmospheric CO 2 without glucose.

Cell counting, cell size determination
Cell number was determined with a Nexcelcom Cellometer X2 via the following method. Sample OD 750 was adjusted to one, brightfield images were captured and the cell number was determined by the Nexcelcom software. In order to exclude the visual glitches falsely recognized as cells by the software, only the four most populous cell size group was averaged. Typically, three thousand cells were counted per plate.

Glucose determination
Glucose concentration of the spent media was determined spectrophotometrically with the commercial High Sensitivity Glucose Assay Kit (Sigma-Aldrich, U.S.). Prior to measurements, the cell suspension was centrifuged at 5000 g for 10 min and the supernatant was filtered through a 0.2 µm filter.

MIMS measurements
Gas fluxes of intact cells were measured using membrane inlet mass spectrometry. The inhouse built system consists of a DW-1 oxygen electrode chamber (Hansatech Ltd., U.K.) connected to the vacuum line of a mass spectrometer (Prima PRO model, Thermo Scientific, U.S.). The sample was separated from the vacuum line by a Hansatech S4 PTFE membrane (Hansatech Ltd., U.K.). The chl a concentration of the sample was adjusted to 10 mg ml − 1 . Prior to measurements, the sample was enriched with 98 % 18 O 2 heavy isotope (CK Isotopes Limited, U.K.), the dissolved total carbon concentration was adjusted to 1.5 mM by adding NaHCO 3 , and then 10-15 min dark adaptation was applied. The measurement was performed in a semi-closed cuvette at 30°C with constant stirring. The light source was a 150 Watt, 21 V, EKE quartz halogen-powered fiber-optic illuminator (Fiber-Lite DC-950, Dolan-Jenner, U.S.). Rates were calculated as described previously (Beckmann et al 2009).

Clark-type electrode measurements
PSII activity was tested in the presence of 0.5 mM 2,6-dichloro-p-benzoquinone (DCBQ) with a Clark-type oxygen electrode and chamber (Hansatech Ltd., U.K.). Cells were maintained at 30°C, with constant stirring, under 1000 µmol photons m − 2 s − 1 illumination using a Fiber-Lite DC-950 light source. Prior to the measurements, cells were resuspended in BG-11 (pH 8.2) supplemented with 10 mM glucose and the chl a concentration was adjusted to 7.5 µg ml − 1 . Rates of oxygen production was calculated using the Hansatech software.

Chl fluorescence and P700 oxidoreduction measurements
Whole cell chl fluorescence was measured simultaneously with P700 with a pulse amplitudemodulated fluorometer (Dual-PAM-100, Walz, Germany). Prior to measurements, cells were resuspended in BG-11 (pH 8.2) supplemented with 10 mM glucose and the chl a concentration was adjusted to 15 µg ml −1 . Measurements were performed at 30°C, and samples were initially incubated in darkness for 15 minutes with stirring. To determine P m , 30 s strong far-red light (720 nm, 40 W m −2 ) and red multiple turnover saturating pulses (MT) were applied. MT pulses were set to an intensity of 5000 µmol photons m −2 s −1 (width: 500 ms). Red (635 nm) actinic light was at an intensity of 50 µmol photons m −2 s −1 was used as background illumination. Photosynthetic parameters were calculated as described previously (Klughammer et al 2008 a,b).
Relaxation of flash-induced fluorescence yield was monitored using a fluorometer (FL3500, PSI Instruments, Czech Republic) outlined previously (Allahverdiyeva et al 2003). Prior to the measurement, cells were resuspended in BG-11 (pH 8.2) supplemented with 10 mM glucose, adjusted to 5 µg chl a ml −1 and dark adapted for 5 min. Curves were normalized to F 0 and F m .
Fluorescence emission at 77K was determined using a USB4000-FL-450 fluorometer (Ocean Optics, USA). Prior to measurements, samples were adjusted to 7.5 mg chl ml −1 and light-adapted to identical conditions under which the cells were cultivated. Samples were excited at 440 nm. The curves were then normalized to their respective PSI emission peak at 723 nm.

Western blotting
Total protein extraction, electrophoresis and immunoblotting was performed as described previously (Huokko et al 2019).

MS analysis: sample preparation, data-dependent analysis, protein identification and quantitation
For data analysis, we used the proteome of Synechocystis   Table S1. List of oligonucleotides used in this study.