In the Quest for Potent and Selective Malic Enzyme 3 Inhibitors for the Treatment of Pancreatic Ductal Adenocarcinoma

Abstract

The genome of pancreatic ductal adenocarcinoma (PDAC) is associated with frequent deletion of the tumor suppressor gene SMAD family member 4 (SMAD4) with collateral deletion of its chromosomal neighbor malic enzyme 2 (ME2). In SMAD4^–/–/ME2^–/– PDAC cells, ME3 takes over the function of the ME2 enzyme, and hence therapeutic targeting of ME3 is expected to arrest tumor growth. Hitherto no selective small molecule inhibitor of ME3 has been reported in the context of PDAC. Based on the molecular docking studies and structure–activity relationships with the reported ME1 inhibitor, several analogues of 6-piperazin-1-ylpyridin-3-ol amides have been synthesized and screened for their ME inhibition activity. Among them, compound 16b is identified as the most potent and selective ME3 inhibitor with an IC₅₀ of 0.15 μM on ME3, and with 15- and 9-fold selectivity over ME1 and ME2, respectively. In the cell viability assay, compound 16b exhibited an IC₅₀ of 3.5 μM on ME2-null PDAC cells, viz., BxPC-3.

Pancreatic ductal adenocarcinoma (PDAC) is an aggressive pancreatic cancer originating from exocrine cells and accounts for more than 90% of pancreatic tumors.¹ PDAC patients remain asymptomatic at early stages, and a majority of them are diagnosed only at unresectable stages.² Due to its poor prognosis, it is one of the leading causes of cancer-related mortalities with a five year survival rate of less than 11%.³ Chemotherapeutic combinations, viz., gemcitabine and paclitaxel or FOLFIRINOX (a combination of 5-fluorouracil, irinotecan, oxaliplatin, and folinic acid), are the current major standards of care for PDAC. However, there is an unmet clinical need for novel targeted therapies.⁴

Activating mutations in the Kirsten rat sarcoma virus (KRAS) oncogene is associated with more than 90% of PDAC cases along with subsequent deletion of tumor-suppressor genes like INK4A/ARF, tumor protein 53 (TP53), and SMAD family member 4 (SMAD4).⁵ In PDAC, oncogenic KRAS mutation (KRas^G12D) driven activation of downstream signaling pathways like MAPK and PI3K-mTOR mobilizes uncontrolled proliferation and survival of cancer cells. Rapid proliferation of neoplastic cells requires an adequate supply of energy and biosynthetic precursors as cellular building blocks.⁶ This enhanced energetic and anabolic requirement is satisfied through metabolically rewired processes which include aerobic glycolysis, glutaminolysis, and de novo fatty acid synthesis. Advancement of glycolysis and fatty acid synthesis is supported by bioprecursors like pyruvate and cofactors like nicotinamide adenine dinucleotide (NAD) phosphate NAD(P)⁺ and NAD(P)H.

Expression of enzymes involved in the mitochondrial tricarboxylic acid (TCA) cycle is also upregulated in numerous cancers leading to metabolic alterations to meet the enhanced demand for bioenergy and biomass for chronic proliferation of cancer cells. In this context, malic enzymes (MEs) play a crucial role in catalyzing oxidative decarboxylation of l-malate to form pyruvate and CO₂ while simultaneously reducing NAD(P)⁺ to NAD(P)H.⁷ Both pyruvate and NADPH have an important role in energy production in cells, for the maintenance of redox balance and for the production of building blocks for biosynthesis. In mammalian context, three isoforms of malic enzyme have been identified. These are classified based on their cofactor specificities and subcellular localization: cytosolic NADP⁺-dependent ME (ME1), mitochondrial NAD(P)⁺-dependent ME (ME2), and mitochondrial NADP⁺-dependent ME (ME3).⁸ These enzymes are essential for the maintenance of mitochondrial health and related energetics.

ME1 is a potential oncogene product linked to energy metabolism, redox status, and epithelial to mesenchymal transition (EMT) in squamous cell carcinoma (oral, head, and neck), breast cancer, and gastric cancer.⁹⁻¹¹ ME2 is overexpressed in human primary oral squamous cell carcinoma tissue and in lung carcinoma tissue for helping tumor cell proliferation and colony formation.^12,13 ME2 is also known to promote cell proliferation, invasion, and migration of glioma cells in glioblastoma.¹⁴ ME3, a paralogous isoform of ME2, is localized in mitochondria, and its aberrant expression contributes to the carcinogenesis of pancreatic cancer.¹⁵

In nearly one-third of PDAC cases, the tumor-suppressor gene SMAD4 is homozygously deleted, and neighboring gene ME2, which is in the same locus, is codeleted conferring “collateral lethality”. Through systematic in vitro and in vivo genetic experiments, Dey et al. have demonstrated that, in such a situation, the paralogous enzyme ME3 takes over the role of ME2, and depletion of this enzyme creates cancer specific metabolic vulnerability leading to cell death.¹⁶ Based on their compelling data, it was conceived that targeting ME3 with selective small molecule pharmacological inhibitors could be an attractive therapeutic strategy for specific PDAC patients having the SMAD4/ME2 deletions. Herein, we report the design and synthesis of small molecules that potently and selectively inhibit ME3.

In order to choose appropriate chemotype(s) for the development of novel ME3 inhibitors, the reported inhibitors A–E (Table 1) were synthesized and screened for their inhibitory potency against all three ME isoforms.¹⁷⁻²⁰ Among these, compound A showed superior inhibition on ME3 with an IC₅₀ of 0.15 μM. Hence, compound A was selected as a lead chemotype for the design and optimization of potent and selective ME3 inhibitors.

Table 1. ME Inhibition Data for Reference Molecules.

To understand and to analyze the possible binding interactions of compound A with the ME3 enzyme, docking studies needed to be carried out. Since the crystal structure of ME3 either in the apo form or with a substrate/inhibitor and cofactor (NADP⁺) was not available in the protein data bank (PDB), it was crystallized as an oxalate whose crystal structure was resolved (resolution: 1.8 Å). Using this crystal structure, the in silico binding mode of compound A and its analogues in the malate binding site of the ME3 enzyme was established by the induced fit docking (IFD) protocol of Schrödinger’s Maestro. In silico docking of compound A (Glide Score −10.03 kcal/mol) in the malate binding pocket of ME3 (Figure 1) revealed that the phenolic hydroxyl group in ring A undergoes H-bond donor–acceptor interactions with Asp304 and Asn489. Further, it was seen that ring A was a having cationic···π interaction with the Lys208 side chain; the pyrrolidine-2,5-dione ring had an H-bond interaction with Tyr107; and ring B had a π–π stacking interaction with Tyr107. Identical binding modes and interactions existed with either of the enantiomers (R and S) of compound A.

Figure 1.

In silico binding mode of compound A in the malate binding pocket of the ME3 enzyme.

Based on the observed binding mode of compound A, several new compounds were designed whereby the pyrrolidine-2,5-dione ring of compound A was replaced with a diverse acyclic linkers connecting the piperazine moiety and ring B (Scheme 1). These designed compounds, when docked in the established IFD grid accommodated compound A, indicating that the inhibitory activity on ME3 would be retained.

Scheme 1. Design of New Molecules as ME3 Inhibitors.

As per the conceived design, supported by their in silico predictions, the synthesis of several new compounds was undertaken. In the first instance, compounds 6a to 11a were synthesized (Scheme 2).

Scheme 2. General Scheme for the Synthesis of NCEs.

Reagents and conditions: (i) DMF, powdered K₂CO₃, benzyl bromide, 25 °C, 4 h, 78–90%; (ii) N-Boc piperazine, sodium tert-butoxide, tetrakis-(triphenylphosphine)palladium(0), S-Phos, toluene, 90 °C, 3 h, 60–76%; (iii) 1,4-dioxane, conc. HCl, 0–25 °C, 6 h, 90–95%; (iv) 5% Pd–C, H₂, THF-methanol, 25 °C, 6 h, 85–90%; (v) DMAc, EDC·HCl, R-COOH, 25 °C, 6 h, 50–70%.

The synthesis of these compounds involved the Buchwald–Hartwig amination (BHA) reaction, the palladium catalyzed cross-coupling reaction of aryl halides and primary or secondary amines yielding aryl amines, between 4-benzyloxy-1-bromobenzene/heteroarene 2a−2e and N-BOC piperazine to yield the intermediates 3a–3e, which were double deprotected to give the key intermediates 5a–5e. Compound 5a, on condensation with appropriately substituted carboxylic acids, yielded the compounds 8a to 11a. While the compounds 6a and 7a were prepared after derivatization of aniline with needed functionalities and coupling them with 4a followed by debenzylation.

All these compounds (6a–11a), when screened for in vitro activity on all the three ME isoforms, exhibited a promising inhibition (Table 2). Among them, compound 11a was the most active with an IC₅₀ of 0.08 μM, while the remaining compounds 6a to 10a were equipotent to compound A on the ME3 enzyme. The results showed that the compounds 6a to 11a are pan ME inhibitors.

Table 2. ME Isoforms Inhibition and BxPC-3 Cell Growth Inhibition Data for the Designed Compounds.

In the cell viability assay study on ME2-null BxPC-3 cells, the reference compound A did not show any activity up to 10 μM concentration. Compounds 8a and 10a, having an amide linker, showed a good cell growth inhibition as compared to compound 7a with the urea linker. Hence, further optimizations were carried out focusing on the compounds having the amide linkers.

An in silico binding mode study of compound 6a (Glide Score −7.78 kcal/mol) showed that, besides retaining all of the critical interactions observed for compound A, there was an additional H-bond interaction (N–H···O) with Gly203 (Figure 2a). Compound 8a (Glide Score −8.77 kcal/mol) also showed a similar binding mode; the essential interactions were retained with an additional π–π stacking of ring B with Tyr107 (Figure 2b).

Figure 2.

In silico binding mode of ligands in malate binding pocket of ME3 enzyme. (a) Compound 6a. (b) Compound 8a.

Moreover, it was observed that the ring B was directed toward the hydrophobic pocket hinting that the introduction of a lipophilic group at the para position of the phenyl ring B could improve enzyme potency, and by the virtue of increased lipophilicity, the cell permeability could as well get enhanced. Taking a clue from these in silico indications, compounds 12a to 16a were designed, synthesized, and characterized. Their synthesis involved coupling of the key intermediate 5a with appropriate carboxylic acids (Scheme 2).

These compounds were also subjected to in vitro activity studies on ME isoforms. The results of this study are summarized in Table 2. The compound 15a, having n-butyloxy substitution, showed an enhanced inhibition of the ME3 enzyme with an IC₅₀ of 0.06 μM. Incorporation of gem-dimethyl groups at the β-position to the amide linkage (compound 16a) had no additional advantage in terms of ME3 inhibition potency compared to the parent compound 10a having the unsubstituted spacer. However, the presence of the gem-dimethyl group in this compound could enhance the metabolic stability.

The in silico binding mode of compound A in the malate binding pocket of ME3 (Figure 1) showed that the phenolic hydroxyl group in ring A forms a hydrogen bond with Asp304 and Asn489. To investigate the importance of the phenolic hydroxyl group, the next set of the molecules (17 to 22) with the other polar functionalities was designed, synthesized, and characterized.

For the synthesis of compound 17, ethyl 4-bromobenzoate, N-BOC piperazine, and phenylacetic acid were employed. 17 on the reaction with hydroxylamine hydrochloride gave 18, while compound 19 was prepared by the reaction of (4-nitrophenyl)piperazine and phenylacetic acid, followed by reduction and methylsulfonation of the resulting aniline derivative. The reaction of phenylacetic acid with (3-hydroxyphenyl)piperazine yielded compound 20. The compounds 10b, 21, and 22 were prepared by employing appropriately substituted pyridine derivatives and a BHA reaction with the protected piperazine followed by a reaction with phenylpropanoic acid.

The compounds 17–22 thus synthesized were subjected to ME3 inhibition and BxPC-3 cell viability studies. From the in vitro screening data (Table 3), it was revealed that the phenolic hydroxyl group at position-4 was essential and critical for binding with ME3. Bioisosteric replacement of the same with a carboxylic group, hydroxamic group, or methylsulphonamide group (17, 18, or 19) resulted in a loss of the activity on ME3. The change in position of the phenolic hydroxyl group (i.e., compound 20) abolished the activity on the ME3. Replacement of the hydroxyl group with a methoxy or fluoro group (i.e., 21 or 22) also resulted in a complete loss of the activity on ME3. To understand these results, in silico study of these compounds was undertaken. It revealed that these compounds lacked either the required binding mode and consistency or the H-bonding with Asp304/Asn489 of the ME3 enzyme. Further, in the growth inhibition screen on BxPC-3 cells, compounds 21 and 22 exhibited a loss of activity, suggesting that the target engagement with ME3 is essential for the inhibitory activity.

Table 3. Influence of Replacement of Phenolic Hydroxyl on ME3 Inhibition.

Compound	R	X	R1	R2	% ME3 inhibition (1 μM)	% ME3 inhibition (10 μM)	% BxPC-3 cell growth inhibition (10 μM)	BxPC-3 IC50 (μM)
8a	PhCH2–	C	–OH	–H	88	100	80	6
17	PhCH2–	C	–COOH	–H	00	19
18	PhCH2–	C	–C(O)NHOH	–H	00	15
19	PhCH2–	C	–NHSO2Me	–H	04	32
20	PhCH2–	C	–H	–OH	06	00
10b	PhCH2CH2–	N	–OH	–H	97	100	92	5.10
21	PhCH2CH2–	N	–OMe	–H	00	27	13
22	PhCH2CH2–	N	–F	–H	10	39	00	>30

Since no specific binding interactions of the piperazine ring were observed in the in silico binding mode of compound 8a in the malate binding pocket of ME3 (Figure 2b), compounds 23–25 (Table 4) having variants for the piperazine ring were prepared and screened in vitro for ME3 inhibitory activity.

Table 4. Influence of Replacement of Piperazine Nitrogens on ME3 Inhibition.

The synthesis of 23 involved the coupling of phenylacetic acid with (4-piperidyl)phenol, while 24 was prepared from methyl piperidine-4-carboxylate, bezyloxybenzene (2a), and aniline as starting materials and using the BHA reaction for N-arylation as a key step. Similarly, the coupling of 2a, N-BOC piperazine, and phenyl acetic acid led to the synthesis of 25.

These compounds were then subjected to in vitro screening for ME3 inhibition activity, the results of which are summarized in Table 4. They indicated that the N-4 nitrogen of piperazine in compound 8a was critical for ME3 inhibition since its replacement with carbon (X = C, compound 23) resulted in a significant loss of the activity. However, replacement of the piperazine ring in compound 8a with either an isonipecotic moiety (Y = C, compound 24) or with 4-aminopiperidine moiety (Y = C, compound 25) did not have any impact on the activity on ME3 enzyme. A plausible explanation is that the most stable conformation in compound 8a is locked between the piperazine and ring A due to participation of the nitrogen lone pair electrons, while in the case of compound 23, rotation around C4 of the piperidine and ring A causes conformational changes leading to loss of the activity on the ME3 enzyme.

Selectivity for ME3 is crucial for preferential activity toward tumor cells over normal cells where ME2 is operative. With an aim to achieve the selectivity for ME3, the compounds with nitrogen containing heterocycles replacing the phenolic phenyl ring (ring A) (viz. 8b–8d, 10e) were designed and synthesized.

The synthesis of compounds 8b to 8d was achieved by employing appropriately positioned chloro and hydroxy substituents on pyridine and pyridazine heterocycles, while 10e was prepared using 2-chloro-5-hydroxypyrimidine employing the synthesis sequence described for 10a.

These newly synthesized compounds (8b–8d and 10e) were subjected to in vitro activity study on all the three isoforms of ME; the results of the same are summarized in Table 5.

Table 5. Influence of Heterocyclic Modification of Ring A on Selectivity for ME3^a.

Compound	R	X	Y	Z	ME3 IC50 (μM)	ME2 IC50 (μM)	ME1 IC50 (μM)
8a	PhCH2–	C	C	C	0.10	0.19	0.18
8b	PhCH2–	N	C	C	0.20	0.98	1.60
8c	PhCH2–	C	C	N	IA	IA	IA
8d	PhCH2–	N	C	N	IA	IA	IA
10a	PhCH2CH2–	C	C	C	0.10	0.27	0.27
10b	PhCH2CH2–	N	C	C	0.23	1.72	1.50
10e	PhCH2CH2–	N	N	C	IA	IA	IA

^aNote: IA = Inactive.

The results of this study revealed that the replacement of the 4-hydroxyphenyl ring in compound 8a with either 2-hydroxy-5-pyridyl or 3-hydroxy-6-pyridazinyl moieties (compounds 8c, 8d) resulted in a complete loss of the activity. Interestingly, replacement with 3-hydroxy-6-pyridyl ring (compound 8b) retained the potency on ME3 and exhibited nearly 5-fold and 8-fold selectivity over ME2 and ME1, respectively. Similarly, the 5-hydroxy-2-pyrimidinyl analogue (compound 10e) was found to be inactive on ME3 while the 3-hydroxy-6-pyridyl analogue (compound 10b) showed both potency and selectivity for ME3 over the other isoforms.

Following these observations, compound 10b was subjected to the docking study with the ME3 enzyme. The in silico binding mode of this compound (Glide Score −9.51 kcal/mol) showed that ring B was aligned toward the hydrophobic pocket of ME3 (Figure 3).

Figure 3.

In silico binding mode of compound 10b in the malate binding pocket of the ME3 enzyme.

Prior to the optimization of compound 10b, it was essential to study its mode of inhibition and target engagement with ME3. Keeping in view that ME3 has two substrates, i.e., l-malate and NADP⁺, the mode of action study was performed. The results indicated that compound 10b was a competitive inhibitor referring to l-malate and was an uncompetitive inhibitor with respect to NADP⁺. Details of the experimental procedures and analytical data are included in the Supporting Information.

In order to ascertain the target (ME3) engagement by compound 10b, a cell based thermal shift assay (CETSA) was performed using the BxPC-3 cells (ME2^–/– pancreatic cancer cell line). It was observed that compound 10b improved the thermal stability of ME3 in a dose dependent manner as a consequence of binding to ME3. A T_m shift of nearly 5 °C was observed with live cells in the presence of compound 10b (Figure 4). A similar experiment was carried out using the BxPC-3 cell lysate where a T_m shift of nearly 7 °C was detected with 10b. Details of the experimental procedures and analytical data are provided in the Supporting Information.

Figure 4.

CETSA–melt curve analysis for compound 10b.

With an intent to improve the potency of compound 10b, various analogues (15b, 16b, 26b–34b) with modifications at ring B were designed, prepared, and characterized. Their synthesis involved coupling of the key intermediate 5b with appropriately substituted carboxylic acids.

These newly designed compounds were screened for their ME inhibitory activity (Table 6). The results indicated that butyloxy-, fluoro-, methoxy-, and 2-methoxyethoxy- substitutions at the para position of ring B (cf. compounds 15b, 27b, 28b, and 29b) did not have any significant advantage, while chloro substitution (compound 26b) exhibited an enhanced potency on ME3 compared to the lead compound 10b. Incorporating gem-dimethyl groups at the benzylic position of ring B (compound 16b) exhibited greater potency and selectivity for ME3; however, further substitutions at position 4 of ring B with fluorine or chlorine (compounds 30b and 31b) resulted in decreased potency.

Table 6. ME Isoforms Inhibition and BxPC-3 Cell Growth Inhibition Data for Designed Compounds.

Bioisosteric replacement of phenyl ring B with a thiophene ring (compound 32b) improved the potency on ME3, while replacement with benzothiophene or 5-phenylthiophene (compounds 33b and 34b) retained the activity on ME3, similar to that of compound 10b.

From the detailed investigation of the structure–activity relationship (SAR), the structural features conducive to the potency and selectivity for ME3 are summarized in Figure 5.

Figure 5.

SAR summary for ME3 enzyme inhibition.

With an aim to evaluate inhibitory effects of selective ME3 inhibitors on SMAD4^–/–/ME2^–/– PDAC cell growth as a proof of the concept, selected NCEs showing promising and selective inhibition of ME3 were screened for growth inhibition on the BxPC-3 cells and the results are included in Table 6.

Among the total of 40 new chemical entities studied, compound 10b exhibited BxPC-3 cell growth inhibition with an IC₅₀ of 5.10 μM. As was predicted, incorporating a gem-dimethyl group (compound 16b) resulted in a better inhibitory potency with an IC₅₀ of 3.50 μM attributed to an improved log P of 3.41. The thiophene analogue (compound 32b) also exhibited BxPC-3 cell growth inhibition with an IC₅₀ of 4.60 μM. These results demonstrate that inhibition of ME3 creates cancer specific metabolic vulnerability leading to BxPC-3 cell growth attenuation. These compounds are currently under evaluation on the other SMAD4^–/–/ME2^–/– PDAC cell lines and will be followed by translation in in vivo xenograft models.

The structural feature present in the ME3 active NCEs has a 1,4 relationship between the phenolic hydroxyl group and the tertiary nitrogen substitution on the aryl ring (ring A). Are they cytotoxic, as they may get converted to quinones? To look into this possible alternative mechanism, a series of the truncated compounds (35b–38b, 5b) was prepared and subjected to in vitro ME3 enzyme inhibition activity as well as a BxPC-3 cell growth inhibition study (Table 7). The results clearly ruled out the quinonoid hypothesis, as these compounds failed to produce cytotoxicity on BxPC-3 cells despite having an equal probability of their conversion to quinones.

Table 7. ME Isoforms Inhibition and BxPC-3 Cell Growth Inhibition Data for Designed Compounds.

It was observed in the docking study that the para-hydroxyl group undergoes H-bond interactions with Asp304 and Asn489 of ME3. The CESTA study, as described earlier, also supported binding of the compounds with the enzyme with the retention of the hydroxyl functionality.

Moreover, selected compounds showing significant inhibition of BxPC-3 cells were screened in the nononcogenic HCE-T (Human corneal epithelial cell- transformed) cells (Table 2, Table 6). They were found to be at least 5–6-fold more selective toward BxPC-3 over HCE-T cells. Similarly, compounds 10b and 16a were also found to be selective toward BxPC-3 over HUVEC (Human Umbilical Vein Endothelial Cells) cells (Table S2 in the Supporting Information).

In summary, we have identified potent and cell-active ME3 inhibitors that show selectivity over the other ME isoforms for the first time. Moreover, robust target engagement of ME3 by one of the lead compounds (10b) was also demonstrated using CETSA. This observed selectivity (9–15 fold) has a scope of further advancement for any practical application in the treatment of the cancerous cell growth.

This study provides an insight into the structural elements that are critical for ME3 inhibition. It may enable the design of new inhibitors with enhanced potency and greater selectivity.

Experimental Section

The general synthesis route followed for all of the compounds is outlined in Scheme 2. Details of the experimental procedures and analytical data are provided in the Supporting Information.

Safety statement: No unexpected or unusually high safety hazards were encountered during the present work.

Glossary

Abbreviations

ME3

malic enzyme 3

ME2

malic enzyme 2

ME1

malic enzyme 1

PDAC

pancreatic ductal adenocarcinoma

NADP

nicotinamide adenine dinucleotide phosphate

NADH

nicotinamide adenine dinucleotide

KRAS

Kirsten rat sarcoma virus

TP53

tumor protein 53

SMAD4

SMAD family member 4

TCA

tricarboxylic acid

BHA

Buchwald–Hartwig amination

CETSA

cellular thermal shift assay

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.2c00369.

• Conditions and additional results of the in vitro biological assays, synthesis procedures, and analytical data for all of the new compounds including the NMR spectra for 10b, 16b, 32b, 10a, 11a, and 15a (PDF)

Author Contributions

The scientific work and the manuscript preparation involved contributions from all of the authors. All of the authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.

Dedication

Dedicated to Professor Sukh Dev on his 99th birthday.