Synthesis and Antiproliferative Evaluation of 3-Chloroazetidin-2-ones with Antimitotic Activity: Heterocyclic Bridged Analogues of Combretastatin A-4

Antimitotic drugs that target tubulin are among the most widely used chemotherapeutic agents; however, the development of multidrug resistance has limited their clinical activity. We report the synthesis and biological properties of a series of novel 3-chloro-β-lactams and 3,3-dichloro-β-lactams (2-azetidinones) that are structurally related to the tubulin polymerisation inhibitor and vascular targeting agent, Combretastatin A-4. These compounds were evaluated as potential tubulin polymerisation inhibitors and for their antiproliferative effects in breast cancer cells. A number of the compounds showed potent activity in MCF-7 breast cancer cells, e.g., compound 10n (3-chloro-4-(3-hydroxy-4-methoxy-phenyl)-1-(3,4,5-trimethoxyphenyl)azetidin-2-one) and compound 11n (3,3-dichloro-4-(3-hydroxy-4-methoxyphenyl)-1-(3,4,5-trimethoxyphenyl)-azetidin-2-one), with IC50 values of 17 and 31 nM, respectively, and displayed comparable cellular effects to those of Combretastatin A-4. Compound 10n demonstrated minimal cytotoxicity against non-tumorigenic HEK-293T cells and inhibited the in vitro polymerisation of tubulin with significant G2/M phase cell cycle arrest. Immunofluorescence staining of MCF-7 cells confirmed that β-lactam 10n caused a mitotic catastrophe by targeting tubulin. In addition, compound 10n promoted apoptosis by regulating the expression of pro-apoptotic protein BAX and anti-apoptotic proteins Bcl-2 and Mcl-1. Molecular docking was used to explore the potential molecular interactions between novel 3-chloro-β-lactams and the amino acid residues of the colchicine binding active site cavity of β-tubulin. Collectively, these results suggest that 3-chloro-2-azetidinones, such as compound 10n, could be promising lead compounds for further clinical anti-cancer drug development.


Introduction
Microtubules play an essential role in many cellular functions, such as cell division and mitosis, and are investigated as attractive drug targets in anti-cancer chemotherapy. Many Trimethoxyphenyl-1,2,3-triazole hybrids, such as 4b, containing the coumarin fragment, inhibit human gastric MGC803 cancer cell growth, induce G2/M phase arrest by down-regulating the expression of CDK1, promote apoptosis by regulating Death Receptor 5 (DR5) and the Bcl-2 family of proteins and inhibit tubulin polymerisation by interacting with the colchicine site [23]. The quinaldinyl-iso-carbazolyl compound 5a is more active than CA-4 1a and isoCA-4 3c against A549, lung adenocarcinoma epithelial cells It is interesting to observe that the orally active diketopiperazine plinabulin 7 selectively targets and binds to the colchicine-binding site of tubulin [28], blocks tumour growth [29] and provides early protection against severe neutropenia induced by chemotherapy in patients with advanced NSCLC (Figure 1) [30]. BKM120 (Buparlisib) 8 is one of the most advanced phosphoinositide 3-kinase (PI3K) inhibitors for the treatment of cancer, but it has been shown to also interfere with microtubule polymerisation as an off-target effect [31]. Although considerable progress has been achieved in the discovery of targeted cancer therapies, both innate and acquired mechanisms of resistance are commonly observed for many successful cancer drugs [32].
We have reported the synthesis, antiproliferative and tubulin-binding effects of a series of 2-azetidinones (β-lactams), containing the structural features of CA-4, while retaining the necessary cis configuration of Rings A and B [33,34]. β-Lactam compounds containing aryl and heterocycles such as thiophene located at C-3 were found to be particularly effective [35][36][37]. In addition, both the anti-angiogenic and anti-migratory effects observed in MDA-MB-231 breast adenocarcinoma cells suggest a potential anti-metastatic role for these compounds [38]. We have identified the β-lactam heterocycle as a potential scaffold for the development of new anti-tumour agents and wished to establish the structural requirements for substituents at C-3.
We now report a series of novel 3-chloro-2-azetidinone and 3,3-dichloro-2-azetidinone compounds with an interesting profile, particularly in triple negative breast cancer, which could be considered for potential development as tubulin destabilising agents in preclinical studies of breast cancer (see Figure 1, target structures). A library of 1,4-diarylazetidin-2-ones that contain halogen substituents chloro, dichloro or bromo at C-3 was prepared for evaluation. The β-lactam ring forms a rigid scaffold for the hydrophobic CA-4 aryl rings A and B required for interaction with the colchicine binding site of tubulin. The effect of these C-3 halogen substituents on the biological activity of these compounds when the cis configuration (Rings A and B) is constrained into the 4-membered azetidin-2-one ring structure was investigated. The synthesis of the phosphate ester prodrugs of the most potent 3-chloro-1,4-diarylazetidin-2-one was also examined, to increase the potential bioavailability of the compound. The introduction of this halogen substituent at C-3 also allowed us to examine potential structure-activity relationships for the series, and to rationalise the effect of the introduction of the C-3 chlorine on the interaction with the colchicine binding site. We have now investigated a new series of novel 3-halo-2-azetidinone compounds with an improved biochemical profile, e.g., in triple negative breast cancer for potential development in the treatment of breast cancer as tubulin destabilising agents. These novel heterocyclic structures were further investigated for their effects on cell viability, cell cycle and tubulin polymerisation in MCF-7 breast cancer cells. 9a: R 1 = R 2 =R 3 =H 9b: R 1 =Cl, R 2 =R 3 =H 9c: R 1 =Br, R 2 =R 3 =H 9d: R 1 =NO 2 , R 2 =R 3 =H 9e: R 1 =OCH 3 , R 2 =R 3 =H 9f: R 1 =OCH 2 CH 3 , R 2 =R 3 =H 9g: R 1 =OPh, R 2 =R 3 =H 9h: R 1 =OCH 2 Ph, R 2 =R 3 =H 9i: R 1 =SCH 3 , R 2 =R 3 =H 9j: R 1 =H, R 2 R 3 =CH=CH-CH=CH 9k: R 1 R 2 =CH=CH-CH=CH, R 3 =H 9l: R 1 =OCH 3 , R 2 =OH, R 3 =H 9m: R 1 =OCH 3 , R 2 =OTBDMS, R 3 =H 9n: R 1 =OCH 3 ; R 2 =NO 2, R 3 =H 9o: R 1 =SCH 2 CH 3 , R 2 =R 3 =H 9p: R 1 =OCH 3 , R 2 = CH 3 , R 3 =H 9q: R 1 =OCH 3 , R 2 = F, R 3 =H 9r: R 1 =OCH 3 , R 2 = Cl, R 3 =H 9s: R 1 =OCH 3   The Staudinger reaction ([2+2] ketene-imine cycloaddition reaction) is a v method for the synthesis of 2-azetidinones. Ketenes are usually formed in situ by tion of acyl halides with tertiary amines. However, 2-azetidinones are also directly sible from imines and carboxylic acids via mixed anhydrides [65], using activating such as methoxymethylene-N,N-dimethyliminium salt, [66], the Vilsmeier reage Mukaiyama reagent and triphosgene [67]. 3-Chloro-3-thioaryl-β-lactams, obtained chlorination of 3-thioaryl-β-lactams with sulfuryl chloride, are reported as suitab strates for Lewis acid catalysed nucleophilic substitution reactions [68]. In the work, the Staudinger reaction of the imines 9a-9k and 9m-9s with chloroacetyl chlo the presence of triethylamine (Scheme 2) afforded the β-lactam products 10a-m a as racemic mixtures in yields of 3-57%. Compound 10l was also obtained in a rea the imine 9l with chloroacetic acid with triphosgene as the acid activating agent. tection of the silyl ether 10l with TBAF afforded the phenolic product 10n. Low y some of the Staudinger reactions were due to the degradation of the imine com observed in these reaction mixtures.
The additional 3-chloro-β-lactam products 12a-12c, 14a and 14b were also o in a similar Staudinger reaction of the appropriate imines 9t-9v, 9w and 9x with c cetyl chloride. The coupling constant for H-3 and H-4 of the azetidinone ring is ge useful in the identification of the stereochemistry of 2-azetidinones, with J3,4 usual 6 Hz for the cis and J3,4 1-2 Hz for the trans stereoisomers. The structural assignme monochloro β-lactam 10n was particularly interesting for the assignments of H-3 4 protons of the β-lactam compound. Two doublet signals were observed in the 1 H spectrum for H-4 and H-3 at δ 4.61 and 4.89 ppm, respectively (J3,4 = 1.52 Hz), whi firmed the trans isomer assignment) and were more downfield due to the adjacent e withdrawing chlorine substituent. In the 13 C NMR spectrum of 10n, the correspond and C4 carbons of β-lactam appeared at 62.67 (C3) and 65.63 (C4) ppm, respectivel the characteristic signal of carbonyl group of the β-lactam at δ 160.22 ppm. Th pounds were isolated exclusively as the trans isomer; the only exception was com 10e, where both the trans and cis compounds were obtained (trans:cis ratio 1.9:1; J3 Hz cis, J3,4 = 2 Hz trans).  Single crystal X-ray analysis was obtained for compounds 10e and 10o (recrystallised from dichloromethane/n-hexane), and the crystal structure is shown in Figure 3. The crystal data and structure refinement for the 3-chloro compounds 10e and 10o are displayed in Tables 1 and 2. The trans stereochemistry for compounds 10e and 10o was confirmed from the X-ray crystal structures. The aryl rings at N-1 (Ring A) and C-4 are in a non- The additional 3-chloro-β-lactam products 12a-12c, 14a and 14b were also obtained in a similar Staudinger reaction of the appropriate imines 9t-9v, 9w and 9x with chloroacetyl chloride. The coupling constant for H-3 and H-4 of the azetidinone ring is generally useful in the identification of the stereochemistry of 2-azetidinones, with J 3,4 usually of 4-6 Hz for the cis and J 3,4 1-2 Hz for the trans stereoisomers. The structural assignment Pharmaceuticals 2021, 14, 1119 9 of 53 of 3-monochloro β-lactam 10n was particularly interesting for the assignments of H-3 and H-4 protons of the β-lactam compound. Two doublet signals were observed in the 1 H NMR spectrum for H-4 and H-3 at δ 4.61 and 4.89 ppm, respectively (J 3,4 = 1.52 Hz), which confirmed the trans isomer assignment) and were more downfield due to the adjacent electron withdrawing chlorine substituent. In the 13 C NMR spectrum of 10n, the corresponding C3 and C4 carbons of β-lactam appeared at 62.67 (C 3 ) and 65.63 (C 4 ) ppm, respectively, with the characteristic signal of carbonyl group of the β-lactam at δ 160.22 ppm. The compounds were isolated exclusively as the trans isomer; the only exception was compound 10e, where both the trans and cis compounds were obtained (trans:cis ratio 1.9:1; J 3,4 = 5.00 Hz cis, J 3,4 = 2 Hz trans).
Single crystal X-ray analysis was obtained for compounds 10e and 10o (recrystallised from dichloromethane/n-hexane), and the crystal structure is shown in Figure 3. The crystal data and structure refinement for the 3-chloro compounds 10e and 10o are displayed in Tables 1 and 2. The trans stereochemistry for compounds 10e and 10o was confirmed from the X-ray crystal structures. The aryl rings at N-1 (Ring A) and C-4 are in a non-coplanar cis-like arrangement, while the phenyl ring at C4 (Ring B) and the 3-chloro substituent are in a trans configuration on opposite sides of the β-lactam ring. For compound 10e, the distance between the centroid of ring A and ring B is~5.2 Å, while the distance between ring B and the chloro group is~5.0 Å. For compounds 10e and 10o, the torsional angle value ring B/C was calculated as 116.0(1) • and 112.0(3) • , respectively, which is consistent with the small trans coupling constant observed in the 1 H NMR spectrum of 2.00 and 2.20 Hz, respectively, for H3/H4 in these compounds. The β-lactam C=O bond lengths are 1.2122(18) Å and 1.2122 (19) Å for compounds 10e and 10o, respectively, which is consistent with the data previously reported for the carbonyl bond length of monocyclic β-lactams of 1.217(3) Å [69] and 1.207(2) Å [70]. The torsional angles (Ring A/B) observed for compounds 10e and 10o were calculated to be 60.7(2) • and 62.7(2) • , respectively; these values are slightly greater than the corresponding torsional angles for Ring A/B of 55 • and 53 • reported in DAMAcolchicine and CA-4, respectively [6,71]. The numbers in parentheses refer to the second crystallographically independent molecule in the asymmetric unit.     Compound Structure X-Ray representation 10e 10o 11o 16g Figure 3. ORTEP representation of the X-ray crystal structure of compounds 10e, 10o, 11o and 16g with heteroatoms labelled and thermal ellipsoids set at 50% probability. Structure 11o shows the majority occupied disordered 4-methoxyphenyl moiety (81% occupied). For structures 10o and 16g, the 3 position is substituted by both chloride (75%) and bromide (25%); distances restrained, and atomic displacement constrained.
The most potent compound in this series was identified as 10n, with the characteristic CA-4 3-hydroxy-4-methoxyphenyl Ring B substitution (IC50 = 0.017 μM), which compares favourably with CA-4 (IC50 0.004 μM). Analysis of the results from the 3,3-dichloro compound series 11a-n showed that the compounds displayed a similar SAR profile to the 3chloro compounds but with reduced potency, with compounds 11e, 11f, 11i, 11k and 11m displaying IC50 values in the range 0.119-0.353 μM. The most potent compound in this series was identified as 11n, again with the characteristic CA-4 3-hydroxy-4-methoxyphenyl Ring B substitution (IC50 = 0.031 μM). Interestingly, the prodrug 18 (the phosphate Figure 3. ORTEP representation of the X-ray crystal structure of compounds 10e, 10o, 11o and 16g with heteroatoms labelled and thermal ellipsoids set at 50% probability. Structure 11o shows the majority occupied disordered 4-methoxyphenyl moiety (81% occupied). For structures 10o and 16g, the 3 position is substituted by both chloride (75%) and bromide (25%); distances restrained, and atomic displacement constrained.
As an extension to this study, a further series of 3,3-dichloro-β-lactams (11a-11o, 13a-c, 15a, 15b) with similar substituents in rings A and B as that described for the 3-chloro β-lactams was prepared (Scheme 2). The phenolic product 11n was obtained by treating the silyl ether 11l with TBAF. The compounds were obtained in moderate yields (27-63%), apart from compounds 15a and 15b (with 3,5-dimethoxy substitution pattern in the aryl A ring), which were isolated in low yields, 13 and 7%, respectively. When comparing the 3,3-dichloro-β-lactams with the mono chloro β-lactam compounds, the 1 H NMR spectrum for the 3,3-dichloro β-lactam compound 11n is relatively simple; the characteristic signal for 3,3-dichloro β-lactam compounds appeared as a singlet for the H4 proton at δ 5.39 ppm and it is further downfield than in the corresponding 10n because of the electron withdrawing properties of the two chlorine atoms. In the 13 C NMR spectrum, the high resonance signals at δ 83.68 and 73.54 ppm were assigned to C3 (with the dichloro substituent) and C4, respectively, while the resonances for C3 and C4 were observed at 62.67 (C 3 ) and 65.63 (C 4 ) ppm in the 3-monochloro compound 10n. The X-ray crystal structure and data for the 3,3-dichloro β-lactam compound 11o are presented in Figure 3 and Table 1 and show that rings A and B (located at N-1 and C-4 of the β-lactam ring) are not coplanar, with a torsional angle of 68.9 • (see Table 2). A significant difference in the distances from the centroid of ring B to each of the two C3 chloro substituents (3.7 Å and 5.4 Å) was observed. This may be relevant in rationalising the differences in the antiproliferative activity observed for the monochloro and dichloro-β-lactam compounds.
As the 3-chloro-β-lactams exhibited an excellent antiproliferative profile, a further series of related 3-bromo β-lactams was investigated by using the Staudinger procedure with bromoacetyl chloride (Scheme 3). The preparation of 3-bromo β-lactams using bromoacetyl bromide [72], bromination of 3-azetidinone [48] and a ring expansion of aziridines with triphenylphosphine/NBS or triphenylphosphine dibromide has been reported [73]. The 3-bromo β-lactams (16a-16i) were initially obtained as a mixture with corresponding 3-chloro-β-lactams in a ratio of 1:2 in most cases in yields of 5-31%, due to the halogen exchange with the chlorinated solvent (dichloromethane), following purification by either recrystallisation or from the gradient column chromatography. The presence of the trans isomer of 3-bromo-β-lactam 16a was confirmed from the 1 H NMR spectrum that shows H-3 at δ 5.05 and H-4 at δ 4.65 (J 1.96 Hz) (in comparison, H-3 and H-4 of the 3-chloro β-lactam 10e were observed at δ 4.95 and δ 4.63, J 2.00 Hz). The phenolic product 16j was obtained by treating the silyl ether 19i with TBAF. The asymmetric synthesis of 16j (3S,4S) was previously reported [48]. The X-ray crystal structure and data for the 3-bromo-β-lactam compound 16g ( Figure 3, Table 1) again demonstrates that rings A and B (located at N-1 and C-4 of the β-lactam ring) are not coplanar (see Table 2 To improve the solubility and bioavailability of the compounds, the phenol 11n w selected for phosphate ester prodrug preparation. The esterification of the phenol with dibenzyl phosphite using diisopropylethylamine and dimethylaminopyridine forded dibenzyl phosphate β-lactam ester 17 (Scheme 4 To improve the solubility and bioavailability of the compounds, the phenol 11n was selected for phosphate ester prodrug preparation. The esterification of the phenol 11n with dibenzyl phosphite using diisopropylethylamine and dimethylaminopyridine afforded dibenzyl phosphate β-lactam ester 17 (Scheme 4). The subsequent hydrogenation of 17 with a palladium/carbon catalyst removed the dibenzyl protecting groups to afford the phosphate ester 18 in a 75% yield, while the β-lactam ring remained intact. To improve the solubility and bioavailability of the compounds, the phenol 11n was selected for phosphate ester prodrug preparation. The esterification of the phenol 11n with dibenzyl phosphite using diisopropylethylamine and dimethylaminopyridine afforded dibenzyl phosphate β-lactam ester 17 (Scheme 4). The subsequent hydrogenation of 17 with a palladium/carbon catalyst removed the dibenzyl protecting groups to afford the phosphate ester 18 in a 75% yield, while the β-lactam ring remained intact.
HPLC stability studies at three different pH systems were performed on a representative compound 16a to determine the stability at acidic pH 4, pH 7.4 and basic pH 9 (acid pH found in the stomach, basic found in the intestine and pH 7.4 in the plasma). The compound was stable at these buffered pH systems with a half-life of 18 h (pH 4), 20 h (pH 9) and 22 h (pH 7.4).   HPLC stability studies at three different pH systems were performed on a representative compound 16a to determine the stability at acidic pH 4, pH 7.4 and basic pH 9 (acid pH found in the stomach, basic found in the intestine and pH 7.4 in the plasma). The compound was stable at these buffered pH systems with a half-life of 18 h (pH 4), 20 h (pH 9) and 22 h (pH 7.4).

Activity of β-Lactam Compounds in MCF-7 Human Breast Cancer Cell Line
The antiproliferative potential of the β-lactams 10a-o, 11a-o, 12a-c, 13a-c, 18, 14a, 14b, 15a, 15b and 16a-h was initially evaluated in the CA-4 sensitive oestrogen receptor positive MCF-7 human breast cancer cell line. CA-4 was used as the control compound in the assay (Table 3), together with β-lactam compounds that we had previously reported [34,74]. The IC 50 value obtained for CA-4 (0.0039 µM for MCF-7) is in good agreement with reported values [75]. The introduction of the halogen substituent at C-3 was examined in an effort to investigate the effect on the activity of this substituent, and subsequently optimising the cytotoxic effects against MCF-7 human breast cancer cells. The most potent analogues in MCF-7 cells were further screened in the MDA-MB-231, Hs578T and Hs578Ts(i)8 triple negative breast cancer cell lines, multiple myeloma (U266), acute myeloid leukaemia (HL60) and colon cancer (HT-29 and SW480) cell lines using the AlamarBlue assay. Compounds were initially assessed for antiproliferative activity in MCF-7 cells to determine the structure-activity relationship for these halogenated compounds and to identify the most potent compounds to progress for further investigation. tween zero and two, lipophilicity (AlogP) appeared in the range 2.67-4.76 (apart from the 4-naphthyl compound 11k, AlogP 5.84) and the number of rotatable bonds in the range 5-8. The calculated TPSA was between 48 and 130 Å2, which suggested good intestinal absorption. The pharmacokinetics results indicate that these compounds satisfy the criteria for good drug likeness parameters and good bioavailability. The compounds were free from alerts for Pan Assay Interfering substances (PAINS) [76] and are predicted to have excellent drug-like properties (e.g., metabolic stability, permeability, blood-brain barrier partition, plasma protein binding and human intestinal absorption properties), which encouraged us to perform further in vitro studies.  Selected β-lactam compounds were next evaluated in the triple negative cell line MDA-MB-231 (Table 4). A total of 10-15% of breast cancers are classified as triple-negative breast cancers (TNBC) and include any breast cancer that does not express the genes for oestrogen and progesterone receptors (ER/PR) and HER2. In addition, MDA-MB-231 cells possess mutant p53. These cancers are difficult to treat since they are generally not responsive to hormone therapies such as the selective oestrogen receptor modulator (SERM) tamoxifen, or aromatase inhibitors such as anastrozole, or to the monoclonal antibody In the series of compounds with 3-chloro substituent 10a-o, a number of varied substituents were introduced at C-4 of Ring B, while retaining the 3,4,5-trimethoxy substitution for Ring A usually present in many colchicine binding-site type ligands [19]. The introduction of nitro (10d), chloro (10b) and bromo (10c) at C-4 of Ring B resulted in increased activity when compared with the unsubstituted 10a, with the 4-methoxy 10e and 4-thiomethyl 10i showing excellent potency (IC 50 34 and 73 nM, respectively [74]).
The cis isomer of 10e demonstrated a nine-fold reduction in activity (IC 50 0.317 µM) when compared with the trans isomer (IC 50 34 nM). The bulkier 4-phenoxy 10g and 4-benzyloxy 10h substituents resulted in significantly reduced activity (IC 50 64.07 and 59.91 µM, respectively). Interestingly, the 2-naphthyl 10k was much more potent (IC 50 0.20 µM) than the 1-naphthyl compound 10j (IC 50 14.66 µM), possibly due to the steric interference by the 1-naphthyl at the colchicine binding site, while the 2-naphthyl was more easily accommodated. Compounds 10m and 10o, with an additional substituent at the meta position of B ring (10m nitro and 10o chloro), retained moderate activity with IC 50 values of 3.088 and 0.433 µM, respectively.
The most potent compound in this series was identified as 10n, with the characteristic CA-4 3-hydroxy-4-methoxyphenyl Ring B substitution (IC 50 = 0.017 µM), which compares favourably with CA-4 (IC 50 0.004 µM). Analysis of the results from the 3,3-dichloro compound series 11a-n showed that the compounds displayed a similar SAR profile to the 3-chloro compounds but with reduced potency, with compounds 11e, 11f, 11i, 11k and 11m displaying IC 50 values in the range 0.119-0.353 µM. The most potent compound in this series was identified as 11n, again with the characteristic CA-4 3-hydroxy-4-methoxyphenyl Ring B substitution (IC 50 = 0.031 µM). Interestingly, the prodrug 18 (the phosphate ester of the phenol 11n) retained potent activity with IC 50 = 0.077 µM. The introduction of the 3,4,5-trimethoxyaryl ring A at C-4 of the β-lactam in both the 3-chloro and 3,3-dichloro series (compounds 12a-c, 13a-c) resulted in significant decrease in activity, e.g., compound 10e (IC 50 = 0.034 µM (trans) and 0.317 µM (cis)) compared with 12a (IC 50 = 14.81 µM). The poor activity of compounds 12a-c and 13a-c where the 3,4,5-trimethoxyphenyl group A is located at C-4 could be due to the bulkiness of Ring A, which is unable to fit correctly in the target binding pocket of tubulin, in agreement with previous reports findings [33].
In compounds 14a, 14b, 15a and 15b, the 3,5-dimethoxyaryl ring is located at N-1 of β-lactam and replaces the usual 3,4,5-trimethoxyaryl ring A. Compound 14b with the para OEt substituent in Ring B produces a remarkably better antiproliferative effect (IC 50 = 0.045 µM) than the para OMe for both 3-chloro and 3,3-dichloro compounds. 14b was identified as of particular interest and only slightly less potent than the corresponding 3,4,5-trimethoxy analogue 10e (IC 50 = 0.034 µM). The 3,4,5-trimethoxy substituted A Ring of CA-4 plays an important role in inhibiting tubulin polymerisation, confirmed by the crystal structure of CA-4 in tubulin [19]. It is interesting to see that the removal of the 4-methoxy group results in the retention of activity in the 3,5-dimethoxyaryl ring A compound 14b. The introduction of the 3-bromo substituent to replace the chloro at C-3 of the β-lactam resulted in a significant reduction in the antiproliferative effect of the compounds in the series, e.g., comparing compound 16a (IC 50 = 0.579 µM) with the corresponding 3-chloro compound 10e (IC 50 = 0.034 µM) resulted in a 17-fold decrease in activity.
The physicochemical properties and metabolic stability of the panel of compounds synthesised were evaluated to probe into the drug-relevant properties (see Supplementary Information Tables S1 and S2 for Tier 1 profiling screen). The physicochemical properties of the compounds complied with the Lipinski's rule of five, thus ensuring a good lipophilic-hydrophilic balance and adequate membrane permeability. Most of the compounds followed Lipinski and Veber rules, i.e., molecular weight ranges from 347 to 457, hydrogen bond acceptor range between four and nine, hydrogen bond donor range between zero and two, lipophilicity (AlogP) appeared in the range 2.67-4.76 (apart from the 4-naphthyl compound 11k, AlogP 5.84) and the number of rotatable bonds in the range 5-8. The calculated TPSA was between 48 and 130 Å 2 , which suggested good intestinal absorption. The pharmacokinetics results indicate that these compounds satisfy the criteria for good drug likeness parameters and good bioavailability. The compounds were free from alerts for Pan Assay Interfering substances (PAINS) [76] and are predicted to have excellent drug-like properties (e.g., metabolic stability, permeability, blood-brain barrier partition, plasma protein binding and human intestinal absorption properties), which encouraged us to perform further in vitro studies.

Antiproliferative Activity of Selected β-Lactam Compounds in the MDA-MB-231, Hs578T and Hs578Ts(i)8 Triple Negative Breast Cancer Cell Lines
Selected β-lactam compounds were next evaluated in the triple negative cell line MDA-MB-231 (Table 4). A total of 10-15% of breast cancers are classified as triple-negative breast cancers (TNBC) and include any breast cancer that does not express the genes for oestrogen and progesterone receptors (ER/PR) and HER2. In addition, MDA-MB-231 cells possess mutant p53. These cancers are difficult to treat since they are generally not responsive to hormone therapies such as the selective oestrogen receptor modulator (SERM) tamoxifen, or aromatase inhibitors such as anastrozole, or to the monoclonal antibody Herceptin, which targets the HER2 receptor. There are fewer treatment options available for TNBC compared with ER+, PR+ and HER2+ breast cancers and the prognosis is poorer [77,78]. The 3,3-dichloro-β-lactam 11n (with 3-hydroxy-4-methoxyphenyl Ring B substitution) was identified as the most potent with IC 50   Compound 10e was further evaluated in the triple-negative Hs578T breast cancer cell line together with its isogenic subclone Hs578Ts(i)8 cells. Hs578Ts(i)8 cells are three-fold more invasive than the parental cell line (Hs578T) and 2.5-fold more migratory. Hs578Ts(i)8 cells display enhanced invasive properties with 30% more CD44+/CD24-/low cells. They show an increased capacity to proliferate, migrate, invade through the extracellular matrix and produce tumours in nude mice [80]. Compound 10e demonstrated significant antiproliferative activity at nanomolar concentrations in Hs578T cells (IC 50 124 nM) with increased potency in the invasive Hs578Ts(i)8 cells (IC 50 = 61 nM). These results compared favourably with CA-4 (IC 50 = 8 nM in Hs578T and 20 nM in Hs578Ts(i)8 cells) and indicated the potential ability of the compound to inhibit tumour invasion and angiogenesis, which are characteristic features of tumour growth and metastasis in aggressive breast cancers.

Antiproliferative Activity of Selected β-Lactam Compounds in Multiple Myeloma (U266), Acute Myeloid Leukaemia (HL60) and Colon Cancer (HT-29 and SW480) Cell Lines
Compounds 10n and 10e were next evaluated for antiproliferation in multiple myeloma (U266) cells. Multiple myeloma, also known as plasma cell myeloma, is a malignant haematological disease characterised by the proliferation of clonal plasma cells predominantly in the bone marrow. U266 cells are considered to be the least sensitive multiple myeloma cells to nucleoside drug cladribine compared to RPMI8226 and MM1.S cells [81]. Compounds 10n and 10e as well as CA-4, demonstrated potent antiproliferative activity in the nanomolar range, with an IC 50 value of 77 nM for the 3-chloro compound 10e (with 4-methoxyphenyl Ring B substitution) and a more potent result for the 3-chloro analogue (with 3-hydroxy-4-methoxyphenyl Ring B substitution) 10n, with IC 50 = 31 nM, which compares favourably with CA-4 (IC 50 = 35 nM) in this cell line, Table 5. These results demonstrated the sensitivity of U266 cells toward CA-4 and its 3-chloroazetidinone analogues 10e and 10n.   (Table 5) and are in agreement with the reported values for CA-4 in MCF-7 and human breast cancer and leukaemia cell lines [74,82,83]. The corresponding values for 10e in these cell lines were 161 nM (HL-60), 55 nM (SW480) and 135 nM (HT-29). The 3,3-dichloro β-lactam 11n was slightly less potent than 10n in these cell lines with IC 50 values of 16 nM (HL-60), 44 nM (SW480) and 941 nM (HT-29), respectively. These results are interesting as the control drug CA-4 (IC 50 value of 4.165 µM) is much less active than compounds 10n and 11n in the chemoresistant HT-29 cell line. This effect may be due to the inactivation of CA-4 by glucuronidation in HT-29 cells, as previously reported [74]. SW480 colon cells expressed low levels of the UDP-glucuronosyltransferase (UGT) protein compared to expression levels in HT-29 cells. The 3-bromoazetidinones 16a-h and 16j demonstrated significantly less potent activity in the SW480 colon cancer cells, with 16a, 16b and 16j being the most effective with a 52, 50 and 48% inhibition of cell viability when evaluated at a 10 µM concentration, Table 3. The antiproliferative results for the most potent compound 10n in cell lines MCF-7, HL-60, SW480, HT-29, HL-60 and U266 are summarised in Table 5, together with the lead compound 10e and CA-4. The novel 3-haloazetidinone compounds 10e, 11n and 16d were selected by NCI for further biological evaluation in the NCI 60 cell line screen following an initial Tier 1 profiling screen [84]. The results obtained for the 3-haloazetidinone compounds 10e, 11n and 16d in the NCI 60 cancer cell line screening (GI 50 values, five doses) are shown in Table 6. (GI 50 is the concentration of the compound required to produce 50% of the maximal inhibition of cell proliferation.) The compounds showed broad-spectrum antiproliferative activity against leukaemia, non-small cell lung, colon, CNS, melanoma, ovarian, renal, breast and prostate cancer cell lines and confirmed the results obtained with compounds 10e, 11n and 16d in MCF-7 cells in our laboratory with GI 50 (Table 7), and were greater than 100 µM in all but one cell line for 10e and greater than 100 µM in all but four cell lines examined for compound 11n, indicating low toxicity and suggesting that these compounds may be suitable for therapeutic development.
From the results obtained above, it is apparent that the inclusion of the chloro substituent on the β-lactam scaffold (as in compound 10n) results in greater antiproliferative effects in the MCF-7 cell line (IC 50 = 0.0175 µM) than the corresponding 3,3-dichloro compound (11n) (IC 50 = 0.031 µM). By comparison, the introduction of a 3-bromo substituent on the β-lactam scaffold resulted in decreased antiproliferative activity, e.g., compound 16a (IC 50 = 0.579 µM) compared with the corresponding 3-chloro compound 10e (IC 50 = 0.034 µM). The antiproliferative activity of the most potent β-lactam compounds 10e, 10n and 11n may be correlated to the logP values for some compounds (see Table 3 and Supplementary Information, Tables S1 and S2). The most potent 3-chloro compound 10n (IC 50 = 0.0175 µM) has a lower logP (2.666) when compared to the 3,3-dichloro compound 11n (IC 50 = 0.031 µM) (logP 3.85); and the 3-bromo compound 16a (IC 50 = 0.579 µM) (logP 3.543) has a higher logP value when compared with the corresponding 3-chloro compound 10e (IC 50 = 0.034 µM) logP = 3.40, suggesting that compounds with higher logP values displayed poorer activity. However, 3-chloro-4-(2-naphthyl) compound 10k (IC 50 = 0.20 µM) and 3,3-dichloro-4-(2-naphthyl) compound 11k (IC 50 = 0.322 µM) retained potency, although with higher logP values (4.66 and 5.84, respectively). Interestingly, the 2-naphthyl 10k was much more potent (IC 50 = 0.20 µM) than the 1-naphthyl compound 10j (IC 50 = 14.66 µM), both with the same logP value (4.66); a similar trend was observed in 3,3-dichloro-1-naphthyl compounds 11j (IC 50 = 7.990 µM) and 2-naphthyl compound 11k (IC 50 = 0.322 µM), which is possibly related to the steric difficulty in accommodating the 1-naphthyl ring at the colchicine binding site, whereas the 2-naphthyl ring is a better substitute for the 3,4,5-trimethoxyaryl ring A. This effect is also reported for the 1-and 2-naphthyl analogues of CA-4 [86].  respectively, while the 3,3-dichloro compound 11e demonstrated increased cytotoxicity (10%). Compound 10n (3-hydroxy-4-methoxyphenyl Ring B), which was the most potent compound in the cell proliferation assays, resulted in increased cytotoxicity (17% cell death), while the similarly substituted 3,3-dichloro compound 11n displayed lower cytotoxicity of an 8% cell death in this assay. CA-4 (positive control) resulted in a 12% cell death at 10 μM.        The GI 50, TGI and LC 50 results of β-lactams 10e, 11n and 16d are summarised in Table 7. The COMPARE algorithm was used to compare the GI 50 and TGI results for compounds 10e, 11n and 16d with compounds of a known mechanism of action in the NCI Standard Agents Database (Table S3, Supplementary information) and allows correlations in drug sensitivities and molecular targets for biologically active compounds. The highest correlations for potent compounds 10e and 11n were obtained for tubulin-targeting agents, including the clinically used vinca alkaloids vincristine sulfate and vinblastine sulfate, together with maytansine. ADC T-DM1 contains an analogue mertansine conjugated with trastuzumab and is used in the treatment of metastatic HER-2 positive breast cancer [3]. The toxicity and selectivity of 10n towards normal cells was investigated in the non-tumourigenic cell line HEK-293T (normal human epithelial embryonic kidney cells). The cell viability of HEK-293T cells was significantly higher than MCF-7 cells following treatment with concentrations of 10n of 10, 1 and 0.5 µM for 72 h (Figure 4). The IC 50 value of 10n in HEK-293 cells (5.5 µM) compared favourably with that observed against the MCF-7 cell line (IC 50 = 17.5 nM), demonstrating that β-lactam 10n was less toxic to human normal cells than cancer cells. These data suggested the compound 10n could be developed as a broad-spectrum anti-cancer agent with lower cytotoxicity to normal cells compared with MCF-7 cancer cells. The cytotoxic effect of selected 3-chloro-β-lactams 10e, 10f, 10i, 10n and 3,3-dichloro-β-lactams 11e and 11n in MCF-7 cells was initially determined using the lactate dehydrogenase (LDH) assay [87]. The 3-chloro-β-lactams produced low cytotoxicity (at 10 µM) over 24 h with 5, 4 and 3% cell death, for compounds 10e, 10f and 10i, respectively, while the 3,3-dichloro compound 11e demonstrated increased cytotoxicity (10%). Compound 10n (3-hydroxy-4-methoxyphenyl Ring B), which was the most potent compound in the cell proliferation assays, resulted in increased cytotoxicity (17% cell death), while the similarly substituted 3,3-dichloro compound 11n displayed lower cytotoxicity of an 8% cell death in this assay. CA-4 (positive control) resulted in a 12% cell death at 10 µM.   apoptosis level of 7% observed with the vehicle ethanol at 72 h. The percentage of cells in the G 0 G 1 phase was observed at 8.5% (500 nM), while the untreated cells were 50.7% at 72 h, indicating that the cells are coming out of the G 0 G 1 phase and are undergoing G 2 /M followed by apoptosis. Similar effects on the cell cycle of MCF-7 cells were observed for the control drug CA-4 with a significant increase in the percentage of cells in G 2 M arrest (52%, 100 nM) with an increase in apoptosis (sub-G 1 ) (9.4%) [88]. In summary, compound 10n was found to induce G 2 /M arrest in MCF-7 cells in a time dependent manner, followed by apoptosis.
To further investigate the effects of compound 10n on the induction of cellular apoptosis, MCF-7 cells were treated with compound 10n for 48 h and then stained with Annexin Vfluorescein isothiocyanate (FITC)/propidium iodide (PI). Following analysis using flow cytometry, differentiation between live cells (annexin-V − /PI − ), early apoptotic cells (annexin-V + /PI − ), late apoptotic cells (annexin-V + /PI + ) and necrotic cells (annexin-V − /PI + ) is possible with dual staining with Annexin-V and PI, see Figure 6. Compound 10n induced both early and late apoptosis in MCF-7 cells in a concentration-dependent manner when compared to the untreated control cells ( Figure 6). When MCF-7 cells were treated with 10n, the total apoptotic cells (Annexin V-stained positive cells) increased in a dose-dependent manner from 29.8% at 50 nM to 37% at 500 nM. In contrast, only 5.0% of the total apoptotic cells were detected in the control cells (0.1% ethanol (v/v) treated sample). In comparison, the Annexin V-stained positive cells (total apoptotic) cells for CA-4 were determined as 34.6% in MCF-7 cells at 50 nM, as shown in Figure 6. These results demonstrated that compound 10n induced cell apoptosis in MCF-7 breast cancer cells.

Effects of Compound 10n on Tubulin Polymerisation in MCF-7 Cells
The tubulin binding activities of potent compounds 10e and 11n evaluated in MCF-7 cells were carried out using a tubulin polymerisation assay kit from Cytoskeleton (BK006P) [89]. In this assay, light is scattered by microtubules to an extent that is proportional to the concentration of the microtubule polymer. The compounds were tested at 10 µM concentration with purified and unpolymerised tubulin. The results for the selected β-lactam compounds are shown in Table 8. The initial assay performed established the effects of compounds 10e and 11n on tubulin polymerisation for 30 min (Figure 7, Table 8). Ethanol and CA-4 (2a) were used as a vehicle and a positive control, respectively. CA-4 is one of the most effective anti-tubulin natural products. Both compounds 10e and 11n showed moderate tubulin polymerisation inhibition effects, although they were less effective than CA-4. When evaluated at a 10 µM concentration, the 3-chloro and 3,3-dichloro compounds reduced the V max value for the rate of tubulin polymerisation 1.7-fold for compound 11n and 1.8-fold for compound 10e, whereas CA-4 induced a 6.3-fold reduction ( Figure 7, Table 8 and Supplementary Information Figure S1). In general, the tubulin polymerisation inhibition activities of selected compounds depend on the substitutions at the C-3 and C-4 position of β-lactam core. The compounds with chloro and dichloro substituents at the C-3 position, such as 10e and 11n, exhibited moderate tubulin polymerisation inhibition and demonstrated some correlation with the antiproliferative effects of these compounds.

Effects of Compound 10n on Tubulin Polymerisation in MCF-7 Cells
The tubulin binding activities of potent compounds 10e and 11n evaluated in MCF-7 cells were carried out using a tubulin polymerisation assay kit from Cytoskeleton (BK006P) [89]. In this assay, light is scattered by microtubules to an extent that is proportional to the concentration of the microtubule polymer. The compounds were tested at 10 μM concentration with purified and unpolymerised tubulin. The results for the selected β-lactam compounds are shown in Table 8. The initial assay performed established the effects of compounds 10e and 11n on tubulin polymerisation for 30 min (Figure 7, Table  8). Ethanol and CA-4 (2a) were used as a vehicle and a positive control, respectively. CA-4 is one of the most effective anti-tubulin natural products. Both compounds 10e and 11n showed moderate tubulin polymerisation inhibition effects, although they were less effective than CA-4. When evaluated at a 10 μM concentration, the 3-chloro and 3,3-dichloro compounds reduced the vmax value for the rate of tubulin polymerisation 1.7-fold for compound 11n and 1.8-fold for compound 10e, whereas CA-4 induced a 6.3-fold reduction   polymerisation inhibition activities of selected compounds depend on the substitution the C-3 and C-4 position of β-lactam core. The compounds with chloro and dichloro stituents at the C-3 position, such as 10e and 11n, exhibited moderate tubulin polymer tion inhibition and demonstrated some correlation with the antiproliferative effect these compounds.   Subsequently, the effect of the representative compound 10n on the organisation of microtubule cytoskeleton of MCF-7 cells was also determined by confocal microscopy using an anti-tubulin antibody ( Figure 8). As expected, the MCF-7 cells exhibited a wellorganised microtubular network (stained green) when treated with the vehicle control (0.1% ethanol). The MCF-7 cell nuclei (stained blue) were also clearly observed ( Figure 8A). In contrast, the fibrous microtubule structures were disorganised, and their densities were also significantly reduced by treating the MCF-7 cells with compound 10n. The paclitaxeltreated sample ( Figure 8C) showed the hyper-polymerisation of tubulin, while the extensive depolymerisation of tubulin was demonstrated in the CA-4-treated sample Figure 8B. Cells treated with the β-lactam compound 10n (0.05, 0.1 and 0.5 µM) displayed a disorganised microtubule network with similar effects to CA-4, together with multinucleation ( Figure 8D-F). The treatment of MCF-7 breast cancer cells with CA-4 tubulin-targeting agents has been reported to result in the formation of multiple micronuclei and mitotic catastrophes [90][91][92]. These immunofluorescence studies for the visualisation of the microtubule network in MCF-7 cells confirmed that compound 10n could directly inhibit the tubulin polymerisation.
A colchicine-site binding assay was performed to evaluate the interaction of compound 10n at the colchicine binding site of tubulin using N,N-ethylenebis(iodoacetamide) (EBI) [93,94]. EBI crosslinks the Cys-239 and Cys-354 residues of the colchicine binding site of β-tubulin, alkylating the sulfhydryl group of cysteine. This covalent EBI adduct occupies the colchicine binding site of β-tubulin and can be detected using Western blotting as it appears at a lower position than tubulin, indicating that Cys239 and Cys354 amino acids of β-tubulin are crosslinked with EBI. Microtubule targeting compounds binding at the colchicine site, e.g., colchicine and CA-4, prevent the formation of the β-tubulin-EBI adduct. MCF-7 cells were treated with 10n (10 µM) or CA-4 (10 µM), followed by EBI (100 µM). Control samples (ethanol 0.1% (v/v)) indicated the formation of the β-tubulin-EBI adduct at a slightly lower position (Figure 9). Tubulin EBI adduct formation was inhibited in the MCF-7 cells treated with CA-4 and 10n, indicating that both CA-4 and 10n bind to the colchicine binding site of tubulin. These tubulin polymerisation inhibitors act on the colchicine site of tubulin and compete with EBI to bind the colchicine binding site and hinder the cross-linking of EBI with β-tubulin.

Effects of Compound 10n on Expression Levels of Apoptosis-Associated Proteins Bax, Bcl-2 and Mcl-1 in MCF-7 Cells
The effects of compound 10n on the expression of the pro-apoptotic protein Bax and anti-apoptotic proteins Bcl-2 and Mcl-1 were next investigated using Western blot analysis. The Bcl-2 family of proteins controls and regulates the intrinsic or mitochondrial apoptotic pathway. Pro-and anti-apoptotic members of the Bcl-2 family can oligomerise at the mitochondrial outer membrane to regulate permeabilization, which is a central event in the intrinsic apoptotic pathway. The pro-apoptotic protein Bax, together with Bak, is a key member of the Bcl-2 family and is a core regulator of the intrinsic pathway of apoptosis [95]. The effects of compound 10n on the expression of the pro-apoptotic protein Bax and anti-apoptotic proteins Bcl-2 and Mcl-1 were next investigated using Western blot analysis. The Bcl-2 family of proteins controls and regulates the intrinsic or mitochondrial apoptotic pathway. Pro-and anti-apoptotic members of the Bcl-2 family can oligomerise at the mitochondrial outer membrane to regulate permeabilization, which is a central event in the intrinsic apoptotic pathway. The pro-apoptotic protein Bax, together with Bak, is a key member of the Bcl-2 family and is a core regulator of the intrinsic pathway of apoptosis [95]. Western blot analysis ( Figure 10 The apoptosis regulating proteins Bcl-2 and Mcl-1 were next investigated. The antiapoptotic or pro-survival Bcl-2 protein is also a member of the Bcl-2 family. It prevents the release of a pro-apoptotic AIF (apoptosis inducing factor) and cytochrome c from the mitochondria into the cytoplasm [96] and prevents apoptosis by sequestering caspases (apoptosis promoters). The level of the anti-apoptotic protein Bcl-2 was downregulated in a dose-dependent manner after treating MCF-7 cells with compound 10n (0.05, 0.1 and 0.5 μM) for 48 or 72 h. (Figure 9). Mcl-1 protein (an induced myeloid leukaemia cell differentiation protein) is another key anti-apoptotic member of the Bcl-2 family and is localised in the mitochondrial outer membrane [97]. It binds and sequesters the pro-apoptotic Bax/Bak proteins and, thus, prevents the release of cytochrome c [98]. Overexpression of the anti-apoptotic factors Mcl-1, Bcl-2 and Bcl-xL in acute myeloid leukaemia [99] and acute lymphocytic leukaemia [100] may be associated dysregulation of apoptosis. The level of the anti-apoptotic protein Mcl-1 was downregulated in a dose-dependent manner after treating MCF-7 cells with compound 10n (0.05, 0.1 and 0.5 μM) for 48 or 72 h. ( Figure  10). Apoptosis can be triggered by a reduction in the expression levels of Mcl-1 and Bcl-2 (e.g., by drug treatment). The increase in the percentage of cells observed in the sub-G1 peak, together with the flow cytometry analysis of Annexin V/PI-stained cells support the pro-apoptotic mechanism of action proposed for these compounds (Figures 5 and 6). The apoptosis regulating proteins Bcl-2 and Mcl-1 were next investigated. The antiapoptotic or pro-survival Bcl-2 protein is also a member of the Bcl-2 family. It prevents the release of a pro-apoptotic AIF (apoptosis inducing factor) and cytochrome c from the mitochondria into the cytoplasm [96] and prevents apoptosis by sequestering caspases (apoptosis promoters). The level of the anti-apoptotic protein Bcl-2 was downregulated in a dosedependent manner after treating MCF-7 cells with compound 10n (0.05, 0.1 and 0.5 µM) for 48 or 72 h. (Figure 9). Mcl-1 protein (an induced myeloid leukaemia cell differentiation protein) is another key anti-apoptotic member of the Bcl-2 family and is localised in the mitochondrial outer membrane [97]. It binds and sequesters the pro-apoptotic Bax/Bak proteins and, thus, prevents the release of cytochrome c [98]. Overexpression of the anti-apoptotic factors Mcl-1, Bcl-2 and Bcl-xL in acute myeloid leukaemia [99] and acute lymphocytic leukaemia [100] may be associated dysregulation of apoptosis. The level of the anti-apoptotic protein Mcl-1 was downregulated in a dose-dependent manner after treating MCF-7 cells with compound 10n (0.05, 0.1 and 0.5 µM) for 48 or 72 h. (Figure 10). Apoptosis can be triggered by a reduction in the expression levels of Mcl-1 and Bcl-2 (e.g., by drug treatment). The increase in the percentage of cells observed in the sub-G 1 peak, together with the flow cytometry analysis of Annexin V/PI-stained cells support the pro-apoptotic mechanism of action proposed for these compounds (Figures 5 and 6).

Computational Modelling of β-Lactam Compounds 10n, 11n and 14b
Computational docking calculations using MOE 2019.01 [101] were undertaken on both enantiomers of the potent compounds trans-3-chloro-1-(3,4,5-trimethoxyphenyl)β-lactam 10n, 3,3-dichloro-1-(3,4,5-trimethoxyphenyl)-β-lactam 11n together with the 3-chloro-1-(3,5-dimethoxyphenyl)-β-lactam 14b, using the X-ray structure of bovine tubulin co-crystallised with N-deacetyl-N-(2-mercaptoacetyl)-colchicine (DAMA-colchicine) 1SA0 [6], Figure 11. 1 H NMR analysis determined that only the trans isomers of the compounds 10n and 14b were isolated; therefore, we modelled only the 3S,4S and 3R,4R enantiomeric pairs. In all cases, the S,S enantiomers were more highly ranked than the corresponding R,R enantiomeric pair; therefore, only they will be discussed. All trimethoxy compounds overlaid their B-rings on the C-ring of DAMA-colchicine (forming HBA interactions with Lys352), co-located the 3,4,5-trimethoxyphenyl substituted A-rings and were able to position the halogens in an open region of the tubulin binding site at the monomer interface. The predicted affinity ranking from best ranked to worst was 10nSS, 14bSS, 11nS, 10nRR, 11nR and 14bRR. Generating conformers with OMEGA [1,2] and running docking with FRED [3] also gave the same preference for SS over RR enantiomers. Docking studies are not ideal for studying changes in cellular efficacy associated with different halogen substituents. While the SS enantiomer of the dimethoxy analogue 14b presented a comparable binding mode to other analogues in the tubulin site, it did not overlap fully with DAMA-colchicine ( Figure 11, Panel C). The lack of a hydroxy group in Ring B of 14b to potentially hydrogen bond with Lys352 and the added steric bulk, resulting from the substitution of 4-methoxy with 4-ethoxy on the B-ring, forced the molecule deeper into the binding site, resulting in the less favourable docking scores. The increase in lipophilicity could also decrease the cell permeability of 14b, causing a slight loss in efficacy against MCF-7 cells. Figure 11 illustrates the best ranked binding pose of each compound, showing the shared binding mode across the analogues.  Figure 11. Overlay of the X-ray structure of tubulin co-crystallised with DAMA-colchicine (PDB entry 1SA0) on the best ranked docked poses of the S enantiomers of the three studied beta-lactams: (A) 10n, (B) 11n and (C) 14b. Ligands are rendered as tube and amino acids as line. Tubulin amino acids and DAMA-colchicine are coloured by atom type: carbon = grey, hydrogen = white, oxygen = red, nitrogen = blue, sulphur = yellow, bromine = dark red, chlorine = dark green. The beta-lactams are depicted with a green backbone. The atoms are coloured by element type, Key amino acid residues are labelled, and multiple residues are hidden to enable a clearer view.

Chemistry
All chemicals were commercially purchased and were used without further purification unless otherwise indicated. Solvents were either purchased dry or purified by distillation in accordance with literature methods. Dichloromethane was dried by distillation from calcium hydride prior to use. Tetrahydrofuran (THF) was distilled immediately prior to use from Na/Benzophenone under nitrogen. Toluene was dried by distillation from calcium hydride and stored on activated molecular sieves (4 Å). Melting points were determined on a Gallenkamp SMP 11 melting point apparatus and are uncorrected. Infra-red (IR) spectra were recorded as KBr discs, thin films on NaCl disk or ATR on a Perkin Elmer FT-IR Paragon 1000 spectrometer. 1 H and 13 C nuclear magnetic resonance (NMR) spectra were recorded at 20 • C on a Bruker DPX 400 spectrometer (400.13 MHz, 1 H; 100.61 MHz, 13  High resolution mass spectrometry scans were also performed using Electrospray Ionisation operated in negative and positive ion modes on an LTQ/Orbitrap Discovery Mass Spectrometer, and samples were dissolved in CH 3 OH, with a mass accuracy of <±5 ppm. Low resolution mass spectra (LRMS) were obtained on a Hewlett-Packard 5973 MSD GC-MS system in electron impact (EI) mode. Thin layer chromatography was performed with Merck silica gel 60 TLC aluminium sheets using a fluorescent indicator visualising at 254 nm in UV. Merck Kiesegel 60 (particle size 0.040-0.063 mm) was used for flash column chromatography. Preparative chromatography was also carried out on a Biotage SP4 instrument. All products isolated were homogenous on TLC. Microwave experiments were performed with the Biotage Initiator and Discover CEM microwave synthesisers. Purity of the finial compounds was achieved using analytical high-performance liquid chromatography (HPLC) using a Waters 2487 Dual Wavelength Absorbance detector, Waters 1525 binary HPLC pump, Waters In-Line Degasser AF and Waters 717plus Autosampler, with a Varian Pursuit XRs C18 reverse phase 250 × 4.6 mm column and detection at 254 nm. Samples were analysed using acetonitrile (60%):water (40%) with 0.1% (v/v) TFA over 10 min and a flow rate of 1 mL/min . Imines 9a-9j, 9m-9s, 11f, 11i, 12e, 12f and 12i were prepared as previously reported [34,35,74,102]. The details for the preparation of 3-azetidinones 10f, 10i, 10o, 11e,  11f, 11i and 11o were as previously reported [34,74] are provided in the Supplementary Information.
3.1.1. General Method 1A: PREPARATION of Imines with Ethanol as Solvent (9k, 9l, 9t-v): The appropriately substituted benzaldehyde (10 mmol) and corresponding substituted aniline (10 mmol) were heated at reflux in ethanol (40 mL) for 4 h. The reaction solvent was then reduced to approximately 10 mL in vacuo and the reaction mixture was allowed to stand for 12 h. The precipitated product was filtered and then recrystallised from ethanol.

Stability Study for Compound 16a
A stability study for compound 16a was performed by analytical HPLC using a Symmetry ® column (C18, 5 mm, 4.6 × 150 mm), a Waters 2487 Dual Wavelength Absorbance detector, a Waters 1525 binary HPLC pump and a Waters 717 plus Autosampler (Waters Corporation, Milford, MA, USA). Samples were detected at λ 254 nm using acetonitrile (70%)/water (30%) as the mobile phase over 15 min and a flow rate of 1 mL/min. A stock solution of the compound was prepared using 10 mg of compound 16a in 10 mL of mobile phase (1 mg/mL). A calibration curve was prepared using a solution of 0.5, 0.25, 0.125, 0.0625, 0.03125, 0.015625 and 0.0078 mg/mL. (i) Stability of 16a in phosphate buffers: Phosphate buffers at pH values 4, 7.4 and 9 were prepared according to the British Pharmacopoeia 2020. A total of 300 µL of stock solution (1 mg/mL ACN) for 16a was added to a vial containing 9.7 mL of buffer, mixed and pre-heated to 37 • C. A total of 1 mL of the solution was added to the HPLC glass vial and 10 µL was injected, followed by hourly injections for a 24-h period. Samples were withdrawn and analysed at time intervals of t = 0 min, 5 min, 30 min, 60 min and hourly for 24 h. The analysis was performed in triplicate. (ii) Thermal stability: 16a (1 mg) was placed in a vial (for the solution, 1 mL of stock solution was used) at 60 • C for 4 h on a heating block. The sample was then cooled, diluted with ACN and analysed using HPLC. (iii) Photostability study: A solution of compound 16a (1 mL of the stock solution) was placed in a vial and exposed to UV light for 4 h. The sample was directly analysed using HPLC. (iv) Stability in acidic condition: The stock solutions (0.8 mL) of 16a were placed in a vial and HCl (0.1 M, 0.2 mL) was added. The vial was vortexed to ensure a homogeneous mixture and left to stand at room temperature. A sample from the vial was taken and neutralised with NaOH (0.1 M, 0.2 mL) every hour for 4 h. Once neutralised, the samples were analysed using HPLC. (v) Stability of 16a in basic (alkaline) conditions: The stock solution (0.8 mL) was placed in a vial and NaOH (0.1 M, 0.2 mL) was added. The vial was vortexed to ensure a homogeneous mixture and left to stand at room temperature. A sample from the vial was neutralised with HCl (0.1 M, 0.2 mL) every hour for the 4 h. Once neutralised, the samples were analysed using HPLC. (vi) Stability of 16a in oxidising conditions: The stock solution (0.8 mL) was placed in a vial and H 2 O 2 (3%, 0.2 mL) was added. The vial was vortexed to ensure a homogeneous mixture and left to stand for 4 h at room temperature. A sample from the vial was taken every hour over 4 h and analysed using HPLC.

Biochemical Evaluation of Activity
All biochemical assays were performed in triplicate on at least three independent occasions for the determination of mean values reported. All the reagents including foetal bovine serum (FBS) and cell culture growth medium (MEM, DMEM and RPMI-1640) were purchased from BD Biosciences. CA-4 was purchased from Sigma Aldrich. with GlutaMAXTM-I in the absence of non-essential amino acids. Human breast cancer MCF-7 cells and multiple myeloma U266 cells were cultured in Minimum Essential Media (MEM) with GlutaMAX™-I, supplemented with 1% (v/v) non-essential amino acids, 10% 2(v/v) foetal bovine serum (FBS) and 1% (v/v) penicillin/streptomycin 5000 U/mL. MDA-MB-231 cells were maintained in DMEM supplemented with 10% (v/v) foetal bovine serum, 2 mM L-glutamine and 100 µg/mL penicillin/streptomycin (complete medium). Colon cancer HT-29 and SW-480 and triple negative breast cancer Hs578T and its invasive variant Hs578T8i were cultured in DMEM with GlutaMAX™-I, with the same supplement in the absence of non-essential amino acids. Leukaemia HL-60 cancer cells were cultured in Roswell Park Memorial Institute Media (RPMI-1640) with GlutaMAX™-I, supplemented with 10% FCS media, and 100 µg/mL penicillin/streptomycin as above. Cell numbers were monitored using a haemocytometer. Cell culture flasks were incubated in a humidified incubator (5% CO 2 /95% air) at 37 • C. All cell lines were sub-cultured three times per week with trypsinisation using TrypLE Express (1X) required for adherent cell lines.

Cell Viability Assay
A stock solution of each β-lactam compound was prepared (10 mM) and serial 100× dilutions were made with ethanol for compounds to have working dilutions of 1 nM, 10 nM, 100 nM, 500 nM, 1µM, 5 µM, 10 µM and 50 µM. CA-4 was dissolved in ethanol to obtain a 10 mM stock solution. All stock solutions and serial dilution in ethanol/DMSO were stored at -20 • C. All cells were seeded at a density of 2.5 × 10 4 cells/mL in a 96-well plate (200 µL per well). The cells were incubated in a 95% O 2 /5% CO 2 atmosphere at 37 • C for 24 h, then treated with test compound (2 µL of stock solutions per 200-microlitre well) in ethanol to obtain a concentration range of 1 nM-200 µM for the study. The plates were then re-incubated for a further 72 h. Control wells contained an equivalent volume of the vehicle ethanol or DMSO (1% v/v). MTT cell viability assay: The culture medium was removed, and the cells were washed with phosphate buffered saline (PBS, 100 mL). MTT (dissolved in PBS, 50 mL) was added to obtain a final concentration of MTT (1 mg/mL). Cells were incubated at 37 • C for 3 h in the dark. Solubilisation was commenced by the addition of DMSO (200 mL) and the cells were maintained at 20 • C in the dark for 20 min before reading the absorbance to ensure complete colour diffusion. The absorbance value of control cells (no added compound) was set to 100% cell viability and absorbance versus cell density per well was determined to assess cell viability using Graph-Pad Prism software. AlamarBlue cell viability assay: Cells were seeded in 96-well plates (e.g., MCF-7, 5 × 10 3 cells/well) and (HT-29 cells, 1 × 10 3 cells/well) and (HL-60 cells, 1 × 10 4 cells/well) with a total volume per well of 200 µL. After 24 h, cells were treated in triplicate with serial dilutions of CA-4 or β-lactam analogues (0.001-100 µM), medium alone or vehicle (1% ethanol (v/v)). Ethanol or DMSO were used as vehicle control and cells were treated with no more than 1% ethanol (v/v) or 0.1% DMSO in all experiments. Cell proliferation for cells was analysed using the AlamarBlue assay (Invitrogen Corp.) following the manufacturer's instructions. After 72 h, AlamarBlue (10% (v/v) (20 µL)) was added to each well and plates were incubated in the dark at 37 • C for 3-5 h. The blank consisted of the appropriate medium (according to cell type) with the addition of AlamarBlue. Plates were analysed on the 96-well fluorimeter Spectramax Gemini plate reader with excitation at 530 nm and emission at 590 nm using a SOFTmax Pro version 4.9 (Molecular Devices, Sunnyville, C.A) software package and the percentage viability relative to vehicle control was recorded. Results were plotted using GraphPad Prism 5 software and analysed using a non-liner, sigmoidal dose response curve to determine the relative IC 50 values. All assays were performed in triplicate for the determination of mean values reported.

Lactate Dehydrogenase Assay for Cytotoxicity
The cytotoxicity of selected compounds was determined using the CytoTox 96 nonradioactive cytotoxicity assay (Promega Corporation, Madison, WI, USA) [104]. Briefly, MCF-Pharmaceuticals 2021, 14, 1119 44 of 53 7 cells were seeded in a 96-well plate (200 µL per well), at a density of 2.5 × 10 4 cells/mL, and incubated for 24 h. The cells were then treated with selected β-lactam compounds as described above for the cell viability assay. After 72 h, 'lysis solution (10X)' (20 µL) was added to the plate and incubated for a further 1 hr to ensure 100% death. Supernatant (50 µL) was removed from each well to a 96-well plate. Reconstituted 'CytoTox 96 ® Reagent (50 µL) was added to each well and the plate was placed in the dark at 20 • C for 30 min. 'Stop solution' (50 µL) was added to each well and the samples were analysed at 490 nm using a Dynatech MR5000 plate reader. The percentage cell death at 10 µM was calculated.

Cell Cycle Analysis
Flow cytometric analysis was used to determine DNA level in any given cell that had been stained with propidium iodide (PI) [105]. In this experiment, adherent and detached cells were collected by trypsinisation and centrifuged at 800× g for 15 min. Cells were then washed three times with ice-cold PBS and fixed with slow addition of ice-cold 70% ethanol overnight at −20 • C. The cells were then centrifuged (800× g) for 15 min; the pellet was re-suspended in PBS (400 µL) and transferred to LP5 FACS tubes. Cells were then stained with PI (50 µg/mL), containing DNase-free RNase A (50 µg/mL) at 37 • C for 30 min, which degrades any double-strand RNA. The DNA content of the cells (10,000 cells/experimental group) was analysed by flow cytometry at 488 nm using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA, USA). Results were presented as mean ± SEM. The statistical analysis of experimental data was performed using the program Prism GraphPad 5. A two-way ANOVA (Bonferroni post-test) was used to test for statistical significance (**, p < 0.05, ***, p < 0.001). A value of p < 0.05 was considered to be significant.

Annexin V/PI Apoptotic Assay
The Annexin V/Propidium Iodide (PI) assay was used to detect both early-and late-stage apoptosis using flow cytometry, as previously reported [34]. Early apoptosis is detected by the presence of phosphatidylserine (PS) on the outer surface of the cell membrane. PS is a phospholipid normally found on the cytoplasmic surface of the cell membrane. In apoptosis, PS is translocated to the outer surface of the cell membrane, and is exposed to the extracellular environment [106]. MCF-7 cells were seeded in 6-well plates (1 × 10 5 cells/mL) and treated with either vehicle (0.1% (v/v) EtOH), CA-4 (50 nM) or β-lactam compound 10n (50 and 500 nM) for 48 h. Cells were then analysed using flow cytometry. Cells were first washed in 1X binding buffer (20X binding buffer: 0.1 M HEPES, pH 7.4; 1.4 M NaCl; 25 mM CaCl 2 diluted in dH 2 O) and treated for 30 min on ice in Annexin V-containing binding buffer (1:100) in the dark. Cells were washed in binding buffer and then re-suspended in PI-containing binding buffer (1:1000). Samples were analysed without delay using the BD Accuri flow cytometer and the data analysed with GraphPad Prism software.

In Vitro Tubulin Polymerisation Assay
The assembly of purified bovine tubulin was monitored using a kit, BK006, purchased from Cytoskeleton Inc. (Denver, CO, USA) [89] as we have previously reported [34]. Briefly, purified bovine brain tubulin (>99%, 3 mg/mL) in a buffer (80 mM PIPES (pH 6.9), 0.5 mM EGTA, 2 mM MgCl 2 , 1 mM GTP and 10% glycerol) was incubated at 37 • C in the presence of either vehicle (2% (v/v) ddH 2 O) or β-lactam compounds 10e and 11n (10 µM). A reference control experiment with CA-4 was also used (See Supplementary information). Light was scattered proportionally, dependent on the concentration of polymerised microtubules in the sample. Tubulin assembly was monitored turbidimetrically at 37 • C in a Spectramax 340 PC spectrophotometer (Molecular Devices, Sunnyvale, CA, USA) at 340 nm. The absorbance was measured at 30-s intervals for 60 min.
Data for 10e were collected on a Rigaku Saturn 724 (Mo Kα radiation, λ = 0.71073 Å) equipped with a Rigaku X-Stream low temperature device. The sample was mounted on a Hampton cryoloop and data were collected at 93(2) K. Data were measured using 0.3 • scans per frame for 20 s. A total of 852 frames were collected with a final resolution of 0.77 Å. Data reduction and correction for Lorenz, polarisation and absorption were performed using the CrystalClear software. Absorption corrections were applied using REQAB (Rigaku Inc., 2007). Structures 10o/16g and 11o were solved with the SHELXT structure solution program [109] using Intrinsic Phasing and 9o and 10e with SHELXS with direct methods. All were refined using Least Squares method on F 2 with SHELXL. All non-hydrogen atoms were refined anisotropically. Hydrogen atoms were assigned to calculated positions using a riding model with appropriately fixed isotropic thermal parameters. Molecular graphics were generated using OLEX2 [110]. In the structure of 10o, the halogen substituted 3 position on the lactam ring was modelled between Cl and Br with 75:25% of each, respectively. The occupancy was freely refined then fixed. C-Cl and C-Br distances were restrained (DFIX) and the atomic displacement of both halides was also constrained (EADP). In 11o, the 4 substituent, the Me(OMe)Ph ring, was modelled in two locations with restraints (SADI) and constraints (EADP). The refined occupancies of each moiety were 81:19%. Crystallographic data for the structures in this paper have been deposited with the Cambridge Crystallographic Data Centre as supplementary publication nos. 2077515-2077518. Copies of the data can be obtained, free of charge, on application to CCDC, 12 Union Road, Cambridge CB2 1EZ, UK (fax: +44-(0)1223-336033 or e-mail: deposit@ccdc.cam.ac.uk).

Computational Procedure for Molecular Docking
The 1SA0 X-ray structure of bovine tubulin co-crystallised with N-deacetyl-N-(2-mercaptoacetyl)-colchicine (DAMA-colchicine) was downloaded from the PDB website [6]. A UniProt Align analysis confirmed a 100% sequence identity between human and bovine β tubulin. The crystal structure was prepared using QuickPrep (minimised to a gradient of 0.001 kcal/mol/Å), Protonate 3D, Residue pKa and Partial Charges protocols in MOE 2019 with the MMFF94x force field. Compounds 10n, 11n and 14b were drawn in MOE, saved as an mdb and processed in MOE [101]. Both trans enantiomers of the compounds 10n, 11n and 14b were examined. For each compound, MMFF94x partial charges were calculated, and each was minimised to a gradient of 0.001 kcal/mol/Å. Default parameters were used for docking except that 300 poses were sampled for each compound and the top 50 docked poses were retained for subsequent analysis. Default settings of OMEGA [111,112] were used to generate 50 conformers of each compound prior to running rigid docking with FRED [113], included in the OEDocking suite [111].

Conclusions
Microtubule-targeting drugs such as taxanes and vinca alkaloids are very effective therapeutic agents in the treatment of various types of cancers. Interestingly, the antiviral activity of the bis-indole microtubule targeting drug sabizabulin (VERU-111) against SARS CoV-2 was recently reported [114]. Sabizabulin binds to tubulin and disrupts the intracellular transport of the SARS CoV-2 virus; it also demonstrates an effective anti-inflammatory effect. In this work, a novel series of heterocyclic combretastatin CA-4 compounds based on the β-lactam scaffold were designed and synthesised as tubulin-targeting agents. All the novel compounds were initially evaluated in the MCF-7 breast cancer cell line and of particular interest were compounds 10e, 10n and 11n, which displayed antiproliferative activity in the nanomolar range, e.g., 10e (IC 50 = 34 nM), 10n (IC 50 = 17.5 nM) and 11n (IC 50 = 31 nM) in MCF-7 cells. These compounds were identified for further studies to provide a better understanding of their mechanism of action in breast cancer cells. Minimal cytotoxicity was observed on the treatment of the most potent compound 10n in the nontumourigenic cell line HEK-293T, demonstrating the selectivity of the compounds toward cancer cells. The compounds were evaluated in the NCI 60 cancer cell line panel and demonstrated significant antiproliferative activity at nanomolar concentrations in a range of human cancer cell lines. Cell cycle analysis of compound 10n in MCF-7 cells demonstrated that this compound induces G 2 /M arrest and apoptotic cell death. The induction of apoptosis in MCF-7 cells by compound 10n was confirmed using flow cytometric analysis of Annexin V/PI-stained cells. An alteration of the expression levels of apoptosis-related proteins Bax, Bcl-2 and Mcl-1 in MCF-7 cells was shown using Western blot analysis. To examine whether the antiproliferative activities might be related to the depolymerisation of tubulin, the inhibitory effects of compounds 10e and 11n on tubulin polymerisation were confirmed with the suppression of in vitro tubulin polymerisation. The tubulin depolymerisation effects of compound 10n were confirmed when MCF-7 cells treated with the β-lactam 10n displayed a disorganised microtubule network with similar multinucleation effects to CA-4. Tubulin EBI-adduct formation was inhibited in MCF-7 cells treated with 10n, indicating an interaction with the colchicine binding site of tubulin. Our data strongly indicate that this class of β-lactams represent interesting lead molecules with the potential for a design of potent microtubule-targeting agents and further clinical anti-cancer drug development.