2-substituted Nγ-glutamylanilides as novel probes of ASCT2 with improved potency

Abstract

Herein, we report the discovery and structure-activity relationships (SAR) of 2-substituted glutamylanilides as novel probes of the steric environment comprising the amino acid binding domain of alanine-serine-cysteine transporter subtype 2 (ASCT2). Focused library development led to three novel, highly potent ASCT2 inhibitors, with N-(2-(morpholinomethyl)phenyl)-L-glutamine exhibiting the greatest potency in a live-cell glutamine uptake assay. This level of potency represents a three-fold improvement over the most potent, previously reported inhibitor in this series, GPNA. Furthermore, this and other compounds in the series exhibit tractable chemical properties for further development as potential therapeutic leads.

Emerging evidence implicates oncogenic signaling pathways with nutrient uptake in cancer cells. The natural amino acid glutamine is essential for cell growth and proliferation. In addition to glucose, cancer cells utilize glutamine as a carbon source for ATP production and biosynthesis. Mammalian cells can internalize glutamine through an evolutionary redundant repertoire of cell surface transporters, though a primary sodium-dependent transporter of glutamine, ASCT2 (gene symbol SLC1A5), stands out as a promising target for probe development. In cancer cells, SLC1A5 expression is associated with oncogenic MYC^1,2 and KRAS^3,4, suggesting its relevance in many clinically important tumors, including those of the lung, colon, and pancreas.^5–7 Demonstrating that ASCT2/SCL1A5 activity might be ‘actionable’ in variety of settings in oncology, Fuchs and co-workers first demonstrated that SLC1A5 antisense RNA triggered apoptosis in human hepatocellular carcinoma cells.⁸ Furthermore, Hassanein et al. more recently reported that SLC1A5 was expressed in 95% of squamous cell carcinomas (SCC), 74% of adenocarcinomas (ADC), and 50% of neuroendocrine tumors. In those studies, siRNA down-regulation of ASCT2 in lung cancer cells resulted in significant growth inhibition⁹. Collectively, these studies suggest the potential fruitfulness of developing small molecules capable of inhibiting ASCT2 activity as precision cancer medicines.

To date, few pharmacological inhibitors of ASCT2 have been reported and none appear to be optimal for advancing as therapeutic leads. As an early entrant to the field, in 2004, Esslinger and co-workers described L-γ-glutamyl-p-nitroanilide (GPNA) as a commercially available probe of the ASCT2 amino acid binding site.¹⁰ While this work illustrated that GPNA could inhibit glutamine uptake in cells at millimolar levels and ascribes certain potential electronic requirements possessed by GPNA and similar analogues from that series, this work did not address the steric requirements for binding to ASCT2 within this compound class.

To discover ASCT2 inhibitors with greater potency and to elucidate SAR around this target, we merged structure-based design with technology-enabled medicinal chemistry and high-throughput screening to identify novel ASCT2 probes with improved potency. We also sought to explore the steric environment of the ASCT2 amino acid binding pocket to encourage future probe development. Since the crystal structure of human ASCT2 has not been elucidated, we employed computational approaches similar to the approach of Albers et al.¹¹ to explore potential points of intermolecular interaction and binding pockets accessible to candidate probes. From a homology model based on the open structure of the bacterial aspartate transporter GltPh in complex with inhibitor D,L-threobenzyloxyaspartate (TBOA), PDB ID 2NWW, a number of targetable structural motifs were identified including a lipophilic pocket adjacent to the amino acid zwitterion binding site and potential hydrophilic points of contact within a loop region that was displaced by the inhibitor in the open form of the transporter. Based upon these structural elements, we expanded a focused library of candidate small molecules based on the Nγ-glutamylanilide series to generate novel chemical matter to test the hypothesis that targeting at least a portion of these elements would result in ASCT2 inhibitors with greater potency. In support of this structure-based approach, we herein report several novel leads from this series that exhibit potency similar to or significantly greater than GPNA in live cell assays.

Initially, we developed an improved synthetic scheme to yield target Nγ-glutamylanilides. The previously reported synthesis of GPNA and related analogs required 6 steps starting from L-glutamate in overall yields ranging from 10–54%.¹⁰. In order to achieve a more facile synthesis, we took advantage of microwave-assisted organic synthesis (MAOS) to rapidly generate Nγ-glutamylanilides analogs in just two steps starting from the commercially available Boc-L-glutamic acid-tert-butyl ester. with typical yields ranging from 23–75% over two steps (Figure 1).¹²

Figure 1.

Synthetic route towards 2-substituted N_γ-glutamylanilides.

Initial compound libraries focused on 2, 3, and 4 substituted Nγ-glutamylanilides with aryl, alicyclic, and heterocyclic substitutions (Table 1). To evaluate the biological activity, compounds were initially screened at a single concentration for their ability to inhibit ³H-glutamine uptake in live HEK-293 cells, an established model suitable for evaluating ASCT2 activity ^14,15. Full concentration response curves were developed for compounds that exhibited evidence of glutamine inhibition; inactive compounds were not pursued further. From early library development efforts, we were able to prioritize the 2-substitution as a determinant of ASCT2 activity among this series. For example, N-(2-morpholinophenyl)-L-glutamine (Table 1, compound 3)¹³ emerged as a potential lead compound of interest, exhibiting a potency roughly equivalent to that of GPNA (Table 1, compound 1). In contrast, the analogous 4-morpholinophenyl and 3-morpholinophenyl anilides proved to be inactive, leading us to pursue 2-substituted glutamyl anilides for the remainder of the study. Further development of the 2-substituted series led to three novel compounds with significantly greater potency than GPNA, N⁵-(2-(benzo[d]thiazol-2-yl)phenyl)-L-glutamine (Table 1, compound 4)¹⁶, N-(2-(morpholinomethyl)phenyl)-L-glutamine (Table 1, compound 5)¹⁷, and N5-(2-((4-methylpiperazin-1-yl)methyl)benzyl)-L-glutamine (Table 1, compound 20). Furthermore, 4 novel compounds among the series exhibited potencies equivalent to GPNA.

Table 1.

SAR of N_γ-glutamylanilides. Compounds deemed inactive exhibited IC₅₀ ≥ 1mM.

Compound	IC50 (μM) ± SEM
1	954 ± 140
2	inactive
3	664 ± 87
4	436 ± 87
5	312 ± 106
6	inactive
7	inactive
8	inactive
9	inactive
10	inactive
11	inactive
12	514 ± 58
13	inactive
14	inactive
15	inactive
16	832 ± 164
17	inactive
18	inactive
19	852 ± 80
20	425 ± 76

Biologically active compounds were also evaluated computationally in the open human ASCT2 model. The best scoring poses for the most potent compounds identified demonstrated a compatible fit with the human ASCT2 model and, interestingly, a tendency to exhibit points of interaction with both the amino acid zwitter ion binding site and an adjacent hydrophobic pocket (Figure 2).¹⁷

Figure 2. Docking of potential leads into ASCT2 homology model.

The most potent lead, compound 5 (brown all atom colored, capped sticks) fits the homology model (protein ribbons) generated for the open form of human ASCT2 and is consistent with the displacement of a key loop region (grey, right). The docked pose shown represents the best scoring SurflexDock conformation for compound 5 (IC₅₀ = ~312μM) that contains a morpholino moiety occupying the hydrophobic pocket adjacent to the amino acid zwitterion binding site. One of the two reported sodium binding sites in the SLC1A5 family (purple van der Waals dotted surface) is shown centered beneath the amino acid binding site. A potential weak hydrogen bond between the morpholine oxygen of compound 5 to ASCT2 residue Cys 467 sulfhydryl side chain (yellow dashed line) is highlighted.

In summary, we report three novel Nγ-glutamylanilides as inhibitors of cellular glutamine uptake via ASCT2 with significantly greater potency than GPNA. Evaluation of this chemical series within the context of ligand docking to a homology model of human ASCT2 revealed reasonable compatibility with the ASCT2 binding site based on SurflexDock Total Scores. Based upon our data, we anticipate that compounds with the greatest potency may interact with multiple structural elements within the ASCT2 binding site, including the amino acid zwitter ion binding site and the adjacent hydrophobic pocket. Ongoing efforts employing a combination of these effects may lead to compounds with even greater potency. Uniquely, previous work in the Nγ-glutamylanilide series suggested that reduction of the glutamine amide pK_a was required for ASCT2 inhibition¹⁰; we did not observe this trend in our study. Advances from these studies will be reported in due course.