Biochemical Pharmacology, Vol. 55, pp. 1541–1549, 1998.
`© 1998 Elsevier Science Inc. All rights reserved.
`
`ISSN 0006-2952/98/$19.00 1 0.00
`PII S0006-2952(97)00684-9
`
`COMMENTARY
`Mechanism of Action of Diazaborines
`Clair Baldock,* Gert-Jan de Boer,† John B. Rafferty,* Antoine R. Stuitje†
`and David W. Rice*‡
`*KREBS INSTITUTE FOR BIOMOLECULAR RESEARCH, DEPARTMENT OF MOLECULAR BIOLOGY AND BIOTECHNOLOGY,
`THE UNIVERSITY OF SHEFFIELD, SHEFFIELD S10 2TN, U.K.; AND †DEPARTMENT OF GENETICS, INSTITUTE OF
`MOLECULAR BIOLOGICAL STUDIES (IMBW), VRIJE UNIVERSITEIT, BIOCENTER AMSTERDAM, DE BOELELAAN 1087,
`1081 HV AMSTERDAM, THE NETHERLANDS
`
`ABSTRACT. The diazaborine family of compounds have antibacterial properties against a range of
`Gram-negative bacteria. Initially, this was thought to be due to the prevention of lipopolysaccharide synthesis.
`More recently, the molecular target of diazaborines has been identified as the NAD(P)H-dependent enoyl acyl
`carrier protein reductase (ENR), which catalyses the last reductive step of fatty acid synthase. ENR from
`Mycobacterium tuberculosis is the target for the front-line antituberculosis drug isoniazid. The emergence of
`isoniazid resistance strains of M. tuberculosis, a chronic infectious disease that already kills more people than any
`other infection, is currently causing great concern over the prospects for its future treatment, and it has
`reawakened interest in the mechanism of diazaborine action. Diazaborines only inhibit ENR in the presence of
`the nucleotide cofactor, and this has been explained through the analysis of the x-ray crystallographic structures
`of a number of Escherichia coli ENR–NAD1– diazaborine complexes that showed the formation of a covalent
`bond between the boron atom in the diazaborines and the 29-hydroxyl of the nicotinamide ribose moiety that
`generates a noncovalently bound bisubstrate analogue. The similarities in catalytic chemistry and in the
`conformation of the nucleotide cofactor across the wider family of NAD(P)-dependent oxidoreductases suggest
`that there are generic opportunities to mimic the interactions seen here in the rational design of bisubstrate
`analogue inhibitors for other NAD(P)H-dependent oxidoreductases.
`BIOCHEM PHARMACOL 55;10:1541–1549,
`1998. © 1998 Elsevier Science Inc.
`
`KEY WORDS. enoyl reductase; diazaborine; isoniazid; lipid biosynthesis; bisubstrate analogues
`
`Diazaborines represent a group of antibacterial drugs of
`which the important structural element is a heterocyclic
`1,2-diazine ring containing a boron as a third hetero atom
`(Fig. 1). Although the antimicrobial activity of these
`compounds was first described in the late 1960s, their
`biological target remained obscure until the 1980s. The first
`clue that diazaborines may interfere with membrane bio-
`synthesis was obtained by Hogenauer and Woisetschlager
`[1]. Using specific mutant strains of Escherichia coli and
`Salmonella typhimurium, they showed that diazaborine in-
`hibits the incorporation of radioactive galactose into the
`LPS§ of these bacteria. This result fitted nicely with earlier
`observations that the antibacterial activity of diazaborines
`is confined almost exclusively to Gram-negative bacteria,
`indicating that they specifically inhibit LPS synthesis,
`which is an integral part of the outer membrane of this
`group of bacteria. This notion triggered the search for new
`diazaborine derivatives and analogues in the treatment of
`bacterial infections, despite the inherent toxicity of boron-
`containing compounds [2].
`
`‡ Corresponding author. Tel. 44-114-222-4242; FAX 44-114-272-8697.
`§ Abbreviations: LPS, lipopolysaccharide; ACP, acyl carrier protein;
`ENR, enoyl ACP reductase; and FAS, fatty acid synthetase.
`
`STRUCTURE–ACTIVITY RELATIONSHIPS
`The schematic structure for compounds that are generally
`referred to as diazaborines is given in Fig. 1A. The more
`systematic name for these compounds is 1,2-dihydro-1-
`hydroxy-2-(organosulfonyl)-areno[d][1,2,3]diazaborines
`(arene 5 benzene, naphthalene, thiophene, furan, pyrrole).
`Systematic syntheses of these compounds, by a reaction of
`(organosulfonyl)hydrazones of arene aldehydes or ketones
`with tribromoborane in the presence of ferric chloride, were
`first described by Grassberger et al. [2]. In this study, the
`activities of approximately 80 different diazaborine deriva-
`tives against bacteria in vitro and in vivo (E. coli septicae-
`mia) were determined. Although, in general, thieno-diaza-
`borines were found to be the most potent inhibitors,
`followed by benzo-diazaborines and furo-diazaborines,
`whereas pyrrolo-diazaborines were totally inactive (Fig.
`1B), this classification oversimplifies the extensive data of
`Grassberger et al. [2].
`To facilitate a more comprehensive understanding of
`the significance of the organosulfonyl side chain, the
`arene group, and various substitutions on this group in
`relation to antibacterial activity of diazaborines, we have
`summarized the most relevant data, in this respect, in
`Table 1.
`
`CFAD v. Anacor, IPR2015-01776
`ANACOR EX. 2009 - 1/10
`
`

`
`1542
`
`C. Baldock et al.
`
`FIG. 1. (A) Structural formulae of a number of different classes of diazaborine; and (B) the MIC values for the best inhibitors in each
`class of diazaborine 1A, 1B, 2, 3 and 4 in an in vivo antibacterial assay against E. coli [2]. MIC 5 minimum inhibitory concentration.
`
`Importance of the Organosulfonyl Side Chain (SO2R2)
`As depicted in Table 1, irrespective of the nature of the
`arene group, diazaborines with propylsulfonyl side chains
`show the highest antibacterial activity. Decreasing the
`chain length of the alkyl group greatly reduces activity
`(Table 1, No. 6 vs No. 7 or No. 11 vs No. 13). Derivatives
`with benzylsulfonyl side chains show considerable activity,
`whereas substitutions on the benzene ring are generally not
`favourable (No. 9 vs No. 8 and 10).
`
`Importance of the Arene Group (X)
`Irrespective of the nature of the organosulfonyl group,
`thieno-diazaborines are more potent than benzo-diazabo-
`rines (No. 1 vs No. 11, No. 12 vs No. 3 and 10). In the
`thieno-diazaborine series, however, the thieno[2,3-d]diaza-
`borines (No. 1– 4) are generally slightly more active than
`their thieno[3,2-d] counterparts (No. 5–10).
`Substitution of a methyl for a hydrogen in position 6
`(Fig. 1) of thieno-diazaborine significantly increases biolog-
`ical activity (e.g. No. 6 vs No. 5), whereas replacement of
`a methyl by bromine has little effect, with the exception of
`the 6-bromo derivative of thieno[2,3-d]diazaborine, which
`is totally inactive (e.g. No. 4 vs No. 9 and 10). Substitution
`of hydrogen in position 7 is probably not advisable, since
`replacement with bromine resulted in a complete loss of
`biological activity (not shown).
`Benzo-diazaborines are generally less active than thieno-
`diazaborines, and substitutions by methyl or halogen (F, Cl,
`Br) on the benzene ring have no marked influence on the
`bacterial activities in vitro. In general, benzo-diazaborine
`derivatives with methyl or halogen in position 5 or 7 (Fig.
`
`1) are less active than the unsubstituted parent compound
`(No. 14; not shown). Substitution (methyl, halogen) in
`position 6 slightly increases antibacterial activity (No. 14 vs
`No. 12). Substitution with polar groups (OH, NH2, NR2,
`NHCOCH3, COOH) in position 6 or 7 generally leads to
`a complete loss of activity (not shown).
`Only a few furo- and pyrrolo-diazaborines (Fig. 1) were
`analysed in the work of Grassberger et al.
`[2]. While
`pyrrolo-diazaborines were found to be completely inactive,
`2 out of the 3 furo-diazaborines tested showed substantial
`antibacterial activity. Both biologically active furo-diazabo-
`rines have a tosyl group attached to the sulfonyl moiety
`(Table 1, No. 15), and a methyl or bromine substitution in
`position 6. Although alkylsulfonyl derivatives were not
`included in the study, it seems very likely that substitution
`of a propyl moiety instead of a tosyl moiety will further
`enhance the biological activity of furo-diazaborines.
`
`BORON-FREE ANALOGUES
`A major problem with diazaborines is their inherent toxic
`potential, which is probably due to the arenoboronic acid
`amide moiety [2]. To help design boron-free analogues,
`Grassberger et al. [2] tried to address the question as to
`whether the bicyclic areno-diazaborines themselves are the
`active species or whether hydrolytic cleavage at the BN
`bond to give the corresponding (dihydroxy)arenes is essen-
`tial
`for biological activity of the compounds. For this
`purpose they synthesized the carbacyclic analogue of benzo-
`diazaborine (No. 14). This isoquinoline (Fig. 2A), how-
`ever, was inactive in all test systems. On the other hand,
`the stable N-methyl derivative (Fig. 2B), which served as a
`
`ANACOR EX. 2009 - 2/10
`
`

`
`Mechanism of Action of Diazaborines
`
`1543
`
`TABLE 1. Antibacterial activity of selected diazaborines*
`
`Structure
`
`No.
`
`ED50
`(mg/kg)
`
`MIC† values (mg/mL)
`E. coli E. aerogenes S. typhimurium K. pneumoniae P. mirabilis N. gonorrhoeae
`
`2
`
`4
`
`9
`
`1.25
`
`1.56
`
`2.5
`
`5
`
`6.25
`
`25
`
`0.19
`
`0.31
`
`1.56
`
`0.31
`
`0.31
`
`1.25
`
`1.56
`
`1.25
`
`0.5
`
`0.5
`
`1.25
`
`0.25
`
`.300
`
`.50
`
`.50
`
`.50
`
`.50
`
`.50
`
`.8
`
`4.5
`
`1.56
`
`3.12
`
`28
`
`6.25
`
`25
`
`0.78
`
`3.12
`
`0.39
`
`0.78
`
`1.56
`
`25
`
`1
`
`8
`
`113
`
`.50
`
`.50
`
`.50
`
`.50
`
`.50
`
`.8
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`49
`
`3.12
`
`12.5
`
`9
`
`;20
`
`3.12
`
`12.5
`
`10
`
`;15
`
`6.25
`
`11
`
`73
`
`6.25
`
`12
`
`;15
`
`12.5
`
`13
`
`42
`
`14
`
`;25
`
`25
`
`25
`
`25
`
`10
`
`50
`
`.50
`
`.50
`
`15
`
`;10
`
`12.5
`
`25
`
`0.78
`
`0.78
`
`3.12
`
`2.5
`
`3.12
`
`0.31
`
`0.78
`
`1.56
`
`1.25
`
`1.56
`
`0.39
`
`1.56
`
`1.56
`
`1.56
`
`3.12
`
`0.5
`
`1
`
`0.5
`
`2
`
`1
`
`12.5
`
`6.25
`
`25
`
`.8
`
`6.25
`
`3.12
`
`3.12
`
`12.5
`
`1.56
`
`3.12
`
`2
`
`1
`
`*Data excerpted from Ref. 2. Reprinted with permission from J Med Chem 27: 947–953, 1984. © 1984 American Chemical Society.
`†MIC 5 minimum inhibitory concentration that inhibited visible growth of bacteria in trypticase soy broth (in mg/mL). Genus and species not specified in text: E. aerogenes,
`Enterobacter aerogenes; K. pneumoniae, Klebsiella pneumoniae; and P. mirabilis, Proteus mirabilis.
`
`ANACOR EX. 2009 - 3/10
`
`

`
`1544
`
`C. Baldock et al.
`
`FIG. 2. Structural formulae of: (A) the car-
`bacyclic analogue of benzo-diazaborine; and
`(B) the ring-opened derivative of benzo-
`diazaborine.
`
`stable ring-opened analogue of benzo-diazaborine (No. 14),
`showed some biological activity on E. coli, but was at least
`ten times less active than the corresponding diazaborine.
`Although these results seem to support to some extent the
`hypothesis that ring opening is essential
`for biological
`activity, no firm conclusions could be made.
`
`CELLULAR TARGET OF DIAZABORINES
`The antibacterial activity of diazaborines is confined almost
`exclusively to Gram-negative bacteria. A rationale behind
`this observation was first presented by Hogenauer and
`Woisetschlager [1], who demonstrated that a specific thi-
`eno-diazaborine (Table 1, No. 5) inhibited LPS biosynthe-
`sis in both E. coli and Salmonella. In a later paper of
`Turnowsky et al. [3], it was shown that the primary target of
`inhibition was not LPS biosynthesis itself, but rather an
`earlier step in lipid A precursor biosynthesis, such as de novo
`fatty acid biosynthesis or the acyl transfer to the UDP-N-
`acetylglucosamine moiety. In addition, these authors car-
`ried out a thorough molecular genetic analysis of diazabo-
`rine resistance in both E. coli and Salmonella and demon-
`strated that resistance results from a point mutation in the
`envM genes of these bacteria. In an earlier study, the envM
`gene of E. coli was identified as an essential gene, since an
`allelic mutant form (envM392) results in a temperature-
`sensitive growth phenotype [4]. In fact, the similarity of the
`effects on E. coli cells seen after treatment with diazaborine
`or by incubating a temperature-sensitive envM mutant at
`
`42° strongly suggested that the envM protein was the actual
`target of the drug.
`
`BIOCHEMICAL TARGET OF DIAZABORINES
`Despite the demonstration that both inhibition of wild-
`type E. coli cells by diazaborine and shifting the envM392
`mutant to the nonpermissive temperature result in imme-
`diate cessation of fatty acid biosynthesis, the actual func-
`tion of the envM protein in fatty acid biosynthesis re-
`mained obscure until more recently. The notion that the
`envM protein might actually be the ENR component of the
`bacterial fatty acid synthetase (FAS II) was only recognized
`a few years ago on the basis of amino acid sequence
`homology with the purified ENR protein and the corre-
`sponding cDNA sequence from oilseed rape (Brassica na-
`pus) [5, 6]. ENR catalyses the last reductive step in the
`cyclic process of fatty acid elongation, as depicted in Fig. 3.
`Direct evidence that the E. coli envM gene encodes a
`diazaborine-sensitive ENR was provided independently by
`two laboratories in 1994 [7, 8]. In addition, it was estab-
`lished that the envM392 (ts) allele encoded an extremely
`temperature-sensitive ENR and that diazaborine is a spe-
`cific inhibitor of this E. coli enzyme [8]. The enzyme studies
`also showed that NAD1 is required as a cofactor for both
`the inhibition and the binding of diazaborine to the ENR
`enzyme [8]. Based on the fact that the envM gene encodes
`the ENR component of the E. coli fatty acid synthetase, this
`gene was recently renamed fabI [7].
`
`FIG. 3. Schematic representation of the fatty
`acid synthetase cycle.
`
`ANACOR EX. 2009 - 4/10
`
`

`
`Mechanism of Action of Diazaborines
`
`1545
`
`FIG. 4. Protein sequence alignment of E. coli, B. napus (rape
`seed) and M. tuberculosis ENRs in the region of the Gly93Ser
`and Ser94Ala mutations, associated with diazaborine and isoni-
`azid resistance in E. coli and M. tuberculosis ENR, respectively.
`
`MOLECULAR GENETICS OF
`DIAZABORINE RESISTANCE
`The demonstration that ENR is the target of diazaborine
`and the observation that all diazaborine-resistant mutants
`isolated thus far have the same Gly93Ser (see Fig. 4) amino
`acid substitution in this gene justify the conclusion that the
`primary effect of diazaborine is on core fatty acid biosyn-
`thesis, whereas the effects on,
`for example, membrane
`integrity and LPS biosynthesis are of a pleiotropic nature
`[8]. It was also shown that both the rape seed ENR and the
`equivalent E. coli enzyme encoded by the diazaborine-
`resistant allele are insensitive to the drug [8]. In this
`respect,
`it is important to realize that both rape seed
`plantlets and E. coli strains with a resistant fabI allele are
`still sensitive to diazaborine at concentrations above 20 and
`200 mg/mL, respectively (Stuitje AR, unpublished observa-
`tion). This observation suggests that other targets for
`diazaborine may exist in these organisms.
`The insensitivity of the plant ENR towards diazaborine
`has facilitated gene replacement experiments, demonstrat-
`ing that the coding sequence of the essential E. coli fabI
`gene can be replaced by a cDNA sequence encoding rape
`seed ENR. Although the resulting E. coli strain shows
`slightly different growth characteristics and membrane fatty
`
`acid composition, it is viable under laboratory conditions,
`demonstrating that plant ENR can functionally replace its
`counterpart in the bacterial multi-enzyme FAS system [8].
`
`SENSITIVITY OF ENR TO OTHER DRUGS
`Recently, it has become evident that besides diazaborines,
`ENR is a potential target for at least two other drugs (see
`Fig. 5), the bleaching herbicide diflufenican [N-(2,4-
`difluorophenyl)-2-[3-(trifluoromethyl)phenoxy]-3-pyridin-
`ecarboxyamide] and the antituberculosis drug isoniazid
`[INH; isonicotinic acid hydrazide].
`Although the primary mode of action of diflufenican is
`on carotenoid biosynthesis, with phytoene synthase being
`the main target, it was also reported to inhibit plant fatty
`acid synthetase in vitro [9]. In fact, Ashton et al. [10]
`demonstrated that both plant and E. coli ENR are also a
`target for diflufenican. Although it has structural similari-
`ties, in part, with pyridine nucleotides, the mode of action
`of diflufenican is still obscure, since it is clearly not a
`general
`inhibitor of pyridine nucleotide-dependent en-
`zymes. The NADPH-dependent b-keto reductase compo-
`nent of FAS, for example, is not inhibited by diflufenican
`[10].
`Isoniazid has been used since 1952 as one of the most
`effective drugs for the treatment and prophylaxis of tuber-
`culosis. A single missense mutation in the inhA gene of
`Mycobacterium tuberculosis can confer resistance to this
`drug. Based on the similarities of the corresponding inhA
`protein to E. coli (40%) and rape seed ENR (37%), it was
`demonstrated recently that this gene also encodes an
`NADH-specific ENR [11]. Despite the fact that the crystal
`structure of the inhA protein is solved and that the
`mutation that leads to resistance to isoniazid, Ser94Ala
`(Fig. 4), maps to a region close to the nucleotide binding
`site [11], little is known about the mode of action of
`isoniazid at the molecular level. Further mechanistic studies
`on enzyme inhibition by isoniazid have been seriously
`hampered, because it is now recognized that isoniazid is a
`prodrug that is activated by mycobacteria to an as yet
`
`FIG. 5. Structural formulae of diflufenican and
`isoniazid.
`
`ANACOR EX. 2009 - 5/10
`
`

`
`1546
`
`C. Baldock et al.
`
`FIG. 6. (A) Schematic diagram of the ENR tetramer with the cofactor bound. The ribbon trace of E. coli ENR is shown coloured in
`red, white, green, and cyan for each subunit, and the NAD1 is shown in an all-atom representation, coloured by atom (produced using
`MIDAS [14]). (B) Schematic diagram of a single subunit of the ENR–NAD1 complex. The ribbon trace of E. coli ENR is shown
`coloured in rainbow colours from red at the N-terminus to blue at the C-terminus, and the NAD1 is shown in an all-atom
`representation, coloured by atom (produced using MIDAS [14]). (C) Photograph of an ENR–NAD1– diazaborine crystal. (D)
`Schematic diagram of a single subunit of the ENR–NAD1–thieno-diazaborine complex. The ribbon trace of E. coli ENR is shown in
`green; the NAD1 and diazaborine are shown in an all-atom representation and coloured cyan and blue, respectively. The loop that
`orders on diazaborine binding is highlighted in red (produced using MIDAS [14]). (E) Active site of the ENR–NAD1–thieno-
`diazaborine complex. The Ca backbone trace is shown in cyan, with the NAD1 and diazaborine shown in an all-atom representation
`and coloured by atom (produced using MIDAS [14]). (F) Fourier map of the NAD1–thieno-diazaborine complex at 2.2 Å resolution
`with the final refined structure superimposed (produced using BOBSCRIPT [15]).
`
`unknown species. This activation process is thought to
`involve the catalase-peroxidase KatG [12].
`
`molecular basis by which diazaborine inhibits the E. coli
`enzyme.
`
`MOLECULAR BASIS OF
`DIAZABORINE ACTION
`Given the importance of ENR as a drug target, recent
`crystallographic studies have sought to understand the
`
`Structure and Mechanism of ENR
`E. coli ENR is a homo-tetramer of subunit Mr of approxi-
`mately 28,000 and, following crystallization [13], its struc-
`ture was solved to 2.1 Å resolution (Fig. 6A), by a
`
`ANACOR EX. 2009 - 6/10
`
`

`
`Mechanism of Action of Diazaborines
`
`1547
`
`Structural Studies on the E. coli
`ENR–NAD1–Diazaborine Complexes
`To understand the mode of diazaborine action, crystals of
`two E. coli ENR–NAD1– diazaborine complexes were pre-
`pared (Fig. 6C) [in the presence of a thieno (No. 5) and a
`benzo-diazaborine (No. 14)], and the structures were solved
`independently by molecular replacement using the struc-
`ture of E. coli ENR–NAD1 as a search model to 2.2 and 2.6
`Å, respectively [17]. Unlike the situation for the ENR–
`NAD1 complex, in both the diazaborine complexes the
`loop between Leu195 and Met206 is well defined (Fig. 6D)
`and provides two residues, Ile200 and Phe203, whose side
`chains are in van der Waals’ contact with the non-boron-
`containing 5- and 6-membered rings of the thieno- and
`benzo-diazaborines, respectively.
`
`Binding of Diazaborine
`Analysis of the diazaborine binding sites showed that both
`the diazaborine compounds bind in a closely related man-
`ner, adjacent to the nicotinamide ring of the cofactor, in a
`pocket formed by the side chains of Tyr146, Tyr156,
`Met159, Ile200, Phe203, Leu100, Lys163, and the main
`chain peptide between Gly93 and Ala95. The bicyclic rings
`of the diazaborines form a face-to-face interaction with the
`nicotinamide ring, allowing the formation of extensive
`p–p stacking interactions with additional van der Waals’
`interactions between the rings and the side chains of
`Tyr156, Tyr146, Phe203, and Ile200. The major difference
`between the binding of the two diazaborines is that their
`respective tosyl and propyl groups occupy subtly modified
`positions. The tosyl moiety lies perpendicular to the bicy-
`clic ring and interacts with the main chain peptide between
`Gly93 and Ala95 and the side chain of Leu100, whereas the
`propyl moiety folds back onto the planar bicyclic ring
`system in a manner reminiscent of a scorpion’s tail and
`forms interactions with the side chain of Met159 and
`Ile200 and the main chain peptide of both Gly93 and
`Phe94 (Fig. 6E). Additional interactions made by both
`drugs include hydrogen bonds between the boron hydroxyl
`and the phenolic hydroxyl of Tyr156 and between a
`nitrogen in the boron-containing ring and an ordered
`solvent molecule.
`
`MECHANISM OF DIAZABORINE RESISTANCE
`Previous studies have shown that a Gly93Ser mutation in
`E. coli ENR leads to diazaborine resistance [3, 8] and
`analysis of the structure shows that the alpha carbon of
`Gly93 lies close to the sulphonyl group of the diazaborine.
`Modelling studies show that in the absence of changes to
`the main-chain torsion angles in the Gly93Ser mutant, the
`Cb atom of the serine side-chain would be unacceptably
`close to the two oxygens of the sulphonyl group of the
`diazaborine (2.1 and 2.6 Å, respectively). Therefore, resis-
`tance to diazaborine can be explained by the serine side-
`
`FIG. 7. Proposed catalytic mechanism of reduction of the double
`bond in an enoyl substrate by ENR.
`
`combination of isomorphous replacement and molecular
`replacement using the B. napus ENR structure (Protein
`Data Bank entry 1ENO; [16]) as a search model [17]. A
`notable feature in the structure of this enzyme is a region of
`10 amino acid residues from Leu195 to Met206 that forms
`a disordered loop between strand b6 and helix a6, which
`borders the nucleotide binding site (Fig. 6B). In the B.
`napus structure, residues in the equivalent region (residues
`Ala240 to Thr249) have high temperature factors and have
`been implicated in substrate binding [16]. Therefore, this
`disorder may reflect the fact that the acyl substrate is not
`present in the crystals of the E. coli enzyme.
`The ENR subunit comprises a single domain of dimen-
`sions 45 3 48 3 50 Å and is formed from seven b-strands
`(b1–b7), creating a parallel b-sheet with seven flanking
`a-helices (a1–a7). The b-sheet is flanked on one side by
`helices a1, a2, and a7 and on the other by helices a3, a4,
`and a5 with a6 sitting along the “top edge” of the b-sheet
`above the COOH-terminal ends of strands b6 and b7. This
`fold is highly reminiscent of the Rossmann fold commonly
`found in dinucleotide binding enzymes [18]. The cofactor is
`bound in a similar and extended conformation to that
`observed in other dehydrogenases at the COOH-terminal
`end of the b sheet with the nicotinamide ring lying deep in
`a pocket on the enzyme surface.
`A plausible mechanism for the catalytic activity of ENR
`has been described previously [16]. This involves the
`hydride transfer from the C-4 position of the NADH to the
`C-3 position at the double bond in the enoyl substrate. This
`leads to the formation of an enolate anion intermediate
`that can then be protonated on the oxygen to form an enol;
`subsequent tautomerization of the enol would then lead to
`the production of the reduced acyl product. Residues
`implicated in this mechanism include Lys163 whose role is
`thought to be to stabilize the positive charge of the
`transition state and Tyr156, thought to be the proton donor
`to the enolate anion (Fig. 7). Both of these residues are
`conserved in the sequences of the ENRs from E. coli, M.
`tuberculosis, B. napus, Anabaena spheroides, and Haemophilus
`influenzae.
`
`ANACOR EX. 2009 - 7/10
`
`

`
`1548
`
`C. Baldock et al.
`
`chain of the mutant encroaching into the drug-binding site
`and causing severe steric hindrance. In B. napus ENR, the
`equivalent residue to Gly93 is an alanine, and the B. napus
`enzyme is insensitive to diazaborine, presumably because a
`steric clash with the methyl group of the alanine prevents
`diazaborine from binding. In M. tuberculosis ENR, the
`Ser94Ala mutation, which leads to resistance to isoniazid
`[19], maps to a similar region of the structure as Gly93 in E.
`coli ENR. However, in contrast to the evidence supporting
`the steric mechanism responsible for diazaborine resistance
`in E. coli ENR, Dessen and coworkers [11] proposed that
`the serine to alanine mutation in M. tuberculosis ENR
`causes the disruption of a hydrogen-bonding network asso-
`ciated with the cofactor binding, which leads to a lower
`affinity for the cofactor and a consequent reduction in the
`binding of the isoniazid-derived inhibitor.
`
`MECHANISM OF DIAZABORINE ACTION
`Analysis of the drug complex showed that the distance
`between the boron atom of the diazaborine and the 29OH
`of the nicotinamide ribose was approximately 1.7 Å, com-
`parable with a B—O covalent bond length of 1.6 Å. The
`quality of the electron density map (particularly for the
`thieno-diazaborine complex at 2.2 Å) implies that the
`errors in coordinates are very small, and thus the interac-
`tion between these two atoms is covalent. This is further
`supported by continuous electron density between the
`29OH of the nicotinamide ribose and the boron and the
`unambiguous identification of the position of the hydroxyl
`oxygen to which the boron is linked, which can be seen to
`form part of a tetrahedral, rather than a trigonal, arrange-
`ment as required if the boron forms four covalent bonds
`(Fig. 6F). This finding provides a clear explanation for the
`strong inhibitory properties of the diazaborines and for the
`requirement of NAD1 for diazaborine binding. The forma-
`tion of a covalent bond by the boron atom in diazaborines
`is similar to the mechanism of the boronic acid inhibitors of
`serine proteases, which chemically modify the active site
`serine to give a tetrahedral adduct [20].
`The position of the aromatic bicyclic ring of the diaza-
`borine above the nicotinamide ring strongly resembles the
`proposed model for the binding of the enoyl substrate
`suggested from studies on B. napus ENR [16] with the
`proposed position for the negatively charged oxygen of the
`enolate anion of the substrate close to that of the boron
`atom in the drug. The amino group of the putative catalytic
`lysine (Lys163) of ENR is only 4.1 Å from the boron atom,
`and, therefore, this residue may afford partial stabilization
`of the negatively charged boron, in a manner similar to its
`proposed role in the stabilization of the enolate anion
`during catalysis. Thus, it is clear that the inhibitor action of
`the diazaborines derives,
`in part,
`from their structural
`resemblance to the substrate of the enzyme.
`The formation of a covalent bond between the enzyme-
`bound NAD1 and diazaborine generates a tight, non-
`covalently bound bisubstrate analogue. In this respect,
`
`diazaborines are similar to inhibitors of pyridoxal phos-
`phate-containing enzymes (e.g. gabaculine), which co-
`valently modify the cofactor [21]. The best analogy is
`perhaps with 5-fluoro-2-deoxyuridylic acid, which acts as a
`potent inhibitor of thymidylate synthase [22]. This inhibi-
`tor also modifies one of the substrates (the deoxynucleo-
`side) to covalently modify the other substrate (methylene
`tetrahydrofolate) to form a bisubstrate analogue. Potent
`bisubstrate inhibitors of other enzymes with nucleotide
`substrates have also been described (for example, the
`polyoxin inhibitors of chitin synthase [23] and various
`synthetic inhibitors of protein kinase C [24]). Hitherto, no
`such good bisubstrate inhibitors have been described for
`NAD(P)-dependent oxidoreductases. Examples of this type
`designed to inhibit 3-hydroxy-3-methyl glutaryl CoA re-
`ductase were only very weak inhibitors of cholesterol
`biosynthesis, possibly not only because of the lack of a
`moiety to mimic the adenosine diphosphoribose but also,
`quite probably, because of steric problems in the active site
`associated with the nature of the linkage that utilized the
`C4 atom of the nicotinamide ring [25]. Therefore, an
`important feature of the diazaborine study is that it indi-
`cates the type of linkage that might be used to create a
`bisubstrate analogue with the necessary geometry to occupy
`the active site cleft within a pyridine nucleotide-dependent
`enzyme. This may prove crucial in the design of a new
`generation of antibacterial agents against a range of drug-
`resistant organisms including M. tuberculosis. Furthermore,
`the broad spectrum antibacterial activity of diazaborines
`against organisms such as Enterobacter, Neisseria gonor-
`rhoeae, Proteus, and Salmonella [2] suggests that bisubstrate
`analogues designed from the structure/activity profiles of
`the diazaborines may be effective as broad spectrum anti-
`biotics and may even be targeted against organisms such as
`the multi-drug-resistant strains of staphylococcus that are
`proving to be a problem for the current range of antibiotics.
`However, given the inherent toxic potential of boron [2], it
`is obviously important to consider the possibility of design-
`ing a pre-formed bisubstrate analogue substituting the
`boron for another atom in order to develop a range of
`inhibitors with minimal side-effects.
`
`GENERIC APPLICATIONS
`The similarities in chemistry catalysed by the family of
`NAD(P)-dependent oxidoreductases give rise to generic
`opportunities for the creation of a series of novel enzyme
`inhibitors based on a related chemistry. Across the family of
`enzymes that belong to this class, the catalytic cycle of
`oxidation/reduction necessarily leads to a situation where
`the p electron system of a substrate approaches the face of
`the nicotinamide ring. Furthermore, the glycosidic bond
`between the nicotinamide ring and its associated ribose
`moiety generally adopts only one of two conformations that
`differ by a rotation of approximately 180° and that lead to
`the presentation of either the pro-4R or pro-4S hydrogen of
`the NADH to the active site. These conformations merely
`
`ANACOR EX. 2009 - 8/10
`
`

`
`Mechanism of Action of Diazaborines
`
`1549
`
`result in a shift in the position of the carboxyamide moiety
`of the nicotinamide ring on the enzyme surface and do not
`affect the relative positions of the nicotinamide ring to its
`ribose group. Moreover, the structural analysis of members
`of this family has shown that the active site in these
`enzymes is consistently positioned in the same relative
`orientation to the nicotinamide ribose. Thus, for the subset
`of dehydrogenases where there is sufficient space in the
`structure around the 29OH group of the nicotinamide
`ribose, there is an excellent opportunity to mimic the
`chemistry seen in the diazaborine–NAD1 complex in the
`synthesis of new enzyme inhibitors. A number of NAD(P)-
`dependent oxidoreductases are known to be drug targets,
`including dihydrofolate reductase, the target for the anti-
`cancer agent methotrexate [26, 27], steroid 5a-reductase,
`the target for finasteride, used to treat benign prostatic
`hyperplasia [28], and inosine monophosphate dehydroge-
`nase (IMPDH), the target for the immunosuppressant,
`mycophenolic acid (MPA) [28]. For example, there is
`considerable similarity between the ENR–NAD1– diazabo-
`rine complex and the orientation of the inosine-59-mono-
`phosphate thioimidate intermediate and the inhibitor
`MPA in IMPDH [29], suggesting that the 29-hydroxyl of
`the inosine may be linked to the inhibitor in a manner
`similar to that seen in the diazaborine–NAD1 complex.
`This suggests that the utilization of the ribose hydroxyl to
`create a bisubstrate analogue might find important applica-
`tions in other areas of medicinal chemistry.
`
`This work is supported by grants from the Biotechnology and Biological
`Sciences Research Council (BBSRC). C. B. is funded by a Zeneca
`Agrochemicals supported CASE award. J. B. R. is a BBSRC David
`Phillips Research Fellow. The Krebs Institute is a designated BBSRC
`Biomolecular Science Centre.
`
`References
`1. Hogenauer G and Woisetschlager M, A diazaborine deriva-
`tive inhibits lipopolysaccharide biosynthesis. Nature 293:
`662– 664, 1981.
`2. Grassberger MA, Turnowsky F and Hildebrandt J, Preparation
`and antibacterial activities of new 1,2,3-diazaborine deriva-
`tives and analogues. J Med Chem 27: 947–953, 1984.
`3. Turnowsky F, Fuchs K, Jeschek C and Hogenauer G, envM
`genes of Salmonella typhimurium and Escherichia coli. J Bacteriol
`171: 6555– 6565, 1989.
`4. Egan AF and Russell RRB, Condition