Development of Anthraquinone Analogues as Phosphoglycerate Mutase 1 Inhibitors

Phosphoglycerate mutase 1 (PGAM1) coordinates glycolysis and biosynthesis to promote cancer cell proliferation, and is believed to be a promising target for cancer therapy. Herein, based on the anthraquinone scaffold, we synthesized 31 anthraquinone derivatives and investigated the structure−activity relationship (SAR). The 3-substitient of sulfonamide on the anthraquinone scaffold was essential for maintaining potency and the modifications of the hydroxyl of alizarin would cause a sharp decrease in potency. In the meantime, we determined the co-crystal structure of PGAM1 and one of the anthraquinone inhibitors 9i with IC50 value of 0.27 μM. The co-crystal structure revealed that F22, K100 and R116 of PGAM1 were critical residues for the binding of inhibitors which further validated the SAR. Consistent with the crystal structure, a competitive assay illustrated that compound 9i was a noncompetitive inhibitor. In addition, compound 9i effectively restrained different lung cancer cells proliferation in vitro. Taken together, this work provides reliable guide for future development of PGAM1 inhibitors and compound 9i may act as a new leading compound for further optimization.


Introduction
In 1924, Warburg discovered that cancer cells tended to metabolize glucose through aerobic glycolysis rather than oxidative phosphorylation, even if there was sufficient oxygen [1,2]. This character of cancer cells was distinct from normal differentiated cells and the phenomenon was thereafter named the "Warburg effect". However, the reason why cancer cells adopt such an altered metabolism remained unclear. In the past decades, reprogramming energy metabolism was considered as an emerging hallmark of cancer cells [3] and the study of cancer metabolism has drawn great interest [4][5][6][7][8]. During aerobic glycolysis of cancer cells, adenosine 5 -triphosphate (ATP) is produced inefficiently in this metabolic pattern, whereas building blocks such as nucleotides, and amino acids are generated in large amounts, satisfying the demands of anabolic biosynthesis which are essential for rapid cell proliferation [9][10][11][12]. Nowadays, various enzymes playing critical roles in cancer metabolism have been reported, including isocitrate dehydrogenase [13][14][15]. Inhibitors targeting these enzymes have been approved by Food and Drug Administration [14,15] which further demonstrated the potential of modulating cancer metabolism for cancer therapy.
Phosphoglycerate mutase 1 (PGAM1) is a key enzyme in the 8th step of glycolysis pathway converting 3-phosphoglycerate (3PG) to 2-phosphoglycerate (2PG) [16]. Previously we found that PGAM1 could balance the intracellular concentrations of its substrate 3PG and its product 2PG to control tumor growth [17]. In detail, 3PG inhibits 6-phosphogluconate dehydrogenase in Since only a limited amount of anthraquinone inhibitors were reported, the structure−activity relationship (SAR) of anthraquinone derivatives and PGAM1 was not well studied yet. Besides, up to these days, the molecular interactions of PGAM1 and anthraquinone inhibitor were not observed in crystal structure. However, anthraquinones exhibit a variety of pharmacological activities including anticancer [33][34][35], antifungal [36], and antibacterial properties [37] and so on. What's more, there are many clinically approved anthraquinone drugs such as doxorubicin and mitoxantrone with strong potency in suppressing cancer cell proliferation [38]. Here we designed and synthesized a series of new anthraquinone compounds and evaluated their biological activity on PGAM1 and cancer cell proliferation. In addition, we solved the co-crystal structure of PGAM1 and compound 9i which interpreted the molecular mechanism of compound 9i interacting with PGAM1 and further confirmed the SAR we investigated. Combined with the crystal structure, additional competitive assay demonstrated the noncompetitive binding of compound 9i with the 3PG substrate. Since only a limited amount of anthraquinone inhibitors were reported, the structure−activity relationship (SAR) of anthraquinone derivatives and PGAM1 was not well studied yet. Besides, up to these days, the molecular interactions of PGAM1 and anthraquinone inhibitor were not observed in crystal structure. However, anthraquinones exhibit a variety of pharmacological activities including anticancer [33][34][35], antifungal [36], and antibacterial properties [37] and so on. What's more, there are many clinically approved anthraquinone drugs such as doxorubicin and mitoxantrone with strong potency in suppressing cancer cell proliferation [38]. Here we designed and synthesized a series of new anthraquinone compounds and evaluated their biological activity on PGAM1 and cancer cell proliferation. In addition, we solved the co-crystal structure of PGAM1 and compound 9i which interpreted the molecular mechanism of compound 9i interacting with PGAM1 and further confirmed the SAR we investigated. Combined with the crystal structure, additional competitive assay demonstrated the noncompetitive binding of compound 9i with the 3PG substrate.
As shown in Scheme 3, the 4-nitro group was reduced to an amino group after the nitrification of compound 6c, followed by sulfonylation of the amino group and hydrolysis of the ester as described above. Finally, compound 9i was employed as a starting material, and compounds 14a-g were obtained by substitution with different reagents and further hydrolysis of the ester or amidation of the carboxyl group (Scheme 4).

SAR Exploration of the Anthraquinone Derivatives Against PGAM1
Previously we reported several anthraquinone compounds as PGAM1 inhibitors [17]. Due to the limited amount of compounds and the lack of chemical diversity, the SAR of anthraquinone derivatives and PGAM1 remained poorly understood. Here we synthesized four classes of compounds on the base of bioisosterism to study the SAR of anthraquinone derivatives against PGAM1 in detail (Figure 2), using an optimized enzymatic assay in which PGMI-004A served as a control with an IC50 value of 1.2 ± 0.3 μM. Firstly, we found that different modifications of the hydroxyl of alizarin (6a-e) showed weak PGAM1 inhibition activity at 10 μM (Table 1). Finally, compound 9i was employed as a starting material, and compounds 14a-g were obtained by substitution with different reagents and further hydrolysis of the ester or amidation of the carboxyl group (Scheme 4). As shown in Scheme 3, the 4-nitro group was reduced to an amino group after the nitrification of compound 6c, followed by sulfonylation of the amino group and hydrolysis of the ester as described above. Finally, compound 9i was employed as a starting material, and compounds 14a-g were obtained by substitution with different reagents and further hydrolysis of the ester or amidation of the carboxyl group (Scheme 4).

SAR Exploration of the Anthraquinone Derivatives Against PGAM1
Previously we reported several anthraquinone compounds as PGAM1 inhibitors [17]. Due to the limited amount of compounds and the lack of chemical diversity, the SAR of anthraquinone derivatives and PGAM1 remained poorly understood. Here we synthesized four classes of compounds on the base of bioisosterism to study the SAR of anthraquinone derivatives against PGAM1 in detail (Figure 2), using an optimized enzymatic assay in which PGMI-004A served as a control with an IC50 value of 1.2 ± 0.3 μM. Firstly, we found that different modifications of the hydroxyl of alizarin (6a-e) showed weak PGAM1 inhibition activity at 10 μM (Table 1).

SAR Exploration of the Anthraquinone Derivatives against PGAM1
Previously we reported several anthraquinone compounds as PGAM1 inhibitors [17]. Due to the limited amount of compounds and the lack of chemical diversity, the SAR of anthraquinone derivatives and PGAM1 remained poorly understood. Here we synthesized four classes of compounds on the base of bioisosterism to study the SAR of anthraquinone derivatives against PGAM1 in detail ( Figure 2), using an optimized enzymatic assay in which PGMI-004A served as a control with an IC 50 value of 1.2 ± 0.3 µM. Firstly, we found that different modifications of the hydroxyl of alizarin (6a-e) showed weak PGAM1 inhibition activity at 10 µM (Table 1).
In order to explore the SAR of 3-substituents of the anthraquinone scaffold against PGAM1, compounds 9a-q were synthesized by reversing the sulfonamide group of PGMI-004A. Compounds 9a-e with different substituents on the phenyl ring displayed comparable inhibition activity towards PGAM1, with IC 50 values of 0.5 µM or so (Table 2). They were much more potent than the compounds without the sulfonamide group, suggesting the sulfonamide group was essential for maintaining potency. Then, we studied the impact of substituents of different halogens on the phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC 50 values below 0.5 µM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC 50 value of 0.27 µM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. derivatives and PGAM1 remained poorly understood. Here we synthesized four classes of compounds on the base of bioisosterism to study the SAR of anthraquinone derivatives against PGAM1 in detail (Figure 2), using an optimized enzymatic assay in which PGMI-004A served as a control with an IC50 value of 1.2 ± 0.3 μM. Firstly, we found that different modifications of the hydroxyl of alizarin (6a-e) showed weak PGAM1 inhibition activity at 10 μM (Table 1). the limited amount of compounds and the lack of chemical diversity, the SAR of anthraquinone derivatives and PGAM1 remained poorly understood. Here we synthesized four classes of compounds on the base of bioisosterism to study the SAR of anthraquinone derivatives against PGAM1 in detail (Figure 2), using an optimized enzymatic assay in which PGMI-004A served as a control with an IC50 value of 1.2 ± 0.3 μM. Firstly, we found that different modifications of the hydroxyl of alizarin (6a-e) showed weak PGAM1 inhibition activity at 10 μM (Table 1). the limited amount of compounds and the lack of chemical diversity, the SAR of anthr derivatives and PGAM1 remained poorly understood. Here we synthesized four c compounds on the base of bioisosterism to study the SAR of anthraquinone derivative PGAM1 in detail (Figure 2), using an optimized enzymatic assay in which PGMI-004A se control with an IC50 value of 1.2 ± 0.3 μM. Firstly, we found that different modificatio hydroxyl of alizarin (6a-e) showed weak PGAM1 inhibition activity at 10 μM (Table 1). In order to explore the SAR of 3-substituents of the anthraquinone scaffold against PGAM1, compounds 9a-q were synthesized by reversing the sulfonamide group of PGMI-004A. Compounds 9a-e with different substituents on the phenyl ring displayed comparable inhibition activity towards PGAM1, with IC50 values of 0.5 μM or so (Table 2). They were much more potent than the compounds without the sulfonamide group, suggesting the sulfonamide group was essential for maintaining potency. Then, we studied the impact of substituents of different halogens on the phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. In order to explore the SAR of 3-substituents of the anthraquinone scaffold against compounds 9a-q were synthesized by reversing the sulfonamide group of PGMI-004A. Co 9a-e with different substituents on the phenyl ring displayed comparable inhibition activity PGAM1, with IC50 values of 0.5 μM or so ( Table 2). They were much more potent compounds without the sulfonamide group, suggesting the sulfonamide group was ess maintaining potency. Then, we studied the impact of substituents of different halogen phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compou Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F su (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which discussed below.

SAR Exploration of the Anthraquinone Derivatives Against PGAM1
Previously we reported several anthraquinone compounds as PGAM1 inhibitors [17]. Due to the limited amount of compounds and the lack of chemical diversity, the SAR of anthraquinone derivatives and PGAM1 remained poorly understood. Here we synthesized four classes of compounds on the base of bioisosterism to study the SAR of anthraquinone derivatives against PGAM1 in detail (Figure 2), using an optimized enzymatic assay in which PGMI-004A served as a control with an IC50 value of 1.2 ± 0.3 μM. Firstly, we found that different modifications of the hydroxyl of alizarin (6a-e) showed weak PGAM1 inhibition activity at 10 μM (Table 1). In order to explore the SAR of 3-substituents of the anthraquinone scaffold against PGAM1, compounds 9a-q were synthesized by reversing the sulfonamide group of PGMI-004A. Compounds 9a-e with different substituents on the phenyl ring displayed comparable inhibition activity towards PGAM1, with IC50 values of 0.5 μM or so ( Table 2). They were much more potent than the compounds without the sulfonamide group, suggesting the sulfonamide group was essential for maintaining potency. Then, we studied the impact of substituents of different halogens on the phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. In order to explore the SAR of 3-substituents of the anthraquinone scaffold against PGAM1, compounds 9a-q were synthesized by reversing the sulfonamide group of PGMI-004A. Compounds 9a-e with different substituents on the phenyl ring displayed comparable inhibition activity towards PGAM1, with IC50 values of 0.5 μM or so ( Table 2). They were much more potent than the compounds without the sulfonamide group, suggesting the sulfonamide group was essential for maintaining potency. Then, we studied the impact of substituents of different halogens on the phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below.  In order to explore the SAR of 3-substituents of the anthraquinone scaffold against PGAM1, compounds 9a-q were synthesized by reversing the sulfonamide group of PGMI-004A. Compounds 9a-e with different substituents on the phenyl ring displayed comparable inhibition activity towards PGAM1, with IC50 values of 0.5 μM or so ( Table 2). They were much more potent than the compounds without the sulfonamide group, suggesting the sulfonamide group was essential for maintaining potency. Then, we studied the impact of substituents of different halogens on the phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. In order to explore the SAR of 3-substituents of the anthraquinone scaffold against PGAM1, compounds 9a-q were synthesized by reversing the sulfonamide group of PGMI-004A. Compounds 9a-e with different substituents on the phenyl ring displayed comparable inhibition activity towards PGAM1, with IC50 values of 0.5 μM or so ( Table 2). They were much more potent than the compounds without the sulfonamide group, suggesting the sulfonamide group was essential for maintaining potency. Then, we studied the impact of substituents of different halogens on the phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. In order to explore the SAR of 3-substituents of the anthraquinone scaffold against PGAM1, compounds 9a-q were synthesized by reversing the sulfonamide group of PGMI-004A. Compounds 9a-e with different substituents on the phenyl ring displayed comparable inhibition activity towards PGAM1, with IC50 values of 0.5 μM or so ( Table 2). They were much more potent than the compounds without the sulfonamide group, suggesting the sulfonamide group was essential for maintaining potency. Then, we studied the impact of substituents of different halogens on the phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. In order to explore the SAR of 3-substituents of the anthraquinone scaffold against PGAM1, compounds 9a-q were synthesized by reversing the sulfonamide group of PGMI-004A. Compounds 9a-e with different substituents on the phenyl ring displayed comparable inhibition activity towards PGAM1, with IC50 values of 0.5 μM or so ( Table 2). They were much more potent than the compounds without the sulfonamide group, suggesting the sulfonamide group was essential for maintaining potency. Then, we studied the impact of substituents of different halogens on the phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. In order to explore the SAR of 3-substituents of the anthraquinone scaffold against PGAM1, compounds 9a-q were synthesized by reversing the sulfonamide group of PGMI-004A. Compounds 9a-e with different substituents on the phenyl ring displayed comparable inhibition activity towards PGAM1, with IC50 values of 0.5 μM or so ( Table 2). They were much more potent than the compounds without the sulfonamide group, suggesting the sulfonamide group was essential for maintaining potency. Then, we studied the impact of substituents of different halogens on the phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. In order to explore the SAR of 3-substituents of the anthraquinone scaffold against PGAM1, compounds 9a-q were synthesized by reversing the sulfonamide group of PGMI-004A. Compounds 9a-e with different substituents on the phenyl ring displayed comparable inhibition activity towards PGAM1, with IC50 values of 0.5 μM or so ( Table 2). They were much more potent than the compounds without the sulfonamide group, suggesting the sulfonamide group was essential for maintaining potency. Then, we studied the impact of substituents of different halogens on the phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below.     We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, followed by nitrification, reduction, and sulfonylation. Probably due to the modification of the hydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, decreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good solubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized compounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition activity toward PGAM1 to different extents (Table 3).
In order to exclude the inhibition of the other three enzymes (enolase-pyruvate kinase-LDH) in the coupled assay, we performed counter screening of selected compounds with IC50 value below 1 μM on PGAM1. We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, followed by nitrification, reduction, and sulfonylation. Probably due to the modification of the hydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, decreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good solubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized compounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition activity toward PGAM1 to different extents (Table 3).
In order to exclude the inhibition of the other three enzymes (enolase-pyruvate kinase-LDH) in the coupled assay, we performed counter screening of selected compounds with IC50 value below 1 μM on PGAM1. We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, followed by nitrification, reduction, and sulfonylation. Probably due to the modification of the hydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, decreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good solubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized compounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition activity toward PGAM1 to different extents (Table 3).
In order to exclude the inhibition of the other three enzymes (enolase-pyruvate kinase-LDH) in the coupled assay, we performed counter screening of selected compounds with IC50 value below 1 μM on PGAM1. We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, followed by nitrification, reduction, and sulfonylation. Probably due to the modification of the hydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, decreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good solubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized compounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition activity toward PGAM1 to different extents (Table 3).
In order to exclude the inhibition of the other three enzymes (enolase-pyruvate kinase-LDH) in the coupled assay, we performed counter screening of selected compounds with IC50 value below 1 μM on PGAM1.  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, followed by nitrification, reduction, and sulfonylation. Probably due to the modification of the hydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, decreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good solubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized compounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition activity toward PGAM1 to different extents (Table 3).
In order to exclude the inhibition of the other three enzymes (enolase-pyruvate kinase-LDH) in the coupled assay, we performed counter screening of selected compounds with IC50 value below 1 μM on PGAM1.  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, followed by nitrification, reduction, and sulfonylation. Probably due to the modification of the hydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, decreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good solubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized compounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition activity toward PGAM1 to different extents (Table 3).
In order to exclude the inhibition of the other three enzymes (enolase-pyruvate kinase-LDH) in the coupled assay, we performed counter screening of selected compounds with IC50 value below 1 μM on PGAM1. We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, followed by nitrification, reduction, and sulfonylation. Probably due to the modification of the hydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, decreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good solubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized compounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition activity toward PGAM1 to different extents (Table 3).
In order to exclude the inhibition of the other three enzymes (enolase-pyruvate kinase-LDH) in the coupled assay, we performed counter screening of selected compounds with IC50 value below 1 μM on PGAM1. We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, followed by nitrification, reduction, and sulfonylation. Probably due to the modification of the hydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, decreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good solubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized compounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition activity toward PGAM1 to different extents (Table 3).
In order to exclude the inhibition of the other three enzymes (enolase-pyruvate kinase-LDH) in the coupled assay, we performed counter screening of selected compounds with IC50 value below 1 μM on PGAM1.  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, followed by nitrification, reduction, and sulfonylation. Probably due to the modification of the hydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, decreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good solubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized compounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition activity toward PGAM1 to different extents (Table 3).
In order to exclude the inhibition of the other three enzymes (enolase-pyruvate kinase-LDH) in the coupled assay, we performed counter screening of selected compounds with IC50 value below 1 μM on PGAM1. We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, followed by nitrification, reduction, and sulfonylation. Probably due to the modification of the hydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, decreasing 10% and 34% of PGAM1 activity at 10 µM, respectively. Finally, considering the good solubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized compounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition activity toward PGAM1 to different extents (Table 3).
In order to exclude the inhibition of the other three enzymes (enolase-pyruvate kinase-LDH) in the coupled assay, we performed counter screening of selected compounds with IC 50 value below 1 µM on PGAM1.
As shown in Table 4, the compounds displayed almost no inhibition on the three downstream enzymes at 0.5 µM which further confirmed the on-target effect of the compounds. As shown in Table 4, the compounds displayed almost no inhibition on the three downstream enzymes at 0.5 μM which further confirmed the on-target effect of the compounds.

Binding Mode of Compound 9i with PGAM1
To further understand the molecular mechanism of the anthraquinone derivatives interacting with PGAM1, we determined the X-ray structure of PGAM1 in complex with compound 9i at resolution of 1.98 Å (Table 5). Compound 9i occupied a novel allosteric site adjacent to substrate binding site with nice electron density ( Figure 3A and Figure 3B). The allosteric pocket was surrounded by the residues of F22, R90, K100, R116 and R191. In detail, the anthraquinone scaffold and sulfonamide of compound 9i interacted with the main chain carbonyl of K100 through water bridges ( Figure 3C). In addition, a hydrophobic interaction was observed between F22 and chlorine-substituted phenyl ring of compound 9i ( Figure 3C). Compound 9i also engaged in a As shown in Table 4, the compounds displayed almost no inhibition on the three downstream enzymes at 0.5 μM which further confirmed the on-target effect of the compounds.

Binding Mode of Compound 9i with PGAM1
To further understand the molecular mechanism of the anthraquinone derivatives interacting with PGAM1, we determined the X-ray structure of PGAM1 in complex with compound 9i at resolution of 1.98 Å (Table 5). Compound 9i occupied a novel allosteric site adjacent to substrate binding site with nice electron density ( Figure 3A and Figure 3B). The allosteric pocket was surrounded by the residues of F22, R90, K100, R116 and R191. In detail, the anthraquinone scaffold and sulfonamide of compound 9i interacted with the main chain carbonyl of K100 through water bridges ( Figure 3C). In addition, a hydrophobic interaction was observed between F22 and chlorine-substituted phenyl ring of compound 9i ( Figure 3C). Compound 9i also engaged in a As shown in Table 4, the compounds displayed almost no inhibition on the three downstream enzymes at 0.5 μM which further confirmed the on-target effect of the compounds.

Binding Mode of Compound 9i with PGAM1
To further understand the molecular mechanism of the anthraquinone derivatives interacting with PGAM1, we determined the X-ray structure of PGAM1 in complex with compound 9i at resolution of 1.98 Å (Table 5). Compound 9i occupied a novel allosteric site adjacent to substrate binding site with nice electron density ( Figure 3A and Figure 3B). The allosteric pocket was surrounded by the residues of F22, R90, K100, R116 and R191. In detail, the anthraquinone scaffold and sulfonamide of compound 9i interacted with the main chain carbonyl of K100 through water bridges ( Figure 3C). In addition, a hydrophobic interaction was observed between F22 and chlorine-substituted phenyl ring of compound 9i ( Figure 3C). Compound 9i also engaged in a As shown in Table 4, the compounds displayed almost no inhibition on the three downstream enzymes at 0.5 μM which further confirmed the on-target effect of the compounds.

Binding Mode of Compound 9i with PGAM1
To further understand the molecular mechanism of the anthraquinone derivatives interacting with PGAM1, we determined the X-ray structure of PGAM1 in complex with compound 9i at resolution of 1.98 Å (Table 5). Compound 9i occupied a novel allosteric site adjacent to substrate binding site with nice electron density ( Figure 3A and Figure 3B). The allosteric pocket was surrounded by the residues of F22, R90, K100, R116 and R191. In detail, the anthraquinone scaffold and sulfonamide of compound 9i interacted with the main chain carbonyl of K100 through water bridges ( Figure 3C). In addition, a hydrophobic interaction was observed between F22 and chlorine-substituted phenyl ring of compound 9i ( Figure 3C). Compound 9i also engaged in a  As shown in Table 4, the compounds displayed almost no inhibition on the three downstream enzymes at 0.5 μM which further confirmed the on-target effect of the compounds.

Binding Mode of Compound 9i with PGAM1
To further understand the molecular mechanism of the anthraquinone derivatives interacting with PGAM1, we determined the X-ray structure of PGAM1 in complex with compound 9i at resolution of 1.98 Å (Table 5). Compound 9i occupied a novel allosteric site adjacent to substrate binding site with nice electron density ( Figure 3A and Figure 3B). The allosteric pocket was surrounded by the residues of F22, R90, K100, R116 and R191. In detail, the anthraquinone scaffold and sulfonamide of compound 9i interacted with the main chain carbonyl of K100 through water bridges ( Figure 3C). In addition, a hydrophobic interaction was observed between F22 and chlorine-substituted phenyl ring of compound 9i ( Figure 3C). Compound 9i also engaged in a

Binding Mode of Compound 9i with PGAM1
To further understand the molecular mechanism of the anthraquinone derivatives interacting with PGAM1, we determined the X-ray structure of PGAM1 in complex with compound 9i at resolution of 1.98 Å (Table 5). Compound 9i occupied a novel allosteric site adjacent to substrate binding site with nice electron density ( Figure 3A,B). The allosteric pocket was surrounded by the residues of F22, R90, K100, R116 and R191. In detail, the anthraquinone scaffold and sulfonamide of compound 9i interacted with the main chain carbonyl of K100 through water bridges ( Figure 3C). In addition, a hydrophobic interaction was observed between F22 and chlorine-substituted phenyl ring of compound 9i ( Figure 3C).
Compound 9i also engaged in a π-cation interaction with R116 ( Figure 3C), which explains why modifications of the hydroxyl group led to decreased potency [39]. To validate the binding mode revealed by the co-crystal structure, we tested the activity of PGAM1 mutants (Supplementary Data, Figure S1) and the inhibition activity of compound 9i on different mutations of PGAM1. Compound 9i failed to inhibit mutations of PGAM1 (F22A, R116H and R191H) as effectively as the wild type at concentration of 5 µM which agreed with the results from crystal structure. Furthermore, a substrate competitive assay demonstrated that compound 9i held a non-competitive property with substrate 3PG which was also consistent with the binding mode revealed by X-ray structure. The co-crystal structure together with the molecular biological assays illustrated the binding mode of the anthraquinone inhibitor with PGAM1 and provided useful information for further optimization. mutations of PGAM1 (F22A, R116H and R191H) as effectively as the wild type at concentration of 5 μM which agreed with the results from crystal structure. Furthermore, a substrate competitive assay demonstrated that compound 9i held a non-competitive property with substrate 3PG which was also consistent with the binding mode revealed by X-ray structure. The co-crystal structure together with the molecular biological assays illustrated the binding mode of the anthraquinone inhibitor with PGAM1 and provided useful information for further optimization.   The data are presented as mean ± s.d.

Inhibition Activity of Selected Compounds on Cancer Cell Proliferation
Given PGAM1 plays a crucial part in cancer metabolism, inhibition of PGAM1 by small molecules is supposed to suppress cancer cell proliferation. We selected the compounds with IC 50 values below 10 µM toward PGAM1 to inhibit H1299 cells proliferation. Then the potent inhibitors with IC 50 values below 20 µM on H1299 cells were further tested on A549 and PC9 cells. In general, proliferation inhibition of the compounds was correlated with PGAM1 inhibition and the inhibitors performed similarly among different cancer cells, with IC 50 values ranging from approximately 6-50 µM (Tables 6 and 7). Notably, compound 9i effectively suppressed the three different cancer cells proliferation which was also potent towards PGAM1. Table 6. Inhibition activity of compounds 9a-q on cancer cell proliferation.
with IC50 values below 20 μM on H1299 cells were further tested on A549 and PC9 cells. In general, proliferation inhibition of the compounds was correlated with PGAM1 inhibition and the inhibitors performed similarly among different cancer cells, with IC50 values ranging from approximately 6-50 μM ( Table 6 and Table 7). Notably, compound 9i effectively suppressed the three different cancer cells proliferation which was also potent towards PGAM1. phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below. maintaining potency. Then, we studied the impact of substituents of different halogens on the phenyl ring (9f-q) on the inhibition activity towards PGAM1 (Table 2). Generally, compounds with Cl, Br and I substituents (9i-q) with IC50 values below 0.5 μM performed better than F substituents (9f-h). Among them, we discovered one of the most potent PGAM1 inhibitors 9i with an IC50 value of 0.27 μM and we solved the co-crystal structure of compound 9i with PGAM1 which would be discussed below.  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized ompounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition 24  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized ompounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition 31  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized ompounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized ompounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized ompounds 14a-g. However, diverse modifications of the hydroxyl caused a loss of inhibition 23  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized 34  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized 6  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized 28  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized 50 ± 0.6 n.d. n.d.  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized 24  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized  We next moved the sulfonamide to the C4 position of alizarin by substituting the hydroxyl, ollowed by nitrification, reduction, and sulfonylation. Probably due to the modification of the ydroxyl and replacement of the sulfonamide, compounds 12 and 13 failed to maintain potency, ecreasing 10% and 34% of PGAM1 activity at 10 μM, respectively. Finally, considering the good olubility and potency of inhibitor 9i, we kept the 4-chlorobenzene group and synthesized 26.4 ± 3.8 n.d. n.d.

General Procedures
All reagents were purchased commercially. 1 H-NMR and 13 C-NMR spectra were recorded on Bruker AC400 and Bruker AC600 NMR spectrometers, respectively (Billerica, MA, USA). Low-resolution mass spectra were recorded on a 6120 Quadrupole mass spectrometer (Agilent, Santa Clara, CA, USA) equipped with electrospray ionization (ESI). High-resolution mass spectra were determined on triple TOF 5600 + MS/MS system (AB Sciex, Concord, ON, Canada) in negative ESI mode. The purity of target compounds was determined by high-performance liquid chromatography (Agilent, Santa Clara, CA, USA, DIKMA Diamonsil Plus C18, 250 × 4.6 mm, 5 µm, 25 • C, UV 290 nM). All the biologically tested compounds achieved ≥95% purity.

Constructions of Mutations of PGAM1
The wild-type plasmid was a gift from Department of Chemistry and Institute for Biophysical Dynamics, University of Chicago (Chicago, IL, USA). To obtain PGAM1 F22A/R116H/R191H mutant construct, the quick change kit (#KOD-401, TOYOBO, Osaka, Japan) was used to make site-specific mutagenesis, which confirmed by standard DNA sequencing methods. The primers used for the mutation were listed below: PGAM1 F22A Forward: GAACCTGGAGAACCGCGCCAGCGGCTGGTACGAC PGAM1 F22A Reverse: GTCGTACCAGCCGCTGGCGCGGTTCTCCAGGTTC PGAM1 K116H Forward: CAGGTGAAGATCTGGCACCGCTCCTATGATGTCC PGAM1 K116H Reverse: GGACATCATAGGAGCGGTGCCAGATCTTCATCTG PGAM1 K119H Forward: CATGGCAACAGCCTCCACGGCATTGTCAAGCAT PGAM1 K119H Reverse: ATGCTTGACAATGCCGTGGAGGCTGTTGCCATG

Protein Purification, Crystallization, Data Collection and Structure Determination
The C-terminal His 6 -tagged PGAM1 was expressed and affinity purified following the reported protocols [22]. The crystals of PGAM1 were obtained using the hanging drop vapor-diffusion method at 16 • C in a crystallization buffer containing of 8% (w/v) PEG3350 and 100 mM MES 6.0. To obtain the co-crystal of PGAM1 with compound 9i, the crystals of PGAM1 were soaked in stock solution containing 500 µM 9i for 2 h. Crystals were then cryo-protected by brief soaking in a mixture solution of crystallization buffer:glycerol (76:24) and quickly cooled in liquid nitrogen. Diffraction data were collected at beamline BL17U1, BL18U1 and BL19U1 in the Shanghai Synchrotron Radiation Facility (SSRF, Shanghai, China) using an X-ray beam of wavelength 0.97851 Å. The data was handled with HKL3000 [41] and the structure was determined by using molecular replacement in CCP4 [42] via an initial model of PGAM1 derived from PDB entry 4GPZ [22]. The model was then refined to 1.98 Å resolution using Phenix. The ligand restraints were generated using eLBOW in Phenix and manual rebuilding of the model was completed using the molecular graphics program COOT [43] according to the electronic density. All the graphs were plotted by Pymol (DeLano Scientific LLC, San Carlos, CA, USA).

Cell Viability Assays
The H1299 cell line was obtained from Guangzhou Jenniobio Biotechnology Co., Ltd. (Guangzhou, China), PC9 and A549 cells were gifted from Deng Jiong's lab in the School of Medicine, Shanghai Jiao Tong University, Shanghai, China). H1299, PC9 cells were cultured in RPMI-1640 medium and A549 cells were cultured in F-12K medium containing 10% fetal bovine serum (FBS), 100 units/mL of Penicillin and 100 µg/mL Streptomycin. In cell viability assays, 2000 H1299 cells, A549 cells or 1000 PC9 cells per well were seeded in 96-well plate. After attachment for 24 h, the cells were treated with indicated inhibitor for 72 h. After incubation with 0.5 mg/mL methylthiazolyldiphenyl-tetrazolium bromide (MTT) for 4 h, cell viability was measured as the OD at 570 nm.

Conclusions
In summary, we designed and synthesized 31 anthraquinone derivatives as PGAM1 inhibitors. The SAR study focused on the effect of 1,2-substituents, 3-substituents, 2,4-substituents, and 2,3-substituents on the anthraquinone core scaffold activity against PGAM1. Among them, 3-sulfonamide substituents but not 4-substituents on the anthraquinone scaffold were essential for maintaining potency, which was further revealed by the X-ray structure of PGAM1 complexed with compound 9i with IC 50 value of 0.27 µM. In addition, modifications of the hydroxyl of alizarin disrupted the π-cation interaction of PGAM1 and the anthraquinone derivatives which led to the loss of potency. Taken as a whole, the SAR study of anthraquinone derivatives against PGAM1 provided useful information for further discovery of PGAM1 inhibitors.
Supplementary Materials: The following are available online, Figure S1: The activity of PGAM1 mutant.