Design and Synthesis of Sulfamoyl Benzoic Acid Analogues with Subnanomolar Agonist Activity Specific to the LPA₂ Receptor

Abstract

Lysophosphatidic acid (LPA) is a growth factor-like mediator and a ligand for multiple GPCR. The LPA₂ GPCR mediates antiapoptotic and mucosal barrier-protective effects in the gut. We synthesized sulfamoyl benzoic acid (SBA) analogues that are the first specific agonists of LPA₂, some with subnanomolar activity. We developed an experimental SAR that is supported and rationalized by computational docking analysis of the SBA compounds into the LPA₂ ligand-binding pocket.

Introduction

Exposure to direct ionizing radiation can result in programmed cell death due to the accompanying DNA damage. Lysophosphatidic acid (LPA, 1-acyl-2-hydroxy-sn-3-glycerophosphate) (1, Figure 1) is a growth factor-like lipid mediator with antiapoptotic actions elicited via a set of G protein coupled receptors. LPA GPCR activate antiapoptotic kinase pathways, inhibit apoptosis, promote cell regeneration, and augment DNA repair, altogether.¹⁻⁴ Our group has been interested in developing LPA analogues for the attenuation of programmed cell death elicited by radiation and chemotherapeutic agents.^2,5−7 In the course of our studies, we discovered octadecenyl thiophosphate (OTP) (2, Figure 1), a synthetic analogue of LPA, which showed strong radioprotective action by rescuing cells from apoptosis and mice irradiated with lethal doses of γ-irradiation.⁷ Although OTP is highly effective in protecting animals from radiation injury, it activates multiple LPA GPCRs, including LPA₁ that has been linked to cause apoptosis via anoikis.^8,9 Recently, our group identified novel nonlipid compounds that are selective, although not specific, agonists of LPA₂ and also inhibit LPA₃.¹⁰ Among these nonlipid compounds, GRI977143 (3, Figure 1), although much less potent than LPA, nevertheless showed antiapoptotic efficacy in cell-based models of apoptosis and it was modestly efficacious against radiation-induced apoptosis in mice suffering from the acute hematopoietic radiation syndrome.¹⁰ In continuation of this line of research, we designed and synthesized sulfamoyl benzoic acid (SBA) analogues of 3 by isoteric replacement of −S– group with −NH-SO₂ using a medicinal chemistry approach guided by computational modeling.^11,12 All 14 new SBA analogues were tested for agonist and antagonist activity at the LPA_1/2/3/4/5 GPCR. We found several nonlipid specific agonist of LPA₂, one of which is 5-chloro-2-(N-(4-(1,3-dioxo-1H-benzo[de]isoquinolin-2(3H)-yl)butyl)sulfamoyl)benzoic acid (11d) (Scheme 4), with picomolar activity (EC₅₀ = 5.06 × 10^–3 ± 3.73 × 10^–3 nM vs LPA 18:1; EC₅₀ = 1.40 ± 0.51 nM) (Figure 2).

Figure 1.

Chemical structures of LPA (1), OTP (2), GRI977143 (3), and SBA analogues.

Scheme 4. Synthesis of Head Group Modified SBA Analogues (Compounds 11a–d).

Figure 2.

Fura-2AM Ca²⁺ assay of 11d versus LPA 18:1. Fura-2AM-loaded LPA₂ DKO MEF cells (closed symbols) or empty vector DKO MEF cells (open symbols) were treated with LPA 18:1 (circles) or 11d (squares).

Results and Discussion

The synthesis of new novel SBA analogues, compounds 4–6, 7a–b, 8a–c, 9–10, and 11a–d, is outlined in Schemes 1–4, respectively.

Scheme 1. Synthesis of Carbon Linker Modified SBA Analogues (Compounds 4–6).

2-(Bromoalkyl)benzo[de]isoquinoline-1,3-dione (15a–c) was obtained in high yield following a similar approach of that already reported procedure.¹³ Compound 15a–c was reacted with 2-sulfamoylbenzoic acid ethyl ester (16) using K₂CO₃ in DMF to get compounds 4–6 in moderate yield (Scheme 1). 2-[4-(1,3-Dioxo-1,3-dihydroisoindol-2-yl)butylsulfamoyl]benzoic acid (7a) and 2-(N-(4-(2,3-dioxoindolin-1-yl)butyl)sulfamoyl)benzoic acid (7b) were synthesized by the reaction of commercially available 2-(3-bromobutyl)isoindole-1,3-dione (17) and 1-(4-bromobutyl)indoline-2,3-dione (19) (which was prepared from 14b and isatin 18) using 16, respectively (Scheme 2).

Scheme 2. Synthesis of Tail Group Modified SBA Analogues (Compounds 7a,b).

2-(N-(4-(1,3-Dioxo-1H-benzo[de]isoquinolin-2(3H)-yl)butyl)sulfamoyl)-substituted benzoic acid (8a–c) was accomplished by the reaction of 15b and substituted benzo[d]isothiazol-3(2H)-one-1,1-dioxides (20a–c) in the presence of K₂CO₃ and DMF under reflux conditions in moderate yield. Reaction of 15b with 3-oxo-2,3-dihydrobenzo[d]isothiazole-6-carboxylic acid 1,1-dioxide (21) produced 2-(4-(1,3-dioxo-1H-benzo[de]isoquinolin-2(3H)-yl)butyl)-3-oxo-2,3-dihydrobenzo[d]isothiazole-6-carboxylic acid 1,1-dioxide (9), which was then treated with aqueous 1N NaOH to procure the desired compound 2-(N-(4-(1,3-dioxo-1H-benzo[de]isoquinolin-2(3H)-yl)butyl)sulfamoyl)terephthalic acid (10) (Scheme 3).

Scheme 3. Synthesis of Head Group Modified SBA Analogues (Compounds 8–10).

N-Butylamino-1,8-naphthalimide (22)¹⁴ was coupled with substituted sulfonyl chloride 23a–e using triethylamine in THF to furnish corresponding ester derivatives (12a–e). Upon treatment of compound 12b–c with LiOH in THF–water (1:1) and compound 12d–e with K₂CO₃ in DMF afforded the final acid derivatives 11a–b and 11c–d, respectively, in moderate to quantitative yield (Scheme 4).

LPA receptor activation leads to a transient mobilization of Ca²⁺ that can be quantified by measuring the fluorescence of the indicator dye Fura-2AM. All compounds were tested using cell lines established previously to overexpress a single LPA GPCR subtype.^10,11 For the human LPA₁, LPA₂, and LPA₃ GPCR, we used mouse embryonic fibroblasts (MEF) derived from LPA₁xLPA₂ double knockout mice (DKO) that lack endogenous LPA₃ expression and do not respond to LPA with Ca²⁺ transients (for assay details and the LPA₄ and LPA₅ cells, see the Supporting Information). The compounds were evaluated for agonist and antagonist activity (Table 1). For agonists compounds, their dose–response was compared to that elicited by LPA 18:1 using a FLEX-Station III robotic plate reader as described in our previous publication.¹⁰

Table 1. LPA Receptor Ligand Properties of Sulfamoyl Benzoic Acid (SBA) Analogues.

	LPA2		LPA1		LPA3		LPA4		LPA5		vectorc
compd	EC50 (μM)	Emax (%)	EC50 (μM)	Emax (%)	EC50 (μM)	Emax (%)	EC50 (μM)	Emax (%)	EC50 (μM)	Emax (%)	EC50 (μM)	Emax (%)
LPA 18:1	0.002	100	0.35	100	0.83	100	0.26	100	0.51	100	NE
GRI-977143	3.30	75	NEa	NE	NEb	NE	NE	NE	NE	NE	NE	NE
4	2.16	100	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
5	0.11	100	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
6	1.42	100	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
7a	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
7b	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
8a	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
8b	0.05	100	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
8c	0.06	104.15	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
9	3.60	100	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
10	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
11a	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
11b	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
11c	1.5 × 10–4	119	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE
11d	5.06 × 10–6	124	NE	NE	NE	NE	NE	NE	NE	NE	NE	NE

^aNE: no detectable effect up to 10 μM concentration.

^bGRI-977143 has no agonist activity at LPA₃, however, it is an antagonist with an IC₅₀ concentration of 6.6. μM against LPA 18:1

^cVector control cells were DKO MEF (LPA₂), RH7777 (LPA_1/3), CHO (LPA₄), and B103 (LPA₅) transfected with the human LPA receptor orthologue (in parentheses) or transduced with appropriate empty vector. EC₅₀ concentration is given in μM for dose–response curves covering the 0.0001 nM to 10 μM range. For determination of antagonist action, dose–response curves were generated using an ∼E_max50 concentration of LPA 18:1 for any given LPA receptor subtype, and the ligand was coapplied in concentrations ranging from 30 nM to 10 μM. Means represent the mean values of three assays.

We embarked on a scaffold derivatization approach steered by structure-based pharmacophore design. Studies were initiated with GRI977143 (3) as a template scaffold with the aim of achieving LPA₂ subtype agonist specificity by eliminating LPA₃ antagonist activity and improving its potency through molecular modifications. Our approach was to prioritize the scaffold variants through several iterative steps. Ten different scaffolds based on 3 were flexibly docked into the LPA₂ homology model^10,11 and first ranked by lowest energy docked score. Each scaffold was then ranked based on the protein–ligand interaction fingerprint, and a final ranking was constructed based on visual inspection of pharmacophore match. Using this strategy in combination with perceived ease of synthesis, 2-(N-(3-(1,3-dioxo-1H-benzo[de]isoquinolin-2(3H)-yl)propyl)sulfamoyl)benzoic acid (4) was selected. Compound 4 had virtually identical overlay with respect to the shape of 3 where sulfur was replaced by a sulfamoyl moiety (Figure 1). Molecular modeling studies comparing compounds 3 and 4 suggested that the improved potency and selectivity of compound 4 was attributed to the increased binding affinity of compound 4 (−8.53 kcal/mol) compared to compound 3 (−7.94 kcal/mol). While both compounds can adopt similar conformations in the binding pocket, compound 4 has a higher conformation strain at 12.39 kcal/mol than compound 3 at 4.32 kcal/mol (Figure 3A). The pharmacological characterization of compound 4 at LPA₂ showed that, in spite of its low potency (EC₅₀ LPA₂ ∼ 2 μM), it was a specific agonist of LPA₂ without activating or inhibiting LPA_1/3/4/5 receptors. Of note is that, unlike GRI977143, compound 4 lacked LPA₃ antagonist activity up to 10 μM, the highest concentration tested (Table 1). Therefore, this scaffold 4 was deemed an optimal template for guiding lead optimization medicinal chemistry. Medicinal chemistry focused SAR to increase the potency by synthesizing derivatives of compound 4 by modifying at three target regions: the head group, chain linker, and tail group (Figure 1). We docked and evaluated the chain linker of compound 4 by increasing the carbon numbers to four carbon (compound 5) and five carbon (compound 6) linkers, which are similar in length to the high-potency natural ligand LPA 18:1 of the LPA₂. Computational models of the docked linker analogues predicted the overall length of compounds of these analogues at 14, 15, and 16 Å, respectively. Pharmacological characterization of these three compounds 4, 5, and 6 showed that compound 5 with a four carbon linker had the highest potency, lowering the EC₅₀ value of the compound from the micromolar to the nanomolar range. Thus, the optimal length of these analogues is at or near 15 Å that is the same as of LPA 18:1 (Scheme 1 and Table 1). In continuation of our SAR, we explored the modifications of compound 4 by keeping the linker length fixed at four carbons. We replaced the 1H-benzo[de]isoquinolyl-1,3(2H)-dione tail group with isoindolyl-1,3-dione (17) or indolyl-2,3-dione (18) to obtain compounds 7a and 7b, respectively. These two changes abolished the activation of LPA₂ up to 10 μM, the highest concentration tested. These tail groups are predicted to occupy a highly hydrophobic region in the LPA binding pocket. Modeling of the tail groups of compounds 7a and 7b in the LPA₂ binding pocket predicts several missing π–π interactions (Supporting Information Figure 1), which is the reason for their lack of activity. In our final SAR modifications, we modified the phenyl head group of compound 4. We replaced the hydrogen, which is para to the carboxy group in compound 4 by an electron donating methoxy group or withdrawing bromo group yielded compounds 8a and 8b, respectively. Compound 8a showed very weak agonist activity, whereas compound 8b was more potent than compound 4. We also introduced dicarboxy groups on the head group to improve the potency because previous analysis of intramolecular pH predicted that two negative charges of the LPA phosphate interact with the receptor.¹⁰ During the synthesis of the dicarboxy analogue (compound 10), we obtained a cyclized intermediate compound 9, which was further hydrolyzed to dicarboxy analogue 10. Compound 10 was considerably less potent than the compound 9. Next we turned our attention to introduce the electron withdrawing groups. Fluoro, chloro, and bromo substitutions in meta position to the carboxy group of compound 4 produced compounds 11c, 11d, and 8c, respectively. Among these three analogues, compound 11d exhibited picomolar activity (EC₅₀ (nM): 5.06 × 10⁻³ ± 3.73 × 10^–3 vs LPA 18:1; 1.40 ± 0.51), whereas compounds 11c and 8c showed 0.15 ± 0.02 nM and 60.90 ± 9.39 nM, respectively.

Figure 3.

Docked positions of compound 4 (A) and compound 11d (B), shown overlaid on compound 3 (purple) in the LPA₂ binding pocket. Critical residues for receptor activation are shown in stick. Compounds were docked and energy minimized using the Molecular Operating Environment, version 2009.10 software (Chemical Computing Group Inc., 2012).

Molecular docking simulations revealed that the addition of halogens at the para position would lead to compounds with a conformation strain similar to compound 3. Compound 11d is predicted to share approximately the same distance in critical interactions with R3.28 and Q3.29 as compounds 3 and 4, however, compound 11d binding affinity is −9.21 kcal/mol and conformation strain is 6.50 (Figure 3B). To complete our SAR studies, finally we prepared additional two compounds where we changed the position of carboxy group on the head group of phenyl ring of compound 4 from ortho to meta (compound 11a) and para (compound 11b) positions that abolished the activity at LPA₂. The corresponding methyl carboxy esters of compounds 5 and 11a–d (compound 12a–e) were also inactive at activating LPA₂ up to 10 μM, the highest concentration tested.

Conclusion

Here we designed and synthesized new SBA analogues using a structure-based pharmacophore. Starting from our active scaffold 4, the detailed structure–activity relationship profile was established to understand the effect of particular substituents on head and tail groups along with carbon linker. We concluded that three important structural requirements are critical to have potent and specific LPA₂ agonistic activity; (i) four carbon chain linker, (ii) keeping the same tail group 1H-benzo[de]isoquinolyl-1,3(2H)-dione, and (iii) introducing the electron withdrawing group such as chloro meta to the carboxy group of compound 4. This approach yielded compound 11d, which is the first nonlipid specific agonist of LPA₂ with picomolar affinity.

Experimental Section

Chemistry

All compounds were analyzed by ¹H NMR, ¹³C NMR, MASS, and HRMS. The purity of final compounds was ≥95% as measured by HPLC. Full experimental procedures can be found in the Supporting Information.

Analytical Data for Compound 11d

Yield 53%. ¹H NMR (400 MHz, DMSO-d₆) d 9.13 (bs, 1H), 8.52–8.44 (m, 4H), 7.88–7.85 (m, 2H), 7.71–7.68 (m, 1H), 7.61 (d, J = 2.4 Hz, 1H), 7.43 (dd, J = 2.0 Hz, 2.4 Hz, 1H), 3.99 (t, J = 7.2 Hz, 2H), 2.74 (q, J = 6.4 Hz, 2H), 1.65–1.57 (m, 2H), 1.47–1.39 (m, 2H). ¹³C NMR (400 MHz, DMSO-d₆) d 167.67, 163.38, 136.53, 135.18, 134.29, 131.28, 130.73, 130.19, 129.23, 127.35, 127.20, 127.12, 122.01, 42.72, 40.11, 26.73, 25.01. MS (ES−) m/z 485 (M – H)⁻. HPLC purity: t_R = 2.23, 98.09%. HRMS calcd for C₂₃H₁₉ClN₂O₆S, 487.0731 (M – H)⁻; found, 487.0737.

Glossary

Abbreviations Used

LPA

lysophosphatidic acid

K₂CO₃

potassium carbonate

DMF

dimethylformamide

THF

tetrahydrofuran

RT

room temperature

SBA

sulfamoyl benzoic acid

SAR

structure–activity relationship

Supporting Information Available

Experimental details of chemical synthesis, computational and biological tests. This material is available free of charge via the Internet at http://pubs.acs.org.

Author Contributions

^∥ R.P. and J.I.F. contributed equally to this work.

The authors declare the following competing financial interest(s): G.T. and D.M. are founders and stock holders of RxBio Inc.

Funding Statement

National Institutes of Health, United States