Evaluation of Fusobacterium nucleatum Enoyl-ACP Reductase (FabK) as a Narrow-Spectrum Drug Target

Fusobacterium nucleatum, a pathobiont inhabiting the oral cavity, contributes to opportunistic diseases, such as periodontal diseases and gastrointestinal cancers, which involve microbiota imbalance. Broad-spectrum antimicrobial agents, while effective against F. nucleatum infections, can exacerbate dysbiosis. This necessitates the discovery of more targeted narrow-spectrum antimicrobial agents. We therefore investigated the potential for the fusobacterial enoyl-ACP reductase II (ENR II) isoenzyme FnFabK (C4N14_ 04250) as a narrow-spectrum drug target. ENRs catalyze the rate-limiting step in the bacterial fatty acid synthesis pathway. Bioinformatics revealed that of the four distinct bacterial ENR isoforms, F. nucleatum specifically encodes FnFabK. Genetic studies revealed that fabK was indispensable for F. nucleatum growth, as the gene could not be deleted, and silencing of its mRNA inhibited growth under the test conditions. Remarkably, exogenous fatty acids failed to rescue growth inhibition caused by the silencing of fabK. Screening of synthetic phenylimidazole analogues of a known FabK inhibitor identified an inhibitor (i.e., 681) of FnFabK enzymatic activity and F. nucleatum growth, with an IC50 of 2.1 μM (1.0 μg/mL) and a MIC of 0.4 μg/mL, respectively. Exogenous fatty acids did not attenuate the activity of 681 against F. nucleatum. Furthermore, FnFabK was confirmed as the intracellular target of 681 based on the overexpression of FnFabK shifting MICs and 681-resistant mutants having amino acid substitutions in FnFabK or mutations in other genetic loci affecting fatty acid biosynthesis. 681 had minimal activity against a range of commensal flora, and it was less active against streptococci in physiologic fatty acids. Taken together, FnFabK is an essential enzyme that is amenable to drug targeting for the discovery and development of narrow-spectrum antimicrobial agents.


■ INTRODUCTION
−5 In periodontal disease, F. nucleatum assists periodontal pathogens such as Porphyromonas gingivalis by promoting their integration into the oral microbiota community.In the context of CRC, genomic analysis of patient biopsies has shown a frequent association of F. nucleatum with tumor tissues, as opposed to normal tissues. 6,7F. nucleatum expresses virulence factors that contribute to the development of a pro-tumorigenic environment, which includes suppression of the immune system, 8,9 promotion of chemoresistance, 10 and induction of inflammation. 4,11−16 The advantage of eliminating F. nucleatum was demonstrated in a study by Bullman et al., 12 wherein mice treated with the antimicrobial metronidazole exhibited a decrease in F. nucleatum positive xenograft CRC tumors.Highlighting the importance of this work, there are two ongoing clinical trials exploring the utility of metronidazole as an adjuvant to standard of care approaches for CRC. 17,18However, metronidazole has limitations as it is a broad-spectrum antianaerobic antibiotic that is likely to promote dysbiosis during the treatment of CRC.Thus, there is a need for narrowspectrum antimicrobial agents that target F. nucleatum while minimizing collateral damage of the normal microbiota.Recent research has identified different potential strategies, including aspirin, that kill F. nucleatum and shows efficacy in Apc Min/+ mice; 13 bacteriophages that target F. nucleatum; 19−22 and derivatives of the natural product higenamine that inhibits F. nucleatum growth. 23To address the need for F. nucleatumspecific antimicrobial agents, we reviewed prior literature 24,25 on antimicrobial drug targets with the view of identifying large-scale differences between F. nucleatum and common gut flora (i.e., Bacteroides and Firmicutes). 26This led us to explore enoyl-ACP reductase (ENR; Figure 1), which catalyzes the rate-limiting step of the bacterial type II fatty acid synthesis pathway (FAS-II) since these enzymes are attractive targets for narrow-spectrum drugs. 27FAS-II is required for bacteria to synthesize their phospholipid membranes.Additionally, Gram-negative bacteria like F. nucleatum depend on FAS-II to synthesize β-hydroxyacyl-ACPs, which are essential precursors for the lipid component of lipopolysaccharide outer membranes. 28,29In contrast, humans adopt the type I fatty acid synthesis pathway that is conducted by a large multifunctional enzyme. 30Among bacteria, there are primarily four ENR isoenzymes, of which FabI, FabL, and FabV are structurally related short-chain dehydrogenases that use NADPH in their enoyl reduction reactions. 31Conversely, the fourth isoform, FabK, is a flavoenzyme that uses FMN and NADPH. 32The occurrence of four isoenzymes makes ENRs attractive targets for antimicrobial agents with a narrower spectrum of action.The feasibility of this concept is exemplified by afabicin, a selective inhibitor for FabI that is in clinical development for Staphylococcus aureus skin and skin structure infections, and isoniazid that inhibits the mycobacterial ENR (InhA) and is an antituberculosis drug. 33,34ur bioinformatic analysis of F. nucleatum genomes revealed that they encode FabK as their only ENR (Figure 1).We, and others, have reported that FabK enzymes can be selectively inhibited by small molecules, with the phenylimidazole class of compounds representing potential drug candidates.Studies of a lead phenylimidazole, designated as 296, inhibited Clostridioides difficile FabK in cellular and enzymatic assays and was efficacious in treating C. difficile infection (CDI) in mice. 26,35,36mportantly, 296 did not appear to significantly disrupt the gut microbiome of mice, in contrast to the standard-of-care drugs for CDI, vancomycin, and fidaxomicin. 35The FabK proteins from F. nucleatum (FnFabK) and C. difficile (CdFabK) are closely related, with 56% overall identity and 74% overall similarity, and their active sites differ by only a single residue at amino acid position 46; Figure S1.This prompted us to investigate whether FnFabK could be a target.In this study, we report that FnFabK is a druggable antimicrobial target that is essential for the growth of F. nucleatum.

■ RESULTS AND DISCUSSION
Bioinformatic Analysis.The F. nucleatum FAS-II pathway was assembled from DELTA-BLAST searches against the genome of strain 23726 (Figure 1; Table S1; accession no.SAMN00001497).Given the phylogenetic relatedness of F. nucleatum to Gram-positive firmicutes, we adopted enzymes from the FAS-II cycle of Enterococcus faecalis FA2−2 (accession no.SAMN22569088) to identify F. nucleatum FAS-II enzymes, using criteria of >80% query coverage and >35% identity.−40 We identified homologues for each initiation and elongation step of FAS-II.Crucially, FnFabK was the only ENR in strain 23726 and all other annotated F. nucleatum genomes (n = 51) studied from GenBank.No homologues of FabI, FabL, or FabV were found.All FnFabK contained the distinctive FMN binding motif that is absent in other ENR isoenzymes and had >90% sequence identity.Hence, we further studied FnFabK, considering that the homologous CdFabK is a drug target in C. difficile. 26enetic Essentiality of FnFabK.The genetic essentiality of FnFabK was examined through gene deletion trials and gene silencing.Initially, we attempted to delete fabK by allelic exchange using a galactose kinase (GalK)-based genetic system in F. nucleatum 23726ΔgalK (a galK-deletion mutant). 41The suicide vector containing the fabK allelic cassette was integrated into the chromosome adjacent to fabK, and cells were treated with 2-deoxy-D-galactose to counter-select and identify the  S1.
second crossover mutants as those with potential fabK deletions.However, PCR analysis of colonies from two independent experiments (n = 10 each) showed that they all carried wild-type fabK.This indicated that fabK was not disrupted, and it was likely to be indispensable for F. nucleatum growth.
Since our deletion approach utilized a genetically modified strain lacking galK and we did not attempt to delete the gene while the strain was complemented with a codon-altered fabK, we next employed gene silencing in the wild-type strain to confirm that F. nucleatum fabK is essential.Thus, we adopted gene silencing as a known target validation approach 24,42−44 by adapting a paired-termini gene silencing method to inhibit fabK mRNA translation with antisense nucleotides that are directed toward the ribosome binding site. 26The F. nucleatum antisense vector (pHFK2) was constructed by cloning the paired-termini and anhydrotetracycline (ATc)-inducible pTET promoter sequences from the Clostridial vector pMSPT into the F. nucleatum cloning vector pCWU6 (Figure S2).Next, two different antisense nucleotides (100 bp) were cloned into pHFK2, i.e., specifically, FabK antisense 1 (FabKAS1) targeted the 50 bp upstream and downstream regions of the fabK start codon, whereas FabK antisense 2 (FabKAS2) targeted 25 bp upstream and 75 bp downstream of the start codon.Following induction with ATc, both FabKAS1 and FabKAS2 arrested the growth of F. nucleatum 23726 in an ATc concentrationdependent manner (Figure 2A).RT-qPCR confirmed that gene silencing inhibited the fabK mRNA ribosome binding region, as significantly fewer transcripts were detected for the antisense regions (Figure 2B).Interestingly, exogenous fatty acids could not relieve F. nucleatum growth inhibition caused by the antisenses (Figure 2C), indicating that fabK remained essential even when F. nucleatum cultures were provided with host lipids.We adopted a mixture of fatty acids of varying chain lengths and saturation (described in the Materials and Methods; Table S2) that are based on the composition of free fatty acids in the oral cavity 45 and 0.1% v/v of Tween-80 (as a source of oleic acid, C18:1 cis-9; Figure S3).Thus, under these test conditions, fabK appears to be indispensable for F. nucleatum growth irrespective of the presence of exogenous fatty acids.Nonetheless, further genetic experiments may be warranted to determine whether F. nucleatum fabK is essential under different conditions (e.g., gene deletion in strains complemented with codon-altered fabK cloned with or without a conditional promoter).
681 Inhibits FnFabK and Directly Interacts with Key Residues in the Catalytic Site.Compounds 681 and 701 inhibited the enoyl reductase activity of purified FnFabK in a dose-dependent manner, with mean IC 50 s of 2.1 μM (equivalent to 1.0 μg/mL; 681) and 10.6 μM (equivalent to 4.6 μg/mL; 701) from three replicates, respectively (Figure 3A,B).The Hill coefficient of 681 was 1.01 (with a 95% confidence interval of 0.70−1.48),which is close to the ideal value of 1.000 for the Hill slope.Similarly, the Hill coefficient of 701 was 1.06 (95% confidence interval of 0.59−1.75).These were comparable to previously reported inhibition of CdFabK with a 681 IC 50 of 0.45 μM (equivalent to 0.2 μg/mL; Hill slope 0.97) and a 701 IC 50 of 1.88 μM (equivalent to 0.8 μg/mL; Hill slope 0.94). 36nterestingly, the fivefold higher IC 50 of 701 compared to 681 roughly correlates with the fourfold difference in MIC between the two compounds.Hence, the difference in activity between 681 and 701 may likely be due to binding affinity rather than penetration into cells, as was initially speculated; however, drug penetration was not tested and was beyond the scope of the study.
Interactions of 681 with FnFabK were examined using a homology model of the enzyme that was based on the reported crystal structure of Streptococcus pneumoniae FabK (Figure 3C). 47FnFabK and SpFabK proteins are 56% identical with 71% similarity.Their active site residues are 94% identical, differing only in the presence of an ASN residue at the FnFabK GLN-46 position.The docking pose of 681 closely matched the binding orientation of the phenylimidazole compound (an analogue of 296) that was experimentally determined in the SpFabK cocrystal structure (PDB ID: 2Z6J). 47The phenylimidazole ring system of 681 occupied the same position in the FnFabK active site as the homologous ring system of the 296 analogue in the SpFabK structure.Similarly, the thiazole ring and urea groups of 681 engaged in a pi-stacking interaction with the FMN cofactor in a similar orientation as the 296-analogue in the SpFabK structure.The methyl sulfone occupied the same region of the active site as the pyridine ring of the 296-analogue, potentially engaging in a hydrogen-bond interaction with ASN-46 (i.e., F. nucleatum numbering) in SpFabK and CdFabK enzymes.However, this interaction is not possible in FnFabK  since it carries a GLY-46 residue, which may explain why 681 has a more potent IC 50 (0.45 μM) against CdFabK. 36ompounds 681 and 701 differ only by the presence of a methyl-sulfone group at the 6-position of the benzothiazole ring system in 681.While this functional group does not appear to directly engage the FnFabK active site in an electrostatic interaction, as discussed above, the electron-withdrawing character of the functional group likely improves 681 binding by lowering the pK a of the urea nitrogen adjacent to the benzothiazole ring (pK a ∼8), allowing the compound to coordinate an active site metal as an anion that is present in FabK enzymes.We previously discussed this structure−activity relationship (SAR) observation for inhibitors of CdFabK 36 and have noted this in our deposited CdFabK cocrystal structure (PDB ID: 7L00), which carries an active site sodium.Further, the lipophilic character of the methyl sulfone of 681 may improve binding affinity as these inhibitors are fatty acid substrate mimics.These chemical properties likely explain the difference in biochemical activity noted between the two compounds (2.1 μM for 681 vs 10.6 μM for 701).Since 681 had stronger inhibition in both the enzymatic and the whole-cell MIC assays, it was selected for further characterization.
Third, we generated 681-resistant mutants by serially passaging F. nucleatum 23726 in increasing concentrations of compound until there was at least an eightfold increase in the MIC.This corresponded to 4−5 passages across three separate cultures.Mutants isolated from each experiment (i.e., JR1-JR4) had MICs of 3.1−12.5μg/mL and were genome sequenced to identify key resistance-associated mutations, which were confirmed by Sanger sequencing (Table 3); the parent strain was also sequenced.All genomes were deposited in NCBI under accession number PRJNA1028625.Across the four mutants, three classes of resistance-associated mutations arose as follows: in the ORF for fabK, its upstream regulatory region, and/or the upstream region of C4N14_08920 (which encodes a putative nitronate monooxygenase that is homologous to the Helicobacter pylori FabX, a bifunctional dehydrogenase/isomerase, which synthesizes unsaturated fatty acids). 49C4N14_08920 aligns with the H. pylori FabX, sharing 50.8% identity and 69.6% similarity.C4N14_08920 also possesses the key cysteine residues for the formation of the [4Fe−4S] clusters and residues that are important for FMN binding (Figure S4).Structural modeling of C4N14_08920 further indicated that it was homologous to H. pylori FabX (Figure S5).Other mutations acquired during serial passage are reported in Table S3.
Structural Analysis of Amino Acid Substitutions in FnFabK.Two 681-resistant mutants had mutations in FabK (i.e., Gly96Ser and Ala132Thr in JR1 and JR4, respectively).Homology modeling (Figure 3C) revealed that glycine-96 neighbors the phenylimidazole binding region where the adjacent alanine-97 (backbone amide) forms an H-bond with the imidazole ring of the inhibitor.The Gly96Ser mutation is, therefore, predicted to interfere with the drug-binding pocket.On the other hand, alanine-132 is distantly located from the drug-binding pocket (∼13 Å from the ligand).To better understand the contribution of the Ala132Thr substitution in resistance, we cloned the wild-type alanine-132 and threonine-132 mutants into F. nucleatum 23726 under the wild-type fabK promoter and determined the 681 MIC (n = 4 replicates/test strain).The threonine-132 mutant conferred low-level resistance to 681 (MIC range of 2.5−5 μg/mL) when compared to the wild-type FabK (MIC = 1.3 μg/mL) and the empty vector controls (MIC = 0.6 μg/mL); there was no variation in the replicate MIC values against the wild-type or the empty vector controls.These results suggest that resistance to 681 can develop by direct alterations to FabK, supporting our hypothesis that FabK is the intracellular target.
Genetic Analysis of Mutations in fabK Regulatory Regions.In addition to the Gly96Ser mutation, mutant JR1 (MIC = 3.1 μg/mL) evolved a cytosine to thymine mutation   S3.
upstream of fabK.To define the location of this mutation, we compared the mutation site with the genome of F. nucleatum 25586 because this strain was previously used to create a comprehensive global transcriptional map of F. nucleatum that is found on the Fuso-Base platform (hosted by the Helmholtz Institute for RNA Infection Research). 50This identified that the mutation occurred 1 bp from the transcription start site (i.e., TSS + 1) of fabK.Further assessment of the mutation region in Fuso-Base and in the bacterial promoter prediction software BPROM revealed that the mutation site was 7 and 30 bp downstream of the −10 (5′ TGATAATAT 3′) and −35 (5′ TTGACA 3′) boxes, respectively.Mutant JR3 (MIC = 12.5 μg/ mL) also acquired the TSS + 1 mutation found in JR1.
Mutations in regulatory regions of ENR genes are known to enhance transcription, resulting in resistance to inhibitors, as seen with the FabI inhibitor triclosan. 26,51Furthermore, as described above, when FabK was overexpressed from the pTET promoter, F. nucleatum was resistant to 681.We therefore suspected that the TSS + 1 mutation increased fabK mRNA, and this was confirmed by RT-qPCR (Figure 4A).As expected, JR1 and JR3, with the TSS + 1 mutation, showed a significant 7−19fold increase in fabK mRNA transcripts compared to the wildtype transcripts (Figure 4A).In contrast, fabK transcript levels for JR2 and JR4, lacking the TSS+1 mutation, resembled those of wild-type (Figure 4A).We confirmed that the increased expression of fabK was not due to nonspecific upregulation of FAS-II by showing that there was no effect on fabG transcription (C4N14_09755; 3-ketoacyl-ACP reductase) (Figure S6).Genetic Analysis of Mutations Upstream of fabX (C4N14_08920).Mutant JR2 (MIC = 6.3 μg/mL) had a 12 bp deletion (5′ CAAATTATAGTT 3′) that was identified upstream of the putative fabX start codon.According to Fuso-Base, fabX in F. nucleatum 25586 is part of an operon with Fn0663 52 (i.e., C4N14_08915 in strain 23726), a hypothetical protein that is poorly transcribed when compared to fabX. 50nalysis of the operon in BPROM indicated that there are two transcriptional start sites and two promoters upstream of fabX.The first corresponds to the TSS and promoter identified by Ponath et al. 50that is upstream of C4N14_08915, while the second predicted TSS and promoter are located in C4N14_08915 at 107 bp upstream of the fabX start codon.The 12 bp deletion occurred in C4N14_08915, resulting in an in-frame deletion of four amino acids (Ser121-Ser124), which did not affect the second promoter but shortened the distance from the promoter to the start of the gene.This 12 bp mutation also arose in JR4 (MIC = 6.3 μg/mL) that carried the Ala132Thr substitution in FabK.JR2 and JR4 exhibited a 5−13-fold increase in fabX transcription when compared to wild-type (Figure 4A); increased expression of fabX also did not affect fabG transcription (Figure S6).To determine if increased fabX expression caused resistance to 681, the gene was expressed from the pTET promoter in wild-type F. nucleatum 23726.When fabX was induced, the MIC of 681 increased to 3.1 μg/mL (Table 2) when compared to the empty vector (MIC = 0.4 μg/ mL).FabX had a significantly lower effect on 681-activity than FabK when both were similarly expressed from the pTET promoter, consistent with FabK being the drug target (i.e., 3.1 μg/mL [for fabX] versus >100 μg/mL [for fabK]).Interestingly, the pTET overexpression of fabX did not impact growth (Figure S7), suggesting that the effects on growth that were seen in JR2 and JR4 (Figure 4B) may be due to the mutation in the hypothetical protein C4N14_08915.
Antimicrobial Spectrum of Activity of 681.We evaluated the antimicrobial spectrum of 681 against fusobacteria and other gut flora (Table 4).The compound was inactive (MIC > 100 μg/mL) against representative gut/oral species, including Bacteroides ovatus and Bifidobacterium breve, but was active (MICs < 1.6 μg/mL) against all tested Streptococci.To examine if Streptococci can use environmental fatty acids to bypass 681inhibition, we tested the commensals Streptococcus pyogenes and Streptococcus salivarius in comparison to F. nucleatum (Figure 5A−C).Figure 5B,C shows that when both species were supplemented with fatty acids, they partly bypassed inhibition by 681.In the presence of 681 and exogenous fatty acids, respectively, S. pyogenes and S. salivarius showed 4−5-fold and twofold increases in biomass (based on OD 600 nm).In contrast, the same fatty acid composition did not rescue F. nucleatum growth, either in 681 or the control thiolactomycin (Figure 5A).This finding supports the above observations that the silencing of fabK mRNA was not rescued by fatty acids (Figure 2C).Noteworthily, the protection by fatty acids of Streptococci seems weaker when compared to previous work, 26,53 but this may be due to us using a lower concentration of fatty acids to reflect physiological concentrations in the oral cavity. 45Compound 681 was also tested for cytotoxicity against Vero cells (a monkey kidney epithelial cell line) and HCT116 cells (a colonic epithelial cell line).The IC 50 for 681 was above the maximum tested concentration (>100 μg/mL) in both cell lines, while the cytotoxic controls doxorubicin (Vero) and staurosporine (HCT116) had IC 50 s of 8.1 ± 1.0 μg/mL and 32.6 ± 14.2 μg/mL, respectively.Together, these findings indicate that 681 has promising antibacterial selectivity and minimal cytotoxicity.

■ CONCLUSIONS
Narrow-spectrum antimicrobial agents have become increasingly attractive therapeutic concepts for diseases that are characterized by an underlying dysbiosis, infections involving monospecies, and to avoid collateral damage to the gut microbiome. 54,55This concept has been focused on ESKAPE pathogens, C. difficile, and Mycobacterium tuberculosis. 55Building on our previous studies establishing CdFabK as a drug target, 26,36 we identified that F. nucleatum FabK is also a target for FabK inhibitors.We presented several lines of evidence that FnFabK is a drug target, including the inability to delete fabK; the observed attenuation of growth due to gene silencing of fabK; its druggability with inhibitors (e.g., 681 and 701); and the demonstration that 681 targets the enzyme intracellularly.When considered together, the above-mentioned results strongly support our hypothesis that fabK is essential to F. nucleatum growth under the media conditions tested, with or without exogenous host fatty acids at 37 °C.During these studies, we discovered that F. nucleatum encodes a homologue of H. pylori FabX and that it can partly compensate for FabK inhibition when dysregulated, as 681-resistant mutants arose by overexpressing fabX.FabX is likely involved in the synthesis of unsaturated fatty acids, but since cells require saturated fatty acids to make phospholipids then it is inconceivable that FabX can replace the role of FabK.681-resistant mutants also arose from amino acid substitutions in FabK and overexpression of fabK due to mutations in the promoter region.Supplementation with exogenous fatty acids did not protect F. nucleatum against FabK inhibition or the FabB/FabF inhibitor thiolactomycin, meaning that the organism is unlikely to bypass FAS-II inhibition.This is unsurprising given the importance of FAS-II intermediates to the synthesis of Gram-negative outer membranes. 29Based on the MICs of 681 against different commensal species, we predict that FnFabK inhibitors could counteract F. nucleatum without significantly causing collateral damage.However, further work is needed to examine the efficacy of FnFabK inhibitors in a disease context, such as colorectal cancer or periodontal disease, and their effect on microbiomes.Nonetheless, 681 could represent a good starting point for chemical optimization to discover drug candidates for infections involving F. nucleatum.

■ MATERIALS AND METHODS
Strains and Growth Conditions.F. nucleatum subsp.nucleatum ATCC 23726 was routinely cultured in Columbia broth or Columbia agar with 5% v/v defibrinated sheep's blood (HemoStat Laboratories); thiamphenicol (5 μg/mL) was added to maintain plasmids.All strains, plasmids, and media conditions are in Tables S4 and S5.Anaerobic cultures were grown at 37 °C in a Don Whitley A35 anaerobic chamber; aerobic cultures were grown at 37 °C in 5% CO 2 in a humidified incubator.When required, media was supplemented with fatty acids that were from the following sources [Sigma (oleic acid and stearic acid), Acros Organics (linoleic acid, myristic acid, and palmitic acid), and Chem Impex (linolenic acid)].
FAS-II Inhibitors.The synthesis of 681 and all other FabK inhibitors was fully described by Norseeda et al. 36 Compound structures are listed in Table S6.Cerulenin was from the Cayman Chemical Company, and thiolactomycin was from Toronto Research Chemicals.
Bioinformatics Analysis.This was performed by using the F. nucleatum genome of strains 23726 and 25586.KEGG and GenBank annotations were reviewed, with confirmation of annotated genes by searching with both DELTA-BLAST and PHYRE2.FnFabK was further confirmed as the active site residue, His-145, aligned to the active histidine of the E. faecalis FA2−2 FabK (accession no.SAMN22569088).FnFabK also had identical residues for each of the FMN binding residues (Gly-19, Ala21, Asn-69, Glu-137, Gly142, Gly-170, Gln189, Gly-191, and Thr-192), except Gly-19, which is an alanine in FnFabK.Since FabK was identified as the sole ENR of F. nucleatum 23726, the analysis was expanded to investigate the ENRs of other Fusobacterium spp.
Genetic Manipulation of F. nucleatum.The transformation of F. nucleatum 23726 was performed via electroporation as previously described, 41 except that Columbia broth  S4; the encoded ENR and major FAS-II regulators of each species are indicated.If empty, then no regulator was identified.MICs were determined from three biological replicates.b The active site of F. necrophorum 25286 FabK differs from F. nucleatum FabK by the former having an alanine at position 119 instead of a proline in the latter.c These strains contain the FabT regulator.d Incomplete inhibition was observed, with a trailing growth effect from 6.3 to 100 μg/mL; hence, the MIC for complete inhibition was >100 μg/mL; FabK of Pg strain F0569 is 100% identical to that of strain 33277.was used.(i) Antisense vector: F. nucleatum gene silencing vector, pHFK2, was developed by inserting the ATc inducible paired-termini antisense sequences from the clostridial vector pMSPT into pCWU6, a F. nucleatum vector. 56pCWU6 was amplified by PCR (CloneAmp polymerase) and recombined with the BamHI and KpnI fragments of pMSPT containing the Tet promoter and paired-termini sequences. 26Antisense nucleotides targeting either 50 bp upstream and downstream of the fabK start codon (FabKAS1) or 25 bp upstream and 75 bp downstream (FabKAS2) were cloned into SphI and XhoI restriction sites of pHFK2.(ii) Complementation vectors: these were generated by PCR amplifying the fabK promoter and gene and introducing KpnI and HindIII restriction sites at 5′ and 3′ ends, respectively.Purified PCR amplicons were inserted into the KpnI and HindIII sites of pCWU6.(iii) Overexpression vectors: the corresponding gene, fabK or fabX, was amplified via PCR, introducing HindIII or SacI restriction sites to the 5′ end, respectively.BamHI restriction sites were introduced to the 3′ end of both.Purified PCR amplicons were inserted into the respective restriction sites of pHFK2.(iv) Allelic exchange vector: vector, pCWU8-FabK, was generated using a previously described protocol, 41,56 whereby 1-kb regions upstream and downstream of fabK were cloned into the suicide vector pCWU8 using a Gibson Assembly Cloning Kit (New England Biolabs).The recombinant vector was transformed into strain F. nucleatum 23726ΔgalK, and successful integrants were collected that were resistant to thiamphenicol (5 μg/mL).Since the suicide vector expresses GalK, which is toxic in the presence of 2deoxy-D-galactose (2-DG), the compound 2-DG (0.25% w/v) was added to counter-select for mutants that underwent a second homologous recombination.Ten resulting colonies were then tested for deletion of fabK via PCR, using primers in Table S7.The experiment was independently performed twice for a total of 20 colonies screened for fabK deletion.All experiments were done anaerobically at 37 °C in tryptic soy broth supplemented with 1% w/v peptone and 0.25% w/v autoclaved L-cysteine hydrochloride (TSPC) or TSPC agar with thiamphenicol or 2-DG.
Determination of Minimum Inhibitory Concentration (MICs).Compounds were twofold serially diluted (100−0.2μg/ mL) across a 96-well plate in 100 μL of media used for the specific organism (Table S4).For anaerobes, plates were incubated anaerobically at 37 °C for 1 h and then inoculated with approximately 10 5 to 10 6 CFU/well.Plates were incubated at 37 °C for up to 48 h, and MICs were read by visual inspection.
Growth Kinetics.To plates containing twofold serial dilutions of compounds, exponential phase cultures were added to a final OD 600 nm of 0.1.Plates containing F. nucleatum in Columbia broth were incubated at 37 °C inside of a Tecan Infinite 200 Pro, which was inside an anaerobic chamber; growth kinetics for Streptococcus spp. in brain heart infusion broth were done aerobically in the same Tecan instrument.
Transcription Analysis.For antisense analysis, midlogarithmic cultures (OD 600 nm 0.45−0.65) in Columbia broth (75 mL) were split into two 30 mL aliquots and treated with either 62.5 ng/mL ATc or vehicle (ethanol).Aliquots (5 mL) were recovered at time 0 and at 1 and 3 h after treatment with ATc or vehicle.The aliquots were treated with a recipe of RNALater, 57 and cell pellets were collected by centrifugation at 4000g for 10 min at 4 °C and stored overnight at −80 °C.After mechanical cell lysis, RNA was isolated using the Qiagen RNEasy mini kit.cDNA was synthesized from 200 ng of RNA using qScript cDNA SuperMix (Quanta Biosciences), and RT-qPCR was performed with SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) in an Applied Biosystems QuantStudio 6 Flex.Results were calculated using the 2 −ΔΔCT method 58 and normalized to the 16S rRNA.Gene expressions in mutants were analyzed similarly.Primers used for RT-qPCR analysis can be found in Table S7.
Generation of 681-Resistant Mutants.Overnight cultures of F. nucleatum 23726 were diluted 1:50 in fresh Columbia broth, which was used to inoculate 96-well plates containing twofold serial dilutions of compound 681.After 48 h of incubation, cultures were recovered from wells one dilution below the MIC, and this was used to establish the inocula for the subsequent passage into 96-well plates containing 681.This process was repeated until cultures grew at 8× the original MIC for two consecutive passages.Cultures were then plated onto Columbia blood agar plates and incubated for 72 h at 37 °C.Single colonies were collected into Columbia broth with 0.39 μg/mL of 681 and cultured for up to 48 h and then plated onto selective Columbia blood agar plates to further purify putative 681-resistant colonies.
Genome Sequencing and Analysis.DNA was extracted from overnight cultures (15 mL) using the Qiagen QIAamp DNA mini kit according to the manufacturer's protocol.Whole genome sequencing was performed by SeqCenter as follows.Sample libraries were prepared using the Illumina DNA Prep kit and IDT 10 bp UDI indices and sequenced on an Illumina NextSeq 2000, producing 2 × 151 bp reads.Demultiplexing, quality control, and adapter trimming were performed with bclconvert* (v3.9.3) [*bcl-convert: a proprietary Illumina software for the conversion of bcl files to basecalls].FastQ files were uploaded to the Bacterial and Viral Bioinformatics Resource Center.First, the parent strain was annotated using the Genome Annotation tool.Next, the variation analysis tool was used to identify genetic differences in the mutant genomes.The identified mutations were confirmed via sanger sequencing.
Protein Expression and Purification.Codon-optimized F. nucleatum fabK was cloned into the pET15b (N-term hexa-HIS tag) at NdeI and BamHI restriction sites by Azenta Life Sciences, and the vector was transformed into Escherichia coli XL2-Blue for storage and BL21(DE3) for expression.Cells were grown to an OD 600 of ∼0.6 in 500 mL of Terrific Broth with 100 μg/mL ampicillin at 37 °C with shaking at 250 rpm; 0.1 mM isopropyl β-D-1-thiogalactopyranoside was used to induce expression, while 0.5 mM flavin mononucleotide (FMN) was also added to improve protein stability and yield.Cells were then incubated for an additional 18 h at 18 °C and 220 rpm before being harvested by centrifugation at 10,000 rpm for 15 min at 4 °C.Pellets were resuspended in lysis buffer containing 50 mM Tris pH 8.0, 100 mM NH 4 Cl, 10% v/v glycerol, 100 μM FMN, 5 mM imidazole, 0.5 mg/mL lysozyme, 0.5% v/v Triton-X 100, 5 mM MgCl 2 , 25 mM sucrose, 2 mM dithiothreitol (DTT), 1 mg/ mL DNase, and 1.5 mL protease inhibitor cocktail per 100 mL.For every gram of cells, 10−15 mL of lysis buffer was used and stirred at 4 °C for 1 h before being sonicated at 50% amplitude for a total of 8 min using a cycle of 8 s on and 24 s off.The lysate was centrifuged at 18,000 rpm at 4 °C for 15 min, after which the supernatant was passed through a 0.45 mm filter.The first step of purification uses His Trap affinity chromatography on a 5 mL HP column (Cytiva Life Sciences).The binding buffer contained 50 mM Tris pH 8.0, 100 mM NH 4 Cl, 10% v/v glycerol, 100 μM FMN, 5 mM imidazole, and 2 mM DTT, and the elution buffer contained the same components with an increased imidazole concentration of 250 mM.Eluted protein was further purified by gel filtration on Superdex 200 pg (Cytiva Life Sciences) using a running buffer containing 50 mM HEPES pH 8.0, 300 mM NH 4 Cl, 10% v/v glycerol, 2 mM DTT, and 100 μM FMN.After purification, concentrated protein was stored at −80 °C in 35% v/v glycerol.
FnFabK Enzymatic Assay.The FnFabK assay was conducted via the following protocol: reactions were carried out at 25 °C in assay buffer (100 mM HEPES pH 8.0, 500 mM NH 4 Cl, 10% v/v glycerol) with 150 μM crotonyl-CoA (butenoyl-CoA) and 160 μM NADH.FabK enzyme was diluted using 2.5 mg/mL gamma globulin in assay buffer to a working stock of 30 nM and was incubated with 681 threefold serial dilutions ranging from 100 μM to 1.69 nM.Incubation lasted for a total of 10 min before the addition of crotonyl-CoA substrate, and the reaction was initiated by the addition of NADH for a final assay volume of 100 μL.The oxidation of NADH to NAD+ was measured by tracking fluorescence (340 nm/460 nm) with a Biotek Synergy H1 microplate reader in 15 s intervals for a total of 10 min to monitor the rate of reaction.The reaction was conducted in Greiner Bio-One 384-well μClear Bottom Polystyrene Microplates.For IC 50 calculations, linear slopes were calculated from the first 5 min and used to determine the reaction rates.Measurements were conducted in triplicate, and IC 50 's were calculated via GraphPad Prism 9.1.2using fourparameter logistic (Hill) curve analysis using equation Y = bottom + (top − bottom)/[1 + 10 (Log IC 50 − X) × Hill Slope], where X is the logarithm of the dose and Y is the response.The assay concentrations of the cofactor, NADH, and substrate, crotonyl-CoA, were also determined via GraphPad Prism 9.1.2using the substrate inhibition model to determine K m(app) .
FnFabK 681 Homology Modeling.The homodimer homology model of FnFabK was generated from the S. pneumoniae FabK experimental structure (PDB ID: 2Z6J), 47 using Prime software 59,60 in the 2023 Schrodinger modeling suite (Schrodinger Release 2023-1: Prime, Schrodinger, LLC, New York, NY, 2023), and included placement of the FMN cofactor and inhibitor compound.The faster, knowledge-based model-building method was selected due to the high sequence similarity.Default values were used for other settings.Structure and loop position refinements were performed by using iterative energy calculations.
Cytotoxicity.This was done as described 36 using colorectal carcinoma HCT116 and Vero epithelial cells cultured at 37 °C in 5% CO 2 in Dulbecco's modified Eagle's medium (DMEM) with 4.5 g/L glucose and L-glutamine.The medium was supplemented with 10% v/v FetalGro (Rocky Mountain Biologicals).Approximately 25,000/well were aliquoted into black 96-well plates (Greiner Bio-One) and incubated overnight, before being resuspended in DMEM supplemented with 20% Gibco KnockOut Serum Replacement (Thermo Fisher Scientific).Compounds (100−1.6 μg/mL) were then added to the cells and incubated for 24 h at 37 °C and 5% CO 2 .Cell viability was analyzed using resazurin as described. 36tatistical Analysis.All statistical analysis was performed using GraphPad Prism version 10.1.1.

Figure 1 .
Figure1.Overview of fatty acid synthesis in F. nucleatum.The initiation and elongation cycles of FAS-II are annotated with the genes encoding FAS-II proteins from F. nucleatum strain ATCC 23726.Highlighted in red is FabX, which is homologous to C4N14_08920; FabX is thought to synthesize unsaturated fatty acids.Thiolactomycin is a commercially available FabF inhibitor, while 681 is an analogue of a known phenylimidazole FabK inhibitor.The corresponding enzyme names are given in TableS1.

Figure 2 .
Figure 2. Gene silencing of F. nucleatum fabK inhibits growth.(A) Antisenses FabKAS1 and FabKAS2 were cloned into F. nucleatum 23726 and induced with indicated concentrations of anhydrotetracycline (ATc) to evaluate effects on growth compared with the empty vector control.Growth was inhibited in an ATc concentration-dependent manner; data are plotted as the mean of three biological replicates; error bars indicate the SEM.(B) RT-qPCR confirmed that FabKAS1 and FabKAS2 blocked the mRNA site targeted by the antisenses.After ATc induction or ethanol (EtOH) control treatment for 1 and 3 h, RNA was collected for RT-qPCR.The data was normalized to the 16S rRNA and fold-change is relative to the uninduced (EtOH) corresponding strain.Each data point is from three biological replicates, with each having two technical replicates, error bars indicate the standard deviation.Statistical analysis was done using a two-way ANOVA on the dCT values, comparing induced to uninduced for each strain and time point.ns, p > 0.05; **, p < 0.01; ****, p < 0.0001.(C) Supplementation with a mixture containing different fatty acids (FAs) did not reinstate growth when the antisenses were induced with ATc (62.5 ng/mL); the FAs were myristic acid (C14), palmitic acid (C16), stearic acid (C18), oleic acid (C18:1Δ9), linoleic acid (C18:2Δ9,12), and linolenic acid (C18:3Δ9,12,15); concentrations of FAs can be found in Table S2; data are plotted as the mean of three biological replicates, and error bars indicate the SEM.

Figure 4 .
Figure 4. Physiological effects of resistance to 681.Mutants exhibiting resistance to 681 were characterized in terms of expression of fabK and fabX and effects on growth.(A) RT-qPCR comparing the transcript levels of fabK and fabX in the wild type and each of the mutants.The data was normalized to 16S rRNA and fold-change is relative to the wild-type strain.The data was obtained from three biological replicates with each having two technical replicates.Statistical analysis was performed using a two-way ANOVA on the dCT values, comparing the mutants to the wild-type strain.ns, p > 0.05; **, p < 0.01.(B) Growth rates of the 681-resistant mutants, compared to the wild-type strain; data is shown as the mean of 7 biological replicates; error bars indicate the SEM.

Table 1 .
Activity of Previously Reported C. difficile FabK Inhibitors against F. nucleatum 23726 and E. faecalis FA2-2 Strains; Results Are from Three Biological Replicates a Shown in parentheses is the previously published nomenclature for the compounds.b F. nucleatum (Fn) 23726.c E. faecalis (Ef) fabK+ possesses only FabK.d Ef fabI+ possesses only FabI.e Ef wild-type possesses both FabI and FabK.

Table 2 .
Activities of 681 and Thiolactomycin (TLM) against F. nucleatum 23726 Transformed with the Empty Vector Control, the Vector Containing fabK or fabX Expressed from the ATc-Inducible Promoter a a Expression was induced with 62.5 ng/mL of ATc (anhydrotetracycline).MICs were determined from three biological replicates.

Table 3 .
Activity of 681 against Inhibitor-Resistant Mutants of F. nucleatum 23726 MICs were determined from three biological replicates.b The C > T mutation occurs 1 bp downstream of the fabK TSS (transcription start site) determined by Ponath et al. 50The 12 bp mutation occurs 99− 111 bp upstream of fabX start codon, inside a hypothetical protein encoded by Fn0663, causing an in-frame deletion of Ser121-Ser124.The full list of mutations occurring during serial evolution in 681 are in Table a

Table 4 .
Activity of 681 against a Panel of Species Isolated from the Digestive Tract a Source information can be found in Table Gene nomenclature for F. nucleatum ATCC 25586 and the corresponding gene name in F. nucleatum ATCC 23726, composition of fatty acid mixture supplemented during bypass experiments, other mutations identified in the 681-resistant F. nucleatum 23726 mutants, strains and corresponding media used in this study; plasmids used in this study, structures of phenylimidazole FabK inhibitors, primers used in this study, alignment of the F. nucleatum 23726 FabK to the C. difficile FabK, schematic overview of cloning strategy for knockdown vector, growth rates of F. nucleatum antisense strains when induced with 62.5 ng/ mL ATc and supplemented with 0.1% Tween-80, alignment of C4N14_08920 to H. pylori FabX, structural homology of C4N14_08920 to H. pylori FabX, fabG transcript levels of 681-resistant mutants, and growth rates of F. nucleatum FabK or FabX overexpression strains (PDF) Julian G. Hurdle − Center for Infectious and Inflammatory Diseases, Institute of Biosciences and Technology, Department of Translational Medical Sciences, Texas A&M Health Science Center, Houston, Texas 77030, United States; orcid.org/0000-0003-2214-8105; Email: jhurdle@tamu.edu