Halting Tumor Progression via Novel Non-Hydroxamate Triazole-Based Mannich Bases MMP-2/9 Inhibitors; Design, Microwave-Assisted Synthesis, and Biological Evaluation

Matrix metalloproteinases (MMPs) are key signaling modulators in the tumor microenvironment. Among MMPs, MMP-2 and MMP-9 are receiving renewed interest as validated druggable targets for halting different tumor progression events. Over the last decades, a diverse range of MMP-2/9 inhibitors has been identified starting from the early hydroxamic acid-based peptidomimetics to the next generation non-hydroxamates. Herein, focused 1,2,4-triazole-1,2,3-triazole molecular hybrids with varying lengths and decorations, mimicking the thematic features of non-hydroxamate inhibitors, were designed and synthesized using efficient protocols and were alkylated with pharmacophoric amines to develop new Mannich bases. After full spectroscopic characterization the newly synthesized triazoles tethering Mannich bases were subjected to safety assessment via MTT assay against normal human fibroblasts, then evaluated for their potential anticancer activities against colon (Caco-2) and breast (MDA-MB 231) cancers. The relatively lengthy bis-Mannich bases 15 and 16 were safer and more potent than 5-fluorouracil with sub-micromolar IC50 and promising selectivity to the screened cancer cell lines rather than normal cells. Both compounds upregulated p53 (2–5.6-fold) and suppressed cyclin D expression (0.8–0.2-fold) in the studied cancers, and thus, induced apoptosis. 15 was superior to 16 in terms of cytotoxic activities, p53 induction, and cyclin D suppression. Mechanistically, both were efficient MMP-2/9 inhibitors with comparable potencies to the reference prototype hydroxamate-based MMP inhibitor NNGH at their anticancer IC50 concentrations. 15 (IC50 = 0.143 µM) was 4-fold more potent than NNGH against MMP-9 with promising selectivity (3.27-fold) over MMP-2, whereas 16 was comparable to NNGH. Concerning MMP-2, 16 (IC50 = 0.376 µM) was 1.2-fold more active than 15. Docking simulations predicted their possible binding modes and highlighted the possible structural determinants of MMP-2/9 inhibitory activities. Computational prediction of their physicochemical properties, ADMET, and drug-likeness metrics revealed acceptable drug-like criteria.


Introduction
Medicinal chemistry research is always focusing on modulating the tumor microenvironment, particularly its extracellular matrix, which innately confers cancer growth and metastasis [1,2]. Various crucial cancer progression signaling pathways are dependent on a plethora of proteases released into the tumor extracellular matrix [3]. Matrix metalloproteinases (MMPs) are among the most studied proteases given the fact that most of them have been found dysregulated in nearly all human malignancies [4][5][6]. Twenty-six MMPs have been characterized and correlated as a family of zinc-dependent endopeptidase [7]. The MMPs family was subclassed as collagenases; MMP-1, -8, -13, -18, gelatinases; MMP-2, -9, stromelysins; MMP-3, -10, matrilysins; MMP-7, -26, membrane-type MMPs; MMP-14, -15, -16, -17, -24, -25, and others [8]. All the MMPs family members share nearly the same catalytic domain structure with three a-helixes and five ß-sheets. The domain is characterized by critical active-site zinc coordinated by three histidine residues and five calcium ions. It is divided into C-terminal and N-terminal subdomains by a shallow cleft comprising six binding pockets (S1-S3, and S1 -S3 ). S1 subsite is considered the domain's selectivity pocket due to its amino acid sequence variations among MMPs [7,9]. A large body of evidence elects MMPs as attractive anticancer targets, being promotors of extracellular matrix turnover, tumor growth, and metastasis [10][11][12]. Their expression levels are tied to the tumor stage and the patient's prognosis [5]. Among MMPs, gelatinases (MMP-2 and -9) have become the focus of many anticancer research programs being validated as druggable targets for halting cancer progression at different stages [13][14][15][16]. With that, numerous MMPs inhibitors have been introduced over the last decades [14][15][16][17]. Early inhibitors were broad-spectrum mimics of the enzyme's endogenous ligands capped with the prototypic zinc-binding group hydroxamic acid [18]. Despite their outstanding anticancer potency [18], these peptidomimetics failed in the clinic due to hydroxamic acid-related pharmacokinetics challenges [19] and side effects [13,20,21]. The doubts about hydroxamates suitability directed research strategies to diversify the zinc-binding groups [22]. Although some inhibitors reached clinical trials, they eventually failed to achieve complete success [23]. Consequently, a fundamentally different design approach, utilizing information about the crystal structures of MMPs and computer-aided drug design, adopted introducing non-zinc binding inhibitors for avoiding previous failures. The new molecules can bind to the enzymatic active site while lacking the typical zinc chelation mode [24,25]. Non-zinc binding MMP inhibitors share certain thematic features, such as being long molecules with planar or aromatic linkers. While these architectures tend to be hydrophobic, some hydrophilic interactions are favored by incorporating amino and carbonyl groups [24,26,27]. Accordingly, various lead gelatinases inhibitors were introduced [14][15][16]28]. In the current study, we tailored triazole-based scaffolds mimicking the general theme of MMPs inhibitors while sparing the hydroxamate moiety with its associated drawbacks. It is worth mentioning that over the last few decades, triazoles have received a significant amount of attention [29]. This is because this unique scaffold possesses superior chemical and biological properties. The rational design of such heterocycles in drug discovery purposes has become a prominent area of research, defining medicinal chemistry [30,31]. Hence, we presume to develop new hybrid molecules combining 1,2,4-triazole and 1,2,3-triazoles and incorporating Mannich bases, employing molecular hybridization and click chemistry concepts under both conventional and microwave methods, as part of our ongoing research into the design, synthesis, and biological investigation of such hybrid molecules [32][33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48].

Design Rationale
The design protocol utilized the isomeric triazoles as privileged motifs of many reported lead MMP-2/9 inhibitors [49][50][51][52][53][54][55] (Figure 1). Besides its unique chemical and biological properties, 1,2,3-triazole was rationalized as a non-classical amide isostere [56]. Thus, the designed scaffold comprised three linked aromatic rings with possible hydrophilic interactions. It was also amenable to diversifying its electronic, steric nature via incorporating various aromatic and aliphatic substituents. Importantly, the study was extended to probe the effect of increasing the molecule length on activity through the synthesis of dimeric derivatives ( Figure 1). They were preliminarily screened for their potential anticancer activities against breast (MDA-MB 231) and colon (Caco-2) cancers, following an investigation of their safety profiles on normal human lung fibroblasts (Wi-38) via an MTT assay as reported [57][58][59]. The promising derivatives were evaluated for their in vitro MMP-2/9 inhibitory potential. Docking simulations were conducted to predict the structural determinants contributing to receptors interactions. Finally, their physicochemical parameters, ADMET, and drug-likeness profiles were computationally predicted.
Hydrazinolysis of ester 3 with hydrazine hydrate under MW irradiation for 2 min gave the acid hydrazide 4 in excellent yield (97%). The thermal hydrazinolysis required heating under reflux for 4 h to afford the same acid hydrazide 4 in 89% yield. The acid thiosemicarbazides 5 and 6 were obtained in 95-96% yield through the microwave-assisted condensation of the synthesized acid hydrazide 4 with phenyl and/or ethyl isothiocyanates for 3 min. Under conventional heating for 4 h, compounds 5 and 6 were obtained in 88 and 86%, respectively. Furthermore, intramolecular dehydrative cyclization of the resulted acid thiosemicarbazides 5 and 6 was carried out in basic media (10% NaOH), under reflux for four hours, to get the desired 1,2,4-triazoles 7 and 8 in good yields (88-86%). Microwave-assisted intramolecular cyclization required only 4 min to afford the targeted 1,2,3-triazole-1,2,4-triazole molecular hybrids 7 and 8 in excellent yields (96-97%) ( Table  1).    The structures of 7 and 8 were elucidated from their NMR spectra, where the 1 H-NMR spectra showed the presence of a distinct singlet at δ H 14.11-13.93 ppm attributed to the 1,2,4-triazolyl-NH proton. The 13 C-NMR spectra showed the presence of a characteristic downfield signal at δ C 169.15-167.60 ppm corresponding to the C=S group; these data confirmed the presence of the 1,2,4-triazoles 7 and 8 in their thione form rather than the thiol analogs. Moreover, the methylene protons between the two triazole rings were resonated as a singlet peak at δ H 5.93-5.66 ppm, which are correlated to its carbon at δ C 44.99-44.55 ppm in the 13 C-NMR spectra. Additionally, H-5 of the 1,2,3-triazolyl ring was observed as a singlet peak at δ H 8.71-8.25 ppm correlated to C-5 at 122.68-122.55 ppm.
The structure elucidation of the synthesized Mannich bases 9-14 was established based on their 1 H-and 13 C-NMR spectra. All their 13 C-NMR spectra revealed the presence of a characteristic signal corresponding to the C=S at δ C 170.10-168.37 ppm that distinguished the thione form of the synthesized compounds.
Moreover, the investigation of their 1 H-NMR spectra supported the formation of such bases by the disappearance of the 1,2,4-triazolyl-NH protons of 9-14 compared to their starting material 7 and 8, and the appearance of new singlet around δ H 5.11-4.98 ppm belonging to the N-CH 2 -N, which were resonated at δ C 70.40-68.90 ppm in their 13 C-NMR spectra, confirming the success of the aminomethylation reaction. The remaining protons and carbons were recorded in their expected area; the methylene (-CH 2 ) linking the two triazolic rings were resonated as a singlet at δ H 6.00-5.71 ppm, correlated with the signal observed at δ C 44.82-44. 35. In addition, H-5 of the 1,2,3-triazolyl proton was observed as a singlet between at δ H 8.76-8.21 ppm in a correlation to a signal at δ C 122.75-122.50 ppm in their 13 C-NMR spectra, respectively. Other protons and carbons corresponding to the phenyl ring or ethyl group in addition to the secondary amine (piperidine, morpholine, and N-methyl piperazine) moiety were investigated in the experimental section.
Under the same optimized conditions, the reaction of two equivalents of triazole 7 and/or 8 with one equivalent of piperazine and formaldehyde in refluxing ethanol for 20 h afforded the respective bis-aminomethylated Mannich bases 15 and 16 in 81-82% yield (Scheme 3). Under microwave irradiation, only 6 min were required to yield the same Mannich bases 15 and 16 in 92% yield. The NMR spectra of the bis aminomethylated Mannich base 15 showed the presence of an extra signal for some protons and carbons, which may be due to the unsymmetrical conformation of the structure with regard to its minimized energy. Thus, the investigation of its spectral data revealed the presence of two distinguishable singlets resonating at δ H 5.71 and 5.68 ppm assigned to the two CH 2 linker groups between the 1,2,3-and 1,2,4-triazole moieties and correlated with the signal resonated at δ C 42.91 ppm in the 13 C-NMR spectrum. In addition, a characteristic singlet was observed at δ H 4.92 ppm, integrating with four protons assigned to the two N-CH 2 -N groups and correlated with the signal appeared at δ C 53.59 ppm. Moreover, two singlets were recorded at δ H 8. 46 and 8.28 ppm attributed to the H-5 protons of the two 1,2,3-triazole rings and associated to signals at δ C 123.01 and 122.61 ppm. The presence of two signals at δ C 169.67 and 169.11 ppm proved the presence of two thione carbons (C=S) supporting the N-aminomethylation rather than S-aminomethylation. On the other hand, the NMR spectra of compound 16 showed equivalent signals, confirming its symmetric structure, where a singlet peak integrating two protons was observed at δ H 8.73 ppm, corresponding to H-5 of the two 1,2,3-triazole moieties and correlated to the signal resonating at δ C 122.71 ppm. The linker methylene protons (CH 2 ) which combine the 1,2,3-triazole and the 1,2,4-triazole rings were resonated at δ H 5.97 ppm as a singlet peak integrating with four protons and correlated with signal resonated at δ H 44.35 ppm in its 13 C-NMR spectrum. The N-CH 2 -N protons were observed at δ H 4.98 ppm as a singlet and correlated with that appeared at δ C 68.93 ppm. The signal recorded at δ C 168.40 ppm was attributed to two equivalent thione (C=S) carbons, confirming the symmetrical structure of 16. The remaining protons and carbons were investigated from the NMR spectra and represented in the experimental section. It should be noted that the unsymmetric conformation of the bis-Mannich base 15 may be due to the presence of the N-phenyl groups that affect the stability of the structure and, consequently, its conformation due to their steric hindrance rather than the ethyl groups in compound 16.

Cytotoxicity Screening
All the newly synthesized derivatives were firstly screened for cytotoxic activities on normal lung fibroblasts (Wi-38) for evaluating their safety profiles. Then, they were subjected to anticancer revaluation against two selected human tumors (MDA-MB 231 and Caco-2) compared 5-fluorouracil as reference chemotherapy via MTT assay ( Table 2). All the studied compounds were more active than the reference against MDA-MB 231, with 15 and 16 at the top of the list recording submicromolar IC 50 , followed by 13, 12, 10, 9, 14, and 11, respectively. On the other hand, only 15 (IC 50 ; 0.26 µM) and 16 (IC 50 ; 0.39 µM) were superior to 5-fluorouracil against Caco-2 cells. Moreover, 13 (IC 50; 0.61 µM) was nearly comparable to the reference. The remainder compounds were relatively moderate with IC 50 , ranging from 1.63-3.31 µM. The assessment of the compounds' selectivity towards tumor cells rather than the normal ones is key to real evaluation. Herein, the selectivity index (SI) was calculated for each compound as the ratio of its IC 50 against normal and cancer cells ( Table 2). It is generally accepted that SI ≥3 acknowledges considerable selectivity of the compound to cancer cells [64]. Interestingly, the most active derivative among the group 15 recorded the highest SI were against Caco-2 (SI; 4.13) and MDA-MB 231 (SI; 3.56) cells, exceeding the acceptable SI limit against both cancer cell lines. Accordingly, the two compounds 15 and 16 were selected for further mechanistic studies, whereas other derivatives were considered beyond the acceptable selectivity.  Figure 2). The treated cancer cells lost their characteristic shape. Severe shrinkage was observed reflecting the potent anticancer potential of the studied compounds [65].

Apoptosis Induction Evaluation
The apoptotic induction efficacy of 15 and 16 was evaluated by quantifying the elevation in the tumor suppressor gene p53 expression and suppression in oncogene (cyclin D)-mediated cell cycle progression via quantitative real-time PCR analysis.

Quantitative Real-Time PCR Analysis of p53 Gene
The tumor suppressor p53 gene is one of the main apoptosis induction markers. It is mutated in 50% of human cancers, including breast and colon. Its upregulation leads to induction of cell cycle arrest and apoptosis [66,67]. Although MDA MB-231 cells carry a mutated p53 gene (R280K mutation), mechanistic studies demonstrated that treatment with anticancer agents inducing p53-dependent intrinsic apoptosis (6-gingerol) resulted in p53 expression and quick apoptosis execution, supporting the hypothesis that p53 in MDA MB-231 is indeed a gain-of-function mutant [68][69][70]. Similarly, the expression of apoptosis regulatory genes evaluated by PCR following treatment of Caco-2 cells with some apoptotic inducers (Drimia calcarata bulb extracts) revealed p53 upregulation, even though p53 was not detected in the untreated cells [71].

Apoptosis Induction Evaluation
The apoptotic induction efficacy of 15 and 16 was evaluated by quantifying the elevation in the tumor suppressor gene p53 expression and suppression in oncogene (cyclin D)-mediated cell cycle progression via quantitative real-time PCR analysis.

Quantitative Real-Time PCR Analysis of p53 Gene
The tumor suppressor p53 gene is one of the main apoptosis induction markers. It is mutated in 50% of human cancers, including breast and colon. Its upregulation leads to induction of cell cycle arrest and apoptosis [66,67]. Although MDA MB-231 cells carry a mutated p53 gene (R280K mutation), mechanistic studies demonstrated that treatment with anticancer agents inducing p53-dependent intrinsic apoptosis (6-gingerol) resulted in p53 expression and quick apoptosis execution, supporting the hypothesis that p53 in MDA MB-231 is indeed a gain-of-function mutant [68][69][70]. Similarly, the expression of apoptosis regulatory genes evaluated by PCR following treatment of Caco-2 cells with some apoptotic inducers (Drimia calcarata bulb extracts) revealed p53 upregulation, even though p53 was not detected in the untreated cells [71].
with anticancer agents inducing p53-dependent intrinsic apoptosis (6-gingerol) resulted in p53 expression and quick apoptosis execution, supporting the hypothesis that p53 in MDA MB-231 is indeed a gain-of-function mutant [68][69][70]. Similarly, the expression of apoptosis regulatory genes evaluated by PCR following treatment of Caco-2 cells with some apoptotic inducers (Drimia calcarata bulb extracts) revealed p53 upregulation, even though p53 was not detected in the untreated cells [71].

MMP-2/9 Inhibition
The inhibitory potential of the hit anticancer derivatives 15 and 16 against MMP-2 and -9 was evaluated in terms of the percentage of inhibition at 0.4 µM (≈anticancer IC50; Table 2) in comparison to the reference prototypic inhibitor N-Isobutyl-N-(4-methoxyphenylsulfonyl)glycyl hydroxamic acid (NNGH) (Figure 1) through a colorimetric assay in a 96-well microplate format using a chromogenic substrate (Ac-PLG-[2-mercapto-4-me-  Table 2) in comparison to the reference prototypic inhibitor N-Isobutyl-N-(4-methoxyphenylsulfonyl)glycyl hydroxamic acid (NNGH) (Figure 1) through a colorimetric assay in a 96-well microplate format using a chromogenic substrate (Ac-PLG-[2-mercapto-4-methyl-pentanoyl]-LG-OC 2 H 5 ) ( Table 3). An MMP-2 or MMP-9 cleavage site bond is replaced by a thioester in the included thiopeptide, which affords sulfhydryl group upon hydrolysis. The sulfhydryl moiety reacts with Ellman's reagent to produce 2-nitro-5-thiobenzoic acid, which is detectable at 412 nm according to the kits manufacturers' protocol. Then, the same above-mentioned procedure was performed at serial concentrations of 15, 16, and the reference inhibitor NNGH to calculate their IC 50 values against MMP-2 and -9 (Table 3). Interestingly, 15 recorded higher MMP-9% inhibition than both 16 and the reference inhibitor NNGH at its anticancer IC 50 concentration. At similar concentrations, 16 was comparable to NNGH. As for MMP-2, an opposite inhibition pattern was detected, where 16 was superior to 15, and both were less active than NNGH. IC 50 determination results provided more clarity about the inhibition profiles, where 15 was approximately 4-fold more active than 16 and NNGH against MMP-9, while 16 was comparable to NNGH. On the other hand, 16 was slightly (1.2-fold) more active than 15 against MMP-2 inhibitor. In light of the mentioned results, 15 was 3.27-fold more active against MMP-9 than MMP-2, highlighting its promising selectivity towards MMP-9 compared to 16 and NNGH.

Docking Simulations
Docking was performed by MOE 2015.10 [73] on the promising compounds 15 and 16 to predict their major determinants towards MMP-2/9 binding sites and the structural features crucial for activity and selectivity. The crystal structures of the two gelatinases, MMP-2 (PDB code: 1HOV [74]) and MMP-9 (PDB code: 1GKC [75]), were retrieved from the protein data bank to derive the receptor models. Unnecessary residues and solvent molecules were eliminated, then the MOE "QuickPrep" module was utilized for structure preparation with the default settings. The studied compounds 15 and 16 were built in silico and energy minimized according to the default geometry optimization settings. Taking into account that the MMP active site may be biased to the hydroxamate-based cocrystallized ligand, the possible binding sites encompassing the key residues were located using the MOE 'Site Finder' feature. Docking of the studied compounds was firstly run with various fitting protocols to record the best docking scores and binding interactions. The best binding modes were computed by rigid docking employing the Triangle Matcher placement method and London dG scoring function. The selected protocol was validated by re-docking the co-crystallized ligand in the active site and reproducing most of the key experimental interactions at acceptable RMSD.
In MMP-2 docking (Figure 5), the computed scores (binding energies) of the top docking poses corresponding to the lowest energy conformers of 15 and 16 were approximately similar (∆G = −12.133 and −11.900 Kcal/moL, respectively). Both compounds resided well in the active site towards the S1 pocket forming π-π stacking between His120 and the triazole rings, H-π interactions between Tyr 142 and the phenyl rings substituents of the triazoles, as well as interactions between the active site zinc ion and the piperazinyl nitrogens. In 15, additional π-π stacking between His120 and the phenyl ring on the 1,2,4triazole ring and H-bonding between Glu121 and the triazole-piperazine linker methylene were detected.
with various fitting protocols to record the best docking scores and binding interactions. The best binding modes were computed by rigid docking employing the Triangle Matcher placement method and London dG scoring function. The selected protocol was validated by re-docking the co-crystallized ligand in the active site and reproducing most of the key experimental interactions at acceptable RMSD.
In MMP-2 docking (Figure 5), the computed scores (binding energies) of the top docking poses corresponding to the lowest energy conformers of 15 and 16 were approximately similar (ΔG = −12.133 and −11.900 Kcal/moL, respectively). Both compounds resided well in the active site towards the S1′ pocket forming π-π stacking between His120 and the triazole rings, H-π interactions between Tyr 142 and the phenyl rings substituents of the triazoles, as well as interactions between the active site zinc ion and the piperazinyl nitrogens. In 15, additional π-π stacking between His120 and the phenyl ring on the 1,2,4triazole ring and H-bonding between Glu121 and the triazole-piperazine linker methylene were detected. As for MMP-9 ( Figure 6), the computed scores of the top docking poses for 15 and 16 well resembled the experimental outcome so far, (ΔG = −10.102 and −9.522 Kcal/moL, respectively). Both compounds fitted into the MMP-9 catalytic domain and looped into the proximity of the S1′ pocket. They exhibited hydrogen bonding interactions between the 1,2,4-triazole thione, Leu188, and Ala189. H-π interactions were also detected between the 1,2,3-triazole phenyl and Arg424. Additional H-π interactions were predicted between the methylene, linking the isomeric triazoles of 16 and the active site His 401. As for MMP-9 (Figure 6), the computed scores of the top docking poses for 15 and 16 well resembled the experimental outcome so far, (∆G = −10.102 and −9.522 Kcal/moL, respectively). Both compounds fitted into the MMP-9 catalytic domain and looped into the proximity of the S1 pocket. They exhibited hydrogen bonding interactions between the 1,2,4-triazole thione, Leu188, and Ala189. H-π interactions were also detected between the 1,2,3-triazole phenyl and Arg424. Additional H-π interactions were predicted between the methylene, linking the isomeric triazoles of 16 and the active site His 401.
As for MMP-9 (Figure 6), the computed scores of the top docking poses for 15 and 16 well resembled the experimental outcome so far, (ΔG = −10.102 and −9.522 Kcal/moL, respectively). Both compounds fitted into the MMP-9 catalytic domain and looped into the proximity of the S1′ pocket. They exhibited hydrogen bonding interactions between the 1,2,4-triazole thione, Leu188, and Ala189. H-π interactions were also detected between the 1,2,3-triazole phenyl and Arg424. Additional H-π interactions were predicted between the methylene, linking the isomeric triazoles of 16 and the active site His 401.

ADMET and Drug-Likeness Prediction
In silico computation of ADMET and drug-likeness parameters has attracted considerable interest in the recent medicinal chemistry research programs as a useful tool for identifying lead compounds. Herein, various physicochemical properties formulating the drug-likeness parameters were predicted for our hit compounds 15 and 16 utilizing Swis-sADME [76], Pre-ADMET [77], and PROTOX [78] servers (Table 4). Both compounds showed acceptable predicted drug-like bioavailability according to the parameters used by Lipinski [79]. Two violations were detected according to the parameters used by Veber [80] and only one according to the parameters used by Egan [81]. Both compounds were predicted to possess high intestinal absorption (>98%), moderate CNS absorption, medium Caco2 model permeability, low MDCK model ones, high plasma proteins binding, and weak aqueous solubility. They lack cytochromes P450 2D6 (CYP2D6) inhibition activities but not CYP3A4 according to Pre-ADMET [77]. Moreover, the toxicity predictor software PROTOX [78]

ADMET and Drug-Likeness Prediction
In silico computation of ADMET and drug-likeness parameters has attracted considerable interest in the recent medicinal chemistry research programs as a useful tool for identifying lead compounds. Herein, various physicochemical properties formulating the drug-likeness parameters were predicted for our hit compounds 15 and 16 utilizing SwissADME [76], Pre-ADMET [77], and PROTOX [78] servers (Table 4). Both compounds showed acceptable predicted drug-like bioavailability according to the parameters used by Lipinski [79]. Two violations were detected according to the parameters used by Veber [80] and only one according to the parameters used by Egan [81]. Both compounds were predicted to possess high intestinal absorption (>98%), moderate CNS absorption, medium Caco2 model permeability, low MDCK model ones, high plasma proteins binding, and weak aqueous solubility. They lack cytochromes P450 2D6 (CYP2D6) inhibition activities but not CYP3A4 according to Pre-ADMET [77]. Moreover, the toxicity predictor software PROTOX [78] computed their acute oral toxicity in rodents in terms of their average lethal dose (LD 50 ) as 1000 mg/Kg; hence, both compounds were classified according to the Globally Harmonized System of Classification and Labeling of Chemicals (GHS) as class IV. Collectively, both 15 and 16 could be viewed as promising drug-like molecules.

Structure-Activity Relationship
The cytotoxicity pattern of the screened Mannich bases 9-16, reported in Table 2, suggests a key role of the molecule length and hydrophobicity in determining potency and selectivity. This pool of compounds showed promising anticancer activities. Notably, the bis Mannich bases 15 and 16 displayed the highest activity and selectivity compared to their monomeric precursors as well as the reference chemotherapy. The phenyl substituted bis Mannich base 15 was superior to the ethyl substituted derivative 16. In most cases, a similar observation was detected when comparing the cytotoxic activities of the phenyl substituted Mannich bases 9, 11, and 13, with the ethyl substituted analogs 10, 12, and 14, highlighting the preference of the electronic, hydrophobic, and steric effects of aromatic substituents to anticancer activity in these particular positions. Another driver of the anticancer potency was the heterocyclic amine at the 1,2,4-triazole motif, where N-methyl piperazine conferred, in most derivatives, superior activities especially against Caco-2 cells compared to morpholine and piperidine. Docking simulations of the most promising derivatives 15 and 16 into the MMP-2 and -9 active sites demonstrated that both compounds resided well, and highlighted the significance of the compound's length on activity ( Table 3). The 4-phenyl substituents on the 1,2,4-triazole motifs conferred higher MMP-9 selectivity to the bis Mannich base scaffold compared to the ethyl substituents. This may be attributed to the interactions with the active sites posed by the phenyl ring, as predicted by docking simulations.

Chemistry
All reactions were monitored by thin layer chromatography (TLC) on silica gel using 60 F254 aluminum sheets and were visualized under UV lamp at λ = 254 nm. The melting points were recorded and are uncorrected using a Stuart Scientific SMP1 device. The IR spectra were recorded on SHIMADZU FTIR-Affinity-1S spectrometer using KBr disks. The 1 H NMR (400 MHz) and 13 C NMR (100 MHz) spectra were recorded using a Bruker spectrometer (400 MHz) and TMS as an internal standard to calibrate the chemical shifts (δ) reported in ppm (See Supplementary Material Data). The microwave-assisted reactions were carried out in a programmable single-mode microwave reactor (CEM Discovery) that included a magnetic stirrer, pressure, temperature, and power controls, as well as a nitrogen cooling system. The reactor's maximum operational pressure was 2106 Pa, and its power was set to 300 W. A CHN elemental analyzer was used to perform the elemental analysis. Conventional Method (CM): To a solution of phenyl acetylene 1 (0.11 g, 1 mmoL) in t-BuOH (10 mL) was added a solution of copper sulfate (0.10 g, 0.4 mmoL) and sodium ascorbate (0.15 g, 0.75 mmoL) in water (10 mL). Then, ethyl-2-azidoacetate (2) (0.14 g, 1.1 mmoL) was added and the reaction mixture was stirred at room temperature for 6 h. TLC (hexane-ethyl acetate) was used to monitor the reaction, and after it was completed, iced water was added. The precipitate thus formed was collected by filtration, washed with a saturated solution of ammonium chloride, and recrystallized from ethanol to afford the targeted 1,2,3-triazole 3 with ester functionality.

Synthesis and
Microwave Method (MWI): A solution of ester 3 (0.12 g, 0.5 mmoL) in ethanol (10 mL) and hydrazine hydrate (0.05 g, 1 mmoL) was irradiated by MWI in a closed borosilicate glass vessel fitted with a silicone cap for 2 min. The reaction was conducted according to the conventional procedure stated earlier to produce the same 1,2,3-triazole 4.
Microwave Method (MWI): A solution of the acid hydrazide 4 (0.1 g, 0.5 mmoL) in ethanol (10 mL) and appropriate isothiocayanate derivative (0.6 mmoL) was irradiated by MWI in a closed borosilicate glass vessel fitted with a silicone cap for 3-4 min. The same acid thiosemicarbazide derivatives 5 and 6 were produced by handling the reaction mixture, as indicated in the conventional procedure. Conventional Method (CM): A solution of the acid thiosemicarbazide derivative 5 and/or 6 (1 mmoL) in aqueous sodium hydroxide 2 N (10 mL) was refluxed for 6 h. After cooling to room temperature, the solution was acidified with a diluted solution of hydrochloride acid to form the desired 1,2,4-triazoles 7 and/or 8 as precipitate and was filtered, washed with water, and recrystallized from ethanol.
Microwave Method (MWI): A solution of acid thiosemicarbazide derivative 5 and/or 6 (1 mmoL) in aqueous sodium hydroxide 2 N (10 mL) was irradiated by MWI in a closed borosilicate glass vessel fitted with a silicone cap for 3-8 min using a microwave reactor. The reaction mixture was handled according to the conventional approach to give the same 1,2,4-triazoles 7 and/or 8.

Synthesis and Characterization of Mannich Bases 9-16
Conventional Method (CM): The appropriate secondary amine (1 mmoL) was added to a solution of triazole derivative 7 and/or 8 (1 and/or 2 mmoL) in ethanol (20 mL) in the presence of formaldehyde (1 and/or 2 mmoL), and then the reaction mixture was refluxed for 16-20 h (monitored using TLC). After completing the reaction, the solution was evaporated under reduced pressure, and the solid that resulted was recrystallized from ethanol to get the corresponding Mannich bases 9-16.
Microwave Method (MWI): A mixture of triazole derivative 7 and/or 8 (1 mmoL), appropriate secondary amine (1 and/or 2 mmoL), and formaldehyde (1 and/or 2 mmoL) in ethanol (10 mL) were placed in a sealed borosilicate glass vessel with a silicone cap and subjected to microwave radiation for 5-6 min. The reaction was treated conventionally to provide the same Mannich bases 9-16.

MMP-2/9 Inhibition Assays
MMP-2/9 inhibitory activities were detected for the studied compounds according to the inhibitor screening assay colorimetric kit (Abcam, Catalog No. ab139446 for MMP-2 and ab139448 for MMP-9) instructions. Briefly, the studied compounds and the reference inhibitor (NNGH) at serial concentrations were added to assay buffer and MMP, then incubated for 60 min at 37 • C. After that, the chromogenic substrate was added, and the absorbance (∆A) was recorded per min for 10 min at 412 nm. The inhibition percentage of MMP-9 for each compound was calculated as follows: 100-([velocity (∆A/min) of compound/velocity (∆A/min) of control (no inhibitor added)] * 100). IC 50 values were computed using Graphpad Prism.

Molecular Modeling Docking Simulations
Docking simulations were conducted employing MOE 2015.10 [73]. Briefly, the coordinates of MMP-2 and -9 active sites were retrieved from the protein data bank. Unwanted molecules were deleted. The protein structures were prepared via the default "Quick-Prep" module settings. The studied derivatives were subjected to the MMFF94x forcefield and gradient: 0.05 for energy minimization after adding hydrogens and applying partial charges. The binding sites encompassing key amino acid residues were located via the 'Site Finder' feature. Rigid docking was simulated by MOE employing the Triangle Matcher placement method, Rescoring 1: London dG, Refinement: Forcefield, and Rescoring 2: Affinity dG. After docking, the conformations were visualized and selected based on the protein-ligand interactions and docking scores.

Data Analysis
Biological data were expressed as mean ± SEM. Statistical significance was estimated by the multiple comparisons Tukey post-hoc analysis of variance (ANOVA) employing SPSS16 and differences were considered significant at p < 0.05. IC 50 values were computed using Graphpad Prism.

Conclusions
The current study portrays the design, synthesis, and evaluation of non-hydroxamate MMP-2/9 inhibitors based on tethered isomeric triazoles Mannich bases with varying substituents and lengths. The rationale design relied on mimicking the non-hydroxamate inhibitors structural outlines. Among the synthesized series, the bis-Mannich bases 15 and 16 were superior to the reference chemotherapy 5-fluorouracil regarding safety, selectivity, and anticancer activity (submicromolar IC 50 ) against MDA-MB 231 and Caco-2 cells. In the studied cancers, 15 and 16 upregulated p53, where 15 exhibited higher p53 induction potential (up to 5.6-fold) than that recorded for 16 (up to 2.9-fold). Additionally, they suppressed cyclin D expression (down to 0.2-fold) in the treated cancer cell lines, therefore inducing apoptosis. Moreover, 15 was also superior to 16 in terms of cyclin D suppression leading to 0.23-and 0.38-fold relative decrease in cyclin D in the treated breast and colon cancer cells, respectively, compared to 0.49-and 0.87-fold, respectively, in the 16-treated cancer cells. Both compounds exhibited promising MMP-2/9 inhibitory activities, comparable to the reference MMP inhibitor NNGH, at their anticancer IC 50 doses. Moreover, 15 was 4-fold more active against MMP-9 than NNGH, with 3.27-fold selectivity over MMP-2. The less potent derivative 16 was comparable to NNGH. On the other hand, 16 was 1.2-fold more active than 15 against MMP-2. Docking simulations of 15 and 16 into MMP-2 and -9 catalytic domains predicted suitable fitting of the compounds into the active sites at the proximity of S1 pockets with considerable key interactions. In silico ADMET and drug-likeness prediction recorded acceptable metrics.