Toll-like receptors 7 and 8 (TLR7/8) are broadly expressed on antigen-presenting cells, making TLR7/8 agonists likely candidates for the development of new vaccine adjuvants. We previously reported the synthesis of a new series of 8-oxoadenines substituted at the 9-position with a 4-piperidinylalkyl moiety and demonstrated that TLR7/8 selectivity and potency could be modulated by varying the length of the alkyl linker. In the present study, we broadened our initial structure-activity relationship study to further evaluate the effects of N-heterocycle ring size, chirality, and substitution on TLR7/8 potency, receptor selectivity, and cytokine (IFNα and TNFα) induction from human peripheral blood mononuclear cells (PBMCs). TLR7/8 activity correlated primarily to linker length and to a lesser extent to ring size, while ring chirality had little effect on TLR7/8 potency or selectivity. Substitution of the heterocyclic ring with an aminoalkyl or hydroxyalkyl group for subsequent conjugation to phospholipids or antigens was well tolerated with the retention of both TLR7/8 activity and cytokine induction from human PBMCs.
Toll-like receptors 7 and 8 (TLR7/8) are broadly expressed on antigen-presenting cells, making TLR7/8 agonists likely candidates for the development of new vaccine adjuvants. We previously reported the synthesis of a new series of 8-oxoadenines substituted at the 9-position with a 4-piperidinylalkyl moiety and demonstrated that TLR7/8 selectivity and potency could be modulated by varying the length of the alkyl linker. In the present study, we broadened our initial structure-activity relationship study to further evaluate the effects of N-heterocycle ring size, chirality, and substitution on TLR7/8 potency, receptor selectivity, and cytokine (IFNα and TNFα) induction from human peripheral blood mononuclear cells (PBMCs). TLR7/8 activity correlated primarily to linker length and to a lesser extent to ring size, while ring chirality had little effect on TLR7/8 potency or selectivity. Substitution of the heterocyclic ring with an aminoalkyl or hydroxyalkyl group for subsequent conjugation to phospholipids or antigens was well tolerated with the retention of both TLR7/8 activity and cytokine induction from human PBMCs.
Immunization is one
of the most successful and cost-effective public
health interventions to prevent infections. According to the World
Health Organization, vaccinations currently prevent between 2 and
3 million deaths every year. Despite this success, much development
work is still needed to provide effective vaccines for populations
with impaired immune function (pediatric and elderly populations),
prevent or treat global diseases where no effective vaccine exists
(human immunodeficiency virus, tuberculosis), and provide effective
solutions for newly emerging pathogens such as Zika and Chikungunya
viruses. The effectiveness of vaccines can be improved using adjuvants
to enhance, accelerate, and/or prolong a specific immune response.
Despite many years of research and development, few adjuvants are
currently approved for human use. Alum,[1] the oil-in-water emulsions MF59[2] and
AS03,[3] monophosphoryl lipid A,[4] AS01,[5] AS04,[6] CpG,[7] and virosomes[8] have been approved in Europe and/or the USA and
have an acceptable safety profile and are considered effective. The
water-in-oil emulsion Montanide ISA51[9] and
the synthetic TLR4 agonist CRX-529[10] have
also been approved in Cuba and Argentina, respectively. While these
adjuvants are effective at increasing humoral immunity, adjuvants
capable of eliciting durable cell-mediated immune responses are still
needed.Pattern-recognition receptor (PRR) ligands have been
extensively
investigated as vaccine adjuvants[11,12] because of
their critical role in innate immunity and their ability to shape
downstream adaptive immunity. PRRs are differentially localized on
many cell types and act as sentinels to recognize a wide range of
exogenous pathogen-associated molecular patterns (PAMPs) and endogenous
damage-associated molecular patterns. Among the five major families
of PRRs[13] characterized so far, Toll-like
receptors (TLRs), the first PRRs to be discovered,[14] are the most widely investigated. Each TLR is composed
of an ectodomain with leucine-rich repeats that mediate PAMP recognition,
a transmembrane domain, and a cytoplasmic Toll/IL-1 receptor domain
that initiates downstream signaling. Upon the recognition of PAMPs,
TLRs induce a signal transduction cascade leading to the production
of cytokines and the activation and recruitment of cellular mediators,
critical for initiating innate and adaptive immune responses. Among
the 10 known TLRs in humans, TLR1, 2, 4, 5, and 6 recognize bacterial
components, while TLR3, 7, and 8 and TLR9 detect viral RNA and unmethylated
DNA, respectively.[15] Among the different
TLRs, TLR7 and TLR8, which recognize uridine- and/or guanosine-rich
viral ssRNA, are broadly expressed on dendritic cells (DCs) and other
antigen-presenting cells, making TLR7/8 agonists especially valuable
for the development of vaccine adjuvants.[16] HumanTLR7 is mainly expressed on plasmacytoid DC and B cells and
its activation induces IFNα via the IRF7 pathway[17] and proinflammatory cytokines via the NFκB
pathway. HumanTLR8 is mainly expressed on monocytes, macrophages,
neutrophils, and conventional DCs, including a small population of
BDCA3+ cells shown to be very effective at cross-presenting
exogenous antigens to CD8 T cells,[18,19] and its activation
induces proinflammatory cytokines such as TNFα and IL-12p70.
While some TLR7/8 agonists are effective vaccine adjuvants in vivo,[16,20−22] other vaccine studies have shown a lack of adjuvant
activity with some small-molecule TLR7/8 agonists.[16,23,24] Additionally, conventional vaccination routes
(subcutaneous, intramuscular, and intranasal), and oral and topical
preparations of small-molecule TLR7/8 agonists can lead to serious
side effects due to systemic cytokine distribution, and some clinical
trials have been suspended over safety concerns.[25−29] Consequently, the development of safer and more effective
TLR7/8 agonists as vaccine adjuvants is needed.Several small-molecule
TLR7/8 agonists mimicking the natural viral
ssRNA ligands have been identified,[30−32] including 1H-imidazo[4,5-c]quinolines[33] (Resiquimod, Figure ) and 8-oxoadenines[34] (SM360320, Figure ). Numerous reports
on the structure–activity relationship (SAR) of 8-oxoadenines
substituted at the 9-position with arylmethyl and heteroarylmethyl
groups have been published.[35−42]
Figure 1
Representative
structures of imidazoquinolines and oxoadenines,
and oxoadenines 1–8.
Representative
structures of imidazoquinolines and oxoadenines,
and oxoadenines 1–8.Except for a few oxoadenines substituted at the 9-position
with
cycloalkyl groups and reported as inducing weak or diminished IFN
induction,[34,38] no systematic studies had been
performed on oxoadenines substituted with heterocyclic groups until
a recent publication describing the development of a highly potent
oxoadenineTLR7 agonist for the treatment of asthma (GSK2245035, Figure ).[43] This work investigated the effect of N- and O-heterocycles
at the N-9-position of the oxoadenine scaffold on IFNα and TNFα
induction. Except for two derivatives, the nitrogen atom of the N-heterocycle
was linked to the oxoadenine or substituted with an alkyl group. As
reported herein, we synthesized and evaluated N-heterocyclic oxoadenines
of general structures 1–8 (Figure ) containing an unsubstituted
amine or hydroxyl group. Substitution of the aromatic amino group
abolishes IFN induction,[44] whereas the
NH (or OH) group in oxoadenines 1–8 provides an alternative option for subsequent N- or O-derivatization
and conjugation. Conjugation of TLR7/8 agonists to proteins,[23,45−49] peptides,[50] and other molecules[51,52] has been used to improve immune responses and decrease the toxicity
of the TLR7/8 agonists. Furthermore, since TLR7 and TLR8 are located
in endosomal/lysosomal compartments,[53] cellular
uptake is a prerequisite for activation by TLR7/8 ligands. Thus, there
is considerable interest in strategies that will increase the penetration
of the TLR7/8 ligand into DCs and other immune cells while concomitantly
reducing systemic toxicity. Lipid conjugation of nucleoside drugs
including TLR7/8 agonists[54,55] is one strategy known
to facilitate endocytosis and decrease toxic side effects. Such nucleolipids
can be incorporated into liposomes and other biodegradable nanoparticles
to help protect the drug from degradation, target the TLR7/8 ligand
to DCs, and further reduce toxicity through a depot effect.[56]In the course of our own SAR program aimed
at developing safe and
effective TLR7/8 agonists as vaccine adjuvants, we previously synthesized
and evaluated a series of seven new 8-oxoadenines 1 (Figure ) substituted at
the N-9 with a 4-piperidinylalkyl moiety (n = 0–6).[57] We demonstrated that TLR7/8 selectivity/potency
and cytokine induction could be modulated by varying the length of
the alkyl linker. In this paper, we expanded upon our initial SAR
study in the 2-butoxy-oxoadenine series by investigating the effect
of different N-heterocycles (azetidine, pyrrolidine, piperidine, and
piperazine), N-heterocycle chirality, and substitution in combination
with different linker lengths on TLR7/8 potency and receptor selectivity
and on cytokine (IFNα and TNFα) induction from adult human
peripheral blood mononuclear cells (hPBMCs). The goal of this study
was to optimize the substituent at the N-9-position of the oxoadenine
scaffold and select lead TLR7-selective and TLR7/8 active oxoadenines
for further optimization of the substituent at the 2-position and
subsequent lipidation to generate new and safe TLR7/8 agonists.
Results
and Discussion
Chemistry
8-Oxoadenines 2–4 and 5(58) were synthesized
in two steps by alkylation of the common advanced intermediate 2-n-butoxy-8-methoxyadenine 9 with the requisite N-t-butoxycarbonyl (Boc)-protected alkyl
bromides or iodide (11a), either purchased commercially
or prepared from the corresponding alcohols using Appel conditions
(Ph3P/CBr4; 12c, 12f), and acidic deprotection (Scheme ), as previously described.[57] The common advanced intermediate 9 was prepared[57] in six steps from commercially available 2,6-dichloropurine.
Alkylation of 9 with alkyl bromides led predominantly
to the N-9-alkylated product (based on numerous reports
in the literature indicating that alkylation of 6-amino-adenines with
alkyl/aryl bromides in the presence of potassium carbonate predominantly
leads to the N-9 regioisomer[36,38−40,43,59]and to a lesser extent to the N-7-regioisomer).
The formation of up to 30% of the N-7-regioisomer
was also observed. In most cases, the N-7-regioisomer
was removed by chromatography after the alkylation step or after the
acidic deprotection step (for compounds 2a,b, 4a, 4d,e). The N-substituted 4-piperidinylmethyloxoadenines 6a–d were prepared in
two steps by alkylation of the 4-piperidinylmethyl oxoadenine 1a(57) with the corresponding O-t-butyldimethylsilyl (TBS)- or N-Boc-protected alkyl bromides, and acidic deprotection
(Scheme ).
Scheme 1
Synthesis
of Oxoadenines 3–5
Reagents
and conditions: (i)
K2CO3, Br(CH2)NHBoc, dimethylformamide (DMF), 50 °C, 16 h, 56–70%;
(ii) 4 N HCl/dioxane, CH3OH, rt, 1 h, 50–86%; (iii)
K2CO3, DMF, 50 °C, 16 h, 38–93%.
Scheme 2
Synthesis of Oxoadenines 6a–d
Reagents and conditions: (i)
K2CO3, Br(CH2)pR, DMF, 50 °C,
2–7 h, 67–83%; (ii) 4 N HCl/dioxane, CH3OH,
rt, 1 h, 74–100%.
Synthesis
of Oxoadenines 3–5
Reagents
and conditions: (i)
K2CO3, Br(CH2)NHBoc, dimethylformamide (DMF), 50 °C, 16 h, 56–70%;
(ii) 4 N HCl/dioxane, CH3OH, rt, 1 h, 50–86%; (iii)
K2CO3, DMF, 50 °C, 16 h, 38–93%.
Synthesis of Oxoadenines 6a–d
Reagents and conditions: (i)
K2CO3, Br(CH2)pR, DMF, 50 °C,
2–7 h, 67–83%; (ii) 4 N HCl/dioxane, CH3OH,
rt, 1 h, 74–100%.The piperazinylethyloxoadenine 7a was prepared by
alkylation of 9 with the requisite N-Boc-protected piperazinylethyl bromide followed by acidic deprotection
(Scheme ). Synthesis
of the piperazinylbutyl oxoadenine 7b by alkylation of 9 with N-Boc-protected piperazinylbutyl bromide
could not be attempted because bromination (Appel conditions) of N-Boc-protected piperazinylbutyl alcohol (prepared in a
48% yield by alkylation of N-Boc-piperazine with
1-bromo-4-butanol) failed. Instead, the synthesis of 7b was accomplished in three steps by first introducing the butyl linker
in 9 by alkylation with 1-bromo-4-chlorobutane, followed
by alkylation of the resulting chloro-adenine 15 with N-Boc-protected piperazine, and subsequent acidic deprotection
(Scheme ). In the
same manner, oxoadenines 8a,b were prepared in three
steps by alkylation of 9 with 1,2-dibromoethane followed
by alkylation of the resulting bromoadenine 17 with the
requisite 4-N-Boc-protected- or 4-hydroxy-piperidine
and acidic deprotection (Scheme ).
Scheme 3
Synthesis of Oxoadenines 7 and 8
Reagents and conditions: (i)
K2CO3, DMF, 50 °C, 16–120 h, 48–96%;
(ii) 4 N HCl/dioxane, CH3OH, rt, 1 h, 77–98%.
Synthesis of Oxoadenines 7 and 8
Reagents and conditions: (i)
K2CO3, DMF, 50 °C, 16–120 h, 48–96%;
(ii) 4 N HCl/dioxane, CH3OH, rt, 1 h, 77–98%.
Biological Activity
Oxoadenines 1–8 were assessed for human (h) TLR7 and
TLR8 specificity and
potency in cell lines and primary human cells and compared to the
known benchmarks imidazoquinoline R848 (Resiquimod, a dual TLR7/8
agonist) and Sumitomo oxoadenine SM360320 (TLR7 agonist). HumanTLR7/8
potency and specificity were evaluated 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. The HEK
reporter assay only measures activation through the NFκB pathway
and thus the IRF7 pathway activation by TLR7 agonists is not detected
in this assay. Of note, our previously published report with some
of the compounds described herein used in-house derived stably transfected
HEK293 cells.[57] The current study uses
commercially obtained HEK293-hTLR7 and hTLR8NFkB-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.
Freshly isolated primary human PBMCs were used to confirm HEK assay
responses and assess the cytokine induction (IFNα and TNFα)
ability of oxoadenines 1–8 and benchmarks.
We first focused our efforts on the synthesis and evaluation of unsubstituted
N-heterocyclic oxoadenine derivatives 2–5 to investigate the effect of linker length (0–4 carbons),
ring chirality, and ring size (4- to 6-member ring) on both hTLR7
and hTLR8 activity and cytokine induction.
Carbon Linker Length
Regardless of the N-heterocycle
size, hTLR7 activity and potency increased with increasing linker
length in the HEK293-hTLR7NFκB reporter assay (Table and Figure S1). Going from no linker to 1-carbon linker in the azetidine
(3a to 3b) and pyrrolidine (4a,d to 4b,e) series increased
TLR7 potency by 15- to 28-fold, and adding a second carbon (3c, 4c, and 4f) further increased
TLR7 potency by 2- to 5-fold (Table ). In the piperidine series (1), this
effect was more pronounced with a 22-fold potency increase between
the 1C- and 2C-linker oxoadenines (1a and 1b) and another 1.4-fold increase between the 2C- and 4C-linker oxoadenines
(1b and 1c). We have previously reported
that the 0C-linker derivative in the piperidine series was inactive
and observed a similar trend in potency increase with linker length.[57] This observation held true in the aminoalkyl
series with a 158-fold increase in potency between the aminoethyl
and aminobutyl oxoadenines 2a and 2b (Table and Figure S1).
Table 1
Chemical Structures,
TLR7 and TLR8
Activities, and Cytokine Induction of 8-Oxoadenines
PL (peak level) at 10 μM.
PL at 2 μM. MEC (minimum effective
concentration; lowest dose tested that induced cytokine), and PL and
PC (peak concentration) are from one out of three donors. TLR7 and
TLR8 EC50s are the mean values and standard deviation of
three independent experiments.
PL (peak level) at 10 μM.PL at 2 μM. MEC (minimum effective
concentration; lowest dose tested that induced cytokine), and PL and
PC (peak concentration) are from one out of three donors. TLR7 and
TLR8 EC50s are the mean values and standard deviation of
three independent experiments.The aminobutyl (2b), piperidinyl-ethyl (1b) and -butyl (1c), and pyrrolidinylethyl (4c) oxoadenines were as potent or more potent than the imidazoquinoline
R848 and oxoadenine SM360320 benchmarks. As expected, oxoadenines 1–4 were weaker hTLR8 agonists with EC50 >100 μM (Table and Figure S1), except
for the
aminobutyl oxoadenine 2b displaying an EC50 of 59 μM in the HEK293-hTLR8NFκB reporter assay. A
trend toward higher hTLR8 activity for the 1C-linker (1a, 3b, and 4e) was observed.Regardless
of the N-heterocycle, the minimum effective concentration
(MEC) for IFNα induction decreased with increasing linker length
(Table and Figure S2), although the peak concentration (PC)
and level (PL) of IFNα did not correlate with the MEC values
(Table ). We[57] and others[43] have
previously shown optimal IFNα induction from 8-oxoadenines bearing
4- to 6-carbon linkers at the N-9-position. Interestingly, although
the ethyl and butyl derivatives (1b,c, 3c, 4c, and 4f,) were more potent IFNα
inducers (lower MEC), the methyl derivatives (1a, 3b, 4b, and 4e) induced higher IFNα
peak levels. Oxoadenines 1–5 were
more potent IFNα inducers than SM360320 and R848, except oxoadenines 2a (aminoethyl), 3a,b (azetidine), and 4d (pyrrolidine) that were less potent IFNα inducers
than R848. Oxoadenines with no carbon linker did not induce measurable
TNFα in the dose range evaluated (Table and Figure S2), indicating that azetidinyl and pyrrolidinyl oxoadenines 3a, 4a, and 4d are specific inducers
of IFNα over TNFα in the dose range evaluated. We previously
reported that the 0C-linked piperidine oxoadenine was unable to induce
either IFNα or TNFα in hPBMC.[57] Although the 1C-linker derivatives 1a and 3b and the aminobutyl oxoadenine 2b were the most potent
hTLR8 agonists of this series (EC50 < 200 μM)
in the HEK293-hTLR8NFκB reporter assay, longer carbon linkers
led to lower MEC and higher TNFα peak levels (Table and Figure S2). This trend was previously reported in other piperidinylalkyl
oxoadenine series with 5- or 6-carbon linker oxoadenines being more
potent TNFα inducers.[43,57] This result is not
unexpected since the longer linker derivatives were the most potent
NFκB inducers in the HEK293-hTLR7 reporter assay. Thus, TNFα
induction observed in hPBMCs for oxoadenines 1b,c most
likely originates from the hTLR7/NFκB pathway, while TNFα
induced by oxoadenine 2b most likely originates from
both hTLR7/NFκB and hTLR8 pathways. Oxoadenines 1–4 were less inflammatory than R848, and only
piperidinyl oxoadenines 1b,c and aminobutyl oxoadenine 2b were more inflammatory than SM360320. These trends were
shared among the three donors tested, although donor-to-donor differences
in terms of peak level and minimum effective dose were observed.
Heterocycle Chirality
Both (R) and
(S) isomers of 3-pyrrolidinylalkyl (4a,d, 4b,e, 4c,f) and 3-piperidinylalkyl (5a,b)
oxoadenines were synthesized and evaluated to investigate the effect
of the N-heterocycle chirality on biological activity. In the pyrrolidine
series, the (R) isomers 4a–c were
1.6- to 4-fold more potent in the HEK293-hTLR7NFκB reporter
assay than the corresponding (S) isomers 4d–f (Table and Figure S1). In the piperidinemethyl
series, the (S) isomer 5b was about
2.4-fold more potent than the corresponding (R) isomer 5a. These results indicate that the chirality of the 9-substituent
has no significant influence on hTLR7 activity. Comparing the 3-piperidinemethyl
isomers 5a and 5b to the 4-piperidinemethyl 1a also indicates that the position (3 versus 4) of the heterocyclic
N atom has little influence on hTLR7 activity. While the ring chirality
had little effect on hTLR8 activity (HEK293-hTLR8NFκB reporter
assay) in the pyrrolidine series (4a–f), the (S)-3-piperidinylmethyl derivative 5b was more hTLR8 active than the corresponding (R) isomer 5a. The position of the N atom in the piperidine
ring (3 versus 4) had a larger influence on hTLR8 activity with the
4-piperidinemethyl oxoadenine 1a being more hTLR8 potent
than either 3-piperidinemethyl isomers 5a and 5b.In the pyrrolidine series, the 0C-linker (R) isomer 4a was more potent at inducing IFNα from
human PBMCs (lower MEC and higher PL, Table and Figure S2) than the corresponding (S) isomer 4d, but this potency increase for the (R) isomer vanished
when increasing the linker length to 1C and 2C (4b/4e and 4c/4f). In the piperidinemethyl
series, both (R) and (S) isomers 5a and 5b displayed the same MEC and PL for IFNα induction,
while the 4-piperidinemethyl 1a, which had the same MEC
as the 3-piperidinemethyl 5a and 5b, induced
the highest IFNα peak level (Table and Figure S2). No significant difference in TNFα secretion was observed
among the isomer pairs. In conclusion, the chirality of the N-heterocyclic
substituent had a negligible effect on hTLR7/8 activity in the HEK293-hTLR7/8NFκB reporter assay and on IFNα/TNFα induction from
primary human PBMCs.
Ring Size
We also examined the effect
of the ring size
(azetidine, pyrrolidine, and piperidine) on biological activity. In
the ethyl linker series, there was a clear increase in hTLR7 potency
(HEK293-hTLR7NFκB reporter assay) with increasing ring size
(Table and Figure S3). Replacing the primary amine group
in 2a with an azetidine ring (3c) increased
hTLR7 potency by 42-fold. Increasing the ring size from azetidine
(3c) to pyrrolidine (4c,f)
further increased hTLR7 potency by 3-fold. This trend held true for
both azetidine and pyrrolidine derivatives regardless of the linker
length (n = 0–2) with the pyrrolidine derivatives 4a,d and 4b,e being
2- to 7-fold more potent than the corresponding azetidine derivatives 3a and 3b (Table and Figure S3).There
was a correlation between IFNα induction and ring size. The
MEC and PC for IFNα induction decreased with increasing ring
size, while IFNα PL increased with increasing ring size, regardless
of the carbon linker (Table and Figure S4). In contrast, the
ring size had a negligible effect on TNFα induction from primary
human PBMCs.
Heterocycle Substitution
After investigating
the effect
of the linker length, ring chirality, and size on hTLR7/8 activity
and cytokine induction, we selected the piperidine-substituted oxoadenine
scaffold to further investigate the effects of substituting the piperidine
ring with amino- and hydroxyl-alkyl groups. We are interested in introducing
a functional group onto the oxoadenine scaffold that will retain or
improve the desired hTLR7 activity while allowing subsequent conjugation
to phospholipids or antigens to improve efficacy and safety when used
as a vaccine adjuvant. The 4-piperidinemethyl oxoadenine 1a was first substituted at the 1-position of the piperidine ring with
a hydroxyethyl (6a), hydroxybutyl (6b),
hydroxyhexyl (6c), or aminoethyl (6d) group.
Introduction of a hydroxyalkyl group on 1a had little
effect on HEK293-hTLR7 activity with both hydroxyethyl- and hydroxybutyl-substituted
oxoadenines 6a,b being similarly potent
than unsubstituted oxoadenine 1a and hydroxyhexyl oxoadenine 6c being 1.6-fold more potent than 1a (Table and Figure S5). Within the hydroxyalkyl series 6a–c, there was a slight trend toward increased
hTLR7 potency with increasing alkyl length. The aminoethyl oxoadenine 6d was about 3-fold less potent than the unsubstituted oxoadenine 1a and the hydroxyethyl oxoadenine 6a, indicating
that replacing the hydroxyl group in 6a with an amine
group slightly decreases TLR7 potency. Substituted oxoadenines 6a,b,d were the most potent hTLR8 agonist of the series with
EC50 < 50 μM. Oxoadenine 6a, the
most potent hTLR8 agonist of the series (EC50 = 12.5 μM),
was also the only TLR8-selective agonist of the series (Table ).The hydroxyhexyl derivative 6c had the lowest MEC for IFNα induction from PBMCs,
indicating that a longer hydroxyalkyl substituent increased IFNα
potency (Table and Figure S6). While both hydroxyethyl and hydroxybutyl
derivatives 6a,b displayed similar MEC as
unsubstituted oxoadenine 1a, they induced lower IFNα
PL. Interestingly, despite displaying the highest IFNα MEC,
the aminoethyl derivative 6d induced IFNα PL as
high as PL induced by 1a. Oxoadenines 6a–d induced much higher IFNα peak levels
than benchmarks R848 and SM360320. Both hydroxyl and aminoethyl oxoadenines 6a and 6d were the most potent TNFα inducers
of the series, inducing about 2.5-fold higher TNFα PL than unsubstituted
oxoadenine 1a. TNFα induction decreased with increasing
alkyl chain length, indicating a correlation between alkyl chain length
and TNFα induction. Oxoadenines 6a–d were weaker TNFα inducers than R848. In conclusion,
the substitution of the piperidine ring had a minimal effect on hTLR7/NFκB
activity while increasing hTLR8 activity in the HEK293 reporter assay.
Substitution of the piperidine ring had also a pronounced effect on
cytokine induction. The longer hydroxyhexyl group (6c) led to the most active IFNα inducer and lowest TNFα
inducer of the series, while both amino- and hydroxy-ethyl groups
(6a and 6d) led to higher levels of TNFα.We next investigated the introduction of an amine or hydroxyl group
at the 4-position of the piperidine ring N-linked to the oxoadenine
scaffold via an ethyl linker (8a and 8b). 8a and 8b were 21- and 16-fold less active in
the HEK293-hTLR7 cells than the 4-ethyl piperidine 1b derivative (Table and Figure S5). Since we did not test
the N-linked piperidineethyl analog of 1b, we cannot
conclude if the drop in hTLR7 potency observed for 8a and 8b is due to the introduction of the hydroxyl or
amine group, the presence of an N-linked versus a C-linked piperidine
ring, or a combination of both. The hydroxy-substituted oxoadenine 8b was more active than the corresponding amino derivative 8a in the HEK293-hTLR8 reporter assay and was one of the most
active hTLR8 agonists of the series with an EC50 of 50.4
μM (Table ).
Both 8a and 8b were weak inducers of IFNα
from primary hPBMCs when compared to 1b with an MEC for
IFNα 125-fold higher than 1b (Table and Figure S6). Although we did not test the N-linked analog of 1b, the two homologous N-linked and C-linked piperidinebutyloxoadenines have been previously described and their IFNα pEC50 reported as 8.2 and 9.2, respectively,[43] indicating a 10-fold decrease in IFNα potency when
switching a C-linked to an N-linked piperidine. In light of this observation,
it is tempting to speculate that the lower IFNα levels detected
with 8a and 8b compared to those with 1b were the result of introducing an amino or hydroxyl group
in combination with linking the piperidine ring via the N atom. The
amino derivative 8a displayed a higher IFNα MEC
than 1b but induced similar IFNα PL as 1b. Both hydroxyl and amine derivatives 8a and 8b were lower activators of TNFα (higher MEC) than 1b (Table and Figure S6).
Nitrogen Number in N-Heterocycle
We finally investigated
the replacement of the piperidine ring (1b,c) with a piperazine ring (7a,b). Both piperazineethyl
(7a) and piperazinebutyl (7b) were 12-fold
and 5-fold less active in the HEK293-hTLR7 reporter assay than the
corresponding piperidine analogs 1b and 1c, respectively (Table and Figure S5). As previously noted in
the piperidine series, increasing the linker length from an ethyl
to a butyl in the piperazinealkyl series also increased hTLR7 potency
by 3-fold. Both piperazine derivatives displayed low hTLR8 activity
in the HEK293-hTLR8 reporter assay (Figure S5). While increasing the linker length in the piperidine series (1a–c) led to more potent IFNα induction
from human PBMCs, the linker length in the piperazine series had no
effect on IFNα induction with both ethyl and butyl derivatives 7a and 7b displaying near-identical IFNα
MEC and PL (Table and Figure S6). Both 7a and 7b were less potent IFNα inducers than the ethyl piperidine 1b but more active than R848 and SM360320. As expected from
both HEK293-hTLR7 and hTLR8 assays, both piperazine derivatives also
induced lower levels of TNFα than the piperidine derivatives
(Table and Figure S6). The lower activity of the piperazine
derivatives compared to that of the corresponding piperidine derivatives
might be explained by their lower basicity. Since TLR7 receptors are
localized in endosomal compartments[30] and
require endosomal acidification for ligand recognition and signaling,[60] less basic compounds might accumulate at lower
concentrations in these acidic compartments leading to lower potencies.
Conclusions
Twenty-four oxoadenines substituted at
the N-9-position with various
N-heterocycles were synthesized in two to four steps from a common advanced intermediate to investigate
the effect of the linker length and N-heterocycle ring size, substitution
and chirality on TLR7/8 specificity, potency, and cytokine (IFNα
and TNFα) induction. Most oxoadenines 1–8 activated hTLR7 in the micromolar range and six oxoadenines
(2b, 6a–d, and 8b) activated hTLR8 with EC50 < 50 μM.
Oxoadenine 6a was also found to be TLR8-selective. The
results show that hTLR7/8 activity correlated primarily to linker
length and to a lesser extent to ring size, while ring chirality had
little effect on TLR7/8 activity. This activity was confirmed when
evaluating IFNα and TNFα induction from PBMC isolated
from healthy adult humandonors. Substitution of the heterocyclic
ring with an aminoalkyl or hydroxyalkyl group, for future conjugation
to phospholipids or antigens, was well tolerated with the retention
of TLR7 activity and slight enhancement of TLR8 activity. The broad
tolerability of a wide variety of substituents at the N-9-position
of the oxoadenine scaffold has been previously reported.[43,61] Among the structural features evaluated in this study, combining
a butyl linker with a piperidine ring led to the most potent IFNα
and TNFα inducers, while directly attaching an azetidinyl or
pyrrolidinyl ring to the oxoadenine scaffold led to selective IFNα
release in the dose range evaluated. The N-heterocycle derivatives
reported herein further advance the SAR of this important and clinically
relevant TLR7/8 agonist scaffold. In addition, the synthesis and biological
evaluation of phospholipid derivatives of certain oxoadenines presented
in this study (manuscript in preparation) and advancement toward in
vivo testing in models of vaccine adjuvant and immunotherapy will
provide additional insight into the unique biological properties of
the compounds reported herein.
Experimental Section
Solvents
and reagents were purchased and
used without further purification. Moisture- or air-sensitive reactions
were conducted under nitrogen atmosphere in oven-dried (120 °C)
glassware. Solvents were removed under reduced pressure using rotary
evaporators. Thin-layer chromatography (TLC) was performed on precoated
EMD Millipore TLC Silica Gel 60 F254 glass plates with
visualization by UV light (254 nm) and by staining with an ethanolic
acidicvanillin solution. Flash chromatography purifications were
performed using Reveleris Standard Silica Flash Cartridges on a Büchi
Reveleris X2 flash chromatography system. NMR (1H and 13C) spectra were recorded on a Varian 400-MR DD2 magnetic
resonance system in the noted solvent using tetramethylsilane as an
internal standard. Purity for all final compounds was confirmed to
be greater than 95% by LC–MS using an ACE 3 C8 column (50 ×
3.0 mm2, with a flow rate of 0.5 mL/min, a temperature
of 40 °C, and a detection at 254 nm) with a 0–100% gradient
of 0.5 M ammonium carbonate/water to 0.5 M ammonium carbonate/2-propanol
in an Agilent 6220 accurate-mass TOF equipped with an Agilent 1100
HPLC. High-resolution mass spectra were obtained on the same instrument.
Compounds were formulated at 3–5 mg/mL in 2% glycerol–10%
dimethyl sulfoxide (DMSO) in water. Up to 50% DMSO was added for 6c and 10a,b. SM360320 was formulated in DMSO.
General Procedure for N-Alkylation of CAI 9
Potassium carbonate (325 mesh, 3.0 equiv) was added
to a solution of 9 in DMF (0.25 M). The reaction mixture
was sonicated several seconds to obtain a fine suspension and stirred
at 60 °C for 1 h. After cooling to 50 °C, the alkyl bromide
or iodide (1.2 equiv) was added, and the reaction mixture was stirred
overnight at 50 °C. After cooling to rt and aqueous workup (ethyl
acetate or methyl tert-butyl ether/water), the resulting
crude was purified by chromatography on silica gel (gradient 0–10%
CH3OH in CHCl3). In some cases, the desired
9-alkylated product was isolated with up to 30% of the corresponding
7-alkylated regioisomer. These two isomers were easily separated after
the subsequent acidic deprotection step.
General Procedure for Acidic
Deprotection
The purified
alkylation product was dissolved in methanol (0.1 M) and reacted with
4 N HCl in dioxane (6.0 equiv) at rt for 1 h. The reaction mixture
was concentrated and dried under vacuum, and the residue was purified
by chromatography on silica gel (0–100% CHCl3/CH3OH/H2O 75/25/1.0 in CHCl3/CH3OH/H2O 85/15/0.5).
The title
compound was prepared according to the general N-alkylation
procedure using commercially available 1,1-dimethylethyl (2-bromoethyl)carbamate. 10a was isolated with approximately 30% of the corresponding
7-alkylated isomer as an off-white solid in an 86% yield. 1H NMR (CDCl3/CD3OD, 400 MHz), δ 5.37
(1H), 5.12 (s, 2H), 4.28 (t, J = 6.5 Hz, 2H), 4.12
(s, 3H), 4.08 (t, J = 5.4 Hz, 2H), 3.49 (m, 2H),
1.77 (p, J = 6.6 Hz, 2H), 1.51 (q, J = 7.6 Hz, 2H), 1.38 (s, 9H), 0.97 (t, J = 7.3 Hz,
3H). HRMS calculated for C17H28N6O4 [M + H]+ 381.2250; found 381.2233.
The title
compound was prepared according to the general N-alkylation
procedure using commercially available 1,1-dimethylethyl (4-bromobutyl)carbamate. 10b was isolated with approximately 30% of the corresponding
7-alkylated isomer as an off-white solid in quantitative yield. 1H NMR (CDCl3, 400 MHz), δ 5.08 (s, 2H), 4.56
(s, 1H), 4.27 (t, J = 6.6 Hz, 2H), 4.11 (s, 3H),
3.95 (t, J = 6.9 Hz, 2H), 3.15 (m, 2H), 1.76 (m,
4H), 1.36–1.53 (m, 13H), 0.96 (t, J = 7.3
Hz, 3H). HRMS calculated for C19H32N6O4 [M + H]+ 409.2563; found 409.2537.
The
title compound was prepared as an off-white solid in a 62% yield,
according to the general N-alkylation procedure using
commercially available 1-tert-butoxycarbonyl-3-iodoazetidine. 1H NMR (CDCl3, 400 MHz), δ 5.37 (s, 2H), 5.20
(m, 1H), 4.59 (t, J = 7.2 Hz, 2H), 4.27 (m, 4H),
4.14 (s, 3H), 1.75 (m, 2H), 1.42–1.54 (m, 11H), 0.96 (t, J = 7.3 Hz, 3H). HRMS calculated for C18H28N6O4 [M + H]+ 393.2250;
found 393.2261.
The title compound was prepared according to the general N-alkylation procedure using commercially available (S)-3-bromo-pyrrolidine-1-carboxylic acid tert-butyl ester. 12a was isolated with approximately 30%
of the corresponding 7-alkylated isomer as a light yellow solid in
a 54% yield. 1H NMR (CDCl3, 400 MHz), δ
5.11 (bs, 2H), 5.01 (m, 1H), 4.26 (t, J = 6.6 Hz,
2H), 4.12 (s, 3H), 3.74 (m, 2H), 3.45 (m, 2H), 2.68 (m, 1H), 2.21
(m, 1H), 1.77 (m, 2H), 1.47 (s, 11H), 0.96 (t, J =
7.4 Hz, 3H). HRMS calculated for C19H30N6O4 [M + H]+ 407.2407; found 407.2431.
Triphenylphosphine (1.2 equiv) was slowly added to a cold
solution (0 °C) of (R)-3-hydroxyethyl-pyrrolidine-1-carboxylic
acid tert-butyl ester and carbon tetrabromide (1.6
equiv) in methylene chloride (0.45 M), and the resulting solution
was stirred at rt for 45 min. The concentrated reaction mixture was
directly purified by chromatography on silica gel (0–50% ethyl
acetate in heptane) to give (R)-3-bromoethyl-pyrrolidine-1-carboxylic
acid tert-butyl ester as a colorless oil in an 83%
yield. 1H NMR (CDCl3, 400 MHz), δ 3.35–3.65
(m, 5H), 3.28 (m, 1H), 2.90 (m, 1H), 2.35 (m, 1H), 1.87–2.10
(m, 3H), 1.46 (m, 9H). The title compound was prepared as a viscous
light yellow oil in an 87% yield, according to the general N-alkylation procedure using (R)-3-bromoethyl-pyrrolidine-1-carboxylic
acid tert-butyl ester. 1H NMR (CDCl3, 400 MHz), δ, 5.10 (s, 2H), 4.27 (t, J = 6.6 Hz, 2H), 4.12 (s, 3H), 3.96 (t, J = 5.8 Hz,
2H), 3.37–3.54 (m, 2H), 3.22 (m, 1H), 2.85 (m, 1H), 2.04 (m,
2H), 1.85 (m, 2H), 1.77 (m, 2H), 1.70 (m, 1H), 1.41–1.54 (m,
11H), 0.96 (t, J = 7.3 Hz, 3H).
The title compound was prepared
as a light yellow solid in a 52% yield, according to the general N-alkylation procedure using commercially available (R)-3-bromo-pyrrolidine-1-carboxylic acid tert-butyl ester. 1H NMR (CDCl3, 400 MHz), δ
5.14 (bs, 2H), 5.01 (m, 1H), 4.26 (t, J = 6.6 Hz,
2H), 4.12 (s, 3H), 3.75 (m, 2H), 3.45 (m, 2H), 2.68 (m, 1H), 2.20
(m, 1H), 1.75 (m, 2H), 1.47 (s, 11H), 0.96 (t, J =
7.4 Hz, 3H). HRMS calculated for C19H30N6O4 [M + H]+ 407.2407; found 407.2402.
1,1-Dimethylethyl (3R)-[(6-Amino-2-butoxy-8-metloxy-9H-purin-9-yl)methyl]-1-pyrrolidine
Carboxylate 12e
The title compound was prepared
according to the general N-alkylation procedure using
commercially available (S)-3-bromomethyl-pyrrolidine-1-carboxylic
acid tert-butyl ester and isolated with about 10%
of the corresponding 7-regioisomer as a light yellow solid in an 80%
yield. 1H NMR (CDCl3, 400 MHz), δ 5.11
(s, 2H), 4.27 (t, J = 6.6 Hz, 2H), 4.11 (s, 3H),
3.92 (m, 2H), 3.47 (m, 2H), 3.30 (m, 1H), 3.13 (m, 1H), 2.75 (m, 1H),
1.90 (m, 1H), 1.76 (m, 2H), 1.67 (m, 1H), 1.42–1.54 (m, 11H),
0.96 (t, J = 7.3 Hz, 3H). HRMS calculated for C20H32N6O4 [M + H]+ 421.2563; found 421.2554.
Triphenylphosphine (1.2 equiv)
was slowly added to a cold solution (0 °C) of (S)-3-hydroxyethyl-pyrrolidine-1-carboxylic acid tert-butyl ester and carbon tetrabromide (1.6 equiv) in methylene chloride
(0.45 M), and the resulting solution was stirred at rt for 45 min.
The concentrated reaction mixture was directly purified by chromatography
on silica gel (0–50% ethyl acetate in heptane) to give (S)-3-bromoethyl-pyrrolidine-1-carboxylic acid tert-butyl ester as a colorless oil in a 79% yield. 1H NMR
(CDCl3, 400 MHz), δ 3.35–3.65 (m, 5H), 3.28
(m, 1H), 2.90 (m, 1H), 2.35 (m, 1H), 1.87–2.10 (m, 3H), 1.46
(m, 9H). The title compound was prepared as a viscous yellowish oil
in a 90% yield, according to the general N-alkylation
procedure using (S)-3-bromoethyl-pyrrolidine-1-carboxylic
acid tert-butyl ester. 1H NMR (CDCl3, 400 MHz), δ 5.17 (s, 2H), 4.27 (t, J = 6.6 Hz, 2H), 4.12 (s, 3H), 3.96 (t, J = 5.9 Hz,
2H), 3.60–3.74 (m, 1H), 3.40 (m, 1H), 3.23 (m, 1H), 2.84–2.99
(m, 1H), 2.04 (m, 2H), 1.85 (m, 2H), 1.71–1.81 (m, 3H), 1.43–1.57
(m, 11H), 0.96 (t, J = 7.4 Hz, 3H).
The title compound was
prepared as a white solid in a 79% yield by alkylation of 1a with commercially available (2-bromoethoxy)-tert-butyldimethylsilane according to the general N-alkylation
procedure and a 3 h reaction time. 1H NMR (CDCl3/CD3OD, 400 MHz), δ 4.26 (t, J =
6.6 Hz, 2H), 3.72–3.80 (m, 4H), 2.97 (d, J = 11.1 Hz, 2H), 2.54 (t, J = 6.2 Hz, 2H), 2.11
(t, J = 11.0 Hz, 2H), 1.93 (s, 1H), 1.75 (m, 2H),
1.67 (m, 2H), 1.36–1.53 (m, 4H), 0.98 (t, J = 7.4 Hz, 3H), 0.89 (s, 9H), 0.06 (s, 6H). HRMS calculated for C23H42N6O3Si [M + H]+ 479.3166; found 479.3170.
The title compound was
prepared as a white solid in an 82% yield by alkylation of 1a with commercially available (6-bromohexyloxy)-tert-butyldimethylsilane according to the general N-alkylation
procedure and a 2 h reaction time. 1H NMR (CD3OD, 400 MHz), δ 4.28 (t, J = 6.6 Hz, 2H),
3.74 (d, J = 6.9 Hz, 2H), 3.63 (t, J = 6.3 Hz, 2H), 3.11 (d, J = 11.1 Hz, 2H), 2.53
(m, 2H), 2.23 (m, 2H), 2.00 (m, 1H), 1.71–1.76 (m, 4H), 1.29–1.61
(m, 12H), 0.98 (t, J = 7.4 Hz, 3H), 0.90 (s, 9H),
0.05 (s, 6H). HRMS calculated for C27H50N6O3Si [M + H]+ 535.3792; found 535.3784.
The title compound was
prepared as a white solid in a 67% yield by alkylation of 1a with commercially available 2-[(tert-butoxycarbonyl)amino]ethylbromide according to the general N-alkylation procedure
and a 7 h reaction time. 1H NMR (CDCl3/CD3OD, 400 MHz), δ 4.25 (t, J = 6.5 Hz,
2H), 3.73 (d, J = 6.3 Hz 2H), 3.38 (m, 2H), 3.20
(m, 2H), 2.88 (m, 2H), 2.44 (m, 2H), 1.94 (m, 3H), 1.75 (m, 2H), 1.66
(m, 2H), 1.34–1.53 (m, 13H), 0.97 (t, J =
7.2 Hz, 3H). HRMS calculated for C22H37N7O4 [M + H]+ 464.2985; found 454.2986.
The title compound was prepared as
a pale yellow oil in a 96% yield by alkylation of 9 with
1-bromo-4-chlorobutane (1.1 equiv) according to the general N-alkylation procedure and stirring at rt for 66 h after
bromide addition. 1H NMR (CDCl3, 400 MHz), δ
5.80 (s, 2H), 4.27 (t, J = 6.7 Hz, 2H), 4.11 (s,
3H), 3.97 (t, J = 6.7 Hz, 2H), 3.58 (t, J = 6.4 Hz, 2H), 1.93 (m, 2H), 1.76 (m, 4H), 1.48 (m, 2H), 0.96 (t, J = 7.4 Hz, 3H). HRMS calculated for C14H22ClN5O2 [M + H]+ 328.1540;
found 328.1537.
15 was reacted with DIPEA (2.0 equiv) and tert-butyl piperazine-1-carboxylate (1.2 equiv) in DMF (0.1
M) for 5
days at 60 °C. After aqueous workup (water/ethyl acetate), the
crude was purified by chromatography on silica gel (0–1.5%
CH3OH in CHCl3), and 16b was isolated
as a yellow viscous oil in a 48% yield. 1H NMR (CDCl3, 400 MHz), δ 5.09 (s, 2H), 4.27 (t, J = 6.6 Hz, 2H), 4.11 (s, 3H), 3.94 (t, J = 7.0 Hz,
2H), 3.37–3.47 (m, 4H), 2.34 (m, 6H), 1.76 (m, 4H), 1.62 (m,
2H), 1.43–1.52 (m, 11H), 0.96 (t, J = 7.4
Hz, 3H). HRMS calculated for C23H39N7O4 [M + H]+ 478.3142; found 478.3150.
The title compound was
prepared as
a pale yellow solid in a 54% yield by alkylation of 9 with 1,2-dibromoethane (1.0 equiv) according to the general N-alkylation procedure. 1H NMR (CDCl3, 400 MHz), δ 5.24 (s, 2H), 4.25–4.34 (m, 4H), 4.13
(s, 3H), 3.66 (t, J = 6.9 Hz, 2H), 1.76 (m, 2H),
1.50 (m, 2H), 0.97 (t, J = 7.4 Hz, 3H). HRMS calculated
for C12H18BrN5O2 [M +
H]+ 344.0722; found 344.0725.
The
title compound was prepared as a white solid in a 98% yield by acidic
deprotection of 18b, according to the general acidic
deprotection procedure. 1H NMR (CD3OD, 400 MHz),
δ 4.27 (t, J = 6.5 Hz, 2H), 4.22 (t, J = 5.5 Hz, 2H), 3.96 (m, 1H), 3.43–3.57 (m, 4H),
3.31 (m, 2H, obscured by methanol), 2.07 (m, 2H), 1.84 (m, 2H), 1.73
(m, 2H), 1.49 (m, 2H), 0.98 (t, J = 7.4 Hz, 3H). 13C (CD3OD, 100 MHz), δ 162.0, 154.9, 150.6,
150.2, 100.0, 68.2, 56.6, 36.2, 32.2, 20.3, 14.2. HRMS calculated
for C16H26N6O3 [M + H]+ 351.2145; found 351.2142.
Biology
hTLR7/8 NFκB
Reporter Assay
Humanembryonic kidney
(HEK) 293 cells expressing humanTLR7 or TLR8 and an NFκB-responsive
SEAP reporter gene were obtained from Novus Biologicals (Littleton,
CO) or Invivogen (San Diego, CA), respectively. These cells were maintained
in culture media of Dulbecco’s modified Eagle’s medium
(HyCone, Logan, UT), 10% heat-inactivated FBS (Corning, Manassas,
VA), and selection antibiotics (Invivogen). HEK293 cells were stimulated
for 24 h with increasing concentrations of the indicated compounds,
and culture supernatants were analyzed for NFκB activation using
the colorimetric SEAP detection kit QuantBlue (Invivogen). Values
are expressed as fold change in OD650 over vehicle-only
treated samples, and EC50s are mean values and standard
deviation of three or four independent experiments. EC50s were calculated by nonlinear curve fitting (four parameters) after
generating dose–response curves in GraphPad Prism 7.03.
Measurement
of Cytokines from hPBMCs
Human whole blood
was collected from normal healthy donors at the University of Montana
(Missoula, MT) using an institutional review board approved protocol.
All donors provided written informed consent prior to participation.
Peripheral blood mononuclear cells (PBMCs) were isolated via a Ficoll
Hypaque 1.077 gradient separation and cultured at 0.5 × 106 cells/well in 96-well tissue culture plates with RPMI-1640
media (HyCone, Logan, UT), Pen/Strep/Glutamine (HyCone, Logan, UT),
and 10% heat-inactivated FBS (Corning, Manassas, VA). Human PBMCs
were stimulated for 24 h with increasing concentrations of the indicated
compounds. Culture supernatants were analyzed for TNFα and IFNα
levels using human TNFα DuoSet (R) ELISA kit (R&D Systems,
Minneapolis, MN) and humanIFNα VeriKine ELISA kit (Pestka Biomedical
Laboratories, Inc., Piscataway, NJ).
Authors: Alyson J Smith; Yufeng Li; Hélène G Bazin; Julien R St-Jean; Daniel Larocque; Jay T Evans; Jory R Baldridge Journal: Vaccine Date: 2016-07-09 Impact factor: 3.641
Authors: Alicja Misiak; Rosanna Leuzzi; Aideen C Allen; Bruno Galletti; Barbara C Baudner; Ugo D'Oro; Derek T O'Hagan; Mariagrazia Pizza; Anja Seubert; Kingston H G Mills Journal: Vaccine Date: 2017-08-18 Impact factor: 3.641
Authors: Keith Biggadike; Mahbub Ahmed; Doug I Ball; Diane M Coe; Deidre A Dalmas Wilk; Chris D Edwards; Bob H Gibbon; Charlotte J Hardy; Stephen A Hermitage; Joanne O Hessey; Aimee E Hillegas; Stephen C Hughes; Linos Lazarides; Xiao Q Lewell; Amanda Lucas; David N Mallett; Mark A Price; Fiona M Priest; Diana J Quint; Poonam Shah; Anesh Sitaram; Stephen A Smith; Richard Stocker; Naimisha A Trivedi; Daphne C Tsitoura; Victoria Weller Journal: J Med Chem Date: 2016-02-10 Impact factor: 7.446
Authors: Christina C N Wu; Tomoko Hayashi; Kenji Takabayashi; Mojgan Sabet; Donald F Smee; Donald D Guiney; Howard B Cottam; Dennis A Carson Journal: Proc Natl Acad Sci U S A Date: 2007-02-21 Impact factor: 11.205
Authors: Hélène G Bazin; Laura S Bess; Mark T Livesay; Yufeng Li; Van Cybulski; Shannon M Miller; David A Johnson; Jay T Evans Journal: Bioorg Med Chem Lett Date: 2020-01-22 Impact factor: 2.823
Figure 1
Scheme 1
Scheme 2
Scheme 3