Cyclobutanone analogues of ‐lactam : ‐lactamase inhibitors with untapped potential?

Prarthana Devi and Peter J. Rutledge*

Dr Prarthana Devi, Prof. Dr. P. J. Rutledge, School of Chemistry, The University of Sydney, Sydney, NSW 2006 Australia, E‐mail: [email protected]; Tel.: +61 2 9351 5020; Fax: +61 2 9351 3329.


‐Lactam antibiotics have been used for many years to treat bacterial infections. However the effective treatment of an increasing range of microbial infections is threatened by bacterial resistance to ‐lactams: the prolonged, widespread and at times reckless use of these drugs has spawned widespread resistance, which renders them ineffective against many bacterial strains. The cyclobutanone ring system is isosteric with ‐lactam: in cyclobutanone analogues, the eponymous cyclic is replaced with an all‐carbon ring, the amide N substituted by a tertiary C–H  to a . Cyclobutanone analogues of various ‐lactam antibiotics have been investigated over the last thirty‐five years, initially as prospective antibiotics in their own right and inhibitors of the ‐lactamase enzymes that impart resistance to ‐lactams, more recently as inhibitors of other serine proteases and mechanistic probes of ‐lactam biosynthesis. Cyclobutanone analogues of the penam ring system are the first reversible inhibitors to demonstrate moderate activity against all classes of ‐lactamase, while other compounds from this family inhibit Streptomyces R61 DD‐carboxypeptidase/transpeptidase, human neutrophil elastase (HNE) and porcine pancreatic elastase (PPE). But has their potential as enzyme inhibitors been fully exploited? Challenges in synthesising diversely functionalised derivatives mean only a limited number and structural diversity of cyclobutanone ‐lactam analogues have been made and evaluated to date. This review surveys the different synthetic approaches that have been taken to these compounds, investigations to evaluate their biological activity, and prospects for future developments in this area.

Key words ‐lactam, ‐lactamase inhibitor, cyclobutanone, antibiotics, antimicrobial resistance, , , carbapenam

1 Introduction

The emergence of antimicrobial resistance to antibiotics as a major global problem is well documented in both the scientific literature and the popular press.[1‐7] ‐Lactamases and ‐lactam antibiotics are key players in this emerging epidemic: ‐lactams are amongst the most widely used and best understood classes of antibiotics, ‐lactamases pose an ever‐increasing threat to their continued efficacy as they hydrolyse ‐lactams before they can exert an effect.[8‐10]

‐Lactam antibiotics have been characterised in great detail over the almost 90 years since

Fleming’s serendipitous discovery of Penicillium‐mediated antibiosis:[11] their biosynthesis and modes of action, mechanisms of resistance to ‐lactams, and methods for overcoming that resistance have all been investigated in depth.[12‐15] Myriad structural variants have been conceived and tested for activity as antibiotics and as ‐lactamase inhibitors to combat resistance.

The gamut of ‐lactam ring systems characterised ranges from the penam 1 first discovered to oxycyclic 2, unsaturated 3, oxapenem 4 and 5 ring systems, the ring‐ expanded 6, 7 and 8 (also unsaturated) and the monocyclic 9 and nocardicins 10 (Scheme 1). These subfamilies of compounds all retain the ‐ lactam ring and vary elsewhere in their structure.

The chemistry and biochemistry of these systems has been reviewed comprehensively elsewhere.[15‐20] The focus of this review is ‐lactam analogues that are not themselves ‐lactams: cyclobutanone derivatives designed to mimic the structure and activity of ‐lactams without actually incorporating the signature four‐membered cyclic amide.

Cyclobutanone is isosteric with the ‐lactam ring, with the quintessential ‐lactam nitrogen replaced by a carbon (Scheme 1). Since the ‐lactam nitrogen is effectively pyramidal by virtue of ring fusion in the penam and ring systems, Gordon Lowe and others hypothesised that

2 replacing this nitrogen with an sp3 hybridised carbon should “retain stereochemical compatibility with the active site of the transpeptidases and D,D‐carboxypeptidases involved in bacterial biosynthesis.”[21] Thus cyclobutanone analogues of ‐lactams have been explored since the early

1980s, as potential antibiotics in their own right,[21‐24] as ‐lactamase and serine protease inhibitors,[21, 24‐28] and as mechanistic probes.[29‐31]

Scheme 1: General structures for the major subfamilies of ‐lactams studied to date. The penam 1, clavam 2, carbapenem 5, cephem 6, 9 and nocardicin 10 ring systems are naturally occurring ‐lactams, while 3, oxapenems 4, oxacephems 7 and carbacephems 8 are synthetic ‐lactam subfamilies. The inset shows a generalized cyclobutanone analogue 11 of the penam 1: the tertiary nitrogen of the ‐lactam is replaced by a carbon (CH). R, R’ and R” = various.

It has been proposed that cyclobutanones – such as the generalised penam analogue 11 – have the potential to inhibit serine ‐lactamases via formation of an enzyme bound hemiketal with the active site serine (Scheme 2a), and metallo ‐lactamases via oxygen coordination to the metal ions

(Scheme 2b).[21, 27, 28] Further, it has been postulated that cyclobutanone analogues may inhibit serine proteases more generally, by hemiketal formation akin to that shown for the serine ‐ lactamase in Scheme 2a.[22, 25, 32]


Scheme 2: Proposed activity of cyclobutanones as broad‐spectrum ‐lactamase inhibitors, shown for a generalized cyclobutanone‐penam 11: inhibition of a. serine ‐lactamases via hemiketal formation, and b. metallo‐‐lactamases via oxygen coordination to zinc. Adapted with permission from Johnson et al. J. Org. Chem. 2008, 73, 6970‐6982. Copyright 2008 American Chemical Society. R = various.

Cyclobutanone analogues of the penam, clavam and carbapenam ring systems, and various monocyclic analogues, have all been synthesised and evaluated as inhibitors of transpeptidases, ‐ lactamases and other proteases. Early efforts in this field were reviewed by Jungheim and

Ternansky in the early 1990s, part of their chapter on Non‐‐lactam mimics of ‐lactam antibiotics[33] in the comprehensive text The Chemistry of ‐Lactams.[34] Much has happened since, and the aim of this review is to bring together almost forty years of collective endeavour, surveying and evaluating the synthetic approaches developed to make these challenging targets, the application of these compounds as enzyme inhibitors and mechanistic probes, and possible avenues for further investigation.


A variety of synthetic routes to cyclobutanone‐‐lactam analogues have been developed. All routes to bicyclic derivatives exploit [2+2] cycloaddition reactions to construct the four‐membered cyclic ketone.[22, 35]

Early Routes Gordon et al., reported the first synthesis of cyclobutanone analogues of penam and penem rings systems in 1981;[22] these were carbapenam and carbapenem analogues (Scheme 3), conceived as

DD‐carboxypeptidase and transpeptidase enzyme inhibitors. 4Carboxybicyclo[3.2.0]heptane‐6‐

4 one 12 was conceived as a simplified isosteric model of penicillanic acid 13, with both the ‐lactam

N and S replaced by carbons, while 4carboxybicyclo[3.2.0]hept‐2‐ene‐6‐one 14 is the cyclobutanone isostere of the unfunctionalised carbapenem 15.

Scheme 3: First generation cyclobutanone analogues 12 and 14, conceived as mimics of penicillanic acid 13 and the simple carbapenem 15 respectively.[22]

Both 12 and 14 were synthesized in racemic form from 6,6‐bis‐methylmercaptofulvene 16 and dichloroketene (generated in situ from dichloroacetyl chloride and an base) via a [2+2] cycloaddition reaction to form cycloadduct 17 (Scheme 4).[35] Cleavage of the mercaptofulvene group in 17 using formic acid and two equivalents of mercuric chloride produced both the 18 and thioester 19 in a 1:4 ratio, and 16% combined yield over three steps from cyclopentadiene, precursor to fulvene 16. Carboxylic acid 18 and thioester 19 were each isolated as a single diastereomer (Scheme 4), their regiochemistry and relative stereochemistry confirmed by comparison of 1H NMR data with literature data for related compounds, and X‐ray crystallographic analysis of 18. From the carboxylic acid 18, reductive dechlorination using zinc and acetic acid followed by catalytic hydrogenation afforded carbapenam‐cyclobutanone 12. From thioester 19, using a mixture of concentrated hydrochloric acid and glacial acetic acid, then the zinc/ acetic acid dechlorination, gave carbapenem analogue 14.

5 Scheme 4: Synthesis of first generation cyclobutanone analogues 12 and 14 (racemic). Reagents and

conditions: (i) Cl2CHCOCl, Et3N, Et2O, ‐10 °C; (ii) HCOOH, HgCl2, reflux, 15 min, 16% over 3 steps, of a

~1:4 mixture of 18:19; (iii) Zn/HOAc, 70 °C, 45 min, 66%; (iv) H2, Pd/C, EtOAc, 82%; (v) HCl/HOAc, 27%; (vi) Zn, HOAc, 70 °C.[35]

Meth‐Cohn et al. extended this methodology to make the gem‐dimethyl analogue 20 (Scheme

5).[23] Their approach used the [2+2] cycloaddition of dimethylketene to cyclobutadiene 21 to generate the fused bicyclic system 22, but then deployed a cyanide‐mediated rearrangement and a series of reductive functional group interconversions (FGIs) such that the carbon atoms of the ketene ultimately reside in the five‐membered ring, and not the cyclobutanone. Thus bicyclic adduct 23 was first prepared following literature methods,[36, 37] via the cycloaddition of dimethylketene to cyclopentadiene 21. Then reaction of cycloadduct 22 with N‐bromoacetamide and benzyl alcohol afforded bromo‐ether 23. The key rearrangement used potassium cyanide and sodium methoxide to generate the desired bicyclic product 24 as a ~1:1 mixture of epimers at C4:

29% exo nitrile (cyano cis to benzyloxy), 32 % endo nitrile (cyano trans to benzyloxy). While the two epimers of 24 were readily separated, this proved unnecessary as both gave rise to the thermodynamically more stable exo‐acid 25 after nitrile hydrolysis. Hydrolysis of the epimeric nitriles was achieved stepwise, using phase‐transfer conditions with aqueous sodium hydroxide and peroxide to form the amide, which was efficiently converted to the carboxylate using sodium hydroxide in triethylene glycol. Reduction of the ketene‐derived carbonyl under Huang‐Minlon conditions,[38] hydrogenolysis with Pd/C to remove the benzyl protecting group, and oxidation of the cyclobutanol with pyridine/ sulfur trioxide afforded the penicillanic acid analogue 20 (Scheme

5). The same group also described the synthesis of an O‐methyloxime derivative 26 from ketone

25, designed to optimise ‐lactamase inhibition (vide infra).


Scheme 5: Racemic synthesis of gem‐dimethyl penam analogue 20. Reagents and conditions: (i)

(CH3)2CHCOCl, Et3N, CHCl3, 20‐25 °C, 15 h, 76%; (ii) PhCH2OH, AcNHBr, 15 °C, 18 h, 78%; (iii) KCN,

CH3OH, NaOCH3, heat, 36 h, 61%; (iv) NaOH/H2O2, Bu4N(HSO4), CH2Cl2, rt, overnight, 44‐68%; (v) NaOH,

H2O, triethyleneglycol, 170 °C, 45% (vi) Huang‐Minlon reduction, 79%; (vii) H2, Pd/C, 93%; (viii)

pyridine‐SO3, DMSO, Et3N, 79%; (ix) H2, Pd/C, 90%; (x) NH2OMe, 80%; (xi) pyridine‐SO3, DMSO, Et3N, 74%.[23]

Introducing Functionality  to the Cyclobutanone Carbonyl

The majority of bioactive ‐lactams incorporate an acylamino functionality at the 7‐position on the ring (adjacent to the ‐lactam carbonyl, Scheme 1), but the early cyclobutanone analogues 12, 14 and 20 reported by Gordon and Meth‐Cohn are unfunctionalised at this key position. Lowe and

Swain developed methodology to introduce acylamino functionality at the 7‐position of cyclobutanone analogues, and used it to synthesise the cyclobutanone‐clavam analogues 27 and

28 (Scheme 6).[21, 24] Cyclobutanone 28 was sought as an isostere of 1‐oxabisnorpenicillin G 29, itself an active antibiotic in vivo.[39]

Scheme 6: First examples of cyclobutanone‐‐lactam analogues functionalised at the 7‐position on the ring, bearing Cl (27) and acylamino (28) functionality; 29 is 1‐oxabisnorpenicillin G, the‐lactam on which cyclobutanone analogue 28 is based.

7 Lowe’s strategy used dichloroketene in the key [2+2] cycloaddition step, generating this partner in situ and reacting with the 2,3‐dihydro‐3‐furoic acid 30 to give 7,7‐dichloroadduct 31 (Scheme

7). Dechlorinating 31 with zinc in acetic acid removed both chlorines to generate cyclobutanone

32, unfunctionalised at C7. (Alternatively, treating 31 to catalytic hydrogenation with palladium on carbon removed only one chlorine, and afforded mono‐chloro cyclobutanone 27.)

To form a C–N bond to the carbonyl proved tricky, and a somewhat circuitous route was developed. Reduction of the cyclobutanone carbonyl in 32 with L‐selectride gave cyclobutanol 33 with the relative stereochemistry shown. (Using Bakers’ yeast to carry out this reduction allowed access to enantiomerically pure diastereomers of 33, opening the way to an asymmetric synthesis, although this possibility was not pursued further.) Reaction with phosgene converted 33 to the chloroformate derivative in quantitative yield, from which azidoformate 34 could be generated by reaction with sodium azide in DMF. The key C–N bond‐forming step invokes thermal decomposition of the acyl azide to generate a nitrene, by heating a dichloromethane solution of

34 to 135 °C in a sealed tube. The nitrene inserts into the C–H bond at C7 to give cyclic carbamate

35, in which the nitrene has been delivered from above, as dictated by the stereochemistry of the precursor alcohol.

Base‐catalysed hydrolysis and acylation with phenylacetyl chloride in situ gave the phenylacetamido‐derivative 36, from which the benzyl ester has also been cleaved. Finally oxidation with sulfur trioxide–pyridine yielded cyclobutanone 28 as the major component of a 2:1 mixture epimeric at C7, the first example of cyclobutanone ‐lactam analogue incorporating both a carboxy substituent at C4 and an acylamino side‐chain at C7.


Scheme 7: Synthetic route to cyclobutanone penam analogues with acylamino functionality at the 7‐ position; this approach is racemic, but Bakers’ yeast may be used in place of lithium selectride to reduce 32, opening the possibility of a stereoselective synthesis. Reagents and conditions: (i)

Cl2CHCOCl, Et3N, 59%; (ii) Zn, HOAc, 88%; (iii) LiB(CHMeEt)3H, 59%; (iv) phosgene, pyridine; (v) NaN3,

DMF,64% over two steps; (vi) CH2Cl2, 135 °C, sealed tube, 3.5 h, 29%; (vii) aqueous KOH, dioxane, then [21, PhCH2COCl, 52%; (viii) SO3‐pyridine, DMSO, Et3N. 48% of a 2:1 mixture of 28:C7 epimer; R= CH2Ph. 24] Several groups have incorporated aminoacyl functionality at the corresponding position of monocyclic cyclobutanones, thus making cyclobutanone analogues of monobactams 9 and nocardicin 10. Ghosez and co‐workers prepared enantiomerically pure ‐aminocyclobutanones 37 and 38 and a wider range of cyclobutanol precursors 39–42 (Scheme 8) as putative serine protease inhibitors.[26]

Scheme 8: Cyclobutanones 37 and 38 and related cyclobutanol structures 39–42, carbocyclic analogues of monocyclic ‐lactams like monobactams 9 and nocardicin 10; R, R’ = various.

This approach exploits the [2+2] cycloaddition reaction between ketene 43 and Seebach’s oxazoline (tert‐butyl (R)‐2‐(tert‐butyl)oxazole‐3(2H)‐carboxylate, 44)[40] which yielded the bicyclic cyclobutanone 45 as the major regioisomer (Scheme 9).[41, 42] Regioisomer 45, in which the ketene is  to the oxazoline nitrogen, is favoured 20:1. The reaction is also highly

9 stereoselective: the ketene adds exclusively to the face of the oxazoline that is away from the tert‐ butyl group, and the chloro substituent ends up on the exo face of the bicyclic system. Reduction of the ketone, oxidation of the exocyclic primary alcohol, and a series of protecting group manipulations led via the bicyclic systems 46 and 47 to monocyclic cyclobutanols like 39 and 40.

Reorienting the molecule so the oxazoline‐derived oxygen becomes the cyclobutanone carbonyl and oxidation with TEMPO affords enantiomerically pure monocylic cyclobutanones 37 and 38 which incorporate ‐aminoacyl functionality.[26] The absolute configuration of these compounds derives directly from L‐serine (used to generate oxazoline 44) and the stereochemical course of the cycloaddition step.

Scheme 9: Enantioselective synthesis of carbocyclic analogues of monocyclic ‐lactams. Reagents and

conditions: (i) Et3N, cyclohexane, 75%; (ii) NaBH4, CeCl3, THF, 60%; (iii) CH2=CHCH2Br, KH, THF, 90%; (iv)

TBAF, THF, 92%; (v) TEMPO, NaClO, NaClO2, Na3PO4, MeCN/ H2O, 89%; (vi) TEMPO, NaClO, NaHCO3, [41, 42] toluene, EtOAc, H2O, 85%. R = Me or Bn.

Reid and co‐workers prepared the epimeric ‐aminoacylated monocyclic cyclobutanones 48 and

49, in which the four‐carbon ring is derived directly from trans‐aminocyclobutanol (Scheme 10).[25]

By coupling this amine to N‐Cbz‐L‐proline 50 and then extending the nascent peptide 51, they generated peptide‐functionalized cyclobutanols and oxidised these to the corresponding cyclobutanones 48 and 49, which could be separated by HPLC.


Scheme 10: Synthesis of diastereomeric ‐aminoacylated monocyclic cyclobutanones from L‐proline and trans‐aminocyclobutanol. Reagents and conditions: (i) N‐methyl morpholine, isobutyl

chloroformate, THF, ‐78 °C, 10 min, then trans‐2‐amino‐cyclobutanol, THF, RT, 4 h, 45%; (ii) H2, Pd/C,

MeOH, RT, 4 h; (iii) N‐Cbz‐L‐alanyl‐L‐alanine, N‐methyl morpholine, isobutyl chloroformate, THF, ‐78 °C, [25] 10 min, then 52, RT, 4 h, 28%; (iv) (COCl)2, DMSO, Et3N, CH2Cl2, ‐78 °C to RT, 90 min, 92%.

Several groups have reported efficient syntheses of simple N‐functionalised ‐ aminocyclobutanones 53 (Scheme 11). Vederas and co‐workers made the racemic N‐Cbz derivative (53, R=H, R’=CBz) by converting dimethyl succinate 54 to 1,2‐ bis(trimethylsilyloxy)cyclobutene 55,[43] then treating this with benzyl carbamate and HCl.[44]

Becker and his group extended the approach using a range of nitrogen nucleophiles including carbamates, , sulfonamides, and anilines in place of benzyl carbamate to make a library of substituted ‐aminocyclobutanone derivatives 53 in racemic form.[32] Frongia and co‐workers recently reported an enantioselective organocatalyzed route to N‐functionalised ‐ aminocyclobutanones 53.[45, 46] Their approach combines racemic α‐hydroxycyclobutanone 56 and

N‐substituted anilines or N‐alkyl‐α‐ in a tandem condensation/ keto‐enol tautomerisation process catalysed by the cinchona alkaloid derivative (DHQD)2PHAL (Scheme 11).

Scheme 11: Synthetic approaches to simple N‐functionalised ‐aminocyclobutanones include both racemic (from 55) and enantioselective (from 56) routes. Reagents and conditions: (i) Na, THF, TMSCl,

ultrasound, 0–4 ˚C, 16 h, 60%; (ii) RNHR’, HCl, Et2O, 80 ˚C, 4 h, 10–90%; (iii) RNHR’, (DHQD)2PHAL (30 mol%), toluene, 0 ˚C, 36–72 h, 30–90%, ee up to 82%. R, R’ = various.[32, 44‐46]

11 More Complex Systems

Cocuzza and Boswell synthesised cyclobutanones 57–59, analogues of the carbapenem ring systems (Scheme 12), by extending the ketene [2+2] cycloaddition chemistry developed by Lowe and Swain.[21, 24] N‐Acetyldeazathienamycin 57 is a cyclobutanone analogue of 60, while the S‐oxidised variants 58 and 59 incorporate the sulfoxide or sulfone substituent to activate the cyclobutanone carbonyl to nucleophilic attack.[47]

Scheme 12: Cyclobutanone analogues of carbapenem antibiotics: compound 57 is an N‐acatylated cyclobutanone version of thienamycin 60, while 58 and 59 are simpler S‐oxidised variants in which the sulfoxide/ sulfone electron withdrawing group was conceived to activate the cyclobutanone carbonyl to nucleophilic attack.

This approach combined dichloroketene and the silyloxyfulvene 61 in the key [2+2] cycloaddition step (Scheme 13), followed by cleavage of the silyl enol ether to unmask the aldehyde, oxidation to the carboxylic acid, esterification and dechlorination to afford the simple carbapenem analogue

62. Michael addition of thiophenol and oxidation (with sulfuryl chloride) to reintroduce the double bond gave key intermediate 63. Varying the temperature of peracid oxidation allowed access to either sulfoxide 58 or sulfone 59, and the phenylsulfonyl group of 59 was readily displaced with various nucleophiles to prepare a range of analogues 64–66.

Cocuzza and Boswell used aldol chemistry to functionalise at C7 of the bicyclic cyclobutanone in their route to thienamycin analogue 57 (Scheme 13). Thus ketone 63 was converted to its zirconium enolate and combined with acetaldehyde to give alcohol 67 (one half of a 1:1 mixture of trans aldol products). S‐Oxidation to the sulfone and displacement of the phenylsulfonyl group

12 with sodium 2‐aminoethylthiolate, followed by a series of protecting group manipulations completed synthesis of N‐acetyldeazathienamycin 57.

Scheme 13: Racemic route to N‐acetyldeazathienamycin 57, cyclobutanone analogue of thienamycin

60. Reagents and conditions: (i) Cl2CHCOCl, Et3N, RT, 1.5 h; (ii) 50% HF, MeCN, reflux, 35 min; (iii) Jones

reagent, acetone, 20 °C, 45 min; (iv) Ph2CN2, EtOAc, RT, 1 h; (v), Zn, THF, HOAc, RT, 6 h; (vi) PhSH, Et3N,

THF, RT, 1 h; (vii) SO2Cl2, Pyridine, CH2Cl2, ‐60 °C, 30 min; (vii) LiN(SiMe3)2, Cp2ZrCl2, CH3CHO, THF, ‐78 [47] °C, 1 h. R= CHPh2.

To access the thiabicyclo ring system and thus make cyclobutanone analogues of penams and penems (e.g. 68–73, Scheme 14), Dmitrienko and co‐workers first developed a route to the 3‐ carboxy‐substituted 2,3‐dihydrothiophene ring system, then used this in the [2+2] cycloaddition reaction with dichloroketene.[48]

Scheme 14: Cyclobutanone analogues of penam (68–71) and penem (72, 73) ring systems.

Dmitrienko made 2,5‐dihydrothiophene‐3‐carboxylic acid 74 in five steps from ethyl bromoacetate,[27] then generated the required 2,3‐dihydrothiopene‐3‐carboxylate regioisomer 75 by treating 74 with ethyl chloroformate and triethylamine to migrate the out of conjugation.[27, 48] The [2+2] cycloaddition of dichloroketene to 75 proceeded smoothly to form the key thiabicyclo adduct 76, which can be elaborated to a range of derivatives (Scheme 15).[27, 28]

Ester hydrolysis and reductive dechlorination affords the simple penam analogues 69 and thence

13 68. Regioselective chlorination of 76 with sulfuryl chloride generates trichloro derivative 77, a stereoselective reaction which favours the 3‐ chloride. Reaction of 77 with a range of nucleophiles (including water, methanol, and isopropanol) opens the way to various 3‐alkoxy derivatives (S,O‐acetals, e.g. 78, 79), the corresponding 3‐hydroxy compounds (thiolactols), and the elimination product 80.[27]

To separate the 3 (70) and 3 (71) epimers of the 3‐methoxy penam analogue, it proved expedient to work with the corresponding benzhydryl esters 81 and 82, which were separated using a combination of fractional crystallisation and chromatography and then deprotected by reacting with TFA (Scheme 15).[28] Penem analogue 73 was generated by treating a mixture of epimeric acetals 70 and 71with methanesulfonic acid to eliminate methanol; dechlorination of 73 using the established zinc/ acetic acid conditions yielded 72 in low yield.[28]

Scheme 15: Synthesis of cyclobutanone analogues of penams and penems (racemic). Reagents and

conditions: (i) ClCO2Et, Et3N, CH2Cl2, 94%; (ii) Cl2CHCOCl, Et3N, hexane, 65%; (iii) 6 M HCl, dioxane, 80

°C, 18 h, 79%; (iv) Zn, AcOH, 88%; (v) SO2Cl2, CH2Cl2, 0 °C to RT, 4 h, 100%; (vi) MeOH, AgOTf, 3 Å MS,

CH2Cl2, 0 °C to RT, 6 h, 96% of a 5.5:18:1 mixture (78:79:80); (vii) Ph2CN2, EtOAc, rt, 2 h, 98%; (viii) N‐

Chlorosuccinimide, CH2Cl2, rt, 24 h; (ix) MeOH, rt, 24 h, 25% (81), 5% (82) over two steps; (x) TFA,

14 [27, anisole, RT, 2 h, 79% (70), 61% (71); (xi) MsOH, CH2Cl2, RT, 1 h 67%; (xii) Zn, HOAc, 85 °C, 14 h, 9%. 28]

Analogues of Penicillin N and Isopenicillin N

Baldwin and co‐workers took this chemistry further to make cyclobutanones 83 and 84, direct analogues of penicillin N 85 and isopenicillin N (IPN) 86 respectively (Scheme 16).[30, 31]

Scheme 16: Cyclobutanones 83 and 84 are direct isosteres of the biosynthetic ‐lactams penicillin N 85 and isopenicillin N 86 respectively, and (with thienamycin analogue 57) are the most elaborate ‘cyclobutanone ‐lactams’ made to date.[30, 31]

To achieve this, the 2,2‐dimethyl‐2,3‐dihydrothiophene‐3‐carboxylate 87 (made in four steps from commercially available triethylphosphono acetate) was used in the key [2+2] cycloaddition reaction with dichloroketene (Scheme 17).[29‐31] The resulting cycloadduct was readily dechlorinated to 88, then converted to acyl azide 89 and tricyclic carbamate 90 by adapting Lowe’s nitrene insertion chemistry to the thiabicyclo system. N‐Boc protection and opening of the cyclic carbamate under basic conditions[49] afforded cyclobutanol 91. From this key intermediate, a protecting group exchange enables peptide coupling (EDCI/ HOBt) with a suitably protected ‐ aminoadipic acid monoester, then oxidation of the cyclobutanol (IBX/DMSO),[50] and global

[30] deprotection affords either 83 (when D‐‐aminoadipic acid is used in the peptide coupling step)

[31] or 84 (from L‐‐aminoadipic acid).

Scheme 17: Racemic approach to precursors of cyclobutanone penicillin N and isopenicillin N (IPN)

analogues: deprotection of 91, amide coupling with ‐aminoadipic acid (D or L) and oxidation of the

15 cyclobutanol affords cyclobutanones 83 and 84. Reagents and conditions: (i) Cl2CHCOCl, Et3N, CCl4, 5

days, 96%; (ii) Zn, HOAc, reflux, 83%; (iii) NaBH4, MeOH 83%; (iv) triphosgene, pyridine, CCl4, 50 ˚C,

quantitative; (v) NaN3, DMF, 50 ˚C, 67%. (vi) TCE, 147 ˚C, 15 min, 28%; (vii) Boc2O, Et3N, DMAP, THF, RT, [29‐31] 1 h, 79%; (viii) Cs2CO3, EtOH, RT, 3 h, 90%.

Chemical Biology

Inhibition of Transpeptidases and ‐Lactamases

Scheme 18: First generation cyclobutanone ‐lactam analogues, designed and tested as inhibitors of R‐

TEM ‐lactamase and DD‐carboxypeptidase R61 (12 and 14),[22] Streptomyces R61 DD‐ carboxypeptidase, E. coli R‐TEM and B. cereus I ‐lactamases (27 and 28).[21, 24]

Early cyclobutanone analogues of ‐lactam antibiotics were devised as potential antibiotics and ‐ lactamase inhibitors. The first such compounds made (12 and 14, Scheme 18) were tested as inhibitors of the R‐TEM ‐lactamase[51] and R61 DD‐carboxypeptidase/transpeptidase.[52] No significant inhibition was observed, leading the authors to conclude that the ‐lactam nitrogen plays a key role in how these proteins recognize and bind ‐lactam antibiotics.[22]

In contrast, both the C7‐functionalised clavam analogues chloroketone 27 and acylamido derivative 28 (as the ~2:1 mixture of C7 epimers detailed above, Scheme 7) showed “slow, time‐ dependent inhibition of the E. coli RTEM‐2 ‐lactamase and ‐lactamase type I from B. cereus

[24] strain 568/H.” These compounds also demonstrated weak inhibition of Streptomyces R61 DD‐

‐1 carboxypeptidase (50% inhibition (IC50) at ca. 260 mg L , i.e. 1.36 mM for 27 and 0.90 mM for 28).

However these compounds did not show any significant antibacterial activity when tested against a range of Gram‐positive and Gram‐negative bacteria, Haemophilis, Pseudomonas spp., and

‐1 Bacteriodes spp., at a maximum concentration of 128 mg L (i.e. 0.67 mM for 27, 0.44 mM for 28).

16 Lowe and Swain postulated that the ‐lactamase and DD‐carboxypeptidase inhibition observed may be associated with the slow formation of a tetrahedral adduct between the inhibitor 27/ 28 and the enzyme.

The derivative 26 of gem‐dimethyl analogue 20 was devised to activate the cyclobutanone carbonyl to nucleophilic attack (Scheme 19), and thus optimise the inhibitory activity versus serine

‐lactamase.[23] However no biochemical or biological data for either 20 or 26 has been disclosed.

Scheme 19: Oxime derivative 26 of carbapenam analogue 20 was conceived to optimise the susceptibility of the cyclobutanone carbonyl to nucleophilic attack at the active site of serine ‐ lactamases (Enz–OH).[23]

With thienamycin analogues 57–66, Cocuzza et al. first demonstrated the acylating ability of these compounds in a model reaction with benzylamine (Scheme 20).[47] The unactivated systems 62–64 did not react and were recovered unchanged. The sulfoxide 58, sulfone 59, nitrile 65 and ketone

66, each of which incorporates an electron withdrawing group to activate the cyclobutanone to nucleophilic attack, were efficient acylating agents and reacted quickly with benzylamine.

[47] Scheme 20: Reagents and conditions: (i) PhCH2NH2, CH2Cl2, RT. R= CHPh2.

17 When these compounds were tested for antibacterial activity, sulfoxide 58, sulfone 59, and sulfoxide 92 (derived from S‐oxidation of intermediate 67, see Scheme 13) were active against

‐1 Staphlococcus aureus with MICs in the range 25–50 g mL (i.e. 0.05–0.11 mM). Cocuzza et al. also deduced inhibition of ‐lactamase indirectly, observing that these compounds showed “synergism with penicillin G against ‐lactamase producing strains of S. aureus.” However anti‐‐lactamase activity was not directly demonstrated. The key thienamycin analogue 57 was not active in either assay.

Dmitrienko, Strynadka and co‐workers demonstrated conclusively that cyclobutanone ‐lactam analogues can inhibit all classes of ‐lactamases in 2010, when they tested the activity of thiabicyclo analogues 68–73 against a panel of ‐lactamases (Table 1).[28] This screen comprised

Klebsiella pneumonia carbapenemase 2 (KPC‐2), a class A extended‐spectrum ‐lactamase (ESBL) that hydrolyses all classes of ‐lactams,[53] IMP‐1, a class B metallo‐‐lactamase with a wide spectrum of activity (its name derives from the fact that it is ‘active on ’),[54] GC1, a class

C ESBL from Enterobacter cloacae,[55] and OXA‐10, a class D ‐lactamase.[56, 57]

† Table 1: Inhibition of ‐lactamases by thiabicyclo cyclobutanone analogues of ‐lactams (IC50, M ). Adapted with permission from Johnson et al. J. Am. Chem. Soc. 2010, 132, 2558‐2560. Copyright 2010 American Chemical Society..

Compound % hydrate in Class A Class B Class C Class D ‡ D2O KPC‐2 IMP‐1 GC1 OXA‐10 68 0 117 ± 13 235 ± 14 44 ± 3 1135 ± 33 69 74 76 ± 8 >1000 25 ± 3 268 ± 8 70 >98 58 ± 2 122 ± 5 6.5 ± 1.4 156 ± 6 71 6 99 ± 5 nd§ 38 ± 4 547 ± 19 72 <2 170 ± 2 >500 34 ± 3 >1000 73 93 26 ± 2 213 ± 21 4.5 ± 0.3 370 ± 15

† determined by monitoring nitrocefin hydrolysis ‡ determined by NMR § nd = not determined 18 Cyclobutanones 68–73 inhibited all classes of ‐lactamase; the class C enzyme GC1 was inhibited most strongly, then class A carbapenemase KPC‐2, and class D enzyme OXA‐10. Activity against the metallo‐‐lactamase IMP‐1was also observed, with analogues 70, 73 and 68 showing strongest inhibition of this protein target. While the observed anti‐‐lactamase activity is modest, these cyclobutanones are the first class of compound to demonstrate inhibitory activity against all classes of ‐lactamase.[28]

Activity against the three serine ‐lactamases (KPC‐2, GC1, OXA‐10) correlates with the extent to which the cyclobutanone carbonyl is hydrated in aqueous solution.[27, 28] Thus the dichloro compounds 69 and 73, which are hydrated to a greater extent in solution than the corresponding dechlorinated compounds 68 and 72 (Table 1), were correspondingly more active inhibitors of

KPC‐2, GC1, and OXA‐10. Moreover, the strongest inhibition of these serine ‐lactamases was seen with compounds that prefer the exo conformation over an endo envelope (Scheme 21).

Comparing the three dichlorinated compounds 69, 70 and 71, the 3‐methoxy analogue 70, which favours the exo envelope whereas 69 and 71 prefer endo,[27] is the most potent of these inhibitors against all three serine ‐lactamases.

Hemiketal formation is key to the proposed mechanism of serine ‐lactamase inhibition by cyclobutanones and related compounds such as peptide aldehydes[58, 59] and ‐fluoroketones.[60]

In these other compound classes, the relative stability of enzyme‐bound tetrahedral intermediates has been shown to correlate with rates of hydration and hemiketal formation at the carbonyl carbon.[61, 62]. Thus Dmitrienko and co‐workers have looked in detail at the conformational preferences of thiabicyclo systems, and the influence of conformation on the relative stability of the corresponding hemiketals and hydrates.[27] Computational and NMR experiments with the 3‐ methoxy compounds 78 (exo) and 79 (endo), the ethyl esters of inhibitors 70 and 71 respectively

19 reveal the effects of 3‐substitution on cyclobutanone conformation and the stability of hemiketals at the cyclobutanone carbonyl (Scheme 21).

Scheme 21: Reagents and conditions: (i) MeOH, CH3CN, 48 h, rt, 75% (78), 24% (79); (ii) MeOD, NMR tube. In comparing the exo and endo conformations, note in particular the relative position of ring carbon C3, which bears the OR group.[27]

Computational analyses indicated that bicyclic cyclobutanones with 3substituents (e.g. 78) favour an exo envelope, while analogues with 3 substituents (e.g. 79) prefer the endo conformation, a consequence of anomeric effects which favour positioning a C3 alkoxy or hydroxy substituent in an axial orientation. The rate and extent of hemiketal (or hydrate) formation at the cyclobutanone carbonyl then depends on the preferred conformation, and so ultimately on the stereochemistry at C3. Monitoring hemiketal formation in NMR experiments with methanol‐d4

Dmitrienko and co‐workers noted that the exo cyclobutanone 78 was converted almost completely to hemiketals 93 and 94 (98%), while endo analogue 79 gave much lower levels (15‐

40%) of the corresponding hemiketals 95 and 96. This difference is thought to arise from adverse steric interactions between the 3‐methoxy group and the hemiketal in 95 and 96 (Scheme 21), and provides explanation as to why 3‐methoxy analogue 70 is a more potent inhibitor of serine

‐lactamases than 3‐methoxy epimer 71 or des‐methoxy analogue 69 (which also adopts an endo conformation, in the absence of anomeric effects).[27, 28]

20 Taking this one step further, Dimetrienko, Strynadka and co‐workers succeeded in crystallising the

3‐methoxy analogue 70 in the active site of OXA‐10, and thus directly visualizing the inhibitor in action against a serine ‐lactamase (Figure 1). The resulting X‐ray crystal structure confirmed that the cyclobutanone of inhibitor 70 is covalently linked to the active site serine via a hemiketal. It was proposed that 70 is bound to OXA‐10 as the hemiketal/ alkoxide, stabilized by the oxyanion hole at the enzyme active site,[28] on the basis of this crystal structure and recent findings with other serine hydrolase enzymes.[63, 64]

Figure 1: X‐ray structure of OXA‐10 with cyclobutanone 70 as a serine‐bound hemiketal: A. stereoview of the active site; B. Electron density at the hemiketal linkage; C. Interactions of OXA‐10 with 70; D. alternate stereoview of the active site. Reproduced with permission from Johnson et al. J. Am. Chem. Soc. 2010, 132, 2558‐2560. Copyright 2010 American Chemical Society.

Inhibition of Other Serine Proteases

Broadening the scope of cyclobutanone β‐lactam analogues beyond transpeptidase and β‐ lactamase inhibition, Reid and co‐workers tested their ‐pepitamino monobactam analogues 48 and 49 as inhibitors of the serine proteases human neutrophil elastase (HNE) and porcine pancreatic elastase (PPE).[25] The activity of cyclobutanones 48 and 49 was compared to acyclic 21 counterparts (e.g. 97) and a similarly elaborated trifluoromethyl ketone 98 (Scheme 22). All compounds tested inhibited PPE more strongly than HNE, an expected outcome as the L‐Ala‐L‐Ala‐

L‐Pro backbone of 48 and 49 was optimised for the PPE binding site, which prefers small aliphatic side chains.[65] Importantly, the cyclobutanones 48 and 49 were up to two orders of magnitude more potent than the corresponding acyclic analogues. Pauls, Cheng and Reid posited that this difference in potency arises either from conformational restriction in the cyclic analogues (i.e. entropic factors), or because internal ring strain (I strain) in the cyclobutanone promotes formation of an sp3 hybridised (i.e. tetrahedral) hemiketal in the four‐membered ring.[25, 66] By comparison, peptidyl trifluoromethyl like 98 inhibit HNE in the nanomolar range[59, 67, 68] by virtue of their ability to form highly stabilised hemiketal‐like structures within the enzyme active site.[69, 70] It follows that cyclobutanones like 48 and 49 will undergo nucleophilic attack more easily than their acyclic analogues, as the I strain of the cyclobutanone is released by formation of the hemiketal in the enzyme active site. The monocyclic ‐aminocyclobutanones prepared by Ghosez, Becker, Frongia and their co‐workers (e.g. 37, 38 and 53) have yet to be tested against serine protease targets,[26, 32, 45, 46] but the simple N‐Cbz derivative (53, R = H, R′ =

Cbz) did not inhibit the 3C cysteine protease from hepatitis A virus when used as a negative control.[44]

Scheme 22: The monocyclic ‐peptamidocyclobutanone 49 is a more potent inhibitor of both human neutrophil elastase (HNE) and porcine pancreatic elastase (PPE) than the acyclic analogue 97, but a weaker inhibitor of both enzymes than trifluoromethyl ketone 98.[25]

22 Cyclobutanones as Mechanistic Probes of ‐Lactam Biosynthesis

The bicyclic cyclobutanones 83 and 84 – analogues of penicillin N 85 and IPN 86 respectively

(Scheme 16) – were devised as mechanistic probes of penicillin and cephalosporin biosynthesis.[30,

31] Isopenicillin N synthase (IPNS) and deacetoxycephalosporin C synthase (DAOCS) catalyse key steps in the biosynthesis of and respectively (Scheme 23). [15][71] [72]

Scheme 23: Isopenicillin N synthase (IPNS) catalyses the oxidative cyclization of ‐(L‐‐aminoadipoyl)‐L‐cysteinyl‐

D‐valine (ACV, 99) to isopenicillin N 86. Subsequent epimerisation at the remote ‐aminoadipoyl stereocentre affords penicillin N 85, which is the substrate for ring expansion by deacetoxycephalosporin C synthase (DAOCS) to form the cephem 100. A wide variety of other penicillins can also be made from IPN 86 via transacylation of the ‐aminoadipoyl sidechain with other carboxylic acids.[71]

Following the crystallisation of the IPNS complex with its substrate ‐(L‐‐aminoadipoyl)‐L‐

[73] cysteinyl‐D‐valine (ACV, 99) and DAOCS complexed with its co‐substrate 2‐oxoglutarate

(2OG),[74] there was considerable interest in the elucidation of structures for an IPNS:product complex,[75, 76] and a DAOCS:2OG:substrate complex.[77] However the ‐lactams required for these experiments – penicillin N 85 as the substrate of DAOCS, IPN 86 as the product of IPNS catalysis – are prone to hydrolysis in the aqueous buffers used for co‐crystallisation.[73, 74] Thus cyclobutanones 83 and 84 were conceived as hydrolytically stable analogues of penams 85 and 86, for co‐crystallisation with DAOCS and IPNS respectively.

The L‐‐aminoadipoyl cyclobutanone analogue 84 was successfully co‐crystallised with IPNS, proving to be an effective isostere of IPN 86 in binding to IPNS, and stable to the conditions required for crystallisation.[31] By virtue of the racemic approach used to construct its bicyclic core

(Scheme 17), 84 was introduced to IPNS as a mixture of diastereomers (prepared by coupling enantiopure L‐‐aminoadipic acid with the racemic cyclobutanol derived from 91). The protein

23 selected the required stereoisomer from the crystallisation buffer to afford an IPNS‐Fe(II)‐84 complex (Figure 2) which closely mirrors the structure of IPNS‐Fe(II)‐IPN.[75] The bicyclic ring system takes up a very similar aspect in both structures, with the L‐‐aminoadipoyl carboxylate forming a salt bridge to Arg87 and the exo‐4‐carboxylate hydrogen bonding to several proximate protein sidechains. The bicyclic ring is more puckered in cyclobutanone 84 than in IPN 86, a consequence of replacing the β‐lactam nitrogen with carbon, and this positions that exo‐4‐ carboxylate at a steeper angle relative to the plane of the molecule. The active site iron atom is bound by the usual triad of amino acid residues from the protein (His214, Asp216 and His270) and a water molecule occupies the site trans to His214 in the IPNS‐Fe(II)‐84 complex, as in previous

IPNS crystal structures.[75, 78‐80]

Figure 2: Stereo representations showing the active site region of the IPNS–Fe(II)–84 complex: A. The IPNS–Fe(II)–84 complex; B. an overlay structure showing IPN 86 and analogue 84 at the IPNS active site. Reproduced with permission from Stewart et al. ChemBioChem 2007, 8, 2003‐2007.

As it transpired, the challenge of generating crystalline complexes of IPNS and DOACS with hydrolytically labile ‐lactams was resolved in different ways: using high‐pressure oxygenation to trigger turnover of ACV 99 and other substrates in the IPNS active site,[75, 76, 81] and by soaking

24 alternative penicillin substrates (penicillin G and ) into pre‐grown crystals of apo‐DAOCS while carefully controlling temperature, pH and soak time.[77] Nonetheless the co‐crystallisation of cyclobutanone 84 with IPNS afforded further insight into the geometry of product binding at the active site, in particular, changes to the position of amino acid sidechains (e.g. Val100, Phe211) associated with substrate turnover and product departure from the active site.[31] The IPNS‐Fe(II)‐

IPN structure solved after reacting ACV 99 in crystallo was complicated by residual electron density due to unreacted ACV 99, still present in ca. 30% occupancy, a convolution not present in the IPNS‐Fe(II)‐84 structure. The co‐crystallisation of cyclobutanone 84 with IPNS also demonstrated conclusively the utility of such compounds as structural analogues of bicyclic penams and mechanistic probes of ‐lactam biosynthesis.


Although considerable effort has been invested in the synthesis of cyclobutanone ‐lactam analogues and establishing them as effective ‐lactamase inhibitors, this has not yet been achieved. Cyclobutanones are the first class of reversible inhibitors to deliver time‐dependent inhibition of all classes of ‐lactamase, inhibitory activity for which a plausible unified mechanism of action has been proposed and partially demonstrated. However the activity of the analogues reported to date is inconsistent and generally modest, and there appears considerable scope for optimisation. There are potential selectivity issues to be explored too, given the demonstrated capacity of other cyclobutanone derivatives to act on some human proteases.

Overall, one is left with a sense that the full potential of cyclobutanone analogues of ‐lactam antibiotics has not yet been fully tapped. The synthesis of these compounds is challenging. As a result only a relatively small number of analogues have been prepared to date. Most of the analogues that have been made are simplified structures lacking the pendant functionality and second ring diversity that is seen in bioactive ‐lactams. This stands in vivid contrast to the myriad

25 natural and semi‐synthetic variants of all the major subfamilies of ‐lactams that have been explored (Scheme 1), exploiting the promiscuity of key biosynthetic enzymes such as IPNS, DAOCS and isopenicillin N acyltransferase.[15]

Furthermore, the generally underwhelming results of initial screening assays against transpeptidases and ‐lactamases have conspired to discourage further investigation and development of more efficient synthetic routes to more diversely functionalised cyclobutanone analogues. It is tempting to conclude that should a robust and efficient general route be developed, enabling access to an array of more complex cyclobutanones functionalised with the same diverse array of sidechains and second rings as known for bioactive ‐lactams, this would at last enable a comprehensive investigation of this compound class, and in particular, evaluation of their potential to work synergistically with existing antibiotics against ‐lactam‐resistant bacteria, in the ongoing and increasingly urgent battle against antimicrobial resistance.


PD was supported by the Henry Bertie and Florence Mabel Gritton Postgraduate Research

Scholarship and a University of Sydney World Scholars Award. PJR thanks Professor David Spring,

The Department of Chemistry, and Trinity College Cambridge for supporting a sabbatical visit to the University of Cambridge, Cambridge, UK.


[1] J. O'Neill, Review on Antimicrobial Resistance. Antimicrobial Resistance: Tackling a Crisis for the Health and Wealth of Nations, https://amr‐ 2014. [2] M. Woolhouse, J. Farrar, Nature 2014, 509, 555–557 [3] World Health Organisation, Antimicrobial resistance: Global report on surveillance WHO, 2014. [4] Department of Health (UK), Annual Report of the Chief Medical Officer 2011: Volume Two ,‐medical‐officer‐annual‐report‐volume‐2 2013. [5] A. H. Holmes, L. S. P. Moore, A. Sundsfjord, M. Steinbakk, S. Regmi, A. Karkey, P. J. Guerin, L. J. V. Piddock, Lancet 2016, 387, 176‐187. [6] A. R. McCullough, S. Parekh, J. Rathbone, C. B. Del Mar, T. C. Hoffmann, J. Antimicrob. Chemother. 2016, 71, 27‐33.

26 [7] R. Laxminarayan, P. Matsoso, S. Pant, C. Brower, J. A. Rottingen, K. Klugman, S. Davies, Lancet 2016, 387, 168‐175. [8] C. Walsh, Nature Rev. Microbiol. 2003, 1, 65‐70. [9] J. F. Fisher, S. O. Meroueh, S. Mobashery, Chem. Rev. 2005, 105, 395‐424. [10] D. M. Livermore, J. Antimicrob. Chemother. 2009, 64, i29‐i36. [11] A. Fleming, Brit. J. Exp. Pathol. 1929, 10, 226‐236. [12] R. P. Elander, Appl. Microbiol. Biotechnol. 2003, 61, 385‐392. [13] D. J. Waxman, J. L. Strominger, Annu. Rev. Biochem. 1983, 52, 825‐869. [14] A. Zapun, C. Contreras‐Martel, T. Vernet, FEMS Microbiol. Rev. 2008, 32, 361‐385. [15] R. B. Hamed, J. R. Gomez‐Castellanos, L. Henry, C. Ducho, M. A. McDonough, C. J. Schofield, Nat. Prod. Rep. 2013, 30, 21‐107. [16] J. D. Buynak, Expert Opin. Ther. Pat. 2013, 23, 1469‐1481. [17] G. S. Singh, Mini. Rev. Med. Chem. 2004, 4, 93‐109. [18] J. M. T. Hamilton‐Miller, J. Antimicrob. Chemother. 1999, 44, 729‐734. [19] W. Qin, M. Panunzio, S. Biondi, Antibiotics 2014, 3, 193‐215. [20] C. N. Wivagg, R. P. Bhattacharyya, D. T. Hung, J. Antibiot. 2014, 67, 645‐654. [21] G. Lowe, S. Swain, J. Chem. Soc., Chem. Commun. 1983, 1279‐1281. [22] E. M. Gordon, J. Pluscec, M. A. Ondetti, Tetrahedron Lett. 1981, 22, 1871‐1874. [23] O. Meth‐Cohn, A. J. Reason, S. M. Roberts, J. Chem. Soc., Chem. Commun. 1982, 90‐92. [24] G. Lowe, S. Swain, J. Chem. Soc., Perkin Trans. 1 1985, 391‐398. [25] H. W. Pauls, B. Cheng, L. S. Reid, Bioorg. Chem. 1992, 20, 124‐134. [26] G. Yang, L. Ghosez, Eur. J. Org. Chem. 2009, 1738‐1748. [27] J. W. Johnson, D. P. Evanoff, M. E. Savard, G. Lange, T. R. Ramadhar, A. Assoud, N. J. Taylor, G. I. Dmitrienko, J. Org. Chem. 2008, 73, 6970‐6982. [28] J. W. Johnson, M. Gretes, V. J. Goodfellow, L. Marrone, M. L. Heynen, N. C. J. Strynadka, G. I. Dmitrienko, J. Am. Chem. Soc. 2010, 132, 2558‐2560. [29] D. H. Martyres, J. E. Baldwin, R. M. Adlington, V. Lee, M. R. Probert, D. J. Watkin, Tetrahedron 2001, 57, 4999‐5007. [30] A. C. Ferguson, R. M. Adlington, D. H. Martyres, P. J. Rutledge, A. Cowley, J. E. Baldwin, Tetrahedron 2003, 59, 8233‐8243. [31] A. C. Stewart, I. J. Clifton, R. M. Adlington, J. E. Baldwin, P. J. Rutledge, ChemBioChem 2007, 8, 2003‐ 2007. [32] N. Armoush, P. Syal, D. P. Becker, Synth. Commun. 2008, 38, 1679‐1687. [33] L. N. Jungheim, R. J. Ternansky, in The Chemistry of ‐Lactams (Ed.: M. I. Page), Springer Netherlands, Dordrecht, 1992, pp. 306‐324. [34] M. I. Page, Ed., The Chemistry of ‐lactams, Blackie: Glasgow, 1992. [35] S. M. Ali, T. V. Lee, S. M. Roberts, Synthesis 1977, 155‐166. [36] M. Rey, S. M. Roberts, A. Dieffenbacher, A. S. Dreiding, Helv. Chim. Acta 1970, 53, 417‐432. [37] Z. Grudzinski, S. M. Roberts, J. Chem. Soc., Perkin Trans. 1 1975, 1767‐1773. [38] G. D. Han, Z. Y. Ma, Chinese J. Org. Chem. 2009, 29, 1001‐1017. [39] L. D. Cama, B. G. Christensen, Tetrahedron Lett. 1978, 4233‐4236. [40] D. Seebach, B. Lamatsch, R. Amstutz, A. K. Beck, M. Dobler, M. Egli, R. Fitzi, M. Gautschi, B. Herradon, a. et, Helv. Chim. Acta 1992, 75, 913‐934. [41] J. R. Cagnon, F. Le Bideau, J. Marchand‐Brynaert, L. Ghosez, Tetrahedron Lett. 1997, 38, 2291‐2294. [42] L. Ghosez, G. Yang, J. R. Cagnon, F. Le Bideau, J. Marchand‐Brynaert, Tetrahedron 2004, 60, 7591‐ 7606. [43] P. Bisel, E. Breitling, A. W. Frahm, Eur. J. Org. Chem. 1998, 1998, 729‐733. [44] M. S. Lall, Y. K. Ramtohul, M. N. G. James, J. C. Vederas, J. Org. Chem. 2002, 67, 1536‐1547. [45] D. J. Aitken, P. Caboni, H. Eijsberg, A. Frongia, R. Guillot, J. Ollivier, P. P. Piras, F. Secci, Adv. Synth. Catal. 2014, 356, 941‐945. [46] A. Frongia, N. Melis, I. Serra, F. Secci, P. P. Piras, P. Caboni, Asian. J. Org. Chem. 2014, 3, 378‐381. [47] A. J. Cocuzza, G. A. Boswell, Tetrahedron Lett. 1985, 26, 5363‐5366. [48] G. Lange, M. E. Savard, T. Viswanatha, G. I. Dmitrienko, Tetrahedron Lett. 1985, 26, 1791‐1794.

27 [49] P. Yuan, M. R. Driscoll, S. J. Raymond, D. E. Hansen, R. A. Blatchly, Tetrahedron Lett. 1994, 35, 6195‐ 6198. [50] M. Frigerio, M. Santagostino, S. Sputore, G. Palmisano, J. Org. Chem. 1995, 60, 7272‐7276. [51] N. Datta, M. H. Richmond, Biochem. J. 1966, 98, 204‐209. [52] J. M. Frere, J. M. Ghuysen, H. R. Perkins, M. Nieto, Biochem. J. 1973, 135, 463‐468. [53] P. Nordmann, G. Cuzon, T. Naas, Lancet Infect. Dis. 2009, 9, 228‐236. [54] T. R. Walsh, M. A. Toleman, L. Poirel, P. Nordmann, Clin. Microbiol. Rev. 2005, 18, 306‐325. [55] G. V. Crichlow, A. P. Kuzin, M. Nukaga, K. Mayama, T. Sawai, J. R. Knox, Biochemistry 1999, 38, 10256‐10261. [56] M. Paetzel, F. Danel, L. de Castro, S. C. Mosimann, M. G. P. Page, N. C. J. Strynadka, Nat. Struct. Mol. Biol. 2000, 7, 918‐925. [57] D. Golemi, L. Maveyraud, S. Vakulenko, J.‐P. Samama, S. Mobashery, Proc. Natl. Acad. Sci. USA 2001, 98, 14280‐14285. [58] D. O. Shah, D. G. Gorenstein, Biochemistry 1983, 22, 6096‐6101. [59] R. L. Stein, A. M. Strimpler, P. D. Edwards, J. J. Lewis, R. C. Mauger, J. A. Schwartz, M. M. Stein, D. A. Trainor, R. A. Wildonger, M. A. Zottola, Biochemistry 1987, 26, 2682‐2689. [60] L. A. Reiter, G. J. Martinelli, L. A. Reeves, P. G. Mitchell, Bioorg. Med. Chem. Lett. 2000, 10, 1581‐ 1584. [61] M. H. Gelb, J. P. Svaren, R. H. Abeles, Biochemistry 1985, 24, 1813‐1817. [62] B. Imperiali, R. H. Abeles, Biochemistry 1986, 25, 3760‐3767. [63] M. Mileni, J. Garfunkle, C. Ezzili, F. S. Kimball, B. F. Cravatt, R. C. Stevens, D. L. Boger, J. Med. Chem. 2010, 53, 230‐240. [64] T. Tamada, T. Kinoshita, K. Kurihara, M. Adachi, T. Ohhara, K. Imai, R. Kuroki, T. Tada, J. Am. Chem. Soc. 2009, 131, 11033‐11040. [65] J. C. Powers, P. M. Tuhy, Biochemistry 1973, 12, 4767‐4774. [66] H. C. Brown, R. S. Fletcher, R. B. Johannesen, J. Am. Chem. Soc. 1951, 73, 212‐221. [67] D. A. Trainor, Trends Pharmacol. Sci. 1987, 8, 303‐307. [68] C. H. Hassall, W. H. Johnson, A. J. Kennedy, N. A. Roberts, FEBS Lett. 1985, 183, 201‐205. [69] E. F. Meyer, Jr., R. Radhakrishnan, G. M. Cole, L. G. Presta, J. Mol. Biol. 1986, 189, 533‐539. [70] L. H. Takahashi, R. Radhakrishnan, R. E. Rosenfield, Jr., E. F. Meyer, Jr., D. A. Trainor, M. Stein, J. Mol. Biol. 1988, 201, 423‐428. [71] J. E. Baldwin, C. Schofield, in The Chemistry of ‐Lactams (Ed.: M. I. Page), Springer Netherlands, Dordrecht, 1992, pp. 1‐78. [72] P. J. Rutledge, in 2‐Oxoglutarate‐Dependent Oxygenases (Eds.: R. P. Hausinger, C. J. Schofield), The Royal Society of Chemistry, 2015, pp. 414‐424. [73] P. L. Roach, I. J. Clifton, C. M. H. Hensgens, N. Shibta, C. J. Schofield, J. Hajdu, J. E. Baldwin, Nature 1997, 387, 827‐830. [74] K. Valegard, A. C. T. van Scheltinga, M. D. Lloyd, T. Hara, S. Ramaswamy, A. Perrakis, A. Thompson, H.‐J. Lee, J. E. Baldwin, C. J. Schofield, J. Hajdu, I. Andersson, Nature 1998, 394, 805‐809. [75] N. I. Burzlaff, P. J. Rutledge, I. J. Clifton, C. M. Hensgens, M. Pickford, R. M. Adlington, P. L. Roach, J. E. Baldwin, Nature 1999, 401, 721‐724. [76] J. M. Elkins, P. J. Rutledge, N. I. Burzlaff, I. J. Clifton, R. M. Adlington, P. L. Roach, J. E. Baldwin, Org. Biomol. Chem. 2003, 1, 1455‐1460. [77] K. Valegard, A. C. Terwisscha van Scheltinga, A. Dubus, G. Ranghino, L. M. Oester, J. Hajdu, I. Andersson, Nat. Struct. Mol. Biol. 2004, 11, 95‐101. [78] A. R. Grummitt, P. J. Rutledge, I. J. Clifton, J. E. Baldwin, Biochem. J. 2004, 382, 659‐666. [79] A. R. Howard‐Jones, P. J. Rutledge, I. J. Clifton, R. M. Adlington, J. E. Baldwin, Biochem. Biophys. Res. Commun. 2005, 336, 702‐708. [80] W. Ge, I. J. Clifton, A. R. Howard‐Jones, J. E. Stok, R. M. Adlington, J. E. Baldwin, P. J. Rutledge, ChemBioChem 2009, 10, 2025‐2031. [81] P. J. Rutledge, N. I. Burzlaff, J. M. Elkins, M. Pickford, J. E. Baldwin, P. L. Roach, Anal. Biochem. 2002, 308, 265‐268.

28 Table of Contents

Cyclobutanone is isosteric with ‐lactam, the eponymous cyclic amide replaced by an all‐carbon ring. Cyclobutanones have been investigated as ‐lactam analogues since the 1980s, and were the first class of inhibitor to demonstrate activity against all classes of ‐lactamase. But challenges with the synthesis of appropriately functionalised cyclobutanone β‐lactam analogues mean that their potential has yet to be fully tapped: their activity as ‐lactamase inhibitors is not yet optimised, their wider potential not fully explored.

29 Author Biographies

Dr Peter Rutledge received his BSc and MSc from the University of Auckland and his DPhil from the University of Oxford, where he worked with Professor Sir Jack Baldwin on the mechanism of penicillin biosynthesis. He held a post‐doctoral fellowship at the Dyson Perrins Laboratory in Oxford then a lectureship in the Centre for Synthesis and Chemical Biology at University College Dublin before moving to the University of Sydney in 2006. His current research interests include antibiotics discovery, natural products chemistry, molecular probes and sensors, biosynthesis and biocatalysis.

Dr Prarthana Devi completed her BSc from Dibrugarh University, India and MSc from the Indian Institute of Technology Guwahati. She worked in several research institutes and pharmaceutical companies in India before moving to Australia to pursue her PhD at The University of Sydney. She worked with Dr Peter Rutledge on the development of new synthetic pathways to generate cyclobutanone ‐lactam analogues as antibiotics and ‐lactamase inhibitors.