Transcriptomics and methylomics study on the effect of iodine- containing drug FS-1 on Escherichia coli ATCC BAA-196

Background: Recent studies showed promising results on application of iodine-containing nanomicelles, FS-1, against antibiotic resistant pathogens. The effect was studied on Escherichia coli ATCC BAA-196. Materials & methods: RNA sequencing for transcriptomics and the complete genome sequencing by SMRT PacBio RS II technology followed by genome assembly and methylomics study were performed. Results & conclusions: FS-1 treated E. coli showed an increased susceptibility to antibiotics ampicillin and gentamicin. The analysis of differential gene regulation showed that possible targets of iodine-containing particles are cell membrane fatty acids and proteins, particularly cytochromes, that leads to oxidative, osmotic and acidic stresses. Cultivation with FS-1 caused gene expression alterations towards anaerobic respiration, increased anabolism and inhibition of many nutrient uptake systems. Identification of methylated nucleotides showed an altered pattern in the FS-1 treated culture. Possible role of transcriptional and epigenetic modifications in the observed increase in susceptibility to gentamicin and ampicillin were discussed.

at 230 nm wavelength and compared to the values of eluted antibiotics used as standards (Supplementary Figures S1-3).

Radiological study of FS-1 complexing with chromosomal DNA
To carry out the radiological study, isotope 131 I was used for synthesis of FS-1. The documented radioactivity of the isotope 20 MBq/ml was controlled at the beginning of the synthesis. Radioactivity was measured by β-spectrometer Hidex 300 SL (Finland).
Bacteria were incubated in a thermo-shaker overnight on MH medium at 37°C. The culture growth was centrifuged at 5,000 g and the cells were resuspended in saline to achieve

Genome assembly and annotation
The complete genome assembly was performed using SMRT Link 6.0.0 pipeline as published previously [13]. Two genome variants, NC and FS, were obtained by assembly of the joint pools of DNA reads of three sets of DNA samples generated from NC and FS cultures of E. coli . The NC (CP042865) and FS (CP042867) chromosomes were 4,682,561 and 4,682,572 bp, respectively. The genomes also comprise large plasmids of 266,396 and 279,992 bp, for NC (CP042866) and FS (CP042868) variants, respectively. Moreover, FS-1 treated strain contains two smaller plasmids CP042869 and CP042870 of 44,240 bp and 11,153 bp, respectively. The completeness of the final assemblies was evaluated using the benchmarking universal singlecopy orthologous (BUSCO) software [32]. Genome annotation was performed using the RAST Server (http://rast.nmpdr.org/; [33]) and then manually corrected. Involvement of the predicted genes in metabolic pathways was predicted by the Pathway Tools 23.0 [34] and the EcoCyc Database [29]. Genes involved in antibiotic resistance were predicted by RGI 5.1.0 / CARD 3.0.5 Web-portal [35]. Locations of horizontally transferred genomic islands were identified and the general DNA sequence parameters such as GC-content and GC-skew were calculated by the program SeqWord Genome Island Sniffer [36] in 8 kbp sliding windows stepping 2 kbp along the genome sequence. The same program was used for identification of replication origins and termini on the bacterial chromosomes by GC-skew between the leading and lagging strands [37].
The obtained genomes were deposited in NCBI with the accession numbers CP042865-

RNA extraction and sequencing
Isolation of the total RNA from the NC and FS cultures after overnight cultivation (the mid logarithmic growth phase controlled by OD) on the respective media was performed as it was published before [9]. Shortly, the RiboPure Bacteria Kit (Ambion, Lithuania) was used according to the developer's guidelines. The quality and quantity of the resulting RNA were determined using the Qubit 2.0 Fluorometer (Thermo Scientific, USA) and Qubit RNA Assay

Differential gene expression analysis
The differential expression was done using the R-3.4.4 software. Firstly, a reference index was built for each reference genome using the buildindex function available in the Rsubreads package (Bioconducter). For each bacterium, the obtained RNA fragments were aligned to the relevant reference genomes with the use of the "align" function. The aligned BAM files and relevant GFF annotation files were then used as input for the featureCounts function to obtain gene counts. The R packages DESeq2 (Bioconducter) and GenomicFeatures were then used in R studio for the differential expression analyses. The visualization of the volcano plots of differential expression in the studied genomes was performed by an in-house Python script using the output excel files generated by R package DESeq2. Networks of regulated genes were constructed using the Web based tool PheNetic [38] based on the regulation network available from the PheNetic Web site, which was designed for the strain E. coli K12 (NC_000913. 2).
Pairs of homologous genes in the genomes K12 and BAA196 (Supplementary Table S1) were identified using the program GET_HOMOLOGUES [39].

Profiling of epigenetic modifications
Epigenetic modifications of nucleotides were predicted using the standard SMRT Link DNA modification prediction protocol, ds_modification_motif_analysis, as it was explained in the previous publication [9]. DNA reads generated from every repetition of DNA sampling (three NC and three FS samples) were processed separately. The program calculates several statistical parameters such as IPDRatio of exceeding of the base call delay over the expectation and base call quality values to estimate the base modification (BM) scores representing the likelihood of modification of a nucleotide at the given strand and location. BM scores above 21 (p-value ≥ 0.01) were used to select statistically reliable sites of epigenetically modified nucleotides. Only those sites which showed BM scores ≥ 21 in all three DNA sample repeats were selected for further consideration. Average BM scores were calculated.
Base modification motifs were predicted by the program motifMaker. Visualization of profiles of epigenetic modifications was performed by an in-house Python script.

Antibiotic susceptibility trials
This study was performed to evaluate the effect of iodine-containing nano-micelle drug FS-1 on antibiotic susceptibility of the reference multidrug resistant strain, E. coli ATCC BAA-196, which is characterized by resistance to the beta-lactam antibiotic ampicillin and the aminoglycoside antibiotic gentamicin. The experimental culture (FS) was cultivated for 10 passages with the sub-bactericidal concentration of FS-1 (500 g/ml). In parallel, a negative control (NC) culture was cultivated for 10 passages in Mueller-Hinton (MH) liquid medium without FS-1 and antibiotics.
Minimal bactericidal concentrations (MBC) of the antibiotics were determined for FS and NC cultures by serial dilution in 96-well plates [40]. Cultivation with FS-1 reduced MBC for ampicillin from 2 -2.6 mg/ml recorded for the culture NC to 1.6 -2 mg/ml recorded for the culture FS (Fig. 1A). For gentamicin, these numbers were 0.3 -0.35 mg/ml and 0.2 -0.25 mg/ml for NC and FS cultures, respectively (Fig. 1B). This difference in susceptibility to antibiotics between the initial and FS-1 treated cultures was statistically reliable as illustrated by OD records for the plates with ampicillin and gentamicin in Fig 1. Influence of the treatment with FS-1 on intracellular antibiotic permeability It was found that 30 min treatment of E. coli cells with FS-1 and antibiotics amoxicillin, ampicillin and piperacillin increased the intracellular concentration of antibiotics for several orders of magnitude compared to the treatment of the cells solely with antibiotics (Table 1, HPLC graphs for amoxicillin, ampicillin and piperacillin are shown in Supplementary Figures   S1-3). It was hypothesized that iodine molecules released from FS-1 micelles halogenate cell wall and cytoplasmic membrane proteins and/or fatty acids disrupting their barrier functions. The list of regulated genes is given in Supplementary Table S1. The regulated genes grouped by relevant metabolic pathways predicted by the program Pathway Tools 24.0 are shown in Table 2.
Analysis of the regulated genes showed that up-and down-regulation of these genes is strongly associated with three conditions: strict anaerobic growth, low pH and osmotic/oxidative stresses.
PheNetic network of transcriptional regulation of differentially expressed genes is shown in to sudden changes in the environment to allow bacterial cells to adapt to this change [41]. The expression of this sigma-factor and controlled genes was strongly down-regulated, while the genes suppressed by rpoS were up-regulated indicating that the long cultivation with FS-1 required other defense mechanisms rather than the immediate stress response.
Adaptation of E. coli to the permanent presence of FS-1 in the medium involved a significant alteration of the central metabolic pathways (Fig 4B). An unexpected observation was that the metabolic processes were switched to anaerobic respiration dependent on formate and glycerol-3-phosphate as electron donors, and nitrates and fumarate as terminal electron acceptors. The later reactions are controlled by HydN electron carrier protein, Sn-glycerol-3-phosphate anaerobic dehydrogenase complex GlpABC, selenocysteine containing formate dehydrogenase FdhF and formate-hydrogenlyase complex HycBCDEFG, which showed no expression at the negative control condition but were highly expressed on the medium with FS-1. All systems associated with the aerobic lifestyle, such as the aerobic formate dehydrogenase complex FdoGHI, were strongly inhibited either directly by ArcA repressor or by associated regulatory pathways ( Fig 4A). Expression of all genes of the cytochrome bo terminal oxidase complex, cyoA, cyoB, cyoC and cyoE, was strongly suppressed. It may result from damaging of cytochrome molecules by iodine released from FS-1.
An alternative hypothesis may be that the oxidative phosphorylation pathway was restrained to decrease the oxidative stress caused by the halogen. In a previous study on Staphylococcus aureus, the oxidative stress was reported after an exposure of the culture to FS-1 for 5 min [9,42]. Many systems preventing cell damaging by free radicals, which are byproducts of the oxidative phosphorylation, were down-regulated in the culture cultivated with FS-1. These down-regulated genes include tpx thiol peroxidase; osmC peroxiredoxin; glutathione reductase gor; yodD, ychH and ygaM genes. Other systems probably aiding the cells to cope with halogen oxidation were activated including periplasmic disulfide oxidoreductase dsbC and membrane bound superoxide:ubiquinone oxidoreductase cybB.
All genes of the tricarboxylic acid cycle (TCA) were strongly inhibited in the FS-1 treated E. coli except for MaeA/SfcA NAD-dependent malate dehydrogenase that decarboxylates malate to pyruvate. However, in E. coli strains blocked in the fermentative pyruvate utilization, this gene is overexpressed to support cell growth by catalyzing the normally nonphysiological reductive carboxylation of pyruvate to malate [43] for further use in gluconeogenesis. Two enzymes of the gluconeogenic pathway (fructose 1,6-bisphosphatase GlpX and glucose-6-phosphate isomerise Pgi) and two enzymes shared by the gluconeogenesis and glycolysis (glyceraldehyde-3phosphate dehydrogenase GapA and triosephosphate isomerase TpiA) were upregulated.
Anaerobic fermentation is less effective in terms of energy production compared to the oxidative phosphorylation and thus the cell requires an intensification of the glycolytic pathways and TCA cycle [44]. It contradicts to the observed inhibition of TCA enzymes and the activation of gluconeogenesis. E. coli has several alternative glycolytic pathways bypassing glycolysis: Entner Doudoroff (ED), Embden-Meyerhof-Parnas (EMP) and oxidative pentose phosphate (OPP) pathways [45]. The genes of the ED pathway, phosphogluconate dehydratase edd and ketohydroxyglutarate-aldolase eda, were 1.8 and 1.6-fold upregulated. Genes pfkA and pfkB encoding subunits of the EMP pathway enzyme, 6-phosphofructokinase, were not regulated at this condition. Gnd 6-phosphogluconate dehydrogenase starting the OPP pathway was 3-fold upregulated in E. coli at the presence of FS-1. Upregulation was observed also for two other genes of this pathway: ribose-phosphate pyrophosphokinase prsA and 6phosphogluconolactonase ybhE. Activation of the OPP pathway redirects the glycolytic fluxes towards the ED pathway [45] activated by FS-1. It may be concluded that the ED pathway was used by E. coli as the main glycolytic pathway under the effect of FS-1.
All the pathways of acetate catabolism including acetate uptake transporters and the enzymes of the fatty acid -oxidation (aerobic and anaerobic) and the glyoxylate pathway were strongly inhibited in the FS-1 treated culture. It was unexpected as the fatty acid beta-oxidation was reported as a critical pathway during the anaerobic growth of E. coli on several sugars, particularly on xylose [46]. Accumulation of acetate and other organic acids produced by anaerobic fermentation and due to acetate uptake inhibition could lead to an acidic stress indicated by upregulation of several acidic stress response genes: yhjX, ymgD, yfiD and adiABC.
In consistence with this was the observed upregulation of arginine decarboxylase AdiABC and lactate dehydrogenase LdhA, which are activated in anaerobic conditions in response to extremely acidic environments acidified by carbohydrate fermentation products [47,48].
Protein and fatty acid biosynthesis pathways generally were activated in the cells treated with FS-1. Proteins PrmC and YjiA controlling the accuracy of protein translation [49,50] and the elongation factor P (efp) were activated, while the ribosomal inhibitors ygiU/mqsR and rmf [51,52] were strongly downregulated. At the same time, the uptake and transportation of many amino acids, organic acids, sugars and polypeptides were inhibited. Iodine in the FS-1 complex is bound to polypeptide and oligosaccharide micelles [6,8]. Transmembrane polypeptide and amino acid transporters, such as low affinity tryptophan transporter TnaB, oligopeptide transporter OppADF, dipeptide transporter DppA, proline transporter PutP, glutamate/aspartate ABC transporter GltI and glutamine ABC transporter GlnH could serve as entry points for bound iodine atoms and thus were downregulated in the FS-1 treated E. coli.
Activation of heavy metal homeostasis efflux pumps cueO and copA, and potassium uptake ATPase kdpAB involved in osmotic stress response [53,54]  Hyperactivation of transposases and integrases located on the plasmid may explain the fact that two smaller plasmids were found only in the E. coli population grown on the medium with FS-1.
Gene comparison showed that these plasmids originated from the large plasmid by genetic rearrangements (Fig 2).
Despite of the activation of the plasmid born -lactamases, viability of FS-1 treated E. coli decreased and susceptibility to antibiotics increased (Fig 1). It may result from the observed downregulation of many other genes associated with drug resistance: yejG [56], uspF [57] and ygaM [58], and also with the oxidative stress and increased penetrability of the cell membrane.
Strong inhibition of the genes involved in biofilm formation was observed in the FS-1 treated culture that potentially can reduce pathogenicity of the bacterium. Expression of quorum sensing regulators lsrR and csrB [59][60][61] was strongly downregulated. Directly or indirectly, the inhibition of regulatory elements has affected many other genes involved in bacterial motility, adhesion and biofilm formation, which include the AI-2 quorum sensing signal procession protein [52], AriR transcriptional regulator [62], inducer of biofilm formation BolA [63], stress induced protein uspF [57], biofilm-stress-motility lipoprotein BsmA/YjfO [64] and the regulator of biofilm formation through signal secretion protein BssS/YseP [65].
It should be noted that the increased susceptibility of FS-1 treated E. coli to gentamicin and ampicillin (Fig 1)  Frequencies of different methylation motifs in the sequenced genomes are shown in Table 3.  (Table 3). Bipartite adenine methylation in sequence motifs GATC controlled by S-adenosylmethionine depended orphan (not associated with any restriction enzymes) Dam methylase is common in E. coli [66,67]. GATC sites were often associated with a more complex pattern of nucleotide modifications, CRGKGATC, where the leading cytosine residue was also methylated. While this type of cytosine methylation was not the most abundant, these sites of modified cytosines showed the highest BM scores. Notably, while the total numbers of CRGKGATC sites in NC and FS genomes remain similar (respectively 105 and 106 out of 745 available CRGKGATC motifs), they were found at different loci on the chromosomes (Fig 6A-B).
Another type of bipartite methylation at palindromic motifs AAC(N6)GTGC and GCAC(N6)GTT is under control of type-1 EcoKI-like methylase [68]. In the sequenced genomes, the most likely candidate for this methylase is chromosomal gene hsdM within an operon comprising also type I restriction enzyme hsdR and specificity determinant hsdS. This operon is followed by three other restriction-modification genes, uncharacterized gene yjiV and two subunits of 5-methylcytosine-specific restriction enzyme McrBC, which were upregulated in the culture FS (78-fold for mcrB), while the operon hsdRMS was constantly expressed in both cultures. Methylation at motifs AAC(N6)GTGC / GCAC(N6)GTT protects DNA from cleavage by the cognate restriction enzyme. However, approximately 25% of these palindromic motifs were methylated only at one DNA strand (Table 3). Either one strand methylation was sufficient to protect DNA from cleavage, or it could be that the semi-palindromic nature of these motifs excludes its recognition by protein HsdS on both DNA strands, or these sites could be protected from both, methylation and restriction, by some regulatory elements (DNA binding proteins or small RNA molecules). Also, the stringent selection of only those modified sites which were predicted in all three repeats of the experiment potentially could increase the rate of falsenegative predictions.
Other predicted cytosine methylation motifs were CCAGGRAH and WCCCTGGYR.
Methylation at CCAGG Dcm motif is typical for E. coli [66]; however, motifMaker has predicted in NC and FS genomes a similar but more complex methylation motif CCAGGRAH.
Only a fraction of methylated sites CCAGG was associated with 3'-end sequence conservation; however, it may explain why this palindromic sequence was methylated only at one DNA strand.
Only 5% of available CCAGG sites were methylated on the chromosomes and plasmids. While the number was constant, different sites were methylated in the NC and FS genomes. The distribution of methylated CCAGG motifs in the plasmid sequences of two genomes is shown in  (Table 3).
CCAGGRAH and WCCCTGGYR sequences share the same central motif CCWGG.
Nevertheless, they show different patterns of distribution and are likely controlled by different methyltransferases. Methylation of CCWGG motifs is controlled by Dcm cytosine methylase also referred to as Mec methylase [69]. Two alleles of dcm genes associated with EcoRII-like restriction-modification systems were found in the genome of E. coli ATCC BAA-196 with chromosomal and plasmid locations: BAA196NC_1741 and BAA196NC_RS24085, respectively. Both these genes are expressed at NC and FS conditions (plasmid gene showed 4fold transcriptional activation; however, it was not statistically reliable). Chromosomal gene dcm is followed on the chromosome by short patch repair DNA mismatch endonuclease var. It shows that this methylation may be associated with DNA repair mechanisms rather than with DNA cleavage prevention [67]. Contrary, DNA cytosine methyltransferase located on the plasmid is neighbored with a type II restriction endonuclease BAA196NC_RS24090.
Only a small fraction of CCWGG motifs were methylated. It may be explained by the fact that EcoRII restriction enzymes do not cleave DNA at a single recognition site but require binding to an additional target site serving as an allosteric effector [70]. Thus, restriction of CCWGG sites in E. coli may depend on the spatial conformation of the chromosomal and plasmid chromatin, which may be guided by methylation of CCWGG sites.
On the chromosomes NC and FS, the total numbers of adenine residues with BM scores > 21 including sporadic sites not associated with the recognized motifs were 43,361 and 43,760, respectively. Among them, 41,352 sites (95%) were located exactly at the same positions in both genomes. The numbers of modified adenine residues with BM scores > 80 were 39,098 in NC and 39,083 in FS, and 39,045 (99%) were the same loci in both genomes. Only 53 high scored modified adenine residues were found in NC and 38 in FS. NC-specific adenine methylation sites were found within sequences of 16 genes, but only one of them was significantly downregulated -yebV encoding an uncharacterized protein. Seven genes in the genome FS contained genome-specific high scored methylated adenine residues. One of these genes, nitrite reductase nitB, was significantly upregulated.
There were 10,915 methylated cytosine residues in the genome NC and 11,220 in the genome FS with only 6,001 methylated sites (54%) shared by these two genomes indicating that cytosine methylation patterns were more dependent on the growth conditions. Numbers of modified cytosine residues with BM scores > 80 were 201 and 203 in NC and FS genomes, respectively; among which 119 sites (59%) were common for both genomes. NC-specific modified cytosine residues were found in 33 genes. Two of these genes, citrate-ACP transferase citF and DNAbinding transcriptional activator gadW, were significantly upregulated. One CRGKGATC site with NC-specific cytosine methylation was near the promoter region of the NADP-dependent malate dehydrogenase maeB, which was strongly downregulated by the presence of FS-1 in the medium. In contrast to FS-1 activated NAD-dependent malate dehydrogenase MaeA, MaeB activity is associated with acetate metabolism [71] that generally was downregulated in the FS-1 treated E. coli.
FS-specific cytosine methylation was found in 39 genes with three of them upregulated: biodegradative arginine decarboxylase adiA, DNA-directed RNA polymerase rpoC and glucose-6-phosphate isomerase pgi; and two genes strongly downregulated: bifunctional isocitrate dehydrogenase aceK and acyl-CoA dehydrogenase fadE.

Complexation of FS-1 with bacterial DNA
Epigenetic changes in bacterial chromosomes under the effect of the iodine-containing nanomolecular complex FS-1 suggested a possible complexation of DNA with nanoparticles and/or with released iodine, which potentially can halogenate chromosomal nucleotides. Iodine may halogenate DNA nucleotides mostly at thymine residues [72], however, other computational simulations showed that FS-1 micelles may interact with purine residues [8]. The number of thymine residues with BM scores > 21 increased insignificantly from 3,199 sites in NC to 3,306 in FS. To study the ability of FS-1 to reach the chromosome and create complexes with DNA, the drug was synthesized with the radioactive isotope 131 I (20 MBq/ml). After 1 h cultivation with the radioactive labeled FS-1, bacterial cells were washed twice to remove the remaining micelles of FS-1 and the residual radioactivity was measured. Then the residual radioactivity was measured in DNA samples extracted from the washed cells. It was found that radioactivity of the extracted DNA was 43.46 ± 13.895 Bq/ng that constituted 0.46% ± 0.15 of the residual radioactivity of the treated bacterial cells after washing. This insignificant residual radioactivity could be associated with either complexing of FS-1 with the DNA or a direct halogenation of nucleotides by iodine isotopes that potentially can damage DNA.
The analysis of expression of the genes responsible for DNA repair showed that the genes of the SOS-response, recA and lexA, were downregulated. Strong upregulation was detected for the gene yjiA, which is expressed in response to mitomycin C treatment causing DNA damage [73].
Genes iraD/yjiD encoding an inhibitor of the  S activity in response to DNA damage [74] also were upregulated. Genes of the DNA repair RecFOR complex were differentially regulated.
RecF was strongly upregulated while RecO and RecR were downregulated. RecFOR independent double-strand break repair protein YegP was downregulated. It indicates that damaging of the chromosomal DNA of E. coli most likely was not the major target of the drug FS-1 and the DNA of this bacterium was protected from the direct iodination. The activated DNA repair genes may be responsible for the DNA protection from iodine. help but use the glycolysis to feed the activated fermentation pathways as this bacterium, in contrast to E. coli, has no alternative glycolytic pathways such as the Entner Doudoroff pathway that was activated in the FS-1 treated E. coli. This may be another factor making S. aureus more susceptible to FS-1. Another noteworthy difference in gene regulations by FS-1 in S. aureus compared to E. coli was a strong activation in the former organism of multiple chaperons and DNA repair genes that was not the case with E. coli [9]. Contrary, S. aureus genes encoding cytochromes were not as much inhibited as in the FS-1 treated E. coli.
Long-lasting effects on gene regulation may be associated with different patterns of methylation of DNA nucleotides. E. coli in contrast to S. aureus comprised a bigger number of methyltransferases catalyzing the methylation of both adenosine and cytosine residues.
This study showed that the treatment of E. coli with FS-1 altered the pattern of methylation of the chromosomal and plasmid DNA that in the case with S. aureus was observed to a smaller extent [9]. In the strain E. coli BAA-196, the dominant type of DNA modification was adenine methylation at GATC motifs recognized by an orphan Dam methylase. While this methylation usually covers almost 100% available motifs, demethylation of a single nucleotide at one DNA strand may cause a significant impact on gene regulation and bacterial virulence [56,57]. In this work, it was shown for the first time that the abundant in gamma-Proteobacteria GATC Dam methylation was associated in this bacterium with a more complex pattern of tripartite cytosine and adenine methylation at CRGKGATC motifs. A recently published overview of the epigenomic landscape of prokaryotes [67] did not mention any combinatorial cytosine and adenine methylations at common motifs. In contrast to GATC methylation, methylation of cytosine residues at CRGKGATC motifs was fractional and unequally distributed on the NC and FS chromosomes (Fig 6A-B). There are several candidate methyltransferases to perform this methylation. One orphan uncharacterized S-adenosylmethionine-dependent methyltransferase, BAA196NC_RS22825, is located on the pathogenicity plasmid and is significantly expressed in the culture FS. Further study is needed to check if this type of methylation is specific for the plasmid bearing E. coli. Three uncharacterized orphan methyltransferases, yafS, yafE and yfiF, are located on the chromosome and expressed in both cultures, NC and FS. Another methyltransferase, yhdJ, is believed to be responsible for type-II ATGCAT methylation [77].
This gene is expressed in both cultures; however, only one methylation at this motif was found in NC, and five instances of this methylation out of 1,706 available motifs were found in FS culture. Either this methyltransferase is inhibited post-transcriptionally that is common for prokaryotes [67], or this methyltransferase may have another recognition site.
Methylation of the recognition sites protects the host DNA from cleavage by cognate endonucleases; however, other authors reported that alternative patterns of methylation by orphan methylases may be naturally observed in long-term stationary phase and may prevent DNA replication [78,79], participate in gene regulation [80] and even reduce resistance to antibiotics [81].
The global pattern of cytosine methylation at motifs CRGKGATC, CCAGGRAH and WCCCTGGYR shows more variations between the NC and FS chromosomes and plasmids. The role of CCWGG-methylation in gene regulation in both, eukaryotes and prokaryotes, has been reported in many publications [67]. In E. coli, CCWGG-methylation is a regulator of the stationary phase and stress response [69]. However, no phenotypic changes were reported for E. coli cells with under-and over-production of the Dcm methyltransferase suggesting that the number of methyltransferase molecules does not affect the methylation pattern [66]. The Stable but alternative patterns of adenine and cytosine methylation in the NC and FS genomes discovered in this study demonstrates an importance of DNA methylation for adaptation of bacteria to the presence of iodine-containing nano-micelles in the medium and may be associated with the observed increase in susceptibility of the treated E. coli to gentamicin and ampicillin.
However, it should be admitted that the altered pattern of epigenetic modifications cannot prove by itself that these variations play any specific role in the bacterial adaptation to halogen containing environments and/or in antibiotic resistance reversion. These hypotheses must be proved in additional studies focused on individual methylation sites and its role in gene regulation.

Conclusion
Iodine is one of the oldest antibacterial medicines used by humans starting from its discovery by French chemist Bernard Courtois in 1811 [82]. Acquired resistance to iodine, in contrast to antibiotics, has never been reported. Application of iodine-containing nano-molecular complexes, such as FS-1, allows a broader use of iodine against antibiotic resistant pathogens.
This study showed a profound alteration of the gene expression profile in the antibiotic resistant strain E. coli ATCC BAA-196 cultivated with a sub-bactericidal concentration of FS-1 that led to an increase in susceptibility to gentamicin and ampicillin. The analysis of differential gene regulation suggested that possible targets of iodine-containing particles are cell membrane fatty acids and proteins, particularly the cytochrome molecules, that leads to oxidative, osmotic and acidic stresses. Damaging of the cell wall and membrane structures increased penetrability of bacterial cells by toxic compounds and heavy metal ions accompanied with a reduction of the membrane potential due to the loss of intracellular potassium ions. Bacteria responded by an increased activity of anion pumps and the potassium uptake system, and by inhibiting of permeases to reduce the iodine flux. This general inhibition may complicate uptake of nutrients into cells, which already are stressed by damaged cell wall and cellular membrane structures. All these factors lead to an increased susceptibility to antibiotics that persists even after removal of FS-1 from the medium. Inability of a quick restoration of antibiotic resistance suggests an involvement of epigenetic and epistatic mechanisms in adaptation to FS-1. This hypothesis was supported by discovering a steadily altered DNA methylation pattern in the FS-1 treated E. coli.
Another hypothesis that iodine may halogenate directly the bacterial DNA may be true for S. aureus but hardly for E. coli.

Future Perspective
This work was performed as a part of a bigger project on studying the mechanisms of reversion of susceptibility to antibiotics by treatment of multidrug resistant bacteria with iodine-containing nano-micelles. This approach looks promising for combating antibiotic resistant infections. This phenomenon first was observed on multidrug resistant tuberculosis [11] but also was reproduced on S. aureus [9,12] and E. coli (the current work). All these studies followed the same experimental protocol to allow comparison of the results. Further studies aimed at development of new drugs against antibiotic resistant nosocomial infections will be conducted on multiple clinical isolates, which were recently collected in clinics in Kazakhstan [83].

Summary Points
 Treatment of the antibiotic resistant E. coli with iodine-containing nano-micelles increased its susceptibility to gentamicin and ampicillin.
 The profound alteration of the gene expression pattern of the FS-1 treated culture was observed.
 All aerobic metabolic pathways were strongly inhibited in the treated culture.
 Many nutrient uptake transporters were inhibited possibly as iodine entrance points.
 Specific gene regulation showed that the treated culture suffered from osmotic, oxidative and acidic stresses.