COMMENTARY

Repurposing libraries of eukaryotic kinase inhibitors for antibiotic discovery

Christopher T. Walsha,1 and Michael A. Fischbachb aDepartment of Biological Chemistry and Molecular , Harvard Medical School, Boston, MA 02115; and bDepartment of Molecular Biology, Massachusetts General Hospital, Boston, MA 02114

ith the rise in resistance NH2 N that inevitably follows the Br O O O O N clinical deployment of an O P O P O P O N N antibiotic, there is a con- N N N O O O O Br O W H N N N NH H N N N NH H N N tinual need for new antibiotic discovery, N N 2 2 2 2 2 OH OH development and approval. Food and 1 2 3 ATP Drug Administration approvals for Synercid (quinupristin/dalfopristin) in Biotin Carboxylase Carboxyltransferase 1999, Zyvox (linezolid) in 2000, and Cu- bicin (daptomycin) in 2003 have ad- O O dressed life-threatening infections from HN NH drug-resistant Gram-positive bacteria S CoA BCCP andrimid such as Staphylococcus aureus, Strepto- S acetyl-CoA coccus pneumomiae, and Enterococcus O biotinyl-BCCP O O faecalis. However, because these antibiot- HO S CoA O ATP ADP O O O ics are not active against an emerging O malonyl-CoA class of nosocomial pathogens (multidrug- HO OH HO O P O HN N OH resistant Gram-negative bacteria, includ- O bicarbonate carboxyphosphate BCCP ing strains of Klebsiella, Acinetobacter S O 1,2,3 and Pseudomonas) there is renewed fo- N-carboxybiotinyl-BCCP lipid cus on developing treatments for infec- biosynthesis tions caused by Gram-negative bacteria. To this end, in this issue of PNAS Fig. 1. inhibitors of acetyl-CoA carboxylase. (Upper) The chemical structures of molecules Miller et al. (1) report a novel approach 1–3 and ATP are shown. (Lower) A schematic of the reaction catalyzed by acetyl-CoA carboxylase, divided into the portions catalyzed by its biotin carboxylase and carboxyltransferase subunits. to discovering new classes of antibiotics. Although bacterial genome sequenc- ing and genetics have identified essential molecules with anti-Gram-negative activity olite platensimycin as a potent inhibitor of genes as potential targets for new antibi- from the outset. the FAS chain elongation enzyme (4). otics, efforts to screen synthetic chemi- Miller et al. (1) identified promising The BC subunit of acetyl-CoA car- cal libraries have been disappointingly hits from a series of pyridopyrimidines boxylase carries out this enzyme’s first unproductive (2). One possible explana- and determined that drug-resistant mu- half-reaction: activating bicarbonate by tion for the low yield is the bias, histori- tants (1 in 109) mapped to the biotin cleaving ATP to ADP and transiently cal and contemporary, of pharmaceuti- carboxylase (BC) subunit of the multi- generating the mixed anhydride carboxy- cal companies’ synthetic small molecule subunit enzyme acetyl-CoA carboxylase, phosphate. This activated carboxyl moi- libraries for eukaryotic rather than pro- ety is then captured by biotin [which is karyotic targets. Miller et al. (1) sought which catalyzes the first committed step in fatty acid biosynthesis (Fig. 1). In- tethered to the biotin carboxyl carrier to turn this paradigm on its head by protein (BCCP)], to yield N-carboxybi- making a virtue of the depth of such a triguingly, the pseudopeptide natural product andrimid is a nanomolar inhibi- otinyl-BCCP (Fig. 1). A second subunit, library, in this case composed of protein the carboxyltransferase, generates the kinase inhibitors. Even though protein tor of the carboxyltransferase subunit of the same enzyme (3), raising the pros- thioester enolate of acetyl-CoA and cat- kinases are much less widespread in bac- alyzes its attack on the biotin-tethered terial metabolism, Miller et al. set out to pect of the synergistic action of 2 drugs against this essential enzyme. carboxyl group to generate malonyl- determine whether their library con- CoA, the building block for fatty acid tained ATP analogs that could kill bac- Bacterial fatty acid biosynthesis has long been thought to be an underex- synthesis. The BC subunit, which had terial cells potently and selectively. been crystallized (5), is a member of the ploited pathway for antibiotics. The team (1) made use of the ATP-grasp enzyme superfamily. The Although 2 inhibitors of the enoylreduc- company’s screening file, comprising some crystal structures of library compounds tase component of fatty acid synthase 1.6 million compounds. A significant num- 1 and 2 in complex with the BC subunit (FAS) (isoniazid, used in combination ber of these compounds were heterocycles confirmed that they bind to its ATP- designed to be ATP-competitive ligands therapies for , and triclosan, binding site. for the ATP binding site of a eukaryotic an antiseptic) have seen use in humans, . They screened their library no other clinically-used antibiotics target against a membrane-compromised (imp), this pathway. However, recent efforts to Author contributions: C.T.W. and M.A.F. wrote the paper. efflux pump-deficient (tolC) strain of identify inhibitors of fatty acid biosyn- The authors declare no conflict of interest. Escherichia coli to find molecules that en- thesis have yielded promising leads: a See companion article on page 1737. tered and killed whole cells. Their choice screen of natural product extracts at 1To whom correspondence should be addressed. E-mail: of indicator strain imparted a focus on Merck turned up the Streptomyces metab- christopher࿝[email protected].

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0813405106 PNAS ͉ February 10, 2009 ͉ vol. 106 ͉ no. 6 ͉ 1689–1690 Downloaded by guest on September 26, 2021 Hit molecules from the screen were temic model of murine infection with H. allosterically activate ATP-dependent bactericidal against E. coli and the con- influenzae. Anticipating that a BC inhib- chambered proteases (7). sequential Gram-negative respiratory itor might be a candidate for combina- Both DNA gyrase and BC are in the pathogens Haemophilus influenzae and tion therapy, compound 1 was tested for ATP-grasp superfamily, members of Moraxella catarrhalis, but not against which cleave ATP to ADP. D-Ala-D-Ala Gram-positive bacteria. Compound 1 ligase, an ATP-grasp enzyme involved in has a Kd of 0.8 nM against the purified Miller et al. report a muropeptide elongation, is targeted with BC subunit from E. coli, and the selec- modest affinity by members of a syn- tivity profile of compounds 1 and 2 is novel approach to thetic screening library (8). It may be encouraging: no significant inhibition of worthwhile to test all bacterial ATP- 30 eukaryotic protein kinases, including discovering new classes grasp enzymes with libraries of synthetic VEGFR-2 and FGFR-1, the targets for kinase inhibitors. In this vein, the pre- which these heterocyclic scaffolds had of antibiotics. dominant form of protein kinases in been optimized. Thus, the ATP-binding bacteria are the autophosphorylating sites of eukaryotic and prokaryotic en- histidine kinases involved in 2- zymes may be an expanding class of antagonism or synergy with a several component signaling pathways. Al- valuable targets for molecules with antibiotics; the antiseptic triclosan and though no promising inhibitor candi- nanomolar potency and selectivity. the DNA gyrase inhibitor ciprofloxacin dates have emerged against this class to One of the ATP-binding site muta- were synergistic. Because andrimid in- date, Miller et al.’s results (1) suggest tions in BC that conferred pyridopyrimi- hibits the other catalytic subunit of that rescreening with eukaryotic target- dine resistance was I437T. BC subunits acetyl-CoA carboxylase (Fig. 1), it might directed ATP-mimetic libraries would be from Gram-positive bacteria harbor a be of interest to evaluate it in combina- worthwhile. It is unclear how many threonine at this position, providing a tion with the pyridopyrimidines for po- ATP-using enzymes in the bacterial pro- possible explanation for why compounds teome, both phosphoryl and tential synergy. 1 and 2 are selective for Gram-negative adenylyl transferases, could be targeted In addition to spotlighting BC as a bacteria. Further examination of the li- by existing eukaryotic kinase libraries. new antibacterial target with subnanmo- brary, leveraging the prior medicinal However, it is promising that ATP- chemical investment by Pfizer in eukary- lar lead compounds from a synthetic mimetic scaffolds have been particularly otic inhibitors, showed library, one of the values of Miller et malleable for exploring target protein al.’s study (1) is the renewed impetus it that compound 3, with a Kd of 150 nM, selectivity while maintaining high was only 14-fold less potent against the places on the evaluation of other bacte- potency. I437T BC mutant. This result gives a rial enzymes that use ATP (or GTP) as Finally, the question arises whether a clear path for future medicinal chemical potential targets for new antibiotics. To similar strategy of using synthetic chemi- efforts to tailor the pyridopyrimidine wit: the target of ciprofloxacin and cal libraries developed and iteratively scaffold to get higher efficacy against other fluoroquinolones is the ATP- improved for other classes of eukaryotic BCs from Gram-positive pathogens such consuming topoisomerase DNA gyrase, might be profitably screened as S. aureus. an inhibitor of the M. tuberculosis ATP against bacterial targets in whole-cell Initial pharmacokinetic studies in synthase (6) is in late-stage clinical eval- assays. Two target classes that might be mice led to the demonstration of effi- uation, and semisynthetic derivatives of worth particular attention are ion chan- cacy in both a thigh model and a sys- the ADEP family of natural products nels and prenyltransferases (9).

1. Miller JR, et al. (2009) A class of selective anti- carboxylase inhibitors with antibacterial activity. J Biol 7. Brotz-Oesterhelt H, et al. (2005) Dysregulation of bac- bacterials derived from a protein kinase inhibitor Chem 279:26066–26073. terial proteolytic machinery by a new class of antibiot- pharmacophore. Proc Natl Acad Sci USA 106:1737– 4. Wang J, et al. (2006) Platensimycin is a selective FabF inhib- ics. Nat Med 11:1082–1087. 1742. itor with potent antibiotic properties. Nature 441:358–361. 8. Triola G, et al. (2008) ATP competitive inhibitors of 2. Payne DJ, Gwynn MN, Holmes DJ, Pompliano DL 5. Mochalkin I, et al. (2008) Structural evidence for sub- D-alanine–D-alanine ligase based on protein kinase (2007) Drugs for bad bugs: Confronting the chal- strate-induced synergism and half-sites reactivity in inhibitor scaffolds. Bioorg Med Chem, 10.1016/ lenges of antibacterial discovery. Nat Rev Drug Dis- biotin carboxylase. Protein Sci 17:1706–1718. j.bmc.2008.02.046. cov 6:29–40. 6. Andries K, et al. (2005) A diarylquinoline drug active on 9. Liu CI, et al. (2008) A cholesterol biosynthesis inhibitor 3. Freiberg C, et al. (2004) Identification and character- the ATP synthase of Mycobacterium tuberculosis. Sci- blocks Staphylococcus aureus virulence. Science ization of the first class of potent bacterial Acetyl-CoA ence 307:223–227. 319:1391–1394.

1690 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0813405106 Walsh and Fischbach Downloaded by guest on September 26, 2021