Discovery of Novel Small-Molecule FAK Activators Promoting Mucosal Healing

Abstract

Gastrointestinal mucosal wounds are common to patients injured by factors as diverse as drugs, inflammatory bowel disease, peptic ulcers, and necrotizing enterocolitis. However, although many drugs are used to ameliorate injurious factors, there is no drug available to actually stimulate mucosal wound healing. Focal adhesion kinase (FAK), a nonreceptor tyrosine kinase, induces epithelial sheet migration and wound healing, making FAK a potential pharmacological target in this regard. In our previous research, we found a lead compound with drug-like properties, ZINC40099027, which promotes FAK phosphorylation, inducing mucosal healing in murine models. Herein we describe the design and optimization of a small library of novel FAK activators based on ZINC40099027 and their applications toward human intestinal epithelial wound closure and mouse ulcer healing.

Gastroduodenal ulcer disease is a source of significant morbidity and mortality all over the world. Approximately two-thirds of patients with gastroduodenal ulcer disease are asymptomatic or minimally symptomatic¹ until initial presentation. The prevalence of gastroduodenal ulcer disease in the United States is estimated to be 8.4%.² Despite the reduction in Helicobacter pylori prevalence in many parts of the world due to antibiotic treatment, the prevalence of gastroduodenal ulcer disease is actually increasing in the aged population, while nonulcer gastritis and duodenitis are increasingly common. These increases are mainly due to the growing use of nonsteroidal anti-inflammatory drugs (NSAIDs) not only to manage acute and chronic pain and arthritis but also, importantly, for the primary and secondary prevention of inflammation and cardiovascular events such as stroke and myocardial infarction.^3,4 More than 50% of patients taking NSAIDs have some mucosal injury in the small bowel,⁵ but NSAID-related enteropathy often presents with subtle findings.⁶ The gut mucosa is also injured in other settings such as inflammatory bowel diseases (IBDs) like Crohn’s disease and ulcerative colitis. More than 1 million people in the U.S. and 2.5 million in Europe are estimated to have IBDs,⁷ and 51 200 patients with IBDs died in the U.S. in 2013 alone.⁸

Gastrointestinal mucosal healing is very important to enhance the quality of life for patients suffering from peptic ulcers, necrotizing enterocolitis, and IBDs. Potential strategies for the prevention of peptic ulcers and their complications when the use of NSAIDs is necessary include cotherapy of NSAIDs with an antisecretory agent, substitution of nonselective NSAIDs with COX-2-selective NSAIDs, and combination of a COX-2-selective NSAID with a gastroprotective agent.^9,10 However, there is no known therapy that promotes mucosal healing directly. Moreover, recent data suggest that coprescription of NSAIDs with an antisecretory agent, like a proton pump inhibitor, to reduce the synergistic mucosal damage caused by gastric acid can actually potentiate NSAID injury to the distal small bowel because of changes in the microbiome and how it processes the NSAID; this is no longer routinely recommended.¹¹

Gut wound healing is a complex process. For deeper wounds, healing involves a complex time-dependent interplay among epithelial, endothelial, and inflammatory cells and fibroblasts, all driven by cell–matrix interactions.¹²⁻¹⁴ However, the sine qua non of mucosal repair, whether for deep or superficial mucosal injury, is epithelial sheet migration, a process by which the epithelial cells at the edge of the wound assume a squamous morphology and migrate across the defect to reseal it.¹⁵

Focal adhesion kinase (FAK), a 125 kDa nonreceptor tyrosine kinase, plays an important role in cell motility, proliferation, and epithelial sheet migration.¹⁶ FAK is ubiquitously expressed and could be a potential therapeutic target with pharmacological applications. The proportion of FAK that is phosphorylated and active (FAK-Tyr-397, in its initial activated form) is actually decreased in both migrating intestinal epithelial cells in vitro(17) and around human mucosal gastric and colonic ulcers in vivo,¹⁸ making this an even more attractive target for activation to overcome this deficit. Although others have extensively explored FAK inhibitors,¹⁹ little attention has been paid to the possibility of activating FAK. We previously identified two commercially available small molecules that enhance FAK activation in vitro(20) and demonstrated that one of these molecules stimulates small intestinal mucosal wound repair in two mouse models.²¹ Following up on these encouraging results, herein we report a library of novel FAK small-molecule activators that induce epithelial sheet migration, with potential therapeutic application to stimulate epithelial sheet migration and gut mucosal wound healing.

While searching for small molecules that would mimic a key subdomain of the N-terminal FERM domain of FAK and therefore competitively inhibit FAK–AKT binding, we serendipitously identified two small molecules, ZINC40099027 (which we abbreviate here as Zn27) and ZINC25613745 (Zn45), that may actually activate FAK.^20,21 While our previous studies had shown that Zn27 activates FAK even in poorly differentiated cancer cells and Zn45 did not achieve a statistically significant effect in those cells, further studies had shown that Zn27 does activate FAK in human Caco-2 intestinal epithelial cells (a common model for the study of intestinal mucosal healing in vitro) at concentrations as low as 10 nM,²¹ and new studies showed that Zn45 activates FAK comparably to Zn27 at 10 nM (Figure 1).

Figure 1.

Zn27 and Zn45 (10 nM) activate FAK (assessed as FAK-Tyr-397 phosphorylation) in Caco-2 intestinal epithelial cells (*, p < 0.05; n = 8).

Biological evaluation of Zn27in vitro suggested that it potently activates FAK and promotes intestinal epithelial sheet migration in monolayer wound closure. For example, Zn27 specifically increased FAK-Tyr 397 phosphorylation within intact cells while activating neither Pyk2, a close paralogue of FAK, nor Src, another key nonreceptor kinase within the focal adhesion complex.²¹Zn27 stimulated monolayer wound closure and intestinal mucosal healing in vivo. The in vitro effects on wound closure were blocked by FAK inhibition, while the in vivo effects were accompanied by increased FAK activation at the ulcer edge.²¹ Encouraged by this FAK activation by Zn27, we designed and implemented follow-up structure–activity relationship (SAR) studies using commercially available analogues (SAR by commerce). These compounds show high structural similarity to Zn27 and suitable in silico-predicted drug-like properties (Figure 2).²²

Figure 2.

Library of commercially available molecules tested as FAK activators.

Among these molecules that are structurally similar to Zn27, the novel compounds 3, 6, 11, and 12 displayed increased FAK-Tyr-379 phosphorylation versus DMSO vehicle control, while the rest of the compounds caused little change in FAK phosphorylation (Figure 3). Further studies performed with 3, 6, 11, and 12 in Caco-2 cells showed that these molecules are very potent dose-dependent FAK activators at nanomolar to picomolar levels (Figures 4 and 5). After obtaining these encouraging results, we proceeded to test the ability of 3, 6, 11, and 12 to promote Caco-2 epithelial monolayer wound closure. As shown in Figure 6, compared with the DMSO vehicle control, 3, 6, 11, and 12 stimulated wound closure by approximately 10–26% over baseline wound closure rates. Compound 7 (used as an internal control) at different concentrations did not promote wound closure compared with compounds 3, 6, 11, and 12. Addition of hydroxyurea (HyU), which prevents cell proliferation,²³ did not hinder wound closure in the presence of 3, 6, 11, and 12. These results suggest that 3, 6, 11, and 12 not only activate FAK but also stimulate epithelial sheet migration.

Figure 3.

(a) 3, 6, 11, and 12 activate FAK-Tyr-379 phosphorylation at 10 nM, whereas (b, c) the other compounds did not show a statistical difference toward the activation of FAK-Tyr-379 phosphorylation (*, p < 0.05; n > 6).

Figure 4.

(a) 3, (b) 6, (c) 11, and (d) 12 activate FAK-Tyr-379 phosphorylation in a dose-dependent fashion at nanomolar concentrations (*, p < 0.05; n > 5).

Figure 5.

Dose-dependent activation of FAK-Tyr-379 phosphorylation by (a) 3, 6, and 11 and (b) 12 (*,p < 0.05; n > 5).

Figure 6.

(a–d) Aggregated results from multiple wounds made in multiple cell monolayers suggest that 3, 6, 11, and 12 promote Caco-2 monolayer wound closure (*, p < 0.05; n > 5). (e) 7 was used as an internal control. (f) 3, 6, 11, and 12 enhance wound closure in the presence of hydroxyurea. (g) Typical images from wound closure assays demonstrating stimulation of Caco-2 cell monolayer wound closure by FAK activators 3, 6, 11, and 12 and the internal control 7 (10 nM).

Following the encouraging in vitro results, we proceeded to perform in vivo intestinal epithelial wound closure studies on compound 3 as our lead compound. For our initial study, a 900 μg/kg dose of compound 3 (diluted in 100 μL of normal saline after first being dissolved in DMSO) was administered intraperitoneally to C57BL/6 mice (the final concentration of DMSO in the injected intraperitoneal solution was 2.5%). The plasma concentration of compound 3 peaked at 1 h and decayed time-dependently to baseline after 6 h (Figure 7a). Ischemic jejunal ulcers were then created artificially in mice.^24,25 After 24 h of ulcer formation, compound 3 or DMSO vehicle control was administered intraperitoneally every 6 h for 3 days. On day 4, mice that had received compound 3 exhibited a substantially smaller ulcer size compared with those treated with the DMSO vehicle (Figure 7b). Although all of the mice lost weight, as one would expect from this surgical and stressful model, mice treated with compound 3 lost less weight than the DMSO-treated mice (Figure 7c). At sacrifice, the plasma concentration of compound 3 was similar to the concentration that was effective in the in vitro model (data not shown).

Figure 7.

(a) Compound 3 shows the highest concentration at 1 h. (b) Typical images of ischemic ulcers (1.57 ± 0.45 mm² for compound 3 vs 3.26 ± 0.47 mm² for DMSO; p < 0.05). (c) Loss of weight in the presence of compound 3 (6.33 ± 1.06%) vs DMSO (9.16 ± 0.62%) (p < 0.05).

On the basis of the enhancement of both FAK activation and wound closure by this novel set of molecules in vitro as well as in vivo, we proceeded to synthesize a small library of novel FAK activators using 3 and 6 as a basis for our further SAR studies. Physicochemical and other predicted properties of 3, 6, 11, and 12 were computed using QikProp, industry-standard Schrödinger software for successful lead discovery (Table S1).²⁶ The only noted potential issue is the predicted borderline hERG liability for compound 6. This liability, and possible others, could be further addressed as the series evolves with compounds with a better pharmacological profile. At this point, our focus is to enhance the potency for FAK phosphorylation, even though this process would not take into account the presence of chiral centers. Based on scaffold similarity among the compounds, our preliminary conclusion is that the 1-(2-morpholino-5-(trifluoromethyl)phenyl)urea moiety is essential for FAK phosphorylation as seen in Figure 2. For example, comparison of the FAK activation by 3 to those by 8 and 9 clearly shows that the lack of the morpholine ring does not enhance FAK activation.

With these results in hand, we further modified other ring moieties by rational isosteric modifications.^27,28 Our synthetic strategy for diversified compound preparation was based on a previously reported approach for synthesizing ureas²⁹ via treatment of a phenyl carbamate with the corresponding cyclic or aliphatic amine derivative (Scheme 1). Similarly, we further explored whether removal of the trifluoromethyl group is essential for FAK activation. When required, the hydrochloride salts of these analogues were pursued following standard experimental procedures. To further explore and assess the importance of the morpholine moiety, in parallel we also proceeded to synthesize novel analogues of 3 bearing N-methylpiperazine, 2,6-dimethylmorpholine, and thiomorpholine dioxide, respectively (Scheme 2). When required, 4-fluoro-3-nitrobenzotrifluoride was alkylated with 2,6-dimethylmorpholine through an S_NAr reaction, followed by reduction to its amine derivative. Afterward, commercially available amines as well as the one bearing the 2,6-dimethylmorpholine, were converted to ureas in a similar fashion via phenyl carbamate as described above.

Scheme 1. Synthesis of Novel Analogues of 3, 6, and 11 as FAK Activators.

Scheme 2. Synthesis of Novel Analogues of 3 with the Morpholine Ring Replaced as FAK Activators.

These novel compounds were tested in a biochemical assay to assess the potency to stimulate Caco-2 (by Western blotting for FAK-397) and HIEC-6 FAK activation (by a commercially available ELISA for FAK-Y-397). As shown in Table 1, 3 and its analogues were compared with regard to effectiveness in phosphorylating FAK, measured as fold change to FAK. In this series, 3 was the most potent FAK activator.

Table 1. Cell-Based FAK-Tyr-397 Activation^a.

^aCaco-2 cells were treated with compounds at 10 nM concentration in at least eight separate experiments with similar results, and the results were pooled for analysis. Data shown represent mean ± standard error.

In order to enhance the solubility of 3 and at the same time facilitate treatment of cells or potentially intact organisms to induce FAK activation, we synthesized its hydrochloride salt 14 (Table S2). However, 14 showed reduced activation compared with 3. Similarly, to reduce the lipophilicity, we also pursued the removal of the CF₃ moiety to obtain 15 and synthesized its corresponding hydrochloride salt 16, but neither of those compounds induced FAK phosphorylation. Compound 17, the desmethoxy analogue of 6, was also synthesized and evaluated, but it did not stimulate FAK phosphorylation. Finally, 18, an analogue of 12 bearing a hydroxyl moiety, was synthesized with the goal of increasing the uptake in CaCo-2 cells by glucurodination, thereby enhancing FAK phosphorylation. However, 18 also did not promote FAK phosphorylation as seen with 12. All of the new compounds were tested at 10 nM. However, none of these novel analogues stimulated FAK phosphorylation compared with 3, 6, and 12.

By replacing the morpholine group with N-methylpiperazine (to obtain 19), 2,6-dimethylmorpholine (to obtain 20), or thiomorpholine dioxide (to obtain 21), we sought to enhance either the solubility or metabolic stability. However, as can be seen in Table 2, these compounds showed decreased activity toward FAK phosphorylation compared with 3. The lack of FAK phosphorylation activation with 19 could be due to quick clearance before activation could be induced. In the case of 20 and 21, the bigger size of the substituent could lead to steric hindrance and thus decreased activity. Up to this point, morpholine appears to be a key fragment to enable 3 to exert its biological activity.

Table 2. Cell-Based FAK-Tyr-397 Activation^a.

^aCaco-2 cells were treated with compounds at 10 nM concentration in at least eight separate experiments with similar results, and the results were pooled for analysis. Data shown represent mean ± standard error.

While the regulation of intestinal epithelial sheet migration involves a complex web of interconnected signals and feedback loops, reducing FAK activity by pharmacological inhibition, siRNA interference, or transfection of dominant-negative FAK mutants decreases epithelial sheet migration in vitro despite compensation by other kinases, while changes in FAK activity are associated with changes in the intestinal epithelial migratory phenotype in vivo.^{18,25,30−33} Indeed, our previous study with Zn27 makes precisely this point.²¹ For the first time, in that paper, we were able to study the specific effects of FAK activation without activating Pyk2 or Src using Zn27, and we observed increased epithelial sheet migration in vitro and mucosal healing in vivo. That is not to say that Src, Pyk2, and other kinases cannot also play an important role in epithelial sheet migration and mucosal healing. However, it is clear that changing levels of FAK activity alone is sufficient to alter this process. Therefore, the absence of an effect of these molecules on Pyk2 or Src represents a relative potential advantage from the clinical perspective over less specific growth factors or tyrosine phosphatase inhibitors or other agents that globally stimulate many kinases, since these molecules can stimulate epithelial sheet migration and mucosal healing without side effects that might be evoked by the activation of other off-target kinases.

It was originally hypothesized that synthesizing the corresponding hydrochloride salt, reducing the lipophilicity by removing the CF₃ moiety, adding a hydroxyl group, or replacing the morpholine moiety would lead to enhancement of uptake inside Caco-2 cells and thus increase FAK activation. However, we must recall that the ability of a small molecule to exert its biological activity inside the cell requires the molecule to overcome diverse issues, including bioavailability, cell membrane transportation, pharmacokinetics, and others. How these various issues affect the fate of these molecules when they are added to cultured cells awaits investigation. Nevertheless, we were encouraged by discovering promising FAK activators such as 3, 6, 11, and 12. These molecules not only stimulate FAK phosphorylation at remarkably low concentrations but also promote the healing of intestinal epithelial wounds in vitro. In addition, 3 showed reasonable drug-like properties based on in vitro, in vivo, and in silico results. We are currently working on further exploration of a new series based on the scaffold of 3 with improved pharmacological profile and its possible mechanism of action.

Experimental Procedures

Zn27, Zn45, and the set of screened compounds 1–13 were obtained from Enamine Ltd. (Monmouth Junction, NJ). All other precursors for chemical synthesis were of the highest purity available. Compounds 6 and 12 were resynthesized (see the Supporting Information).

Glossary

Abbreviations

FAK

focal adhesion kinase

IBD

inflammatory bowel disease

NSAID

nonsteroidal anti-inflammatory drug

Zn27

ZINC40099027

Zn45

ZINC25613745

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.0c00311.

• Synthesis and characterization of compounds 3 and 14–21 and biochemical, cell, and animal studies (PDF)

Author Contributions

^⊥ Q.W. and R.G.-M. are co-first authors. Q.W., S.K.M., A.A.R.E., R., and S.O. performed the biological assays and experiments described here. R.G.-M. synthesized the novel compounds described and verified their purity. M.Y.G. conducted the assays of plasma levels of the molecule under study. M.D.B., V.J.G., and R.G.-M. analyzed the prototypical molecular structures and made decisions about the experimental direction. V.J.G. directed the synthetic work, and M.D.B. directed the biological work and provided overall guidance to the study. All of the authors participated in revising the manuscript.

The authors declare the following competing financial interest(s): M.D.B. and V.J.G. are co-inventors on a patent application filed by the University of North Dakota and the University of Minnesota describing the use of FAK-activating small molecules to promote mucosal healing.