Optimization of 8-oxoadenines with toll-like-receptor 7 and 8 activity.

Abstract

Toll-like receptors 7 and 8 (TLR7/8) agonists are potent immunostimulants that are attracting considerable interest as vaccine adjuvants. We recently reported the synthesis of a new series of 2-O-butyl-8-oxoadenines substituted at the 9-position with various linkers and N-heterocycles, and showed that TLR7/8 selectivity, potency and cytokine induction could be modulated by varying the alkyl linker length and the N-heterocyclic ring. In the present study, we further optimized the oxoadenine scaffold by investigating the effect of different substituents at the 2-position of the oxoadenine on TLR7/8 potency/selectivity, cytokine induction and DC maturation in human PBMCs. The results show that introducing a 1-(S)-methylbutoxy group at the 2-position of the oxoadenine significantly increased potency for TLR7/8 activity, cytokine induction and DC maturation.

Graphical abstract

Despite the success of vaccines in eradicating or reducing the incidence of some infectious diseases such as smallpox, polio and measles, there is still an urgent need for the development of effective vaccines against endemic pathogenic viruses (HIV, influenza), newly emerging diseases (Zika, Chikungunya viruses) and global killers (tuberculosis and cancer). The development of new adjuvants that are able to shape, enhance and prolong specific immune responses are critical to the success of new vaccines. Adjuvants currently approved as part of licensed vaccines¹ are safe and effective at increasing humoral immunity, but have limited capacity to generate potent and durable cell-mediated immune response, which is critical for the protection against many viral diseases and cancers.²

Toll-like receptors (TLR) are expressed on immune cells and recognize many different exogenous pathogen-associated molecular patterns (PAMPs). Because TLRs play a crucial role in innate and adaptive immunity, their agonists are being extensively investigated as vaccine adjuvants.^3,4 Upon recognition of PAMPs, TLRs trigger the induction of cytokines and costimulatory molecules, leading to the recruitment of cellular mediators critical for initiating innate and adaptive immune responses. There are ten known human TLRs including five TLRs (1, 2, 4, 5, 6) that sense bacterial components and four TLRs (3, 7, 8, 9) that are located in cytoplasmic compartments and recognize viral RNA (TLRs 3, 7, 8) and unmethylated DNA (TLR9).⁵ Human TLR7 is mostly found on plasmacytoid DC and B cells and upon activation triggers the IRF7 and NFκB pathways, leading to the induction of IFNα and proinflammatory cytokines, respectively. Expression of human TLR8 mainly occurs on monocytes, macrophages, neutrophils and conventional DCs and its activation leads to proinflammatory cytokines. 1H-imidazo[4,5-c]quinolines⁶ (Imiquimod and Resiquimod, Figure 1) and 8-hydroxyadenines⁷ (SM360320 and GSK2245035, Figure 1) are small molecule mimetics of ssRNA that activate TLR7/8. Imiquimod is the only approved TLR7/8 agonist and is used for the topical treatment of certain skin conditions.⁸ The use of imiquimod and resiquimod as vaccine adjuvants has led to mixed results.⁴ Some clinical trials have also been suspended over safety concerns^9–13 following serious side-effects observed upon administration of TLR7/8 agonists due to rapid systemic cytokine distribution. Thus there is a need to develop safer and more effective TLR7/8 agonists as vaccine adjuvants.

Figure 1.

Structures of known imidazoquinolines and oxoadenines, and oxoadenines 1–4.

In the course of our own program aimed at developing safe and effective TLR7/8 agonists as vaccine adjuvants, we synthesized and evaluated a series of twenty-seven new 2-O-butyl 8-oxoadenines substituted at the N-9 position with different N-heterocycles linked to the oxoadenine via alkyl linkers.^14,15 We demonstrated that TLR7/8 selectivity/potency and cytokine induction could be modulated by varying the length of the alkyl linker and correlated to the N-heterocycle ring size while the ring chirality had little effect on biological activity. We also reported that the N-heterocycle ring could be further substituted with amino- or hydroxyl-alkyl groups (Scheme 1, 3a and 4a) without negatively impacting biological activity.¹⁵ We are interested in developing such N-heterocyclic oxoadenines for several reasons. In addition to their water-soluble salt forming ability, the NH and OH groups in oxoadenines 1–4 (Figure 1) can be further derivatized. Derivatization that can increase the cellular uptake of TLR7/8 agonists into endosomal/lysosomal compartments of DCs where TLR7 and TLR8 reside¹⁶ are of considerable interest. To this end, we are investigating lipid conjugation of imidazoquinolines^17,18 and oxoadenines since lipid conjugation of nucleoside drugs, including TLR7/8 agonists,^19,20 is known to facilitate endocytosis and decrease toxic side effects. Additionally, formulation of such nucleolipids into liposomes and other nanoparticle systems is expected to help protect the drug from degradation, further reduce toxicity and improve immunogenicity via a depot effect.²¹

Scheme 1.

Reagents and conditions: (i) t-BuONa, R-OH, 100 °C (6a,d) or 70 °C (6b-c), 1–4 days, 58–85%; or TBSO(CH₂)₄OH, t-BuONa, 55 °C, 2.5 days, 63% (6e); or n-BuNH₂, n-PrOH, 130 °C, 8 h, 96% (6f); (ii) NBS, CHCl₃, 0 °C to rt, 4 h; (iii) NaOMe, CH₃OH, Δ, 4 h, 75–78% (2 steps); (iv) TFA, CH₃OH, rt, 3 days, 83–90%; (v) K₂CO₃, 4-bromoalkyl-N-Boc-piperidine, DMF, 50 °C, 16 h; (vi) 4 N HCl/dioxane, CH₃OH, rt, 1 h, 61–90% (2 steps); (vii) K₂CO₃, Br(CH₂)₂OTBS or Br(CH₂)₂NHBoc, DMF, 50 °C, 16 h, 53–70%.

In this letter, we report further optimization of the 8-oxoadenine scaffold by investigating the effect of different substituents at the C-2 position of the 8-oxoadnine scaffold on TLR7/8 potency and selectivity, and on cytokine (IFNα and TNFα) induction in human peripheral blood mononuclear cells (hPBMCs). The ability of these oxoadenines to induce DC maturation was also evaluated by measuring expression levels of major histocompatibility complex (MHC) molecules²² (MHC-I and HLA-DR) and co-stimulatory ligands²³ (CD80 and CD86) in hPBMC.

We first evaluated five 2-substituted analogs of the 2-O-butyl-9-methylpiperidinyl-8-oxoadenine 1a.¹⁴ The substituents investigated are 1-(S)-methylbutoxy (1b), 1-(R)-methylbutoxy (1c), methoxyethoxy (1d), 4-hydroxybutoxy (1e) and n-aminobutyl (1f). We choose to limit the linear length of the 2-substituent to 4 and 5 atoms since optimal activity of similar oxoadenines has been observed with chain length ranging from 4 to 6 atoms.^7,24 Oxoadenines 1a-f were prepared in two steps by alkylation of the corresponding intermediate 2-substituted-8-methoxyadenines 6a-f with the requisite N-t-butoxycarbonyl (Boc)-protected 1-bromomethylpiperidine and acidic deprotection (Scheme 1). Intermediates 6a-f were prepared from 2-chloro-6-amino-9-tetrahydropyran adenine 5²⁵ in 4 steps following published procedures^26–28 (Scheme 1).

Figure 3.

Fold-change of (A) HLA-DR and (B) CD86 levels in mDCs; (C) HLA-DR and (D) CD80 levels in pDCs, after 6 h stimulation with oxoadenines 1–4 compared to unstimulated cells. Data is mean values of three independent experiments in three different donors (2 donors for R848 except for CD80 with 1 donor). Error bars indicate SD.

Oxoadenines 1a-f were first assessed for human (h) TLR7 and TLR8 activity, and for cytokine induction (IFNα and TNFα) in human PBMCs, and compared to the known benchmarks imidazoquinoline resiquinod (R848, a dual TLR7/8 agonist) and Sumitomo oxoadenine SM360320 (TLR7 agonist). The hTLR7 and hTLR8 activity of the compounds was assessed by a reporter gene assay using HEK293 cells stably transfected with either hTLR7 or hTLR8 and the NFκB SEAP (secreted embryonic alkaline phosphatase) reporter. This assay measures NFκB mediated SEAP production following TLR7- or TLR8-specific activation. Since the HEK reporter assay only measures the NFκB pathway, additional assay systems would be necessary to evaluate the IRF7 pathway activation by TLR7 agonists. Of note, the TLR7/8 potency values we previously published¹⁴ for 1a and 2a were determined using an in-house derived stably transfected HEK293 cell line. The current study uses commercially obtained HEK293-hTLR7 and -hTLR8 NFkB-SEAP reporter cells from Novus Biologicals (Littleton, CO) and Invivogen (San Diego, CA), respectively, resulting in some changes in EC₅₀ values and in TLR7/8 specificity from the previously published findings.

Replacing the O atom at the 2-position (1a) with a N atom (1f) led to a two-fold reduction in hTLR7 EC₅₀ (Table 1 and Figure S1). Substituting the third C atom in the 2-O-butoxy group (1a) with an oxygen (1d) or adding a terminal hydroxyl group to the 2-O-butoxy group (1e) led to a significant decrease in hTLR7 activity, and the corresponding EC₅₀ for 1e and 1f could not be calculated within the μM dose range evaluated. Introducing a (S)-methyl group α to the 2-O atom (1b) increased hTLR7 potency seven-fold while the corresponding R-isomer 1c was about two times less potent than 1a. Oxoadenines 1a-f were less hTLR7/NFκB potent than the benchmark TLR7/8 ligands R848 and SM360320. Similar results were observed for hTLR8 activity. The 1-(S)-methyl-butoxy oxoadenine 1b was again the most potent oxoadenine of the series with over a 100-fold increase in hTLR8 potency compared to the n-butyloxy derivative 1a (Table 1 and Figure S1). Oxoadenine 1c was about two-fold less potent than 1a while oxoadenines 1d-f had negligible activity and their EC₅₀s could not be calculated in the μM range evaluated. The much larger increase of potency for TLR8 (>100-fold) over TLR7 (7-fold) observed when adding the (S)-methyl substituent to 1a indicates that changes at the 2-position of the oxoadenine scaffold have a much greater effect on the hTLR8 receptor than on the hTLR7 receptor. Oxoadenine 1b was about two-fold more hTLR8 potent than R848. As expected, these compounds did not activate hTLR3 or hTLR9, as shown by the lack of SEAP induction observed after stimulation of HEK293-hTLR3 or HEK293-hTLR9 cell lines with compounds 1a-e (Figure S2), confirming the TLR7/8 selectivity of the compounds evaluated.

Table 1.

Chemical Structures, TLR7 and TLR8 Activity and Cytokines induction of 8-Oxoadenines

Compound	R	n	R’	TLR7 EC50 (μM)	TLR8 EC50 (μM)		IFNα		TNFα
						MEC (μM)	PC (μM)	PL (pg/mL)	MEC (μM)	PLa (pg/mL)
R848	--	--	--	0.75±0.08	5.87±0.40	0.08	10	1408	2.0	19750
SM360320		--	--	1.17±0.30	--	0.08	0.08	1444	2.0	1663
1a		1	H	19.8±2.6	326±626	0.08	0.4	5899	2.0	2861
1b		1	H	2.68±0.57	2.53±1.31	0.0032	0.016	5473	0.4	12456
1c		1	H	35.8±4.6	619±115	0.08	0.4	6210	10	185
1d		1	H	>500	>500	10	10	2240	--	--
1e		1	H	>500	--	--	--	--	--	--
1f		1	H	40.0±15.2	>500	0.08	0.4	5983	--	--
2a		5	H	0.45±0.10	--	0.00064	0.016	4678	0.4	2874
2b		5	H	0.24±0.08	--	<0.00001	0.0032	2642	0.08	3015
3a		1		23.1±2.2	12.5±3.9	0.4	0.4	4731	10	2738
3b		1		4.06±0.56	2.14±0.51	0.016	0.08	4746	2.0	10257
4a		1		30.0±4.8	13.0±5.4	0.4	0.4	3045	2.0	6133
4b		1		6.71±1.13	2.13±0.61	0.016	0.08	4415	2.0	19307

EC₅₀s are mean values and SD of three (TLR7) or four independent experiments (TLR8) conducted with HEK293-hTLR7 or HEK293-hTLR8 cells treated for 24 hours with the indicated compounds.

^aPL (peak cytokine level) at 10 μM; MEC (minimum effective concentration; lowest dose tested that induced cytokine), PC (peak concentration, dose inducing the highest PL). PL, MEC and PC are representative data from one out of three independent donors.

The 1-(S)-methylbutoxy oxoadenine 1b was the most potent IFNα inducer of the series with a minimum effective concentration (MEC) 25-fold lower than observed for the butyloxy oxoadenine 1a (Table 1 and Figure S2). Butyloxy (1a), 1-(R)-methylbutoxy (1c) and butylamine- (1f) analogs displayed similar IFNα induction profile (similar MEC, peak concentration (PC) and peak level (PL)) while the methoxyethoxy oxoadenine 1d induced IFNα only at the highest dose tested. These results are consistent with the literature data.^7,24,27 The hydroxybutyloxy derivative (1e) was inactive across the dose range evaluated. Oxoadenines 1a-c,f were more potent IFNα inducers than R848 and SM360320. The IFNα production from hPBMCs, which occurs via activation of the TLR7/IRF7 pathway, cannot be correlated to the hTLR7 activity observed in the HEK293-hTLR7 assay because the HEK assay only provides a read-out for the TLR7/NFκB pathway. In terms of TNFα induction from primary human PBMCs, the 1-(S)-methylbutoxy oxoadenine 1b was also the most potent oxoadenine of the series (Table 1 and Figure S2) followed by the butyloxy oxoadenine 1a. The 1-(R)-methylpentoxy oxoadenine 1c induced a very small amount of TNFα at the highest dose tested while the remaining oxoadenines were inactive. These results mirror the hTLR8 (NF-κB) responses from the HEK293 cells. Comparing cytokine induction observed for 1a and 1f indicates that when changing the heteroatom at the 2-position from oxygen to nitrogen, IFNα induction is maintained while TNFα induction is abrogated. Oxoadenines 1a-f were less inflammatory than the benchmark imidazoquinoline compound R848.

In order to investigate if the enhancement of TLR7/8 potency and cytokine induction observed upon introducing a (S) methyl group to the 2-butyloxy chain is maintained across different 9-N substituted oxoadenines, we also synthesized the 1-(S)-methylbutoxy analogs (2b-4b) of oxoadenines 2a,¹⁴ 3a^15,29 and 4a.^15,29 Oxoadenine 2b³⁰ was prepared by N-alkylation of intermediate 6b with the requisite 1,1-dimethylethyl 4-(5-bromopentyl)-1-piperidine carboxylate¹⁴ followed by acidic deprotection (Scheme 1). Oxoadenines 3b and 4b were prepared in two steps from 1b by alkylation of the piperidinyl N atom with (2-bromoethyl)-t-butyldimethylsilane or 2- [(t-butoxycarbonyl) amino] ethyl bromide and acidic deprotection (Scheme 1). Introducing a (S)- methyl group to the 2-butyloxy chain of oxoadenines 2a-4a increased hTLR7 potency as shown by the 4- to 6-fold reduction in EC₅₀s of oxoadenines 2b-4b (Table 1 and Figure S1). This potency increase was similar to the one (7-fold) observed between 1a and 1b. hTLR8 potency of oxoadenines 3b and 4b was also enhanced 6-fold. hTLR8 activity of 2a and 2b was very low (Figure S1) and the corresponding hTLR8 EC₅₀s could not be calculated in the μM dose range evaluated. Oxoadenines 3a, 3b, 4a and 4b were slightly TLR8-selective. Both 1-(S)-methylbutoxy oxoadenines 3b and 4b had an IFNα MEC 25-fold lower than the corresponding butyloxy oxoadenines 3a and 4a, similar to the MEC decrease (25-fold) observed between 1a and 1b. While IFNα peak levels for 1-(S)-methylpentoxy oxoadenines 1b, 3b and 4b were similar or higher than the corresponding butyloxy oxoadenines 1a, 3a and 4a, the most potent 1-(S)-methylpentoxy oxoadenines 2b induced lower IFNα peak levels than the corresponding butyloxy oxoadenine 2a. Oxoadenines 2-4 were all more potent IFNα inducer than R848 and oxoadenines 2a and 2–4b were more potent IFNα inducer than SM360320. Although oxoadenines 3b and 4b displayed higher MEC than SM360320, these oxoadenines induced higher IFNα peak levels. As observed for 1a/1b, introduction of a (S)-methyl group to oxoadenines 2a and 3a led to a 5-fold decrease in TNFα MEC (Table 1 and Figure S2). While introducing the (S) methyl group to 4a did not decrease the TNFα MEC concentration of 4b, this small substitution drastically increased TNFα PL induced by 4b, and 4b was equivalent to R848 for TNFα induction in the dose range tested. Oxoadenines 2a-b had the lowest TNFα MEC but also induced the lowest TNFα peak levels.

Oxoadenines 1–4 were also evaluated for their ability to induce DC maturation at a single 0.1 μM dose in hPBMCs, by measuring the levels of major histocompatibility complex (MHC) molecules²² (MHC-I and HLA-DR) and co-stimulatory ligands²³ (CD80 and CD86) by flow cytometry. As expected, the least TLR7/8 active oxoadenines 1d,e did not stimulate the production of any of the four DC maturation markers investigated in either mDCs or pDCs (Figures 2 and 3). In mDCs, increase of MHC-I stimulation was only observed for oxoadenines 2a,b, albeit at a modest level (1.3 to 1.4-fold, Figure 2A). Oxoadenines 1b, 2a,b and 3b also led to a 1.6– 2.7- and 1.5-fold increase of CD80 stimulation in mDCs, respectively (Figure 2B), although the large variance observed between the three donors tested would require testing in additional donors to draw a firm conclusion. Oxoadenines 1–4 induced no HLA-DR or CD86 in mDCs (Figures 3A–B). While R848 was also ineffective at inducing HLA-DR (Figure 3A) and CD80 in mDCs (Figure 2B), it induced a 1.4- and 1.9-fold increase in MHC-I (Figure 2A) and CD86 (Figure 3B) levels in mDCs, respectively.

Figure 2.

Fold-change of (A) MHC-I and (B) CD80 levels in mDCs; (C) MHC-I and (D) CD86 levels in pDCs, after 6 h stimulation with oxoadenines 1–4 compared to unstimulated cells. Data is mean values of three independent experiments in three different donors (2 donors for R848 except for CD80 with 1 donor). Error bars indicate SD.

The low DC maturation activity induced by only a few of the oxoadenines tested on mDCs is not unexpected since mDCs express TLR8 and most of the oxoadenines tested are weak TLR8 agonists. In general, oxoadenines 1–4 were more active in pDCs, which was expected since these oxoadenines are TLR7 agonists and human pDCs express TLR7. In pDCs, oxoadenines 1a-c and 2–4 induced a 1.2- to 2.4-fold increase in MHC-I (Figure 2C) and HLA-DR levels (Figure 3C). Introducing a 1-(S)-methyl group to the 2-butyloxy chain of oxoadenines 3a and 4a increased MHC-I and HLA-DR stimulation by 1.5–1.7-fold. Compounds 2a,b induced a similar 1.7–2.3-fold increase in MHC-I and HLA-DR stimulation but introduction of the 2-α methyl group in 2b did not increase stimulation levels as observed with 3-4. Oxoadenines 1c-e, 3a and 4a were mostly inactive for CD86 (Figure 2D) and CD80 (Figure 3D) activation, while 1a and 1b induced a 1.4- and 2.0-fold increase of CD86 levels, respectively. As previously observed for MHC-I and HLADR upregulation, adding the 2-α methyl group to 3a and 4a led to a 1.7- to 2.3-fold increase in CD86 levels, but did not further increase the activity of 2b. The most potent oxoadenines tested were as potent as R848 with respect to MHCI and II upregulation in pDCs but less potent than R848 with respect to CD86 upregulation.

In conclusion, the data described herein show that the TLR7/8 activity of oxoadenines can be modulated by the substituent at the C-2-position. Replacing the oxygen atom at the 2-position with a nitrogen atom led to a less potent oxoadenine and abrogation of TNFα induction while introducing an oxygen or hydroxyl in the C-2 butyl side-chain drastically decreased TLR7/8 activity. Introducing an α-(S) methyl group on the 2-butyloxy side chain increased TLR7/8 potency, cytokine induction and upregulation of some DC maturation markers. The α-(S) methylated oxoadenines 3b and 4b were the most TLR8 active oxoadenines of the series with EC₅₀ < 5 μM. Oxoadenine 3b has been selected for further phospholipidation and the resulting phospholipidated oxoadenine is being evaluated as a vaccine adjuvant.