Structure–Activity Relationships (SARs) of α-Ketothioamides as Inhibitors of Phosphoglycerate Dehydrogenase (PHGDH)

For many years now, targeting deregulation within cancer cells’ metabolism has appeared as a promising strategy for the development of more specific and efficient cancer treatments. Recently, numerous reports highlighted the crucial role of the serine synthetic pathway, and particularly of the phosphoglycerate dehydrogenase (PHGDH), the first enzyme of the pathway, to sustain cancer progression. Yet, because of very weak potencies usually in cell-based settings, the inhibitors reported so far failed to lay ground on the potential of this approach. In this paper, we report a structure–activity relationship study of a series of α-ketothioamides that we have recently identified. Interestingly, this study led to a deeper understanding of the structure–activity relationship (SAR) in this series and to the identification of new PHGDH inhibitors. The activity of the more potent compounds was confirmed by cellular thermal shift assays and in cell-based experiments. We hope that this research will eventually provide a new entry point, based on this promising chemical scaffold, for the development of therapeutic agents targeting PHGDH.


Introduction
The ability of cancer cells to reprogram their metabolism was demonstrated in numerous studies over the past two decades [1][2][3]. These metabolic modifications are used by cancer cells in order to proliferate in harmful environments (hypoxic zones, deficiencies in the external supply of glucose and amino acids) by producing certain metabolites and antioxidants more efficiently. It quickly became apparent that the development of treatments specifically targeting these deregulated metabolic pathways would reduce tumor development while significantly limiting the side effects associated with conventional non-targeted therapies [4].
In this purpose, among the many pathways studied, the serine synthesis pathway appeared to be a promising target [5,6]. Indeed, this pathway plays a central role in one-carbon metabolism, methylation processes, and allows the cells to produce the building blocks required for their proliferation [7]. Serine can also be used to produce glutathione and therefore offers resistance against the oxidative stress associated with intense proliferation [8]. In addition, the three enzymes constituting this pathway (phosphoglycerate dehydrogenase (PHGDH), PSAT-1, and PSPH) are overexpressed in many cancer types (breast cancer, melanoma, glioma) and often associated with poor prognosis in the patient [9].
Phosphoglycerate dehydrogenase (PHGDH) is the first enzyme to catalyze the rate-limiting step of the serine synthesis pathway. Recently, its study highlighted key roles in diverting glycolytic flow to the serine pathway leading to a tumor invasion process. PHGDH is overexpressed and/or genomically amplified in more than 16% of cancer lines [10]. Moreover, the overexpression of this protein appears to lead to greater metastatic invasion in animal models and could also be correlated with greater aggressiveness and lower survival rates [11,12]. Finally, it was demonstrated that inhibition or silencing of this enzyme can selectively impact tumor growth while sparing healthy cells [13]. Therefore, in recent years, there has been an increased interest in the development of small molecule PHGDH inhibitors.
Since 2015, several chemical series were identified to inhibit PHGDH enzymatic activity ( Figure 1). Two of them are NAD + -competitive and thus interact with the Rossmann fold (BI-4924 and compound 18). These compounds exhibit PHGDH inhibition in the submicromolar range in isolated enzyme assays but have poor cellular efficacy notably because of the competition with a high intracellular NAD + concentration [14,15]. The other series of inhibitors (CBR-5884, NCT-503, disulfiram, Azacoccone, PKUMLD-WQ-220, Ixocarpolactone) have an allosteric mode of action, and although less effective than the NAD + -competitive compounds, they are usually characterized by a better inhibition profile on cancer cell lines in in vitro and in vivo xenograft models. However, for most of these allosteric inhibitors, their precise sites of action are yet to be determined or have been concluded from docking studies so far [16][17][18][19][20][21]. proliferation [7]. Serine can also be used to produce glutathione and therefore offers resistance against the oxidative stress associated with intense proliferation [8]. In addition, the three enzymes constituting this pathway (phosphoglycerate dehydrogenase (PHGDH), PSAT-1, and PSPH) are overexpressed in many cancer types (breast cancer, melanoma, glioma) and often associated with poor prognosis in the patient [9].
Phosphoglycerate dehydrogenase (PHGDH) is the first enzyme to catalyze the rate-limiting step of the serine synthesis pathway. Recently, its study highlighted key roles in diverting glycolytic flow to the serine pathway leading to a tumor invasion process. PHGDH is overexpressed and/or genomically amplified in more than 16% of cancer lines [10]. Moreover, the overexpression of this protein appears to lead to greater metastatic invasion in animal models and could also be correlated with greater aggressiveness and lower survival rates [11,12]. Finally, it was demonstrated that inhibition or silencing of this enzyme can selectively impact tumor growth while sparing healthy cells [13]. Therefore, in recent years, there has been an increased interest in the development of small molecule PHGDH inhibitors.
Since 2015, several chemical series were identified to inhibit PHGDH enzymatic activity ( Figure  1). Two of them are NAD + -competitive and thus interact with the Rossmann fold (BI-4924 and compound 18). These compounds exhibit PHGDH inhibition in the submicromolar range in isolated enzyme assays but have poor cellular efficacy notably because of the competition with a high intracellular NAD + concentration [14,15]. The other series of inhibitors (CBR-5884, NCT-503, disulfiram, Azacoccone, PKUMLD-WQ-220, Ixocarpolactone) have an allosteric mode of action, and although less effective than the NAD + -competitive compounds, they are usually characterized by a better inhibition profile on cancer cell lines in in vitro and in vivo xenograft models. However, for most of these allosteric inhibitors, their precise sites of action are yet to be determined or have been concluded from docking studies so far [16][17][18][19][20][21].
Despite the interest in targeting PHGDH and the increasing number of reported inhibitors, the weak in cellulo and in vivo efficacy of the inhibitors hamper the understanding of the role PHGDH plays in cancer progression. To date, the developed compounds were mainly used to study the serine synthesis pathway in restoring the sensitivity of cancer cells to conventional chemotherapy (bortezomib, doxorubicin, sorafenib) [22][23][24]. The development of more effective PHGDH inhibitors to investigate the serine synthesis pathway in cancer biology hence remains necessary.  Despite the interest in targeting PHGDH and the increasing number of reported inhibitors, the weak in cellulo and in vivo efficacy of the inhibitors hamper the understanding of the role PHGDH plays in cancer progression. To date, the developed compounds were mainly used to study the serine synthesis pathway in restoring the sensitivity of cancer cells to conventional chemotherapy (bortezomib, doxorubicin, sorafenib) [22][23][24]. The development of more effective PHGDH inhibitors to investigate the serine synthesis pathway in cancer biology hence remains necessary.
For the modification of the central linker, syntheses of compounds are outlined in Scheme 2. Compounds 22, 24, 28, and 30 were synthesized by nucleophilic acyl substitution. In the first step, different carboxylic acids were acylated with oxalyl chloride to obtain the corresponding acid chlorides. These chlorides then reacted with a morpholine molecule, or 4-amino-morpholine in the case of compound 30, to obtain the desired compounds. Benzaldehyde or acetophenone were used in the Willgerodt-Kindler reaction described above to lead to the synthesis of compounds 23 and 25. The compound 26 was obtained thanks to an SN2-type reaction between morpholine and bromoacetophenone and the alcohol 27 was obtained after the NaBH4 reduction of 24. The reaction of commercially available ethyl morpholine-4-carboxylate with POCl3 afforded the corresponding acyl chloride, which was converted to compound 29 by reaction with aniline. The reaction of 4aminomorpholine and benzaldehyde furnished the desired methylethanimine derivative 31. The reaction of morpholine with isothiocyanatobenzene or isocyanatobenzene afforded the Scheme 1. Synthetic pathway of α-ketothioamide derivatives 1-21, 38, and 39 a . a Reagents and conditions: (i) Br 2 , CHCl 3 , TBAB, r.t., for 2 h; (ii) S 8 , appropriated amino group, DMF, r.t., 8%-73%. R represents aryl or alkyl moieties and R' and R" represent different amino group as detailed in § 2.3 SAR's Investigation.
For the modification of the central linker, syntheses of compounds are outlined in Scheme 2. Compounds 22, 24, 28, and 30 were synthesized by nucleophilic acyl substitution. In the first step, different carboxylic acids were acylated with oxalyl chloride to obtain the corresponding acid chlorides. These chlorides then reacted with a morpholine molecule, or 4-amino-morpholine in the case of compound 30, to obtain the desired compounds. Benzaldehyde or acetophenone were used in the Willgerodt-Kindler reaction described above to lead to the synthesis of compounds 23 and 25. The compound 26 was obtained thanks to an SN2-type reaction between morpholine and bromo-acetophenone and the alcohol 27 was obtained after the NaBH 4 reduction of 24. The reaction of commercially available ethyl morpholine-4-carboxylate with POCl 3 afforded the corresponding acyl chloride, which was converted to compound 29 by reaction with aniline. The reaction of 4-aminomorpholine and benzaldehyde furnished the desired methylethanimine derivative 31. The reaction of morpholine with isothiocyanatobenzene or isocyanatobenzene afforded the hydrazinecarbothioamide 32 and the urea derivative 33, respectively, with good yield (>76%). Compound 34 was synthesized by a two-step reaction. Firstly, CS 2 and morpholine were used in order to produce the corresponding morpholinium salt. Then, this salt was reacted with (thiocyanatomethyl)benzene in acetonitrile to afford the desired derivative 34. Coupling between the commercially available alcohol and morpholine-4-carbonyl chloride furnished the desired compound 35. Finally, the reaction of morpholine or 4-amino-morpholine with a benzenesulfonyl chloride generated compounds 36 and 37.
Pharmaceuticals 2020, 13, 20 4 of 21 hydrazinecarbothioamide 32 and the urea derivative 33, respectively, with good yield (>76%). Compound 34 was synthesized by a two-step reaction. Firstly, CS2 and morpholine were used in order to produce the corresponding morpholinium salt. Then, this salt was reacted with (thiocyanatomethyl)benzene in acetonitrile to afford the desired derivative 34. Coupling between the commercially available alcohol and morpholine-4-carbonyl chloride furnished the desired compound 35. Finally, the reaction of morpholine or 4-amino-morpholine with a benzenesulfonyl chloride generated compounds 36 and 37.

PHGDH Biochemical Assay Optimization
Prior to inhibitor evaluation, we undertook the optimization of our biochemical assay in order to achieve reliable results. Our initial PHGDH assay contained solely the dehydrogenase and its substrates; the enzyme activity was quantified by directly measuring NADH fluorescence, hence yielded a low fluorescent signal ( Figure 2B).

PHGDH Biochemical Assay Optimization
Prior to inhibitor evaluation, we undertook the optimization of our biochemical assay in order to achieve reliable results. Our initial PHGDH assay contained solely the dehydrogenase and its substrates; the enzyme activity was quantified by directly measuring NADH fluorescence, hence yielded a low fluorescent signal ( Figure 2B).
PHGDH activity was monitored by measuring a continuous increase of fluorescence (Ex 544 nm/Em 590 nm) in a coupled-enzyme system including three enzymes: PHGDH, PSAT1, and diaphorase ( Figure 2A). The addition of PSAT1 promotes accumulation of NADH by the PHGDH reaction and generates a significant increase in the fluorescent signal ( Figure 2C). Moreover, by coupling PHGDH to diaphorase, which utilizes NADH to produce the fluorescent molecule resorufin from resazurin, the assay red-shifted into a spectral region that reduced interference from compounds, as well as greatly increased.
In optimized conditions, Michaelis-Menten constants were estimated for substrates of PHGDH (3-PG and NAD + ) and substrates of PSAT1 (3-PSer and α-KG). Km values for 3-PG, NAD + , 3-PSer, and α-KG were determined to be 294.3, 30.4, 25.1, and 35 μM, respectively (Table 1). The optimized fully coupled assay provides robust measurement of PHGDH activity, suggesting it was amenable to evaluation of our inhibitors. PHGDH activity was monitored by measuring a continuous increase of fluorescence (Ex 544 nm/Em 590 nm) in a coupled-enzyme system including three enzymes: PHGDH, PSAT1, and diaphorase ( Figure 2A). The addition of PSAT1 promotes accumulation of NADH by the PHGDH reaction and generates a significant increase in the fluorescent signal ( Figure 2C). Moreover, by coupling PHGDH to diaphorase, which utilizes NADH to produce the fluorescent molecule resorufin from resazurin, the assay red-shifted into a spectral region that reduced interference from compounds, as well as greatly increased.

SARs Investigation
Although a structural variety has been introduced at part A of the α-ketothioamide scaffold in our previous work, the substitution by alkyl groups has not been previously studied in detail. Hence, Pharmaceuticals 2020, 13, 20 6 of 22 we began our PHGDH SAR investigations with a focus on part A. As depicted in Table 2, only aryl groups led to PHGDH inhibition (1, 2, and 4) whereas the introduction of alkyl groups in this position resulted in inactive derivatives. Moreover, substitutions on the aryl moiety are only tolerated in the para position with the chlorinated derivative 2 being the most potent. Table 2. Results of PHGDH inhibition for compounds 1-11. All experiments were performed with at least triplicates and IC 50 values were determined in two or more independent experiments. * Compounds previously published in Ravez et al. [25].
Pharmaceuticals 2020, 13, 20 6 of 21 Although a structural variety has been introduced at part A of the α-ketothioamide scaffold in our previous work, the substitution by alkyl groups has not been previously studied in detail. Hence, we began our PHGDH SAR investigations with a focus on part A. As depicted in Table 2, only aryl groups led to PHGDH inhibition (1, 2, and 4) whereas the introduction of alkyl groups in this position resulted in inactive derivatives. Moreover, substitutions on the aryl moiety are only tolerated in the para position with the chlorinated derivative 2 being the most potent. To expand the SARs, the introduction of different amino moieties on part C was undertaken keeping fixed parts A and B (Table 3). Except for the 4-methylpiperidine analog (14), the introduction of a bulky group in part C of the molecule abolished the inhibitory potency. Cyclohexyl Ethyl >150 10 Adamantyl To expand the SARs, the introduction of different amino moieties on part C was undertaken keeping fixed parts A and B (Table 3). Except for the 4-methylpiperidine analog (14), the introduction of a bulky group in part C of the molecule abolished the inhibitory potency. Although a structural variety has been introduced at part A of the α-ketothioamide scaffold in our previous work, the substitution by alkyl groups has not been previously studied in detail. Hence, we began our PHGDH SAR investigations with a focus on part A. As depicted in Table 2, only aryl groups led to PHGDH inhibition (1, 2, and 4) whereas the introduction of alkyl groups in this position resulted in inactive derivatives. Moreover, substitutions on the aryl moiety are only tolerated in the para position with the chlorinated derivative 2 being the most potent. To expand the SARs, the introduction of different amino moieties on part C was undertaken keeping fixed parts A and B (Table 3). Except for the 4-methylpiperidine analog (14), the introduction of a bulky group in part C of the molecule abolished the inhibitory potency. Although a structural variety has been introduced at part A of the α-ketothioamide scaffold in our previous work, the substitution by alkyl groups has not been previously studied in detail. Hence, we began our PHGDH SAR investigations with a focus on part A. As depicted in Table 2, only aryl groups led to PHGDH inhibition (1, 2, and 4) whereas the introduction of alkyl groups in this position resulted in inactive derivatives. Moreover, substitutions on the aryl moiety are only tolerated in the para position with the chlorinated derivative 2 being the most potent. Table 2. Results of PHGDH inhibition for compounds 1-11. All experiments were performed with at least triplicates and IC50 values were determined in two or more independent experiments. * Compounds previously published in Ravez et al. [25].
To expand the SARs, the introduction of different amino moieties on part C was undertaken keeping fixed parts A and B (Table 3). Except for the 4-methylpiperidine analog (14), the introduction of a bulky group in part C of the molecule abolished the inhibitory potency. Although a structural variety has been introduced at part A of the α-ketothioamide scaffold in our previous work, the substitution by alkyl groups has not been previously studied in detail. Hence, we began our PHGDH SAR investigations with a focus on part A. As depicted in Table 2, only aryl groups led to PHGDH inhibition (1, 2, and 4) whereas the introduction of alkyl groups in this position resulted in inactive derivatives. Moreover, substitutions on the aryl moiety are only tolerated in the para position with the chlorinated derivative 2 being the most potent. Table 2. Results of PHGDH inhibition for compounds 1-11. All experiments were performed with at least triplicates and IC50 values were determined in two or more independent experiments. * Compounds previously published in Ravez et al. [25].
To expand the SARs, the introduction of different amino moieties on part C was undertaken keeping fixed parts A and B (Table 3). Except for the 4-methylpiperidine analog (14), the introduction of a bulky group in part C of the molecule abolished the inhibitory potency.

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC 50 > 150 µM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC 50 = 106 µM) or modification by sulfonohydrazide (37, IC 50 = 92 µM).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and 39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and 39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A). Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and 39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and 39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A). Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and 39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A). Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and 39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A). Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and 39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A). Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and  39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and  39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and  39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and  39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and  39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and  39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and  39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and  39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and  39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

>150
Subsequently, we shifted our SAR analysis to the linker portion B (Table 4). This feature of the pharmacophore proved essential to activity as the replacement led to significant loss of PHGDH activity (IC50 > 150 μM). However, two modifications to this region of the molecule tolerated such shortening of the linkage from the thiocarbonyl (23, IC50 = 106 μM) or modification by sulfonohydrazide (37, IC50 = 92 μM). Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and  39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

[58.4-145.2]
Based on the modification of the linker portion, two 4-chloroderivatives carrying a thiocarbonyl (38) or a sulfonohydrazide (39) instead of a ketothioamide linker were designed and synthesized (Scheme 2). As expected, substitution of the para position by a chlorine atom resulted in an improvement in PHGDH inhibition ( Figure 3B). The most promising compounds (1, 2, 23, 38, 37, and 39) were validate in a counter screen against PSAT1 and the diaphorase. Based on the preliminary study reported earlier and the present work, the following SARs can be highlighted ( Figure 3A).

PHGDH Engagement Investigated by Cellular Thermal Shift Assay (CETSA)
Next, in order to confirm that the most potent inhibitors target PHGDH in more complex biological models, we measured their ability to stabilize PHGDH in cell lysates using the cellular thermal shift assay (CETSA). In this assay, HL-60 (leukemia cancer cells) lysates were incubated with DMSO or with 80 μM of the inhibitor. The lysate was then heated to a pre-determined temperature, at which most of PHGDH was denaturated (60 °C). The fraction of non-denatured PHGDH was quantified by Western blot and allowed to identify the compounds that bind and stabilize PHGDH. The percentages of stabilization compared to the DMSO control and non-denatured PHGDH (heated to 37 °C) of the six most potent PHGDH inhibitors are depicted in Figure 4. While in untreated cells PHGDH was denaturated at 60 °C, the addition of compounds 1, 23, and 37 leads to a small stabilization of the protein (1.2%, 0.8%, and 1.1%, respectively) thus proving

PHGDH Engagement Investigated by Cellular Thermal Shift Assay (CETSA)
Next, in order to confirm that the most potent inhibitors target PHGDH in more complex biological models, we measured their ability to stabilize PHGDH in cell lysates using the cellular thermal shift assay (CETSA). In this assay, HL-60 (leukemia cancer cells) lysates were incubated with DMSO or with 80 µM of the inhibitor. The lysate was then heated to a pre-determined temperature, at which most of PHGDH was denaturated (60 • C). The fraction of non-denatured PHGDH was quantified by Western blot and allowed to identify the compounds that bind and stabilize PHGDH. The percentages of stabilization compared to the DMSO control and non-denatured PHGDH (heated to 37 • C) of the six most potent PHGDH inhibitors are depicted in Figure 4.

PHGDH Engagement Investigated by Cellular Thermal Shift Assay (CETSA)
Next, in order to confirm that the most potent inhibitors target PHGDH in more complex biological models, we measured their ability to stabilize PHGDH in cell lysates using the cellular thermal shift assay (CETSA). In this assay, HL-60 (leukemia cancer cells) lysates were incubated with DMSO or with 80 μM of the inhibitor. The lysate was then heated to a pre-determined temperature, at which most of PHGDH was denaturated (60 °C). The fraction of non-denatured PHGDH was quantified by Western blot and allowed to identify the compounds that bind and stabilize PHGDH. The percentages of stabilization compared to the DMSO control and non-denatured PHGDH (heated to 37 °C) of the six most potent PHGDH inhibitors are depicted in Figure 4. While in untreated cells PHGDH was denaturated at 60 °C, the addition of compounds 1, 23, and 37 leads to a small stabilization of the protein (1.2%, 0.8%, and 1.1%, respectively) thus proving While in untreated cells PHGDH was denaturated at 60 • C, the addition of compounds 1, 23, and 37 leads to a small stabilization of the protein (1.2%, 0.8%, and 1.1%, respectively) thus proving the interaction between these compounds and PHGDH in cell lysates. More interestingly, Pharmaceuticals 2020, 13, 20 9 of 22 the para-chlorinated derivatives (2, 38, and 39) present a greater stabilization of the protein compared to the non-halogenated compounds (2.8%, 3.1%, and 3.2% respectively); hence, this is in agreement with increased PHGDH inhibition in the enzyme activity assay.

Cell-Based Evaluation
We assessed the ability of para-chlorinated derivatives, 38 and 39, to inhibit tumor proliferation. To this end, breast cancer cells overexpressing PHGDH (BT-20), breast cancer cells that do not express PHGDH (MDA-MB-231) and healthy fibroblast cells as control (BJ-5ta) were treated with compounds 1, 2, 38, 39 and the reference inhibitor NCT-503, used for comparison purpose. The IC 50 values determined are depicted on Table 5. Table 5. Cell proliferation inhibition data (IC 50 ). All experiments to determine IC 50 values were performed with at least triplicates at each compound dilution, and all IC 50 values were averaged when determined in two or more independent experiments. * Values previously published in Ravez et al. [25].

PHGDH-Dependent PHGDH-Independent Normal Cell Line
BT-20 MDA-MB-231 BJ-5ta Except for compound 39, which was deprived of any inhibition of tumor proliferation up to 100 µM probably due to poor membrane permeability, compounds 1, 2, and 38 selectively inhibit BT-20 growth without affecting healthy BJ-5ta or the PHGDH-independent tumor cell line, MDA-MB-231. Interestingly, these inhibitions were similar to the one observed with NCT-503. Again, these results demonstrate the potential of this compound class as a starting point for the design of new therapeutic agents targeting PHGDH.

Conclusions
The development of new PHGDH inhibitors is emerging as a promising strategy to develop new targeted therapy. We previously reported the discovery of compound 1, a PHGDH inhibitor with robust cancer cell inhibition. Here, we have detailed the SARs investigation around this α-ketothioamide motif. Replacement of the aryl and modulation of the amino groups were not tolerated in terms of PHGDH inhibition. However, modification of the ketothioamide linker by a thiocarbonyl or by a sulfonohydrazide led to the identification of novel inhibitors. As expected, the addition of a chlorine atom in para position on the most promising molecules resulted in the discovery of novel micromolar range inhibitors (38 and 39). Target engagement was confirmed in cellular thermal shift assay (CETSA) and the anticancer potential was validated in cell-based experiments. Interestingly, the identification of the sulfonohydrazide motif as a new series of inhibitors of this enzyme offers interesting prospects in the development of future PHGDH inhibitors, this function being reported in several works as non-toxic [29,30]. It should be noted that recent mass spectrometry studies suggest that some of the inhibitors discussed here would interact at the c-terminal end of PHGDH. This will be detailed in a forthcoming publication.

General Chemistry
All reagents were purchased from chemical suppliers and used without purification. Thin-layer chromatography (TLC) was performed using silica gel 60 F254 plates, with observation under UV when necessary. Melting points were recorded on an Electrothermal IA9000 melting point system. 1 H NMR spectra were recorded on an AVANCE II 400 MHz Bruker spectrometer with CDCl 3 or DMSO-d 6 as the solvent. 13 C NMR spectra were recorded at 100 MHz. All coupling constants were measured in hertz (Hz), and the chemical shifts (δH and δC) were quoted in parts per million (ppm) relative to TMS (δ0), which was used as the internal standard. Data were reported as follows: chemical shift, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, br = broad, m = multiplet), integration, and coupling constant (Hz). High-resolution mass spectroscopy (HRMS) analyses were carried out on an LTQ-Orbitrap XL hybrid mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). Data were acquired in positive ion mode using full-scan MS with a mass range of 100-1000 m/z. The orbitrap resolution was set at 30,000 (fwhm definition). All experimental data were acquired using daily external calibration prior to data acquisition. Appropriate tuning of the electrospray ion source was done. The following electrospray inlet conditions were applied: flow rate, 100 µL/min; spray voltage, 5 kV; sheath gas (N 2 ) flow rate, 20 arbitrary unit (au); auxiliary gas (N 2 ) flow rate, 10 au; capillary temperature, 275 • C; capillary voltage, 45 V; tube lens, 80 V. High-performance liquid chromatography (HPLC) analyses were performed on an LC system using a YMC-Triart C-18 (250 mm × 4.6 mm, 5 µm) column as the stationary phase. Mobile phase contained water/CH 3 CN (30:70, v/v) and was maintained isocratically at the flow rate of 1 mL/min. The column temperature was maintained at room temperature. The peaks were monitored at a wavelength of 215 nm. The purity of all compounds tested was greater than 95%, as determined by HPLC and 1 H NMR.
Compounds 1-11 were synthesized according to General Procedure I.