Flavodoxins as Novel Therapeutic Targets against Helicobacter pylori and Other Gastric Pathogens

Flavodoxins are small soluble electron transfer proteins widely present in bacteria and absent in vertebrates. Flavodoxins participate in different metabolic pathways and, in some bacteria, they have been shown to be essential proteins representing promising therapeutic targets to fight bacterial infections. Using purified flavodoxin and chemical libraries, leads can be identified that block flavodoxin function and act as bactericidal molecules, as it has been demonstrated for Helicobacter pylori (Hp), the most prevalent human gastric pathogen. Increasing antimicrobial resistance by this bacterium has led current therapies to lose effectiveness, so alternative treatments are urgently required. Here, we summarize, with a focus on flavodoxin, opportunities for pharmacological intervention offered by the potential protein targets described for this bacterium and provide information on other gastrointestinal pathogens and also on bacteria from the gut microbiota that contain flavodoxin. The process of discovery and development of novel antimicrobials specific for Hp flavodoxin that is being carried out in our group is explained, as it can be extrapolated to the discovery of inhibitors specific for other gastric pathogens. The high specificity for Hp of the antimicrobials developed may be of help to reduce damage to the gut microbiota and to slow down the development of resistant Hp mutants.


Metabolism
The shikimic acid pathway uses erythrose-4-phosphate and phosphoenol pyruvate to produce chorismic acid, the precursor of aromatic amino acids, folate cofactors, ubiquinone, and vitamins E and K. This biosynthetic route involves four Hp essential enzymes that are absent in mammals: 3-dehydroquinate dehydratase, shikimate dehydrogenase, shikimate kinase, and chorismate synthase [30][31][32]34]. The biosynthesis of coenzyme A (CoA), an essential bacterial cofactor, is achieved with participation of phosphopantetheine adenylyltransferase (PPAT), the inactivation of which prevents bacterial viability [35,36]. Fumarate reductase is a key enzyme for aerobic and anaerobic respiration which contains three subunits: FrdA, FrdB, and FrdC. Some fumarate reductase inhibitors used to treat helmintic infection have also shown inhibitory and bactericidal properties against Hp [33,38]. Several other Hp enzymes have been associated with bacterial respiration, rendering them critical for bacterial survival. Some of them: cytochrome c-type biogenesis protein CcdA and cytochrome c oxidase subunits CcoN, CcoO, CcoP, and CcoQ have been described as potential drug targets for Hp infection [37]. Enzymes and electron carrier proteins that take part in pyruvate decarboxylation (such as flavodoxin (Fld), pyruvate:flavodoxin oxidoreductase (POR), and flavodoxin:quinone reductase B (FqrB)) have also been identified as essential proteins for Hp survival [26,28,[41][42][43][44][45][46].

Cell Wall Structure
Peptidoglycan, synthesized by Mur enzymes in a multistep pathway, is an essential component of the bacterial cell wall. MurA, acting on the first step, is targeted by Fosfomycin [50]. For glutamate racemase MurI, transforming L-Glu into D-Glu, several inhibitors have been described [49,50]. The widely used β-lactam antibiotics inhibit peptidoglycan cross-linking which is carried out in the last steps of the pathway [17,49,50]. On the other hand, the succinylase pathway is the only route in Hp for the synthesis of lysine, a required element of the bacterial peptidoglycan cell wall. N-succinyl-L,L-diaminopimelic acid desuccinylase (SDAP-deacylase; DapE), an enzyme of this route, has been identified as essential for Hp survival [47,48].

pH Homeostasis
Helicobacter pylori is able to survive the acidic pH in the stomach thanks to, at least, two enzymes playing a fundamental role in acid acclimation. Urease and carbonic anhydrase maintain neutral pH in the bacterial cytoplasm and periplasm by converting urea and carbon dioxide into ammonia and bicarbonate [54,55,67]. Urease, which is absent in humans, is a critical enzyme for Hp colonization of the host stomach. It is composed of α (UreA) and β (UreB) subunits and its activity requires Ni (II) ions [53,54,56]. The putative nickel-responsive regulator (NikR) regulates urease expression and nickel uptake [54]. This metal goes into the cytoplasm, where urease is localized, through the nickel-cobalt transporter NixA [54,59], and its incorporation to the active sites of urease taking place during enzyme maturation depending on UreD, UreE, UreF, UreG, and UreH accessory proteins [54,59,[61][62][63] and on the HypA and HypB hydrogenase/urease maturation factors [61,63]. Conversely, HspA and Hpn proteins are related to nickel homeostasis, storage, and protection from higher concentrations of metal ions [59,61,64]. The access of urea to urease is restricted by an H + -gated pore (UreI) which regulates the urea entry into the cytoplasmic space [66]. The activity of these proteins is needed for urease function and, thus, for Hp colonization, and some of them (specifically, urease and HspA) can be inhibited by bismuth, an antimicrobial currently used in therapy [59]. On the other hand, two different types of carbonic anhydrase, αand β-, have been identified in Hp periplasm and cytoplasm, respectively. Both metalloenzymes have been described to be essential for acid acclimation, biosynthetic reactions, bacterial survival, and colonization of the stomach and duodenum [67,68,[70][71][72]. They are targeted by sulphonamide antimicrobial agents and phenol-derivatives [68,69,71].

Virulence (Adherence, Motility, and Pathogenicity)
Helicobacter pylori has developed structures and mechanisms contributing to bacterial virulence. Among them are adhesins, pili, flagella, and extracellular polymeric matrix materials such as DNA, polysaccharides, proteins, and lipids [96]. Biofilms are seen as virulence factors, and several mucoactive and antibiofilm substances, such as N-acetylcysteine and erdosteine, have been proposed as new adjuvant agents for therapy [96][97][98]. Motility, a crucial virulence factor needed for persistent Hp infection, is provided by flagella which allow bacteria to travel through the mucus layer from the gastric lumen to the epithelial surface, its site of infection. Flagellar filaments consist of the major (FlaA) and the minor (FlaB) flagellin proteins [53,75,76,78]. Additional related genes, fliD and flgK, are required for flagellar filaments assembly and flagella formation [76][77][78]. For flagellar assembly and motility, flagellin needs to be O-glycosylated with pseudaminic acid (Pse). Thus, the Pse biosynthesis pathway and, in particular, the aminotransferase enzyme PseC have potential as targets [73,74]. Severe complications of bacterial infection have been related to Hp adherence to host cells. Secretion of virulence factors, such as those encoded in the cytotoxin-associated gene pathogenicity island cagPAI (CagA: the cytotoxin-associated antigen, and T4SS: the cag-type IV secretion system), the vacuolating cytotoxin VacA, or the blood group antigen-binding adhesin BabA, have been related to increased epithelial damage and predisposition to gastric carcinogenesis [82]. In particular, BabA enables bacterial contact with the stomach mucosa. Then, VacA delivery and T4SS signaling are induced: VacA leads to membrane pore formation, cellular vacuolation, apoptosis, and inhibition of immune cells [85,86], while the T4SS pathway translocates CagA into host epithelial cells, where it modulates aspects of the host metabolism and provokes inflammation, metaplasia, and neoplastic transformations [80][81][82][83]86]. The adhesin BabA has been associated to disease-related strains and cagPAI and VacA have been linked to increased gastric cancer risk [81,86]. On the other hand, the T4SS route is favored by interaction between the adhesin HopQ and the human carcinoembryonic antigen-related cell adhesion molecules (CEACAMs) [84]. In addition to those indicated above, other virulence factors have been identified in Hp which include adhesins, such as HopZ, OipA, SabA, and AlpA/B [76,80,85], and the HtrA serine protease, an essential periplasmic protein with chaperone and proteolytic activities involved in quality control and stress responses [85]. Specifically, HtrA is involved in the cleavage of the tumor suppressor E-cadherin and so in the disruption of intercellular adhesion and access of bacteria to intracellular spaces [85,87,99,100].

Active Efflux of Metal Ions
The levels of cobalt, zinc, cadmium, and iron need to be regulated in Hp as both too low and too high concentrations can be detrimental for bacterial life. Thus, enzymes that control metal levels (such as cation efflux system protein CusA, cobalt/zinc/cadmium efflux system membrane fusion protein, cobalt/zinc/cadmium resistance proteins CzcA, CzcB, and CzcC, CznABC metal efflux pump, ferrix siderophore transport system TonB periplasmic binding protein, ferric siderophore transport system ExbB biopolymer transport protein, and Haemin uptake system ATP binding protein) are crucial for Hp virulence and adaptation to gastric environment and, therefore, for bacterial survival [37,89].

Protection against Oxidative Stress
Toxic reactive species can cause oxidative stress to Hp, leading to cell death. For this reason, bacterial gene products involved in protection against reactive oxygen species, superoxide, and free radicals have been proposed as therapeutic targets. They include glutathionyl spermidine synthetase, iron-binding ferritin-like antioxidant protein, DNA-binding protein Dps, and superoxide dismutase [37]. Homeostatic stress regulator (HsrA) is an orphan response regulator unique among epsilonproteobacteria. It syncs metabolic functions and virulence with availability of nutrients and cell division. This protein regulates its own expression and that of a large number of genes involved in transcription, translation, energy, and nitrogen metabolism as well as redox homeostasis and oxidative stress defense. Due to the fact of its essentiality, its absence in humans and the availability of an X-ray structure, it has been proposed as a promising therapeutic target against Hp [92][93][94][95].
As can be seen, several Hp gene products have been described that may constitute appropriate targets for the development of novel therapies against this bacterial infection. Among them, flavodoxin stands out as a promising candidate due to the fact of its essentiality for Hp, its absence in humans, and because of structural features that will be discussed below. Stimulated by these facts, our laboratory has undergone several studies aimed at understanding the biophysical properties of Hp flavodoxin and to identify and perfect molecules that could bind to it and interfere with its vital function. The following sections include a detailed description of this flavodoxin, focusing on the properties that make it a fine drug target.

An Overview of Flavodoxins and of the Flavodoxin from Hp
Flavodoxins are acidic proteins that contain a flavin cofactor (flavin mononucleotide, FMN) acting as an electron transfer center [27]. They are small (14.5-23 kDa) α/β proteins with five α-helices packing against a central five-stranded β-sheet, thus forming an αβα sandwich [27,101]. In some bacteria (e.g., Escherichia coli, Azotobacter vinelandii, or Desulfovibrio vulgaris), flavodoxins are constitutive proteins, while in others, such as in several Anabaena strains, flavodoxin synthesis is induced in low iron conditions, where it replaces ferredoxin [102][103][104][105][106][107], a constitutive sulfoferric protein that transfers electrons one by one. Although the FMN in flavodoxin only participates in two-electron transfer reactions when it is free in solution, apoflavodoxins modify the redox potentials of FMN molecules bound to them so that they can accept and donate electrons one by one [27]. Flavodoxins transfer electrons among different partner proteins. In some photosynthetic and/or N 2 -fixing bacteria, flavodoxin shuttles electrons from PSI to NADP + or N 2 via FNR and nitrogenase, respectively [27,108,109]. In an analogous fashion, flavodoxin donates electrons to a variety of partner proteins in different bacteria to perform biosynthetic reactions. For example, flavodoxin activates cobalamin-dependent methionine synthase, pyruvate formate-lyase, and anaerobic ribonucleotide reductase in Escherichia coli [110][111][112][113], as well as biotin synthase in Escherichia coli [111,112] and Bacillus subtilis [114]. Flavodoxin is also involved in nitrate reduction in Azotobacter vinelandii [115] and in the activation of pyruvate formate-lyase by free radicals formation [116]. On the other hand, flavodoxin has been found to function as an electron acceptor of the pyruvate oxidoreductase enzyme complex (POR) which catalyzes the oxidative decarboxylation of pyruvate in Hp [44] (Figure 2). fixing bacteria, flavodoxin shuttles electrons from PSI to NADP + or N2 via FNR and nitrogenase, respectively [27,108,109]. In an analogous fashion, flavodoxin donates electrons to a variety of partner proteins in different bacteria to perform biosynthetic reactions. For example, flavodoxin activates cobalamin-dependent methionine synthase, pyruvate formate-lyase, and anaerobic ribonucleotide reductase in Escherichia coli [110][111][112][113], as well as biotin synthase in Escherichia coli [111,112] and Bacillus subtilis [114]. Flavodoxin is also involved in nitrate reduction in Azotobacter vinelandii [115] and in the activation of pyruvate formate-lyase by free radicals formation [116]. On the other hand, flavodoxin has been found to function as an electron acceptor of the pyruvate oxidoreductase enzyme complex (POR) which catalyzes the oxidative decarboxylation of pyruvate in Hp [44] (Figure 2). The FMN cofactor in flavodoxin is made of an isoalloxazine aromatic ring system connected to a phosphate group by a ribityl chain, and it appears tightly bound at the carboxy-terminal end of the flavodoxin β-sheet. The isoalloxazine moiety usually interacts with aromatic residues, while the phosphate group forms hydrogen bonds with mainly threonine side chains and several main chain The FMN cofactor in flavodoxin is made of an isoalloxazine aromatic ring system connected to a phosphate group by a ribityl chain, and it appears tightly bound at the carboxy-terminal end of the flavodoxin β-sheet. The isoalloxazine moiety usually interacts with aromatic residues, while the phosphate group forms hydrogen bonds with mainly threonine side chains and several main chain NH groups of the protein [27,41]. Those interactions provide high stability to the apoflavodoxin FMN complex, which has a formation that begins by interaction of the isoalloxazin group with the folded apoprotein, followed by docking of the phosphate group [27,117]. The phosphate binding site has been described to be pre-formed in Anabaena PCC7119, Hp [118], and Streptococcus pneumoniae [119] apoflavodoxins, probably due to the presence of bound ions that mimic the phosphate group ( Figure 3). From sequence alignment and structural considerations, flavodoxins can be divided into two groups: long-chain (18-23 kDa) (e.g., those in Anabaena PCC 7119 or Hp) and short-chain flavodoxins (14.5-17 kDa) (e.g., those in Clostridium beijerincki or Desulfovibrio vulgaris). Long-chain flavodoxins contain an extra 20 residue loop intercalated in the β5-strand of the β-sheet [27,120,121] that does not seem to be relevant for protein stability or folding. Its sequence conservation suggests it may play a functional role [120], and it has been suggested to be responsible for the recognition of FNR and methionine synthase in E. coli [122]. Despite their differences, long-and short-chain flavodoxins share a similar three-dimensional structure. In both families, sequence conservation is high at the isoalloxazine binding loops (often referred to as the Y-and W-loops), the phosphate binding loop (P-loop) and, in the long-chain ones, their characteristic long loop. In particular, the phosphate binding site is highly conserved, the consensus sequence being T/S-X-T-G-X-T [27,41,120]. On the other hand, the W-and Y-loops usually contain, respectively, Trp and Tyr residues that are involved in the binding of the isoalloxacin moiety in FMN. Some flavodoxins, nevertheless, contain other residues in those key FMN binding positions (e.g., the tryptophan residue of the W-loop appears replaced by methionine in Clostridium beijerinckii flavodoxin, by leucine in Azobacter vinelandii or by alanine in Hp) [28,41,123].
the W-and Y-loops usually contain, respectively, Trp and Tyr residues that are involved in the binding of the isoalloxacin moiety in FMN. Some flavodoxins, nevertheless, contain other residues in those key FMN binding positions (e.g., the tryptophan residue of the W-loop appears replaced by methionine in Clostridium beijerinckii flavodoxin, by leucine in Azobacter vinelandii or by alanine in Hp) [28,41,123]. The flavodoxin from Hp (Hp-Fld) is involved in a metabolic pathway essential for Hp viability: the oxidative decarboxylation of pyruvate by the pyruvate oxidoreductase complex (POR) [27,44,45,124]. Flavodoxin synthesis in Hp is constitutive and detectable even in dormant forms of Hp which have a significantly reduced metabolic activity [125]. Hp flavodoxin has been related to lowgrade gastric mucosa-associated lymphoid tissue (MALT) lymphoma pathogenesis, and antibodies against this flavodoxin have been found in patients [126]. In anaerobic conditions, flavodoxin is able to activate imidazole antimicrobials (such as metronidazole) [124,127] and transform them into reactive intermediates that cause DNA lethal damage [124,127,128]  The flavodoxin from Hp (Hp-Fld) is involved in a metabolic pathway essential for Hp viability: the oxidative decarboxylation of pyruvate by the pyruvate oxidoreductase complex (POR) [27,44,45,124]. Flavodoxin synthesis in Hp is constitutive and detectable even in dormant forms of Hp which have a significantly reduced metabolic activity [125]. Hp flavodoxin has been related to low-grade gastric mucosa-associated lymphoid tissue (MALT) lymphoma pathogenesis, and antibodies against this flavodoxin have been found in patients [126]. In anaerobic conditions, flavodoxin is able to activate imidazole antimicrobials (such as metronidazole) [124,127] and transform them into reactive intermediates that cause DNA lethal damage [124,127,128]. Mutations in ferredoxin (FdxA), ferredoxin-like protein (FdxB), NAD(P)H flavinnitroreductase (FrxA), oxygen-insensitive NAD(P)H nitroreductase (RdxA), flavodoxin (FldA), the γ-subunit of 2-oxoglutarate oxidoreductase (OorD) or the γ-subunit of pyruvate ferredoxin oxidoreductase (PorD) have been related to metronidazole resistance in Hp [129,130]. Hp flavodoxin is encoded by the fldA gene [41,131], and its 164 residue amino acid sequence is similar to that of other flavodoxins, especially long-chain ones (sequence identities approximately 40%). Sequence differences are noticed at the cofactor binding site, specifically at the phosphate loop which, in Hp, is slightly different (T-D-S-G-N-A) from the general flavodoxin motif (T/S-X-T-G-X-T) [41]. The Hp-Fld structure is similar to that of other known flavodoxins, although the presence of some shorter loops and of an elevated percentage of small side-chain residues makes it slightly more compact [41]. Interestingly, in Hp flavodoxin, a bulky residue located in the W-loop of most flavodoxins (typically a Trp residue), appears replaced by an alanine residue (position 55). This substitution, which lowers the affinity for FMN [41], opens a pocket at the protein surface, near the bound cofactor [41,132], where small organic compounds could bind and inhibit complex formation or electron transfer reactions with partner proteins.
As Hp lacks several essential genes of the glycolysis pathway, pyruvate formation through carbon dioxide fixation may be physiologically favored, because it is the single gluconeogenic pathway in this bacterium [46]. On the other hand, oxidative decarboxylation of pyruvate is a fundamental reaction catalyzed by the pyruvate dehydrogenase complex in most aerobic organisms or by POR in anaerobic ones [133,134]. Thus, Hp POR catalyzes the last step of carbohydrates fermentation as well as the inverse pyruvate oxidative decarboxylation [135]. In this bidirectional electron transfer pathway, another essential enzyme, flavodoxin quinone reductase B (Figure 2), is involved [46].

Flavodoxins in other Pathogenic Bacteria and in the Gut Microbiota
As flavodoxin is present in gastrointestinal pathogens other than Hp, the potential of flavodoxin inhibitor-based therapies against those additional pathogens should be explored, especially in the cases of pathogens for which an essential flavodoxin has been described. On the other hand, flavodoxin is also present in a variety of gut commensal bacteria and the possible negative side effects of Hp flavodoxin inhibitors on the human microbiota should be evaluated too.
Indeed, flavodoxin genes are present in many bacteria (mainly Gram-negative ones), especially in Proteobacteria, Cyanobacteria, Aquificae, Firmicutes, Bacteroidetes, Fusobacteria, and Spirochaetes. We have collected and combined in Table 2 flavodoxin information available in the Uniprot database [136] (searching for "flavodoxin" and refining by "reviewed"), in the NCBI database [137] (searching for "flavodoxin" on the "Protein" tab, then refining by "Bacteria" on the species tag, selecting on the source databases' tag "PDB" and "UniProtKB/Swiss-Prot", and specifying on the sequence length's tag from 130 to 199 residues), and in the flavodoxin-related literature available in PubMed [137].  Proteobacteria Negative a Extracted from Uniprot by searching for "flavodoxin" and refining by "reviewed", from NCBI by searching for "flavodoxin" in the "Protein" tab and refining by "Bacteria" (in the species tag), "PDB and UniProtKB/Swiss-Prot" (in the source databases' tag) and "from 130 to 199 residues" (in the sequence length's tag) and from References [114,119,. Despite the fact that there is a great deal of unreviewed flavodoxin sequences in Uniprot, we chose to include only those that we found as described, which were flavodoxins with an existence that appeared to be firmly established. b Although the length of this sequence is more typical of short-chain flavodoxins, we classified it here as long-chain due to the absence, in sequence alignment with long chain-flavodoxins, of the characteristic 20 residue gap formed in so-aligned short-chain sequences. c Unnamed gene (gene name not reported yet. Alternative sequences may be reported elsewhere). d Although the isolation of a 148 residue flavodoxin from Clostridium pasteurianum has been reported [157,159], we did not find any sequences of such length in Uniprot. On the other hand, the sequence reported in those papers was not complete. Among the sequences in Uniprot, the one which is 140 residues in length (gene name CLPA_c13840) has the highest identity in sequence with the partial sequences reported. e Extracted from the DEG database [160]. This sequence was not identified by following the search pathway used for the rest of the sequences reported in the table. f Its Gram stain classification has been controversial [161,162].
Most of these flavodoxins appear in Proteobacteria, and some of them are known to be essential for bacterial viability. Essential flavodoxins annotated in the DEG database [160] and retrieved by searching for "flavodoxin" by function in the bacteria's tab are shown in Table 3. To anticipate the potential side effects on the human gastric microbiota associated to the use of flavodoxin-targeting therapies, we have revised the presence of flavodoxin in these commensal bacteria. Some of the key organisms present in the human gastric microbiota are shown in Table 4 classified by genus, phylum, gram staining, oxygen requirement, and (non)presence of flavodoxin(s). Essential flavodoxins are indicated, where appropriate. As shown in the table, a variety of bacterial genera from the human gastric microbiota have been identified as flavodoxin-expressing organisms. Particular care should be taken with these microorganisms in flavodoxin-based treatments, especially with Escherichiacoli, Haemophilus influenzae, and Streptococcus agalactiae which have flavodoxins that are essential for bacterial viability. Besides, the flavodoxins from the first two ones share a high percentage of identity (above 40%) with Hp-Fld ( Figure A1, Appendix A). As sequence identity levels higher than 35-40% usually involve substantial structural similarity [163], it is important to ensure that compounds developed against Hp flavodoxin are not able to kill these microorganisms at the minimal inhibitory concentrations (MIC) determined for Hp. On the other hand, it has been proposed that flavodoxin-inhibitors could interact with the protein near its cofactor binding site and then suppress flavodoxin function by modification of the cofactor redox potential or by steric blockage of the interaction between the flavodoxin and its redox partners [28]. Therefore, if Hp-Fld inhibitors bind the protein at the pocket created by the presence of an alanine at position 55, no side effects on the human microbiota would be expected because E. coli and H. influenzae flavodoxins have a bulky tryptophan residue at this position and that of S. agalactiae contains an also bulky tyrosine residue. ( Figure A1, Appendix A) The information related to the bacterial composition of the human gut microbiota was extracted from References [164][165][166][167][168][169][170]. Unreviewed indicates the existence of flavodoxin sequences reported as such in Uniprot. No scientific literature about them has been found. b Essential flavodoxin according to the DEG database [160].
On the other hand, flavodoxin is present in gastrointestinal pathogens, for some of which it is an essential protein. In Table 5, major human stomach pathogens are classified by genus, phylum, gram staining, oxygen requirement, and flavodoxin expression (Yes/No). If flavodoxin is essential for their viability, it is indicated. While different bacteria included in this table (Firmicutes, Bacteroidetes, and Proteobacteria) express flavodoxin, it has been described so far as essential only for some Proteobacteria: Campylobacter, Escherichia, Helicobacter, Salmonella, and Vibrio. As shown in Figure A2 (Appendix A), sequence identities between Hp (strain J99) flavodoxin and those from Campylobacter jejuni (strain ATCC 700819), Vibrio cholerae (strain ATCC 39541), Escherichia coli (strain K12), Salmonella enterica subsp. enterica serovar Typhi (strain Ty2), and Salmonellaenterica subsp. enterica serovar Typhimurium (strain ATCC 700720) are around 40% which is high enough to assume these proteins will show a high structural similarity [163]. Nevertheless, the amino acid sequences of these five flavodoxins, having a tryptophan residue where Hp-Fld carries an alanine one, suggest their tridimensional structures will not display a pocket near the FMN cofactor, and they may not be affected by Hp-Fld inhibitors.

Discovery of Specific Inhibitors of Hp Flavodoxin Using an Approach that Can Be Transferred to Other Pathogens
The eradication rates achieved by the conventional triple and quadruple therapies used for treating Hp infection have been decreasing, so alternative treatments are needed to fight this prevailing infection. Among the different strategies available, development of inhibitors of essential Hp targets seems promising. Following this track, our group has focused on the identification and improvement of small molecule inhibitors to target the essential Hp flavodoxin. As explained above, Hp flavodoxin bears in its surface a peculiar pocket near the FMN binding site that could be exploited to identify molecules that bind there and block function by either modifying the redox potential of the protein or by sterically interfering with the recognition of partner proteins [28]. The Hp-Fld surface pocket is absent in other epsilonproteobacteria, such as Campylobacter jejuni or Wolinella succinogenes, that carry the characteristic Trp residue of the W-loop, or in other flavodoxins where the Trp residue has been replaced by leucine (e.g., in Azotobacter vinelandii [123]) or methionine (e.g., in Clostridium beijerinkii [41]). The pocket is not expected either to be present in several Helicobacter species such as Helicobacter hepaticus (unable to colonize the human gastric mucosa but associated with chronic hepatitis, liver adenocarcinoma in mice, cholecystitis and gallbladder human cancer) [184,185], where its flavodoxin carries a tyrosine replacing the Trp. However, the pocket probably appears in the flavodoxins from Helicobacter acinonychis (a bacteria establishing lifelong infections in the stomach of cheetah and other felines) [186] and Helicobacter felis (which has been related to the development of gastritis in humans) [187] as, similarly to Hp flavodoxin, they carry an alanine residue replacing the Trp. Thus, the pocket initially observed in Hp-Fld as formed by replacement of the bulky Trp residue at the W-loop by an Ala residue is an almost exclusive feature of Hp-Fld which can be used to develop highly selective inhibitors. A priori, such new inhibitors would not give rise to side effects in humans, because flavodoxin is not present in vertebrates. An alternative possibility to use flavodoxin as an antimicrobial target stems from the fact that redox proteins have been suggested to be capable of activating nitro-compounds, as it has been demonstrated for nitroimidazole drugs such as metronidazole. Reduction of nitro groups present in prodrugs could yield cytotoxic products which would be able to act as anti-Hp compounds [124,[127][128][129][188][189][190][191][192].
Our research group has been working on the identification and development of new compounds targeting Hp flavodoxin by following the alternative possibilities of blocking flavodoxin function using binding molecules and that of using compounds that could be activated by their conversion into reactive toxic species after reduction ( Figure 4). First, small molecules, such as benzylamine, were identified as binders to Hp-Fld [28]. Due to the fact of their low binding affinity, they were poor inhibitors. A high-throughput screening method was subsequently implemented to identify them from a chemical library of other small organic molecules which could bind tighter to the protein. Thus, a 10,000 molecule chemical library was screened using pure recombinant Hp-Fld which led to the identification of 29 binding compounds that stabilized the protein as indicated by their capability to increase the temperature of mid-denaturation [42]. Four of those compounds (compounds I, II, III, and IV) were, in addition, able to inhibit the in vitro electron transfer between Fld and their recombinantly produced partner proteins POR and FqrB. Three of them (I, II, and IV) showed bactericidal activity against Hp cells and seemed to be selective for this bacterium. Inhibitors I and II exhibited therapeutic indexes (TI) of around 10, meaning their minimal cytotoxic concentrations (MCC) for eukaryotic cells were 10 times higher than the corresponding minimal inhibitory concentrations (MIC) for Hp. Nonetheless, compound IV showed lower TI due to the higher cytotoxicity and lower efficacy than molecules I and II. On the other hand, those three compounds (I, II, and IV) did not seem to be nephrotoxic or hepatotoxic when they were administered to mice at 10 mg/kg body weight, and they did not produce pathological changes in stomach, liver, heart, lung or kidney at 1 or 10 mg/kg body weight. In an attempt to obtain detailed structural information on the complexes formed by the inhibitors with the target, the interaction of compounds III (the only bacteriostatic compound of the four hits) and IV with Hp-Fld was studied by crystallography and NMR, respectively. Inhibitor III was able to replace FMN and establish hydrogen bonds and hydrophobic interactions with the protein through its nitro group and benzene ring, respectively. However, it was unclear whether the structure solved was showing the functional inhibitory interaction or rather the FMN replacement by the inhibitor occurred in a subsequent step. The NMR analysis indicated that compound IV also appeared to interact with flavodoxin through its nitro group which suggested a role for this functional group in the formation of the inhibitor-flavodoxin complex. Crystallization trials are underway to try to obtain the X-ray structure of the complex between flavodoxin and compound IV and derivatives of it.
In a first round of optimization of the initial inhibitors, 102 new molecules related to inhibitors I, II, and IV were synthetized or acquired, and their toxicity and activity were tested [43]. Among them, 20 compounds were able to bind to flavodoxin with dissociation constants in the micromolar range, some of them with higher affinity than those of the initial hits. Most of these analogues inhibited bacterial growth in vitro and nine of them showed higher therapeutic indexes than those of their parent bactericidal compounds I, II, and IV. Large increases were observed in the therapeutic indexes of the new analogues of II and IV (up to 25 times for derivatives of II and up to 59 times for derivatives of IV). Six of these compounds were further tested, and it was confirmed that they kept a similar binding affinity to flavodoxin and that they also displayed bactericidal properties [43].
A second round of optimization was then carried out. To improve the therapeutic and pharmacokinetic properties of compounds I, II, and IV, new variants carrying modified redox forms of nitro, sulfur and vinyl groups of the lead-molecules were synthesized [193]. Derivatives that contain partially or fully reduced forms of the nitro and/or ethylene groups, or partially or fully oxidized forms of the sulfur atom displayed a considerably lower toxicity against HeLa cells and mice than the corresponding leads. While the therapeutic indexes of derivatives of I or II did not represent a significant improvement, some of the derivatives of IV were effective, according to EUCAST (The European Committee on Antimicrobial Susceptibility Testing) criteria, against Hp clinical isolates resistant to common antibiotics such as metronidazole, clarithromycin, and rifampicin. Furthermore, four of these new compound IV derivatives, used as sole agents, were able to significantly reduce Hp gastric colonization in the mouse model of infection and indeed to eradicate the infection in some mice. At present, we will continue the development of derivatives of compound IV through the design and testing of more soluble variants with optimized metabolic stability and bioavailability, and we are testing their effect on other bacteria including those present in the gastric microbiota. analysis indicated that compound IV also appeared to interact with flavodoxin through its nitro group which suggested a role for this functional group in the formation of the inhibitor-flavodoxin complex. Crystallization trials are underway to try to obtain the X-ray structure of the complex between flavodoxin and compound IV and derivatives of it. In a first round of optimization of the initial inhibitors, 102 new molecules related to inhibitors I, II, and IV were synthetized or acquired, and their toxicity and activity were tested [43]. Among them, 20 compounds were able to bind to flavodoxin with dissociation constants in the micromolar range, some of them with higher affinity than those of the initial hits. Most of these analogues inhibited bacterial growth in vitro and nine of them showed higher therapeutic indexes than those of their parent bactericidal compounds I, II, and IV. Large increases were observed in the therapeutic indexes of the new analogues of II and IV (up to 25 times for derivatives of II and up to 59 times for derivatives of IV). Six of these compounds were further tested, and it was confirmed that they kept a similar binding affinity to flavodoxin and that they also displayed bactericidal properties [43].
A second round of optimization was then carried out. To improve the therapeutic and pharmacokinetic properties of compounds I, II, and IV, new variants carrying modified redox forms of nitro, sulfur and vinyl groups of the lead-molecules were synthesized [193]. Derivatives that contain partially or fully reduced forms of the nitro and/or ethylene groups, or partially or fully oxidized forms of the sulfur atom displayed a considerably lower toxicity against HeLa cells and mice than the corresponding leads. While the therapeutic indexes of derivatives of I or II did not represent a significant improvement, some of the derivatives of IV were effective, according to EUCAST (The European Committee on Antimicrobial Susceptibility Testing) criteria, against Hp clinical isolates resistant to common antibiotics such as metronidazole, clarithromycin, and rifampicin. Furthermore, four of these new compound IV derivatives, used as sole agents, were able to significantly reduce Hp gastric colonization in the mouse model of infection and indeed to eradicate the infection in some mice. At present, we will continue the development of derivatives of compound IV through the design and testing of more soluble variants with optimized metabolic stability and bioavailability, and we are testing their effect on other bacteria including those present in the gastric microbiota.
Our preliminary results indicate the Hp-Fld inhibitors so far developed are highly specific for this bacterium, and that they do not show activity (or very low) against a representative panel of other bacteria from different phyla. This selectivity will be useful to minimize the generation of resistances and suggests these inhibitors will be less damaging to the gut microbiota than broadspectrum antibiotics. Thus, this new family of selective Hp-inhibitors could provide an opportunity for the formulation of therapeutic alternatives to fight Hp-drug resistant strains [193]. As explained, Our preliminary results indicate the Hp-Fld inhibitors so far developed are highly specific for this bacterium, and that they do not show activity (or very low) against a representative panel of other bacteria from different phyla. This selectivity will be useful to minimize the generation of resistances and suggests these inhibitors will be less damaging to the gut microbiota than broad-spectrum antibiotics. Thus, this new family of selective Hp-inhibitors could provide an opportunity for the formulation of therapeutic alternatives to fight Hp-drug resistant strains [193]. As explained, due to the fact of their high selectivity, these Hp-inhibitors will not likely be effective against other pathogens. However, other bacterium-specific flavodoxin inhibitors can be identified anew, though specific screening of chemical libraries against other essential flavodoxins. Such bacterium-specific flavodoxin-inhibitors could also be developed into novel antimicrobials against gastrointestinal pathogens such as Bacillus, Campylobacter, Listeria, Salmonella, Shigella or Vibrio. To do so, the target-based approach followed to discover Hp-Fld inhibitors can be readily applied to identify specific antimicrobials against other flavodoxin-containing pathogens. Moreover, the steps carried out to improve the therapeutic and pharmacokinetic properties of the Hp-Fld inhibitors could also be followed in order to enhance the antimicrobial properties of these compounds against those bacteria.     Hp (strain J99), Escherichia coli (strain K12), and Haemophilus influenzae (strain ATCC 51907). It has been performed with Clustal Omega [194]. Asterisks (*) indicate positions with a single residue; colons (:) indicate conservation between groups of strongly similar properties; dots (.) indicate conservation between groups of weakly similar properties. (b) Sequence identity between Hp (strain J99) flavodoxin and that from Streptococcus agalactiae (ATCC BAA-611), Escherichia coli (strain K12), and Haemophilus influenzae (strain ATCC 51907). It has been calculated with Clustal Omega [194].  Ty2) and Salmonella enterica subsp. enterica serovar Typhimurium (strain ATCC 700720). It has been performed with Clustal Omega [194]. (b) Sequence identity between Hp (strain J99) flavodoxin and that from Campylobacter jejuni (strain ATCC 700819), Vibrio cholerae (strain ATCC 39541), Escherichia coli (strain K12), Salmonella enterica subsp. enterica serovar Typhi (strain Ty2), and Salmonellaenterica subsp. enterica serovar Typhimurium (strain ATCC 700720). It has been calculated with Clustal Omega [194].  Ty2) and Salmonella enterica subsp. enterica serovar Typhimurium (strain ATCC 700720). It has been performed with Clustal Omega [194]. (b) Sequence identity between Hp (strain J99) flavodoxin and that from Campylobacter jejuni (strain ATCC 700819), Vibrio cholerae (strain ATCC 39541), Escherichia coli (strain K12), Salmonella enterica subsp. enterica serovar Typhi (strain Ty2), and Salmonellaenterica subsp. enterica serovar Typhimurium (strain ATCC 700720). It has been calculated with Clustal Omega [194].