The discovery of potent small molecule activators of human STING

Article abstract of DOI:10.1016/j.ejmech.2020.112869

The adaptor protein STING plays a major role in innate immune sensing of cytosolic nucleic acids, by triggering a robust interferon response. Despite the importance of this protein as a potential therapeutic target for serious unmet medical conditions inc

Full text of DOI:10.1016/j.ejmech.2020.112869

Journal Pre-proof
The Discovery of Potent Small Molecule Activators of Human STING
David C. Pryde, Sandip Middya, Monali Banerjee, Ritesh Shrivastava, Sourav Basu,
Rajib Ghosh, Dharmendra B. Yadav, Arjun Surya
PII:
S0223-5234(20)30841-2
DOI:
https://doi.org/10.1016/j.ejmech.2020.112869
EJMECH 112869
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 14 July 2020
Revised Date: 21 September 2020
Accepted Date: 21 September 2020
Please cite this article as: D.C. Pryde, S. Middya, M. Banerjee, R. Shrivastava, S. Basu, R. Ghosh,
D.B. Yadav, A. Surya, The Discovery of Potent Small Molecule Activators of Human STING, European
Journal of Medicinal Chemistry, https://doi.org/10.1016/j.ejmech.2020.112869.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
©
2020 Elsevier Masson SAS. All rights reserved.

Graphical abstract
Cl
F
O
N
N
F
H
O
F
F

The Discovery of Potent Small Molecule Activators
of Human STING
1
,*
2
2
2
2
David C. Pryde , Sandip Middya , Monali Banerjee , Ritesh Shrivastava , Sourav Basu , Rajib
2
2
2
Ghosh , Dharmendra B. Yadav , Arjun Surya
1
Curadev Pharma Ltd., Innovation House, Discovery Park, Ramsgate Road, Sandwich, Kent,
2
CT13 9ND, UK. Curadev Pharma Pvt. Ltd., B-87, Sector 83, Noida 201305, Uttar Pradesh,
India
Keywords
STING, interferon genes, immunotherapy, cytokines.
Corresponding Author
*
David C. Pryde, Curadev Pharma Ltd., Innovation House, Discovery Park, Ramsgate Road,
Sandwich, Kent, CT13 9ND david@curadev.co.uk.
Abstract
The adaptor protein STING plays a major role in innate immune sensing of cytosolic nucleic
acids, by triggering a robust interferon response. Despite the importance of this protein as a
potential therapeutic target for serious unmet medical conditions including cancer and infectious
disease there remains a paucity of STING ligands. Starting with a benzothiazinone series of
1

weak STING activators (human EC50 ~10 µM) we identified several chemotypes with sub-
micromolar STING activity across all the major protein polymorphs. An example compound 53
based on an oxindole core structure demonstrated robust on-target functional activation of
STING (human EC50 185 nM) in immortalised and primary cells and a cytokine induction
fingerprint consistent with STING activation. Our study has identified several related series of
potent small molecule human STING activators with potential to be developed as
immunomodulatory therapeutics.
1
.
Introduction
Immunotherapy is a rapidly advancing field, in which the immune system is harnessed and
directed towards pathogens and malignancies or towards enhanced vaccine performance [1]. For
example, interferon-based regimens have been a mainstay of hepatitis C antiviral therapy for
many years while antibodies to inhibitory immune checkpoints such as CTLA-4 and PD-1 have
shown considerable promise in treating cancers [2].
The human immune system has evolved to recognize and respond to different types of threats
and pathogens to maintain a healthy host [3]. The innate arm of the immune system is mainly
responsible for a rapid initial inflammatory response to danger signals associated with cellular or
tissue damage from bacteria, viruses and other infectious threats via a number of factors such as
cytokines, chemokines and complement factors. An immune response is typically triggered by a
pathogen associated molecular pattern (PAMP) (or DAMP, damage associated molecular
pattern) [4] binding to a component of the immune system called a Pattern Recognition Receptor
(PRR). PRRs include Toll-Like Receptors [5] (TLRs), C-type lectin receptors, retinoic acid
2

inducible gene I (RIG-I like receptors) [6], NOD-like receptors (NLRs) and double stranded
DNA sensors [7].
The protein TMEM173/STING (STimulator of INterferon Genes) has been shown to play a
central role in the innate immune response to nucleic acids [8]. Free cytosolic nucleic acids
(
DNA and RNA) are known PAMPs/DAMPs. The main sensor for cytosolic dsDNA is cGAS
cyclic GMP-AMP synthase) [9] which upon recognition of cytosolic dsDNA, triggers formation
(
of cyclic dinucleotides (CDNs). CDNs are second messenger signalling molecules produced by
diverse bacteria and consist of two ribonucleotides that are connected via phosphodiester bonds
to make a cyclic structure. CDNs cyclo-di(GMP) (c-diGMP), cyclo-di(AMP) (c-diAMP) and
hybrid cyclo-(AMP/GMP) (cGAMP) derivatives all bind strongly to the ER-transmembrane
adaptor protein STING [10] to induce a conformational change and generate an active protein
state. Activated STING then translocates from the ER to the Golgi, where it can be
phosphorylated by TBK1 or IKK kinases to form the STING signalosome and undergo
palmitoylation and full activation. The activated signalosome can then phosphorylate IRF3
Κ
directly or lead to indirect activation of NF B, to aid transcription of cytokines and type-I
interferons.
It has been shown that activation of STING increases T cell infiltration into cold or non-
inflamed tumors leading to significant regression [11]. The ability of STING activators to
generate anti-tumor immune responses in cold tumors in murine models makes them a promising
therapeutic option either as a single agent or in combination with existing or developing
therapies.
STING activators described in the literature fall into two classes – cyclic dinucleotide (CDN)
based and non-CDN small molecules. Currently, the most clinically advanced chemotype
3

reported is the CDN class derived from the natural ligand structure, which is represented in
Phase 1 clinical trials by ADU-S100 [15] (Aduro Biotech/Novartis, 1, Figure 1) and MK-1474
[16] (Merck, structure not disclosed). The CDN class of compounds usually show activity across
STING species. Their physicochemical properties are determined by their nucleotide structure
and synthesis challenges are common for these multi-chiral center macrocyclic compounds.
The small molecule DMXAA 2 (5,6-dimethylxanthenone-4-acetic acid) has demonstrated
immune modulatory activity and disruption of tumor vascularization in a mouse model but failed
human clinical trials [12]. Subsequently DMXAA [13], and a related acridinone CMA [14]), was
shown to be a potent activator of murine STING but not human STING [14].
Figure 1. Selected published STING activators
Other reports in the journal and patent literature have described nucleic acid-derived series
[17] and some small molecule activators [18-20] including a recent report of ‘dimeric’
4

benzimidazole derivatives (e.g. 3) from a group at GSK [21], all with varying levels of potency
as STING activators.
Several groups have solved the crystal structure of truncated STING in its apo form, as well as
bound with ligands [22,8b] and a recent cryo-EM structure of the STING dimer bound to TBK1
[23]. STING forms a symmetrical dimer in apo and ligand-bound states, with all the reported
structures so far showing ligands bound in a pocket at the dimer interface and anchored by a
network of hydrogen bonds. Interestingly, the STING dimer binds one copy of CDN ligands and
the larger ‘dimeric’ class of small molecule, but two copies of the smaller xanthenone/acridinone
ligands. Human and mouse STING are 68% identical [24], though the human protein has several
naturally occurring variant alleles. Five major haplotypes of human STING have been reported
that represent almost 99% of the human population (R232, H232, HAQ, AQ and Q) [25]. One of
the challenges with pharmacological intervention in STING signalling is to reliably achieve high
potency across all major polymorphisms, with the HAQ polymorphism a particular challenge
[26].
Given the importance of STING as a pharmacological target and the limited number of active
small molecule chemotypes reported, we initiated a medicinal chemistry program to seek
activators of the protein. At the outset of our program, we sought to identify a traditional small
molecule drug-like lead series to differentiate it from the natural CDN ligands. We have recently
described our early work on a series of benzothiazinones in which weak leads with little or no
activity at the HAQ polymorphism (e.g. 4) were developed into potent pan-isoform active
compounds e.g. 5 through optimisation of peripheral substitution (Figure 2) [27].
5

Cl
F
O
F
N
O
N
H
F
F
S
Previous work
5
O
R232 STING EC50 122 nM
HAQ STING EC50 944 nM
O
N
S
O
N
H
R
O
4
/N
R
C
Z
X
Current study;
core templates,
substituents +
substitution
vectors
N
Y
R232 STING EC50 ~ 10 M
HAQ STING EC50 > 30 M
H
C/N
n
C
/
N
6
e.g. Z= C, N; Y= S, S(O), C, C(O), N;
X= C, O, S, S(O), N
Figure 2. Design strategy. Previous work explored peripheral substitution to improve potency
within the benzothiazinone chemotype [27]. The current work sought new core templates to
further optimize potency and properties.
In this manuscript, we describe the medicinal chemistry study that was undertaken to explore
several new chemotypes based on the general target structure 6 through modification and
optimization of the central core template. Through this effort, we wished to probe the effect of
different core ring sizes, heteroatom composition and substitution vectors on STING and its
isoforms to ultimately identify suitable potent tool compounds to explore the STING mechanism
more fully.
2
2
.
Results and Discussion
.1 Chemistry
6

The compounds described in the present study were prepared using a variety of
straightforward, robust methods. These are described in detail in the Supplementary Material.
Illustrative example synthetic schemes are shown below for three different core templates.
The benzoxazinone core was built starting from a halo-nitro-ester 7 (Scheme 1). Halide
displacement with a hydroxyester 8 gave the nitro-ether 9, which upon nitro group reduction and
concomitant ring closure gave the fused oxazinone skeleton 10. Core alkylation with a suitable
halide 11 gave the core derivative 12. Ester hydrolysis and amide coupling to the corresponding
acid 13 with a requisite amine 14 gave the final test compound structures 15.
Scheme 1. Synthesis of benzoxazinones. Reagents and conditions: i) Acetonitrile, triethylamine,
0
-5°C, 81-99%; ii) Fe, AcOH, 80°C, 95-99%; iii) 1,2-dichloroethane, NaOH, water,
tetrabutylammonium bromide, room temperature, 61-89%; iv) MeOH, water, THF, LiOH.H O,
2
room temperature, 74-99%; v) DMF, HATU, triethylamine, room temperature, 20-84%.
The tetrahydroquinolinone cores (Scheme 2) were made starting from a commercially
available dihydroquinolin-2(1H)-one core 16. Following a standard base-mediated N-alkylation
to give the substituted quinolinones 17, the core was alkylated either with a single substituent 18
7

(to give 20) or sequentially with two different substituents 18 and 19 (to give 21) using strong
base and a suitable alkyl halide. Ester hydrolysis to the corresponding acid 22 and amide
coupling with a suitable amine 14 gave the final test compounds 23.
Scheme 2. Synthesis of tetrahydroquinolinones. Reagents and conditions: i) 1,2-dichloroethane,
NaOH, water, tetrabutylammonium bromide, room temperature, 68-89%; ii) LiHMDS, THF, -
7
8°C, 17-48%; iii) MeOH, water, THF, LiOH.H O, room temperature, 55-94%; iv) DCM,
2
HATU, triethylamine, room temperature, 53-67%.
Other ring sizes, such as the oxindole core, were accessed using similar methods starting from
an oxindole core, as depicted in Scheme 3. Alkylation of 2-oxindoline-6-carboxylate ester 24
gave the alkylated core 25. Insertion of the N-indoline substituent with a base-mediated
alkylation reaction gave the corresponding N-alkyl analogues 26. Ester hydrolysis 27 and amide
formation with amines 14 as before then gave the final test compounds 28.
8

Scheme 3. Example synthesis of an oxindole core. Reagents and conditions: i) DMF, NaH, -
0°C, 74-79%; ii) DMF, NaH, room temperature, 52-99%; iii) MeOH, water, THF, LiOH.H O,
1
2
room temperature, 57-99%; iv) DCM, HATU, triethylamine, room temperature, 48-85%.
2
.2 Biological testing
Compounds were evaluated in a stable HEK293T STING-expressing luciferase reporter cell
line and a primary screen based on IRF3 reporter activation. Structural changes were assessed for
potency in the HEK293T-hSTING-luciferase assay, tested initially at a single concentration with
the most promising compounds then tested in a dose-response assay. Functional studies to
confirm compound-induced STING phosphorylation used immunoblotting and cell-based
activation was confirmed through multiplexed cytokine measurement. The analogues explored in
this campaign used either the simple benzyl substituent of 4 or similar di-halo benzyl substituents
as found in 5, with either a 2-furanylmethyl or 2,4,6-trifluorobenzyl amide substituent.
2
.3
STING activity of core variations
9

We initially investigated multiple changes to the core benzothiazinone scaffold, using either
the furan or the trifluorobenzyl group as preferred amide substituents (Table 1).
Table 1. Core template analogues of compound 5
4
3
$
Entry
R
Core variant
R
% activation
cLogP
5
27% @ 10µM
30% @ 10µM
2.7
2
9
3.0
2.3
7% @ 100µM
3
3
0
1
105% @ 10µM
-1% @ 10µM
-1% @ 10µM
3.7
2.4
3.1
3
3
2
3
10

4
3
$
Entry
R
Core variant
R
% activation
cLogP
3
3
4
5
-1% @ 10µM
73% @ 10µM
3.9
3.8
5.3
3
3
6
7
116% @ 1µM
0% @ 10µM
3.7
3
rac
8-
2
0% @ 1µM
4.9
4.3
3
rac
9-
67% @ 20µM
4
4
0
1
0% @ 50µM
-1% @ 10µM
2.5
3.4
11

4
3
$
Entry
R
Core variant
R
% activation
cLogP
4
2
6% @ 10µM
3.9
$
Percentage activation relative to the level of activation provided by a positive control. Activation levels were
measured at the stated concentration of compound.
Replacing the thio group with an oxo (29) equivalent was tolerated whilst the carbon version
(31) appeared significantly more active than the benzothiazine core. The oxidised sulfoxide core
3
0 and the expanded oxepine core 32 were, by contrast, inactive. The quinazoline-dione 37 in
which the benzyl substituent was located along a different vector relative to the position of the
amide group was also inactive. We made several indole derivatives, for example the 1-indole 34
and the 3-substituted oxindole 33 which were both inactive. However, the 1-substituted oxindole
3
5 showed appreciable activity, and when the furan amide group was replaced with
trifluorophenyl, compound 36 gained a significant boost in potency, albeit with an increase in
lipophilicity. To ameliorate this increase in logP, we replaced one of the gem-dimethyl groups
with a more polar isostere of Me, for example the OMe 38 and the OH 39. These analogues
became progressively less active, indicating that this region of the pharmacophore required
lipophilicity for high potency. We therefore synthesized several more analogues with various
substituents at the 3-position of the oxindole core. The oxo compound 40 and the
benzimidazolone 41 which featured polar groups at this position were also inactive. However,
when the lipophilic spiro-cyclopropane 42 was also made, it too was found to be inactive,
confirming a very specific tolerance of substituents in this location.
We were aware that in our previous investigations, a dihalo-benzyl substitution pattern at the
3
4
R position and a trifluorobenzyl amide group at R were favoured and we incorporated these
12

combined structural features into the more promising core changes from the compounds in Table
. These structures are shown in Table 2. In the promising tetrahydroquinolinone series, the
1
trifluorobenzyl amide 43 was significantly more potent than the furan equivalent. Core
methylation as in 44 and 46 was tolerated but did not increase activity, while lactam isosteres
such as the thiolactam derivative 45 was less potent. We also found that aza substitution in the
benzene ring of the core was detrimental to activity, for example 47 and 48. In the benzoxazine
series, the trifluorobenzyl amide insertion had little effect on potency, with the aza version 49 of
similar or weaker potency than the original analogue 29. An exocyclic carbonyl linker in 51 was
not tolerated, consistent with our previous observations that methylene linkers tended to be more
potent. Core methylation however, provided a substantial boost in potency with the racemic
methylated derivative 50 now showing similar activity to the quinolinone series of compounds,
while aza substitution in this core (e.g. 52) also led to a sharp drop in compound activity. It
should be noted that attempts to resolve the chiral center in 50 by individual enantiomer
synthesis were thwarted by facile racemization of this position. We were concerned that
separated enantiomers of the final product would also be vulnerable to racemization which led us
to put this series on hold. A description of this work is provided in the Supplementary Material.
Finally, in the oxindole series, optimal substitution of the core provided highly potent STING
activators e.g. 53 which had a measured EC of 185 nM. We found that modest reductions in
5
0
logP could be achieved by replacing the gem-dimethyl substituent with a Me or CN variation,
with compounds 54 and 55 showing similar levels of potency, for example 55 had a measured
EC of 175 nM.
50
3
4
Table 2. Core variations with optimized R and R groupings
13

$
%
activation
Entry
R
cLogP
(
EC )
5
0
107% @ 1µM
5.2
43
4
rac
4-
9
2% @ 1µM
0% @ 1µM
5.8
4.3
4
rac
5-
4
1
02% @ 1µM
5.6
4.6
4
6
7
5
% @ 1µM
4
14

$
%
activation
Entry
R
cLogP
(
EC )
5
0
58% @ 10µM
3.8
48
3
% @ 1µM
3.4
4.5
4
9
0
5
9
5% @ 1µM
rac
2
6% @ 10µM
4.6
3.9
5
1
2
5
6
% @ 1µM
rac
15

$
%
activation
Entry
R
cLogP
(
EC )
5
0
(EC50 185 nM)
5.7
53
1
02% @ 1µM
5.6
5.2
5
rac
4-
(EC50 175 nM)
5
5-
rac
$
Percentage activation relative to the level of activation provided by a positive control. Activation levels were
measured at the stated concentration of compound.
Through this short SAR exercise, we had identified excellent sub-micromolar activity in
several different core structures. We were aware that small molecule ligands in the literature
were typically reported to have activity at just one or two species, and in some cases activity at a
limited number of STING polymorphs. Notably, early examples of benzothiazinones did not
significantly activate the HAQ haplotype [27]. Compounds from within these new core
chemotypes were therefore screened for species and STING mutant activity.
2
.4
Human isoform activity
16

A selection of compounds from the above tables were screened against three of the most
common STING polymorphs, R232, H232 and HAQ (Table 3). All compounds tested showed
excellent pan-polymorph activity across the panel of STING proteins tested against.
Table 3. Activity of selected compounds against STING polymorphs
R232.HEK293T
H232.HEK293T HAQ.HEK293T
Compound
$
$
$
%
act @ 10 µM
% act @ 10 µM
% act @ 10 µM
5
4
4
4
5
4
5
3
3
4
6
5
5
4
92
107
121
132
137
110
128
117
92
86
102
116
114
122
120
113
99
107
98
93
92
$
percentage activation following a challenge with a positive control in a STING expressing HEK293T cell line at
a fixed concentration of each test compound
An example from Table 3, compound 53 was then tested in a dose response assay against all 5
of the major human STING polymorphs (Figure 3). 53 activated the luciferase coupled
ISRE.ISG54 reporter system in R232, H232, HAQ, AQ and Q variants of human STING to
varying degrees in stably transfected HEK293T cells, confirming activity across the range of
polymorphs.
17

Figure 3. Compound 53 activity is STING variant dependent. 53 activates the luciferase coupled
ISRE.ISG54 reporter system in the most common variants of human STING. Cells were
incubated with G10 for 18 hours and luciferase activity was estimated.
2
.5
STING orthologue activity
We were cognisant of the species differences at mouse and human STING proteins of
DMXAA and in other series [18d] and that human STING has distinct sequence variations in the
human population. Human and mouse STING are 68% identical and have 81% similarity, whilst
human and rat are 69% identical and human and cynomolgus monkey more than 90% identical
[28]. Consistent with our previous findings, 53 was not active at mouse STING, but was
similarly potent at both human and cynomolgus monkey (data not shown).
2
.6
We also showed that 53 is a direct activator of STING. This was carried out biochemically,
using a three component assay containing just recombinant STING, TBK1 and test compound,
Compound 53 is a direct STING activator
18

assaying for phosphorylated STING protein by immunoblotting (Figure 4). These data
demonstrated that compound 53 was a potent STING agonist in a cell free pSTING assay.
Figure 4. Compound 53 is a direct and potent STING agonist in a cell-free pSTING assay [27].
Immunoblots showing 53-dependent phosphorylation of recombinant CTT domain of R232
STING by recombinant TBK1 enzyme (Lane 1: Vehicle control only, Lane 2: 53 at 10 µM)
2
.7
Human STING variants are phosphorylated on treatment with a novel STING
agonist
We further showed that 53 was able to phosphorylate various STING polymorphs in
overexpressed HEK293T cells by immunoblotting (Figure 5). Both phosphorylated IRF3 and
phosphorylated STING were produced in a dose-dependent manner when exposed to compound.
19

Figure 5. Compound 53 treatment leads to phosphorylation of variants of human STING in
overexpressed HEK293T cells. Cells were incubated with 53 for 2 hours and then proteins
harvested and analyzed by immunoblotting using human pSTING, STING, pIRF3, IRF3 and
actin antibodies.
We then showed that 53 enabled phosphorylation of endogenous variants of the STING protein
in various tumor cells (Figure 6) at a single concentration. Compound 53 led to STING and
IRF3 phosphorylation across the panel of STING variants tested, except in the STING KO cells
where STING was absent.
20

Figure 6. Compound 53 treatment leads to phosphorylation of the endogenously expressed
human STING variants in various tumor cells. Cells were incubated with 10µM 53 for 2 h and
then proteins harvested and analyzed by immunoblotting using human pSTING, STING, pIRF3,
IRF3 and actin antibodies with the blots shown in the first panel. The second panel shows the
specific SNP pattern contained in each tumor cell line.
2
.8
Treatment with a novel STING agonist leads to phosphorylation of human primary
cells carrying various STING polymorphs
We next showed that 53 enabled phosphorylation of STING in human primary cells from three
different donors at a single concentration (Figure 7).
Figure 7. Compound 53 treatment leads to phosphorylation of the STING variants of human
STING from three different human donors (D1, D2 & D3) PBMCs. Cells were incubated with
21

1
0µM 53 for 2 h and then proteins harvested and analyzed by immunoblotting using human
pSTING, STING, pIRF3, IRF3 and actin antibodies.
2
.9
Human primary cells from donors were stimulated with a single concentration of 53 and levels
of cytokines induced measured at an early (5h) and a late (20h) time point post-treatment
Figure 8).
Cytokine induction following STING activation
(
Figure 8. Compound 53 cytokine induction profile in hPBMC
A panel of ten pro-inflammatory cytokines were measured which included IFNα, IFNβ,
TNFα, IL6, CCL20, IL1β, IL12p40, CXCL10, IFNγ, IL4 as indicators of activation of both the
IRF3 and the NF B pathways. The induction of these cytokines was measured by qPCR and
k
relative normalized expression (fold induction) was determined by comparison with vehicle
treated controls. 53 at 10µM strongly induced the expression of most of the cytokines especially
Type I IFN subtypes such as IFNβ and CXCL10 which is consistent with an activated STING
signature operating through the IRF3 pathway. Cytokine induction was sustained over a period
22

of 24 hours. It is notable that IL6, CCL20 and TNFα were also induced, indicating activation of
the NF B pathway.
k
2
.10 Cytokine induction and dendritic cell maturation following STING activation
Human PBMCs were isolated and treated with IL4 and GMCSF to induce hMO-DC (human
monocyte derived dendritic cells). Cells from a donor harboring the R232 polymorph were
stimulated with 10µM 53 for 20 h and several pro-inflammatory cytokines and activation
markers were monitored as depicted in Figure 9. Six activation cytokines were selected as the
more prominent markers of immune system stimulation.
Figure 9. Cytokine induction in hMO-DCs following treatment with 53
23

To productively present any antigen to T cells, DCs first have to be activated and markers can
be used to indicate DC activation. CD86 and CD83 are two markers of DC activation that
showed increased expression following treatment with 53 (Figure 10).
Figure 10. Dendritic cell maturation marker expression after treatment with 53. The fold
induction of the activation markers were measured by qPCR and relative normalized expression
was determined by comparison with vehicle treated control
The above data showed conclusively that 53 was able to directly activate STING from across
the major protein variants found in the human population and from both recombinant and
primary cells. It was able to induce a broad panel of cytokines and markers of dendritic cell
maturation consistent with its activation of STING.
2
.11 ADMET
24

In order to assess the in vivo pharmacokinetics of 53, we chose a rodent pharmacokinetic
model to evaluate ADME properties. When dosed in mice, 53 was well-absorbed with a short
terminal half-life (Table 4 and Figure 11).
Table 4. Mouse pharmacokinetic profile of 53. Compound was dosed in solution using 40%
PEG, 20% PG & 10% DMA in NS.
Dose
AUC
T
(h)
1/2
C
max
F
(%)
Route
V (mL/kg)
d
(
mpk) (ng/mL*h)
(ng/mL)
IV
PO
5
10
1215
2090
1.41
NA
1867
1723
2689
NA
NA
86
Figure 11. Pharmacokinetic profile of 53 in mouse
At 10 μM concentration, 53 was stable in buffers across the pH range 3-8 and was stable in
mouse and human plasma at 37°C demonstrating excellent chemical stability.The compound
showed low CYP inhibition (Figure 12), did not cause any non-specific anti-proliferation in a
cell-based cytotoxicity assay and possessed a large window over hERG binding (~20% inhibition
at 10 µM).
25

Figure 12. Cytochrome P450 inhibition (panel A) and cytotoxicity assessment (panel B) of 53.
The compound had a low liver microsomal half-life (HLM t1/2 ~ 10 min.), with high human
plasma binding of >99.5%. This stability, absorption, selectivity and overall pharmacokinetic
profile made the compound a suitable tool for further evaluation of the STING mechanism.
3
Conclusion
In summary, we have developed several novel core chemotypes as potent small molecule
human STING activators with excellent functional properties in a battery of human and
preclinical species testing. An example compound, 53, activated all five major STING
polymorphs and showed the expected STING phosphorylation, cytokine induction and dendritic
cell maturation profile consistent with this mechanism. Further studies with compounds from
these series will be reported in due course.
26

4
Experimental
4
.1 General methods for compound synthesis
Compounds were made according to the methods described in WO2018234807, WO2018234805
and WO2018234808 and are described in detail below and in the Supporting Material. All
chemicals, reagents and solvents were purchased from commercial sources and used without
further purification. All reactions were performed under an atmosphere of nitrogen unless
otherwise noted. Nuclear magnetic resonance (NMR) spectra were in all cases consistent with
the proposed structures. Characteristic chemical shifts (δ) are given in parts-per-million
downfield from tetramethylsilane using conventional abbreviations for designation of major
peaks: e.g. s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; br, broad. Spectra were
recorded in deuterodimethylsulphoxide; d -DMSO. Mass spectra, MS (m/z), were recorded using
6
electrospray ionisation (ESI). Where relevant and unless otherwise stated the m/z data provided
1
9
35
79
127
are for isotopes F, Cl, Br and I. Thin layer chromatography (TLC) was carried out on
Merck silica gel 60 plates (5729). All final compounds were >95% pure as judged by the LCMS
or UPLC analysis methods described in the General Purification and Analytical Methods below
unless otherwise stated. Flash column chromatography was carried out using pre-packed silica
gel cartridges in a Combi-Flash platform. Prep-HPLC purification was carried out according to
the General Purification and Analytical Methods described in the Supplementary Material.
4
4
.2 General procedures for synthesis of benzothiazinone and benzoxazinone intermediates
.2.1 Methyl 4-((2-ethoxy-2-oxoethyl)thio)-3-nitrobenzoate
Methyl 4-fluoro-3-nitrobenzoate (10.0 g, 50.2 mmol) was taken up in MeCN (2.0 L) and TEA
7.61 g, 75.38 mmol) was added to it. The reaction mixture was cooled to 0-5 °C and ethyl
(
thioglycolate (7.25 g, 62.7 mmol) was added dropwise. The reaction mixture was stirred for 30
min. at ice cold temperature. Progress of the reaction was monitored by TLC and after
27

completion; the reaction mixture was diluted with EtOAc and washed with saturated solution of
NH Cl and brine. The organic layer was dried over anhydrous Na SO and evaporated under
4
2
4
reduced pressure to dryness to give the title compound (14.0 g, 93% yield) as a yellow solid
which was pure enough to be used in the next step without any further purification. LCMS m/z:
+
3
00.06 [M+H] .
4
.2.2 Methyl 3-oxo-3,4-dihydro-2H-benzo[b-1,4]thiazine-6-carboxylate
To a stirred solution of methyl 4-((2-ethoxy-2-oxoethyl)thio)-3-nitrobenzoate (5.0 g, 16.7 mmol)
in acetic acid (50 mL) was added iron powder (3.73 g, 66.8 mmol). The resulting reaction
o
mixture was stirred at 80 C for 3 h. On completion (monitored by TLC), it was cooled to room
temperature and poured onto 1N HCl (250 mL) and then stirred for 1 h. The white precipitate
was filtered off and washed with water. The obtained solid was re-dissolved in 5% MeOH in
DCM (50 mL) and filtered through a bed of celite. The filtrate was evaporated to dryness to
+
afford the title compound (3.5 g, 91% yield) as a golden white solid. LCMS m/z: 222.05 [M-H] .
4
.2.3 Methyl 4-benzyl-3-oxo-3,4-dihydro-2H-benzo[b][1,4]thiazine-6-carboxylate
To a stirred solution of methyl 3-oxo-3,4-dihydro-2 -benzo[b-1,4]thiazine-6-carboxylate (4.0 g,
7.91 mmol) in EDC (60 mL) was added NaOH solution (1.432 g, 35.82 mmol in 60 mL H O)
H
1
2
portionwise followed by TBAB (576 mg, 1.791 mmol) at RT. The whole was stirred for 30 min
at RT and then benzyl bromide (3.675 g, 21.49 mmol) was added dropwise. The reaction mixture
was then stirred at room temperature for 2 h. The progress of the reaction was monitored by TLC
and after completion the reaction mixture was diluted with DCM, washed with water and then
2 4
brine solution. The organic layer was dried over anhydrous Na SO and evaporated under
28