**1. Introduction**

Proteins containing iron-sulfur (Fe-S) clusters are present in almost all living organisms. They have diverse and often essential functions, for example, electron carriers in redox reactions, in redox sensing, oxidative stress defense, biosynthesis of metal-containing cofactors, DNA replication and repair, regulation of gene expression, and tRNA modification. Fe-S clusters are believed to be among the first catalysts to have evolved [1,2]. With the appearance of oxygenic photosynthesis, increasing oxygen levels drastically decrease iron availability [3]. Besides molecular oxygen, di fferent reactive oxygen species (ROS) appear, which are very harmful to living cells since they can damage proteins, lipids, and nucleic acids. Molecular oxygen and ROS also destabilize Fe-S clusters, leading to the release of Fe2<sup>+</sup> ions that, in turn, potentiate oxygen toxicity by the production of hydroxyl radicals in the Fenton reaction [4]. As a consequence, organisms have to develop multicomponent systems that promote the biogenesis of Fe-S proteins while protecting the cellular surrounding from the deleterious effects of free iron.

Di fferent systems for the assembly of Fe-S clusters into biological macromolecules have evolved: the Isc (**i**ron-**s**ulfur **c**luster) system was identified as the system for generalized Fe-S protein maturation in *Azotobacter vinelandii* [5] and other bacteria. In *Escherichia coli*, the *isc* operon encodes the regulator IscR (**i**ron-**s**ulfur **c**luster **r**egulator), a cysteine desulfurase (IscS), a sca ffold protein (IscU), an A-type carrier protein (IscA), a DnaJ-like co-chaperone (HscB), a DnaK-like chaperone (HscA), and a ferredoxin (fdx) [6]. The Isc machinery is widely conserved from prokaryotes to higher eukaryotes. Later, another operon for Fe-S cluster biogenesis, *suf*, was discovered in *E. coli* [7]: the *suf* operon encodes an A-type protein (SufA), a heterodimeric cysteine desulfurase (SufS and SufE), and a pseudo-ABC transporter (SufB, SufC, and SufD) that could act as a sca ffold. Components of the Suf system are also found in other bacteria, including cyanobacteria and chloroplasts. The number and type of *isc* and *suf* operons, as well as their composition, vary among bacterial species. In *E. coli*, most of the Fe-S cluster biogenesis under non-stress conditions is catalyzed by the housekeeping Isc pathway, while the Suf system functions primarily under oxidative stress and/or iron starvation [8–12].

Fe-S cluster biogenesis systems have to respond to environmental stimuli to maintain and repair the pool of Fe-S proteins under changing environmental conditions. The current understanding of the regulation of genes for Fe-S cluster assembly is mostly limited to the model organism *E. coli* and a few other members of the gamma-proteobacteria. The main regulator of Fe-S assembly in *E. coli* is IscR, but other regulatory proteins, such as Fur (an iron-dependent regulator) and OxyR (an oxidative stress-dependent regulator) also contribute to regulated *isc* and *suf* operon expression [9,10,12]. The first gene of the *iscRSUA-hscBA-fdx* operon encodes the DNA-binding IscR protein that can coordinate a [2Fe-2S] cluster, which is assembled by the Isc system [13,14]. In *E. coli*, IscR regulates at least 40 genes comprising the *isc* and *suf* operons, genes for other Fe-S containing proteins, and genes encoding surface structures (*fim* and *flu*) or of unknown functions [9]. Holo-IscR (IscR containing a Fe-S cluster) represses its gene and the rest of the *isc* operon. This repression is released under conditions unfavorable for Fe-S maturation of IscR [6]. Thus, IscR also functions as a sensor for Fe-S homeostasis. Apo-IscR (protein lacking the Fe-S cluster) activates *suf* operon expression [15] in *E. coli*. Induction of the *isc* and *suf* operons by apo-IscR occurs under iron-limiting conditions and oxidative stress [8,10,11]. A switch from the Isc to the Suf system when the iron is limiting is also promoted by the non-coding sRNA RhyB that is under negative control of Fur (**f**erric **u**ptake **r**egulator) [16]. RhyB base pairs with the Shine-Dalgarno sequence of *iscS*, leading to the degradation of the 3' part of the *iscRSUA-hscBA-fdx* operon mRNA. The 5´ part of the polycistronic mRNA, which contains *iscR*, is stabilized and translated [17]. This favors the formation of apo-IscR, which consequently induces *suf* expression.

Regulation of the *isc* and *suf* operons by stress signals also involves the regulators Fur, OxyR, and RhyB. The iron-sensing regulator protein Fur binds to the *suf* promoter as Fur-Fe2<sup>+</sup> represses the *suf* genes [10]. OxyR, a known sensor protein for oxidative stress in *E. coli* and many other bacteria, also acts as an activator of the *suf* operon [12]. It was also proposed that IscR might directly sense oxidative stress through the destabilization of its Fe-S cluster [6]. Hydrogen peroxide (H2O2) was shown to inactivate the *E. coli* Isc system and to activate the *suf* operon through OxyR [18].

Phototrophic bacteria have a special need for the regulation of the Fe-S cluster assembly. Fe-S clusters are required for some components involved in photosynthesis like enzymes for bacteriochlorophyll synthesis (magnesium chelatase and the dark-operative protochlorophyllide oxidoreductase) or the cytochrome bc1 complex for photosynthetic electron transport. On the other hand, photooxidative stress caused by the formation of ROS by the light excitation of bacteriochlorophyll destroys Fe-S clusters, which in turn leads to a further increase of ROS levels by the Fenton reaction. Latifi and co-workers suggested that elevated levels of ROS upon iron starvation [19], as also determined for *R. sphaeroides* [20], might be a characteristic for photosynthetic bacteria. Therefore, it is important to study the regulation of Fe-S cluster assembly also in phototrophic bacteria that significantly di ffer in their lifestyle from *E. coli*.

*R. sphaeroides* is a facultative photosynthetic bacterium, which can use a variety of metabolic pathways for ATP production. At high oxygen tension, aerobic respiration generates ATP. When the oxygen tension in the environment drops, the synthesis of photosynthetic complexes is induced, while aerobic respiration still takes place. Under anaerobic conditions in the light, anoxygenic

photosynthesis generates ATP, while under dark conditions in the presence of a suitable electron acceptor, anaerobic respiration is performed. Oxygen is a major regulatory factor for the formation of photosynthetic complexes, and several proteins involved in oxygen-mediated gene regulation have been identified [21,22]. Under iron limitation, *R. sphaeroides* loses its purple color due to the loss of photosynthetic complexes and is no longer able to grow phototrophically [23]. Since bacteriochlorophyll synthesis requires iron, and also the reaction center contains iron, no photosynthetic complexes can be formed under iron limitation.

*R. sphaeroides* serves as a model organism to elucidate the response of photosynthetic bacteria to singlet oxygen [24–29] and has also been analyzed in regard to its response to iron limitation in oxic and anoxic environments [20,23,30,31]. The Fur ortholog (Fur/Mur) of *R. sphaeroides* is involved in both iron and manganese homeostasis [23]. Like in other alpha-proteobacteria, the Irr (**i**ron **r**esponse **r**egulator) protein was shown to affect genes of the iron metabolism [30]. Lack of Fur/Mur, as well as lack of Irr, results in stronger induction of *isc*/*suf* genes under iron limitation [23,30]. Two-thirds of the iron-dependent genes in *R. sphaeroides* showed different responses under oxic or anoxic conditions. For some of these genes, including *isc-suf* genes, induction under iron limitation under oxic conditions was mediated by the OxyR protein [20]. *R. sphaeroides* harbors a large operon comprising *isc* and *suf* genes for iron-sulfur cluster assembly. Like in *E. coli*, the product of the first gene, IscR, functions as an iron-dependent repressor of the *isc* genes [31]. To better understand the transcriptional regulation that adjusts *isc-suf* operon expression to changing environmental conditions, we identified promoters responsible for *isc-suf* expression and quantified their activities under different oxygen concentrations in response to iron limitation and oxidative stress. This study resulted in a complex model for *isc-suf* gene regulation in a phototrophic alpha-proteobacterium.

#### **2. Materials and Methods**

#### *2.1. Bacterial Strains and Growth Conditions*

Bacterial strains are listed in Table S1. All *E. coli* strains were cultivated in Standard I medium at 37 ◦C, either in liquid culture by shaking at 180 rpm or on a solid growth medium, which contained 1.6% (*w*/*v*) agar. Depending on the cultivated strain, the antibiotics kanamycin (25 μg mL−1), ampicillin (200 μg mL−1), or tetracycline (20 μg mL−1) were added to the liquid and solid growth media.

*R. sphaeroides* strains were cultivated in 50 mL Erlenmeyer flasks containing 40 mL malate minimal medium [20] with continuous shaking at 32 ◦C (microaerobic growth with a dissolved oxygen concentration of 25–30 μM). Aerobic conditions with 160 to 180 μM dissolved oxygen were achieved by incubating 25 mL of culture in 100 mL Erlenmeyer baffled flasks. The iron-depleted medium used was malate minimal medium without the addition of Fe(III) citrate [20,23] and containing the iron chelator 2,2'-dipyridyl (30 mM, Merck, Darmstadt, Germany). The cells were grown overnight and then transferred to new iron-depleted malate minimal medium three times more before harvesting [20,23]. Cells were harvested at an OD660 of 0.4–0.6. Antibiotics were added to the liquid and solid growth media depending on the cultivated strain at the following concentrations: spectinomycin (10 μg mL−1), kanamycin (25 μg mL−1), gentamicin (25 μg mL−1), tetracycline (2 μg mL−1). For generating oxidative stress, the cultures of *R. sphaeroides* wild type or mutants were grown in iron-repleted malate minimal medium and treated with 1 mM (final concentration) hydrogen peroxide or 100 μM (final concentration) tertiary butyl-alcohol (tBOOH) for 7 min or 30 min at an OD660 of 0.5. After 0, 7, and 30 min, cells were harvested and used for ß-galactosidase measurements.

#### *2.2. Construction of R. sphaeroides RirA Deletion Mutants*

*R. sphaeroides* strain 2.4.1ΔRSP\_3341 was generated by transferring the suicide plasmid pPHU2.4.1ΔRSP\_3341::Sp (RirA homolog 1) into *R. sphaeroides* 2.4.1, and screening for the insertion of the spectinomycin resistance cassette into the chromosome by homologous recombination. Parts of the *rirA* gene (RSP\_3341) of *R. sphaeroides*, together with upstream and downstream sequences, were amplified

by using oligonucleotides 3341up\_f/3341up\_r and 3341dn\_f/3341dn\_r. The amplified PCR fragments were cloned into the *Kpn*I-*BamH*I and *BamH*I-*Hind*III sites of the suicide plasmid pPHU281 [32], generating the plasmid pPHU2.4.1ΔRSP\_3341. A 2.2 kb *BamH*I fragment containing the spectinomycin cassette from pHP45Ω [33] was inserted into the *BamH*I site of pPHU2.4.1ΔRSP\_3341 to generate pPHU2.4.1ΔRSP\_3341::Sp. This plasmid was transferred into *E. coli* strain S17-1 and diparentally conjugated into *R. sphaeroides* 2.4.1 wild-type strain. Conjugants were selected on malate minimal salt agar plates containing spectinomycin. By insertion of the spectinomycin cassette, 438 bp of *R. sphaeroides rirA* gene RSP\_3341 was deleted. *R. sphaeroides* strain 2.4.1ΔRSP\_2888 was generated by transferring the suicide plasmid pPHU2.4.1ΔRSP\_2888::Km (RirA homolog 2) into *R. sphaeroides* 2.4.1, and screening for insertion of the kanamycin resistance cassette into the chromosome by homologous recombination. Parts of the *rirA* gene (RSP\_2888) of *R. sphaeroides*, together with upstream and downstream sequences, were amplified by using oligonucleotides 2888up\_f/2888up\_r and 2888dn\_f/2888dn\_r. The amplified PCR fragments were cloned into the *Kpn*I-*BamH*I and *BamH*I-*Hind*III sites of the suicide plasmid pPHU281 [32], generating the plasmid pPHU2.4.1ΔRSP\_2888. A 2.2 kb *BamH*I fragment containing the kanamycin cassette from pHP45Ω [33] was inserted into the *BamH*I site of pPHU2.4.1ΔRSP\_2888 to generate pPHU2.4.1ΔRSP\_2888::Sp. This plasmid was transferred into *E. coli* strain S17-1 and diparentally conjugated into *R. sphaeroides* 2.4.1 wild-type strain. The conjugants were selected on malate minimal salt agar plates containing kanamycin.

#### *2.3. Constructions of Promoter Fusion Plasmids*

According to the dRNA-seq data, fragments with lengths ranging from 98 bp to 1777 bp containing one of the putative five different single promoters or combined promoters of the *isc-suf* operon, respectively, were amplified by PCR with primers listed in Table S2 The PCR product was ligated into the pJET1.2/blunt cloning vector (Thermo Fisher, Dreieich, Germany) and then transferred into *E. coli* JM109. After confirming the correct sequence, the promoter fragment was cut from the sequenced cloning vector by *Xba*I/*Pst*I or *Spe*I/*Crf* 9 and subsequently ligated into the transcriptional *lacZ* fusion vector pBBR1-MCS3-LacZ [34]. For testing the effect of antisense RNA on the activity of P2, the identical promoter fragment was also ligated into the *Xba*I/*Pst*I sites of the transcriptional *lacZ* fusion vector pBBR1-MCS5-LacZ, which harbors a gentamicin cassette and can be transferred into *R. sphaeroides* together with plasmid pRK4352 [35] and its derivatives.

#### *2.4. Construction of Altered Promoter Sequences by Site-Directed Mutagenesis*

Mutations in promoters of the *isc-suf* operon were constructed by inverse PCR. All primers used are listed in Table S2. For constructions of site-directed mutant P4 and P254, the TTG in the –35 region of the P4 promoter was replaced by AAA by using the plasmids pJET1.2-P4 or pJET1.2-P254 as templates, respectively. Similarly, for the constructions of mutant P5 and P25, the TTG in the –35 region of the P5 promoter was replaced by AAA by using the plasmids pJET1.2-P5 or pJET1.2-P25 as templates, respectively. The PCR products were digested by *Dpn*I and then transferred into JM109. Subsequently, the mutated promoter fragments were cut from the cloning vectors with *Xba*I/*Pst*I and ligated into pBBR-MCS3-LacZ [34].

Overexpression of RNA antisense to promoter P2 and sense promoter 2 fusion plasmids.

To test whether the antisense mRNA affects the activity of sense promoter, a DNA fragment antisense to P2 was amplified by PCR with primers asP2\_fwd/rev listed in Table S2. The PCR product was ligated into the pJET1.2/blunt cloning vector and then transferred into *E. coli* JM109. After sequencing, the cloned fragment was cut out by *Xba*I-*BamH*I and subsequently ligated into vector pRK4352, which contains the strong 16S promoter [35].
