Development of Novel Ecto-Nucleotide Pyrophosphatase/Phosphodiesterase 1 (ENPP1) Inhibitors for Tumor Immunotherapy

The cyclic guanosine monophosphate–adenosine monophosphate synthase–stimulator of interferon genes–TANK-binding kinase 1–interferon regulating factor 3 (cGAS-STING-TBK1-IRF3) axis is now acknowledged as the major signaling pathway in innate immune responses. However, 2′,3′-cGAMP as a STING stimulator is easily recognized and degraded by ecto-nucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1), which reduces the effect of tumor immunotherapy and promotes metastatic progression. In this investigation, the structure-based virtual screening strategy was adopted to discover eight candidate compounds containing zinc-binding quinazolin-4(3H)-one scaffold as ENPP1 inhibitors. Subsequently, these novel inhibitors targeting ENPP1 were synthesized and characterized by NMR and high-resolution mass spectra (HRMS). In bioassays, 7-fluoro-2-(((5-methoxy-1H-imidazo[4,5-b]pyridin-2-yl)thio)methyl)quina-zolin-4(3H)-one(compound 4e) showed excellent activity against the ENPP1 at the molecular and cellular levels, with IC50 values of 0.188 μM and 0.732 μM, respectively. Additionally, compound 4e had superior selectivity towards metastatic breast cancer cells (4T1) than towards normal cells (LO2 and 293T) in comparison with cisplatin, indicating that compound 4e can potentially be used in metastatic breast cancer therapy. On the other hand, compound 4e upgraded the expression levels of IFN-β in vivo by preventing the ENPP1 from hydrolyzing the cGAMP to stimulate a more potent innate immune response. Therefore, this compound might be applied to boost antitumor immunity for cancer immunotherapy. Overall, our work provides a strategy for the development of a promising drug candidate targeting ENPP1 for tumor immunotherapy.


Introduction
Immune checkpoint inhibitors are an important branch of tumor immunotherapy. They activate adaptive regulatory T cells to kill tumor cells by disrupting immunosuppression signals [1,2]. However, immune checkpoint inhibitors have a few limitations and drug resistance, and their response to noninflammatory cancers is very weak [3,4]. The cGAS-STING-TBK1-IRF3 signaling pathway (Figure 1), a significant part of the innate immune system, is thought to be a broader immunotherapy strategy because it enhances tumor immunogenicity [5][6][7][8]. When the dsDNA of tumor cells enters normal cells, it can be promptly sensed and detected by cyclic guanosine monophosphate (GMP)-adenosine monophosphate (AMP) synthase (cGAS) [9,10]. The subsequent binding of dsDNA with cGAS leads to the activation of cGAS and initiates the catalytic synthesis of 2 ,3 -cyclic GMP-AMP (2 ,3 -cGAMP) from guanosine triphosphate (GTP) and adenosine triphosphate (ATP) [11,12]. The cGAS-synthesized 2 ,3 -cGAMP activates the interferon-gene stimulating factor (STING) in ER to issue in a downstream signaling cascade via the recruitment of TANK-binding kinase 1 (TBK1) and the phosphorylation of the interferon regulatory transcription factor 3 (IRF3) [13,14]. The incitement of IRF3 triggers the production of stimulating factor (STING) in ER to issue in a downstream signaling cascade via the recruitment of TANK-binding kinase 1 (TBK1) and the phosphorylation of the interferon regulatory transcription factor 3 (IRF3) [13,14]. The incitement of IRF3 triggers the production of type I IFN and many other pro-inflammatory cytokines, which can activate CD8 + T cells to kill tumor cells [15]. However, 2′,3′-cGAMP, as a STING stimulator, is easily recognized and degraded by ecto-nucleotide pyrophos-phatase/phosphodiesterase 1 (ENPP1), which weakens the effect of tumor immunotherapy [16][17][18]. ENPP1 is involved in different biological processes and hydrolyzes a wide range of phosphodiester bonds, such as nucleoside triphosphates, dinucleotides, cyclic (di-)nucleotides, and nucleotide sugars [19,20]. Li et al. found that ENPP1 hydrolyzed cGAMP (cyclic (di-)nucleotides) and negatively regulated the anti-cancer immune response [17,18]. Moreover, they reported a series of potent ENPP1 inhibitors that could delay tumor growth in a metastatic breast cancer mouse model due to increasing endogenous cGAMP [16]. Furthermore, Bakhoum et al. reported that ENPP1 could hydrolyze extracellular cGAMP to generate AMP and GMP in various metastatic cancers, especially in triple-negative breast cancer (MDA-MB-231 cell line) [21]. The produced AMP was subsequently hydrolyzed into adenosine by CD73 to facilitate cancer-cell migration and extend valid immune suppressive effects [22]. Extracellular cGAMP hydrolysis by ENPP1 transformed an immune stimulatory pathway into an immune-suppressive mechanism that promoted tumor progression. Therefore, finding highly effective ENPP1 inhibitors is a potential therapeutic in metastatic breast cancer.
A few nucleotide-derived ENPP1 inhibitors have been reported to date ( Figure 2). Sevigny et al. found α-borano-β,γ-metATP derivative, which showed high inhibitory activity against ENPP1 for the replacement of one oxygen atom in the α-phosphate by a borano atom [23]. However, nucleotide-based inhibitors had limited applicability due to their high acidity, resulting in poor peroral bioavailability. As a result, non-nucleotide- ENPP1 is involved in different biological processes and hydrolyzes a wide range of phosphodiester bonds, such as nucleoside triphosphates, dinucleotides, cyclic (di-)nucleotides, and nucleotide sugars [19,20]. Li et al. found that ENPP1 hydrolyzed cGAMP (cyclic (di-)nucleotides) and negatively regulated the anti-cancer immune response [17,18]. Moreover, they reported a series of potent ENPP1 inhibitors that could delay tumor growth in a metastatic breast cancer mouse model due to increasing endogenous cGAMP [16]. Furthermore, Bakhoum et al. reported that ENPP1 could hydrolyze extracellular cGAMP to generate AMP and GMP in various metastatic cancers, especially in triple-negative breast cancer (MDA-MB-231 cell line) [21]. The produced AMP was subsequently hydrolyzed into adenosine by CD73 to facilitate cancer-cell migration and extend valid immune suppressive effects [22]. Extracellular cGAMP hydrolysis by ENPP1 transformed an immune stimulatory pathway into an immune-suppressive mechanism that promoted tumor progression. Therefore, finding highly effective ENPP1 inhibitors is a potential therapeutic in metastatic breast cancer.
A few nucleotide-derived ENPP1 inhibitors have been reported to date ( Figure 2). Sevigny et al. found α-borano-β,γ-metATP derivative, which showed high inhibitory activity against ENPP1 for the replacement of one oxygen atom in the α-phosphate by a borano atom [23]. However, nucleotide-based inhibitors had limited applicability due to their high acidity, resulting in poor peroral bioavailability. As a result, non-nucleotidebased inhibitors might be expected to be ideal lead structures for the development of ENPP1 inhibitors as drugs. Langer et al. concluded that isoquinoline derivatives showed inhibitory activity against ENPP1, whose best inhibitor showed an IC 50 value of 360 nM [24]. Meanwhile, they also reported a series of thiadiazolopyrimidone derivatives as ENPP1 inhibitors, the best inhibitor displaying an IC 50 value of 310 nM [25]. Li et al. reported that quinazoline derivatives showed the most potent ENPP1 inhibitors. Among these inhibitors, STF-1084 displayed a K i value of 33 nM against ENPP1, which delays tumor growth in cancer models owing to an increase in the level of endogenous cGAMP [16,17]. based inhibitors might be expected to be ideal lead structures for the development of ENPP1 inhibitors as drugs. Langer et al. concluded that isoquinoline derivatives showed inhibitory activity against ENPP1, whose best inhibitor showed an IC50 value of 360 nM [24]. Meanwhile, they also reported a series of thiadiazolopyrimidone derivatives as ENPP1 inhibitors, the best inhibitor displaying an IC50 value of 310 nM [25]. Li et al. reported that quinazoline derivatives showed the most potent ENPP1 inhibitors. Among these inhibitors, STF-1084 displayed a Ki value of 33 nM against ENPP1, which delays tumor growth in cancer models owing to an increase in the level of endogenous cGAMP [16,17]. As a key component in the early stage of drug discovery, structure-based virtual screening comprised different in silico methodologies to filter a chemical compound library using the three-dimensional structure of the molecular target receptor, which could provide potential candidates for further experimental analysis [26]. CDOCKER is an implementation of a CHARMm-based docking tool using a rigid receptor [27,28], which has attracted increased attention to structure-based virtual screening owing to the advantages of high-precision docking. In this work, we report the discovery of eight ENPP1 inhibitors containing the quinazolin-4(3H)-one scaffold using a structure-based virtual screening strategy, which was carried out by using the CDOCKER protocol with Discovery Studio 2019. However, quinazolin-4(3H)-one is a common heterocyclic alkaloid ring system in medicinal chemistry and was found to possess prominent activity against many cancer cell lines, especially breast cancer and ovarian carcinoma [29,30]. Additionally, quinazolin-4(3H)-one derivatives have been reported to coordinate the catalytic zinc in the active site of metalloenzyme, which was thought to affect the biological role of enzymes [31,32]. Therefore, a series of novel quinazolin-4(3H)-one derivatives were synthesized and characterized by NMR and HRMS. Among these targeted compounds, the most active compound 7-fluoro-2-(((5-methoxy-1H-imidazo [4,5-b]pyri-dine-2-yl)thio)methyl)quinazolin-4(3H)-one(4e) exhibited excellent inhibitory activity at the molecular and cellular levels, with IC50 values of 0.188 μM and 0.732 μM against the ENPP1, respectively. Furthermore, compound 4e was investigated in cytotoxicity studies and pharmacological research in vivo, indicating that compound 4e can be used as a novel drug candidate targeting ENPP1 for tumor immunotherapy. As a key component in the early stage of drug discovery, structure-based virtual screening comprised different in silico methodologies to filter a chemical compound library using the three-dimensional structure of the molecular target receptor, which could provide potential candidates for further experimental analysis [26]. CDOCKER is an implementation of a CHARMm-based docking tool using a rigid receptor [27,28], which has attracted increased attention to structure-based virtual screening owing to the advantages of high-precision docking. In this work, we report the discovery of eight ENPP1 inhibitors containing the quinazolin-4(3H)-one scaffold using a structure-based virtual screening strategy, which was carried out by using the CDOCKER protocol with Discovery Studio 2019. However, quinazolin-4(3H)-one is a common heterocyclic alkaloid ring system in medicinal chemistry and was found to possess prominent activity against many cancer cell lines, especially breast cancer and ovarian carcinoma [29,30]. Additionally, quinazolin-4(3H)-one derivatives have been reported to coordinate the catalytic zinc in the active site of metalloenzyme, which was thought to affect the biological role of enzymes [31,32]. Therefore, a series of novel quinazolin-4(3H)-one derivatives were synthesized and characterized by NMR and HRMS. Among these targeted compounds, the most active compound 7-fluoro-2-(((5-methoxy-1H-imidazo [4,5-b]pyri-dine-2-yl)thio)methyl)quinazolin-4(3H)-one(4e) exhibited excellent inhibitory activity at the molecular and cellular levels, with IC 50 values of 0.188 µM and 0.732 µM against the ENPP1, respectively. Furthermore, compound 4e was investigated in cytotoxicity studies and pharmacological research in vivo, indicating that compound 4e can be used as a novel drug candidate targeting ENPP1 for tumor immunotherapy.

Structure-Based Virtual Screening of Novel Hit ENPP1 Inhibitors
The virtual screening process is schematically shown in Figure 3. Virtual screening was performed using the CDOCKER module of Discovery Studio 2019 [33]. The chemical databases comprised the in-house database (80 compounds). The crystal structure of the ENPP1 in complex with AMP (PDB entry 6WFJ) was used for the docking studies, and the ligand (AMP) binding position was utilized as the active site for docking. Next, virtual screening was performed by docking all the prepared ligands at the defined active site using CDOCKER. The RMSD threshold was set to 0.5 Å to ensure that the docking conformation was as diverse as possible [34]. Based on the -CDOCKER_ ENERGY values (Table 1), all the docked poses were ranked and grouped. The top eight compounds were selected from the ranked compounds by the values (the structure of candidate compounds is shown in Figure 4). Clearly, compound 4d obtained the highest -CDOCKER_ENERGY values (39.904 kcal/mol), indicating that it formed strong-affinity interactions with the ENPP1. However, we elected to synthesize these candidate compounds to carry out biochemical assays.
was performed using the CDOCKER module of Discovery Studio 2019 [33]. The chemic databases comprised the in-house database (80 compounds). The crystal structure of th ENPP1 in complex with AMP (PDB entry 6WFJ) was used for the docking studies, an the ligand (AMP) binding position was utilized as the active site for docking. Next, virtu screening was performed by docking all the prepared ligands at the defined active si using CDOCKER. The RMSD threshold was set to 0.5 Å to ensure that the docking co formation was as diverse as possible [34]. Based on the -CDOCKER_ ENERGY values (T ble 1), all the docked poses were ranked and grouped. The top eight compounds we selected from the ranked compounds by the values (the structure of candidate compound is shown in Figure 4). Clearly, compound 4d obtained the highest -CDOCKER_ENERG values (39.904 kcal/mol), indicating that it formed strong-affinity interactions with th ENPP1. However, we elected to synthesize these candidate compounds to carry out bi chemical assays.   databases comprised the in-house database (80 compounds). The crystal structure of the ENPP1 in complex with AMP (PDB entry 6WFJ) was used for the docking studies, and the ligand (AMP) binding position was utilized as the active site for docking. Next, virtual screening was performed by docking all the prepared ligands at the defined active site using CDOCKER. The RMSD threshold was set to 0.5 Å to ensure that the docking conformation was as diverse as possible [34]. Based on the -CDOCKER_ ENERGY values (Table 1), all the docked poses were ranked and grouped. The top eight compounds were selected from the ranked compounds by the values (the structure of candidate compounds is shown in Figure 4). Clearly, compound 4d obtained the highest -CDOCKER_ENERGY values (39.904 kcal/mol), indicating that it formed strong-affinity interactions with the ENPP1. However, we elected to synthesize these candidate compounds to carry out biochemical assays.     [35]. Next, commercially available 5-methoxy-1H-imidazo [4,5-b]pyridine-2-thiol (compound 3) was treated with compound 2a-h and sodium hydroxide in methanol to produce the final quinazolin-4(3H)-one derivatives 4a-h in acceptable yields [36].

Inhibition of ENPP1 Enzyme Activity by Compound 4a-h
An ENPP1 enzymatic inhibitory activity assay was carried out to evaluate the bio tivity of the candidate compounds. As shown in Table 2, these compounds exhibited inhibition of ENPP1 activity of over 50% at a concentration of 10 μM, which verified t rationality of the molecular docking results. It was obvious that quinazolin-4(3H)-one d rivatives displayed inhibitory potency against ENPP1. Herein, we speculate that the c bonyl O atom of the quinazolin-4(3H)-one group might chelate the catalytic zinc ions the active site to reduce phosphodiesterase activity [37]. According to the results of t enzymatic inhibition, compounds 4d and 4e displayed inhibition of over 90%, with I values of 0.694 μM and 0.188 μM, respectively ( Figure 5A), suggesting that 7-F and CO2CH3 were the optimal substitutions on the quinazolin-4(3H)-one in comparison w the other inhibitors. When the 7-F and 7-CO2CH3 on the quinazolin-4(3H)-one scaffo were replaced by a Cl atom (4f), the ENPP1 activity decreased distinctly due to the absen of hydrogen bonds. Furthermore, compound 4h bearing 8-Br had slightly decreas ENPP1 activity compared with compound 4g bearing 8-CH3, indicating that steric h drance was critical for the ENPP1 inhibitory activity. In a word, compounds 4d and exhibited considerable potency against ENPP1 in enzymatic assays and were chosen further study to discover the most potent ENPP1 inhibitor.

Inhibition of ENPP1 Enzyme Activity by Compound 4a-h
An ENPP1 enzymatic inhibitory activity assay was carried out to evaluate the bioactivity of the candidate compounds. As shown in Table 2, these compounds exhibited an inhibition of ENPP1 activity of over 50% at a concentration of 10 µM, which verified the rationality of the molecular docking results. It was obvious that quinazolin-4(3H)-one derivatives displayed inhibitory potency against ENPP1. Herein, we speculate that the carbonyl O atom of the quinazolin-4(3H)-one group might chelate the catalytic zinc ions in the active site to reduce phosphodiesterase activity [37]. According to the results of the enzymatic inhibition, compounds 4d and 4e displayed inhibition of over 90%, with IC 50 values of 0.694 µM and 0.188 µM, respectively ( Figure 5A), suggesting that 7-F and 7-CO 2 CH 3 were the optimal substitutions on the quinazolin-4(3H)-one in comparison with the other inhibitors. When the 7-F and 7-CO 2 CH 3 on the quinazolin-4(3H)-one scaffold were replaced by a Cl atom (4f), the ENPP1 activity decreased distinctly due to the absence of hydrogen bonds. Furthermore, compound 4h bearing 8-Br had slightly decreased ENPP1 activity compared with compound 4g bearing 8-CH 3 , indicating that steric hindrance was critical for the ENPP1 inhibitory activity. In a word, compounds 4d and 4e exhibited considerable potency against ENPP1 in enzymatic assays and were chosen for further study to discover the most potent ENPP1 inhibitor. The data are presented as the mean ± SD with three independent assays. 72.7 4h 55.5 The data are presented as the mean ± SD with three independent assays.

Cellular ENPP1 Inhibitory Activity by Compound 4a-h
As mentioned above, the ENPP1 was overexpressed in the MDA-MB-231 cell line (TNBC), which facilitated cancer-cell migration by hydrolyzing extracellular cGAMP [21]. In 2018, Nakagawa et al. also reported that MDA-MB-231 cells highly expressed ENPP1 [38]. To our knowledge, there have been no reports on the cellular inhibitory activity of tENPP1 inhibitors in the literature. Herein, we decided to explore the cellular ENPP1 inhibitory activity of compounds 4d and 4e in the MDA-MB-231 cell line. As shown in Figure 5B, compounds 4d and 4e displayed IC50 values of 3.335 μM and 0.732 μM against the ENPP1 in the MDA-MB-231 cell line, respectively, suggesting that these two compounds can be applied to cancer-cell migration by preventing cGAMP from degrading due to ENPP1. However, the results revealed that compound 4e had the best inhibition in the MDA-MB-231 cell line, exhibiting IC50 values of <1 μM against the ENPP1. To sum up, the analyses above led to the discovery of a potent ENPP1 inhibitor (4e) that exhibited higher potency at the molecular and cellular levels.

Binding-Mode Analysis
The binding mode of compound 4e within the active site of the ENPP1 predicted by molecular docking is illustrated in Figure 6A. It was noted that compound 4e was inserted into the catalytic tunnel in a suitable conformation and coordinated with the active-site Zn ion of the ENPP1. The F atom of compound 4e was predicted to engage in a hydrogen bond with Lys-278, which contributed to the increase in the binding affinity. This might explain why compound 4e showed better inhibitory activity compared to the other candidate compounds. The results of the molecular modeling also revealed that a hydrogen

Cellular ENPP1 Inhibitory Activity by Compound 4a-h
As mentioned above, the ENPP1 was overexpressed in the MDA-MB-231 cell line (TNBC), which facilitated cancer-cell migration by hydrolyzing extracellular cGAMP [21]. In 2018, Nakagawa et al. also reported that MDA-MB-231 cells highly expressed ENPP1 [38]. To our knowledge, there have been no reports on the cellular inhibitory activity of tENPP1 inhibitors in the literature. Herein, we decided to explore the cellular ENPP1 inhibitory activity of compounds 4d and 4e in the MDA-MB-231 cell line. As shown in Figure 5B, compounds 4d and 4e displayed IC 50 values of 3.335 µM and 0.732 µM against the ENPP1 in the MDA-MB-231 cell line, respectively, suggesting that these two compounds can be applied to cancer-cell migration by preventing cGAMP from degrading due to ENPP1. However, the results revealed that compound 4e had the best inhibition in the MDA-MB-231 cell line, exhibiting IC 50 values of <1 µM against the ENPP1. To sum up, the analyses above led to the discovery of a potent ENPP1 inhibitor (4e) that exhibited higher potency at the molecular and cellular levels.

Binding-Mode Analysis
The binding mode of compound 4e within the active site of the ENPP1 predicted by molecular docking is illustrated in Figure 6A. It was noted that compound 4e was inserted into the catalytic tunnel in a suitable conformation and coordinated with the active-site Zn ion of the ENPP1. The F atom of compound 4e was predicted to engage in a hydrogen bond with Lys-278, which contributed to the increase in the binding affinity. This might explain why compound 4e showed better inhibitory activity compared to the other candidate compounds. The results of the molecular modeling also revealed that a hydrogen bond was predicted to exist between the methoxyl of compound 4e and Trp-322 in the hydrophobic pocket. Another important hydrogen bond was formed between the NH of the imidazole group and Tyr-371. In addition, the binding mode of compound 4e illustrated π-π stacking interactions between the bicyclic imidazopyridine core and Tyr-340. Altogether, the potency of compound 4e was driven by its zinc-ion binding, π-π stacking interactions with residues in the hydrophobic pocket, and hydrogen bonds with the active side resides of the ENPP1. tive-site Zn1 ion of ENPP1 with a distance of 2.2 Å and formed H-bond interactions with Asp376 and His535 at distances of 3.1 Å and 3.3 Å, respectively. In conclusion, the quinazolin-4(3H)-one of compound 4e formed a solid network of interactions with the zinc and Asp376/His380/His535, which was thought to affect the hydrolysis role of the ENPP1. However, the docking results indicated that ENPP1 inhibitors with a zinc-binding quinazolin-4(3H)-one group exhibit good potential.

Cell Viability Assays
Several studies have also reported that the inhibitory activity of the ENPP1 inhibitors could transfer to tumor-cell lines [41][42][43]. As mentioned above, upregulated ENPP1 expression has been associated with metastatic cancer cells, especially in metastatic breast cancer; thus, ENPP1 inhibitors might be useful for the treatment of metastatic breast cancer [16]. However, the overexpression of wild-type ENPP1 could promote cancer migration and metastasis by hydrolyzing the immunotransmitter cGAMP in the 4T1 model [21]. Therefore, breast-cancer-cell (4T1) proliferation assays with compound 4e were carried out using a standard CCK8 assay. As shown in Table 3, compound 4e showed the same cytotoxic behavior towards the 4T1 cell line as cisplatin, presenting an IC50 value of 2.99 μM. However, compound 4e exerted a nonsignificant cytotoxic effect on the normal cell According to the structural information of the ENPP1, two Zn 2+ ions (Zn1 and Zn2) were tightly bound in the active site among six conserved Asp/His residues (Asp376/ His380/ His535 and Asp218/Asp423/His424), which played a critical role in the catalytic process [20,39,40]. In 2020, Li et al. reported the structure of ENPP1 in a complex with STF-1084 using X-ray crystallography [16]. They found that STF-1084 occupied the lipidbinding hydrophobic pocket and formed extensive interactions with two zinc ions and six conserved Asp/His residues in the catalytic site. Therefore, we elected to explore the interactions between compound 4e and the zinc ions in the active site in more detail. As shown in Figure 6B, the carbonyl O atom of compound 4e was coordinated to the active-site Zn1 ion of ENPP1 with a distance of 2.2 Å and formed H-bond interactions with Asp376 and His535 at distances of 3.1 Å and 3.3 Å, respectively. In conclusion, the quinazolin-4(3H)-one of compound 4e formed a solid network of interactions with the zinc and Asp376/His380/His535, which was thought to affect the hydrolysis role of the ENPP1. However, the docking results indicated that ENPP1 inhibitors with a zinc-binding quinazolin-4(3H)-one group exhibit good potential.

Cell Viability Assays
Several studies have also reported that the inhibitory activity of the ENPP1 inhibitors could transfer to tumor-cell lines [41][42][43]. As mentioned above, upregulated ENPP1 expression has been associated with metastatic cancer cells, especially in metastatic breast cancer; thus, ENPP1 inhibitors might be useful for the treatment of metastatic breast cancer [16]. However, the overexpression of wild-type ENPP1 could promote cancer migration and metastasis by hydrolyzing the immunotransmitter cGAMP in the 4T1 model [21]. Therefore, breast-cancer-cell (4T1) proliferation assays with compound 4e were carried out using a standard CCK8 assay. As shown in Table 3, compound 4e showed the same cytotoxic behavior towards the 4T1 cell line as cisplatin, presenting an IC 50 value of 2.99 µM. However, compound 4e exerted a nonsignificant cytotoxic effect on the normal cell lines (LO2 and 293T) at the highest concentration (50 µM) in the experiment, indicating that compound 4e did not exhibit obvious toxicity. Thus, compound 4e was considered suitable for anti-tumor research in vivo, especially in the 4T1 breast cancer model. Taking the biological data together, compound 4e displayed superior selectivity towards 4T1 breast cancer cells than towards normal cells (LO2 and 293T) and might be used in the treatment of metastatic breast cancer. The data are presented as the mean ± SD with three independent assays.

Molecular Mechanisms of Activity for Compound 4e
Li et al. found that ENPP1 inhibitors could increase the half-life of endogenous cGAMP to enhance the natural anti-tumor response. Furthermore, they also reported that the upregulation of IFN-β meant that the cGAS-STING-TBK1-IRF3 signaling pathway was activated to stimulate the innate immune response [17]. In order to explore the pharmacological mechanisms of compound 4e, we opted to measure the IFN-β from the separated serum by using the ELISA method after the intravenous administration of free cGAMP and cGAMPcompound 4e in mice, respectively. Compared with the model mice, the upregulation of the IFN-β in the free-cGAMP-treated groups suggested that the cGAMP served as the second message to bind STING and, thus, activated the cGAS-STING-TBK1-IRF3 signaling pathway to stimulate an innate immune response (Figure 7). Additionally, it was found that the expression level of the IFN-β in the cGAMP-compound 4e-treated group was higher than those in the free-cGAMP-treated group, indicating that compound 4e could protect cGAMP from degradation to stimulate a more potent innate immune response. As a consequence, compound 4e, as an ENPP1 inhibitor, can be applied in tumor immunotherapy by preventing cGAMP from hydrolyzing. Altogether, compound 4e could stimulate a more potent innate immune response while simultaneously reducing the extracellular levels of immuno-suppressive adenosine that promote metastatic progression.
suitable for anti-tumor research in vivo, especially in the 4T1 breast cancer model. Taking the biological data together, compound 4e displayed superior selectivity towards 4T1 breast cancer cells than towards normal cells (LO2 and 293T) and might be used in the treatment of metastatic breast cancer. The data are presented as the mean ± SD with three independent assays.

Molecular Mechanisms of Activity for Compound 4e
Li et al. found that ENPP1 inhibitors could increase the half-life of endogenous cGAMP to enhance the natural anti-tumor response. Furthermore, they also reported that the upregulation of IFN-β meant that the cGAS-STING-TBK1-IRF3 signaling pathway was activated to stimulate the innate immune response [17]. In order to explore the pharmacological mechanisms of compound 4e, we opted to measure the IFN-β from the separated serum by using the ELISA method after the intravenous administration of free cGAMP and cGAMP-compound 4e in mice, respectively. Compared with the model mice, the upregulation of the IFN-β in the free-cGAMP-treated groups suggested that the cGAMP served as the second message to bind STING and, thus, activated the cGAS-STING-TBK1-IRF3 signaling pathway to stimulate an innate immune response (Figure 7). Additionally, it was found that the expression level of the IFN-β in the cGAMP-compound 4e-treated group was higher than those in the free-cGAMP-treated group, indicating that compound 4e could protect cGAMP from degradation to stimulate a more potent innate immune response. As a consequence, compound 4e, as an ENPP1 inhibitor, can be applied in tumor immunotherapy by preventing cGAMP from hydrolyzing. Altogether, compound 4e could stimulate a more potent innate immune response while simultaneously reducing the extracellular levels of immuno-suppressive adenosine that promote metastatic progression.

Materials and Reagents
The hENPP1-HIS DNA gene was cloned by Shanghai Generay Biotech Co., Ltd. (Shanghai, China). Cancer cell lines were purchased from the Cell Bank of Chinese Academy of Sciences (Shanghai, China). Thymidine 5 monophosphate p-nitrophenyl ester (pNP-TMP) was purchased from Sigma-Aldrich (Shanghai, China). All commercially obtained reagents and solvents were used directly, without further purification, unless otherwise noted. 1 H NMR (400 MHz) and 13 C NMR (101 MHz) spectra were taken on a Bruker AV-400 MHz spectrometer, and chemical shifts were reported in ppm downstream of internal Me 4 Si. High-resolution mass spectra (HRMS) were recorded on a VG ZAB-HS mass spectrometer under electron-spray ionization (ESI). Thin-layer chromatography (TLC) was conducted on aluminum-sheet silica gel Merck 60F254. The spots were visualized using ultraviolet light.

Structure-Based Virtual Screening
In this investigation, the docking method, CDOCKER, was applied on the Discovery Studio 2019 platform. The receptor protein (PDB entry 6WFJ) was prepared by removing the water and some other co-crystallized small molecules, adding hydrogen atoms to the protein, and assigning CHARMm force field in Discovery Studio 2019 software package. After the protein preparation, the binding site of the protein was defined based on volume occupied by the ligand AMP already positioned in the active site. Subsequently, the prepared ligands (according to ligand preparation protocol) from in-house database were docked into the active site of ENPP1. The RMSD threshold was set to 0.5 Å to ensure that the docking conformation was as diverse as possible. Based on the -DOCKER-ENERGY score, all the docked poses were ranked and grouped.

Expression and Purification of Recombinant Human ENPP1
In total, 1 mg hENPP1-HIS DNA using polyethylenimine at a DNA:PEI ratio of 1:3 was transfected into 1 L of 293-F cells in 293 freestyle medium. The medium was harvested after five days and filter-sterilized. Net, the protein was purified on a 10-milliliter Ni affinity column at a flow rate of 10 mL min −1 , washed with 10 column volumes of buffer 1 (350 mM NaCl with 50 mM Tris pH 8.0) with 20 mM imidazole, and then eluted with 250 mM imidazole in buffer 1. Approximately 1 mg of purified protein was obtained from the 1-liter culture. The quality of the protein purification was validated by SDS-PAGE analysis (see Figure S1) [40]. 2-(Chloromethyl)-6-methoxyquinazolin-4(3H)-one (2b). The desired compound was prepared from methyl 2-amino-5-methoxybenzoate 1b (0.724 g, 4 mmol, 1eq) and chloroacetonitrile (0.254 mL, 4 mmol, 1eq) using the procedure described for compound 2a in 56% yield as a brown solid. 1  were calculated by plotting three independent experiments using the program Prism 5.0. Incubation and operation conditions remained the same as those described above [37,44].

Cellular ENPP1 Enzymatic Inhibition Assays
MDA-MB-231 (1.0 × 10 4 cells/well) were grown on 96-well microplates (IWAKI) overnight in the culture medium (100 µL/well) in a humidified incubator containing 5% CO 2 in air at 37 • C for 24 h. The plates were washed with D-Hanks (100 µL) twice, and fresh D-Hanks (80 µL) was added. 5% DMSO or test compound (10 µL) was added to each well, and eventually pNP-TMP (2 mM, 10 µL, final 200 µM) was added. After 4 h incubation at 37 • C, the amounts of released p-nitrophenolate were measured at 405 nm. The IC 50 values of test compounds were calculated by plotting three independent experiments using the program Prism 5.0 [38].

Molecular Docking Study
Docking simulations were performed to study the binding pattern of compound 4e to ENPP1 pocket. ENPP1 (PDB entry: 6WFJ) were downloaded from Protein Data Bank (PDB). The Docking simulation was performed using by CDOCKER. Hydrogen atoms were added to the proteins by using Discovery Studio 2019 (Accelrys Inc., San Diego, CA, USA). Charmm force field was assigned. The binding site was defined as a sphere containing the residues that remained within 10 Å of the ligand, an area large enough to cover the ligand-binding region at the active site. The docking results showed the optimized molecular docking model with the receptor proteins and gave a score. The image was created using PyMOL.

Cell Culture and Proliferation Assays
In total, 1 × 10 4 cells (4T1, LO2 and 293T) were plated in 96-well flat-bottom plates in a final volume of 200 µL DMEM supplemented with 10% FBS and cultured for 24 h. Next, the culture was replaced by fresh DMEM medium (200 µL) and the compound (10 µL) was added. After forty-eight hours, the medium was replaced by PBS (180 µL) mixture with CCK8 solution (20 µL) in each well. Cell viability was evaluated by testing the OD 450 nm. The data were processed by GraphPad Prism 7.0 [45,46].

Pharmacological Research on ENPP1 Inhibitors in Mice
The animal studies were conducted with the approval of the Animal Care and Use Committee at Fudan University. BALB/c nude mice (6−8 weeks) were randomized into three groups (n = 3, with a total tumor number of 9). Free cGAMP (at dose of 5 mg/kg) and cGAMP-compound 4e (at doses of 5 mg/kg cGAMP and 0.5 mg/kg compound 4e), respectively, were injected intravenously (i.v.) into mice via tail vein. After i.v. administration, mouse blood samples were collected from ophthalmic vein after 0.5 h and centrifuged at 500× g for 20 min. The concentration of IFN-β in the serum was determined by mouse ELISA kits, according to the instruction manual [47].

Conclusions
In this study, the structure-based docking strategy was used to screen for novel ENPP1 inhibitors. According to the docking results, eight candidate compounds containing the quinazolin-4(3H)-one scaffold were selected to carry out the bioassays. The recombinant ENPP1 was expressed and purified to evaluate the enzymatic inhibitory activity of the compounds. Of these compounds, compound 4d and compound 4e exhibited excellent inhibitory activity against the ENPP1, with IC 50 values of 0.694 µM and 0.188 µM, respectively. Additionally, two inhibitors also displayed remarkable inhibitory activity against the ENPP1 in the MDA-MB-231 cell line. However, compound 4e showed excellent inhibitory activity against the ENPP1 at the molecular and cellular levels, exhibiting IC 50 values of <1 µM. Through binding-mode analysis, compound 4e was predicted to form a coordination bond, a hydrogen bond, and hydrophobic interaction with the ENPP1, accounting for obvious enzymatic inhibitory activity at the molecular and cellular levels. On the other hand, compound 4e showed high cytotoxic activity against the 4T1 cell line and negligible cytotoxic effects against the normal cell lines (LO2 and 293T), indicating that compound 4e can potentially be employed in metastatic breast cancer therapy. Finally, we found that since compound 4e could enhance the expression levels of IFN-β in vivo by preventing cGAMP from degrading due to ENPP1, it might be applied in tumor immunotherapy.
The inhibition of ENPP1 differs from the pharmacologic activation of STING in a few crucial ways, even though they all elicit anti-tumor immunity through the cGAS-STING-TBK1-IRF3 signaling pathway. On one hand, ENPP1 inhibitors are capable of maintaining cellular endogenous cGAMP levels to inhibit cancer migration and metastasis. On the other hand, ENPP1 inhibitors have a larger therapeutic window, since the systemic administration of STING agonists may cause potential side effects. In summary, compound 4e, as a novel drug candidate targeting ENPP1, could regulate endogenous cGAMP and, thus, deserves further anti-tumor research in vivo. In addition, the synergistic effects of the combination of ENPP1 inhibitors and the STING agonist, cGAMP, should be studied for its therapeutic potential in cancer.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding authors.