On the trail of iron uptake in ancestral Cyanobacteria on early Earth

Cyanobacteria oxygenated Earth's atmosphere ~2.4 billion years ago, during the Great Oxygenation Event (GOE), through oxygenic photosynthesis. Their high iron requirement was presumably met by high levels of Fe(II) in the anoxic Archean environment. We found that many deeply branching Cyanobacteria, including two Gloeobacter and four Pseudanabaena spp., cannot synthesize the Fe(II) specific transporter, FeoB. Phylogenetic and relaxed molecular clock analyses find evidence that FeoB and the Fe(III) transporters, cFTR1 and FutB, were present in Proterozoic, but not earlier Archaean lineages of Cyanobacteria. Furthermore Pseudanabaena sp. PCC7367, an early diverging marine, benthic strain grown under simulated Archean conditions, constitutively expressed cftr1, even after the addition of Fe(II). Our genetic profiling suggests that, prior to the GOE, ancestral Cyanobacteria may have utilized alternative metal iron transporters such as ZIP, NRAMP, or FicI, and possibly also scavenged exogenous siderophore bound Fe(III), as they only acquired the necessary Fe(II) and Fe(III) transporters during the Proterozoic. Given that Cyanobacteria arose 3.3–3.6 billion years ago, it is possible that limitations in iron uptake may have contributed to the delay in their expansion during the Archean, and hence the oxygenation of the early Earth.

productivity (Jiang et al., 2020;Sutak et al., 2020). The uptake of iron by prokaryotes has been extensively studied and yet the understanding of the mechanisms and identification of all the participating receptor components are still unclear (Fresenborg et al., 2020;Qiu et al., 2022). Additionally, most studies investigating iron uptake have focused on iron depleted conditions, usually under the oxidizing environment of our present atmosphere, where mechanisms that dominate utilize siderophores-low-molecular-weight biologic metal chelators that bind free Fe(III) and facilitate its targeted uptake across the prokaryotic cell membranes-dominate (Årstøl & Hohmann-Marriott, 2019;Fresenborg et al., 2020;Kranzler et al., 2011Kranzler et al., , 2014. Early Earth had an anoxic, slightly reducing atmosphere which meant that Fe(II) provided the main source of iron for early life (Canfield, 2005;Catling & Zahnle, 2020). Iron bioavailability was significantly altered with increasing oxygenation, with Fe(II) being oxidized to insoluble Fe(III) (Fresenborg et al., 2020;Jiang et al., 2020;Sutak et al., 2020;Xu et al., 2016). Cyanobacteria, the only presentday prokaryotes capable of conducting oxygenic photosynthesis, are largely accepted to have generated the copious amounts of oxygen required to oxygenate not only the atmosphere, but also the oceans (Jiang et al., 2020;Schopf & Kudryavtsev, 2012).
Cyanobacteria in general contain a higher metal content than chemoheterotrophic micro-organisms (as summarized in reviews by Fresenborg et al., 2020 andQiu et al., 2022). Cyanobacteria can have 25-350 times more atoms of iron per cell than Escherichia coli, depending on strain, cell type, and function (Fresenborg et al., 2020).
The redox status of Cyanobacteria is tightly coupled to the light cycle, with genes encoding high-affinity metal transporters for iron, manganese, and copper following a diurnal expression pattern (Botello-Morte et al., 2014;Saha et al., 2016). For iron to enter the cyanobacterium, it must cross the outer cell membrane, pass through the periplasmic space, and be transported across the inner plasma membrane. An overview of iron specific transporters identified in Cyanobacteria is presented in Figure 1. Summaries of iron transporters in Cyanobacteria are presented in Qiu et al. (2022), Fresenborg et al. (2020 and Jiang et al. (2020). Cyanobacterial porins permit the selective passage of compounds through the outer cell membrane, with an iron-specific porin recently being identified in Synechocystis sp. PCC6803 (Qiu et al., 2021). Most iron in modern-day aquatic systems is bound to organic ligands-siderophores-and crosses the outer membrane via TonB-dependent transporters (TBDT) energized by the ExbB/D system on the inner cell membrane (Figure 1; F I G U R E 1 Inorganic iron uptake in Cyanobacteria. Uncomplexed iron can enter the periplasmic space via TonB-dependent transporters (TBDTs) shown as the orange rectangle (Jiang et al., 2015;Qiu et al., 2018) or outer membrane porins, indicated by a gray square (Qiu et al., 2021). Cyanobacteria make use of specific transporters to facilitate inorganic iron uptake into the cytoplasm: The FutABC system facilitates Fe(III) uptake into the cell (Brandt et al., 2009;Katoh et al., 2001), while FeoB is the primary transporter for Fe(II) (Katoh et al., 2001;Kranzler et al., 2014). The zinc-iron permease (ZIP) (Morrissey & Bowler, 2012), the natural resistance-associated macrophage protein (NRAMP) homologue (Nevo & Nelson, 2006) and FicI (Bennett et al., 2018) are thought to take up Fe(II) and other divalent cationic metals, indicated in red text. Alternative respiratory oxidases (ARTOs) can mediate the redox state of periplasmic Fe (Berry et al., 2002;Hart et al., 2005;Katoh et al., 2000) and are suggested to play a role in periplasmic iron reduction (Kranzler et al., 2011(Kranzler et al., , 2014. The cytosolic protein, fur a, functions as a transcriptional repressor (Kaushik et al., 2016), regulating iron uptake (González et al., 2012(González et al., , 2013(González et al., , 2014(González et al., , 2016. The permease, cFTR1, takes up Fe(III) possibly by re-oxidation of Fe(II) by oxygen (Xu et al., 2016). Potentially the TonB -ExbB/D complex coordinates with TBDTs to take up iron that is bound to organic ligands (Boukhalfa & Crumbliss, 2002;Jiang et al., 2015;Schätzle et al., 2021). ExbB/D was also found to take up inorganic iron directly (Jiang et al., 2015). Oxygen released during oxygenic photosynthesis can also oxidize the periplasmic Fe(II) pool Qiu et al., 2022;Fresenborg et al., 2020;Sutak et al., 2020). The synthesis of siderophores is not prevalent in early diverging lineages of Cyanobacteria (Årstøl & Hohmann-Marriott, 2019), but basic siderophore transporters are commonly found in other Cyanobacterial genomes (reviewed by Qiu et al., 2022;Fresenborg et al., 2020). To date, Cyanobacterial siderophore synthesis studies have focused on aquatic strains grown under iron-depleted conditions (Årstøl & Hohmann-Marriott, 2019;Jiang et al., 2020).
Prior to the appearance of free oxygen, Fe(II) transport mechanisms should have provided the main source of iron for Cyanobacteria (Fresenborg et al., 2020;Jiang et al., 2020;Xu et al., 2016). The FeoB transporter functions as a Fe(II) permease, and its cytosolic Gprotein domain is considered a precursor of eukaryotic G-proteins (Hantke, 2003). Genes putatively identified as feoB analogues are found within many Archaea genomes (Gómez-Garzón et al., 2022;Russum et al., 2021), while pairwise analyses place FeoB in a hierarchical orthologous group that appears at the level of LUCA, with FutB, Ftr1, ZIP, NRAMP, and ExbB/D appearing later in Bacteria (Altenhoff et al., 2018). Cyanobacteria, with their high iron requirements, had to adjust to ever reducing levels of Fe(II) with increasing oxygenation (Fresenborg et al., 2020;Jiang et al., 2020;Qiu et al., 2022). This dramatic change in iron bioavailability may have necessitated the evolution of Fe(III) transporters such as cFTR1 and FutABC within Cyanobacteria (Xu et al., 2016). Given the diversity of iron transporters identified in Cyanobacteria, the expression of iron-specific transporters under iron-replete conditions representing a ferruginous ocean under an anoxic atmosphere is investigated. Whereas most investigations into iron transport in Cyanobacteria have focused on the freshwater, unicellular, feoB carrying Synechocystis sp. PCC6803 (Fresenborg et al., 2020;Jiang et al., 2020;Qiu et al., 2022 and references therein) and, more recently, the filamentous diazotroph, Nostoc sp. PCC7120 (previously Anabaena sp. PCC7120) (Schätzle et al., 2021), we focus on the deeply branching strain of Pseudanabaena sp. PCC7367. It represents a lineage, which diverged from those leading to Synechocystis sp. PCC6803 and Nostoc sp. PCC7120 more than 2 billion years ago (Boden et al., 2021;Sánchez-Baracaldo, 2015;Sánchez-Baracaldo et al., 2017;Schirrmeister et al., 2013), so may offer greater insight into possible processes in the former ferruginous oceans of the Archean.
Recently, it was found that Pseudanabaena sp. PCC7367 was able to survive repeated nocturnal influxes of Fe(II) under anoxic conditions, whereas another deep branching marine strain, Synechococcus sp. PCC7336, did not (Herrmann et al., 2021). Previous analysis of 72 Cyanobacterial genomes by Kranzler et al. (2014; Figure S4) and Qiu et al. (2022), indicated that the genomes of a large number of marine species, including picocyanobacteria and Pseudanabaena sp. PCC 7367, do not encode a FeoB protein for Fe(II) uptake. In this study, we expand upon this research by searching for genes encoding additional iron transporters in 125 Cyanobacteria and reconstructing their evolutionary history. These include the zinc-iron permease (ZIP) (Morrissey & Bowler, 2012) also known as ZupT in Nostoc sp. 7120 (Fresenborg et al., 2020), the natural resistance-associated macrophage protein (NRAMP) homologue MntH, for transporting Mn(II) and Fe(II) into the cytoplasm (Nevo & Nelson, 2006), the Fe(II) and Co(II) transporter (FicI) (Bennett et al., 2018), the Fe(II) transporter, FeoB (Katoh et al., 2001;Kranzler et al., 2014), and the Fe(III) transporters; namely FutABC (Brandt et al., 2009;Katoh et al., 2001) and the iron permease, cFTR1 (Xu et al., 2016) also known as EfeU in Nostoc sp. 7120 (Fresenborg et al., 2020). Furthermore, we screened for siderophore-associated uptake genes encoding TBDTs, TonB, and the ExbB/D complex (Jiang et al., 2015;Qiu et al., 2018;Schätzle et al., 2021). The expression of cftr1, the cytochrome c oxidase gene, cyoC, and the intracellular iron transcriptional regulator gene, furA, in cultures of Pseudanabaena sp. PCC7367 grown in an anoxic atmosphere with 0.2% CO 2 was also investigated. Additionally, we employ phylogenetic and Bayesian molecular clock analyses to estimate when iron-specific transporters for Fe(II) (namely FeoB) and Fe(III) (namely FutB and cFTR1) appeared within the evolutionary history of the Cyanobacteria Phylum.

| Gene screening
In order to understand the differences in the perceived Fe(II) toxicity  (Altschul, 1991(Altschul, , 1993Zhang et al., 2000) with e-value cut-offs less than 0.001 for the following genes linked to iron transporters identified in Synechocystis sp.
The complete bioinformatics search and processing pipeline is illustrated graphically in Figure S7.

| Phylogenetic analyses
Our initial screen indicated a lack of feoB in most basal Cyanobacterial genomes, so the abovementioned gene screen was extended to search for iron transporters in a broader range of genomes representing the full diversity of Cyanobacteria (Boden et al., 2021).
Phylogenetic analyses were then employed to investigate how iron transporters evolved in Cyanobacteria. To do this, amino acid sequences of FeoB, FutB, and Cyanobacterial FTR1 were aligned using MUSCLE (Edgar, 2004) implemented in MegaX version 10.1.8 (Kumar et al., 2018) with the following parameters; gap open −2.9, gap extend 0, hydrophobicity multiplier 1.2, maximum iterations 16, cluster method UPGMA, minimum dialogue length 24. Poorly aligned regions, specifically those with more than 80% gap regions, were removed manually. In order to reconstruct the phylogeny of iron uptake genes, Bayesian phylogenetic trees were generated for FeoB, FutB, and cFTR1 in MrBayes 3.2.7a (Ronquist et al., 2012)

| Genome tree and molecular clock analyses
To estimate when the iron uptake genes specific for Fe(II), Fe(III), and the cFTR1 permease emerged in Cyanobacteria, the evolutionary history of FeoB, FutB, and cFTR1 was compared with the maximum likelihood phylogeny of Boden et al. (2021). Details of how this phylogeny was produced are present in the original paper (Boden et al., 2021), which incorporates information from 139 proteins, 16S rRNA and 23S rRNA collected from >100 strains representing the entire diversity of Cyanobacteria. If topology of this species tree matched the topology of Bayesian phylogenies of FeoB, FutB, or cFTRA, generated in the present study, then the MRCA of that clade was assumed to have utilized the protein. To find out when those ancestors diversified, we cross-referenced them to the Bayesian molecular clock of (Boden et al., 2021). This was made using information from rRNA (16S and 23S) and 6 soft calibrations from fossils and geological records. For further detail, see (Boden et al., 2021).

| Culture conditions and experimental setup
Pseudanabaena sp. PCC 7367 (Pasteur Culture Collection, Paris, France) was maintained in the prescribed ASNIII medium and acclimated to the simulated Archean atmosphere in an anoxic chamber atmosphere (GS Glovebox, Germany) of N 2 gas supplemented with 0.2% CO 2 , 17:9 hr day-night cycle, 65% humidity, and 25 Photosynthetic Photon Flux Density (PPFD [μmols photons · m −2 · s −1 ]) (Herrmann et al., 2021). Triplicate cultures, inoculated at 0.4 μg Chl a · ml −1 from late exponential phase cultures, were set up in acid-washed, sterilized Fernbach flasks containing 600 ml medium equilibrated at the experimental atmosphere. Chl a determination of cell content is routinely used to monitor Cyanobacterial growth and viability. Briefly, Chl a was extracted from a 1.5 ml culture volume on days 1, 3, 6, 9, and 11. Cell pellets were lysed in 90% (v/v) CaCO 3 neutralized methanol by bead beating, quantified as described in Herrmann et al. (2021) and plotted to generate a growth curve for Pseudanabaena sp. PCC 7367 ( Figure S1).
On day 10, Fe(III) was added to the cultures to ensure they were not iron depleted. The following day, an oxygen microsensor (Ox200, UNISENSE, Denmark) was installed to monitor the O 2 levels resulting from oxygenic photosynthesis, in the cultures for the duration of the experiment.

| Spectrophometric ferrozine iron assay
Fe(II) and Fe(III) levels were monitored periodically by means of the spectrophotometric ferrozine iron assay to confim the availability of Fe(II) at night (Herrmann et al., 2021). Briefly, the cultures in the anaerobic glovebox (GS Glovebox, Germany) were gently resuspended and 2 × 1 ml culture volume was removed under sterile conditions from each biological replicate, added to 2 ml reaction tubes (Sarstedt, Germany), and the particulate matter immediately pelleted by cen-

| Primer design and validation
Primers for reverse transcription quantitative PCR (RT qPCR) were designed and validated to detect the following genes of Pseudanabaena sp. PCC 7367: the Cyanobacterial iron permease, cFTR1 (Pse7367_Rs12485), the ferric uptake regulator, FurA (Pse7367_Rs06445), the cytochrome c oxidase (Pse7367_Rs00935), and the reference target gene, rpoC1 (Pse7367_Rs07505), encoding the RNA polymerase gamma subunit (Alexova et al., 2011). The primer sequences, PCR product length, and primer amplification F I G U R E 2 Genomic tree of Cyanobacteria indicating the distribution of iron transporters investigated in this study. The inorganic Fe (II) transporters FeoB, ZIP, NRAMP, and FicI (represented by squares), as well as the Fe (III) transporters (represented by triangles), FutB and Cyanobacterial FTR1, are superimposed on a Bayesian molecular clock adapted from (Boden et al., 2021). Since the TonB, ExbB/D, and TBDT system can also play a role in inorganic iron uptake (Qiu et al., 2018) the presence of these three siderophore associated iron uptake transporters are also indicated. Support values for branching relationships represent ultrafast bootstrap approximations (Hoang et al., 2018). These are equal to 100 unless otherwise stated. The annotations for inorganic iron transporters are as follows: FeoB: Green squares; ZIP, NRAMP, and FicI: Blue squares; FutB: Salmon pink triangles; FTR1: Yellow triangles; ExbBD, TonB, and TBDTs: Dark pink circles. The names of Cyanobacteria isolated from marine habitats are colored black, in comparison to strains from freshwater, terrestrial and geothermal springs, which are gray. Deeply branching lineages (Boden et al., 2021;Sánchez-Baracaldo, 2015) are indicated inside the gray box. Black circles represent calibration points described in Boden et al. (2021), Table 1). The first diversification of crown Cyanobacteria was constrained to occur between 2.32 and 2.7 billion years ago based on evidence of the GOE (Bekker et al., 2004) and stromatolitic laminae characteristic of Cyanobacteria (Bosk et al., 2009). The youngest-bound for FeoB, FutB, and FTR1, based on phylogenetic evidence, are indicated with shapes of the relevant colour in the molecular clock ( Figure 3) efficiencies are presented in Table S3. PCR product integrity was confirmed by melt curve analysis using cDNA of Pseudanabaena sp.
PCC 7367 as template. Phenol v/v; Roth, Germany) and gently inverted to prevent further transcription. Cells were pelleted at 4000 × g for 10 min (Eppendorf 5810R, Germany). Throughout the experiment, the cultures were gently agitated by a magnetic stirrer bar set at 150 rpm to facilitate the release of O 2 from the culture medium prior to the addition of Fe(II) (Herrmann et al., 2021). Pelleted cells were drained and stored at −80°C until RNA extraction.

| RNA extraction and synthesis of copy DNA
RNA was extracted from the thawed pellets using the NucleoSpin® RNA Plant Kit (Macherey-Nagel, Germany) according to the manufacturer's instructions, with a modified cell lysis step (Mironov & Los, 2015). The cell pellets were transferred to a sterile 2 ml tube (Sarstedt, Germany) containing 100 mg RNAsefree 0.1 mm silica beads (Biospec, Germany). RA1 buffer was added (350 μl of RA1 buffer per 100 mg pellet) to the pellet, as well as 1% (v/v) β-Mercaptoethanol (2-Mercaptoethanol, ROTH, Germany). The samples were frozen in liquid nitrogen, allowed to thaw, then were disrupted for 90 sec at setting 6.5 (Fastprep FP120, Thermo systems, USA) followed by an additional freeze/thaw step.
The cell lysates were centrifuged for 1 min at 14,000 × g (Hermle Z233-M2, Germany) to pellet the cell debris and the RNA was extracted from the supernatant using the two column-system of the NucleoSpin® RNA Plant Kit (Macherey-Nagel, Germany). DNA removal was ensured by the on-column DNA digestion according to the manufacturer's description. RNA thus obtained was spectrophotometrically quantified (NanoDrop® Lite, Thermo Scientific, USA) and the quality confirmed by agarose gel electrophoresis, with DNA digestion verified by PCR targeting the housekeeping gene, rpoC1.

| Quantification of gene expression
The levels of expression of the genes encoding FurA, cFTR1, and cytochrome c oxidase and the housekeeping gene for the gamma subunit of the RNA polymerase, rpoC1, were determined via quantitative PCR (qPCR) of the cDNA Nolan et al., 2013).
Each 10 μl reaction was prepared with 5 μl 2x iTaq™ Universal SYBR® Green Supermix, 5 pmol of each primer and 10 ng cDNA template. The volume was adjusted to 10 μl with RNase-free water.
The reactions were performed in triplicate, on three different days, to evaluate transcript abundance relative to the expression of the housekeeping gene (Lü et al., 2018)

using a BIORAD CFX Connect™
Real-Time System thermocycler. Cycling for the qPCR was as follows: activation of the polymerase (50 °C for 10 min), followed by initial denaturation at 95 °C for 5 min and 40 cycles of 95 °C for 10 sec, 20 sec at the primer-pair specific annealing temperature and 72 °C for 10 sec, with a final elongation step at 72 °C for 5 min.
Primer T m and T a , as well as the product size, are listed in Table 1.
Product length was verified by melt curve analysis and the relative gene expression was calculated from the mean fold difference of ∆C q values of the three biological replicates for each timepoint (Guescini et al., 2008;Huggett et al., 2013;Narum, 2006;Rutledge & Stewart, 2008), in Excel (Excel 365, Microsoft, USA).

| Statistical analyses
Statistical analyses were done using the two-tailed, heteroscedastic Student's t-test (Excel 365, Microsoft, USA) to determine the influence of Fe(II) on gene expression levels.

| Iron transporters of Pseudanabaena sp. PCC7367 and other deeply branching Cyanobacteria
Initial similarity searches for a FeoB homologue in Pseudanabaena sp.
In light of the lack of Fe(II) uptake transporters encoded within deeply branching Cyanobacteria, and the potential for the siderophore associated ExbB/D to take up inorganic iron directly (Jiang et al., 2015), further similarity searches for siderophoreassociated uptake genes were conducted. It was confirmed that Pseudanabaena sp. PCC 7367 does not produce siderophores (Fresenborg et al., 2020; this study), but does encode a siderophore uptake system. This includes ExbB/D, TonB protein, and the TonBdependent transporters (TBDTs), fhuE, iutA and fhuA (Table S1; Pseudanabaena sp. PCC 7367 also encodes a few porins that permit the nonspecific entry of substances such as metals into the periplasmic space; however, their functionalities are not well characterized (Table S3). The iron selective porin identified in Synechocystis sp. PCC6803 is not present in Pseudanabaena sp. PCC7367 (Qiu et al., 2021); however, eight potential outer membrane porins were identified (Table S4).

| Phylogeny of Cyanobacterial iron uptake genes FeoB, cFTR1, and FutB
Considering the above observations, a broader range of genomes spanning the Cyanobacterial tree of life (specifically all of those analyzed in Boden et al., 2021) were screened for the presence of iron transporters (Figure 2, Table S2). Genes encoding FeoB, FutB, and cFTR1 were found in a variety of strains from marine and nonmarine habitats (Figure 2), so Bayesian protein phylogenies were generated to describe how these iron transport proteins from different strains of Cyanobacteria are related. and unicellular species, such as Synechocystis sp. PCC6803 (PP 100) ( Figure S1). This lack of relationship between the FeoB homologues of Thermosynechococcus is well-supported (PP 100) and suggests that its two FeoB sequences have different evolutionary origins.
In contrast to FeoB, which was encoded in the genomes of only five of 16 deeply branching strains, the ferric iron transporter, FutB, was identified in 12, most of these strains (Figure 2 Note: These younger-bounds represent the latest (specifically most recent) possible date that each iron transport protein emerged in Cyanobacteria based on phylogenetic evidence. "Upper" and "lower" refer to posterior 95% confidence intervals estimated by the Bayesian molecular clock of

| Dating the Cyanobacterial iron transporters FeoB, FutB, and cFTR1
The differing distributions of FeoB, FutB, and cFTR1 proteins among Cyanobacteria could reflect differences in each strain's metaluptake strategies and environmental history. We therefore searched for congruence between the evolutionary history of each protein and an established molecular clock (Boden et al., 2021) to determine approximately when the feoB, futB, and cftr1 genes were introduced into ancestral Cyanobacteria. Similar methods have been utilized previously to map the origin of nitrogen-metabolizing enzymes (Parsons et al., 2021) and oxygen-utilizing enzymes in bacteria (Jabłońska & Tawfik, 2021   uptake. Expression of futB is known to be constitutive and not influenced by Fe(II) availability (Katoh et al., 2001). In contrast, the cFTR1 transporter was demonstrated to preferentially take up Fe(III) in Synechocystis, with an increase in its expression observed under iron starvation (Xu et al., 2016). As no alternative respiratory terminal oxidase (ARTO) was identified in Pseudanabaena sp.
PCC7367 (Figure 1), we decided to investigate the expression of the gene encoding another terminal oxidase, cytochrome c oxidase to ascertain whether it possibly influenced by the redox state of environmental iron (Schmetterer, 2016). While normally responsible for generating a proton gradient across the thylakoid membrane through the formation of water in the cytoplasm during respiration, cytochrome c oxidase may also be involved in modulating the Fe(II)/Fe(III) pool in the periplasm during respiration (Schmetterer, 2016). Changes in expression of cftr1 and cyoC were monitored in response to an evening influx of Fe(II) under anoxic conditions.
The Fe(II) and oxygen levels in the media were measured over 14 hours and are presented in Figure 4. When oxygen levels dropped to zero, 1 hour after dark, with gentle agitation of the cultures, Fe(II) was added ( Figure 4) and its level tracked using the ferrozine assay ( Figure 4). Fe(II) was gradually oxidized or taken up overnight, but F I G U R E 4 Concentrations of dissolved oxygen and Fe(II) in the Pseudanabeana sp. PCC 7367 cultures, with relative cftr1 and cyoC gene expression levels over the course of a night cycle. The concentration of dissolved oxygen in the culture medium (blue line), was tracked using an oxygen microsensor, while the Fe(II) concentrations (green line) were periodically determined using the ferrozine assay. The first sample of culture biomass for RNA extraction was taken 1 hour before the dark cycle when oxygen was still present in the medium. The following samples for RNA extraction were taken 15 min, 45 min, 2 h, and 7 h after the addition of Fe(II) to a concentration of 240 μM in the dark (green dotted line), when dissolved oxygen levels reached zero. The last sample for RNA extraction was taken 1 hour after the start of the light cycle. The relative expression of cftr1 and cyoC in the cultures to which Fe(II) was added, is plotted at the corresponding time points in comparison to expression of the control cultures without Fe(II) addition ( Figure S6), (n = 3). Stars (*) represent a significant difference to the control culture (p < 0.05; Student's t-test, two-tailed) was still present at 50 μM when the lights went on, after which it was rapidly oxidized as oxygenic photosynthesis commenced.
The relative expression data show a stable expression of cftr1 slightly higher than for the control throughout the experiment, with a significant decrease relative to the control an hour after the lights went on. This decrease corresponds to a rapid decrease in Fe(II) in the medium (Figure 4). The expression of cytochrome c oxygenase increased significantly after the addition of Fe(II), decreasing to late daytime levels 2 hours after the addition of Fe(II).
This is in contrast to the previously studied Synechococcus sp.
Bayesian trees indicated a complex history involving gene duplication and/or other patterns of reticulated evolution, leading to individual strains sometimes harboring more than one gene for a given transporter. For example, two FeoB homologues with different evolutionary trajectories were found in the deeply branching Thermosynechococcus elongatus BP1 ( Figure S1). Furthermore, the FutB homologue of the cyanobacterium, Gloeobacter violaceus PCC 7421 was unrelated to FutB homologues of other basal strains ( Figure S2).
To find out approximately when these Fe transporters emerged in Cyanobacteria, the evolutionary history of each iron transporter was compared to an existing genome tree of Cyanobacteria and compared with a previously published molecular clock (Boden et al., 2021). Overall, this revealed that evolutionary histories of the  (Saito et al., 2003), when lineages with the Fe(III) transporters, FutB and cFTR1, were likely already present ( Figure 3). Whether the incorporation of the feo cluster into the Cyanobacteria lineage coincided with the evolution of reductive iron uptake, whereby Fe(III) reduction to Fe(II) is regulated by the alternative respiratory oxidase (ARTO) on the cell membrane (Kranzler et al., 2011(Kranzler et al., , 2014, is beyond the scope of this study. ARTOs have been proposed to provide reduced Fe(III) to the periplasmic Fe(II) pool accessed by FeoB (Kranzler et al., 2014).
While Synechococcus sp. PCC 7336 encodes an ARTO homologue, Pseudanabaena sp. PCC 7367 does not (Schmetterer, 2016, this study). As there is some evidence for cytochrome c oxidase to be located on the cytoplasmic membrane in Trichodesmium thiebautii (Bergman et al., 1993), its expression in Pseudanabaena sp. PCC 7367 after the addition of Fe(II) was determined (Figure 4). Terminal oxidase expression is known to be increased at night during respiration, regulated by the circadian clock in Synechococcus elongatus PCC7942 (Ito et al., 2009) and the iron uptake regulator, FurA, in Anabaena sp.
PCC7120 (González et al., 2012(González et al., , 2014. If cytochrome c oxidase was involved in Fe(III) reduction, a decrease in its expression after the addition of Fe(II) at night would be expected. Interestingly, expression of the cytochrome c oxidase increased significantly after the addition of Fe(II) before dropping to its original level two hours after Fe(II) addition. Whether this reflects a temporary effect on cellular respiration remains to be determined.
The iron permease of yeast, Ftr1, is tightly coupled to a ferroxidase that oxidses Fe(II) for transport across the cell membrane, whereas prokaryotes were not found to encode ferroxidases to complement their Ftr1 homologues (Banerjee et al., 2022). As 90% of the periplasmic Fe(III) is reduced to Fe(II) during iron uptake (Kranzler et al., 2014), the Fe(III) transporters, namely FutABC and specifically cFTR1, very likely obtain their Fe(III) during the day from re-oxidized Fe(II) after iron reduction (Xu et al., 2016), thereby rendering a spe- This is the first time that cftr1 transcription has been confirmed in a non-Synechocystis species, and under anoxic, ferruginous conditions. Pseudanabaena and Synechocystis, though both Cyanobacteria, are distant relatives, having last shared a common ancestor more than 2 Ga (Figure 2), before evidence for the utilization of cFTR1 appears in the evolutionary tree (Table 1, Figure 3). Previous research has found that most Cyanobacterial strains carry genes encoding siderophore uptake transporters (TBDTs) that pass siderophore bound and free inorganic iron species into the periplasmic space (Årstøl & Hohmann-Marriott, 2019; this study Figure 2). Fe(II) can be mobilized from mineral sources by siderophores, specifically desferroxazine (Bau et al., 2013;Kraemer, 2004), produced by several microorganisms in niches with low iron availability. particularly with respect to metalloenzymes (Dupont et al., 2006(Dupont et al., , 2010. Given the recent support for the evolution of oxygenic pho-

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The sequence data analyzed in this study are available in the open science framework repository, https://osf.io/7x598/ ?view_ only=715cd 38c37 8446b a8c3f 6c924 f9be9f5. All other data are included in the published article and its Supporting Information.