Iron Regulation in Clostridioides difficile

Abstract

The response to iron limitation of several bacteria is regulated by the ferric uptake regulator (Fur). The Fur-regulated transcriptional, translational and metabolic networks of the Gram-positive, pathogen Clostridioides difficile were investigated by a combined RNA sequencing, proteomic, metabolomic and electron microscopy approach. At high iron conditions (15 μM) the C. difficile fur mutant displayed a growth deficiency compared to wild type C. difficile cells. Several iron and siderophore transporter genes were induced by Fur during low iron (0.2 μM) conditions. The major adaptation to low iron conditions was observed for the central energy metabolism. Most ferredoxin-dependent amino acid fermentations were significantly down regulated (had, etf, acd, grd, trx, bdc, hbd). The substrates of these pathways phenylalanine, leucine, glycine and some intermediates (phenylpyruvate, 2-oxo-isocaproate, 3-hydroxy-butyryl-CoA, crotonyl-CoA) accumulated, while end products like isocaproate and butyrate were found reduced. Flavodoxin (fldX) formation and riboflavin biosynthesis (rib) were enhanced, most likely to replace the missing ferredoxins. Proline reductase (prd), the corresponding ion pumping RNF complex (rnf) and the reaction product 5-aminovalerate were significantly enhanced. An ATP forming ATPase (atpCDGAHFEB) of the F₀F₁-type was induced while the formation of a ATP-consuming, proton-pumping V-type ATPase (atpDBAFCEKI) was decreased. The [Fe-S] enzyme-dependent pyruvate formate lyase (pfl), formate dehydrogenase (fdh) and hydrogenase (hyd) branch of glucose utilization and glycogen biosynthesis (glg) were significantly reduced, leading to an accumulation of glucose and pyruvate. The formation of [Fe-S] enzyme carbon monoxide dehydrogenase (coo) was inhibited. The fur mutant showed an increased sensitivity to vancomycin and polymyxin B. An intensive remodeling of the cell wall was observed, Polyamine biosynthesis (spe) was induced leading to an accumulation of spermine, spermidine, and putrescine. The fur mutant lost most of its flagella and motility. Finally, the CRISPR/Cas and a prophage encoding operon were downregulated. Fur binding sites were found upstream of around 20 of the regulated genes. Overall, adaptation to low iron conditions in C. difficile focused on an increase of iron import, a significant replacement of iron requiring metabolic pathways and the restructuring of the cell surface for protection during the complex adaptation phase and was only partly directly regulated by Fur.

Introduction

Clostridioides difficile (formerly Clostridium difficile) is a spore-forming, Gram-positive, anaerobic, toxins-producing pathogen leading to often hospital-acquired infections worldwide (Burke and Lamont, 2014). The phenotypes of C. difficile infections (CDI) range from mild diarrhea to toxic megacolon which ultimately causes death (Bartlett and Gerding, 2008). In the United States over half a million cases of CDI per year with approximately 30,000 deaths are reported, making CDI to one of the most common and also cost-effective healthcare-associated infections (Lessa et al., 2015). Proteins containing iron, [Fe-S]-clusters and iron-coordinated heme are indispensable for the bacterial metabolism. Consequently, iron is an essential element for the growth of all bacteria including C. difficile (Symeonidis, 2012). Despite its abundance in nature, iron is often a growth-limiting nutrient due to the low solubility of the dominating oxidized ferric iron over the soluble ferric form (Braun and Hantke, 2011). To counteract this problem, bacteria have developed high affinity transporters and high affinity chelators, so called siderophores, which are excreted and re-imported after iron acquisition to cope with this limitation (Huang and Wilks, 2017; Khan et al., 2018). Alternatively, ferric reductases are excreted (Schroder et al., 2003). In pathogenic bacteria these iron-uptake mechanisms acquire iron directly from host proteins, including the iron-binding glycoproteins transferrin in serum and extracellular fluid, lactoferrin in mucosal secretions, and heme-containing proteins such as hemoglobin, haptoglobin, and hemopexin (Symeonidis, 2012). C. difficile can utilize different iron salts (FeCl₃, FeSO₄), iron citrate and ferritin as iron source (Cernat and Scott, 2012). In a previous investigation ferritin, hemoproteins and heme were able to sustain growth of C. difficile under iron-limited condition (Cernat and Scott, 2012). However, a cellular overload with iron has to be avoided to prevented reactive oxygen species generation via the Fenton reaction (Cornelis et al., 2011). As a consequence, bacteria have evolved various mechanisms to control iron homeostasis. They carefully adjust their iron uptake and utilization strategies at the transcriptional level (Troxell and Hassan, 2013; Porcheron and Dozois, 2015). Several iron-responsive regulatory proteins (Fur, Irr, RirA, and IscR) have been described in bacteria (Rudolph et al., 2006; Santos et al., 2015; Mandin et al., 2016). The ferric uptake regulator (Fur) protein is a transcriptional repressor of genes in iron uptake and utilization (Troxell and Hassan, 2013; Fillat, 2014; Porcheron and Dozois, 2015). The Fur protein typically contains two structural domains, the N-terminal DNA binding domain and the C-terminal dimerization domain (Deng et al., 2015). Under iron-replete conditions, Fe²⁺ functions as a co-repressor in that the Fur–Fe²⁺ complex binds a conserved DNA site in the promoter of a regulated gene and usually inhibits the expression. In contrast, under iron starvation conditions, the Fur protein is inactive, which allows for the expression of Fur-regulated genes.

The Fur regulons of Clostridium acetobutylicum and C. difficile were determined using DNA microarray-based transcriptome analyses (Vasileva et al., 2012; Ho and Ellermeier, 2015). In C. difficile one transcriptome investigation focused on high iron versus iron-depleted conditions (Hastie et al., 2018), while the second defined the Fur-regulon under high iron conditions (Ho and Ellermeier, 2015). In C. acetobutylicum genes for various siderophore uptake systems (feo, fhu), a flavodoxin (fldX), lactate dehydrogenase (ldh), benzoyl-CoA reductase and riboflavin biosynthesis (rib) were found under Fur-mediated iron control. A Fur binding site of G/T-A/T-T/G-A-A-T-N-A/T-T/A-T/A- T-C-A-T/A-T/A-A/T was proposed (Vasileva et al., 2012). Similarly, in C. difficile genes for 7 putative cation transport systems including various iron uptake systems (fpi, feo, and fhu,) a flavodoxin (fldX), two component regulatory systems and very few metabolic enzymes were found repressed by Fur in an iron-dependent manner. But also a series of Fur induced genes were identified. Furthermore, in vitro DNA binding by Fur was shown (Ho and Ellermeier, 2015). The deduced Fur binding site was A-A-A-T-G-A-T-A-A-T-N-A-A/T-T/A-A/T-T-C-A. A similar binding site A/T-A/T-N- T/A-N-T-G-A-T-A-A-T-G-A-T-T-T-T-C-A-T-T-A/T was proposed by (Dubois et al., 2016). They demonstrated cysteine-dependent regulation of fur and several fur target genes. Finally, the C. difficile Fur regulon was found induced in a hamster infection model control (Ho and Ellermeier, 2015). During the second DNA-array-based transcriptome approach focusing on iron versus iron-depleted conditions, genes for a flavodoxin, enzymes of polyamine and histidine biosynthesis, and flagella formation were found induced under iron limiting conditions (Hastie et al., 2018). Corresponding studies in Clostridium perfringens identified FeoB as the major systems to counteract iron depletion in this bacterium (Awad et al., 2016). Finally, a bioinformatics investigation proposed the DNA binding site for Clostridium botulinum Fur as A/T-T/A- T-N-A/T-T/A- A-A/T-T-A/T-A-T/A-T/A-A-T-T-A/T-T-T (Zhang et al., 2011). A position weight matrix analyses was employed for regulon prediction.

Here we describe a combined RNA sequencing-based transcriptomic, proteomic, metabolomic and electron microscopy approach to characterize multiple functional and metabolic changes induced by the Fur-mediated low iron response. Multiple cellular processes aside of iron transport including mainly energy metabolism, but also flagella formation and motility, cell wall architecture and antibiotic/CAMP resistance were controlled by iron and partly by Fur in C. difficile.

Materials and Methods

Bacterial Strains and Growth Conditions

Escherichia coli DH5a [fhuA2 lac(del)U169 phoA glnV44 Φ80’ lacZ(del)M15 gyrA96 recA1 relA1 endA1 thi-1 hsdR17], DH10B [F^-mcrA Δ(mrr-hsdRMS-mcrBC) Φ80dlacZΔM15 ΔlacX74 endA1 recA1 deoR Δ(ara,leu)7697 araD139 galU galK nupG rpsL λ^-], and CA434 (E. coli HB101 carrying the Incβ conjugative plasmid R702) were grown in LB medium supplemented with 100 mg/l ampicillin or 20 mg/l chloramphenicol as required. C. difficile 630Δerm cells were grown in Brain-Heart-Infusion (BHI) medium (37 g/l) supplemented with 0.1% L-cysteine and 5 mg/ml yeast extract. During mutagenesis experiments C. difficile supplement (250 mg/l D-cycloserine and 8 mg/l cefoxitin) (Sigma Aldrich, Taufkirchen, Germany) and 2.5 mg/l erythromycin (Carl Roth, Karlsruhe, Germany) were added. Growth and Omics experiments were performed in Clostridium difficile minimal medium (CDMM) (Neumann-Schaal et al., 2015) under anaerobic conditions using an anaerobic chamber from Coy Laboratories (Grass Lake, MI, United States). Different iron sources were tested as additives to the medium including 15 μM iron-sulfate, 15 μM iron-chloride, 15 μM iron-citrate, 10 μM hemin, 10 μg/ml ferritin and 10 μg/ml transferrin unless stated otherwise. Cells were harvested (10 min, 8,000 × g) anaerobically using gas-tight polypropylene tubes (TPP, Trasadingen, Switzerland) and harvested cells and/or supernatant were used for transcriptome, proteome and metabolome analyses as described below.

Construction of the C. difficile fur Mutant and a fur Containing Vector for Complementation

The vector for fur inactivation was designed with the help of the ClosTron website¹ using the Perutka algorithm (Perutka et al., 2004). E. coli CA434 was transformed with the resulting pMTL007C-E2_fur274a::intron vector (pMTL007C-E2 retargeted to CD630Δerm fur274a::intron, ermB) for mating with C. difficile 630Δerm cells after a standard protocol (Heap et al., 2007, 2009, 2010). The desired fur mutant was identified using primers Cdi-fur-F (5′-CTGGTTTTAAGATTACGCCAC-3′), Cdi-fur-R (5′-CCATTACACTCGTCACATAGTC-3′), EBSuni-versal (5′CGAAATTAGAAACTTGCGTTCAGTAAA-3′), Erm-RAM-RF (5′-ACGCGTTATATTGATAAAAATAATAGTGGG-3′), ErmRAM-R (5′-ACGCGTGCGACTCATAGAATTATTTCCTCCCG-3′) as described and documented in the Supplementary Material. For complementation of the C. difficile fur mutant, a PCR fragment covering the region from 300 bp upstream to 100 bp downstream of the fur gene (CD630_12870) was amplified using chromosomal C. difficile 630 DNA and the primers Cdi-fur-compl.NotI-F (5′-ATCAGCGGCCGCCAGATATTTATTATATTTGC-3′ and Cdi-fur-compl.HindIII-R (5′-ATCAAAGCTTAATGGAAGAATAGCATAG-3′) digested with NotI and HindIII and cloned into the appropriately cut shuttle vector pMTL82151 to generate pMTL82151_fur (Heap et al., 2009).

Field Emission Scanning Electron Microscopy (FESEM)

Clostridioides difficile 630Δerm and corresponding fur mutant were grown anaerobically in CDMM with and without addition of 15 μM iron-sulfate at 37°C to mid-exponential phase, harvested and fixated with 5% formaldehyde. Afterwards, the cells were washed with TE-buffer (20 mM TRIS, 1 mM EDTA, pH 6.9) before dehydration in a graded series of acetone (10, 30, 50, 70, and 90%) on ice for 15 min for each step. The 100% acetone dehydration step was performed at room temperature. Then, samples were critical-point dried with liquid CO₂ (CPD 30, Bal-Tec, Balzers, Liechtenstein) and covered with a gold-palladium film by sputter coating (SCD 500, Bal-Tec, Balzers, Liechtenstein) before being examined in a field emission scanning electron microscope (Zeiss DSM 982 Gemini, Oberkochen, Germany) using the Everhart Thornley SE detector and the in lens detector in a 50:50 ratio at an acceleration voltage of 5 kV.

RNA Sequencing

Clostridioides difficile 630Δerm and corresponding fur mutant were grown anaerobically in CDMM with and without addition of 15 μM iron-sulfate at 37°C to mid-exponential phase and harvested. Employed CDMM without additions contained 0.2 μM iron. Due to the different growth behavior of both strains the mid-exponential growth rate was reached by both strains at different time points. At these two time points both strains revealed comparable growth rates. Total bacterial RNA was isolated from bacterial cell pellets as described before (Rosinski-Chupin et al., 2014). Residual DNA was removed using TURBO DNase (Ambion, Thermo Fisher Scientific, Waltham, MA, United States). Resulting DNA-free RNA was further purified with phenol:chloroform:isoamylalcohol (25:24:1) extraction. Remaining traces of phenol were removed by washing the samples twice with chloroform:isoamylalcohol (24:1). RNA integrity was assessed using the Agilent RNA 6000 Nano Kit on the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, United States). Transfer RNA was depleted from the total RNA using Microbexpress (Ambion, Thermo Fisher Scientific, Waltham, MA, United States). To 1 μg of rRNA depleted total RNA 1 μl of either 1:10 diluted ERCC ExFold RNA Spike-In Mix 1 or 2 (Ambion, Thermo Fisher Scientific, Waltham, MA, United States) was added. RNA was subsequently treated with tobacco acid pyrophosphatase (TAP) (Epicentre Biotechnologies, Madison, WI, United States). Prior to cDNA library preparation, RNA was further purified with phenol:chloroform:isoamylalcohol (25:24:1), any remaining phenol traces were removed by washing the samples twice with chloroform:isoamylalcohol (24:1), RNA was recovered by ethanol precipitation. Strand-specific RNA-Seq cDNA library preparation and barcode introduction based on RNA adapter ligation was performed as described previously (Nuss et al., 2015). Library quality was validated using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, United States) following the manufacturer’s instruction. Cluster generation was performed using the Illumina cluster station. Single-end sequencing on the Illumina HiSeq2500 followed a standard protocol. The fluorescent images were processed to sequences and transformed to FastQ format using the Genome Analyzer Pipeline Analysis software 1.8.2 (Illumina, San Diego, CA, United States). The sequence output was controlled for general quality features. Sequencing adapter clipping and demultiplexing was done using the fastq-mcf and fastq-multxtool of ea-utils². DNA sequencing output was analyzed using the FastQC tool (Babraham Bioinformatics, Cambridge, United Kingdom). All sequenced libraries were mapped to the C. difficile 630 genome (NC_009089.1) and the corresponding pCD630 plasmid (NC_008226.1) using Bowtie2 (version 2.1.0) (Langmead and Salzberg, 2012) with default parameters. ERCC mapping and analysis were performed after supplier’s instructions. After read mapping, SAMtools (Li et al., 2009) was employed to filter the resulting bam files for uniquely mapped reads (both strands), which were the basis for downstream analyses. Differential gene expression was evaluated using the DESeq2 tool as part of the Bioconductor software package. Throughout the manuscript the data were adapted to the C. difficile 630Δerm annotation. In Table 1 both annotations are given. For the interconversion of the original data shown in Supplementary Tables S1–S3 and also the proteomics data in Supplementary Table S4 in the Supplemental Material an appropriate conversion table (Supplementary Table S5) is provided. The data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (Edgar et al., 2002) and are accessible through GEO Series accession number GSE120189.

Proteomics

Bacteria were grown as outlined for the RNA-seq experiments. Cell pellets were suspended in 700 μl of ice-cold urea-containing buffer (7 M urea, 2 M thiourea, 50 mM dithiothreitol (DTT), 4% (w/v) 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS), 50 mM Tris-HCl). Cell lysis was performed by sonication (probe MS73, Sonoplus, Bandelin, Berlin, Germany) in six cycles of 60s (amplitude 60%, 0.1 s pulse every 0.5 s) on ice. Cell debris was removed by centrifugation at 6,000 g for 10 min at 4°C. Proteins of cell free resulting lysates were precipitated by addition of ice-cold acetone [in a 1:5 ratio (v/v)] for 20 h at -20°C. Subsequently, samples were allowed to warm to room temperature and were centrifuged at 22,000 g for 45 min at room temperature. The supernatant was discarded, the protein pellets were washed in 80% acetone, and subsequently in 100% acetone, before they were air-dried. The protein pellets were solubilized in SDS-containing urea-buffer [7 M urea, 2 M thiourea, 1% (v/v) SDS]. For protein concentration determination, 10 μl of each sample was separated by SDS-PAGE (Criterion TGX Precast Gels 4–20%, Bio-Rad, Hercules, CA, United States). Resulting SDS-gels were fixed for 1 h in 40% (v/v) EtOH, 10% (v/v) glacial acidic acid, washed in H₂O and stained by the Flamingo fluorescent dye (Bio-Rad, Hercules, CA, United States) for 1 h. Resulting fluorescence signals of the samples were measured by a Typhoon TRIO scanner (GE Healthcare, Little Chalfont, United Kingdom), quantified by ImageQuant 5.2 (GE Healthcare, Little Chalfont, United Kingdom) and used for quantitative normalization of protein. Comparable protein amounts (∼30 μg of protein per sample) for each analyzed condition were separated by SDS-PAGE as described above and stained overnight with Colloidal Coomassie. Gel lanes were cut into 10 slices and proteins subjected to in-gel trypsinization as described previously (Lassek et al., 2015).

The eluted peptides were subjected to LC-MS/MS analyses performed on a Proxeon nLC 1000 coupled online to an Orbitrap Elite mass spectrometer (Thermo Fisher Scientific, Waltham, MA, United States). In-house self-packed columns [i.d. 100 μm, o.d. 360 μm, length 150 mm; packed with 1.7 μm Aeris XB-C18 reversed-phase material (Phenomenex, Aschaffenburg, Germany)] were loaded and washed with 10 μl of buffer A [0.1% (v/v) acetic acid] at a maximum pressure of 750 bar. For coupled LC-MS/MS analysis, elution of peptides took place with a non-linear 80 min gradient from 1 to 99% buffer B [0.1% (v/v) acetic acid in acetonitrile] at a constant flow rate of 300 nl/min. Eluting peptides were recorded in the mass spectrometer at a resolution of R = 60,000 with lockmass correction activated. After acquisition of the full MS spectra, up to 20 dependent scans (MS/MS) were performed according to precursor intensity by collision-induced dissociation fragmentation (CID) in the linear ion trap. For protein identification and quantification from raw MS data, the Proteome Discoverer^TM software (version 1.4, Thermo Fisher Scientific Inc., Waltham, MA, United States) was used, and results further evaluated employing Scaffold (version 4.4, Proteome Software Inc., Portland, OR, United States) as previously described in detail (Lassek et al., 2015). In brief, Sequest HT database searches were performed with raw files against a C. difficile 630 protein database containing common contaminations (3804 entries). The following search parameters were used: enzyme type = trypsin (KR), peptide tolerance = 10 ppm, tolerance for fragment ions = 0.6 Da, b- and y-ion series, variable modification = methionine (15.99 Da); a maximum of three modifications per peptide was allowed. Peptide and protein identifications were accepted with a false discovery rate (FDR) of at most 1%, requiring a minimum of at least two unique peptides for protein identification and quantification. Moreover, only proteins that were at least identified in two out of three biological replicates were taken into account. Relative protein quantification was achieved by calculating the normalized area under the curve (NAUC). Identification of statistical differences in relative protein amounts was performed using t-test (p-value < 0.05) including adjusted Bonferroni correction and all possible permutations. Proteome data are summarized in Supplementary Table S4. Data of interest can be easily converted into the C. difficile 630Δerm annotation using conversion Supplementary Table S5. All MS raw data as well as Proteome Discoverer and Scaffold result files have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository (Vizcaino et al., 2016) with data set identifier PXD011161.

Metabolomics

Cells were grown to the mid-exponential growth phase and harvested anaerobically as indicated above for the transcriptome and proteome experiments. The supernatant was removed and the cells were immediately quenched by suspension in pre-cooled isotonic sodium chloride-methanol [50% (v/v), -32°C]. Cells were pelleted at -20°C, 8,000 g for 5 min. The quenching solution was removed and the cells were frozen in liquid nitrogen. Cell lysis and metabolite extraction were performed as described previously (Zech et al., 2009; Dannheim et al., 2017b). One ml of the polar phase was dried in a vacuum concentrator and stored at -80°C prior to analysis. Extracellular samples were prepared as described previously (Neumann-Schaal et al., 2015). Volatile and non-volatile compounds in the culture supernatants and cell free extracts were analyzed via GC/MS as described earlier (Neumann-Schaal et al., 2015). Raw data obtained from GC/MS measurements were processed by applying version 2.2N-2013-01-15 of the in-house developed software MetaboliteDetector (Hiller et al., 2009). The peak identification was performed in a non-targeted manner with a combined compound library. After processing, non-biological peaks and artifacts were eliminated with the aid of blanks. Peak areas were normalized to the corresponding internal standards (o-cresol or ribitol) and derivatives were summarized. Significant changes in metabolite levels were calculated by non-parametric Wilcoxon–Mann–Whitney test (Mann and Whitney, 1947) using Benjamini–Hochberg correction (Benjamini and Hochberg, 1995) to control the false discovery rate. Metabolome data are summarized in Supplementary Table S6.

HPLC/MS-Based Analysis of Coenzyme A-Derivatives

Coenzyme A (CoA)-esters were isolated by cell breakage using a Precellys 24 homogenizer (Peqlab, Erlangen, Germany) at -10°C. The procedure included three cycles of homogenization (6,800 rpm, 30 s with equivalent breaks). The lysate was transferred to 10 ml of ice-cold ammonium acetate (25 mM, pH 6) and centrifuged (5 min at 10,000 g, 4°C). CoA-derivatives were extracted on a Strata XL-AW solid phase extraction column (Phenomenex, Aschaffenburg, Germany) as described previously (Wolf et al., 2016). CoA-derivatives were analyzed on a Dionex ultimate 3000 system (Thermo Scientific Inc., Darmstadt, Germany) coupled to a Bruker MicroTOF QII mass spectrometer (Bruker Daltonik GmbH, Karlsruhe, Germany) equipped with an electrospray ionization interface. The separation and detection was performed as described previously (Peyraud et al., 2009; Wolf et al., 2016). Raw data were processed using the XCMS package (Smith et al., 2006; Benton et al., 2008; Tautenhahn et al., 2008) for R (version 3.0.3) as described previously (Wolf et al., 2016). Significant changes in metabolite levels were calculated by non-parametric Wilcoxon-Mann-Whitney test (Mann and Whitney, 1947) using Benjamini–Hochberg correction (Benjamini and Hochberg, 1995) to control the false discovery rate. Metabolome data are summarized in the Supplementary Table S6.

Bioinformatics

De novo Motif Search

Motif search was performed with the standalone version of MEME (Bailey et al., 2009) on the promoter sequences [-250,0] of 66 genes known to be differentially regulated by Fur and involved in the iron metabolism. MEME was run with option “-anr” and without any restrictions on the motif width. Motif presence was confirmed in 11 out of the 66 promoters.

Genome-Wide Motif Scan

We performed genome-wide motif search in the promoters [-250,0] of C. difficile using the de novo obtained position weight matrix (PWM). The standalone version of the MAST tool available in the MEME package was run with option “-norc” (search only the forward strand) once with default other parameters and once with maximal motif hit P-value of 5.10(-5).

Results

Construction of a C. difficile fur Mutant and Definition of High and Low Iron Growth Conditions

The overarching aim of this study was identification and characterization of the Fur regulon at the transcriptional, translational, metabolomics and the phenotypic level. For this purpose a fur mutant was constructed using the ClosTron technology (Heap et al., 2007, 2009, 2010). The fur gene was identified and characterized before by Ho and Ellermeier (2015) and partially by Dubois et al. (2016). Similar to their approaches, a stabile insertional mutation in open reading frame CDIF630_01441 of the laboratory strain C. difficile 630Δerm (Hussain et al., 2005) was generated. The ClosTron system uses a group II intron to insert an erythromycin resistance cassette into the target gene. C. difficile 630Δerm, an erythromycin-sensitive derivative of C. difficile strain 630, was used as the parental strain and is further referred to as wild type. The insertional mutant was confirmed by PCR analysis (Supplementary Figure S1). The growth behavior of the wild type and the constructed fur mutant in logarithmic growth phase was almost identical when tested in the complex Brain-Heart-Infusion (BHI) medium independent of the addition of iron (Supplementary Figure S2). However, the stationary phase was entered earlier by the fur mutant. Similar observations have been made for the fur mutant grown in complex TY medium before (Ho and Ellermeier, 2015). A different growth behavior was observed in Clostridium Minimal Medium (CDMM). Here we tested high (15 mM) and low (0.2 mM) concentrations of iron, in this case iron sulfate. For this purpose a commercial analytical laboratory (Currenta Analytik, Leverkusen, Germany) investigated the CDMM used throughout this investigation with Inductively-Coupled-Plasma Mass Spectrometry (ICP-MS) for its iron content. Highly reproducible, 0.2 mM iron were determined for the medium and used as low iron conditions. For defining high iron conditions CDMM was titrated with increasing amounts of iron and C. difficile wild type growth stimulation was determined. When the point of no further growth stimulation was reached 9.2 mM iron were measured by ICP-MS in CDMM.

We explicitly circumvented the utilization of the chelator 2,2′-dipyridyl (DPP) to achieve low/no iron conditions. Cernat and Scott failed after DPP treatment of C. difficile to recover the bacterial growth by the addition of alternative iron sources including lactoferrin, transferrin, hemoprotein, and heme (Cernat and Scott, 2012). A high-throughput small molecule screen identified DPP as one of the most potent inhibitors of C. difficile growth, even in a mouse model (Katzianer et al., 2014). Latter indicated the importance of iron for C. difficile growth, but also showed the detrimental effects of DPP treatment. Nevertheless, DPP remains an useful and often used tool to achieve complete iron depletion. To our understanding C. difficile does not encounter completely iron free conditions in its environment, thus, we compared high (15 μM) and low (0.2 μM) iron conditions in all experiments of this study. When growth of the wild type and the fur mutant was compared under both iron concentrations, both strains revealed significant reduced growth under iron limited conditions (Figure 1). Complementation of the fur mutant with a plasmid encoded fur restored growth to almost wild type conditions (Supplementary Figure S3). Furthermore, the fur mutant grew much slower and to lower terminal densities compared to the wild type strain. Obviously, Fur is required for optimal growth under high and low iron conditions (Figure 1).

FIGURE 1

Next, different iron sources were analyzed for the ability to restore iron limited growth of the wild type and the fur mutant (Supplementary Figure S4). Addition of 15 mM iron citrate or iron (II) chloride induced a growth behavior of C. difficile similar to that observed for the addition of iron sulfate (compare Figure 1 and Supplementary Figures S4A,B,D). Addition of 10 μM hemin, or 10 μg/ml transferrin did not significantly improve wild type growth, but slightly enhanced the growth of the fur mutant. Substitution with 10 μg/ml ferritin clearly improved the growth of both strains (compare Figure 1 and Supplementary Figures 4C,E,F). Overall, the basic difference in the growth behaviors of the wild type and the mutant strain remained similar under various tested iron conditions, i.e., the various iron sources did not compensate for the loss of Fur in the mutant strain. Obviously, additional functions besides iron regulation are executed by Fur in C. difficile.

Transcriptome, Proteome and Metabolome of Wild Type and the fur Mutant of C. difficile Grown at Iron-Limiting and Iron-Saturated Conditions

We aimed at a multi-level, holistic view on iron-regulation in C. difficile and the contribution of Fur to these processes. To analyze environmental iron conditions close to the gut habitat, we refrained from DPP treatment of the cultures, rather we compared samples taken from low, iron limiting growth conditions (0.2 μM) with samples of iron saturated (15 μM) growth conditions. The transcriptome (RNA-Seq), cytoplasmic proteome, metabolome, and exo-metabolome of wild type and the fur mutant grown under both conditions were compared. Samples were taken in the exponential growth phase as indicated by arrows in Figure 1. This approach enabled us to functionally identify iron regulated processes at the transcriptional and proteomic level, and to observe their metabolic consequences. Furthermore, the inhibitory and promoting activities of Fur became visible. Certain phenotypes were further investigated using electron microscopy and growth experiments.

The RNA-Seq approach identified 3,156 individual transcripts. First, we compared the 4 different transcriptomes (wild type low/high iron, fur mutant low/high iron) by principal component analyses (Supplementary Figure S5). Interestingly, biological triplicates from wild type/low iron, fur/low iron, and fur/high iron showed a certain degree of overlap, while triplicates for wild type/high iron clustered very distinct. As Fur usually acts as a transcriptional repressor at high iron concentration, global transcriptional changes due to high iron availability were mostly effected by the presence of active Fur. The terms “induced” and “repressed” were used for enriched or depleted RNAs throughout the paper. We are fully aware of the fact that comparative RNA-Seq shows changes in RNA abundancies, which might not always correlate with changes in gene expression. Using a log2 fold change of 1 in transcript abundance (p-value of 0.05) as cutoff, 243 genes were found up- and 303 genes down-regulated in response to iron limitation (Supplementary Tables S1–S3). Comparing wild type and the fur mutant at high iron 369 genes were found up- and 268 genes found down-regulated (Table 1). In order to visualize the differences of the currently available transcriptome wild type (Ho and Ellermeier, 2015; Hastie et al., 2018) the principal component analysis was employed for all available transcriptome data of C. difficile wild type versus fur mutant at high iron growth conditions. Results are summarized in Supplementary Tables S1–S3. We were aware of the fact that highly different transcriptome methods (RNA-Seq versus DNA array) and different low/no iron condition (with and without DPP) were compared. Clear cut differences became visible (Figure 2). The DNA array data of the wild type obtained in the presence of high iron and low/no iron (Hastie et al., 2018) cluster together, nevertheless, with some distance. The DNA array data for the fur mutant obtained at high iron conditions (Ho and Ellermeier, 2015) cluster separate from the RNA-Seq data, however, with the wild type data oriented toward the RNA-Seq wild type data and the fur mutant data toward the RNA-Seq fur mutant data (Figure 2).

FIGURE 2

Analyzing iron limitation in C. difficile with a proteomics approach, a total of 1,639 proteins were identified. A recent investigation of 8 C. difficile proteome yielded 662 quantifiable common proteins (Dresler et al., 2017). Using a cutoff at a log2 fold of 1 (p-value of 0.05) 85 proteins were found down- and 61 up-regulated in response to iron limitation. A total of 170 proteins were not found (OFF) and 85 solely found (ON) under iron limiting conditions (Supplementary Table S5). For the wild type versus fur mutant comparison 1,682 proteins were analyzed. Using the same cutoff 86 proteins were found depleted and 122 enriched in the fur mutant compared to wild type, both grown at high iron conditions. A total of 152 proteins were not detected (OFF) and 128 proteins solely identified (ON) in the fur mutant (Supplementary Table S4). Comparing transcriptome and proteome data, interesting differences were observed, most likely reflecting the delay of the response of the proteome compared to the transcriptome at the analyzed time point (Table 1). These differences will be described and discussed in the context of the various regulated processes below. Furthermore, a significant degree of similarity was observed for the Omics data for high versus low iron and the wild type versus fur mutant at high iron conditions, indicating that major adaptations were controlled directly or indirectly by Fur (Supplementary Table S4).

Combined GC/MS- and LC/MS-based metabolome approaches were employed for the analyses of intracellular metabolites including CoA-esters and for the elucidation of the metabolic composition of the growth medium and corresponding volatiles. Overall, we identified 113 intracellular metabolites including 23 coenzyme A-esters. Extracellularly, 45 metabolites were identified. Using a fold change cutoff of 1.5 and at an adjusted p-value of 0.05, 52 metabolites were found in higher concentration and 13 in lower concentration under low iron conditions (Supplementary Table S5). For the wild type versus fur mutant comparison using the same cutoff 29 metabolites were found more and 31 less abundant in the fur mutant compared to wild type when both were grown at high iron conditions. Overall, the most abundant identified metabolites were dominated by amino acids and their products (5-aminovalerate, glutamate, valine, isoglutamate, leucine, lysine, alanine and more) followed by diverse coenzyme A-esters, cofactors and polyamines (spermine, spermidine). As typically observed for C. difficile, only a few sugars and activated sugars (glucose, glucose-6-phosphate and fructose-1,6-bisphosphate) or intermediates of the central carbon metabolism (2-oxoglutarate) were under the highly abundant metabolites (Supplementary Table S6).

Finally, a bioinformatics approach for the definition of the Fur regulon was taken. Fur binding sites upstream of Fur-regulated genes in C. difficile were combined to define a position weight matric using the MEME motif search tool version 4.11.2. A consensus binding site of TGATAATVAWHWTCA was deduced (Figure 2). Overall, 161 potential strand-specific Fur binding sites were identified up to 250 bp upstream of 147 coding genes/operons. Approximately 20 of these binding sites were found upstream of genes involved in the regulation of the major adaptations processes to low iron condition in C. difficile (Supplementary Table S7).

Fur-Mediated Iron Regulation of Metal Uptake Systems

As expected various iron and other metal uptake systems encoded by fpi, fhu, zupT and the sulfonate transporters encoded by the ssu operon (CDIF630erm_03273–03276) were found more abundant by low iron conditions at the transcriptome and proteome level (Table 1). This response was indirectly mediated by Fur, since a conserved Fur binding site was not detected upstream of the ssu operon. This is in agreement with previously published transcriptome analyses induced (Ho and Ellermeier, 2015; Hastie et al., 2018). In the previous two transcriptome analyses using DPP-treated bacteria as iron depleted condition, the ferrous iron uptake transporter genes feoA1 (CDIF630erm_01641 – 01642) were also found clearly induced (Ho and Ellermeier, 2015; Hastie et al., 2018). Moreover, low induction by iron depletion was observed for feoA5/feoB3 (CDIF630erm_03573 – 03574) and feoA4 (CDIF630erm_01939). None of the FeoA type systems were found more abundant at the transcriptome level in our approach with 0.2 μM iron as low iron conditions. However, the proteome data revealed that FeoAB system encoded by CDIF630erm_01641 – 01643 was induced at low iron condition in a Fur-dependent manner (Table 1). Moreover, the significant differences in the observed fold changes in gene induction between DPP-treated cells (up to 730-fold with the DNA array, over 100-fold for the RNA-seq) and 0.2 μM iron grown cells (around 5-fold) might further explain some of these observations. Possibly, at an iron concentration of 0.2 μM the necessary threshold of iron depletion for the Feo-type systems was not reached. Alternatively, feo gene regulation by low iron with Fur was of transient nature and finished at the time point of sampling. Possibly, the adaptation at this certain time-point was only visible at the proteome level (summarized in Figure 3). Many of the operons/genes (CDIF630erm_01824, CDIF630erm_01827, CDIF630erm_03145, CDIF630erm_03146, CDIF630erm_01641, CDIF630erm_01643, CDIF630erm_01939) involved in iron uptake contain potential Fur binding sites in their upstream region, indicating direct Fur control. Nevertheless, similar results were obtained for the currently available three transcriptome analyses (Ho and Ellermeier, 2015; Hastie et al., 2018). Overall, this response was clearly coordinated directly by Fur, indicated by the multiple potential binding sites (Supplementary Table S7). Some of them were already confirmed by DNA binding studies before (Ho and Ellermeier, 2015).

FIGURE 3

With standard Western diet the iron concentrations in the gut is about 100 mg Fe/g wet weight feces (Pizarro et al., 1987; Lund et al., 1999). However, due to the rising pH in the duodenum and the small intestine solubility of ferric iron decreases and favors the oxidation to ferrous iron in the presence of oxygen. Furthermore, ascorbic acid and citric acid chelate iron and make it available to the microbiome and host. On the other side polyphenols like tannins and catechols from tea or coffee as well phytate from cereals tightly bind iron. Consequently, the amount of iron in the colon lumen that is readily available to bacteria is difficult to estimate (Kortman et al., 2014). The large amounts of different siderophores found in the feces indicate strongly limited access to ferric iron for the gut microbiome (Kortman et al., 2014). In summary, there is always some iron around in the gut. However, the actual iron concentrations might vary with respect to nutrition. Consequently, high and low affinity iron uptake systems are advised. Most likely the differences observed between the proteome and transcriptome data regarding induction of iron-uptake systems of this study as well as the difference to the two transcriptome analyses performed before might be caused by a time-resolved response to iron limitation.

Certain systems found still enhanced at both the RNA and protein level (yclP, ssuA2, ssuB2, CDIF630erm_01231), while other were already formed and the increased abundance became only visible at the protein level (feoA, CDIF630erm_01642; feoB1 CDIF630erm_01643). A co-regulation of the sulfur (ssu operon) and iron metabolism becomes obvious and was observed before (Dubois et al., 2016). This might be explained by the often sulfur-mediated iron coordination in enzymes of C. difficile (see Supplementary Table S8). One of the strongest induced operons at no/low iron conditions in all three transcriptome analyses (Ho and Ellermeier, 2015; Hastie et al., 2018) was the one encoding the catecholate siderophore import system YclNOPQ (CDIF630erm_001824 – 01827). This represents a high affinity iron import system induced at low iron conditions and in the fur mutant which allows uptake of iron at low bioavailability.

The precursor of many catechol siderophores is spermidine (Datta and Chakrabartty, 2014). Interestingly, the spermidine biosynthesis genes speAHEB (CDIF630erm_01008 – 01022) and spermidine/putrescine transporter genes potABCD (CDIF630erm_01160 – 01163) were significantly induced on the transcriptome (spe, Hastie et al., 2018 and pot) and the proteome (only spe) level (Table 1 and Figure 3). In agreement, significantly increased levels of spermidine, spermine and putrescine were detected in the metabolome of iron limited C. difficile cells (Supplementary Table S6). Already in the nineties an increase in polyamines in bacterial cell grown under iron-limited conditions was studied (Bergeron and Weimar, 1991). A similar close relationship between intracellular iron and polyamine content was described for cancer cells (Bae et al., 2018; Lane et al., 2018). Interestingly, siderophores like petrobactin (Lee et al., 2007), alcaligin (Challis, 2005) are formed from polyamines. Others like vibriobactin and vulnibactin contain polyamine backbones (Shah and Swiatlo, 2008; Bergeron et al., 2011). However, the protective function of polyamines during stress situation and their importance for the infection process of many bacteria have been widely described (Shah and Swiatlo, 2008).

The fur Mutant Lost Most of Its Flagella and Motility

Comparative inspection of the wild type and fur mutant C. difficile strains using scanning electron microscopy revealed obvious differences with regard to the presence of flagella. Scanning electron microscopy revealed a significant loss of flagella in the fur mutant compared to the wild type C. difficile (Figure 4B). Negative staining (Figure 4C) also depicted less flagellation of the fur mutant and no detectable other appendage-like structures on the surface like pili or fimbriae. Motility assay revealed in agreement with the electron microscopy analyses, that the fur mutant was highly impaired in motility (Figure 4A). Interestingly, the two flagella operons were also subject to Fur-mediated gene regulation, one (CDIF630erm_00375 – 00395) was found Fur-repressed, while the other (CDIF630erm_00348 - 00361) was identified as Fur-induced. Interestingly, latter operon contained a Fur box upstream of CDIF630erm_00348. The missing Fur-dependent induction of this operon might have caused the observed phenotype. The proteome data partially confirmed this assumption. Most likely, additional unknown factors are required. Remarkably, the gene for the pleiotropic regulator SinR (CDIF630erm_02447) was found overexpressed under iron limiting conditions. One function of the SinR regulator in C. difficile is the induction of flagella formation and motility via the control of c-di-GMP levels (Girinathan et al., 2018). Similarly, proline iminopeptidase (CDIF630erm_02215), catalyzing the removal of N-terminal proline residues from peptides, was described to be involved in Xanthomonas campestris motility via influencing c-di-GMP levels (Khan et al., 2018). The corresponding plp gene and a TetR family transcriptional regulator were found induced under low iron conditions. Pili gene transcription (CDIF630erm_03817 – 03826) was found reduced in the fur mutant (Table 1). This is in agreement with the electron microscopy inspection of the fur mutant (Figure 4C). Due to their extracellular location only one pilus protein was detected by the proteomics approach, but as expected solely in the wild type strain (OFF). Due to a missing potential Fur box upstream the pil operon the observed regulation might be of indirect nature.

FIGURE 4

Low Iron Conditions Induce Major Re-Arrangements of the Energy Metabolism Partially Regulated by Fur

The basic principles of energy generation of Clostridia differs significantly from those of eukaryotes or other prokaryotes. Some of these bacteria mainly generate their energy in form of amino acid fermentation using two coupled reactions previously called Stickland reaction (Stickland, 1934). ATP is formed at the substrate level and using a proton/sodium gradient at the membrane. In principle, during the first oxidative part of the reaction, the first amino acid gets deaminated to form an a-keto acid with the concurrent transfer of the electrons to a carrier like NAD⁺. Subsequently, the decarboxylation of the a-keto acid is linked with the formation of a coenzyme A ester, which in turn is converted into an acyl-phosphate. The final transfer of the activated phosphate residue to ADP yields ATP. In the reductive part of the pathway the second amino acid gets reduced by the formed electrons and deaminated. Sometimes this process is again coupled to ATP generation (Durre, 2014). During the analysis of butyrate formation in C. difficile the enzyme butyryl-CoA dehydrogenase (Bcd) was identified as an electron bifurcating stable complex with the flavoproteins EtfA and EtfB (Aboulnaga et al., 2013). The complex oxidizes NADH and transfers two electrons to the first flavin (β-flavin), which bifurcates one electron to ferredoxin and one electron to a second flavin (α-flavin). After two such rounds the completely reduced a-flavin transfers two electrons further to the third flavin (s-flavin) of the complex, which finally reduces crotonyl-CoA to butyryl-CoA (Chowdhury et al., 2014; Demmer et al., 2017). Most importantly, formed reduced ferredoxins are the substrate of the membrane spanning ferredoxin-NAD⁺ reductase complex (Rnf) which couples the electron transfer from the ferredoxin to NAD⁺ with the generation of a proton or sodium gradient at the membrane (Biegel and Muller, 2010; Tremblay et al., 2012; Mock et al., 2015; Chowdhury et al., 2016). The generated sodium gradient drives ATP generation via a sodium-dependent ATPase (Buckel and Thauer, 2013; Buckel and Thauer, 2018).

Clostridioides difficile possesses three different EtfAB systems. The first is encoded downstream the bcd2 gene encoding butyryl-CoA dehydrogenase (CDIF630erm_01194 – 01199), the second (CDIF630erm_01319 – 01320) in an operon with lactate racemase (LarA, CDIF630erm_01318) and a lactate dehydrogenase (CDIF630erm_01321) and the third downstream of acdB encoding a short chain acyl-CoA dehydrogenase involved in the conversion of 2-enoyl-3-phenylpropionyl-CoA/isocaprenoyl-CoA into 3-phenylpropionyl-CoA/isocaproyl-CoA during phenylalanine/leucine fermentation with formation of 3-phenylpropionate/isocaproate (Elsden and Hilton, 1978; Britz and Wilkinson, 1982; Kim et al., 2006). The selenoprotein D-proline reductase (PrdABCDE) catalyzes the reductive ring cleavage of D-proline to form 5-aminovalerate. As typical Stickland reaction it is coupled to the oxidation of other amino acids, but also formate can serve as electron donor (Kabisch et al., 1999). First proline racemase (PrdF) converts L-proline into D-proline (Wu and Hurdle, 2014). Already in the eighties it was shown that proline reduction is coupled to proton motive force generation (Lovitt et al., 1986). However, the enzyme complex does not reduce ferredoxin and was proposed to directly interact with the membrane-localized, proton/sodium pumping Rnf complex (Jackson et al., 2006). Glycine reductase (GrdABCDE) catalyzes the reductive deamination of glycine to form acetylphosphate and ammonia with the oxidation of thioredoxin (TrxA2, TrxB3) (Andreesen, 2004). The influence of iron on the fermentative metabolism of Clostridium acetobutylicum was already described in the eighties (Bahl et al., 1986).

At low iron conditions the ferredoxin-dependent pathways of phenylalanine, leucine (hadAIBCB, etfBA, CDIF630erm_00523 – 00529), glycine degradation (grdDCBAE,trxBA (CDIF630erm_02587 – 02597) and butyrate fermentation (CDIF630erm_01194 – 01199) with the formation of 3-phenylpropionate, isocaproate, butyrate, 5-methylcaproate, valerate and acetate (CDIF630erm_01194 – 01199, CDIF630erm_02577 – 02583) were significantly reduced, while the proline reductase (prdFEDBARC, CDIF630erm_03533 – 03544) and Rnf complex encoding operons (rnfCDGEAB, CDIF630erm_01284 – 01289) were induced in C. difficile (Table 1). Correspondingly, the highest increase in metabolite concentration was observed for 5-aminovalerate, the product of up-regulated proline utilization, when the lower biomass responsible for its production is taken into account. In agreement, the substrates of the reduced pathways phenylalanine, leucine, glycine and some initial intermediates (phenylpyruvate, 2-oxo-isocaproate, 3-hydroxybutyryl-CoA, crotonyl-CoA) were found accumulated. At the same time some end product like isocaproate and butyrate were found reduced (Figure 3).

Moreover, the synthesis of the flavodoxin (FldX, CDIF630erm_02217) and of enzymes of riboflavin biosynthesis (ribHBAED, CDIF630erm_01882 – 01885) was found significantly enhanced. Interestingly, flavodoxins can replace ferredoxins as electron donors for the proton/sodium ion pumping ferredoxin-NAD⁺ reductase (Rnf) (Chowdhury et al., 2016). Consequently, one explanation for the increased formation of flavodoxins is the replacement of iron-containing ferredoxin as electron donors at the Rnf-complex. The EtfAB (CDIF630erm_01319 - 01321) containing system with nickel-dependent lactate racemase (CDIF630erm_01318) was found induced under low iron conditions (Weghoff et al., 2015). Possibly, the yet unknown function contributes to the overall change or the system uses flavodoxin as natural electron acceptors. Interestingly, a change in ATPase also accompanied the switch from high to low iron conditions. Under low iron conditions, where the directly Rnf complex-coupled proline reductase was found enhanced, an F₀F₁-type, sodium-dependent ATP forming ATPase (atpCDGAHFEB, CDIF630erm_03778 – 03785) was found induced. Under high iron conditions Fur-induces the formation of a V-type, mostly proton-pumping, ATP-consuming ATPase (atpDBAFCEKI, CDIF630erm_03237 – 03245) was preferentially produced. Promoter sequences upstream of the latter genes contained a potential Fur biding site. Clearly, under low iron conditions C. difficile significantly reduced the formation of most iron-requiring, ferredoxin-dependent processes including phenylalanine/leucine utilization via AcdB/EtfA1B1 and butyrate/caproate/valerate formation via Bcd2, EtfA3B3 (Table 1 and Figure 3). Only lactate formation via Ldh/EtfA4B4 was found. Finally, the transcripts for an F₀F₁-type, sodium-dependent ATP forming ATPase (atpCDGAHFEB, CDIF630erm_03778 – 03785) were more abundant, while the formation of a V-type, mostly proton-pumping, ATP-consuming ATPase (atpDBAFCEKI, CDIF630erm_03237 – 03245) was reduced. Obviously, the ferredoxin-independent process of Rnf-complex coupled proline utilization was found enhanced and with it the formation of an ATP forming proton/sodium–driven ATPase. The oligopeptide transporter OppBCAD (CDIF630erm_0972 - 0975) was found reduced.

Looking at the identified potential Fur binding sites, most of the observed changes are not directly regulated by Fur. Two open readings frames upstream of the had operon and the hydroxybutyrate metabolizing enzymes encoding operon possessed potential Fur binding sites (Table 1). Additionally, the flavodoxin gene fldX contained a Fur-box upstream its coding region. Perhaps, known regulators including PrdR, CodY, CcpA or Rex are involved in the detection of the drastic physiological changes accompanying the outlined adaptation process (Bouillaut et al., 2015). Currently, the relationship between membrane protein function and lipid composition becomes true. In this context, the major operon of fatty acid biosynthesis (fapR, plsX, fabHKDG, acpP, fabF, CDIF630erm_01326 – 01333) was up-regulated at the transcriptional level during iron limiting conditions (Table 1). Obviously, a re-structuring of the membrane is required for the overall adaptation of multiple membrane associated metabolic processes.

Iron Requiring Metabolic Processes of the Central Metabolism and of CO Oxidation Are Significantly Downregulated at Low Iron Conditions

Clostridioides difficile is utilizing pyruvate via the radical enzyme pyruvate formate-lyase, which forms in the presence of coenzyme A acetyl-CoA and formate (Figure 3). Formate gets subsequently oxidized by the formate dehydrogenase H to CO₂ and protons. The [NiFe] Hydrogenase Hyd reduces protons to molecular hydrogen (Shafaat et al., 2013; Pinske and Sawers, 2016). Pyruvate formate-lyase (PflD) requires an [4Fe-4S] cluster containing activating enzyme (PflC, PflD1) for the formation of the catalytic glycyl radical (Crain and Broderick, 2014). Formate lyase H (FdhF) is described as a MoCo-containing selenoprotein with a single [4Fe-4S] cluster (Pinske and Sawers, 2016). FdhD is a sulfurtransferase which transfers the sulfur residing on the desulfurase IscS to FdhF (Thome et al., 2012). The [NiFe] Hydrogenase (HydN1AN2) contains 3 different iron-sulfur clusters and heme (Shafaat et al., 2013; Pinske and Sawers, 2016). The whole array of Fe-containing enzymes and their activators were found strictly down-regulated under low iron conditions (Supplementary Table S8). The utilization of glucose via pyruvate-formate lyase (PflD, CDIF630erm_03582 – 03583), with formate dehydrogenase and a hydrogenase (Hyd, Fdh, CDIF630erm_03614 – 03619) was downregulated mainly at the transcript level at low iron condition (Table 1 and Figure 3). Consequently, the whole flux toward formate and hydrogen was significantly blocked at low iron conditions, glucose and pyruvate accumulated (Supplementary Table S5). The overall flux through the glycolysis seemed to be reduced since also glucose accumulated 2.17-fold (Figure 3). In agreement the synthesis of the enzyme for glycogen formation from glucose (GlyCDAP) was also diminished by 3.5-fold. Again, an iron-requiring metabolic pathway was shut down at low iron conditions. Fur-dependent regulation might be mediated via a potential Fur-box found upstream the hydrogenase gene hydN2 and the ATPase gene atpA. CO dehydrogenase (CooSC, CDIF630erm_00832 – 00833) formation was found reduced at the transcriptional and proteomic level. Carbon monoxide dehydrogenase CooSF contains 5 [Fe-S] cluster and catalyzes the oxidation of CO using water with the formation of CO₂ and hydrogen (Dobbek et al., 2001). Like the hydrogen utilizing hydrogenase, carbon monoxide dehydrogenase was down-regulated at the transcriptional and proteomic level.

Cell Wall Restructuring and the Protection Against Antibiotics and CAMPs

Obviously, the low iron stress was counteracted via increased resistance to external attacks. Firstly, the transcription of the dtl operon (dtlCBAD, CDIF630erm_03118 - 03122) involved in the resistance to the collection antimicrobial peptides (CAMP) was enhanced during low iron conditions (McBride and Sonenshein, 2011). The enzymatic system encoded by the corresponding genes is responsible for the D-alanylation of lipoteichoic acids. The D-alanine-poly(phosphoriboto) ligase DltA ligates D-alanine to the carrier protein DltC. Aided by DltB, DltC transferres the D-alanine further to undecaprenyl phosphate and transverses to the membranes. Finally, the D-alanine transferases DltD is involved in the final release of the lipoteichoic acids outside the cell. Similarly, the so called vancomycin resistance gene cluster (vanGYTG, CDIF630erm_01803 – 01805) was found induced at the transcriptional level. The encoded proteins VanG (D-Ala:D-Ser ligase), VanXY (D,D-depeptidase), and VanT (D-Ser racemase) acting on the peptidoglycan, were found all functional in C. difficile before, however, confer only low resistance to vancomycin (Ammam et al., 2013; Peltier et al., 2013).

We challenged the wild type and the fur mutant with below MIC₅₀ amounts of vancomycin and the CAMP polymyxin B as determined before (McBride and Sonenshein, 2011; Ammam et al., 2013). In the presence of 0.3 mg vancomycin/l a delayed growth of the wild type and the fur mutant with an visibly increased sensitivity of the fur mutant strain especially after 18 h to vancomycin treatment was observed. The treatment of both strains with 150 mg/l polymyxin resulted in normal growth of the wild type and significantly inhibited growth of the fur mutant (Figure 5).

FIGURE 5

Additionally, two potential ABC transporter systems of the bacitracin/multidrug family (CDIF630erm_00443 – 00445,CDIF630erm_00938 – 00943) and one multi antimicrobial extrusion protein with a downstream MarR family transcriptional regulator gene (CDIF630erm_03501 – 03502) were also found approximately 2- to 3-fold induced at the transcriptional level (Table 1). Multiple genes encoding enzyme of cell wall biosynthesis and modification (murG, murD, mraY, murF, CDIF630erm_02905 – 02909, manC, pgm, mviN, glmU, prs) were two–fourfold transcriptionally up-regulated under low iron conditions. A mutated murG gene was selected to mediate in decreased susceptibility to vancomycin (Leeds et al., 2014). Deletion of the manC gene in Corynebacterium glutamicum resulted in a slow growing mutant, showing the essential role of the targeted pathway (Mishra et al., 2012). Interestingly, disruption of GDP-mannose synthesis in Streptomyces coelicolor resulted in an increased susceptibility to antibiotics of the bacterium (Howlett et al., 2018). The last gene of the operon encoding the transmembrane virulence factor MviN was shown to be essential in C. difficile (Chu et al., 2016). Antisense RNA mediated down-regulation of mviN resulted in a morphology defects, retarded growth and decreased PSII (integral part of the cell wall anchored glycopolymers) formation and surface deposition (Chu et al., 2016). The bifunctional N-acetyltransferase/uridylyltransferase GlmU (CDIF630erm_03829) catalyzes the transfer of an acetyl from acetyl-coenzyme A to glucosamine 1-phosphate to form N-acetylglucosamine 1-phosphate during cell wall biosynthesis. The protein is necessary for the infection of various pathogenic bacteria, including Mycobacterium tuberculosis, Yersinia pestis, Haemophilus influenzae and Xanthomonas oryzae (Buurman et al., 2011; Min et al., 2012; Patin et al., 2015), it serves as target for the antimicrobial treatment of Mycobacteria (Sharma and Khan, 2017). The glmU gene obviously forms an operon with the prs gene (CDIF630erm_03828) encoding ribose-phosphate pyrophosphokinase that catalyzes the conversion of ribose-5-phosphate into phosphoribosyl pyrophosphate during nucleotide biosynthesis. The prs gene was one of the major up-regulated genes of Bacillus thuringiensis in response to erythromycin treatment (Zhou et al., 2018). Both genes were found up-regulated under low iron conditions. The phosphotransferase uptake system for mannose/fructose/sorbose (CDIF630erm_00408 – 00413) was also found enhanced two–threefold at the transcriptional level. Mannose-derived and guanosine-activated compounds are important constituents of the Gram-positive cell wall. In contrast, the genes for the enzymes N-acetylglucosamine-6-phosphate deacetylase (NagA) and the N-acetylglucosamine-6-phosphate deaminase (NagB) (CDIF630erm_01146 – 01147) involved in cell wall degradation and restructuring were found fourfold down-regulated. The enzyme catalyzes the conversion of N-acetyl-D-glucosamine-6-phosphate via D-glucosamine-6-phosphate to D-fructose-6-phosphate during cell recycling. Interestingly, the promoter of the glmU gene is the only gene regulatory element with a potential Fur binding site of almost all iron regulated genes of cell wall metabolism. Obviously, the cell wall is restructured to protect the bacterium against various external challenges. At the same time cell wall degradation and recycling is stopped. Interestingly, the genes for extracytoplasmic function (ECF) sigma factor s^V (csfV) and the corresponding anti ECF sigma factor RsiV were found enhanced at the transcriptional level. The sigma factor s^V regulates peptidoglycan deacetylation and lysozyme resistance (Ho and Ellermeier, 2011; Ho et al., 2014). An iron-regulated gene encoding a peptidyl-prolyl isomerase is encoded by the gene upstream of both genes. The corresponding promoter carried a potential Fur binding site.

Nucleotide Biosynthesis, CRISPR/Cas System and Prophage Cluster Regulation

Dihydroorotate dehydrogenase (PyrDK), aspartate carbamoyltransferase (PyrB), and orotate phosphoribosyltransferase (PyrE), all enzyme of pyrimidine biosynthesis (pyrBKDE, CDIF630erm_00305 – 00308) were found induced at the transcript level under low iron conditions. Interestingly, in other bacteria dihydroorotate dehydrogenase (PyrDK) channels abstracted electrons directly into electron transfer chains and contributes to proton/sodium gradient formation (Reis et al., 2017). Furthermore, the transcripts from an operon (purECFGNHDL, CDIF630erm_00340 – 00347) involved in purine biosynthesis were found more abundant (Table 1). The purine GTP serves as precursor of riboflavin biosynthesis, which also was found enhanced under low iron conditions. Both operons revealed Fur binding site containing promoters. The bacterial immunity system against phage infections CRISPR/Cas (CDIF630erm_03259 – 03266) was found approximately twofold down-regulated at the transcriptional and proteomic level (Hargreaves et al., 2014). The prophage encoded by CDIF630erm_01522 – 01532 was also found down-regulated. The corresponding promoter of the operon possessed a conserved Fur binding site.

Discussion

The highly specialized energy metabolism of C. difficile mainly relies on multiple ferredoxin-mediated amino acid utilizing reactions, and on pathways harboring various iron-sulfur cluster containing enzymes (see Figure 3 and Supplementary Table S8). Overall, it is highly iron-dependent. In an anaerobic organism, this usually represents a feasible and effective strategy. There are two major drawbacks of this highly specialized lifestyle. (1) Oxygen is inactivating many of the employed processes. (2) Iron is essential for this type of energy metabolism. We investigated here the critical scenario of low iron conditions. It was no surprise that an initial stress response for the acquisition of iron (iron transporter on!) was observed. Maybe, the production of polyamines has something to do with iron storage and acquisition. But in parallel, a major rebuilding of the central energy metabolism occurred. All ferredoxin-dependent amino acid (Phe, Leu, Gly) utilizing processes were drastically reduced. Flavodoxin as an alternative was brought into the game. Similarly, glucose utilization via pyruvate-formate-lyase, formate dehydrogenase, and hydrogenase, all multi-Fe-S-enzymes, was reduced. Instead, proline utilization directly coupled to the sodium ion/proton pumping RNF complex was strongly enhanced. Thus, the switch from more substrate phosphorylation dominated energy generation to membrane potential based processes obviously required the utilization of a different, membrane potential-dependent ATP-forming ATPase. Most likely, even the membrane composition was adjusted appropriately. Finally, the energy consuming process of motility via flagella movement was reduced. However, the transition period for the adaptation to low iron conditions represents a period of metabolic weakness and physical vulnerability. Here, C. difficile “protects the gates,” changing drastically the composition of the cell wall. Protection against antibiotics from other microorganisms of the microbiome, against CAMPs or molecules of the immune system of the host are the major task. And what has Fur to do with all of it? It is the major player, directly and indirectly. Proposed Fur binding sites identified central adaptation processes as directly Fur-controlled. Nevertheless, especially in the complex adaptation of the energy metabolism several indirect regulatory scenarios can be assumed.

In the closely related C. acetobutylicum the strong induction flavodoxins and riboflavin biosynthesis under iron limited conditions was also observed besides the expected increase of iron uptake systems (Vasileva et al., 2012). Additionally, a few metabolic enzymes involved in energy generation were found iron controlled, however, not to the degree observed in this study for C. difficile. The major difference of C. difficile to many other pathogenic bacteria is their aerobic/facultative anaerobic life style. Under these condition iron uptake and storage is connected to ROS formation. Consequently, these bacteria use Fur for the control of superoxide dismutase, catalase, or hydroperoxidase formation (Troxell and Hassan, 2013). Nevertheless, a strict co-regulation of the TCA cycle during virulence by Fur was observed for Staphylococcus epidermidis and Vibrio cholera (Troxell and Hassan, 2013). Finally, multiple other bacteria employ completely different systems (Irr, RirA, and IscR) for their iron response (Rudolph et al., 2006; Santos et al., 2015; Mandin et al., 2016). Consequently, the observed adaptation of C. difficile to low iron conditions partly mediated by Fur is the result of its unique life style and metabolism.

References

• 1

  AboulnagaE.-H.PinkenburgO.SchiffelsJ.El-RefaiA.BuckelW.SelmerT. (2013). Effect of an oxygen-tolerant bifurcating butyryl coenzyme A dehydrogenase/electron-transferring flavoprotein complex from Clostridium difficile on butyrate production in Escherichia coli.J. Bacteriol.1953704–3713. 10.1128/JB.00321-13

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2

  AmmamF.Meziane-CherifD.Mengin-LecreulxD.BlanotD.PatinD.BonecaI. G.et al (2013). The functional vanGCD cluster of Clostridium difficile does not confer vancomycin resistance.Mol. Microbiol.89612–625. 10.1111/mmi.12299

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3

  AndreesenJ. R. (2004). Glycine reductase mechanism.Curr. Opin. Chem. Biol.8454–461. 10.1016/j.cbpa.2004.08.002

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 4

  Argiris Symeonidis and Markos Marangos (2012). “Iron and microbial growth,” inInsight and Control of Infectious Disease in Global Scenario, ed.RoyP. K. (Rijeka: In Tech), 289–332.

  • Google Scholar

• 5

  AwadM. M.CheungJ. K.TanJ. E.McEwanA. G.LyrasD.RoodJ. I. (2016). Functional analysis of an feoB mutant in Clostridium perfringens strain 13.Anaerobe4110–17. 10.1016/j.anaerobe.2016.05.005

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6

  BaeD. H.LaneD. J. R.JanssonP. J.RichardsonD. R. (2018). The old and new biochemistry of polyamines.Biochim. Biophys. Acta18622053–2068. 10.1016/j.bbagen.2018.06.004

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 7

  BahlH.GottwaldM.KuhnA.RaleV.AnderschW.GottschalkG. (1986). Nutritional factors affecting the ratio of solvents produced by Clostridium acetobutylicum.Appl. Environ. Microbiol.52169–172.

  • Pubmed Abstract
  • Google Scholar

• 8

  BaileyT. L.BodenM.BuskeF. A.FrithM.GrantC. E.ClementiL.et al (2009). MEME SUITE: tools for motif discovery and searching.Nucleic Acids Res.37W202–W208. 10.1093/nar/gkp335

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9

  BartlettJ. G.GerdingD. N. (2008). Clinical recognition and diagnosis of Clostridium difficile infection.Clin. Infect. Dis.46(Suppl. 1), S12–S18. 10.1086/521863

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 10

  BenjaminiY.HochbergY. (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing.J. R. Stat. Soc.57289–300. 10.1111/j.2517-6161.1995.tb02031.x

  • CrossRef
  • Google Scholar

• 11

  BentonH. P.WongD. M.TraugerS. A.SiuzdakG. (2008). XCMS2: processing tandem mass spectrometry data for metabolite identification and structural characterization.Anal. Chem.806382–6389. 10.1021/ac800795f

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12

  BergeronR. J.SinghS.BhartiN. (2011). Synthesis of heterobactins A and B and nocardia heterobactin.Tetrahedron673163–3169. 10.1016/j.tet.2011.03.003

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13

  BergeronR. J.WeimarW. R. (1991). Increase in spermine content coordinated with siderophore production in Paracoccus denitrificans.J. Bacteriol.1732238–2243. 10.1128/jb.173.7.2238-2243.1991

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 14

  BiegelE.MullerV. (2010). Bacterial Na⁺-translocating ferredoxin:NAD⁺ oxidoreductase.Proc. Natl. Acad. Sci. U.S.A.10718138–18142. 10.1073/pnas.1010318107

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15

  BouillautL.DuboisT.SonensheinA. L.DupuyB. (2015). Integration of metabolism and virulence in Clostridium difficile.Res. Microbiol.166375–383. 10.1016/j.resmic.2014.10.002

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16

  BraunV.HantkeK. (2011). Recent insights into iron import by bacteria.Curr. Opin. Chem. Biol.15328–334. 10.1016/j.cbpa.2011.01.005

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17

  BritzM. L.WilkinsonR. G. (1982). Leucine dissimilation to isovaleric and isocaproic acids by cell suspensions of amino acid fermenting anaerobes: the Stickland reaction revisited.Can. J. Microbiol.28291–300. 10.1139/m82-043

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 18

  BuckelW.ThauerR. K. (2013). Energy conservation via electron bifurcating ferredoxin reduction and proton/Na⁺ translocating ferredoxin oxidation.Biochim. Biophys. Acta182794–113. 10.1016/j.bbabio.2012.07.002

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 19

  BuckelW.ThauerR. K. (2018). Flavin-based electron bifurcation, a new mechanism of biological energy coupling.Chem. Rev.1183862–3886. 10.1021/acs.chemrev.7b00707

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20

  BurkeK. E.LamontJ. T. (2014). Clostridium difficile infection: a worldwide disease.Gut Liver81–6. 10.5009/gnl.2014.8.1.1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21

  BuurmanE. T.AndrewsB.GaoN.HuJ.KeatingT. A.LahiriS.et al (2011). In vitro validation of acetyltransferase activity of GlmU as an antibacterial target in Haemophilus influenzae.J. Biol. Chem.28640734–40742. 10.1074/jbc.M111.274068

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 22

  CernatR. C.ScottK. P. (2012). Evaluation of novel assays to assess the influence of different iron sources on the growth of Clostridium difficile.Anaerobe18298–304. 10.1016/j.anaerobe.2012.04.007

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23

  ChallisG. L. (2005). A widely distributed bacterial pathway for siderophore biosynthesis independent of nonribosomal peptide synthetases.Chembiochem6601–611. 10.1002/cbic.200400283

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 24

  ChowdhuryN. P.KlomannK.SeubertA.BuckelW. (2016). Reduction of flavodoxin by electron bifurcation and sodium ion-dependent reoxidation by NAD⁺ catalyzed by ferredoxin-NAD⁺ reductase (Rnf).J. Biol. Chem.29111993–12002. 10.1074/jbc.M116.726299

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 25

  ChowdhuryN. P.MowafyA. M.DemmerJ. K.UpadhyayV.KoelzerS.JayamaniE.et al (2014). Studies on the mechanism of electron bifurcation catalyzed by electron transferring flavoprotein (Etf) and butyryl-CoA dehydrogenase (Bcd) of Acidaminococcus fermentans.J. Biol. Chem.2895145–5157. 10.1074/jbc.M113.521013

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 26

  ChuM.MallozziM. J.RoxasB. P.BertoloL.MonteiroM. A.AgellonA.et al (2016). A Clostridium difficile cell wall glycopolymer locus influences bacterial shape, polysaccharide production and virulence.PLoS Pathog.12:e1005946. 10.1371/journal.ppat.1005946

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 27

  CornelisP.WeiQ.AndrewsS. C.VinckxT. (2011). Iron homeostasis and management of oxidative stress response in bacteria.Metallomics3540–549. 10.1039/c1mt00022e

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 28

  CrainA. V.BroderickJ. B. (2014). Pyruvate formate-lyase and its activation by pyruvate formate-lyase activating enzyme.J. Biol. Chem.2895723–5729. 10.1074/jbc.M113.496877

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 29

  DannheimH.RiedelT.Neumann-SchaalM.BunkB.SchoberI.SproerC.et al (2017a). Manual curation and reannotation of the genomes of Clostridium difficile 630Deltaerm and C. difficile 630.J. Med. Microbiol.66286–293. 10.1099/jmm.0.000427

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 30

  DannheimH.WillS. E.SchomburgD.Neumann-SchaalM. (2017b). Clostridioides difficile 630Deltaerm in silico and in vivo – quantitative growth and extensive polysaccharide secretion.FEBS Open Bio7602–615. 10.1002/2211-5463.12208

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31

  DattaB.ChakrabarttyP. K. (2014). Siderophore biosynthesis genes of Rhizobium sp. isolated from Cicer arietinum L.3 Biotech4391–401. 10.1007/s13205-013-0164-y

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32

  DemmerJ. K.Pal ChowdhuryN.SelmerT.ErmlerU.BuckelW. (2017). The semiquinone swing in the bifurcating electron transferring flavoprotein/butyryl-CoA dehydrogenase complex from Clostridium difficile.Nat. Commun.8:1577. 10.1038/s41467-017-01746-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 33

  DengZ.WangQ.LiuZ.ZhangM.MachadoA. C.ChiuT. P.et al (2015). Mechanistic insights into metal ion activation and operator recognition by the ferric uptake regulator.Nat. Commun.6:7642. 10.1038/ncomms8642

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 34

  DobbekH.SvetlitchnyiV.GremerL.HuberR.MeyerO. (2001). Crystal structure of a carbon monoxide dehydrogenase reveals a [Ni-4Fe-5S] cluster.Science2931281–1285. 10.1126/science.1061500

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 35

  DreslerJ.KrutovaM.FucikovaA.KlimentovaJ.HruzovaV.DuracovaM.et al (2017). Analysis of proteomes released from in vitro cultured eight Clostridium difficile PCR ribotypes revealed specific expression in PCR ribotypes 027 and 176 confirming their genetic relatedness and clinical importance at the proteomic level.Gut Pathog.9:45. 10.1186/s13099-017-0194-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 36

  DuboisT.Dancer-ThibonnierM.MonotM.HamiotA.BouillautL.SoutourinaO.et al (2016). Control of Clostridium difficile physiopathology in response to cysteine availability.Infect. Immun.842389–2405. 10.1128/IAI.00121-16

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 37

  DurreP. (2014). Physiology and sporulation in Clostridium.Microbiol. Spectr.2:TBS–0010.2012. 10.1128/microbiolspec.TBS-0010-2012

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 38

  EdgarR.DomrachevM.LashA. E. (2002). Gene Expression omnibus: NCBI gene expression and hybridization array data repository.Nucleic Acids Res.30207–210. 10.1093/nar/30.1.207

  • CrossRef
  • Google Scholar

• 39

  ElsdenS. R.HiltonM. G. (1978). Volatile acid production from threonine, valine, leucine and isoleucine by clostridia.Arch. Microbiol.117165–172. 10.1007/BF00402304

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 40

  FillatM. F. (2014). The FUR (ferric uptake regulator) superfamily: diversity and versatility of key transcriptional regulators.Arch. Biochem. Biophys.54641–52. 10.1016/j.abb.2014.01.029

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 41

  GirinathanB. P.OuJ.DupuyB.GovindR. (2018). Pleiotropic roles of Clostridium difficile sin locus.PLoS Pathog.14:e1006940. 10.1371/journal.ppat.1006940

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 42

  HastieJ. L.HannaP. C.CarlsonP. E. (2018). Transcriptional response of Clostridium difficile to low iron conditions.Pathog. Dis.76:fty009. 10.1093/femspd/fty009

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 43

  HargreavesK. R.KropinskiA. M.ClokieM. R. (2014). Bacteriophage behavioral ecology: how phages alter their bacterial host’s habits.Bacteriophage4:e29866. 10.4161/bact.29866

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 44

  HeapJ. T.CartmanS. T.KuehneS. A.CooksleyC.MintonN. P. (2010). ClosTron-targeted mutagenesis.Methods Mol. Biol.646165–182. 10.1007/978-1-60327-365-7_11

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 45

  HeapJ. T.PenningtonO. J.CartmanS. T.CarterG. P.MintonN. P. (2007). The ClosTron: a universal gene knock-out system for the genus Clostridium.J. Microbiol. Methods70452–464. 10.1016/j.mimet.2007.05.021

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 46

  HeapJ. T.PenningtonO. J.CartmanS. T.MintonN. P. (2009). A modular system for Clostridium shuttle plasmids.J. Microbiol. Methods7879–85. 10.1016/j.mimet.2009.05.004

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 47

  HillerK.HangebraukJ.JagerC.SpuraJ.SchreiberK.SchomburgD. (2009). MetaboliteDetector: comprehensive analysis tool for targeted and nontargeted GC/MS based metabolome analysis.Anal. Chem.813429–3439. 10.1021/ac802689c

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 48

  HoT. D.EllermeierC. D. (2011). PrsW is required for colonization, resistance to antimicrobial peptides, and expression of extracytoplasmic function sigma factors in Clostridium difficile.Infect. Immun.793229–3238. 10.1128/IAI.00019-11

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 49

  HoT. D.EllermeierC. D. (2015). Ferric uptake regulator fur control of putative iron acquisition systems in Clostridium difficile.J. Bacteriol.1972930–2940. 10.1128/JB.00098-15

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 50

  HoT. D.WilliamsK. B.ChenY.HelmR. F.PophamD. L.EllermeierC. D. (2014). Clostridium difficile extracytoplasmic function sigma factor sigmaV regulates lysozyme resistance and is necessary for pathogenesis in the hamster model of infection.Infect. Immun.822345–2355. 10.1128/IAI.01483-13

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 51

  HowlettR.AnttonenK.ReadN.SmithM. C. M. (2018). Disruption of the GDP-mannose synthesis pathway in Streptomyces coelicolor results in antibiotic hyper-susceptible phenotypes.Microbiology164614–624. 10.1099/mic.0.000636

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 52

  HuangW.WilksA. (2017). Extracellular heme uptake and the challenge of bacterial cell membranes.Annu. Rev. Biochem.86799–823. 10.1146/annurev-biochem-060815-014214

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 53

  HussainH. A.RobertsA. P.MullanyP. (2005). Generation of an erythromycin-sensitive derivative of Clostridium difficile strain 630 (630Deltaerm) and demonstration that the conjugative transposon Tn916DeltaE enters the genome of this strain at multiple sites.J. Med. Microbiol.54137–141. 10.1099/jmm.0.45790-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 54

  JacksonS.CalosM.MyersA.SelfW. T. (2006). Analysis of proline reduction in the nosocomial pathogen Clostridium difficile.J. Bacteriol.1888487–8495. 10.1128/JB.01370-06

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 55

  KabischU. C.GrantzdorfferA.SchierhornA.RucknagelK. P.AndreesenJ. R.PichA. (1999). Identification of D-proline reductase from Clostridium sticklandii as a selenoenzyme and indications for a catalytically active pyruvoyl group derived from a cysteine residue by cleavage of a proprotein.J. Biol. Chem.2748445–8454. 10.1074/jbc.274.13.8445

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 56

  KatzianerD. S.YanoT.RubinH.ZhuJ. (2014). A high-throughput small-molecule screen to identify a novel chemical inhibitor of Clostridium difficile.Int. J. Antimicrob. Agents4469–73. 10.1016/j.ijantimicag.2014.03.007

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 57

  KhanA.SinghP.SrivastavaA. (2018). Synthesis, nature and utility of universal iron chelator - Siderophore: a review.Microbiol. Res.212–213, 103–111. 10.1016/j.micres.2017.10.012

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 58

  KimJ.DarleyD.SelmerT.BuckelW. (2006). Characterization of (R)-2-hydroxyisocaproate dehydrogenase and a family III coenzyme a transferase involved in reduction of L-leucine to isocaproate by Clostridium difficile.Appl. Environ. Microbiol.726062–6069. 10.1128/AEM.00772-06

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 59

  KortmanG. A.RaffatelluM.SwinkelsD. W.TjalsmaH. (2014). Nutritional iron turned inside out: intestinal stress from a gut microbial perspective.FEMS Microbiol. Rev.381202–1234. 10.1111/1574-6976.12086

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 60

  LaneD. J. R.BaeD. H.SiafakasA. R.Suryo RahmantoY.Al-AkraL.JanssonP. J.et al (2018). Coupling of the polyamine and iron metabolism pathways in the regulation of proliferation: mechanistic links to alterations in key polyamine biosynthetic and catabolic enzymes.Biochim. Biophys. Acta18642793–2813. 10.1016/j.bbadis.2018.05.007

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 61

  LangmeadB.SalzbergS. L. (2012). Fast gapped-read alignment with Bowtie 2.Nat. Methods9357–359. 10.1038/nmeth.1923

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 62

  LassekC.BurghartzM.Chaves-MorenoD.OttoA.HentschkerC.FuchsS.et al (2015). A metaproteomics approach to elucidate host and pathogen protein expression during catheter-associated urinary tract infections (CAUTIs).Mol. Cell. Proteomics14989–1008. 10.1074/mcp.M114.043463

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 63

  LeeJ. Y.JanesB. K.PassalacquaK. D.PflegerB. F.BergmanN. H.LiuH.et al (2007). Biosynthetic analysis of the petrobactin siderophore pathway from Bacillus anthracis.J. Bacteriol.1891698–1710. 10.1128/JB.01526-06

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 64

  LeedsJ. A.SachdevaM.MullinS.BarnesS. W.RuzinA. (2014). In vitro selection, via serial passage, of Clostridium difficile mutants with reduced susceptibility to fidaxomicin or vancomycin.J. Antimicrob. Chemother.6941–44. 10.1093/jac/dkt302

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 65

  LessaF. C.MuY.BambergW. M.BeldavsZ. G.DumyatiG. K.DunnJ. R.et al (2015). Burden of Clostridium difficile infection in the United States.N. Engl. J. Med.372825–834. 10.1056/NEJMoa1408913

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 66

  LiH.HandsakerB.WysokerA.FennellT.RuanJ.HomerN.et al (2009). The sequence alignment/map format and SAMtools.Bioinformatics252078–2079. 10.1093/bioinformatics/btp352

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 67

  LovittR.KellD. B.MorrisG. J. (1986). Proline reduction by Clostridium sporogenes is coupled to vectorial proton ejection.FEMS Microbiol. Lett.36269–273. 10.1111/j.1574-6968.1986.tb01708.x

  • CrossRef
  • Google Scholar

• 68

  LundE. K.WharfS. G.Fairweather-TaitS. J.JohnsonI. T. (1999). Oral ferrous sulfate supplements increase the free radical-generating capacity of feces from healthy volunteers.Am. J. Clin. Nutr.69250–255. 10.1093/ajcn/69.2.250

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 69

  MandinP.ChareyreS.BarrasF. (2016). A regulatory circuit composed of a transcription factor, IscR, and a regulatory RNA, RyhB, controls Fe-S cluster delivery.mBio7:e00966–16. 10.1128/mBio.00966-16

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 70

  MannH. B.WhitneyD. R. (1947). On a test whether one of the two random variables is stochastically larger than the other.Ann. Math. Stat.1850–60. 10.1214/aoms/1177730491

  • CrossRef
  • Google Scholar

• 71

  McBrideS. M.SonensheinA. L. (2011). The dlt operon confers resistance to cationic antimicrobial peptides in Clostridium difficile.Microbiology1571457–1465. 10.1099/mic.0.045997-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 72

  MinJ.LinD.ZhangQ.ZhangJ.YuZ. (2012). Structure-based virtual screening of novel inhibitors of the uridyltransferase activity of Xanthomonas oryzae pv. oryzae GlmU.Eur. J. Med. Chem.53150–158. 10.1016/j.ejmech.2012.03.051

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 73

  MishraA. K.KrumbachK.RittmannD.BattS. M.LeeO. Y.DeS.et al (2012). Deletion of manC in Corynebacterium glutamicum results in a phospho-myo-inositol mannoside- and lipoglycan-deficient mutant.Microbiology1581908–1917. 10.1099/mic.0.057653-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 74

  MockJ.ZhengY.MuellerA. P.LyS.TranL.SegoviaS.et al (2015). Energy conservation associated with ethanol formation from H₂ and CO₂ in Clostridium autoethanogenum involving electron bifurcation.J. Bacteriol.1972965–2980. 10.1128/JB.00399-15

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 75

  MonotM.Boursaux-EudeC.ThibonnierM.VallenetD.MoszerI.MedigueC.et al (2011). Reannotation of the genome sequence of Clostridium difficile strain 630.J. Med. Microbiol.601193–1199. 10.1099/jmm.0.030452-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 76

  Neumann-SchaalM.HofmannJ. D.WillS. E.SchomburgD. (2015). Time-resolved amino acid uptake of Clostridium difficile 630Deltaerm and concomitant fermentation product and toxin formation.BMC Microbiol.15:281. 10.1186/s12866-015-0614-2

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 77

  NussA. M.HerovenA. K.WaldmannB.ReinkensmeierJ.JarekM.BeckstetteM.et al (2015). Transcriptomic profiling of Yersinia pseudotuberculosis reveals reprogramming of the Crp regulon by temperature and uncovers Crp as a master regulator of small RNAs.PLoS Genet.11:e1005087. 10.1371/journal.pgen.1005087

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 78

  PatinD.BaylissM.Mengin-LecreulxD.OystonP.BlanotD. (2015). Purification and biochemical characterisation of GlmU from Yersinia pestis.Arch. Microbiol.197371–378. 10.1007/s00203-014-1065-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 79

  PeltierJ.CourtinP.El MeoucheI.Catel-FerreiraM.Chapot-ChartierM. P.LemeeL.et al (2013). Genomic and expression analysis of the vanG-like gene cluster of Clostridium difficile.Microbiology1591510–1520. 10.1099/mic.0.065060-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 80

  PerutkaJ.WangW.GoerlitzD.LambowitzA. M. (2004). Use of computer-designed group II introns to disrupt Escherichia coli DExH/D-box protein and DNA helicase genes.J. Mol. Biol.336421–439. 10.1016/j.jmb.2003.12.009

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 81

  PeyraudR.KieferP.ChristenP.MassouS.PortaisJ. C.VorholtJ. A. (2009). Demonstration of the ethylmalonyl-CoA pathway by using 13C metabolomics.Proc. Natl. Acad. Sci. U.S.A.1064846–4851. 10.1073/pnas.0810932106

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 82

  PinskeC.SawersR. G. (2016). Anaerobic formate and hydrogen metabolism.Ecosal Plus7. 10.1128/ecosalplus.ESP-0011-2016

  • CrossRef
  • Google Scholar

• 83

  PizarroF.AmarM.StekelA. (1987). Determination of iron in stools as a method to monitor consumption of iron-fortified products in infants.Am. J. Clin. Nutr.45484–487. 10.1093/ajcn/45.2.484

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 84

  PorcheronG.DozoisC. M. (2015). Interplay between iron homeostasis and virulence: Fur and RyhB as major regulators of bacterial pathogenicity.Vet. Microbiol.1792–14. 10.1016/j.vetmic.2015.03.024

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 85

  ReisR. A. G.CalilF. A.FelicianoP. R.PinheiroM. P.NonatoM. C. (2017). The dihydroorotate dehydrogenases: past and present.Arch. Biochem. Biophys.632175–191. 10.1016/j.abb.2017.06.019

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 86

  Rosinski-ChupinI.SoutourinaO.Martin-VerstraeteI. (2014). Riboswitch discovery by combining RNA-seq and genome-wide identification of transcriptional start sites.Methods Enzymol.5493–27. 10.1016/B978-0-12-801122-5.00001-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 87

  RudolphG.HenneckeH.FischerH. M. (2006). Beyond the Fur paradigm: iron-controlled gene expression in rhizobia.FEMS Microbiol. Rev.30631–648. 10.1111/j.1574-6976.2006.00030.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 88

  SantosJ. A.PereiraP. J.Macedo-RibeiroS. (2015). What a difference a cluster makes: the multifaceted roles of IscR in gene regulation and DNA recognition.Biochim. Biophys. Acta18541101–1112. 10.1016/j.bbapap.2015.01.010

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 89

  SchroderI.JohnsonE.de VriesS. (2003). Microbial ferric iron reductases.FEMS Microbiol. Rev.27427–447. 10.1016/S0168-6445(03)00043-3

  • CrossRef
  • Google Scholar

• 90

  ShafaatH. S.RudigerO.OgataH.LubitzW. (2013). [NiFe] hydrogenases: a common active site for hydrogen metabolism under diverse conditions.Biochim. Biophys. Acta1827986–1002. 10.1016/j.bbabio.2013.01.015

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 91

  ShahP.SwiatloE. (2008). A multifaceted role for polyamines in bacterial pathogens.Mol. Microbiol.684–16. 10.1111/j.1365-2958.2008.06126.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 92

  SharmaR.KhanI. A. (2017). Mechanism and potential inhibitors of GlmU: a novel target for antimicrobial drug discovery.Curr. Drug Targets181587–1597. 10.2174/1389450117666160502152011

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 93

  SmithC. A.WantE. J.O’MailleG.AbagyanR.SiuzdakG. (2006). XCMS: processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification.Anal. Chem.78779–787. 10.1021/ac051437y

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 94

  SticklandL. H. (1934). Studies in the metabolism of the strict anaerobes (genus Clostridium): the chemical reactions by which Cl. sporogenes obtains its energy.Biochem. J.281746–1759. 10.1042/bj0281746

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 95

  TautenhahnR.BottcherC.NeumannS. (2008). Highly sensitive feature detection for high resolution LC/MS.BMC Bioinformatics9:504. 10.1186/1471-2105-9-504

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 96

  ThomeR.GustA.TociR.MendelR.BittnerF.MagalonA.et al (2012). A sulfurtransferase is essential for activity of formate dehydrogenases in Escherichia coli.J. Biol. Chem.2874671–4678. 10.1074/jbc.M111.327122

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 97

  TremblayP. L.ZhangT.DarS. A.LeangC.LovleyD. R. (2012). The Rnf complex of Clostridium ljungdahlii is a proton-translocating ferredoxin:NAD⁺ oxidoreductase essential for autotrophic growth.mBio4:e00406–12. 10.1128/mBio.00406-12

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 98

  TroxellB.HassanH. M. (2013). Transcriptional regulation by Ferric Uptake Regulator (Fur) in pathogenic bacteria.Front. Cell. Infect. Microbiol.3:59. 10.3389/fcimb.2013.00059

  • CrossRef
  • Google Scholar

• 99

  VasilevaD.JanssenH.HonickeD.EhrenreichA.BahlH. (2012). Effect of iron limitation and fur gene inactivation on the transcriptional profile of the strict anaerobe Clostridium acetobutylicum.Microbiology1581918–1929. 10.1099/mic.0.056978-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 100

  VizcainoJ. A.CsordasA.Del-ToroN.DianesJ. A.GrissJ.LavidasI.et al (2016). 2016 update of the PRIDE database and its related tools.Nucleic Acids Res.44:11033. 10.1093/nar/gkw880

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 101

  WeghoffM. C.BertschJ.MullerV. (2015). A novel mode of lactate metabolism in strictly anaerobic bacteria.Environ. Microbiol.17670–677. 10.1111/1462-2920.12493

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 102

  WolfJ.StarkH.FafenrotK.AlbersmeierA.PhamT. K.MullerK. B.et al (2016). A systems biology approach reveals major metabolic changes in the thermoacidophilic archaeon Sulfolobus solfataricus in response to the carbon source L-fucose versus D-glucose.Mol. Microbiol.102882–908. 10.1111/mmi.13498

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 103

  WuX.HurdleJ. G. (2014). The Clostridium difficile proline racemase is not essential for early logarithmic growth and infection.Can. J. Microbiol.60251–254. 10.1139/cjm-2013-0903

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 104

  ZechH.TholeS.SchreiberK.KalhoferD.VogetS.BrinkhoffT.et al (2009). Growth phase-dependent global protein and metabolite profiles of Phaeobacter gallaeciensis strain DSM 17395, a member of the marine Roseobacter-clade.Proteomics93677–3697. 10.1002/pmic.200900120

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 105

  ZhangW.MaJ.ZangC.SongY.LiuP. (2011). The fur transcription regulator and fur-regulated genes in Clostridium botulinum A ATCC 3502.J. Biomed. Biotechnol.2011:934756. 10.1155/2011/934756

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 106

  ZhouP.ChenY.LuQ.QinH.OuH.HeB.et al (2018). Cellular metabolism network of Bacillus thuringiensis related to erythromycin stress and degradation.Ecotoxicol. Environ. Saf.160328–341. 10.1016/j.ecoenv.2018.05.048

  • Pubmed Abstract
  • CrossRef
  • Google Scholar