Fragment-Based Lead Discovery Strategies in Antimicrobial Drug Discovery

Fragment-based lead discovery (FBLD) is a powerful application for developing ligands as modulators of disease targets. This approach strategy involves identification of interactions between low-molecular weight compounds (100–300 Da) and their putative targets, often with low affinity (KD ~0.1–1 mM) interactions. The focus of this screening methodology is to optimize and streamline identification of fragments with higher ligand efficiency (LE) than typical high-throughput screening. The focus of this review is on the last half decade of fragment-based drug discovery strategies that have been used for antimicrobial drug discovery.


Introduction
Bacteria evolved and continue to change in response to environmental stressors including antibiotics. Thus, it is worth realizing that drug resistance is unavoidable, necessitating the development of new strategies to address antimicrobial resistance (AMR). Antimicrobial resistance develops via a number of mechanisms in response to both naturally occurring compounds, i.e., antibiotics, and organically synthesized compounds which are generally referred to as antimicrobials. In order to develop resistance, bacteria have developed multiple mechanisms including drug target modification, efflux pumps, hydrolytic enzyme expression, etc., all of which can be shared with other microbes by the various modes of horizontal gene transmission, i.e., transduction, transformation, and conjugation. [1][2][3]. Multiple approaches to combat drug resistance have been developed, in addition to drugs purified from natural product sources, e.g., soil and plants, and the follow up procedures for chemical modification of these products. One of the latest is fragment-based lead generation (FBLG), or fragment-based lead discovery (FBLD). The focus of this review is on the developments in FBLD over the last five years (2018)(2019)(2020)(2021)(2022).
FBLD involves the detection of new inhibitors of known and novel potential drug targets through unique binding site identification, chemistry, or new mechanisms of inhibition that evade existing modes of resistance. FBLD has broad acceptance with a majority of commercial operations utilizing it as part of their generation of lead molecules [4]. The foundational theories of FBLG were implemented in silico around the end of the 1980s [5,6]. However, it wasn't until the mid-1990s with the introduction of structure activity relationship (SAR) by nuclear magnetic resonance (NMR) that the follow up with experimental screening of fragments towards identification of lead model generation gained widespread usage [7,8]. Utilization of SAR-NMR have propelled the selection and construction of libraries to in-depth identification molecules that show promise with regards to the desired activity in a screening assay, i.e., a hit. For background on other antimicrobial-relevant screening assays for hit discovery and hit-to-lead compound selection, qualification, and subsequent optimization strategies, the reader is directed to the following references [9][10][11][12][13].
A range of biochemical, biophysical, and structural assays are employed to identify ligands that could bind to the designated target. These include, but are not limited to NMR, and more specifically, saturation transfer difference (STD NMR), surface plasmon resonance (SPR), crystallography (X-ray), thermal shift analysis (TSA), microscale thermophoresis (MST), and mass spectrometry (MS), biochemical assays. However, MS and biochemical assays have somewhat limited utility as techniques for ligand identification, thus, are usually used more often in situations where the ligand has binding affinity to the target in the 100 µM range. Each of these methods have their advantages and disadvantages, including solubility requirements for both compound and its ligand. This presents a distinct advantage, since together they provide a robust process for the detection of various dynamic ranges of binding affinity. Often the biophysical/biochemical fragment screens are complimented by virtual screens (VS), i.e., screens in silico [10]. Advances supporting the evolution of FBLD methods in the area of antimicrobial development have been elegantly summarized in a recent review [14].
It is the hope that adaption of the new approach to antibiotic design will make the drugs more resistant to microbial inactivation. However, the distribution of antimicrobial resistance (AMR) is not always predictable and is constantly evolving. For example, a chimeric gene, likely the result of fusion a metallo-β-lactamase gene with a partial aminoglycoside resistance conferring sequence, is the novel gene encoding for the New Delhi metallo-β-lactamase, NDM-1 [15]. These ever-changing targets continue to be one of the challenging areas in drug development.
Antibiotics 2023, 12, x FOR PEER REVIEW 2 of 27 qualification, and subsequent optimization strategies, the reader is directed to the following references [9][10][11][12][13]. A range of biochemical, biophysical, and structural assays are employed to identify ligands that could bind to the designated target. These include, but are not limited to NMR, and more specifically, saturation transfer difference (STD NMR), surface plasmon resonance (SPR), crystallography (X-ray), thermal shift analysis (TSA), microscale thermophoresis (MST), and mass spectrometry (MS), biochemical assays. However, MS and biochemical assays have somewhat limited utility as techniques for ligand identification, thus, are usually used more often in situations where the ligand has binding affinity to the target in the 100 μM range. Each of these methods have their advantages and disadvantages, including solubility requirements for both compound and its ligand. This presents a distinct advantage, since together they provide a robust process for the detection of various dynamic ranges of binding affinity. Often the biophysical/biochemical fragment screens are complimented by virtual screens (VS) i.e., screens in silico [10]. Advances supporting the evolution of FBLD methods in the area of antimicrobial development have been elegantly summarized in a recent review [14].
It is the hope that adaption of the new approach to antibiotic design will make the drugs more resistant to microbial inactivation. However, the distribution of antimicrobial resistance (AMR) is not always predictable and is constantly evolving. For example, a chimeric gene, likely the result of fusion a metallo-β-lactamase gene with a partial aminoglycoside resistance conferring sequence, is the novel gene encoding for the New Delhi metallo-β-lactamase, NDM-1 [15]. These ever-changing targets continue to be one of the challenging areas in drug development.
Iminodiacetic acid (IDA) is a novel pharmacophore that has b NDM-1 inhibitor using the fragments guide lead derivatization FBLD from aspergillomarasmine A (AMA), a natural product recognized a inhibitor of NDM-1 (IC50 4-7 μM). In addition, AMA showed clinica stored meropenem anti-Klebsiella pneumoniae NDM-1 activity in a mo [25,26]. The indiscriminate AMAs metal chelating properties along w thesis of AMA's derivatives has led to the identification of IDA as the of AMA. Based on the IC50 (120 μM) of IDA, a fragment-based libra ( Figure 2) and IDA was converted to inhibitor 2 (IC50 8.6 μM, Ki 2.6 ternary complex with NDM-1 [24]. Similarly, using another natural product, captopril, in silico fragm lar design employing a thiol as a metal chelating motif has led to str [27]. These new inhibitors demonstrated inhibition against NDM-1, a tegron-encoded MBL (VIM) and imipenemase (IMP-1). From the lead 3), a library of 17 compounds was prepared. The compounds with the as NDM-1 inhibitors are shown in Figure 3. Using 19 F NMR to vali selected fragments to NDM-1, four fragments, including 3 and 4, wer ity against two human MBL-fold enzymes involved in DNA repair a repair, with enzymes 1A and B showing partial selectivity for ND meropenem alone has an MIC >128 μg/mL against MBL-producing s K. pneumoniae, in combination with fragment 3, meropenem reduced μg/mL, respectively. Interestingly, when tested alone, fragment 3 ha bacterial activity against the two bacterial NDM-1 producing strains a Similarly, using another natural product, captopril, in silico fragment-based molecular design employing a thiol as a metal chelating motif has led to strong MBL inhibitors [27]. These new inhibitors demonstrated inhibition against NDM-1, as well as verona integronencoded MBL (VIM) and imipenemase (IMP-1). From the lead fragment 3 (Figure 3), a library of 17 compounds was prepared. The compounds with the most favorable SAR as NDM-1 inhibitors are shown in Figure 3. Using 19 F NMR to validate the binding of selected fragments to NDM-1, four fragments, including 3 and 4, were tested for selectivity against two human MBL-fold enzymes involved in DNA repair and DNA cross-link repair, with enzymes 1A and B showing partial selectivity for NDM-1binding. While meropenem alone has an MIC >128 µg/mL against MBL-producing strains of E. coli and K. pneumoniae, in combination with fragment 3, meropenem reduced its MIC to 32 and 16 µg/mL, respectively. Interestingly, when tested alone, fragment 3 has no detectable antibacterial activity against the two bacterial NDM-1 producing strains at 100 µg/mL [27]. In addition, these fragments are also inhibitors of Class A and D serine-β-lactamases, albeit weak [27]. In addition to SPROUT [28], other laboratories utilized both Surface Plasmon Resonance (SPR) and STD NMR in searching for a potent NDM-1 inhibitor [29]. Their compound screening used a 122,500-compound library, of which 2500 fragments were obtained from a commercial source, with the rest obtained from an in-house library. Compounds from the latter library that were outside the 150-350 MW fragment range were excluded from the study. After performing high-throughput virtual screening (HTVS) of the library of compounds, followed by SPR validation, 31 fragments were selected based on the data obtained by these two methods, in addition to considering the presence of suitable metal-binding functional groups. This SPR and NMR analysis led to confirmation of the inhibitory activity against NDM-1, albeit weak, of fragment 9 [29]. These findings formed the basis of the synthesis of derivatives of fragment 9 ( Figure 4).

023, 12, x FOR PEER REVIEW
of the inhibitory activity against NDM-1, albeit wea formed the basis of the synthesis of derivatives of fra Figure 3. Fragments guiding lead derivatization of NDM-1 product captopril making use of the de novo molecular de program for constrained structural generation). The synth analysis of a crystal structure of NDM-1 complexed with h [30]. SPROUT [28] was used to identify active sites for in s identified were adjacent to the following: the Lys224 side c philic hydroxide/water that "bridges" the two zinc ions; an hydrophobically interacts with the aromatic ampicillin C6 lactams to metallo β-lactamases. SAR also identified sever against NDM-1 in the submicromolar range (4-8) [27].
Fragments with quinoline (9)(10)(11)(12), naphthalen demonstrated medium to weak binding affinities to binding affinities of fragments 9-12 and 13-16, Fig  groups: methyl, ethyl, and n-propyl contribute simila . Fragments guiding lead derivatization of NDM-1 inhibitors derived from the natural product captopril making use of the de novo molecular design program SPROUT (a computer program for constrained structural generation). The synthesis of fragment 3 [27] is based on the analysis of a crystal structure of NDM-1 complexed with hydrolyzed ampicillin (PDB ID: 3Q6X) [30]. SPROUT [28] was used to identify active sites for in silico-generated fragments. The sites identified were adjacent to the following: the Lys224 side chain; the Zn-2 metal ion; the nucleophilic hydroxide/water that "bridges" the two zinc ions; and a conserved tryptophan (Trp87) that hydrophobically interacts with the aromatic ampicillin C6 side chain [31], crucial for binding of β-lactams to metallo β-lactamases. SAR also identified several other compounds with activity against NDM-1 in the submicromolar range (4-8) [27].
Fragments with quinoline (9-12), naphthalene (21,22,24), and benzene (13)(14)(15)(16) demonstrated medium to weak binding affinities to NDM-1 ( Figure 4). Comparing the binding affinities of fragments 9-12 and 13-16, Figure 4 indicates that the different R groups: methyl, ethyl, and n-propyl contribute similarly to the affinity, while the branched tert-butyl in the A and D series decrease affinity for NDM-1. Several of these synthesized compounds (10, 11 and 22, Figure 4) demonstrate synergistic antimicrobial activity with meropenem against NDM-1 producing K. pneumoniae [29].  An alternative approach that combines in silico screening of a very large library of compounds with STD NMR has been used to identify another promising NDM-1 inhibitor [32]. The authors of this study started from a HTVS of a large library of more than 700,000 putative NDM-1 inhibitors. The targeted library was built from fragments obtained from "sliced" hit molecules. This library, in addition to an in-house untargeted fragment library was screened by STD NMR. From the 37 STD NMR identified hit fragments, 10 molecules were synthesized to confirm the abovementioned strategies. The core structure on which further studies were based is fragment 26 (8-hydroxyquinolone, 8HQ, Figure 5) [32]. Fragment 26 (8HQ), nanomolar a broad-spectrum inhibitor against VIM-2 and NDM-1, was initially identified by fragment-based screening of 31 compounds from a commercially available non-specific metal chelator library of fragments (MW = 120-250) through biochemical assays methods [33]. Subsequently, fragment 26 binding to NDM-1 was demonstrated by STD NMR [32]. Fragment 28, which has fragment 26 as its core structure ( Figure 5), has been identified by HTVS [32]. Compound 30 was derived from the structure of hit compound 28, an HIV-1 integrase inhibitor, which contains two divalent metal cations in its active site [34]. Compound 30 combines fragments 26 and 27, while compound 31 is based on only one identified fragment (26, Figure 5). Further modifications of the initial scaffolds of compounds 30 and 31 gave the corresponding derivatives 32 and 33, respectively ( Figure 5). Compounds 30 and 32 (Ki = 1.1 µM), which combine two high potential fragments, are only marginally better inhibitors of NDM-1 than phenyl analogue 33 (Ki = 2.2 µM). These results indicate that the two-fragment linking strategy is of less benefit in this case; although, a broader analysis may be needed for confirmation. In addition, indenone 29 ( Figure 5), which combines two identified fragments, also demonstrated NDM-1 inhibitory activity (Ki = 4 µM) [32]. Moreover, these four 8-HQ derivatives (39, 31 and 32, 33) can inhibit IMP-1, whereas they are only moderately inhibitory for VIM-2. Compounds 30 and 31, Figure 5, demonstrated synergistic activity together with meropenem (32 µg/mL) against NDM-1 positive E. coli and K. pneumoniae clinical isolates. Furthermore, the meropenem MIC was decreased 64-and 256-fold together with compound 31, in an isolate-dependent manner, in addition to its effect on VIM-2 and IMP-1. MICs of meropenem alone (64 mg/mL) and in combination with inhibitors 31 (0.25 mg/mL, K. pneumoniae; 1.0 mg/mL, E. coli) and 32 (64 mg/mL, K. pneumoniae; 64 mg/mL, E. coli) demonstrated superiority of 31 when tested in vitro.  Figure 5, demonstrated synergistic activity together with meropenem (32 μg/mL) against NDM-1 positive E. coli and K. pneumoniae clinical isolates. Furthermore, the meropenem MIC was decreased 64-and 256-fold together with compound 31, in an isolate-dependent manner, in addition to its effect on VIM-2 and IMP-1. MICs of meropenem alone (64 mg/mL) and in combination with inhibitors 31 (0.25 mg/mL, K. pneumoniae; 1.0 mg/mL, E. coli) and 32 (64 mg/mL, K. pneumoniae; 64 mg/mL, E. coli) demonstrated superiority of 31 when tested in vitro. To identify and validate fragment interaction with NDM-1, a prepared library of 10 compounds, HTVS, NMR screening and Density Functional Theory (DFT) calculation, and biochemical assays were used [32]. The compounds with estimated To identify and validate fragment interaction with NDM-1, a prepared library of 10 compounds, HTVS, NMR screening and Density Functional Theory (DFT) calculation, and biochemical assays were used [32]. The compounds with estimated optimal putative activity, i.e., Ki and per cent inhibition at 50 mg/mL were determined to be inhibitors of NDM-1 (29)(30)(31)(32)(33) and are shown here. Hit molecule 34 was identified through VS as NDM-1 inhibitor [32].

UDP-3-O-acyl-N-acetylglucosamine Deacetylase (LpxC)
A zinc metalloenzyme, LpxC, catalyzes the first committed step in the biosynthesis of lipid A, a toxic but essential component of the Gram-negative outer membrane [35]. Since LpxC does not have a mammalian homologue, different types of its inhibitors (representatives shown in Figure 6) have been developed, as summarized in references [36][37][38]. Reported compounds' studies extensively contain a hydroxamate functionality. The hydroxamate functionality combines with the zinc ion on the active site of the LpxC. One of these compounds, 35 (ACHN-9758, Figure 6) [39], was tested clinically; however, the trials were halted, presumably due to off-target-related side effects [39]. These off-target effects may be linked to the hydroxamate moiety which was shown in HDAC inhibitors to be mutagenic [40], presumably due to the nonspecific binding of many matrix metalloprotease (MMP) inhibitors [41].

UDP-3-O-acyl-N-acetylglucosamine Deacetylase (LpxC)
A zinc metalloenzyme, LpxC, catalyzes the first committed step in the biosynthesis of lipid A, a toxic but essential component of the Gram-negative outer membrane [35]. Since LpxC does not have a mammalian homologue, different types of its inhibitors (representatives shown in Figure 6) have been developed, as summarized in references [36][37][38]. Reported compounds' studies extensively contain a hydroxamate functionality. The hydroxamate functionality combines with the zinc ion on the active site of the LpxC. One of these compounds, 35 (ACHN-9758, Figure 6) [39], was tested clinically; however, the trials were halted, presumably due to off-target-related side effects [39]. These off-target effects may be linked to the hydroxamate moiety which was shown in HDAC inhibitors to be mutagenic [40], presumably due to the nonspecific binding of many matrix metalloprotease (MMP) inhibitors [41]. Therefore, the development of fragment-based compounds with moieties different from hydroxamate is highly desirable. Recently, two types of functionalities that bind in the active site of LpxC to the Zn atom via different modes have been reported by researchers from Taisho Pharmaceutical Co., and Vernalis (R&D) [42]. The 1152 compounds of the Vernalis fragment library [42] were screened against PaLpxC in STD, water-LOGSY, and CPMG ligand-observed NMR experiments. Fragments (252) were identified that bound to different LpxC sites by this methodology. Four fragments showed binding across the NMR screening methods, representatives of which are shown in Figure 7. Therefore, the development of fragment-based compounds with moieties different from hydroxamate is highly desirable. Recently, two types of functionalities that bind in the active site of LpxC to the Zn atom via different modes have been reported by researchers Antibiotics 2023, 12, 315 7 of 28 from Taisho Pharmaceutical Co., and Vernalis (R&D) [42]. The 1152 compounds of the Vernalis fragment library [42] were screened against PaLpxC in STD, water-LOGSY, and CPMG ligand-observed NMR experiments. Fragments (252) were identified that bound to different LpxC sites by this methodology. Four fragments showed binding across the NMR screening methods, representatives of which are shown in Figure 7.
Of the LpxC bound fragments (Figure 7), molecules competitive with hydroxamatecontaining probes (NMR experiments) exhibited dual binding capability by binding to the zinc ion, and the enzyme active site tunnel, as determined by X-ray crystallography. The synthetic efforts have focused on improving the interactions of the new derivatives in both glycine and imidazole fragments (40 and 41, Figure 7). Derivative 42 of the glycine series inhibited LpxC with IC 50 in the nanomolar range (20 nM). However, it did not affect the antibacterial activity ( Figure 7). Enzyme activity was improved to the nanomolar range with the addition of a sulfonyl group (43, Figure 7). The designed interactions with the protein bound to PaLpxC were confirmed by crystallography [42]. Unfortunately, even this potent derivative of the glycine series in the presence of phenylalanine-arginine β-naphthylamide (PAβN) (an efflux pump inhibitor), demonstrated only minimal antimicrobial activity; therefore, no further development of the glycine series was pursued [42]. After chiral separation of 44 and 45 (Figure 7), it was determined that the S enantiomer (>100× enzyme inhibition) has the higher potency. However, the antibacterial activity of S enantiomer 44 against a range of Gram-negative bacteria was eliminated by addition of human serum albumin (HSA). Compounds 47 to 49 were produced upon further examination of the solubilizing group ( Figure 7). Similar levels of enzyme inhibition were produced by these compounds, with better antimicrobial activity, as compared to 44 and 45. Of the compounds tested, 49 had the best antibacterial activity across a range of bacterial species, but 46 is the least affected by HSA [42]. Compound 46 is currently undergoing further optimization and determination of its in vivo efficacy [42].

Botulinum Neurotoxins' Metalloprotease
Clostridium botulinum, strains of C. butyricum and C. baratii, which are anaerobic Gram-positive spore-forming bacilli, produce neurotoxins (NT) [43]. These toxins are excreted as a ~150 kDa single polypeptide chain that is cleaved to a heavy chain (HC, 100 kDa) and a light chain (LC, ~50 kDa) by extracellular proteases. Clinical symptoms of neurotoxicity occur post-appearance in the cytosol of LC metalloprotease activity (Zn 2+ -de- After chiral separation of 44 and 45 (Figure 7), it was determined that the S enantiomer (>100× enzyme inhibition) has the higher potency. However, the antibacterial activity of S enantiomer 44 against a range of Gram-negative bacteria was eliminated by addition of human serum albumin (HSA). Compounds 47 to 49 were produced upon further examination of the solubilizing group ( Figure 7). Similar levels of enzyme inhibition were produced by these compounds, with better antimicrobial activity, as compared to 44 and 45. Of the compounds tested, 49 had the best antibacterial activity across a range of bacterial species, but 46 is the least affected by HSA [42]. Compound 46 is currently undergoing further optimization and determination of its in vivo efficacy [42].

Botulinum Neurotoxins' Metalloprotease
Clostridium botulinum, strains of C. butyricum and C. baratii, which are anaerobic Grampositive spore-forming bacilli, produce neurotoxins (NT) [43]. These toxins are excreted as a~150 kDa single polypeptide chain that is cleaved to a heavy chain (HC, 100 kDa) and a light chain (LC,~50 kDa) by extracellular proteases. Clinical symptoms of neurotoxicity occur post-appearance in the cytosol of LC metalloprotease activity (Zn 2+ -dependent). Inhibiting metalloprotease activity of different NT serotypes with small molecules has been the focus of numerous studies [43][44][45], including the recent exploration of the quinolinol scaffold [46][47][48][49]. FBLD has been employed for evaluating a 24-compound library, having inhibitors comprised of the 8-hydroxyquinoline (8-HQ) scaffold of botulinum NT serotype F. Compounds were chosen according to computational analysis of~800 molecules [50]. Selected compounds per the 24 8HQ-based library (fluorescence thermal shift, FTS) were then tested using an endopeptidase assay. The surface plasmon resonance (SPR) based Proteon™ XPR 36 system was utilized for binding affinity analysis. This was followed by in vivo efficacy analysis in a mouse model. The analysis of FTS and endopeptidase assays led to identification of 3, 8HQ fragments-50 (NSC1011), 51 (NSC1014), and 52 (NSC84094) ( Figure 8) as inhibitors of the NT serotype F. The highest binding affinity was shown by SPR studies of 51 (NSC1014, K D : 5.58 × 10 −6 ) with BoNT/F-LC. Of the fragments screened, 50 (NSC1011) and 51 (NSC1014) appear to have the highest promise as drugs against BoNT/F intoxication [50].
FBLD-based studies have been emerging in the literature for this relatively new antibacterial drug target [53]. Researchers [53] have characterized a library of compounds as inhibitors of the AtlE (enzyme found in S. aureus.) subfamily of the aforementioned enzymes. However, these inhibitors have low water solubility. To improve the physicochemical properties of the initial inhibitor library, a fragment-based library (containing 216, 472 fragments) compiled from several commercial sources was evaluated through virtual screening followed by SPR [53]. From the initial 24 compounds selected, based on the data from the virtual screen, 12 compounds have been validated by SPR. Most of the fragments contain nitrogen heterocycles (di-and tri-azoles) (Figure 9). The best fragment has KD of 228 μM. N-acetyglucosaminidase ligands identified were demonstrated to bind different allosteric sites, which may lead to the preparation of antimicrobials that exhibit a novel mechanism of action.
FBLD-based studies have been emerging in the literature for this relatively new antibacterial drug target [53]. Researchers [53] have characterized a library of compounds as inhibitors of the AtlE (enzyme found in S. aureus.) subfamily of the aforementioned enzymes. However, these inhibitors have low water solubility. To improve the physicochemical properties of the initial inhibitor library, a fragment-based library (containing 216, 472 fragments) compiled from several commercial sources was evaluated through virtual screening followed by SPR [53]. From the initial 24 compounds selected, based on the data from the virtual screen, 12 compounds have been validated by SPR. Most of the fragments contain nitrogen heterocycles (di-and tri-azoles) (Figure 9). The best fragment has K D of 228 µM. N-acetyglucosaminidase ligands identified were demonstrated to bind different allosteric sites, which may lead to the preparation of antimicrobials that exhibit a novel mechanism of action. the data from the virtual screen, 12 compounds have been validated fragments contain nitrogen heterocycles (di-and tri-azoles) (Figure 9) has KD of 228 μM. N-acetyglucosaminidase ligands identified were de different allosteric sites, which may lead to the preparation of antimi a novel mechanism of action.

UDP-N-Acetylglucosamine Enolpyruvyl Transferase (MurA)
A library of small electrophilic fragments, predominately nitroge geting either one of the Cys (Cys 115 or Cys119) residues in the MurA aureus and E. coli, have also been evaluated as warheads (Figure 10), to use reactive groups for binding to poorly conserved amino acids, i. covalent inhibitors [54]. A crucial enzyme targeted in the cytoplas

UDP-N-Acetylglucosamine Enolpyruvyl Transferase (MurA)
A library of small electrophilic fragments, predominately nitrogen heterocycles, targeting either one of the Cys (Cys 115 or Cys119) residues in the MurA active site from S. aureus and E. coli, have also been evaluated as warheads (Figure 10), which are designed to use reactive groups for binding to poorly conserved amino acids, i.e., they are targeted covalent inhibitors [54]. A crucial enzyme targeted in the cytoplasmic biosynthesis of peptidoglycan precursors is MurA. The function of MurA is to catalyze phosphoenolypyruvate (PEP) transfer to UDP-N-acetylglucosamine (UNAG), which releases inorganic phosphate [55]. Inactivation of MurA is demonstrated to weaken bacterial cell walls, thus increasing the risk for osmotic lysis [56].
Antibiotics 2023, 12, x FOR PEER REVIEW 9 of 2 peptidoglycan precursors is MurA. The function of MurA is to catalyze phosphoenolypy ruvate (PEP) transfer to UDP-N-acetylglucosamine (UNAG), which releases inorgani phosphate [55]. Inactivation of MurA is demonstrated to weaken bacterial cell walls, thu increasing the risk for osmotic lysis [56].   Another sulfhydryl reactive moiety, the chloroacetamid porated in a library of 47 compounds, the majority of which nantly by a one-step process ( Figure 12) [57]. Chloroacetamid show selectivity to different targets, regardless of their electr the more common vinylsulfonamides [58]. Most of the chlor IC50 in the 70-140 μM range. In addition, E. coli MurA inhibit most active inhibitor having a low micromolar IC50 [57]. Another sulfhydryl reactive moiety, the chloroacetamide group, has also been incorporated in a library of 47 compounds, the majority of which were synthesized predominantly by a one-step process ( Figure 12) [57]. Chloroacetamides were chosen because they show selectivity to different targets, regardless of their electrophilicity being higher than the more common vinylsulfonamides [58]. Most of the chloroacetamide fragments have IC 50 in the 70-140 µM range. In addition, E. coli MurA inhibitors were identified, with the most active inhibitor having a low micromolar IC 50 [57]. Fragments containing aliphatic rings were determined to be potential inhibitors of MurA, with 64 (containing a primary aliphatic amine) being the most effective inhibitor (39 μM IC50). Most heteroaromatic compounds (e.g., 67-69) and aromatic compounds (e.g., 70) demonstrated either none or at maximum concentration tested (500 μM) a weak inhibition of MurA. Of the aromatic group of compounds, only 68 had activity (287 μM IC50).

Phosphopantetheine Adenylyltransferase (PPAT, Also Known as CoaD)
Coenzyme A (CoA), a cofactor, is essential in the biosynthesis of bacterial membrane lipids, peptidoglycan, teichoic acids, and lipid A [59][60][61]. The next to last step in CoA biosynthesis is catalyzed by PPAT, a hexameric enzyme. While sharing minimal sequence homology with its human ortholog, PPAT shows a high level of sequence homology between bacterial species. This broad spectrum species homology makes PPAT a target for the development of novel antimicrobials [62,63]. Inhibitors of PPAT have been previously

Phosphopantetheine Adenylyltransferase (PPAT, Also Known as CoaD)
Coenzyme A (CoA), a cofactor, is essential in the biosynthesis of bacterial membrane lipids, peptidoglycan, teichoic acids, and lipid A [59][60][61]. The next to last step in CoA biosynthesis is catalyzed by PPAT, a hexameric enzyme. While sharing minimal sequence homology with its human ortholog, PPAT shows a high level of sequence homology between bacterial species. This broad spectrum species homology makes PPAT a target for the development of novel antimicrobials [62,63]. Inhibitors of PPAT have been previously reported [64]. These include 71 and 72 ( Figure 13), synthesized by AstraZeneca, which showed both in vitro and in vivo activity against Gram-positive bacteria [65]. Using an FBLD approach, Novartis identified inhibitors of PPAT of Gram-negative bacteria [66,67]. Fragments with activity bind at the E. coli PPAT 4 -phosphopantetheine site. With nanomolar IC 50 , lead compounds 73 and 74 ( Figure 13) were identified. Unfortunately, 73 and 74 had limited anti-E. coli ∆tolC activity [66]. Additional series optimization resulted in inhibitors of E. coli WT strain PPAT at picomolar concentrations [67].

Cell Wall Components: Lectins
Most pathogens adhere to host tissue, either through biofilm formation, or receptor ligand binding, as part of the colonization process [68]. Lectins, which are carbohydrate Figure 13. AstraZeneca PPAT inhibitors (71,72), and from Novartis lead compounds (72,74), and fragment hit 75. When fragments co-crystalized with E. coli PPAT were examined, 75, a methoxy tryptamine derivative, partly overlapped with 73, although its interaction is different from fragment 73. Fragments 73 and 74 served the basis for development of more than 50 analogs based on with improved on-target potency, including 76, piperidine carbamate. Analysis of these analogs verified that compounds with an AZ benzimidazole core exhibited an enhanced ability to permeate Gram-negative bacteria. These findings eventually resulted in discovery of compounds, e.g., 77, with anti-E. coli WT activity, which was confirmed by multiple methodologies (biochemical, SPR, and MICs). Regrettably, further progression of this series was halted due to bacterial efflux actions [67].

Cell Wall Components: Lectins
Most pathogens adhere to host tissue, either through biofilm formation, or receptorligand binding, as part of the colonization process [68]. Lectins, which are carbohydratebinding proteins, can participate in colonization development since they have a high affinity for mammalian carbohydrates. Therefore, targeting lectins could prove an effective strategy in the prevention and treatment of bacterial and fungal infections [69]. A fragment library of small molecules lacking carbohydrate residues has been developed targeting the propeller lectin BambL from Burkolederia ambifaria [70]. This Gram-negative bacterium causes chronic infections and is multidrug resistant. The early inhibitors developed were based on carbohydrate moieties, such as methyl a-l-fucopyranoside (MeFuc; K D = 1 µM) and complex carbohydrates (H type 2 tetrasaccharide; K D = 7.5 µM) [71]. The principal drawback in the use of carbohydrate-based inhibitors is their mass, which restricts their bioavailability [72]. Using a recently developed library of small molecules, researchers focused their effort on fluorinated scaffolds, since they are used as the primary methods for binding studies of the fragments to BambL via 19 F and T 2 filtered (CMPG) NMR. They accessed the druggability, i.e., whether a drug discovery project progresses from "hit" to "lead", of b-propeller lectins of 350 fluorinated fragments via 19 F and CPMG NMR followed by the computational pocket prediction algorithm SiteMap, SPR, and TROSY (transverse relaxation optimized spectroscopy) NMR [70]. From this analysis, three potential pockets for drug targeting appear present in BambL, the bacterial b-propeller lectin, in addition to possible secondary binding sites in which fragments 78, 79, and 80 ( Figure 14) could be accommodated. Compounds with the strongest effects in SPR, 19F, and TROSY NMR were assessed by TROSY NMR for a dose-dependent binding. The latter method identified compounds fragments 79-85 ( Figure 14, the three best being 79, 80, and 83).

EcDsbA
FBLD has been developed to aid the efforts on the search for inhibitors with high selectivity for EcDsbA [72]. Initial efforts in developing a fluorine containing FBLD library [73], using a combination of X-ray crystallography and NMR, were performed to charac-

EcDsbA
FBLD has been developed to aid the efforts on the search for inhibitors with high selectivity for EcDsbA [72]. Initial efforts in developing a fluorine containing FBLD library [73], using a combination of X-ray crystallography and NMR, were performed to characterize the initial non-peptide EcDsbA inhibitors. It was determined, through the use of HSQC NMR titration, that 6-phenoxy and 6-benzyl analogues were the strongest binders [74]. Latter efforts [75] focused on fragments based on the benzofuran scaffold ( Figure 15) for evaluation and fine tuning of the binding affinity to the EcDsbA. The synthesis of 2-, 5and 6-subsituted benzofuran derivatives, along with their structural characterization and in vitro assessment of EcDsbA inhibition confirmed the high affinity of the benzofuran analogs for EcDsbA with improved in vitro inhibition.
Antibiotics 2023, 12, x FOR PEER REVIEW 13 of 27 Figure 15. Fragment 86 was identified by NMR and X-ray crystallography as an initial analog for scaffold building. The synthesis started from C-2 to access a more polar region of the binding groove. Confirmation of the appropriateness of C-2 as a starting area for binding pocket access was validated by crystal structural analysis of 87, 88, and 89. Subsequent studies using HSQC showed that relative to parent compounds, C-2 analogues exhibited enhanced KD values. In addition, C-2 analogues inhibited DsbA. This also indicates that there is the potential for development of benzofuran analogues, which could target virulence [73].

VcDsbA
Another use for FBLD has been to identify novel leads built on the benzimidazole scaffold. Ideally, these leads would bind (KD of 446 μM) to oxidized VcDsbA via its hydrophobic groove. Interestingly, this benzimidazole fragment has an ~eight-fold selectivity for VcDsbA over EcDsbA. In addition, it has the capability of binding to oxidized Ec-DsbA, (KD > 3.5 mM) ( Figure 12) [76]. Starting with 500 commercially available fragments, using STD NMR analysis, 15 fragments were identified as hits [76]. The binding of these 15 fragments was confirmed in the presence and absence of each fragment by recording 1 H-15 N HSQC spectra of VcDsbA. Fragment 90 (Figure 16), built on the benzimidazole chemotype, had the best affinity for binding in the hydrophobic grove of VcDsbA according to the 1 H-15 N HSQC NMR perturbation in chemical shift. Therefore, fragment 90 was chosen for further SAR studies. To identify the area(s) of fragment 90 responsible for binding to the VcDsbA, seven fragment 90 analogs were synthesized. Of these seven compounds, fragment 91 proved to have the best affinity for the VcDsbA, and it was chosen for further investigation (Figure 16). Despite the weak binding affinity of 91, in a complex with VcDsbA, the NMR model offers a structural basis for its selectivity and provides a Figure 15. Fragment 86 was identified by NMR and X-ray crystallography as an initial analog for scaffold building. The synthesis started from C-2 to access a more polar region of the binding groove. Confirmation of the appropriateness of C-2 as a starting area for binding pocket access was validated by crystal structural analysis of 87, 88, and 89. Subsequent studies using HSQC showed that relative to parent compounds, C-2 analogues exhibited enhanced K D values. In addition, C-2 analogues inhibited DsbA. This also indicates that there is the potential for development of benzofuran analogues, which could target virulence [73].

VcDsbA
Another use for FBLD has been to identify novel leads built on the benzimidazole scaffold. Ideally, these leads would bind (K D of 446 µM) to oxidized VcDsbA via its hydrophobic groove. Interestingly, this benzimidazole fragment has an~eight-fold selectivity for VcDsbA over EcDsbA. In addition, it has the capability of binding to oxidized EcDsbA, (K D > 3.5 mM) ( Figure 12) [76]. Starting with 500 commercially available fragments, using STD NMR analysis, 15 fragments were identified as hits [76]. The binding of these 15 fragments was confirmed in the presence and absence of each fragment by recording 1 H-15 N HSQC spectra of VcDsbA. Fragment 90 (Figure 16), built on the benzimidazole chemotype, had the best affinity for binding in the hydrophobic grove of VcDsbA according to the 1 H-15 N HSQC NMR perturbation in chemical shift. Therefore, fragment 90 was chosen for further SAR studies. To identify the area(s) of fragment 90 responsible for binding to the VcDsbA, seven fragment 90 analogs were synthesized. Of these seven compounds, fragment 91 proved to have the best affinity for the VcDsbA, and it was chosen for further investigation (Figure 16). Despite the weak binding affinity of 91, in a complex with VcDsbA, the NMR model offers a structural basis for its selectivity and provides a prototype for ongoing fragment development [76].

Quorum Sensing-QS
Virulence factor expression can be controlled by chemical signaling molecu the process of quorum sensing (QS) [77]. FBLD has been applied in the disc optimization of 2-aminopyrimidine QS inhibitors, one of the four QS systems by Pseudomonas species. The Pseudomonas Quinolone Signal (PQS) system was since it is specific to Pseudomonas spp. and Burkholderia spp. [78]. The PQS syste alkylquinolones (AQs), rather than the widespread Gram-negative bacteria molecules N-acylated homoserine lactones (AHLs). This system, a transcription controls production of virulence factors including elastase, pyocyanin, and lect The first inhibitor of PqsR, which was based on the natural ligand HHQ, wa upon [78] to improve its poor physicochemical profiles using SPR technology fragment screenings. This approach resulted in the identification of hydroxam and the 2-amino-oxadiazole 99 ( Figure 15) as PqsR inhibitors. However, furthe to expand the two structures were unsuccessful, since these analogs lack activ P. aeruginosa. Application of an enthalpic efficient approach led to fragment 1 17). This was followed by introducing a flexible linker in fragment 105 ( Figure 1 to compound (106, Figure 17) which completely inhibited pyocyanin (107, Figu aeruginosa at nanomolar concentrations. Additional information on recent dev of lead compounds based on FBLD targeting P. aeruginosa is summarized in a r prehensive review [81].

Quorum Sensing-QS
Virulence factor expression can be controlled by chemical signaling molecules during the process of quorum sensing (QS) [77]. FBLD has been applied in the discovery and optimization of 2-aminopyrimidine QS inhibitors, one of the four QS systems employed by Pseudomonas species. The Pseudomonas Quinolone Signal (PQS) system was focused on since it is specific to Pseudomonas spp. and Burkholderia spp. [78]. The PQS system employs alkylquinolones (AQs), rather than the widespread Gram-negative bacterial signaling molecules N-acylated homoserine lactones (AHLs). This system, a transcription regulator, controls production of virulence factors including elastase, pyocyanin, and lectins [79,80]. The first inhibitor of PqsR, which was based on the natural ligand HHQ, was modified upon [78] to improve its poor physicochemical profiles using SPR technology with two fragment screenings. This approach resulted in the identification of hydroxamic acid 98 and the 2-amino-oxadiazole 99 ( Figure 15) as PqsR inhibitors. However, further attempts to expand the two structures were unsuccessful, since these analogs lack activity against P. aeruginosa. Application of an enthalpic efficient approach led to fragment 100 ( Figure 17). This was followed by introducing a flexible linker in fragment 105 (Figure 17), leading to compound (106, Figure 17) which completely inhibited pyocyanin (107, Figure 17) in P. aeruginosa at nanomolar concentrations. Additional information on recent developments of lead compounds based on FBLD targeting P. aeruginosa is summarized in a recent comprehensive review [81]. P. aeruginosa. Application of an enthalpic efficient approach led to fragment 100 ( Fig  17). This was followed by introducing a flexible linker in fragment 105 (Figure 17), lead to compound (106, Figure 17) which completely inhibited pyocyanin (107, Figure 17) aeruginosa at nanomolar concentrations. Additional information on recent developm of lead compounds based on FBLD targeting P. aeruginosa is summarized in a recent c prehensive review [81].
Recently, PurC has been examined as a potential drug target. Specifically, the PurC of Mycobacterium abscessus (Mab) has been the focus since it proves difficult to treat, particularly in individuals with cystic fibrosis. The fact that the bacterial PurC and its human analog PAICS exhibit significant differences, both structurally and functionally, makes PurC a reasonable target for antimicrobial drug development [85][86][87][88]. The FBLD methodology resulted in identification of a new inhibitor class based on 4-amino-6-(pyrazol-4-yl)pyrimidine [89]. This strategy utilized hits detected by high throughput X-ray (XChem, Diamond Light Source) from several fragment libraries (1853 total fragments screened) as starting points for further development [90]. The fragments' potential as inhibitors of PurC in Mab (MabPurC) were validated via different methodologies including differential scanning fluorimetry (DSF), isothermal titration calorimetry (ITC), and X-ray crystallography. Furthermore, these findings demonstrated that MabPurC is essential for Mab, thus justifying this approach for continuing the development of specific inhibitors [89]. Eight fragments were selected for further derivatization. Fragments identified by X-ray crystallography or XChem screening, respectively, to be in complex with MabPurC were 108 and 109 ( Figure 18). The derivatization of both fragment 108 and fragment 109 produced a little over 30 analogs, from which compound 111 (Figure 18) showed an inhibition of MabPurC in the nanomolar range and a very good LE. Compound 112 (Figure 18) demonstrated the best inhibition and LE for MabPurC [89].
Eight fragments were selected for further derivatization. Fragments identified by X-r crystallography or XChem screening, respectively, to be in complex with MabPurC w 108 and 109 ( Figure 18). The derivatization of both fragment 108 and fragment 109 p duced a little over 30 analogs, from which compound 111 ( Figure 18) showed an inhibiti of MabPurC in the nanomolar range and a very good LE. Compound 112 (Figure demonstrated the best inhibition and LE for MabPurC [89].

tRNA (m1 G37) Methyltransferase (TrmD) from Mycobacterium Abscessus (MabTrmD)
TrmD, a member of the SpoU-TrmD (SPOUT) RNA methyltransferase family, w evaluated as a drug target using FBLD [91]. The methyl transferase TrmD, like MabPu is essential for viability in Gram-positive bacteria, e.g., S. aureus, Gram-negative Figure 18. Fragment 108 demonstrated the highest ligand efficiency (LE). Fragment 108 with MabPurC revealed an interaction at the 6-postion of the pyrimidine in the "ribose binding pocket". This binding position appeared to be tolerated since it also was functional 110 binding. Fragment expansion here permits the addition of the pyridine moiety of fragment 109 into fragment 108. However, the introduction of a flexible linker was necessary for linkage of the fragments since they were almost perpendicular to each other. Synthesis of a library of a dozen compounds, where different linkers were explored, led to identification of compound 112 having the best binding affinity and LE [89].
The X-ray studies of fragment hits led to 27 fragments which bind at the Mab TrmD SAM binding pocket. Using the fragment-merging strategy, compounds were further developed to increase affinity binding for MabTrmD (< four-fold) then screened for in vitro anti-Mab activity in both culture and a human macrophage infection model [91]. Several compounds ( Figure 20   The X-ray studies of fragment hits led to 27 fragments which bind at the Mab TrmD SAM binding pocket. Using the fragment-merging strategy, compounds were further developed to increase affinity binding for MabTrmD (< four-fold) then screened for in vitro anti-Mab activity in both culture and a human macrophage infection model [91]. Several compounds ( Figure 20 [99]. Unfortunately, the compounds displayed little activity against Gram-positive and Gram-negative pathogens, e.g., E. coli efflux mutants and H. influenzae (Figure 19) [99]. The scanning fluorimetry (DSF) primary screen, then thermal shift cut-off value (3 S.D. from the negative control) FLBG approach were used for MabTrmD hit optimization [91], screening an initial fragment library of 960 fragments. The X-ray studies of fragment hits led to 27 fragments which bind at the Mab TrmD SAM binding pocket. Using the fragment-merging strategy, compounds were further developed to increase affinity binding for MabTrmD (< four-fold) then screened for in vitro anti-Mab activity in both culture and a human macrophage infection model [91]. Several compounds ( Figure 20) displayed activity against Mtb [93].

Mycobacterium Thermoresistible (MthIMPDH) Inosine-5′-Monophosphate Dehydrogenase (IMPDH)
The first unique step in the synthesis of guanine nucleotides is catalyzed by IMPDH (Figure 1) [100]. This enzyme has been targeted in the development of immunosuppres sive [101], anticancer [102,103], and antiviral drugs, and now antimicrobials, including Mtb [104][105][106][107][108][109]. In addition, FBLD was utilized for developing inhibitors for another my cobacterial enzyme, the IMPDH from Mycobacterium thermoresistible, MthIMPDH [109] Analogous to the screens used against the Mab enzymes, the screen against MthIMPDH involved the fragment library of 960 fragments. From fragment hits from biochemical as says (18 fragments), 6 were studied by X-ray crystallography. These data suggested an approach for optimization via fragment-linking being most suitable for synthetic modifi cations, namely fragment hits 120 and 121 ( Figure 21) [109]. , which compared to 120 and 121 is significantly more potent (1300-fold); compound 121 X-ray structure with shown MthIMPDH at lower right. The racemate 122 has ~50% the Mth IMPDH ΔCBS inhibition as compared to its (S)-isomer, a pattern similar to that previously reported for other IMPDH inhibitors [109][110][111]. However, anti-Mtb H37Rv activity of the most potent analogues (0-100 μM tested) lacked clinically relevant activity (MIC90 ≥ 50 μM). This lack of activity may be the result of poor cell permeability, metabolic instability, and/or efflux [109]. Addi tional information about FBLD developments in the search for anti-Mtb compounds can be found in these recent reviews [112,113].

Threonyl-tRNA Synthetase from Salmonella Enterica
A crucial member of the aminoacyl-tRNA synthetases (aaRS) family is threonyl tRNA synthetase (ThrRS) which catalyzes amino acids attachment to their tRNAs. Inhibi tion of ThrRS has potential utility in the treatment of infections and cancers. Several of the currently known inhibitors are shown in Figure 22. The three substrates of ThrRS, i.e. tRNAThr, ATP, and L-threonine interact with three corresponding pockets on the ThrRS catalytic domain [114]. The usual binding sites for the known aaRS inhibitors are the amino acids' binding sites and/or ATP. Other active sites of ThrRS have been explored recently by utilization of FBLD to develop new aaRS-based inhibitors. These ThrRS dua inhibitors are based on the chlorinated analog halofuginone (HF) (Figure 23) of the natura product febrifugine [115]. The latter is isolated from Dichroa febrifuga Lour, a medicina plant used in traditional Chinese medicine to treat malaria [115]. HF has been demon strated to be an inhibitor of the plant analog of ThrRS, the prolyl-tRNA synthetase (ProRS). Its binding to ProRS differs from most of the inhibitors shown in Figure 23, since , which compared to 120 and 121 is significantly more potent (1300-fold); compound 121 X-ray structure with shown MthIMPDH at lower right. The racemate 122 has~50% the Mth IMPDH ∆CBS inhibition as compared to its (S)-isomer, a pattern similar to that previously reported for other IMPDH inhibitors [109][110][111]. However, anti-Mtb H37Rv activity of the most potent analogues (0-100 µM tested) lacked clinically relevant activity (MIC 90 ≥ 50 µM). This lack of activity may be the result of poor cell permeability, metabolic instability, and/or efflux [109]. Additional information about FBLD developments in the search for anti-Mtb compounds can be found in these recent reviews [112,113].

Threonyl-tRNA Synthetase from Salmonella Enterica
A crucial member of the aminoacyl-tRNA synthetases (aaRS) family is threonyl-tRNA synthetase (ThrRS) which catalyzes amino acids attachment to their tRNAs. Inhibition of ThrRS has potential utility in the treatment of infections and cancers. Several of the currently known inhibitors are shown in Figure 22. The three substrates of ThrRS, i.e., tRNAThr, ATP, and L-threonine interact with three corresponding pockets on the ThrRS catalytic domain [114]. The usual binding sites for the known aaRS inhibitors are the amino acids' binding sites and/or ATP. Other active sites of ThrRS have been explored recently by utilization of FBLD to develop new aaRS-based inhibitors. These ThrRS dual inhibitors are based on the chlorinated analog halofuginone (HF) (Figure 23) of the natural product febrifugine [115]. The latter is isolated from Dichroa febrifuga Lour, a medicinal plant used in traditional Chinese medicine to treat malaria [115]. HF has been demonstrated to be an inhibitor of the plant analog of ThrRS, the prolyl-tRNA synthetase (ProRS). Its binding to ProRS differs from most of the inhibitors shown in Figure 23, since HF binds simultaneously to the ThrRS L-threonine and tRNAThr binding pockets which increases selectivity and activity [116]. HF binds simultaneously to the ThrRS L-threonine and tRNAThr binding pockets which increases selectivity and activity [116]. Thr-AMS, a non-hydrolyzable synthetic derivative of Thr-AMP [117], and its simplified Thr-AMS analog improved specificity and activity [116]. Mupirocin is another inhibitor of aaRSs, i.e., bacterial isoleucyl-tRNA synthetase (IleRS). This naturally occurring antibiotic is used clinically against methicillin-resistant S. aureus (MRSA) and others [118,119]. AN2690, another FDA approved inhibitor, binds to the eukaryotic leucyl-tRNA synthetase (LeuRS) editing site and is clinically used in the treatment of onychomycosis [120,121]. Another naturally occurring inhibitor of ThrRS is borrelidin, an 18-membered polyketide macrolide isolated from Streptomyces species [122], which competes with all three substrates (tRNAThr, ATP, and L-threonine) and has anti-bacterial, anti-fungal, and anti-cancer activity in the nanomolar range [123].
HF occupies the adenosine 76 (A76) and L-proline binding pockets via its halogenated quinazolinone and hydroxypiperidine ring, respectively, to inhibit ProRS [124,125]. HF activity is enhanced by ATP, usually reaching millimolar concentrations in vivo [126,127]. Thus, the inhibitory mechanism of HF presents an attractive avenue for drug development. An HF analogue series were designed and synthesized as ThrRS inhibitors by utilizing the FBLD approach [115].  Thr-AMS, a non-hydrolyzable synthetic derivative of Thr-AMP [117], and its simplified Thr-AMS analog improved specificity and activity [116]. Mupirocin is another inhibitor of aaRSs, i.e., bacterial isoleucyl-tRNA synthetase (IleRS). This naturally occurring antibiotic is used clinically against methicillin-resistant S. aureus (MRSA) and others [118,119]. AN2690, another FDA approved inhibitor, binds to the eukaryotic leucyl-tRNA synthetase (LeuRS) editing site and is clinically used in the treatment of onychomycosis [120,121]. Another naturally occurring inhibitor of ThrRS is borrelidin, an 18-membered polyketide macrolide isolated from Streptomyces species [122], which competes with all three substrates (tRNAThr, ATP, and L-threonine) and has anti-bacterial, anti-fungal, and anti-cancer activity in the nanomolar range [123].
HF occupies the adenosine 76 (A76) and L-proline binding pockets via its halogenated quinazolinone and hydroxypiperidine ring, respectively, to inhibit ProRS [124,125]. HF activity is enhanced by ATP, usually reaching millimolar concentrations in vivo [126,127]. Thus, the inhibitory mechanism of HF presents an attractive avenue for drug development. An HF analogue series were designed and synthesized as ThrRS inhibitors by utilizing the FBLD approach [115].
When anti-E. coli activity of the compounds was determined, 123 and 124 demonstrated the best activity. Although 123 lacked activity against Gram-positive bacterial species (S. aureus, MRSA, Enterococcus faecalis) and P. aeruginosa, 125 exhibited activity against E. coli and S. enterica 87 (16 mg/mL MIC). Crystal structures of SeThrRS (ThrRS of S. enterica 87) with or without 125 indicated that it interacted with the tRNA binding pocket of SeThrRS via a dual-site induced-fit mechanism. However, more studies are needed to improve tRNA-amino acid dual-site inhibition [115]. ated quinazolinone and hydroxypiperidine ring, respectively, to inhibit Pr HF activity is enhanced by ATP, usually reaching millimolar concentr [126,127]. Thus, the inhibitory mechanism of HF presents an attractive av development. An HF analogue series were designed and synthesized as Th by utilizing the FBLD approach [115].

Primase/SSB-Ct Interaction
Bacterial DnaG primase is involved in short RNA primer synthesis, functioning during chromosomal replication to initiate chain extension by replicative DNA polymerase(s). DnaG of E. coli interacts with several proteins including SSB, an ssDNA-binding protein.
SSB is an interaction hub binding >14 proteins participating in DNA replication, repair, and recombination [128,129]. FBLD was utilized to evaluate screening of 1140 structurally diverse fragments [130] by STD-NMR and SPR assays that detected primase/SSB-Ct interaction inhibitors [131]. SPR competition assay initially identified six fragments able to compete with immobilized SSB-Ct peptide. The SPR hits, as well as combinations incorporating the MIPS library (1140 fragments) [130] were analyzed by STD-NMR [131]. Of the screened fragments, 56 exhibited strong intensity difference with an additional 62 showing moderate intensity difference [130]. The concluding STD-NMR of 80 fragments identified~50 fragments for exclusion [130]. Fragments 126-129 ( Figure 24) which had binding affinities in the 1-3 mM range, as determined by NMR titration experiments, were tested further. When anti-E. coli activity of the compounds was determined, 123 and 124 demonstrated the best activity. Although 123 lacked activity against Gram-positive bacterial species (S. aureus, MRSA, Enterococcus faecalis) and P. aeruginosa, 125 exhibited activity against E. coli and S. enterica 87 (16 mg/mL MIC). Crystal structures of SeThrRS (ThrRS of S. enterica 87) with or without 125 indicated that it interacted with the tRNA binding pocket of SeThrRS via a dual-site induced-fit mechanism. However, more studies are needed to improve tRNA-amino acid dual-site inhibition [115].

Primase/SSB-Ct Interaction
Bacterial DnaG primase is involved in short RNA primer synthesis, functioning during chromosomal replication to initiate chain extension by replicative DNA polymerase(s). DnaG of E. coli interacts with several proteins including SSB, an ssDNA-binding protein. SSB is an interaction hub binding >14 proteins participating in DNA replication, repair, and recombination [128,129]. FBLD was utilized to evaluate screening of 1140 structurally diverse fragments [130] by STD-NMR and SPR assays that detected primase/SSB-Ct interaction inhibitors [131]. SPR competition assay initially identified six fragments able to compete with immobilized SSB-Ct peptide. The SPR hits, as well as combinations incorporating the MIPS library (1140 fragments) [130] were analyzed by STD-NMR [131]. Of the screened fragments, 56 exhibited strong intensity difference with an additional 62 showing moderate intensity difference [130]. The concluding STD-NMR of 80 fragments identified ~50 fragments for exclusion [130]. Fragments 126-129 ( Figure  24) which had binding affinities in the 1-3 mM range, as determined by NMR titration experiments, were tested further. Based on the assumption that the tetrazole moiety in 129 would increase lipophilicity and improve membrane permeability, as compared with 126-128) [131], 129 was chosen for further optimization. Although in silico screening identified tetrazole analogs (10) with potentially advantageous binding poses vis a vis SSB-Ct peptide, only 130 showed any Based on the assumption that the tetrazole moiety in 129 would increase lipophilicity and improve membrane permeability, as compared with 126-128) [131], 129 was chosen for further optimization. Although in silico screening identified tetrazole analogs (10) with potentially advantageous binding poses vis a vis SSB-Ct peptide, only 130 showed any binding to DnaGC with a three-fold increase as compared to 129 (15N-1H HSQC spectra; K D = 1.3 mM) [131]. The only available analogue with structural similarity to the ligands 5 (ZINC database) was found to be 131 ( Figure 24) [132]. Based on chemical shift perturbation (CSP) modeling studies, 130 and 131 form electrostatic and hydrogen bond complexes in the binding pockets. A combination of binding to the C-terminal domain of DnaG primase as well as SSB-interacting is a promising start for the development of drugs with long-lasting potential.

Fragment-Based Lead Discovery (FPLD); Cell-Based Screens for the Identification of Microbial Inhibitors of Leishmania, Plasmodium falciparum, Neisseria, Mycobacterium, and Flaviviruses
Multiple methodologies have been utilized to test fragments as potential antimicrobials in clinically appropriate environments. However, few fragments' screens were conducted without any preconceived expectations relative to the putative mode of action [133][134][135]. The potential advantage of FBLD is that it is a target-based approach wherein lipophilicity and selectivity can be controlled for relative to specific targets. There are successful examples of optimization strategies, such as LE in the area of antimicrobial lead development, through which a lead fragment is developed into a drug with clinical potential [136]. The useful metric of LE becomes less of a gold standard when MICs are the principal gauge of antimicrobial activity. In order to get around the issue of optimization of a lead compound as a potential antimicrobial drug candidate, after optimization of the initial fragment hit activity and efficacy against a specific antimicrobial drug target, FPLD was applied to screen M. tuberculosis to identify compounds that exhibit favorable drug properties in a whole cell screen [130] and mouse models [134]. In addition, a whole cell screen has been used for inhibiting Leishmania parasites [135]. FPLD, however, poses a challenge in determining a SAR when a specific target has not been identified, since MIC activity could be the result of off-target effects. This FPLD screen against Leishmania [135] has been expanded to utilization of FPLD against a variety of microorganisms, such as Plasmodium falciparum, Neisseria, Mycobacterium, and flaviviruses [137]. Several illustrative examples of fragments with activity against P. falciparum and Neisseria meningitidis are shown below (Figure 25) [137]. The fragments with activity against P. falciparum identified through a phenotypic screen [137] have provided an interesting example of fast and effective application of FPLD at the start for lead identification. Fragment hit 132 ( Figure 25) from a library of~1600 compounds [137] has high structural similarity and comparable activity to hit 133 ( Figure 25) which has been identified from a library of 500,000 compounds, and later optimized to hit 134, with activity in the nanomolar range. This example implies that fragment libraries used together with phenotypic screens have the capability to detect hits that can be exploited to obtain clinically relevant activity and concentrations. Phenotypic screening could be an accessible alternate approach for the identification of new antimicrobial leads at the start of an FBLD, especially in institutions with limited resources. activity to hit 133 ( Figure 25) which has been identified from a library of 500,000 compounds, and later optimized to hit 134, with activity in the nanomolar range. This example implies that fragment libraries used together with phenotypic screens have the capability to detect hits that can be exploited to obtain clinically relevant activity and concentrations. Phenotypic screening could be an accessible alternate approach for the identification of new antimicrobial leads at the start of an FBLD, especially in institutions with limited resources.

Conclusions
Different approaches to screening for antibacterial leads continue to be employed-from natural product-inspired scaffolds and their derivatization, to making use of the de novo molecular design program SPROUT [24,27] for whole cell fragment screening [137]. In the case of β-lactamase boronate leads [21], it was fortunate that FLBD-guided modifications improved compounds' target affinity, which also resulted in whole cell activity. Moreover, QPX7728 is now considered as an ultrabroad-spectrum inhibitor of serine and metallo-βlactamases. Regretfully, this, as has been seen in many other antibacterial projects, is rarely the case. The latter might be a good reason to prompt the investigation of the activity of more advanced fragments, whose good target specificity and promise for antibacterial activity have been determined solely based on purified protein, to be evaluated by a whole cell assay. This appears to be the important strategy utilized for the for successful optimization of the cyclic boronates as antibacterial candidates, e.g., compound QPX7728 [23].

Future Direction
Recent developments using fragments to probe-identified unique microbial targets may start to address the challenges associated with antimicrobial resistance. In addition, the identification of new drug targets by using fragments in a phenotypic screen has the potential for opening novel avenues for fragment weaponization. Furthermore, these fragments could be doubly functionalized, as has been demonstrated in attempts to expand the pharmacopoeia of anti-cancer drugs that can be used in mammalian systems [138,139]. However, since this approach presents a significant challenge in screening small-molecule libraries, CRISPR/Cas9 technology may offer some helpful alternatives, particularly if it could be adapted to high throughput settings [140]. This gap in identification of novel ligands and targets could also be filled by the use of fully functionalized fragments (FFFs) together with covalent fragment screening, which are nicely summarized in recent reviews [141,142]. Thus, it may be that the primary phenotypic screens are what will guide the "rational" fragment-based discovery of novel antibiotics in the next several years.