A Novel Target of IscS in Escherichia coli: Participating in DNA Phosphorothioation

Many bacterial species modify their DNA with the addition of sulfur to phosphate groups, a modification known as DNA phosphorothioation. DndA is known to act as a cysteine desulfurase, catalyzing a key biochemical step in phosphorothioation. However, bioinformatic analysis revealed that 19 out of the 31 known dnd gene clusters, contain only four genes (dndB-E), lacking a key cysteine desulfurase corresponding gene. There are multiple cysteine desulfurase genes in Escherichia coli, but which one of them participates into DNA phosphorothioation is unknown. Here, by employing heterologous expression of the Salmonella enterica dnd gene cluster named dptBCDE in three E. coli mutants, each of which lacked a different cysteine desulfurase gene, we show that IscS is the only cysteine desulfurase that collaborates with dptB-E, resulting in DNA phosphorothioation. Using a bacterial two-hybrid system, protein interactions between IscS and DptC, and IscS and DptE were identified. Our findings revealed IscS as a key participant in DNA phosphorothioation and lay the basis for in-depth analysis of the DNA phosphorothioation biochemical pathway.


Introduction
Sequence and stereo specific physiological DNA phosphorothioation occurs in many bacteria [1][2][3][4]. In Streptomyces lividans 1326, a five-gene cluster, dndA-E, determines the modification [1]. Orthologs of these genes were found in 30 bacterial species and one Archaea [2]. The dnd genes are usually located on genomic islands that were probably acquired by horizontal gene transfer [3].
Several of these gene clusters contain dndB-E homologues, but lack a dndA homologue [2,3]. In-frame deletion of dndA in S. lividans showed that the gene is essential for DNA phosphorothioation [1,4]. DndA was then shown to be a cysteine desulfurase involved in the iron-sulfur cluster assembly for apo-Fe DndC [5].
Salmonella. enterica serovar cerro 87 contains dndB-E orthologs that are called dptB-E [6]. There is, however, no dndA ortholog in the entire 20 kb genomic island that contains the dpt genes (Fig. 1A) [2]. Heterologous expression of dptB-E in E. coli DH10B [7] resulted in DNA phosphorothioation [8]. Since DndA is essential for DNA phosphorothioation in S. lividans, we hypothesized that there should be one or more genes in the E. coli genome that could provide the cysteine desulfurase activity known to be necessary for the modification. Searching for a putative dndA orthologue in E. coli BW25113 was easier than in S. enterica because of the availability of a comprehensive library of knockout mutants of all nonessential genes [9]. In E.coli, there are at least three different cysteine desulfurases: IscS, SufS and CsdA [10,11]. Here we show that only one of them, IscS, supports DNA phosphorothioation in E. coli expressing the S. enterica dptB-E gene cluster. Protein interactions, which are likely necessary for DNA phosphorothioation, were detected between IscS and both DptC and DptE.

Bacterial strains, plasmids and primers
Bacterial strains, plasmids, and primers are listed in Table 1, 2  and 3.
The E. coli BW25113 gene replacement mutants listed in Table 1 were obtained from Yale Coli Genetic Stock Center [9]. Among these, the iscS mutant JW2514 was not viable, and was recreated by using the gene knockout method described by Datsenko [12]. For this, the neo-FRT (FLP, recombinase recognition target) cassette was amplified using primer P1 and P2, then H1P1 and H2P2. Successful iscS deletion was confirmed by PCR using the flanking primers U and D ( Fig. 2A).

Detection of DNA phoshorothioation
Phosphorothioate DNA is sensitive to double-strand cleavage by Tris-peracetic acid (TPA) [13]. The phosphorothioation was detected by incubating DNA samples for 30 min at 25uC in TAE buffer (40 mM Tris, 20 mM sodium acetate, 0.8 mM EDTA pH 7.5) supplemented with 1.0% peracetic acid. Phosphorothioate DNA, but not normal DNA, shows Dnd phenotype, producing a

Bacterial two-hybrid analysis
Protein-protein interactions were investigated using the Bacter-ioMatch II two-hybrid system (Stratagene), according to the manual [15] with some modifications. The system features a HIS3-aadA reporter cassette, whose expression allows E. coli growth in the presence of 3-AT (3-amino-1,2,4-triazole), which is a competitive inhibitor of His3 (imidazoleglycerol-phosphate dehydratase), and in the presence of streptomycin.
To test protein-protein interactions, in-frame gene fusions were created in the pBT (bait) or pTRG (target) vectors. PCR primers with suitable restriction sites were constructed and are listed in Table 1. IscS was fused with a bait protein, generating pBT-IscS; DndB-E were fused with target protein, generating pTRG-DptB, pTRG-DptC, pTRG-DptD and pTRG-DptE respectively. The resulting bait and target clones were co-transformed into the reporter strain E. coli XL1-Blue MRF9 Kan (Stratagene/Agilent) and selected on LB agar containing 25 mg/ml chloramphenicol (to select for pBT derivatives), 12.5 mg/ml tetracycline (to select for pTRG derivatives), and 50 mg/ml kanamycin (to maintain F9proAB lacI q ZDM15 Tn5).
To test for resistance to 3-AT, single colonies were inoculated into 1 mL LB containing the three above antibiotics, and kept shaking overnight at 30uC. 500 ml of this overnight culture was then inoculated into 5 mL SOC medium and incubated for 90 min at 37uC. The cells were then spun down at 3500 rpm for 5 min at room temperature, and the supernatant was carefully removed. The cells were then re-suspended in 2 ml M9 + His-drop out broth, collected by centrifugation as described above, and resuspended in 3 mL M9 + His-drop out broth [15]. After incubation for 2 hours at 37uC, three parallel ten-fold dilutions 10 21 -10 27 were prepared and plated 10 0 , 10 21 , 10 22 and 10 23 on Selective Screening Medium (SSM) containing 5 mM 3-AT and 10 24 , 10 25 , 10 26 and 10 27 on Nonselective Screening Medium (NSM) without 3-AT. Colonies were counted after 24 h incubation at 37uC. If there were no visible colonies, the plates were incubated in dark at 25uC for another 16 hours.
Putative positive interactions were verified using Dual Selective Screening Medium containing 5 mM 3-AT + 12.5 mg mL 21 streptomycin.

Strep and GST Pull-down
Ten milliliters of E. coli BL21 (DE3) strain (harboring Strep-iscS, or GST-DptC, or GST-DptE) was inoculated to 1 L and grow at 37uC for 3 hours with shaking (220 rpm). IPTG was then added to a final concentration of 0.2 mM (from 1000 folds stock). The culture was then moved to 16uC and grew for another 24 hours. The cells were collected by centrifuge for 10 minutes at 4uC.
Cell pellet was re-suspended using Buffer S (25 mM Hepes pH7.6, 100 mM KCl, 10% glycerol, 1 to 10 folds (w/w)) and sonicated for 30 minutes. Cell debris was removed by centrifugation at 15000 g for 20 minutes. Equal volume of the extract was mixed (IscS-GST DptC or IscS-GST DptE). Two milliliter of the mixture was incubated with 0.1 ml Streptactin resin (Qiagen) or GST resin (Qiagen) pre-equilibrated using Buffer S. After 1 hour, the resin was spin down (400 g, 3 minutes). Supernatant was removed. The resin was wash 5 times using 2 ml Buffer S. The protein was eluted using 0.3 ml Buffer S supplemented with 2.5 mM Desthiobiotin or 20 mM Glutathione. Western blot was

Expression of S. enterica dptB-E in E. coli BW25113 results in DNA phosphorothioation
Owing to the observation that in Streptomyces lividans, dndA is essential for DNA phosphorothioation, we sought to find the cysteine desulfurase gene in E. coli. The E.coli genome was searched for orthologs of a cysteine desulfurase gene. Fig. 1B shows that there are at least three cysteine desulfurase genes in the E. coli genome [10,11]. Fig. 1C shows that introducing pJTU3510 carrying dptB-E four genes, a low-copy plasmid, into E. coli BW25113 resulted in DNA S-modification (lane 3). We speculated that dptB-E, in cooperation with one or more E. coli desulfurase gene, leads to DNA phosphorothioation.

IscS is responsible for the DNA phosphorothioation in E.coli
For DNA phosphorothioation in E.coli BW25113, it seemed likely that a protein similar to the cysteine desulfurase DndA was needed in addition to the S. enterica dptB-E gene cluster. Individual E.coli BW25113 knockout mutants, DiscS, DsufS and DcsdA were available from the Yale Coli Genetic Stock Center. The iscS mutant did not survive the transport and was reconstructed (Fig.  S1).
E. coli BW25113 and the three cysteine desulfurase mutants were transformed with pJTU3510 expressing dptB-E, and tested for the phosphorothioation status by Dnd phenotypic assay (DNA smear, an indicator of DNA phosphorothioate modification) (Fig. 2). Only the iscS mutant failed to modify its DNA (lane 4), suggesting that only E. coli IscS, but not SufS or CsdA, was responsible for DNA phosphorothioation in E. coli.
To confirm that iscS is responsible for DNA phosphorothioation, iscS was cloned into pET15b, and co-transformed with a dpt gene cluster harboring low copy number plasmid pJTU3510 into the E. coli iscS deletion mutant. Fig. 2 lane 9 shows that DNA phosphorothioation was restored in the strain, proving that IscS, in cooperation with DptB-E, restored DNA phosphorothioation in E. coli. Involvement of IscS in DNA phosphorothioation in E. coli was further confirmed by site-directed mutagenesis. Three conserved cysteine residues in IscS, were mutated to Ala, generating three iscS cysteine mutants (C111A, C170A, and C328A). These mutants were again co-transformed with pJTU3510 (harboring the dpt gene cluster) into the iscS deletion mutant. Fig. 2

IscS might participate in DNA phosphorothioation directly
The cysteine desulfurase IscS is a highly conserved master enzyme initiating sulfur transfer via persulfide to a range of acceptor proteins. IscS is involved in various physiological processes, including Fe-S cluster assembly, tRNA modification, and sulfur-containing cofactor biosynthesis. IscS-interacting partners, including IscU, TusA, ThiI, ThiF and MoeB are sulfur acceptors. Other proteins, such as CyaY, IscA and IscX, also bind to IscS, but their functional roles are not directly related to sulfur transfer [16].
Mutants of cyaY, iscA, iscU, iscX, moeB, tusA, thiF, thiI and thiS, proteins known to interact with IscS in E. coli, were tested for their possibility to participate into DNA phosphorothioation. Fig. 3 shows that none of these genes was required for the modification, as assayed by Dnd phenotype. This suggested that IscS in E. coli might participate directly into the modification process.

Protein-protein interactions between IscS and Dpt proteins
The bacterial two-hybrid system was used to detect interactions between E. coli IscS and DptB, C, D and E. IscS was fused with the bait protein, while DptB, C, D, and E were fused with the target protein.
Strong protein-protein interactions were immediately detected between IscS and DptC (2% surviving cells on 3AT), IscS and DptE (2% surviving cells on 3AT), but not between IscS and DptB and DptD (Fig. 4A). These protein interactions were confirmed further by plating the co-transformed strains on medium containing streptomycin (Fig. 4B).
Protein-protein interaction between IscS and DptC as well as IscS and DptE were further confirmed by pull-down experiments. Fig. 4C shows that Strep tagged IscS can pull-down both GST tagged DptC and DptE. Reciprocally, GST tagged DptC and DptE can also pull-down Strep tagged IscS.

Discussion
IscS is a highly conserved, but functionally versatile pyridoxal-59-phosphate (PLP)-dependent enzyme. It delivers sulfur to players within various metabolic pathways, including iron-sulfur cluster assembly, thiamine and biotin synthesis, tRNA modifications, and molybdopterin biosynthesis [16,17]. We show here that IscS can also participate in DNA phosphorothioation.
The involvement of IscS in DNA phosphorothioation could be direct or indirect. By analyzing the Dnd phenotype and the mutants (Fig. 3), we were able to rule out the possibility that IscS participates indirectly via other pathways. We hypothesized that if IscS is involved in the DNA phosphorothioation process directly, we might be able to detect protein-protein interaction between IscS and the Dnd proteins. In keeping with this hypothesis, protein interaction between IscS and DndE and DndC were detected using the bacterial two hybrid system.
There are two potential functions of IscS in the process of DNA phosphorothioation. One is Fe-S cluster assembly for the DndC protein. It is known that DndA can catalyze apo-Fe DndC to its Fe-S cluster form [5]. Another function might be to transfer sulfur from cysteine to the target DNA via protein interactions with the Dnd proteins, which is reminiscent of tRNA modification [18,19]. These hypothesises are currently under intensive investigation.