2-Amino-2,3-dihydro-1H-indene-5-carboxamide-Based Discoidin Domain Receptor 1 (DDR1) Inhibitors: Design, Synthesis, and in Vivo Antipancreatic Cancer Efficacy

A series of 2-amino-2,3-dihydro-1H-indene-5-carboxamides were designed and synthesized as new selective discoidin domain receptor 1 (DDR1) inhibitors. One of the representative compounds, 7f, bound with DDR1 with a Kd value of 5.9 nM and suppressed the kinase activity with an half-maximal (50%) inhibitory concentration value of 14.9 nM. 7f potently inhibited collagen-induced DDR1 signaling and epithelial–mesenchymal transition, dose-dependently suppressed colony formation of pancreatic cancer cells, and exhibited promising in vivo therapeutic efficacy in orthotopic mouse models of pancreatic cancer.


■ INTRODUCTION
Pancreatic cancer, also known as the "king of cancer" with a 5 year survival rate less than 7%, is lacking effective therapies. 1 Intrinsic and/or acquired chemoresistance to the first-line drug gemcitabine is a major reason for the poor prognosis. Collective studies suggested that desmoplasia and tumor microenvironment (TME) are the key contributors to chemoresistance of pancreatic cancer patients. Desmoplasia (or desmoplastic reaction) is a prominent pathological characteristic of pancreatic cancer, resulting from the rapid expansion of cancer-associated fibroblasts and increased deposition of extracellular matrix (ECM) components. Numerous studies have shown that ECM is heavily involved in pancreatic cancer development, immune evasion, and therapy resistance. Therefore, stroma-targeting therapy becomes an attractive strategy to improve therapeutic response. 2 Collagens are the most abundant components in ECM. 3 Biological functions of collagens are mainly mediated by two types of receptors, integrins and discoidin domain receptors (DDRs). 4,5 DDRs belong to the receptor tyrosine kinase (RTK) family. Two members of DDRs (DDR1 and DDR2) have been identified to date. 6 DDR1 is predominantly expressed in epithelial cells, whereas DDR2 is typically found in cells of connective tissues. Dysregulation of DDR1 is frequently detected in a variety of human cancers and involved in several key cellular processes, such as cell differentiation, proliferation, adhesion, migration, and invasion. 7,8 Studies have shown that DDR1 overexpression mediates prosurvival signals and metastasis in breast cancer and gastric cancer and is also involved in the recurrence of certain types of cancers. 9−11 DDR1 has also been demonstrated to contribute to epithelial− mesenchymal transition (EMT) in pancreatic cancer. 12,13 Pharmacological inhibition of DDR1 by a selective DDR1 inhibitor 7rh (compound 1, Figure 1) successfully slowed tumor progression and enhanced chemosensitivity to standardof-care pancreatic cancer regimens. 14 Therefore, DDR1 is being considered as a novel molecular target for drug discovery against pancreatic cancer.
A number of DDR inhibitors have been reported ( Figure  1), 15−21 among which compound 1 (7rh) 16 represents one of the first disclosed selective DDR1 inhibitors. This compound demonstrated promising therapeutic potential in a variety of human cancer models including pancreatic cancer, 14 nonsmall cell lung cancer, 22 and gastric carcinoma. 10 Most recently, tetrahydroisoquinoline-7-carboxamide derivatives were also designed and synthesized as highly selective DDR1 inhibitors exhibiting in vivo efficacy in mouse models of inflammationmediated pulmonary fibrosis and acute lung injury. 21,23 Compound 6 represents one of the most selective DDR1 inhibitors to date, which exhibited an half-maximal (50%) inhibitory concentration (IC 50 ) value of 15.4 nM against DDR1 kinase, whereas it is significantly less potent for the majority of a panel of 403 other kinases. The major "offtargets" include tropomyosin receptor kinases B and C (TrkB and TrkC, respectively) with corresponding binding affinities (K d values) of 22 and 18 nM, respectively. Further kinase assay determination confirmed that this molecule inhibited TrkA, B, and C, with IC 50 values of 68.4, 36.8, and 30.2 nM, respectively. Trks are a family of receptor tyrosine kinases (RTKs), which are activated by neurotrophin hormones and regulate synaptic plasticity and strength in the mammalian nervous system. 24 Lack of Trks significantly affects the development of central and peripheral nervous systems. 25,26 For instance, the population of corneal sensory neurons is markedly depleted in TrkA (−/−) mice, animals lacking TrkB in parvalbumin-positive cells displayed sexually dimorphic behavioral phenotypes, whereas TrkC knockout mice exhibited profound deficiencies in CNS glial cells. 27 Thus, "off-targeting" Trks may raise some concerns about the potential neurotoxicity issues of the previously reported DDR1 inhibitors. With the aim to improve the target specificity, a series of 2amino-2,3-dihydro-1H-indene-5-carboxamide derivatives were designed and synthesized as new Trks-sparing DDR1 inhibitors. Moreover, the potential therapeutic effects on pancreatic cancer were also investigated by utilizing in vitro and in vivo models ( Figure 2).

■ RESULTS AND DISCUSSION
Compound 6 is a selective DDR1 inhibitor designed and synthesized in our laboratory. However, this compound displayed similar off-target potency against TrkA, TrkB, and TrkC with IC 50 values of 68.4, 36.8, and 30.2 nM, respectively (Table 1). Off-target inhibition against Trks may raise some concern about the potential neurotoxicity issues of the molecule. Aiming to improve the target specificity, we first

Journal of Medicinal Chemistry
Article docked compound 6 into the adenosine triphosphate (ATP)binding site of TrkC [Protein Data Bank (PDB) ID: 3V5Q]. 28 It was shown that the compound could bind well to TrkC with a Asp-Phe-Gly (DFG)-out conformation, which was similar to the binding mode with DDR1 ( Figure 3A,B). 21 The only small difference between binding modes of 6 with DDR1 and TrkC is that the methyl group on tetrahydroisoquinoline occupied a small hydrophobic pocket formed by Val552, Ala570, Lys572, and Phe617 to achieve some hydrophobic interactions in TrkC. Thus, it was predicted that diminishing this interaction might decrease the binding with Trks and improve DDR1 target selectivity of the molecule. Based on this hypothesis, an isoindoline scaffold was first utilized to replace the 1-methyl tetrahydroisoquinoline in 6 to generate compounds 7a and 7b. Disappointingly, both 7a and 7b lost all of the DDR1 inhibitory activity, although the modification indeed abolished the off-target inhibition against Trks (Table 1). Further computational investigation demonstrated that 7a failed to fit nicely into the DDR1 ATP binding pocket (PDB ID: 5FDP, data not shown), 21 whereas compound 7b could form weak hydrogen bonds (HBs) with the hinge residue Met704, but there was no other interaction between the isoindoline scaffold and DDR1 ( Figure S1). The computational study also indicated that the methylene moiety of 7b is accessible to the gatekeeper Thr701 residue, and it could be replaced by a hydrogen bond donor moiety to form a hydrogen bond interaction with Thr701 to potentially improve the DDR1 inhibitory activity. To validate this hypothesis, 2-amino-2,3dihydro-1H-indene-5-carboxamide enantiomers 7c and 7d were designed and synthesized by moving the nitrogen atom out of the original isoindoline ring. It was also predicted that the R-isomer 7c might achieve a more favorable binding with DDR1 than that of the S-isomer 7d to form a hydrogen bond with the gatekeeper residue Thr701 ( Figure S2A,B). Indeed, kinase inhibition determination showed that compound 7c potently inhibited DDR1 with an IC 50 value of 5.6 nM, whereas the corresponding value for 7d was 42.5 nM. Compound 7c exhibited a similar suppressive activity against DDR2 with an IC 50 value of 21.5 nM. It was also noteworthy that 7c almost totally abolished the inhibitory potencies against Trks with IC 50 values greater than 1000 nM. However, introduction of a methyl group at the R 3 -position (7e) caused an approximately 129-fold potency loss, which might be due to the steric hindrance to interfere with potential hydrogen bond formation.

Journal of Medicinal Chemistry
Article shown that the pyrimidinyl moiety of 7c would still be able to form a hydrogen bond (∼3.1 Å) with the key residue Met620 in TrkC ( Figure 3D), but the gatekeeper Phe617 of TrkC altered the three-dimensional (3D) structure of the ATP pocket relative to DDR1, which would both preclude a key hydrogen bond and additionally introduce a steric clash with 7c.
Aiming to improve DDR1 target specificity, further structural optimization was conducted. Superposition of DDR1 (PDB: 6HP9, gray) and a homology model of DDR2 (orange) 29 indicated that DDR1 possessed a small hydrophobic groove formed by Ile685, Thr701, and Asp703, whereas no such pocket existed in DDR2 because of its outward shifting of Glu93 and Ile76 ( Figure 4A). Based on this observation, it was hypothesized that a small hydrophobic substituent might be introduced at the nitrogen atom of 7c to improve the DDR1 selectivity. Thus, compounds 7f−h, in which a methyl, ethyl, or n-propyl group was introduced at the corresponding position, respectively, were designed and synthesized. It was shown that the methylated compound 7i indeed significantly decreased the DDR2 inhibitory potency with an IC 50 of 933 nM, although it also caused a 2.6-fold potency loss for DDR1 with an IC 50 value of 14.9 nM (Table  1). However, both 7g and 7h resulted in 26.8-and approximately 200-fold DDR1 potency decreases, respectively.
Computational investigation supported the idea that a methyl group (7f) could fit into a small hydrophobic pocket formed by Ile685, Thr701, and Asp702, whereas loss of a hydrogen bond with Thr701 would contribute to the DDR1 potency decrease ( Figure 4B).
Further structure−activity relationship investigation also revealed that the trifluoromethyl of 7f could be replaced by the ethyl (7j), isopropyl (7k), or tertiary butyl (7l) group without obviously affecting the DDR1 inhibitory potency. For instance, compounds 7j−l exhibited IC 50 values of 14.7, 15.0, and 14.4 nM, respectively. However, their potencies against DDR2 and Trks were improved, which made the compound less selective. A replacement of the trifluoromethyl with a methyl group (7i) caused an approximately 43-fold decrease in DDR1 potency. When the trifluoromethyl was replaced with a cyclohexyl group, the resulting compound 7m was also 5-fold less potent than the lead molecule 7f. Significantly, the amantadine derivative (7n) totally abolished its activity against DDR1 with an IC 50 value greater than 2.0 μM. Collectively, 7f represents one of the most potent and selective DDR1 inhibitors in the series for further biological investigation.
The binding affinity determination showed that 7f tightly bound to the ATP-binding sites of DDR1 with a K d value of 5.9 nM (DiscoverX, San Diego, CA). 30 Further target specificity of 7f was investigated by KINOMEscan profiling Table 1. In Vitro Kinase Inhibitory Activities of Compounds 7a−n against DDR1, DDR2, TrkA, TrkB, and TrkC a,b a DDR1 and DDR2 inhibition assays were performed using the LANCE ULTRA kinase assay. b TrkA/B/C activity was performed using the fluorescence resonance energy transfer (FRET)-based Z′-Lyte assay. The data are mean values from at least three independent experiments.

Journal of Medicinal Chemistry
Article with a panel of 468 kinases at 1.0 μM ( Figure 5A). The results indicated that 7f displayed an extraordinary target selectivity with S (10) and S (1) scores of 0.015 and 0.007, respectively (Table S3). The major off-targets (inhibition > 90%, ctrl% < 10) included the DDR2, ABL1, KIT, PDGFRB, and TrkC. The inhibitory activities (IC 50 ) of 7f against above-mentioned off-targets were further determined with our in-house kinase assays ( Figure 5B). It was shown that 7f displayed significantly less potencies to all of the off-target kinases.
To investigate the inhibitory effects of 7f on the biological functions of DDR1, we first examined the DDR1 signaling pathway with the drug treatment in a mouse pancreatic cancer cell line Pan02 ( Figure 6). It was found that 7f dosedependently inhibited collagen-induced activation of DDR1 and its downstream signaling proteins, for example, prolinerich tyrosine kinase 2 (PYK2) and pseudopodium-enriched atypical kinase 1 (PEAK1).
The activation of DDR1−PYK2−PEAK1 signaling has been shown to mediate the tumor-promoting functions of collagen, including an important process of EMT called cadherin switching, 31 in which cancer cells specifically upregulate the mesenchymal cell adhesion protein N-cadherin, whereas the epithelial cell adhesion protein E-cadherin may or may not be affected. As a result of N-cadherin upregulation, cancer cells become substantially more motile and aggressive. 32 Therefore, we performed immunofluorescence staining for E-cadherin and N-cadherin in Pan02 and another mouse pancreatic cancer cell line BMF-A3 derived from KPf C (Kras LSL-G12D ; Trp53 f lox/f lox ; Ptf1a Cre/+ ) mice, a genetically engineered mouse model of pancreatic cancer, to explore the effects of 7f on inhibiting DDR1-mediated cadherin switching ( Figure 7A). Consistently, 7f markedly inhibited cadherin switching in each cell line regardless of the different genetic background. Moreover, we also found that 7f significantly inhibited the migratory capability of the cells in a wound-healing assay ( Figure 7B).
Cell heterogeneity is a major cause of therapy resistance in most types of cancers. 33 To understand whether 7f had similar effects on inhibiting cadherin switching in different cancer cell populations derived from the same tumor, we performed a 3D culture experiment with two different cancer cell clones derived from a KPf C tumor (BMF-A3 and CT1A-C11). In the presence of ECM that mimics the in vivo tumor microenvironment. 34 We found that BMF-A3 represented a more epithelial phenotype, which remained as clusters under 3D culture, whereas CT1A-C11 was more mesenchymal and aggressive, with an elongated and fibroblastic morphology. However, we found that 7f strongly inhibited such a mesenchymal phenotype ( Figure 8A). We also examined the effect of 7f on DDR1-induced cadherin switching in the two cell lines by probing for E-cadherin and N-cadherin expressions in cell lysates. We found that although the two cell lines had different phenotypes in 3D culture, 7f inhibited the upregulation of N-cadherin similarly in a dose-dependent manner ( Figure 8B). This suggests that 7f may have effects on many cancer cell populations despite cellular heterogeneity.
In addition, we examined the effects of 7f on the tumorigenicity of pancreatic cancer cells using an in vitro colony formation assay. As shown in Figure 9A,B, 7f dosedependently inhibited colony formation significantly in BMF-A3 and Pan02 cells. However, the direct effect against proliferation of 7f seemed to be moderate, measured by cell proliferation in two-dimensional with BMF-A3 and Pan02 cells showing IC 50 s values of 4.26 and 11.92 μM, respectively ( Figure S4).
Given the effects of 7f on pancreatic cancer cells in vitro, we further studied its effects in vivo. First, we profiled the in vivo pharmacokinetics (PK) of 7f in Sprague-Dawley (SD) rats ( Table 2). 7f displayed an ideal PK profile with an oral area under curve (AUC) value of 80 535 μg/(L h), a T 1/2 value of

Journal of Medicinal Chemistry
Article 1.7 h, and an oral bioavailability of 89.9%, which provides the foundation for an in vivo dosing regimen.
We then established syngeneic models of pancreatic cancer by orthotopically implanting the BMF-A3 and Pan02 cells in C57BL/6 mice and evaluated the efficacy of 7f on pancreatic cancer progression. The animals were orally (po) administered vehicle or 7f [twice per day (bid), 25 and 50 mg(kg day)] for 3 weeks. In both tumor models, 7f inhibited the progression of the tumors significantly without obviously caused animal body weight loss (Table S4), highlighting the in vivo efficacy of the molecule in pancreatic cancer ( Figure 10).

■ CONCLUSIONS
In summary, a series of 2-amino-2,3-dihydro-1H-indene-5carboxamide derivatives were designed and synthesized as novel highly selective DDR1 inhibitors with a structure-based drug design method. Compound 7f strongly suppressed DDR1 with an IC 50 value of 14.9 nM, but it was significantly less potent against a panel of other 403 nonmutated kinases at 1.0 μM. 7f also potently inhibited the collagen-induced cadherin switching event induced by DDR1 and dose-dependently suppressed colony formation of pancreatic cancer cells. Moreover, 7f demonstrated good pharmacokinetic properties and promising therapeutic effect by oral administration in orthotopic syngenic models of pancreatic cancer. Notably, several somatic mutations of DDR1 had been clinically identified, but most of them are located out of the kinase domain. 35 Thus, 7f may achieve similar inhibition against these disease-related mutants. Extensive biological investigation is undergoing to validate 7f as a lead molecule for further development.
■ EXPERIMENTAL SECTION General Chemistry. Reagents and solvents were purchased from commercial sources and used directedly. Flash chromatography was performed using 300-mesh silica gel. Reactions were monitored by thin-layer chromatography using silica gel plates with fluorescence F 254 and UV light visualization. Low-resolution electrospray ionization mass spectrometry (ESI-MS) was performed on an Agilent 1200 highperformance liquid chromatography (HPLC)-mass selective detector mass spectrometer and high-resolution ESI-MS on an Applied Biosystems Q-STAR Elite ESI-LC-MS/MS mass spectrometer. 1 H NMR spectra were performed on a Bruker AV-400 spectrometer at 400 MHz or a Bruker AV-500 spectrometer at 500 MHz. 13 C NMR spectra were performed on a Bruker AV-500 spectrometer at 125 MHz. Coupling constants (J) were expressed in hertz (Hz). Chemical   a SD rats (male, three animals per group) weighted 180−220 g were used for the study.

[M + H] + .
tert-Butyl 5-Bromoisoindoline-2-carboxylate (35). To 5bromoisoindoline 34 (282 mg, 1.42 mmol) in 6 mL of THF were added (Boc) 2 O (1 mL, 4.3 mmol) and NaHCO 3 (360.8 mg, 4.29 mmol). The reaction mixture was stirred at room temperature overnight. The reaction was quenched by water and extracted three times with CH 2 Cl 2 . The combined organic phase was washed with brine, dried over Na 2 SO 4 , concentrated in vacuo, and purified by silica gel column chromatography to give 35 (238 mg, 56%). 1  Methyl 2-(Pyrimidin-5-ylmethyl)isoindoline-5-carboxylate (39). To methyl isoindoline-5-carboxylate hydrochloride 37 (90 mg, 0.42 mmol) in DMF were added 5-chloromethylpyrimidine (65 mg, 0.41 mmol) and potassium carbonate (145 mg, 1.05 mmol). The reaction was stirred overnight at 50°C. Then, water (10 mL) was added to the reaction mixture and extracted with ethyl acetate three times. The combined organic phase was dried over Na 2 SO 4 , concentrated in vacuo, and purified by column chromatography to give 39 (44 mg, 41%). 1  The effects of compounds on the kinases DDR1 and DDR2 were assessed using a LanthaScreen Eu kinase activity assay technology (Invitrogen). Kinase reactions are performed in a 10 μL volume in low-volume 384-well plates. The kinases in the reaction buffer consist of 50 mM HEPES pH 7.5, 0.01% Brij-35, 10 mM MgCl 2 , and 1 mM EGTA; the concentration of the fluorescein−poly-GAT substrate (Invitrogen) in the assay is 100 nM. Kinase reactions were initiated with the addition of 100 nM ATP in the presence of serials of dilutions of compounds. The reactions were allowed to proceed for 1 h at room temperature before a 10 μL preparation of EDTA (20 mM) and Eu-labeled antibody (4 nM) in time-resolved (TR)-FRET dilution buffer are added. The final concentration of antibody in the assay well is 2 nM, and the final concentration of EDTA is 10 mM. The plate is allowed to incubate at room temperature for one more hour before the TR-FRET emission ratios of 665/340 nm were acquired on a PerkinElmer EnVision multilabel reader (Perkin-Elmer, Inc.). Data analysis and curve fitting were performed using GraphPad Prism 7 software.
KINOMEscan. Kinase-tagged T7 phage strains were prepared in an Escherichia coli host derived from the BL21 strain. E. coli were grown to log-phase and infected with the T7 phage and incubated with shaking at 32°C until lysis. The lysates were centrifuged and filtered to remove cell debris. The remaining kinases were produced in HEK-293 cells and subsequently tagged with DNA for quantitative polymerase chain reaction (qPCR) detection. Streptavidin-coated magnetic beads were treated with biotinylated small-molecule ligands for 30 min at room temperature to generate affinity resins for kinase assays. The liganded beads were blocked with excess biotin and washed with blocking buffer [SeaBlock (Pierce), 1% bovine serum albumin (BSA), 0.05% Tween 20, 1 mM dithiothreitol (DTT)] to remove the unbound ligand and to reduce nonspecific binding. Binding reactions were assembled by combining kinases, liganded affinity beads, and compound 7i in 1× binding buffer [20% SeaBlock, 0.17× phosphate-buffered saline (PBS), 0.05% Tween 20, 6 mM DTT]. All reactions were performed in polystyrene 96-well plates in a final volume of 0.135 mL. The assay plates were incubated at room temperature with shaking for 1 h, and the affinity beads were washed with wash buffer (1× PBS, 0.05% Tween 20). The beads were then resuspended in an elution buffer (1× PBS, 0.05% Tween 20, 0.5 μM nonbiotinylated affinity ligand) and incubated at room temperature with shaking for 30 min. The kinase concentration in the eluates was measured by qPCR. For primary screening, compound 6j was screened at the concentration of 1 μM, and the results are reported as "% Ctrl".
For K d determination, an 11-point 3-fold serial dilution of compound 7f was prepared in 100% DMSO at 100× final test concentration and subsequently diluted to 1× in the assay (final DMSO concentration = 1%). Binding constants (K d values) were calculated with a standard dose−response curve using the Hill equation.
Epithelial−Mesenchymal Transition Assay. For immunofluorescent staining, cells were treated with 50 μg/mL collagen and indicated concentration of 7f for 24 h. Cells were fixed with methanol and stained with E-cadherin (24E10, Cell Signaling #3195), Ncadherin (13A9, Cell Signaling #3195), and DAPI. Fluorescent images were captured with a Photometric CoolSNAP HQ camera using NIS Elements AR 2.3 Software (Nikon).
For 3D culture, cells were embedded in ECM consisting of 5 mg/ mL matrigel (BD Biosciences) and 2.1 mg/mL collagen I (BD Biosciences). Cultures were overlaid with DMEM + 10% FBS containing 2% matrigel. For analysis of spheroid morphology, 48 h after plating onto ECM, cells were fixed with methanol and stained with phalloidin and DAPI. Fluorescent images were captured with a Photometric CoolSNAP HQ camera using NIS Elements AR 2.3 Software (Nikon).
Colony Formation Assay. Cells were cultured in six-well tissue culture plates at low density (1000 cells per well) in 2 mL of media with 10% FBS ± 7f at the indicated doses for 10 days. A DMSO control was added to the respective wells to demonstrate the vehicleindependent effect. Cells were then stained with crystal violet.
3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium Bromide Assay. Cells were plated at 2000 cells/well in tissue-culturetreated 96-well plates. The following day, different concentration of 7f was added to each plate in a dilution series across the plate. On day 5, 20 μL of thiazolyl blue tetrazolium blue was added, followed by a 3 h incubation at 37°C, and then the medium was removed, and 100 μL of DMSO was added to each well. The absorbance was read at 540 nm on a plate reader.
Wound-Healing Assay. Cells were cultured in six-well tissue culture plates at high density (∼90% confluence) in 2 mL of media with 5% FBS. Uniform scratches were made down the center of each well with a p20 pipette tip, cells were gently washed with PBS to remove the loose cell debris, and drug was added in media containing 5% FBS. Each condition was in triplicates. Images from the center of each well were taken at 0 and 24 h. The wound width (μm) was measured using NIS Elements AR 2.30 software. The initial wound width was used to verify consistency in scratches.
Determination of Pharmacokinetic Parameters in Rats. All animal studies were performed according to the protocols and guidelines of the institutional care and use committee. The 4−6 week old male Sprague-Dawley rats were purchased from the Shanghai Laboratory Animal Research Center (Shanghai, China). All of the procedures related to animal handling, care, and treatment in this article were performed in compliance with Agreement of the Ethics Committee on Laboratory Animal Care and the Guidelines for the Care and Use of Laboratory Animals in Shanghai, China. Animals were maintained on standard animal chow and water ad libitum, in a climate-controlled room (23 ± 1°C, 30−70% relative humidity, a minimum of 10 exchanges of room air/h and a 12 h light/dark cycle) for one week prior to experiments. The compound was dissolved in the solution containing 2% DMSO, 4% ethanol, 4% castor oil, and 90% ddH 2 O. Pharmacokinetic properties of SD rats (male) were determined following intravenous (iv) and oral administration. Animals were randomly distributed into two experimental groups (n = 3). The oral groups were given 25 mg/kg by gastric gavage. The other group was dosed by injection into the tail vein (5 mg/kg). After a single administration, whole blood samples (100−200 μL) were obtained from the orbital venous plexus at the following time points after dosing: 5, 10, and 30 min and 1, 2, 3, 4, 6, 8, 11, and 24 h (po); 2, 10, and 30 min and 1, 2, 3, 4, 6, 8, 11, and 24 h (iv). Whole blood samples were collected in heparinized tubes. The plasma fraction was immediately separated by centrifugation (8000 rpm, 6 min, 4°C) and stored at −20°C until liquid chromatography−mass spectrometry analysis. The rats were humanely euthanized by carbon dioxide 24 h after experiment without pain. The pharmacokinetics parameters were

Journal of Medicinal Chemistry
Article calculated by analyzing the compound concentration in plasma samples using the pharmacokinetic software DAS.2.0.
In Vivo Anticancer Activity Assay. All animals were housed in a pathogen-free facility with access to food and water ad libitum. C57BL/6 mice were purchased from the UT Southwestern Mouse Breeding Core. Pan02 and BMF-A3 cells (5 × 10 5 ) were injected orthotopically in 6−8 week-old C57BL/6 mice. Fourteen days after tumor cell injection, mice were randomized to receive treatment. Experiments were approved and performed in accordance with the Institutional Animal Care and Use Committee at UT Southwestern. Experiments were stopped after the designated time post tumor cell implantation. Tumors were harvested, and tumor weight was measured. Data sets were analyzed by ANOVA.
Molecular Modeling. All of the procedures were performed in Maestro 11.2 (version 11.2, Schrodinger, LLC, New York, NY, 2017). The DDR1 (PDB code: 5FDP and 6HP9) and TrkC (PDB code: 3V5Q) proteins were processed using the "Protein Preparation Wizard" workflow in Maestro 9.4 (version 11.2, Schrodinger, LLC, New York, NY, 2017) to add bond orders and hydrogens. All hetatm residues and crystal water molecules beyond 5 Å from the het group were removed. Compounds were built by the LigPrep module using the OPLS-2005 force field. The glide module was used as the docking program. The grid-enclosing box was placed on the centroid of the 0LI, which was extracted from the crystal structures of DDR1 and TrkC separately. The standard precision approach of Glide was adopted to dock compounds 6 and 7a−n to DDR1 and compound 7c to TrkC with the default parameters.
Crystallization and Structure Determination. The kinase domain of human DDR1 (Uniprot Q08345, residues 601−913) was expressed as an N-terminal 6× His fusion in Sf9 cells and purified by nickel affinity chromatography followed by tag cleavage with tobacco etch virus protease and then size exclusion chromatography on an S200 column (GE Healthcare). Protein at 12 mg/mL in 50 mM HEPES pH 7.5, 300 mM NaCl, 0.5 mM tris(2-carboxyethyl)phosphine, 2% DMSO was incubated with 1 mM compound 7c for 3 h on ice and then filtered to 0.22 μm. Then, 150 nL of sitting drops was set up with diffracting crystals being obtained from a 1:2 ratio of protein to mother liquor (10% ethylene glycol, 0.2 M sodium sulfate, 25% poly(ethylene glycol) 1500, 0.1 M bis−tris−propane pH 7.1). Crystals were cryoprotected in the mother liquor supplemented with 25% ethylene glycol and flash-frozen in liquid nitrogen. Diffraction was carried out at Diamond Light Source beamline I04 at 100 K. Data were indexed and integrated using XDS 37 and scaled using AIMLESS. 38 Initial phases were identified using molecular replacement in PHASER. 39 The structure was refined using iterative rounds of manual building using COOT 40 and refinement using PHENIX.-REFINE. 41 The refined structure was validated with MolProbity, 42 and the atomic coordinate files were deposited in the Protein Data Bank with Autodep. 43 ■ ASSOCIATED CONTENT

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jmedchem.9b00365.
Docking studies, X-ray structure determination of DDR1-7c, 7f KINOMEscan study, antiproliferation study of 7f, animal body weights in mice with 7f, synthesis route of 7a and 7b, 1 H and 13 C NMR spectra of compounds 7a−n, HPLC traces for the representative compounds (PDF)

Accession Codes
Atomic coordinates and experimental data for the cocrystal structure of 7c with DDR1 (PDB ID: 6HP9) will be released upon article publication.