Dihydropyrrolo[2,3-d]pyrimidines: Selective Toll-Like Receptor 9 Antagonists from Scaffold Morphing Efforts

Abstract

Toll-like receptors (TLRs) play important roles in the innate immune system. In fact, recognition of endogenous immune complexes containing self-nucleic acids as pathogen- or damage-associated molecular patterns contributes to certain autoimmune diseases, and inhibition of these recognition signals is expected to have therapeutic value. We identified dihydropyrrolo[2,3-d]pyrimidines as novel selective TLR9 antagonists with high aqueous solubility. A structure–activity relationship study of a known TLR9 antagonist led to the promising compound 18, which showed potent TLR9 antagonistic activity, sufficient aqueous solubility for parenteral formulation, and druggable properties. Compound 18 suppressed the production of the proinflammatory cytokine IL-6 in CpG-induced mouse model. It is therefore believed that compound 18 has great potential in the treatment of TLR9-mediated systemic uncontrollable inflammatory response like sepsis.

Toll-like receptors (TLRs) are a family of type I transmembrane receptors and the central components of the innate immune system.^1,2 Thirteen TLRs have so far been reported as fundamental in recognition of pathogen-associated molecular patterns (PAMPs), which are expressed by microbial pathogens, or damage-associated molecular patterns (DAMPs), which are transmitted by necrotic or dying cells.^3,4 Among the reported TLRs, TLR1–10 have been identified in human. Human TLR3, TLR7, TLR8, and TLR9 are expressed on endosomal membranes in the cells and recognize pathogen-derived nucleic acid molecular patterns.⁵ However, recognition of endogenous immune complexes containing self-nucleic acids as PAMPs or DAMPs contributes to certain autoimmune diseases, such as systemic lupus erythematosus (SLE),⁶ psoriasis,⁷ arthritis, and multiple sclerosis.⁸ Therefore, inhibition of these recognition signals is expected to have therapeutic value.

TLR9 has been identified as a key receptor for innate immune response to unmethylated CpG-DNA.⁹ Accordingly, a number of TLR9 antagonists have been suggested to be potentially useful in the treatment of diseases characterized by undesired innate immune response, such as systemic autoimmune diseases, sepsis,¹⁰ Graft-versus-host disease,¹¹ and malaria infection.¹² Recently, Plitas and co-workers reported that TLR9 knockout mice with cecal ligation and puncture (CLP), as peritonitis model, showed increased bacterial clearance, decreased serum cytokine production, and increased granulocyte influx in the peritoneum as compared to wild-type animals. The researchers also showed that administration of an inhibitory CpG sequence to block TLR9 signals just before CLP treatment improves mortality in the wild-type animals. These findings have provided a rationale for the pursuit of small molecule TLR9 antagonists as potential candidates for treatment of systemic uncontrollable inflammatory responses, including sepsis. As part of our efforts to find new agents for the treatment of sepsis, we focused on TLR9 selective antagonists. To date, a number of TLR9 antagonists have been reported¹²⁻¹⁵ with promising candidates reaching clinical development.¹⁶ It was recently reported that the hydroxychloroquine 1, a TLR9 antagonist classified as an antimalarial drug and also used for SLE and rheumatoid arthritis therapy, directly blocks interaction between TLR9 and CpG-DNA.¹⁷ In addition, CPG-52364 (2), a small molecule TLR7/8/9 antagonist (the reported ratio of TLR7/TLR9 antagonism was 0.8),^18,22 has recently completed phase 1 clinical trial for SLE therapy (NCT00547014) (Figure 1).

Figure 1.

Structures of small molecule TLR9 antagonists.

Although there has been tremendous progress in the development of selective TLR9 antagonists as a small molecule, most known TLR9 antagonists are reported to also inhibit TLR7. Activation or inhibition of TLR7 and TLR9 is very complex as these receptors have opposing inflammatory and regulatory roles.¹⁹⁻²¹ Therefore, compounds that inhibit the signal cascades of both receptors may induce unwanted immunosuppression. Herein we describe the discovery of a promising TLR9 antagonist obtained by optimization of the known TLR9 antagonist 2. While 2 exhibits strong TLR9 antagonistic activity,²² it is lipophilic and has a large molecular weight to be used as lead compound. Also 2 has insufficient solubility (0.12 mg/mL, pH 7.4) for parenteral formulation. To identify a potential lead compound, we focused on compound lipophilic efficiency (LLE = pIC₅₀ – clogP) as an index because compounds with low clogP generally exhibit high solubility and good physiochemical properties.²³⁻²⁵

The pyrimidine derivatives 5a–e were synthesized as shown in Scheme 1. Displacement of the 4-chloro group in 3a–e with 4-(2-aminoethyl)morpholine afforded 4a–e. Suzuki–Miyaura cross coupling reaction with the corresponding pinacol boronates under microwave heat yielded the desired compounds 5a–e. Dihydropyrrolo[2,3-d]pyrimidine derivatives were prepared as shown in Scheme 2. The diethyl allylmalonate 6 was treated with urea followed by chlorination of 5-allylbarbituric acid using POCl₃ to give trichloro compound 7 in high yield. Oxidative cleavage of the olefin using sodium periodate in the presence of potassium osmate gave the key intermediate 8. Tandem reaction via reductive amination with various alkyl amines followed by cyclization produced the desired dihydropyrrolopyrimidines 9a–e. Suzuki–Miyaura cross coupling reaction with the corresponding pinacol boronates provided compounds 10a–e, 12a–c, and 14. The desired compounds 11a–j and 13a–c were obtained by conversion of the 4-chloro group into a methyl, methoxy, dimethylamino, or hydrogen group. The substituted piperazine derivatives 17 and 18 were synthesized from the dihydropyrrolopyrimidine 9b.

Scheme 1. Synthesis of the Pyrimidine Derivatives.

Reagents and conditions: (a) 4-(2-aminoethyl)morpholine, N,N-diisopropylethylamine or K₂CO₃, i-PrOH or DMF, 60 °C; (b) Pd(PPh₃)₄, 4-(4-methylpiperazin-1-yl)phenylboronic acid pinacol ester, 3 M NaOH, 1,4-dioxane, 120–150 °C (microwave).

Scheme 2. Synthesis of Dihydropyrrolo[2,3-d]pyrimidine Derivatives.

Reagents and Conditions: (a)(1) urea, NaOEt, EtOH, reflux; (2) POCl₃, N,N-dimethylaniline, 120 °C; (b) K₂OsO₄, NaIO₄, acetone, H₂O, rt; (c) R³NH₂, NaBH₃CN, AcOH, MeOH, rt; (d) Pd(PPh₃)₄, 4-(4-methylpiperazin-1-yl)phenylboronic acid pinacol ester, Na₂CO₃, H₂O, 1,4-dioxane, 110–120 °C (microwave); (e) for 11a–d, 2 M MeZnCl or MeZnBr/THF, Pd(t-Bu₃P)₂, THF; for 11e, NaOMe, MeOH, 120 °C (microwave); for 11f, 2 M Me₂NH/THF, N-methylpyrrolidone, 180 °C (microwave); for 11g, 11i, 11j, 10% Pd/C, H₂, trifluoroacetic acid, MeOH, rt; for 11h, 10% Pd/C, HCO₂NH₄, trifluoroacetic acid, MeOH, 50 °C; (f) Pd(PPh₃)₄, substituted phenylboronic acid pinacol ester, 3 M Na₂CO₃, 1,4-dioxane, 110–120 °C (microwave); (g) for 13a, lithium aluminum hydride, THF, reflux; for 13b, (1) 10% Pd/C, HCO₂NH₄, MeOH, 50 °C; (2) lithium aluminum hydride, THF, reflux; for 13c, (1) 4 M HCl/1,4-dioxane, MeOH, rt; (2) 10% Pd/C, HCO₂NH₄, MeOH, 50 °C; (3) 30% formaldehyde solution, NaBH₃CN, AcOH, MeOH, rt; (h)(1) Pd(PPh₃)₄, t-butyl 4-(4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)phenyl)piperazine-1-carboxylate, 3 M Na₂CO₃, 1,4-dioxane, 120 °C (microwave); (2) 4 M HCl/1,4-dioxane, MeOH; (i) 10% Pd/C, HCO₂NH₄, MeOH, 50 °C; (j) R⁴OCH₂CHO, NaBH₃CN, AcOH, MeOH, rt; (k) 1 M TBAF/THF, THF, rt.

As a first step toward exploration of a lead compound, we attempted to identify the minimal core structure of the quinazoline framework for inhibition of TLR9 signal. TLR9 antagonistic activity was evaluated in NF-κB reporter assay as shown in Table 1. This reporter assay was carried out with HEK293-human TLR9 cells transfected with pNF-κB-Luc reporter vectors. To obtain other bicyclic pyrimidines, we replaced the dimethoxyphenyl moiety of 2 with a tetrahydropyran or cyclohexane ring. The obtained fused-ring compounds were less lipophilic compared to the quinazoline core structure (2, clogP 2.96, versus 5a, 0.43, and 5b, 2.15). Although compound 5a showed better LLE value than 2, we had to optimize its selectivity against TLR7 (5a, TLR7: IC₅₀ = 50 nM). Fortunately, compound 5b showed acceptable TLR9 antagonistic activity with good selectivity against TLR7 (5b, TLR9/TLR7 = 71-fold). These findings suggested that lipophilic substituents at the 5- and/or 6-position of the pyrimidine are optimal for generation of lead compounds with high TLR9 selectivity. To reduce molecular weight, the ring of tetrahydroquinazoline 5b was simplified by removal of the cyclohexane ring to give the dimethyl pyrimidine 5e. Compound 5e showed improvement in LLE value with weaker selectivity for TLR7 than 5b (TLR7: IC₅₀ = 928 nM). On the basis of these findings, it was concluded that the pyrimidine derivatives could provide the minimum core structure for TLR9 antagonism. Next, we turned our attention to compounds 5c and 5d, which were obtained by removal of the methyl group at the 5- and/or 6-position of the pyrimidine ring. Both compounds exhibited reduced TLR9 antagonistic activity, likely due to a decrease in hydrophobic interaction and lower basicity of the pyrimidine nitrogen than 5e (Table S2, Supporting Information). Therefore, we focused on substituted or fused pyrimidine derivatives. In general, conformationally locked compounds with appropriate interaction are expected to improve the activity.²⁶ In order to reduce the number of rotatable bonds in the 4-amino group, we cyclized the 5-methyl group of compound 5e with the nitrogen at the 4-position to form the dihydropyrrolopyrimidine 11a. As expected, 11a showed potent TLR9 antagonistic activity with a lower clogP relative to compound 5e. We speculate that a fixed 4-substituted side chain led to strong interaction with the receptor. Furthermore, the metabolic stability of compound 11a was improved compared to that of 5e (human MS: 11a <0.010 vs 5e 0.069 mL/min/mg protein), probably due to interrupted oxidation of the 6-methyl group (metabolic site) by cyclization steric hindrance. As for receptor selectivity, compound 11a showed good selectivity against TLR7 (TLR9/TLR7 = 93-fold). Overall, the dihydropyrrolo[2,3-d]pyrimidine was found to be superior to other core structures in terms of TLR9 selectivity, low lipophilicity, and ligand efficiency.

Table 1. IC₅₀ Values, LLE, and clogP of the Prepared Pyrimidine Analogues.

^aAverage (N = 2).

^bMean (N = 3).

^cCalculated by ACD/LogP.

As we had a good lead compound (11a) in our hand, we looked at the SAR of the side chains. The compounds listed in Table 2 were designed and synthesized to extensively explore suitable substituents at the 7-position (R¹) and 4-position (R²) of the dihydropyrrolo[2,3-d]pyrimidine core structure. Among 11a–d, the order of activity was consistent with the basicity of the terminal amino group. This finding indicated that basicity of the amino group could be an important factor for high TLR9 antagonistic activity (Table S3, Supporting Information). Despite its strong antagonism of TLR9 (IC₅₀ = 3.5 nM), the pyrrolidine derivative 11d displayed slightly lower LLE and metabolic stability than the corresponding morpholine derivative 11b, presumably due to its higher lipophilicity (human MS: 11d 0.035 versus 11b < 0.01 mL/min/mg protein). Next, we evaluated the effects of a substitution at the 4-position of 11d. The chloro (10d), methoxy (11e), and dimethylamino (11f) compounds showed significantly reduced TLR9 antagonistic activity. Although the unsubstituted compound 11g exhibited strong activity, its metabolic stability was not sufficient (human MS: 0.071 mL/min/mg protein). These results suggested that morpholine derivatives were more favorable as drug candidates than pyrrolidine derivatives. As 4-hydogen substitution gave good result for LLE, we replaced the 4-methyl group with a hydrogen in the morpholine compounds 11a and 11b, which had high LLE value and low clogP. Remarkably, compound 11i showed strong TLR9 antagonistic activity with the best LLE value among the compounds in Table 2. In addition, the solubility of 11i (1.1 mg/mL at pH 7.4) was much improved compared to that of compound 2 (0.12 mg/mL at pH 7.4), and its TLR9 selectivity remained over 50-fold that for TLR7 (11i, IC₅₀ = 480 nM). Therefore, we selected 11i as lead compound for further investigation.

Table 2. Optimization of the Substituents in the R¹ -and R²-Positions of the Dihydropyrrolo[2,3-d]pyrimidine.

^aAverage (N = 2).

^bN = 1.

^cN = 3.

^dCalculated by ACD/LogP.

The strong TLR9 antagonistic activity and good aqueous solubility of compound 11i encouraged us to further investigate substitutions on the phenyl ring at the 2-position of the pyrimidine. It is reported that the aqueous solubility of a compound can be improved by disruption of its molecular symmetry.²⁷ Accordingly, we decided to change the position of the piperazine group to the meta- or ortho-position. Unfortunately, the resulting compounds 13a and 13b showed 6- to 7-fold decline in the antagonistic activity. Compound 13c, a keto piperidine derivative, had no TLR9 antagonistic activity. These results suggested that the basic nitrogen (an aniline group) at the para-position of the benzene ring is crucial for interaction with TLR9. Finally, to further improve the solubility of 11i for intravenous administration, the methyl group on the piperazine was replaced by a methoxyethyl or hydroxyethyl group that can act as a hydrophilic group. Among the prepared compounds, the 2-hydroxyethyl derivative 18 showed strong TLR9 antagonistic activity with high LLE value. It is noteworthy to mention here that compound 18 exhibited enough aqueous solubility for parenteral formulation (>10 mg/mL, pH 7.4) with the lowest lipophilicity (clogP = 0.65) (Table 3). Next, we determined compound 18 selectivity for TLR9. Compound 18 IC₅₀ values for inhibition of the off-target receptors TLR7 and TLR8 were 970 and >10000 nM, respectively. Compound 18 IC₅₀ (TLR7)/IC₅₀ (TLR9) ratio was 75-fold. In in vitro experiments, compound 18 inhibited CpG-induced IL-6 production in mouse spleen and human peripheral blood mononuclear cells (PBMC) with IC₅₀ values of 74 and 244 nM, respectively (Table 4). On the basis of these results, we decided to evaluate the efficacy of compound 18in vivo by measuring CpG-induced IL-6 production in the peritoneal lavage fluid (PLF) and plasma of mice. To our delight, compound 18 inhibited CpG-induced IL-6 production in a dose-dependent manner (ED₅₀(plasma) = 0.8 mg/kg; Figures 2, S1 and S2, Supporting Information).

Table 3. Optimization of the Substituents in the R³-Position of the Dihydropyrrolo[2,3-d]pyrimidine.

^aAverage (N = 2).

^bN = 1.

^cCalculated by ACD/LogP.

Table 4. In Vitro Profile of Compound 18.

evaluation	IC50 (nM)a
human TLR9	13
human TLR7	970b
human TLR8	>10000b
CpG-induced IL-6 inhibition in mouse spleen	74
CpG-induced IL-6 inhibition in human PBMC	244b

^aAverage (N = 2).

^bN = 1.

Figure 2.

Compound 18 dose-dependent inhibition of CpG-induced IL-6 production in mice (N = 4).

In summary, we describe here the discovery of dihydropyrrolo[2,3-d]pyrimidine derivatives as novel TLR9 antagonists. The representative compound 18 possessed high TLR9 selectivity with excellent aqueous solubility and showed remarkable efficacy in CpG-induced mouse model. Although further efforts are required to assess the in vivo safety profile of this compound, it is believed that compound 18 has great potential in the treatment of TLR9-mediated systemic uncontrollable inflammatory response like sepsis. Besides, compound 18 would be a useful reagent for studying the physiological roles of TLR9.

Glossary

Abbreviations

TLR9

toll-like receptor 9

PAMP

pathogen-associated molecular patterns

DAMP

damage-associated molecular patterns

SLE

systemic lupus erythematosus

CLP

cecal ligation and puncture

LLE

lipophilic ligand efficiency

TBS

tert-butyldimethylsilyl

THF

tetrahydrofuran

TBAF

tetrabutylammonium fluoride

ACD

advanced chemistry development

PBMC

peripheral blood mononuclear cells

Supporting Information Available

Synthetic procedures, analytical data, and procedures for all biological assays. This material is available free of charge via the Internet at http://pubs.acs.org.

Author Contributions

The manuscript was written through all authors contributions. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.