Antiproliferative and Tubulin-Destabilising Effects of 3-(Prop-1-en-2-yl)azetidin-2-Ones and Related Compounds in MCF-7 and MDA-MB-231 Breast Cancer Cells

A series of novel 3-(prop-1-en-2-yl)azetidin-2-one, 3-allylazetidin-2-one and 3-(buta-1,3-dien-1-yl)azetidin-2-one analogues of combretastatin A-4 (CA-4) were designed and synthesised as colchicine-binding site inhibitors (CBSI) in which the ethylene bridge of CA-4 was replaced with a β-lactam (2-azetidinone) scaffold. These compounds, together with related prodrugs, were evaluated for their antiproliferative activity, cell cycle effects and ability to inhibit tubulin assembly. The compounds demonstrated significant in vitro antiproliferative activities in MCF-7 breast cancer cells, particularly for compounds 9h, 9q, 9r, 10p, 10r and 11h, with IC50 values in the range 10–33 nM. These compounds were also potent in the triple-negative breast cancer (TBNC) cell line MDA-MB-231, with IC50 values in the range 23–33 nM, and were comparable with the activity of CA-4. The compounds inhibited the polymerisation of tubulin in vitro, with significant reduction in tubulin polymerization, and were shown to interact at the colchicine-binding site on tubulin. Flow cytometry demonstrated that compound 9q arrested MCF-7 cells in the G2/M phase and resulted in cellular apoptosis. The antimitotic properties of 9q in MCF-7 human breast cancer cells were also evaluated, and the effect on the organization of microtubules in the cells after treatment with compound 9q was observed using confocal microscopy. The immunofluorescence results confirm that β-lactam 9q is targeting tubulin and resulted in mitotic catastrophe in MCF-7 cells. In silico molecular docking supports the hypothesis that the compounds interact with the colchicine-binding domain of tubulin. Compound 9q is a novel potent microtubule-destabilising agent with potential as a promising lead compound for the development of new antitumour agents.


Introduction
Breast cancer is the most commonly reported cancer among women in developed countries. [1]. Over the last decade, the clinical treatment options for breast cancer patients have significantly improved with the approval of multiple drugs for various indications [2,3]. The hormone receptor (HR) positive/human epidermal growth factor receptor 2 (HER2)negative subtype is the most commonly identified subtype and is found in approximately 70% of breast cancers diagnosed. Endocrine therapies (tamoxifen, fulvestrant, anastrozole, exemestane and letrozole) are conventionally used for treatment of this HR+/HER2-breast cancer subtype. Targeted therapies such as CDK4/6 inhibitors palbociclib, ribociclib and abemaciclib [4] in combination with endocrine therapies show improved therapeutic effects in HR+/HER2-and metastatic breast cancer (MBC) [5]. Alpelisib, a PI3K inhibitor, is approved for patients with HR+, HER2-and PIK3CA-mutated cancers in combination with fulvestrant [6], while the PARP inhibitor olaparib targets BRCA mutations in early breast cancer [7]. Recently approved breast cancer drugs include the antibody-drug conjugate (ADC) trastuzumab deruxtecan [8] and the estrogen receptor degrader (SERD) elacestrant [9].

Chemistry: Synthesis of β-Lactams
β-Lactams are the most widely used group of antibiotic drugs for bacterial infections [55]; β-lactam-containing molecules are also useful as intermediates in organic synthesis, and many synthetic routes are available for the construction of the β-lactam ring [56][57][58], including the recently reported carbonylative formal cycloaddition of alkylarenes with imines [59]. In the present work, the required series of 3-isopropenyl β-lactams with a variety of ring B aryl substituents located at C-4 of the β-lactam ring were obtained by Staudinger reaction of imines with 3,3-dimethylacryloyl chloride. Schiff bases 8a-8r were prepared by the condensation of the appropriately substituted arylaldehyde with the 3,4,5-(See Supplementary Materials Figures S1-S20). The β-lactams 9a-9x demonstrated the characteristic β-lactam IR absorption at 1750 cm −1 .
A series of trans-3-allyl-β-lactams, 10a-s, with a variety of substituents at C4 of the βlactam ring were next synthesised as CA-4 analogues from imines 8a-r by Staudinger reaction using 4-pentenoyl chloride (Schemes 1 and 2). The compounds were obtained exclusively with trans stereochemistry in moderate yields (11-96%) after chromatographic Phosphate ester prodrugs of the phenols 9q and 10p were prepared by controlled esterification [74] to afford the dibenzyl phosphate esters (14a, 14b) (Scheme 3). In the 1 H NMR spectrum of 14a, the H-4 signal was observed at δ 4.68 as a doublet (J 4,3 = 2.52 Hz) and demonstrated trans stereochemistry with H-3 (δ 3.68, J 3,4 = 2.04 Hz). The methyl protons were identified at δ 1.86; the alkene protons were assigned as broad singlets at δ 5.02 and δ 5.05, while the benzylic methylene signals occurring as a multiplet at δ 5. 13-5.17 were confirmed with negative signals in the DEPT experiment (69.48, 69.50 ppm). Treatment of the phosphate esters 14a and 14b with bromotrimethylsilane afforded the desired phosphates 15a and 15c. Reduction of 14a and 14b with hydrogen and palladium/carbon catalyst also allowed removal of the dibenzyl protecting groups; however, reduction of the C-3 alkene occurred to afford compounds 15b and 15d (without decomposition of the main four-member ring). The 1 H NMR spectrum of compound 15a confirms the removal of the benzylic group protons, with the isopropyl methyl signal at δ 1.86, and the β−lactam and vinylic protons overlapping as a multiplet at δ 4.91-5.06 (H-3, H-4, H6). In the reduced product 15b, the methyl (δ 1.00-1.08) and methine (δ 2.06) signals for the C-3 isopropyl substituent are clearly identified together with the β-lactam H-3 and H-4 signals at δ 2.99 and δ 4.64, respectively.
The synthesis of structurally related β-lactams that contain alkene and ester substituents at C-3 was investigated (Scheme 4). Lithium enolates of 3-unsubstituted β-lactams react with numerous electrophiles to provide 3-substituted compounds [75]. In the present work, alkylation of the enolate of 3-unsubstituted β-lactams 16a-c with cinnamyl bromide cleanly afforded the 3-substituted β-lactams (17a-c) (Scheme 4). Similarly, treatment of the enolates of compounds 16a and 16b with ethyl bromoacetate afforded the ester products 17f and 17g, respectively. The enolate chemistry is stereoselective, and compounds 17a-c, 17f and 17g were obtained with exclusively trans stereochemistry. The structural assignment for compound 17a was confirmed from the 1 H NMR, 13 C NMR and HH COSY spectra (see Supplementary Materials, Figures S1-S20). The multiplet signals δ 2.69-2.93 were assigned as H-5 with coupling to H-3, H-6 and H-5, while the corresponding carbon appears as a negative signal at 31.68 ppm in the DEPT 90 spectrum. The multiplet signal δ 3.28-3.33 was attributed to H-3. The doublet at δ 4.69 (J = 2.28 Hz) was assigned to H-4 of the β-lactam ring. The multiplet (δ 6.24-6.31) was assigned to the alkene H-6 with coupling to H-5 and H-7. The signal at δ 6.51 is diagnostic for the alkene H-7 with a trans vicinal coupling constant of 15.8 Hz.

Stability Study of β-Lactams
Stability evaluation of compounds was carried out to avoid subsequent significant loss of pharmacological activity in vivo. The stability of representative β-lactams 9q, 10h, 10q, 10p, 10r, 15a and CA-4 was evaluated by HPLC at relevant in vivo acidic, neutral and basic conditions (pH 4, 7.4 and 9) and in plasma (see Supplementary Materials, Tables S2 and S3 and Figure S21). The half-life (t 1 2 ) for 9q (the most potent compound in the series, containing the phenolic ring B) was determined to be 13 h at both pH 4 and pH 7.4 and 6 h at pH 9. However, the t 1 2 in plasma for compound 9q was greater than 24 h. The t 1 2 of the corresponding phosphate prodrug phosphate ester 15a was greater than 24 h at pH 4 and pH 7.4 and in human blood plasma. CA-4 was stable at pH 4.0, 7.4 and 9.0 and in human plasma for more than 7 h. Based on this stability study, β-lactam 9q and the phosphate 15a would be suitable for further development. The 3-allyl β-lactam compounds 10h and 10q demonstrated poor stability over the pH range studied. The 3-allyl β-lactam phenolic compound 10p demonstrated superior stability at all pH values, with over 95% remaining after 11 days, compared to 10h (24-41%) and 10q (22-26%).

Compound Number
Antiproliferative Activity IC 50  The antiproliferative activity of the amino acid and phosphate prodrugs synthesised was next evaluated in MCF-7 cells. The antiproliferative activity for the phenylalanine prodrug 13a (IC 50 = 4.915 µM) was low, indicating that in vivo hydrolysis of the amide is required to release the active amine 9s (Table 2) [83]. Reduced activity was also observed for amino acid prodrugs 13b-e (Tables 3 and 4). However, the 3-isopropenyl phosphate ester 15a displayed impressive antiproliferative activity, with an IC 50 value of 19 nM (Table 2), compared with parent phenol 9q (IC 50 = 10 nM), suggesting that this compound is a useful prodrug, as in vivo dephosphorylation to produce 9q would be expected to occur as observed for CA-4P [84]. The 3-isopropyl group in compound 15b resulted in reduced activity (Table 2). However, the phosphates 15c and 15d prepared from the 3-allyl phenolic 10p retained excellent activity, with IC 50 values of 94 nM and 32 nM, respectively ( To determine the effect of modification of the alkene substituent at C-3 of the βlactam ring of the designed compounds, a number of structurally related β-lactams were synthesised for evaluation. Compounds 17a-e were designed to determine the effect of the introduction of an aryl substituent at the C-3 allylic position of potent 3-allyl-β-lactam compounds 10h and 10p, while all aryl ring A and ring B methoxy substituents were removed in compound 17c. In compounds 17f and 17g, an ethyl ester group was introduced to replace the C-3 alkene substituent (isopropenyl, allyl and butadiene). In compounds 17d, 17e and 19, a trifluoromethylstyryl substituent was located at C-3 in place of the 3-(prop-1-en-2-yl) and 3-butadienyl series compounds.
The compounds were evaluated in the MCF-7 cell line, and the results are displayed in Table 5. Compounds 17a, 17b and 17c demonstrated reduced antiproliferative effects, compared with the parent compounds 10h and 10p. Interestingly, the ester compounds 17f and 17g demonstrated the most potent antiproliferative effects of the modified series in MCF-7 cells, with IC 50 values of 35 nM and 75 nM, respectively. Compound 19, having a trifluoromethylstyryl substituent at C-3 to replace the alkene, and 17e, having a trifluoroaryl substituent at the C-3 allylic position, both displayed moderate activity, while activity was poor for the ring A unsubstituted compound 17f and the epoxide 20. The antiproliferative activity of modified compounds 21a-d and 22a-d is presented in Table 5. A significant decrease in activity for all β-lactam compounds was observed compared with the parent 3-butadienyl β-lactam compounds and CA-4, which indicated that reduction in antiproliferative activity may result from increasing the substituent group size at the C-3 position of β-lactam ring, in line with our previous observations [85]. The most potent compound of the series was the trans diastereoisomer 22d with IC 50 = 5.33 µM, which was 3-fold more potent than the diastereomer 22c (IC 50 = 15.33 µM).  (Table 6). Triple-negative MDA-MB-231 breast cancer cells do not express the ER, progesterone (PR) or HER2 receptors and possess mutant p53 [86]. Triple-negative breast cancers (TNBCs) account for 10-15% of breast cancers diagnosed, have poor prognosis and are not responsive to hormone therapies, e.g., tamoxifen, to aromatase inhibitors such as anastrozole or to the HER2-receptor-targeting antibody Herceptin. Of the 3-isopropenyl series, the phenolic compound 9q was found to be the most effective, with an IC 50 value of 23.9 nM in the triple-negative MDA-MB-231 cell line. It compared favourably with the positive control CA-4 (IC 50 = 43 nM) in this cell line [87,88] and also with the result for 9q in the MCF-7 cell line (IC 50 = 10 nM). The p-methoxy compound 9h and the phosphate ester 15a were also very effective in the MDA-MB-231 cell line, with IC 50 values of 31.3 nM and 32 nM, respectively. Of the 3-allyl and butadienyl compounds evaluated in the MDA-MB-231 cell line, the phenolic 10p was notable, with IC 50 = 27 nM, together with the amine 10r (IC 50 = 35 nM), the methoxy 11h (IC 50 = 31 nM) and the amine derivative 11r (IC 50 = 33 nM). It is notable that the compounds retained potency in the MDA-MB-231 cell line, which may indicate their potential in development of therapeutic applications for this group of aggressive breast cancers.
In addition, the 3-isopropenyl compound 9q was found to be particularly effective in the chemoresistant HT-29 human colorectal adenocarcinoma cell line, with an IC 50 value of 0.711 µM, in comparison to the control CA-4 (IC 50 = 4.165 µM) in this cell line (Figure 2), indicating the potential of these compounds in chemoresistant colon cancers. Furthermore, 9q was also evaluated in the human leukaemia cell line HL-60 (acute myeloid leukaemia) and the colon adenocarcinoma cell line SW-480, with IC 50 values of 15 nM and 8 nM, respectively. It compares well with the control compound CA-4 (IC 50 = 2 nM and 7 nM, respectively, in these cell lines) (Figure 2). The physicochemical properties of the most potent compounds were next investigated.

Physicochemical Properties
The physicochemical properties and various metabolic properties of eleven selected β-lactam compounds from the panel synthesised were also determined to establish their druggability (see Supplementary Information Tables S4 and S5). The relevant physicochemical and pharmacokinetic properties of the most potent compounds, 9h, 9q, 9s, 10h, 10p, 10r, 11h, 11p, 11r, 15a and 17g, were obtained using Pipeline Pilot Professional [89]. The physicochemical properties of these compounds were found to comply with the requirements of Lipinski and Veber rules, with molecular weight range 445-457, hydrogen bond acceptor range 5-9, hydrogen bond donor range 0-2, 7-10 rotatable bonds and logP range 2.61-3.63. There is some correlation observed between the log P values calculated for the compounds and the antiproliferative activity determined (see Tables 4-6). The most potent compounds, 9q, 9s, 10q, 10r, 11q, 11r and phosphate 15a, with IC50 values in the range 10-61 nM in MCF-7 cells, have a low cLog P values in the range 1.235-2.55. However, the isomeric 1 and 2-naphthyl compounds 9m and 9n, each having higher cLogP values of 4.95, demonstrated a significant difference in potency, with IC50 values of 8.066 µM and

Physicochemical Properties
The physicochemical properties and various metabolic properties of eleven selected β-lactam compounds from the panel synthesised were also determined to establish their druggability (see Supplementary Materials Tables S4 and S5). The relevant physicochemical and pharmacokinetic properties of the most potent compounds, 9h, 9q, 9s, 10h, 10p, 10r, 11h, 11p, 11r, 15a and 17g, were obtained using Pipeline Pilot Professional [89]. The physicochemical properties of these compounds were found to comply with the requirements of Lipinski and Veber rules, with molecular weight range 445-457, hydrogen bond acceptor range 5-9, hydrogen bond donor range 0-2, 7-10 rotatable bonds and logP range 2.61-3.63. There is some correlation observed between the log P values calculated for the compounds and the antiproliferative activity determined (see Tables 4-6). The most potent compounds, 9q, 9s, 10q, 10r, 11q, 11r and phosphate 15a, with IC 50 values in the range 10-61 nM in MCF-7 cells, have a low cLog P values in the range 1.235-2.55. However, the isomeric 1 and 2-naphthyl compounds 9m and 9n, each having higher cLogP values of 4.95, demonstrated a significant difference in potency, with IC 50 values of 8.066 µM and 0.139 µM, respectively, in MCF-7 cells, indicating that the 2-naphthyl compound 9n may be a better fit at the colchicine-binding site.
The pK a H value for potent compound 9q was calculated with Marvin as 9.83, while the phosphate 15a was ionised at physiological pH with pK a H values of 1.62 and 6.59. The calculated topological polar surface area (TPSA) of this panel was in the range 57-103 Å 2 , below the required limit of <140 Å 2 for good intestinal absorption and membrane permeability.
Additionally, three of the compounds, 9h, 10h and 11h, followed the Pfizer rule and GSK rule (MW ≤ 400, log P ≤ 4), with a low log P value (log P < 3) and a low TPSA value (TPSA <75 Å 2 ). The compounds exhibited good drug-likeness parameters with predicted lipophilic-hydrophilic balance, together with high blood-brain barrier (BBB) absorption and plasma-protein-binding properties (>90%), and were not predicted to inhibit the metabolic activity of CYP2D6 (see Supplementary Materials Tables S4 and S5 for details).
Good aqueous solubility (logSw = −3.4130 mol/L) was predicted for the phenolic ester compound 17g, although lower aqueous solubility was predicted for the remaining panel of β-lactam compounds (see Supplementary Materials, Table S4); phosphate esters such as 15a-d and amino acid prodrugs such as 13a-e may result in increased water solubility, as reported for CA-4 and related compounds [72,74]. The most potent compounds identified in the preliminary screening panel [9h, 9q, 9s, 10h, 10p, 10r, 11h, 11p, 11r, 17g] were confirmed as free from pan-assay interference compounds (PAINS) alerts [90] and were identified as suitable candidate compounds for subsequent in vitro biochemical investigations based on the phenotypic screening and Tier-1 profiling of their drug-like properties (Tables S4 and S5 The more potent compounds 9h, 9q, 9s, 10h, 10p, 10r, 11h, 11p, 11r, 15a and 15b were selected for screening for antiproliferative activity by the National Cancer Institute (NCI)/Division of Cancer Treatment and Diagnosis (DCTD)/Developmental Therapeutics Program (DTP) [92] in their drug screening programme and evaluated using approximately 60 different human cancer cell lines. The initial NCI 60 cell line screen used the sulforhodamine B (SRB) protein assay [93], at one dose concentration (10 µM), and subsequently, a five-dose screening in the concentration range 0.01-100 µM was carried out. GI 50 (50% growth inhibition), TGI (total growth inhibition) and LC 50 (50% lethal concentration) were determined (see Table 7 and Supplementary Materials Tables S6-S8). The compounds demonstrated excellent broad-spectrum antiproliferative activity against a range of tumour cell lines contained in the NCI panel of cancer cell lines, e.g., leukaemia, colon cancer, CNS cancer, melanoma, non-small-cell lung cancer, ovarian cancer, renal cancer, breast cancer and prostate cancer, and the results confirmed our in-house evaluations in the MCF-7 cell line. The ring B phenolic 3-isopropenyl compound 9q was identified as the most potent compound synthesised in our panel, with a mean GI 50 value obtained across the entire NCI panel of cell lines screened of 39.81 nM. (See Figure 3 for a heatmap of the activity of compound 9q across the cell lines in the NCI-60 screen). This result compares favourably with the mean GI 50 for CA-4 of 99.30 nM in the NCI-60 cell panel, demonstrating the superior inhibitory potency of the β-lactam analogue (see Table 7 and Supplementary  Materials Tables S6-S8 for further   The ring B phenolic 3-isopropenyl compound 9q was identified as the most potent compound synthesised in our panel, with a mean GI50 value obtained across the entire NCI panel of cell lines screened of 39.81 nM. (See Figure 3 for a heatmap of the activity of compound 9q across the cell lines in the NCI-60 screen). This result compares favourably with the mean GI50 for CA-4 of 99.30 nM in the NCI-60 cell panel, demonstrating the superior inhibitory potency of the β-lactam analogue (see Table 7 and Supplementary Information Tables S6-S8 for further     The mean GI 50 values for the panel of 60 cell lines for the most potent compounds evaluated (9h, 9q, 9s, 10h, 10p, 10r, 11h and 11r) were determined to be <91.20 nM, as shown in Table 7. These results compare very favourably with the mean GI 50 value determined for CA-4 of 99.3 nM in this 60-cell-line panel.
The COMPARE programme developed by the NCI was used to further investigate the mechanism of action of the β-lactam series [95]. The antiproliferative profiles of potent compounds 9q and 9s were compared with compounds possessing known mechanisms of antiproliferative action in the NCI Standard Agents Database. Based on GI 50 , TGI and LC 50 , the compounds 9q and 9s demonstrated a good correlation to tubulin-targeting agents such as maytansine (r = 0.741) and also to the clinically utilised tubulin-targeting anticancer drugs vincristine (r = 0.664), vinblastine (r = 0.632) and paclitaxel (r = 0.768) (see Supplementary Materials, Tables S9 and S10).

Effect of β-Lactam 9q on Non-Carcinogenic HEK-293T Cells
The 3-(prop-1-en-2-yl)azetidin-2-one compounds 9h and 9q were next examined for cytotoxic effects in MCF-7 cells at 10 µM concentration using the lactate dehydrogenase (LDH) assay, in which the release of cytoplasmic LDH is used as a measure of cell lysis [96]. The b-lactams 9h and 9q resulted in 30% and 32% cell death, respectively. CA-4 was used as the positive control in this assay, with 12% cell death at 10 µM.
The cytotoxicity of the most potent example of the 3-isopropenyl series, 9q, was also investigated in the non-tumorigenic cell line HEK-293T (normal human epithelial embryonic kidney cells). The viability of the normal HEK-293T cells was demonstrated to be significantly higher than the treated MCF-7 cells following exposure to compound 9q for 72 h. The cell viabilities observed in the HEK-293T cells were 83%, 60% and 50% at the (LDH) assay, in which the release of cytoplasmic LDH is used as a measure of cell lysis [96]. The b-lactams 9h and 9q resulted in 30% and 32% cell death, respectively. CA-4 was used as the positive control in this assay, with 12% cell death at 10 µM.
The cytotoxicity of the most potent example of the 3-isopropenyl series, 9q, was also investigated in the non-tumorigenic cell line HEK-293T (normal human epithelial embryonic kidney cells). The viability of the normal HEK-293T cells was demonstrated to be significantly higher than the treated MCF-7 cells following exposure to compound 9q for 72 h. The cell viabilities observed in the HEK-293T cells were 83%, 60% and 50% at the concentrations of 0.1 µM, 1 µM and 10 µM, respectively ( Figure 4); this result compares very favourably with the viabilities obtained in MCF-7 cells of 22%, 21% and 11% at the concentrations of 0.1 µM, 1 µM and 10 µM, respectively, with IC50 = 10 nM in the MCF-7 cell line, demonstrating that β-lactam 9q was selectively less toxic to human normal cells (HEK-293T) than cancer cells (MCF-7). These results indicate that compound 9q is suitable for further development as an anticancer agent for breast cancers with selective lower cytotoxicity to normal cells. Additionally, the mean TGI (total growth inhibition) value for the potent compound 9q over the NCI-60 cancer cell line panel was 32 mM (compared with TGI for CA-4 of 10.30 µM), with TGI values >100 µM for 34 of the cell lines tested indicating a wide therapeutic window for the compound (Table 7).
LC 50 values obtained from the NCI screen over the 60 cancer cell lines were also useful in assessing the cytotoxicity of these compounds (Table 7). For compound 9q, LC 50 values were greater than 100 µM in all but two of the cell lines tested (melanoma M14 and colon COLO 205), which indicated minimal toxicity and suggested the potential for this compound in a number of therapeutic applications (Tables S6-S8 Supporting Information). For compound 9h, the LC 50 values were greater than 100 µM in all but one of the cell lines tested (melanoma SK-MEL-5). Similarly low cytotoxicity was also obtained for the related compounds evaluated in the series, e.g., compounds 9s, 15a, 15b, 10r, 10p, 11r and 11p. The antiproliferative activity determined for 9q (IC 50 <10 nM against the MCF-7 cell line) demonstrated a significant therapeutic window between the drug concentration required for cancer cell growth inhibition (GI 50 ) and the concentration that is toxic to these MCF-7 cells, LC 50 (>100 µM). For compound 9q, the mean GI 50 value over all 60 cell lines was 39.81 nM, while the mean LC 50 value was 81.28 µM.

Cell Cycle and Pro-Apoptotic Effects of 3-(Prop-1-en-2-yl)azetidinone 9q
The effects of the β-lactam 9q on the cell cycle in MCF-7 cells were next explored in flow cytometry studies. Tubulin-destabilising agents such as colchicine and CA-4 arrest the cell cycle in the G 2 /M phase, promoting depolymerisation of microtubules and disruption of the cytoskeleton, while the antimitotic action of paclitaxel is to stabilise the microtubule polymer and prevent disassembly. MCF-7 cells were treated with 9q and analysed by propidium iodide staining ( Figure 5). The effect of compound 9q on the mitotic phase G 2 /M (50 µM and 500 µM) at 24 h, 48 h and 72 h was first determined ( Figure 5B). For the 50 µM concentration, the G 2 /M population increased at each of the time points examined to 72% (24 h), 69% (48 h) and 70% (72 h), while this cell population also increased when treated at 500 nM to 85% (24 h), 81% (48 h) and 70% (72 h). In contrast, the G 0 G 1 cell population decreased (10% and 7% (24 h), 10% and 7% (48h) and 8% and 8% (72 h)) when treated with 9q at 50 nM and 500 nM, respectively ( Figure 5A). The vehicle control (ethanol 0.1% v/v) is shown for comparison ( Figure 5). A proapoptotic effect (sub-G 0 G 1 ) was evident for compound 9q (10% and 19% at 50 nM and 500 nM, respectively) at 72 h, compared with vehicle control (6%) ( Figure 5C). To further investigate and confirm the induction of apoptosis by 9q in MCF-7 cells, a dual staining with annexin-V and with propidium iodide (PI) was performed ( Figure 6). Live cells (annexin-V-/PI-), early apoptotic cells (annexin-V+/PI-), late apoptotic cells (annexin-V+/PI+) and necrotic cells (annexin-V-/PI+) can be identified. Compound 9q induced an increase in apoptosis (observed as annexin-V-positive cells) in a concentrationdependent manner when compared to CA-4 ( Figure 6) when examined at 72 h, with 27% To further investigate and confirm the induction of apoptosis by 9q in MCF-7 cells, a dual staining with annexin-V and with propidium iodide (PI) was performed ( Figure 6).

Live cells (annexin-V-/PI-), early apoptotic cells (annexin-V+/PI-), late apoptotic cells (annexin-V+/PI+) and necrotic cells (annexin-V-/PI+) can be identified. Compound 9q
induced an increase in apoptosis (observed as annexin-V-positive cells) in a concentrationdependent manner when compared to CA-4 ( Figure 6) when examined at 72 h, with 27% of cells undergoing apoptosis (early+late) at 50 nM concentration of 9q, and 38.9% at 500 nM. CA-4 (50 nM) induced apoptosis in 34.6% of the MCF-7 cells. Cell death may also be due to autophagy following treatment with 9q [97]. Collectively, these findings indicate that the β-lactam compound 9q demonstrated tubulin-targeting effects, e.g., G 2 /M arrest followed by apoptosis, on cell cycle in MCF-7 cells. The effects of the compounds on tubulin polymerisation were further examined. nM. CA-4 (50 nM) induced apoptosis in 34.6% of the MCF-7 cells. Cell death may also be due to autophagy following treatment with 9q [97]. Collectively, these findings indicate that the β-lactam compound 9q demonstrated tubulin-targeting effects, e.g., G2/M arrest followed by apoptosis, on cell cycle in MCF-7 cells. The effects of the compounds on tubulin polymerisation were further examined.

Tubulin Polymerisation Effects of 3-(Prop-1-en-2-yl)azetidinones, 3-Allylazetidinones and 3-butadienylazetidinones
A panel of β-lactam compounds that demonstrated the most potent antiproliferative effects in vitro were selected for study of their inhibitory effect on tubulin assembly using a light-scattering assay. CA-4 was used as the positive control for the tubulin polymerization experiment, which effectively inhibits the assembly of bovine tubulin; paclitaxel was the positive control for polymerisation, with ethanol and DMSO as the solvent vehicles. (Figure 7). The V max value for the polymerization reaction for each compound was determined (Table 8), in addition to the fold-changes in the V max values for the polymerisation reaction [98]. A panel of β-lactam compounds that demonstrated the most potent antiproliferative effects in vitro were selected for study of their inhibitory effect on tubulin assembly using a light-scattering assay. CA-4 was used as the positive control for the tubulin polymerization experiment, which effectively inhibits the assembly of bovine tubulin; paclitaxel was the positive control for polymerisation, with ethanol and DMSO as the solvent vehicles. (Figure 7). The V max value for the polymerization reaction for each compound was determined (Table 8), in addition to the fold-changes in the V max values for the polymerisation reaction [98].   The ring B phenolic 3-allyl 10p was identified as the most potent polymerisation inhibitor in the series (V max 0.0015 mOD/min), with 4.5-fold reduction in V max value at 10 µM concentration. This result compares favourably to CA-4 (10 µM), for which a 6.3-fold inhibition in the V max for polymerisation was observed. These V max results show correlation with the antiproliferative data recorded for both CA-4 (IC 50 = 4.2 nM) and 10p (IC 50 = 31 nM) in the MCF-7 line and indicate that tubulin is the molecular target for this series of 3-allyl-β-lactams. The most potent antiproliferative compound, 3-(prop-1en-2-yl)azetidinone 9q (IC 50 = 10 nM), was also found to be a potent inhibitor of tubulin polymerization, demonstrating a 4.27-fold reduction in V max value at 10 µM concentration, together with the ring B amino 3-allyl compound 10r (3.4-fold reduction in V max value) and the ring B methoxy-3-butadienyl-β-lactam 11h (2.9-fold reduction in V max value).
In a further investigation of the mechanism of action of the series of β-lactams, the interaction of the most potent antiproliferative compound, 3-(prop-1-en-2-yl)azetidinone 9q, at the colchicine-binding site of tubulin was examined. The colchicine-binding site of tubulin is characterised by the Cys239 and Cys354 thiol-containing residues. The alkylating reagent N,N'-ethylene-bis(iodoacetamide) (EBI) reacts with the thiol-containing amino acid residues of cysteine 239 and cysteine 354 to crosslink [99,100].
In the present work, the β-lactam 9q (10 µM) or CA-4 was used to treat MCF-7 cells for 2 h; EBI was then added, and the cells were incubated for 15 h (Figure 8). Vehicle-treated control samples were observed at a lower position on the gel, confirming the formation of the β-tubulin-EBI adduct and demonstrating that the alkylating reagent EBI had formed cross links on β-tubulin with the cysteine thiol residues Cys239 and Cys354. The formation of the EBI adduct was prevented when the MCF-7 cells were treated with β-lactam 9q and also with CA-4, demonstrating that the β-lactam 9q interacts with tubulin at the colchicine site of tubulin. Evidence of the effects of the β-lactam 9q on the molecular structure of the target tubulin was investigated using immunofluorescence and confocal microscopy studies, through which the changes in the microtubule network structure produced by β-lactam 9q in MCF-7 cells can be observed. MCF-7 cells demonstrated an organised microtubular network structure following staining with α-tubulin mAb (Figure 9) in the presence of the vehicle control (1% ethanol (v/v)). The cells were also treated with paclitaxel, which acts as a microtubule-stabilising agent [101]; paclitaxel clearly induced the formation of microtubule bundles and pseudo-asters. When the MCF-7 cells were treated with CA-4 or βlactam 9q for 16 h, microtubule formation was inhibited and resulted in depolymerised microtubules (Figure 9), with multiple micronuclei present in these cells. Mitotic catastrophe, characterised by the appearance of multinucleated cells, is a type of programmed cell death occurring during mitosis. It results from a combination of deficient cell-cycle checkpoints (in particular, the DNA structure checkpoints and the spindle assembly checkpoint) and also cellular damage [102]. Evidence of the effects of the β-lactam 9q on the molecular structure of the target tubulin was investigated using immunofluorescence and confocal microscopy studies, through which the changes in the microtubule network structure produced by β-lactam 9q in MCF-7 cells can be observed. MCF-7 cells demonstrated an organised microtubular network structure following staining with α-tubulin mAb (Figure 9) in the presence of the vehicle control (1% ethanol (v/v)). The cells were also treated with paclitaxel, which acts as a microtubulestabilising agent [101]; paclitaxel clearly induced the formation of microtubule bundles and pseudo-asters. When the MCF-7 cells were treated with CA-4 or β-lactam 9q for 16 h, microtubule formation was inhibited and resulted in depolymerised microtubules (Figure 9), with multiple micronuclei present in these cells. Mitotic catastrophe, characterised by the appearance of multinucleated cells, is a type of programmed cell death occurring during mitosis. It results from a combination of deficient cell-cycle checkpoints (in particular, the DNA structure checkpoints and the spindle assembly checkpoint) and also cellular damage [102]. Induction of mitotic catastrophe by CA-4 was previously reported in nonsmall-cell lung cancer and breast cancer cells (MCF-7) [103]. The confocal imaging results support the proposed tubulin-targeting action of the 3-(prop-1-en-2-yl)azetidinone 9q.

Computational Modelling of β-Lactam Compounds
The 3-(prop-1-en-2-yl)azetidin-2-one compound 9q was identified as the most potent compound synthesised in this study (IC 50 = 10 nM in MCF-7 breast cancer cells) and also as an inhibitor of tubulin polymerisation. The tubulin-binding and related immunofluorescence studies of 9q indicated that the colchicine-binding site of tubulin is the target for the series of compounds. The structural similarity of the β-lactam compounds with CA-4 was demonstrated by X-ray studies revealing the similar torsional angle observed between rings A and B and a flexible alignment of the 3-butadienyl β-lactam compound 11p with CA-4 (Supplementary Figure S22), which showed excellent overlap with the 3,4,5trimethoxyphenyl ring A and phenolic ring B of both compounds. In a further comparison of the energy-minimised structures of compound 11p and CA-4, the inter-atomic distances of the methoxy oxygens of ring A and ring B is 9.16 Å, while for CA-4, this distance is slightly longer (9.27 Å) (Supplementary Figure S23).
Although the compounds were biologically evaluated as racemates, it was interesting that the 3S,4R enantiomer of each compound was found to be ranked at lower energy in the docking study than the corresponding 3R,4S enantiomer. We had previously reported stereochemical selectivity in docking energy calculated for related β-lactam compounds [54,105]. However, a very small difference was observed in the cellular efficacy of this series of compounds in the modelling study (e.g., IC 50 values in the range 10-61 nM), so it would not be expected to see a large difference in ranking. Indeed, the docking scores only differ by <0.8 from best-to worst-ranked (see Supplementary Materials, Table S11). Thus, docking studies are not ideal for studying changes in cellular efficacy associated with small changes in the β-lactam scaffold substituents located at C-3.
The top three ranked compounds, 3-acetate ester 17e, 3-butadienyl-β-lactam 11p and 3-(prop-1-en-2-yl)-β-lactam 9q, all contained a meta-hydroxyl substituent on the B ring. Figure 10 shows the best docking pose of the top ranked phenolic compounds, 9q, 10p, 11p and 17e. For the 3-(prop-1-en-2-yl) compound 9s, the HBA interaction of the ring B amino group with Lys252 is clearly observed, while for the ring B methoxy compound 11h, the interaction of this methoxy group with the Thr353 residue is also illustrated, together with the β-lactam carbonyl interaction with Ala250 and Leu β248 (hydrophobic), which are residues of the T7 loop H8 helix, observed for all β-lactam compounds and also for colchicine. Although all molecules are located slightly deeper in the binding pocket than DAMA-colchicine, nevertheless, they demonstrate the observed ligand-protein interactions. (See Supplementary Materials, Figures S24-S26 for additional molecular  modelling illustrations for compounds 9h, 9q, 9s, 10h, 11p and 11r). copy confirmed that only the trans isomers of the compounds were obtained in the synthesis. All trimethoxy compounds overlaid their B-rings on the C-ring of DAMA-colchicine (forming HBA interactions with Lys352 and with the 4-methoxy group close to Thr353), co-located the 3,4,5-trimethoxyphenyl-substituted A-rings (with interactions possible with Cys241 and Ala326, Ala317 and Val318) and positioned the 3-alkenyl or ester side chain in an open region of the tubulin-binding site at the dimer interface (all amino acid residues refer to α-tubulin). The predicted docking ranking from best to worst was 17d, 11p, 9q, 11h, 9s, 9h, 10p, 10h, 11r and 10r (see Supplementary Information, 4R enantiomer of 9h, 9q, 10p, 10r, 11h, 17g.
Ligands are rendered as tubes and amino acids as lines. Tubulin amino acids and DAMA-colchicine are coloured by atom type: carbon = grey, hydrogen = white, oxygen = red, nitrogen = blue. The betalactam is 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.
Although the compounds were biologically evaluated as racemates, it was interesting that the 3S,4R enantiomer of each compound was found to be ranked at lower energy in the docking study than the corresponding 3R,4S enantiomer. We had previously reported stereochemical selectivity in docking energy calculated for related β-lactam compounds [54,105]. However, a very small difference was observed in the cellular efficacy of this se-  Figure 10. Overlay of the X-ray structure of tubulin co-crystallised with DAMA-colchicine (PDB entry 1SA0) on the best-ranked docked pose of the 3S ,4R enantiomer of 9h, 9q, 10p, 10r, 11h, 17g. Ligands are rendered as tubes and amino acids as lines. Tubulin amino acids and DAMA-colchicine are coloured by atom type: carbon = grey, hydrogen = white, oxygen = red, nitrogen = blue. The beta-lactam is 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.
We recently developed a novel pharmacophore generation tool, MoPBS [106]. In brief, MoPBS floods the protein-binding site with fragments describing classic molecular interaction features (acetate ion, benzene, methane and methylammonium) and independently minimising their positions within the binding site. Each fragment clusters in the region of complementary amino acids, thereby mapping out preferred interaction locations. By applying K-means clustering algorithms, we, in this case, reduce the clusters to eight pharmacophore features. Application of this tool to the 1SA0 binding site yielded a pharmacophore shown in Figure 11. As expected, many of the features mapped to those present in DAMA-colchicine and 17g, such as the two aromatic cores and the hydrogen bond acceptor (HBA) interacting with Lys352. Interestingly, a unique feature is present in the β-lactam, in that the carbonyl oxygen atom maps to a HBA feature. The pharmacophores also point towards future synthetic possibilities, such as introducing a hydrogen bond donor (F3 HBD) in place of the methoxy group on the compound 17g B-ring, including more hydrophobicity off the trimethoxy phenyl ring (F5) or extending the compound towards the region of space occupied by the F1 HBD feature. rmaceuticals 2023, 16, x FOR PEER REVIEW Figure 11. Mapping of the X-ray structure of tubulin co-crystallised with entry 1SA0) and the best ranked docked pose of the 3S,4R enantiomer 17g created by MoPBS. The ligand and protein colouring is explained in the leg cophore feature colours are HBA, light blue; HBD, pink; aromatic, orange; a

Materials and Methods: Chemistry
Melting points were measured on a Gallenkamp SMP 11 meltin Figure 11. Mapping of the X-ray structure of tubulin co-crystallised with DAMA-colchicine (PDB entry 1SA0) and the best ranked docked pose of the 3S,4R enantiomer 17g, with a pharmacophore created by MoPBS. The ligand and protein colouring is explained in the legend above. The pharmacophore feature colours are HBA, light blue; HBD, pink; aromatic, orange; and hydrophobic, green.

Materials and Methods: Chemistry
Melting points were measured on a Gallenkamp SMP 11 melting point apparatus and are uncorrected. Infrared (IR) spectra were recorded as thin film on NaCl plates or as potassium bromide discs on a Perkin Elmer FT-IR Spectrum 100 spectrometer. 1 H and 13 C nuclear magnetic resonance (NMR) spectra were recorded at 27 • C on a Bruker Avance DPX 400 spectrometer (400.13 MHz, 1 H; 100.61 MHz, 13 C) at 20 • C in CDCl 3 (internal standard tetramethylsilane TMS) or DMSO-d 6 by Dr. John O'Brien and Dr. Manuel Ruether, School of Chemistry, Trinity College Dublin. For CDCl 3 , 1 H-NMR spectra were assigned relative to the TMS peak at 0.00 δ and 13 C-NMR spectra were assigned relative to the middle CDCl 3 triplet at 77.00 ppm. Electrospray ionisation mass spectrometry (ESI-MS) on a liquid chromatography time-of-flight (TOF) mass spectrometer (Micromass LCT, Waters Ltd., Manchester, UK) equipped with electrospray ionization (ES) interface operated in the positive ion mode with high-resolution mass measurement accuracies of <±5 ppm. R f values are quoted for thin-layer chromatography on silica gel Merck F-254 plates. Flash column chromatography was carried out on Merck Kieselgel 60 (particle size 0.040-0.063 mm) and also on Biotage SP4 instruments. All products isolated were homogenous on TLC. Analytical high-performance liquid chromatography (HPLC) for purity determination of products was performed using a Waters 2487 Dual Wavelength Absorbance detector, Waters 1525 binary HPLC pump, Waters In-Line Degasser AF, Waters 717plus Autosampler and Varian Pursuit XRs C18 reverse-phase 150 × 4.6 mm chromatography column with detection at 254 nm. Imines 8a-w and azetidine-2-ones 16a-c and 18 were prepared following the reported procedures [52,53].

General Method II: Preparation of β-Lactams (10a-s)
To a stirring, refluxing solution of the appropriate imine (5 mmol) and triethylamine (6 mmol) in anhydrous dichloromethane (40 mL), a solution of 4-pentenoyl chloride (6 mmol) in anhydrous dichloromethane (10 mL) was added dropwise over 45 min under nitrogen. The reaction mixture was heated at reflux for 5 h, and then stirred at 20 • C for 20 h. The reaction mixture was washed with water (2 × 100 mL), the organic layer was dried (Na 2 SO 4 ), and the solvent was removed under reduced pressure. The crude product was purified by flash chromatography over silica gel (eluent: 4:1 n-hexane: ethyl acetate).  13 13

General Method III: Preparation of β-Lactams (11a-11s)
Sorbic acid (2 mmol) was mixed with 2-chloro-1-methylpyridinium iodide (2.4 mmol) and tripropylamine (6 mmol) in anhydrous dichloromethane (30 mL) under a nitrogen atmosphere at room temperature. The suspension was then heated to reflux for 12 h to afford a clear solution. A solution of the appropriate imine (2 mmol) in anhydrous dichloromethane (10 mL) was added, and reaction mixture was heated at reflux for 24 h. The solution was then cooled and washed with water, HCl (2%, aqueous solution) and water. The organic layer was dried over anhydrous Na 2 SO 4 , and the solvent was removed under reduced pressure. The crude product was purified by column chromatography over silica gel (eluent, 4:1 n-hexane and ethyl acetate).  13 13   Preparation following the general method III above from sorbic acid and (4-bromobenz ylidene)-3,4,5-trimethoxyphenylamine 8c afforded the product as a colourless solid (yield (d, J = 8.04 Hz, 2H), 7.31-7.45 (m, 7H). 13 13 13 13 13 13 13
2-Amino-N-(2-methoxy-5-(1-(3,4,5-trimethoxyphenyl)-4-oxo-3-(prop-1-en-2-yl) azetidin-2-yl)phenyl)-3-phenylpropanamide (13a) Following the general method V above, to amino acid amide 10 (1.56 mmol) in methanol (10 mL)/CH 2 Cl 2 (10 mL) was added 2N NaOH (3.4 mmol) at room temperature, and the mixture was stirred for 24 h. The product was isolated by flash chromatography over silica gel (eluent, dichloromethane-methanol gradient) as a yellow oil, yield 58%  13 13   To a solution of phenol 9q (17 mmol) in acetonitrile (100 mL), cooled to 0 • C, was added carbon tetrachloride (85 mmol). The solution was stirred for 10 min prior, and then diisopropylethylamine (35 mmol) and dimethylaminopyridine (1.7 mmol) were added. The dibenzyl phosphate (24.5 mmol) was then added dropwise to the mixture. When the reaction was complete, 0.5M KH 2 PO 4 (aq) was added and the reaction mixture was warmed to room temperature. The mixture was extracted with ethyl acetate (3 × 50 mL), washed with saturated sodium chloride (aqueous, 100 mL) followed by water (100 mL) and dried (Na 2 SO 4 ). The solvent was reduced in vacuo, and the product was isolated by flash column chromatography over silica gel (n-hexane: ethyl acetate gradient). Yield: 45%, 342 mg, brown oil. IR (NaCl, film)  13 13   Dibenzyl phosphate ester 14a (0.27 mmol) was dissolved in dry dichloromethane (5 mL) under nitrogen at 0 • C. Bromotrimethylsilane (0.59 mmol) was added to reaction mixture and allowed to stir for 45 min. Sodium thiosulphate solution (10%, 5 mL) was added to the reaction, and stirring was continued for 5 min. The aqueous phase was extracted with ethyl acetate (3 × 25 mL). The combined organic phases were concentrated in vacuo and purified by flash chromatography on silica gel (eluent: n-hexane: ethyl acetate, 1:1) to afford the product as an off-yellow solid, yield 57%, Mp > 300 •   The dibenzylphosphate ester 14a (2 mmol) was dissolved in ethanol: ethyl acetate (50 mL; 1:1 mixture) and hydrogenated over 1.2 g of 10% palladium on carbon until complete on TLC, typically for less than 3 h. The catalyst was filtered, the solvent was removed in vacuo, and the product was isolated by flash column chromatography over silica gel (eluent, n-hexane: ethyl acetate gradient) to afford the desired product as a brown oil, 239 mg, yield 98% (HPLC: 73 13   The dibenzyl phosphate ester 14b (2 mmol) was dissolved in ethanol: ethyl acetate (50 mL; 1:1 mixture) and hydrogenated over 1.2 g of palladium (10%) on carbon until complete as monitored by TLC, approx. 3 h. The catalyst was filtered, the solvent was reduced in vacuo, and the product was isolated by flash column chromatography over silica gel (eluent, n-hexane: ethyl acetate gradient) to afford the product as a brown oil, yield 88%,  13 13  All biochemical assays were performed in triplicate for the determination of mean values reported. The human breast carcinoma cell line MCF-7 was purchased from the European Collection of Animal Cell Cultures (ECACC) and was cultured in Eagle's minimum essential medium with 10% fetal bovine serum, 2 mM L-glutamine and 100 µg/mL penicillin/streptomycin. The medium was supplemented with 1% non-essential amino acids. The human breast carcinoma cell line MDA-MB-231 was purchased from the ECACC. MDA-MB-231 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum, 2 mM L-glutamine and 100 µg/mL penicillin/streptomycin (complete medium). HEK-293T normal epithelial embryonic kidney cells were cultured in DMEM with GlutaMAX TM -I in the absence of non-essential amino acids. HL-60 cells were originally obtained from the ECACC. SW-480 cells were a kind gift from Dr. Emma Creagh, School of Biochemistry and Immunology, Trinity College Dublin. SW-480 cells were cultured in DMEM with GlutaMAX-I, with the same supplement in the absence of non-essential amino acids. HL-60 cells were cultured in RPMI-1640 Glutamax 1 medium supplemented with 10% FBS media, and 100 µg/mL penicillin/streptomycin. HT-29 cells originate from a human adenocarcinoma of the colon, were originally obtained from the ECACC and were grown in DMEM Glutamax media. HT-29 media were supplemented with 10% foetal bovine serum (FBS). Cells were maintained at 37 • C in 5% CO 2 in a humidified incubator. All cells were sub-cultured 3 times/week by trypsinisation.

Cell Viability Assay
Cells were seeded at a density of 5 × 10 3 cells/well (MCF-7) in triplicate in 96-well plates. After 24 h, cells were then treated with medium alone, with vehicle (1% ethanol (v/v)) or with selected dilutions of control CA-4 or the synthesised azetidinone compounds in the concentration range 1-50 µM. Cell proliferation for MCF-7 and MDA-MB-231 cells was analysed using the Alamar Blue assay (Invitrogen Corp.). After 72 h, Alamar Blue (10% (v/v)) was added to the contents of each well, and plates were then incubated in the dark for 3-5 h at 37 • C. Fluorescence results were obtained with a 96-well fluorimeter operating with excitation (530 nm) and emission (590 nm), and the results were expressed as viability (%) relative to vehicle control (100%). IC 50 values (concentration of drug resulting in 50% reduction in cell survival) were obtained from the dose response curves using Prism (GraphPad Software, Inc., La Jolla, CA, USA). Experiments were performed in triplicate on at least three separate occasions. vial. A sample (250 µL) was then added to the Eppendorf tube containing ZnSO 4 .7H 2 O solution (500 µL) (2% w/v ZnSO 4 solution in acetonitrile:water, 1:1). The samples were centrifuged (10,000 rpm, 3 min), filtered (0.2 micron filter) and analysed by HPLC as above. Further samples were taken at one-hour intervals.

X-ray Crystallography
Data for 8h and 8i were measured on a Bruker APEX DUO, and data for 10h-11t were measured on a Bruker D8 Quest ECO, using Mo Kα radiation (λ = 0.71073 Å). Each sample was mounted on a MiTeGen cryoloop, and data were collected at 100(2) K using an Oxford Cryosystems Cobra (8h, 8i) or Cryostream (10h-11t) low-temperature device. Bruker APEX [108,109] software was used to collect and reduce data. Absorption corrections were applied using SADABS [110]. Structures were solved with the SHELXT structure solution program [111] using intrinsic phasing. All were refined using least-squares method on F 2 with SHELXL [112]. 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 [113]. Crystal data and details of data collection and refinement are provided in Table S1, Supplementary Materials. In 10h, there are two independent molecules in the asymmetric unit. Part of one beta-lactam ring (C32/C32b) and the vinylic substituent are disordered and modelled in two positions with 49:51% occupancy with displacement restraints (SIMU). In 11t, the donor hydrogens were located and refined, and in 11h, the sample was weakly diffracting, with weak high-angle data resulting in a high R(int The 1SA0 X-ray structure of bovine tubulin co-crystallised with N-deacetyl-N-(2mercaptoacetyl)-colchicine (DAMA-colchicine) was downloaded from the PDB website [18]. 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 2022 with the MMFF94x force field. Each compound was drawn in MOE, saved as an mdb and processed in MOE [104]. 3S,4R trans enantiomers of the compounds were examined. For each compound, MMFF94x partial charges were calculated, and each was energyminimised 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 generated for each compound were retained for subsequent analysis.

Mapping of Protein-Binding Sites (MoPBS) Algorithm
The 1SA0 X-ray structure of bovine tubulin co-crystallised with N-deacetyl-N-(2mercaptoacetyl)colchicine (DAMA-colchicine) [18] was used after the processing above was completed. As described above and in our recent paper [106], the binding site was flooded with 100 copies of each fragment-in this case, a more focused selection of four fragments (acetate ion, benzene, methane, methylammonium) compared to the nine fragments used in the original report. The DBSCAN K-means algorithm was used to cluster eight pharmacophore features from the 400 minimised fragments.

Conclusions
Microtubule-targeting agents (MTA) such as paclitaxel and docetaxel and the vinca alkaloids vincristine, vinblastine and vinorelbine are widely used in chemotherapy for Pharmaceuticals 2023, 16, 1000 56 of 63 a variety of different cancers. Extensive preclinical studies have shown that CBSIs are promising drug candidates for cancer therapy. CA-4-based codrugs (or mutual prodrugs) have recently been demonstrated to improve the therapeutic profile of clinically used drugs such as doxorubicin, floxuridine and tegafur, and they have achieved targeted delivery of drugs to cancer tissues [114]. Colchicine-binding-site-targeting compounds are of interest not only for their potent antimitotic effects [11,[115][116][117]; in a recent study, the potent antimitotic sabizabulin 5 was evaluated in a clinical trial in metastatic breast cancer and also as an antiviral agent for the treatment of hospitalised COVID-19 patients at high risk for acute respiratory distress syndrome [118]. Sabizabulin binds to viral tubulin at the colchicine-binding site and disrupts the intracellular transport of the virus. Therefore, investigation of the non-covalent tubulin-targeting b-lactam compounds developed in the present work as potential antibacterial agents is also of current interest.
Increasing evidence is also available in relation to the non-mitotic effects of tubulintargeting compounds, e.g., as vascular disrupting agents (VDA) and inhibitors of tumour angiogenesis [119]. Many small-molecule dual-targeting tubulin inhibitors have been reported with applications for cancer therapy [120]; e.g., novel CA-4 sulphamate derivatives identified for dual tubulin polymerization and arylsulphatase inhibitors having potential for applications in cancer therapy [121]. Dual CBSIs and Src kinase inhibitors are reported to be effective in regulating the overexpression of tubulin isotypes and can inhibit drug resistance, which is mediated by P-glycoprotein (P-gp), and other multidrug-resistanceassociated proteins (MRP1, MRP2) [122]. Interestingly, it is suggested that microtubules may modulate immune responses, and therefore, the use of microtubule inhibitors together with immune therapy may offer a highly effective option for selected cancer treatments [123].
In this study, a series of ninety-six compounds based on the 2-azetidinone scaffold structure were designed and synthesised as CBSIs. These novel 3-(prop-1-en-2-yl)azetidin-2-one, 3-allylazetidin-2-one and 3-(buta-1,3-dien-1-yl)azetidin-2-one analogues of CA-4 were prepared using an efficient Staudinger procedure, and they allowed introduction of the alkene, allyl and diene substituents at the C-3 position of the β-lactam, together with structurally varied electron-releasing or electron-withdrawing substituent groups on the ring B pharmacophore. These compounds, together with some amino acid and phosphate prodrugs and their Diels-Alder adducts, were evaluated for their antiproliferative activity, cell cycle effects and inhibition of tubulin assembly.
The compounds were shown to have significant in vitro antiproliferative activities in MCF-7 breast cancer cells, particularly compounds 9h, 9q, 9r, 10p 10r, 11h, with IC 50 values in the range 10-33 nM. The phenolic 3-(prop-1-en-2-yl) β-lactam compound 9q was identified as the most potent in MCF-7 breast cancer cells (IC 50 = 10 nM, with drug-like properties. Significant antiproliferative effects were also demonstrated for the compounds in the TNBC cell line MDA-MB-231, with IC 50 values in the range 23-33 µM, which were comparable with the activity of CA-4. The biological target of these compounds was identified as tubulin. They demonstrated significant reduction in tubulin polymerisation in vitro and were shown to interact non-covalently at the colchicine-binding site on tubulin. Cell cycle analysis through flow cytometry demonstrated that compound 9q arrested MCF-7 cells in the G2/M phase, resulting in cellular apoptosis.
Microtubule depolymerisation was confirmed by confocal microscopy, and the immunofluorescence results confirm the antimitotic properties of β-lactam 9q, which was observed to target tubulin and result in mitotic catastrophe. In addition, in silico molecular docking results indicate that the β-lactam compounds can interact with the colchicinebinding site of tubulin, further supporting the observed inhibitory effects of these compounds on tubulin polymerisation. Tubulin-targeted chemotherapy has been clinically successful against a wide spectrum of solid tumours and blood cancers. Compound 9q is representative of a novel class of potent microtubule-destabilising agents having low toxicity with promising small-molecule, drug-like properties and translational potential for the development of new molecules with potent tubulin-inhibitory activity.