Whole cell screen based identification of spiropiperidines with potent antitubercular properties

Article abstract of DOI:10.1016/j.bmcl.2015.05.087

Abstract Whole cell based screens to identify hits against Mycobacterium tuberculosis (Mtb), carried out under replicating and non-replicating (NRP) conditions, resulted in the identification of multiple, novel but structurally related spiropiperidines with potent antitubercular properties. These compounds could be further classified into three classes namely 3-(3-aryl-1,2,4-oxadiazol-5-yl)-1′-alkylspiro[indene-1,4′-piperidine] (abbr. spiroindenes), 4-(3-aryl-1,2,4-oxadiazol-5-yl)-1′-alkylspiro[chromene-2,4′-piperidine] (abbr. spirochromenes) and 1′-benzylspiro[indole-1,4′-piperidin]-2(1H)-one (abbr. spiroindolones). Spiroindenes showed ≥4 log₁₀ kill (at 2-12 μM) on replicating Mtb, but were moderately active under non replicating conditions. Whole genome sequencing efforts of spiroindene resistant mutants resulted in the identification of I292L mutation in MmpL3 (Mycobacterial membrane protein Large), required for the assembly of mycolic acid into the cell wall core of Mtb. MIC modulation studies demonstrated that the mutants were cross-resistant to spirochromenes but not to spiroindolones. This Letter describes lead identification efforts to improve potency while reducing the lipophilicity and hERG liabilities of spiroindenes. Additionally, as deduced from the SAR studies, we provide insights regarding the new chemical opportunities that the spiroindolones can offer to the TB drug discovery initiatives.

Full text of DOI:10.1016/j.bmcl.2015.05.087

Accepted Manuscript
Whole cell screen based identification of spiropiperidines with potent antitu-
bercular properties
Subramanyam J. Tantry, Giulia Degiacomi, Sreevalli Sharma, Lalit kumar Jena,
Ashwini Narayan, Supreeth Guptha, Gajanan Shanbhag, Sreenivasaiah
Menasinakai, Meenakshi Mallya, Disha Awasthy, Gayathri Balakrishnan,
Parvinder Kaur, Deepa Bhattacharjee, Chandan Narayan, Jitendar Reddy, C.N.
Naveen Kumar, Radha Shandil, Francesca Boldrin, Marcello Ventura, Riccardo
Manganelli, Ruben C. Hartkoorn, Stewart T. Cole, Manoranjan Panda, Shankar
D. Markad, Vasanthi Ramachandran, Sandeep R. Ghorpade, Neela Dinesh
PII:
S0960-894X(15)00561-2
DOI:
http://dx.doi.org/10.1016/j.bmcl.2015.05.087
BMCL 22777
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
14 March 2015
23 May 2015
26 May 2015
Please cite this article as: Tantry, S.J., Degiacomi, G., Sharma, S., Jena, L.k., Narayan, A., Guptha, S., Shanbhag,
G., Menasinakai, S., Mallya, M., Awasthy, D., Balakrishnan, G., Kaur, P., Bhattacharjee, D., Narayan, C., Reddy,
J., Naveen Kumar, C.N., Shandil, R., Boldrin, F., Ventura, M., Manganelli, R., Hartkoorn, R.C., Cole, S.T., Panda,
M., Markad, S.D., Ramachandran, V., Ghorpade, S.R., Dinesh, N., Whole cell screen based identification of
spiropiperidines with potent antitubercular properties, Bioorganic & Medicinal Chemistry Letters (2015), doi: http://
dx.doi.org/10.1016/j.bmcl.2015.05.087
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. 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.

Whole cell screen based identification of spiropiperidines with potent antitubercular properties
d,
b,
a,§
Subramanyam J Tantrya,f,ꢀ, Giulia Degiacomi , Sreevalli Sharma , Lalit kumar Jena , Ashwini
Narayanb,§, Supreeth Gupthab,§, Gajanan Shanbhaga,g, Sreenivasaiah Menasinakaia, Meenakshi
Mallyaa, Disha Awasthyb, Gayathri Balakrishnanb, Parvinder Kaurb, Deepa Bhattacharjeeb, Chandan
Narayanb,h, Jitendar Reddyc, C. N. Naveen Kumarc, Radha Shandilc,i, Francesca Boldrind, Marcello
Venturad, Riccardo Manganellid, Ruben C. Hartkoorne, Stewart T. Colee, Manoranjan Pandaa,j,
Shankar D. Markada,k, Vasanthi Ramachandranb, Sandeep R. Ghorpade*,a,l and Neela Dinesh*,b
ꢀ
ꢀ
aDepartment of Chemistry, AstraZeneca India Pvt. Ltd.; bDepartment of Biosciences, AstraZeneca
India Pvt. Ltd.; cDMPK and animal sciences, AstraZeneca India Pvt. Ltd.; dDepartment of Molecular
medicine, University of Padova, Italy; eGlobal health Institute, École polytechnique fédérale de
Lausanne, Switzerland; ꢀFirst authors; §These authors have contributed equally to this manuscript;
fcurrent affiliation: Anthem Biosciences, Bangalore, India; gcurrent affiliation: PI Industries Limited,
h
Udaipur, India; current affiliation: Department of Life Science, Pohang University of Science and
Technology, Republic of Korea; icurrent affiliation: Foundation for Neglected Disease Research,
Bangalore, India; jcurrent affiliation: BMS Biocon Research Centre, Bangalore, India; kcurrent
l
affiliation: Sanofi India Limited, Ankleshwar, India; current affiliation: H3D centre, University of
Cape Town, South Africa; *Address correspondence to Neela Dinesh, neeladin@gmail.com and
Sandeep Ghorpade, sandeepghorpade@hotmail.com
Keywords: whole cell screen, ss18b, antimycobacterial, non-replicating phase, hypoxic conditions,
MmpL3

Abstract:
Whole cell based screens to identify hits against Mycobacterium tuberculosis (Mtb), carried out under
replicating and non-replicating (NRP) conditions, resulted in the identification of multiple, novel but
structurally related spiropiperidines with potent antitubercular properties. These compounds could be
further classified into three classes namely 3-(3-aryl-1,2,4-oxadiazol-5-yl)-1'-alkylspiro[indene-1,4'-
piperidine] (abbr. spiroindenes), 4-(3-aryl-1,2,4-oxadiazol-5-yl)-1'-alkylspiro[chromene-2,4'-piperidine]
(abbr. spirochromenes) and 1'-benzylspiro[indole-1,4'-piperidin]-2(1H)-one (abbr. spiroindolones).
Spiroindenes showed >4 log10 kill (at 2-12 µM) on replicating Mtb, but were moderately active under
non replicating conditions. Whole genome sequencing efforts of spiroindene resistant mutants resulted
in the identification of I292L mutation in MmpL3 (Mycobacterial membrane protein Large), required
for the assembly of mycolic acid into the cell wall core of Mtb. MIC modulation studies demonstrated
that the mutants were cross-resistant to spirochromenes but not to spiroindolones. This report describes
lead identification efforts to improve potency while reducing the lipophilicity and hERG liabilities of
spiroindenes. Additionally, as deduced from the SAR studies, we provide insights regarding the new
chemical opportunities that the spiroindolones can offer to the TB drug discovery initiatives.

Tuberculosis continues to be one of the major killer diseases affecting the world today. It is
increasingly becoming a hard-to-cure disease with the emergence of multidrug resistant (MDR) and
extensively drug resistant (XDR) strains of the causative agent, Mycobacterium tuberculosis (Mtb).
Adding to the complication is the ability of the pathogen to remain as non-replicating persistors (NRP)
that could contribute towards establishment of a chronic and latent infection. Reactivation of the disease
may occur under immuno-compromised conditions.1,2,3 All of these pathologically complex events lead
to difficulties in both diagnosis and treatment of tuberculosis. The current 6-month short-course
chemotherapy comprises of a 2 month-intensive treatment with a 4 drug combination (Isoniazid,
Rifampicin, Pyrazinamide and Ethambutol), followed by a 4-month continuation therapy with Isoniazid
and Rifampicin.4 Non-compliance with this regimen due to its long duration and the side-effects of the
medicines are major challenges to cure tuberculosis.5,6 Hence, there is a desperate need for the
discovery of new drugs that can shorten the treatment duration and be effective against MDR, XDR as
well as latent tuberculosis.
Compounds with cidal properties on replicating and NRP Mtb would be considered attractive for
TB drug discovery. Although, simulating the exact conditions that trigger NRP in latent TB is hard to
achieve under in vitro conditions, various models have been developed that induce actively replicating
Mtb to slow down growth and enter NRP.7,8 The Wayne model mimics the hypoxic conditions observed
in the necrotic lesions, to induce NRP in Mtb under in vitro conditions. In this model, two stages of
NRP are observed, NRP1 is the first stage, where dissolved O2 levels approach 1% saturation and is
characterized by a slow rate of increase in turbidity without a corresponding increase in the number of
CFU or synthesis of DNA. When dissolved O2 content drops below 0.006% saturation, the bacilli shift
to an anaerobic stage, NRP2, where there is no further increase in turbidity. It is hypothesised that the
ability of tubercle bacilli to shift down into one or both of the two NRP stages, is responsible for the
ability of the pathogen to remain dormant in the host for long periods of time.8

Another model for NRP is the streptomycin starved 18b (ss18b) strain, which is a clinical isolate
of Beijing lineage, that is dependent on streptomycin for growth.9 This strain grows actively when the
media is supplemented with streptomycin (18b+) and is comparable to the replicating phase of WT
H37Rv. In the absence of streptomycin, this strain slows down growth and enters an NRP phase,
remaining viable for over 6 months and is referred to as the ss18b. Since the ss18b strain remains non-
replicating under normal aerobic conditions, it is easy to handle and more amenable to use as an NRP
model for High Throughput Screening (HTS).9,10
In this work, we describe a whole cell screen carried out on both replicating and ss18b Mtb,
leading to the identification of novel spiropiperidine class of compounds with potent antitubercular
properties (Figure 1A). Spiropiperdines include three structurally close chemical classes viz. 3-(3-aryl-
1,2,4-oxadiazol-5-yl)-1'-alkylspiro[indene-1,4'-piperidine] (abbreviated as spiroindenes), 4-(3-aryl-
1,2,4-oxadiazol-5-yl)-1'-alkylspiro[chromene-2,4'-piperidine] (abbr. spirochromenes) and 1'-
benzylspiro[indole-1,4'-piperidin]-2(1H)-one (abbr. spiroindolones). Resistant mutant generation
followed by whole genome sequencing (WGS) identified a mutation in MmpL3 that led to resistance
against spiroindene compounds. Previous work in the TB drug discovery area has identified a variety of
scaffolds that target MmpL3, including SQ109,11 AU1235,12 C215,13 BM212,14 THPP,15 Spiros15 and
indolecarboxamides (NITD-349)16 (Figure 1B). SQ109, a 1,2-diamine, is currently in clinical trials for
the treatment of tuberculosis and was recently shown to target MmpL3 (Mycobacterial membrane
protein Large) and thereby interfere with the transport of mycobacterial trehalose monomycolate
(TMM), required for the assembly of mycolic acids into the cell wall core of Mtb.11,17
The mutants raised against spiroindenes were cross-resistant with spirochromene compounds but
not the spiroindolones, suggesting that the latter could bind at a different site or may have a completely
diverse mode of action. Preliminary SAR efforts led to improvement of solubility and reduced the
cytotoxicity liabilities of the spiroindene series. While spirochromenes exhibited similar SAR trends to

spiroindenes, spiroindolone match-pairs of spiroindenes showed better physicochemical properties like
lower logD, improved solubility etc. indicating an opportunity for scaffold-morphing to obtain more
lead-like compounds suitable for antimycobacterial drug discovery.
A.
B.
Figure 1. (A) Spiroindenes, Spirochromenes and Spiroindolones; (B) Known MmpL3 inhibitors
A collection of 320,000 compounds from AstraZeneca corporate library was screened at a single point
concentration of 20 µM against replicating Mtb (H37Rv). At this concentration, 1600 compounds
exhibited at least 80% inhibition of growth. These compounds were further screened in parallel to

determine the inhibitory concentrations against both replicating Mtb (H37Rv) and the non-replicating
ss18b strains, as a 10 point concentration response with two-fold dilution, starting at 50 µM. A cut off
of 10 µM (MIC) was used to select potent compounds for Mtb H37Rv. A higher cut off of 50 µM (IC80)
was used to identify compounds active against the non-replicating strain ss18b, based on the observation
that several fold higher concentrations of compounds was required to inhibit non-replicating Mtb when
compared to their MIC on replicating Mtb.9 Compounds with potent activity on replicating Mtb were
chosen for further analysis, irrespective of their activity on NRP Mtb. This resulted in two categories of
compounds, one set that was active only on replicating Mtb and the other active on both replicating and
NRP Mtb.
Hit evaluation based on potency, chemical structure and lead-like properties resulted in the
identification of 14 clusters prioritized for confirmation of activity. In order to confirm the activity, solid
stocks or resynthesized compounds of representatives from these clusters along with 758 near
neighbours were profiled for MICs. This was followed by determination of Minimum Bactericidal
concentration (MBC) on replicating Mtb H37Rv as well as non-replicating models (hypoxic H37Rv and
ss18b). Cytotoxicity was also evaluated by measuring IC50 against A549 mammalian cells and reported
as the Mammalian MIC (MMIC) (Figure 2). Spiropiperidines consisting of spiroindenes,
spirochromenes and spiroindolones emerged as an attractive series from this screening cascade since
they were cidal against replicating and NRP Mtb, while retaining a good window (>30 fold) between
MMIC and MIC on replicating Mtb (Table 1). Additionally, a second series, 4-aminoquinolone
piperidine amides was also identified as a result of this screening cascade and is discussed separately.18
Since spiroindene was the largest subseries with good potency distribution in the spiropiperidine class;
more detailed evaluation of this series was carried out.

320,000 screened at 20 µM on Mtb H37Rv
≥80% inhibition
10 pt CR for 1600 compounds
H37Rv MIC ≤10 µM
ss18b IC80 ≤50 µM
Hit evaluation, prioritization
14 clusters prioritized
Cluster expansion with 758 NNs
Confirmation with solids and resythesized compounds
H37Rv MIC, MBC, NRP
IC80, NRP cidality, MMIC
12 scaffolds deprioritized
[no cidality, poor selectivity window]
Spiropiperidines
+ 4-AQ
Figure 2. Whole cell Screening Cascade resulting in prioritization of spiropiperidines. NNs, Near
Neighbours; CR, Concentration response; MMIC, IC50 against A549 mammalian cells 4-AQ, 4-
Aminoquinolone amides
Table 1. In vitro properties of representative hits from spiropiperidine class
Compou Structure
nd
Class Mtb Mtb ss18 Hyp MMICa Hu Solu ClogP LogD
MIC MB
b
oxia µM
IC80 IC80
µM µM
ma bilit
n
yd
µM
C
µM
PP µM
Bb
(%
fre
e)

Spiro 0.3
inde
0.8
0.4
>200 >200 68
<1 <1
4.19
4.85
>5.2
>4.8
1
ne
Spiro 0.4
inde
50
100
21
<1
<1
1
3
2
ne
200
Spiro 0.8
chro
1.3
12
50
50
5
3.94
4.34
>4.8
4.3
3
4
mene
Spiro 3
indol
200
>50
ND 4
c
one
aMMIC, IC50 against A549 mammalian cells; bPlasma protein binding; cnot determined; d high-
throughput aqueous solubility generated at pH7.4 using solids from dried DMSO stocks.
Spiroindene compounds were synthesized as described in the scheme 1. The key intermediate 1'-(Boc)
spiro[indene-1,4'-piperidine]-3-carboxylic acid 11 was synthesized in multigram scale using a 8-step
synthesis starting from commercially available bis(2-chloroethyl)amine hydrochloride 5 (Scheme 1). As
a first step towards this synthesis, bis(2-chloroethyl)amine hydrochloride 5 was protected with acid
labile Boc-group. A rapid stirring of 5 with 10% NaOH solution, followed by addition of di-tert-butyl
dicarbonate resulted in tert-butyl bis(2-chloroethyl)carbamate 6 in about 88% yield. Intermediate, tert-
butyl spiro[indene-1,4'-piperidine]-1'-carboxylate 7 was prepared starting from 1H-indene. Addition of
LHMDS onto a cooled solution of 1H-indene in THF resulted in a carbanion which was quenched with

carbamate 6 to give the required spiropiperidine 7 at about 46% yield. Bromination of spiropiperidene 7
at 5ºC resulted in tert-butyl 2,3-dibromo-2,3-dihydrospiro[indene-1,4'-piperidine]-1'-carboxylate 8 in
quantitative yield. The dibromide 8 was subjected to dehydrohalogenation with potassium hydroxide
and the resulting monobromide intermediate 9 was carbonylated with carbon monoxide in ethanol to
give 1'-tert-butyl 3-ethyl spiro[indene-1,4'-piperidine]-1',3-dicarboxylate 10 in about 46% yield.
Carbonylation was carried out using Pd(OAc)2 as catalyst. Optimum results were obtained when the
reaction was carried out in a steel bomb at 100 psi pressure and 110ºC temperature for 12 h. Subsequent
hydrolysis of ester 10 using NaOH in methanol-water led to the desired compound 11 in quantitative
yields.
Refluxing nitrile 12 with hydroxylamine hydrochloride in ethanol resulted in corresponding
amidoximes 13 in quantitative yields. The general applicability of a varied set of aliphatic and aromatic
nitrites was useful in bringing diversity at R2-position. Amidoxime 13 was coupled with 1'-(Boc)
spiro[indene-1,4'-piperidine]-3-carboxylic acid 11 using a simple EDCI/HOBt protocol and the resulting
amido ester 14 was cyclized with potassium acetate in DMF to give corresponding 1,2,4-oxadiazole
substituted spiroindenes 15 in moderate yields. Intermediate 15 was deprotected with HCl in 1,4-dioxan
and was either alkylated with alkyl bromides using potassium carbonate or could be coupled with
corresponding acid employing HATU/DIEA protocol to result in title compounds.

Scheme 1. Synthesis of spiroindenes
Reagents and conditions: (a) Boc anhydride, 10% aq. NaOH, DCM, RT, 6 h.; (b) LHMDS, 1H-indene,
THF, <10oC, 2.5 h.; (c) Br2, THF, <5ºC, 2 h.;
(d) KOH, EtOH, RT, 5 h.; (e) 1,3-
bis(diphenylphosphino)propane, Pd(OAc)2, NEt3, EtOH, carbon monoxide (gas), 100 psi, 110ºC, 12 h.;
(f) NaOH, water, 60ºC, 2 h. (g) NH2OH.HCl, EtOH, K2CO3, reflux, 18 h; (h) EDCI.HCl, HOBt, DCM,
RT 3h.; (i) KOAc, DMF, MW, 120ºC, 25 min.; (j) 1,4-dioxan, HCl (conc) RT, 1h.; (k) R1-Br, K2CO3,
DMF, 120oC, 15 h.; (l) R1-COOH, HATU, DIEA, DMF, RT 2 h.
The key intermediate 1'-(Boc) spiro[chromene-2,4'-piperidine]-4-carboxylic acid 25 was synthesized
starting with Aldol condensation of o-hydroxyacetophenone 16 with 1-Boc-piperidinone 17 in presence

of pyrrolidine (Scheme 2). The resulting α,β-unsaturated ketone intermediate generated in-situ
undergoes intramolecular Michael type addition to result in spirochromone 18 in about 80% yield.19
Reduction of spirochromone 18 with sodium borohydride gave intermediate 19 which on dehydration
with p-toluene sulfonic acid gave spiro[chromene-2,4'-piperidine] 20. During the dehydration stage the
Boc-group on piperidine became de-protected and was protected again with Boc anhydride to yield
intermediate 21. Bromination of spirochromene 21 at 0ºC resulted in tert-butyl 3,4-
dibromospiro[chroman-2,4'-piperidine]-1'-carboxylate 22 in about 88% yield. The dibromide 22 was
subjected to dehydrohalogenation with potassium hydroxide and the resulting monobromide 23
intermediate was carbonylated with carbon monoxide in ethanol to give 1'-tert-butyl 4-ethyl
spiro[chromene-2,4'-piperidine]-1',4-dicarboxylate 24 in about 69% yield. Optimum results were
obtained at 75 psi carbon monoxide pressure and 60ºC temperature in presence of Pd(OAc)2 as catalyst.
Hydrolysis of ester 24 using NaOH in methanol-water provided the desired key intermediate 1'-(tert-
butoxycarbonyl)spiro[chromene-2,4'-piperidine]-4-carboxylic acid 25 which was subjected to protocol
similar to that of spiroindene synthesis (Scheme 1) to obtain the required spirochromenes.

Scheme 2. Synthesis of spirochromenes
Reagents and conditions: (a) Pyrrolidine, EtOH, 0-25oC, 20 h.; (b) NaBH4, MeOH, 0-25oC, 3 h.; (c)
PTSA, Toluene, 130oC, 16 h.; (d) Boc anhydride, 1,4-dioxane-H2O (1:1), NaOH, 0-25oC, 3 h.; (e) Br2,
CCl4, 0oC, 4 h.; (f) KOH, EtOH, 85oC, 16 h.; (g) BINAP, Pd(OAc)2, NEt3, EtOH, carbon monoxide
(gas), 75 psi, 60ºC, 12h.; (h) MeOH, 5% aq. NaOH, 60ºC, 3 h.
Synthesis of spiroindolone 4 and N1-cyclopropylmethyl analogues began with reductive amination of
bis(2-chloroethyl)amine 5 with substituted/unsubstituted benzaldehydes 26 to give tertiary amines 27
(scheme 3). The spiroindolone ring was constructed by the treatment of 2-oxindole with amine 27 using
excess of NaH to give required compounds 4 and 28. Compound 29 was obtained by base mediated
alkylation of compound 4 with cyclopropylmethyl bromide. Similar alkylation of compound 28 gave
intermediate 30 which was used for further transformations. Pd/C catalysed debenzylation of
intermediate 30 to intermediate 31 followed by alkylation of 31 using corresponding alkyl halides in
presence of Hunig’s base gave compounds 32 and 33.
Synthesis of N1-(4-methoxybenzyl) analogues of spiroindolone was achieved following a different
protocol as shown in scheme 3. The commercially available acid 34 was coupled with the 2-
bromoaniline by using 1-propanephosphonic acid cyclic anhydride (T3P) as a promoter in the presence
of a base to give amide 35. The secondary amide 35 was converted into tertiary amide 36 by alkylation
with p-methoxybenzyl chloride using NaH as a base. The required spiropiperidine ring was then
constructed by using intramolecular Buchwald-Hartwig reaction of amide 36 to give intermediate 37 in
good yield. The required alkyl groups were introduced by Boc deprotection of intermediate 37 by
treating with HCl followed by alkylation of intermediate 38 to give compounds 39 and 40.

Scheme 3. Synthesis of spiroindolones
Reagents and Conditions: (a) NaBH(OAc)3, NEt3, DCM; (b) 2-Oxindole, NaH; (c) K2CO3,
cyclopropylmethyl bromide; (d) H2, Pd/C, EtOAc; (e) R1X, DIPEA, CH3CN. (f) 2-bromoaniline,
DIPEA, T3P; (g) NaH, PMBCl, DMF; (h) NaOBut, Pd(dba)3, BINAP, 1,4-dioxane; (i) HCl, Et2O
Out of 83 compounds tested from the spiroindene class, 32 compounds had an MIC ≤2 µM against
replicating Mtb. These were profiled for cidality on replicating Mtb as well as on non-replicating Mtb
using hypoxia and ss18b models (Figure 3). The MBC/MIC ratio was ≤ 4 for all compounds tested
under replicating conditions. IC80 values matched well (less than 4 fold variation) between the two
models for NRP Mtb. The majority of the compounds tested in NRP models had IC80 greater than 10
µM (Figure 3). Most compounds that were active on the NRP models had NRP cidality (MBC) in the
range of 25-200 µM (2 log kill, data not shown) on both ss18b and hypoxic Mtb. Further optimization of
the compounds could result in more potent compounds. Two compounds with IC80 values of 50 µM and
100 µM respectively, were not cidal on hypoxic or ss18b Mtb up to 200 µM. It is possible that they are
cidal at higher concentrations, but this could not be tested due to their limiting solubility. No visible
precipitation of the compound was observed up to the reported IC80 concentrations. Isoniazid and

clofazimine were used as controls in all the assays with NRP models. Clofazimine had MBC of 0.6 µM
against ss18b and 2.6 µM against hypoxic Mtb H37Rv and 0.6 µM against replicating Mtb H37Rv.
Isoniazid did not have MBC against the non-replicating Mtb (ss18b and hypoxic models), but had an
MBC of 0.5 µM on replicating Mtb H37Rv.
35
30
25
20
15
10
5
0
<2
3 to 10
11 to 25
26 to 200
>200
Inhibitory Concentration (µM)
Figure 3. Mtb cidality and NRP activity of spiroindene compounds with MIC ≤ 2µM. ꢁ MBC in
Replicating Mtb, ꢂ MIC in replicating Mtb, ꢃ Hypoxia IC80, ꢄ ss18b IC80.
Compound 1, which was one of the most potent compounds from the initial hits, was selected for
further profiling in in vitro cidality assays. Killing kinetics study under in vitro conditions for compound
1 indicated >4 log10 kill, in 7 days when compared to the start cfu of 4.8 x106 cfu/mL. The study showed
that the activity of spiroindenes was dependent on time of exposure; although some concentration
dependence was observed, it was not significant (Figure 4A). To determine if spiroindenes were active
on intracellular Mtb, compound 1 was tested against macrophage cells infected with Mtb as described in
the methods. The compound was cidal on intracellular Mtb with ~1.3 log10 cfu/ml reduction in cell
number (Figure 4B). The potent cidality on replicating Mtb in addition to the ability to kill intracellular
Mtb suggests that spiroindene series has cidal properties suitable for further lead optimization.

6
5
4
3
2
1
0
12
10
8
B
A
6
4
2
LOQ
LOD
0
0
2
4
6
8
10
12
14
Concentration (µM)
Days
Figure 4. Bactericidal properties of Compound 1. (A) Kill kinetics of H37Rv at various concentrations
(µM). ∆50,ꢅ25, ꢆ12.5, ꢀ6.25, ꢇ3.12, 1.56, ꢃ 0.78, ꢁ No drug. (B) Activity on intracellular
H37Rv. LOQ is the limit of quantification. LOD is the limit of detection. D1 and D7 are Day1 and
Day7 post infection respectively.
The spiroindene compounds were tested against A549 cells, which are adenocarcinomic human
alveolar basal epithelial cells. Cytotoxicity of the compounds was measured as their IC50 on the A549
cells (MMIC) to determine the safety margin compared to the MIC on replicating Mtb. Although
majority of the compounds showed <5 fold window between MMIC and Mtb MIC, a few compounds
(47, 57 and 59, Table 2) with potent MIC had MMIC >100 µM (pMMIC < 4) giving a 500-1000 fold
safety window (Figure S1 in the supporting information). This suggests that there is sufficient scope to
reduce the safety risk against mammalian cells without compromising on the potency against Mtb.
The series had potent antimycobacterial properties (<2 µM MBC). However, most of the compounds
(Mtb MIC ≤ 1 µM, pMIC ≥ 6) were highly lipophilic (ClogP >4) resulting in very low aqueous
solubility (<1 µM) and poor plasma free fractions (<1% free) (Table 1 and 2, Figure S2 in the supporting
information). Also, several of the compounds exhibited potent inhibitory activity on mammalian cells
which would potentially pose a safety risk. Hence, chemistry attempts were focused on reducing the
lipophilicity of the compounds, with the hypothesis that this would result in an improvement of

physicochemical properties including solubility, plasma protein binding and reduced toxicity on
mammalian cells (Table 2).
Polar side-chains were added at the R1 position (Table 2) to increase the polarity of the compounds.
Hydroxylated side-chains at R1 position resulted in potent MICs with reduced lipophilicity, but
unfortunately these compounds had potent MMICs (Table 2, compounds 41and 42) and hence, were
unsuitable for further progression. Compounds 44 and 45 with carboxylic acid and sulphonamide in
side-chain at R1 position respectively, were inactive on Mtb whereas compound 46 with a primary amide
in the side-chain had a weak MIC against Mtb. Methylation of the primary amide to a secondary or
tertiary amide in the side-chain led to better MICs without any improvement in either solubility or
plasma protein binding (compounds 47 and 48 respectively). Compound 49 with isosteric replacement
of amide in the side-chain with 1,2,4-oxadiazole showed weaker MIC. Direct attachment of an amide to
the piperidine N led to compounds with much weaker MICs on Mtb (compounds 50 and 51), hence these
were not useful.
Modifications were made at the R2 position to reduce lipophilicity. Replacement of the aryl group with
methyl resulted in a completely inactive compound 52, indicating essentiality of an aryl group at this
position. Replacing aryl with benzyl led to compound 53 with modest potency without significant
improvement in the properties. Isosteric replacement of the aryl group with thiophene resulted in a
potent compound 54, with similar physicochemical properties; a more polar and smaller pyrazole group
as in compound 55 was not tolerated at this position. Pyridines were well tolerated at the R2 position
(compounds 56-61) with a modest improvement in solubility and plasma protein binding. Compound 60
with 4-methoxy-3-pyridyl group showed potent Mtb MIC of 0.4 µM, but was highly insoluble due to its
crystalline nature, and could not be progressed further. Other variation like replacing oxadiazole with
amide groups or direct attachment of a pyridine ring to the indene led to loss of Mtb MICs.

Among all the pyridines tested, compound 57 with 3-pyridyl and compound 59 with 4-methyl-3-
pyridyl group showed Mtb MICs of 3.1 and 1.6 µM respectively and both the compounds were not
cytotoxic on the A549 cells (MMIC >100 µM).
Table 2. SAR to reduce lipophilicity to improve solubility and PPB
Com R1
R2
Mtb
Hypoxia MMICa ClogP LogD Human Solub Predi hER
G
poun
d
MIC
IC80
(µM)
_PPBb
ilityd cted
IC50
(µM) (µM )
(%
pKa
µM
free)
<1
(µM)
0.3
>200
68
4.19
>5.2
<1
8.6
65
1
12
25
2
3.21
3.52
4.18
3.2
<1
3
8.6
9.2
8.9
NDc
NDc
NDc
41
42
43
0.8
0.4
100
50
3
NDc
NDc
NDc
NDc
NDc
NDc
19

>100 >200
>100 >200
>100
NDc
60.5
>100
12
1.24
2.65
2.81
3.19
3.50
3.33
3.43
3.76
3.9
1.6
<1
1
2.4,
>33.3
44
45
46
47
48
49
50
51
10.4
>4.7
>5
<1
<1
3
10.7, >33.3
8.2
50
25
<1
7.3
6.7
7.3
6.2
-
>33.3
>33.3
>33.3
>33.3
27
12
>200
50
>5.3
>5.3
>5.5
>5
<1
6.3
50
<1
8
>200
>200
50
64
NDc
<1
<1
<1
6
100
50
37
4
3.7
<1
9.3
24
>200 >200
>200
10.2
2.32
4.04
2.9
9.2
<1
895
3
8.6
8.6
1.6
1.4
52
53
25
50
>4.5
1.1
100
12
68
3.98
2.56
NDc
3.3
<1
11
2
8.6
8.6
62
54
55
>100 >200
578
5.0

6.3
>200
75.7
2.89
3.8
2.3
16
8.6
3.2
56
3.1
200
>100
56.3
2.89
3.10
3.8
3.5
3.2
5.1
18
8.6
8.6
6.7
5.5
57
58
*
*
12.5
>200
123
N
N
N
1.6
0.4
100
>100
NDc
3.24
3.56
>4.5
NDc
1.5
11
8.6
8.6
4.3
59
60
NDc
NDc
ND
*
>200
O
6.3
200
31.5
2.94
>4.2
1.4
11
8.6
8.8
61
aMMIC, IC50 against A549 mammalian cells; bplasma protein binding; cNot determined; d high-
throughput aqueous solubility generated at pH7.4 using solids from dried DMSO stocks.
Most of the spiroindenes showed potent inhibition of hERG (human ether-à-go-go-related channel),
presumably due to the basic, lipophilic nature of the scaffold (Table 2). But, compounds with a less
basic piperidine nitrogen (e.g. Compounds 46, 47, 48, 49 with a basic pka of 6.3 -7.2 compared to pka
8.6 for compound 1) showed much lower inhibition on hERG channels (hERG IC50 >33 µM, Table 2).
This observation may be useful during lead optimization phase to allay cardiac risk due to hERG
inhibition.
Compounds 1, 57 and 59 were screened for MIC against single drug resistant clinical isolates of Mtb
as well as a set of non-mycobacterial strains including Staphylococcus aureus ARC517, Streptococcus
pneumoniae ARC548, Haemophilus influenzae ARC446, Escherichia coli ARC523, Pseudomonas

aeruginosa ARC545 and Klebsiella pneumoniae ARC1865. The spiroindene compounds were inactive
(MIC >100 µM) on all of the non-mycobacterial strains tested. The compounds retained activity on all
the clinical strains tested suggesting that this series has the potential to act on resistant strains of Mtb
(Table 3).
Table 3. Activity of spiroindenes on drug resistant strains of Mtb.
STR INH RIF EMB OFL
MIC (µM)
1
57
59
Strain Source
Origin
M. tuberculosis ATCC
2794
Reference strain
Reference strain
0.5 0.2 0.008 1.2 0.14 0.39 1.56 1.56
M. tuberculosis Beijing
(E-47/94)
0.5 0.2 0.008 1.2 0.14 0.39 0.78 0.39
0.5 0.2 0.008 1.2 0.14 0.39 1.56 1.56
0.5 0.2 0.008 1.2 0.14 0.78 0.78 1.56
M. tuberculosis D-211
Sputum Isolate
Streptomycin
dependent
M. tuberculosis 18b+
M. tuberculosis STRR
136570
Sputum Isolate
Sputum Isolate
Sputum Isolate
>55 0.2 0.008 1.2 0.14 0.78 0.78 1.56
0.5 >29 0.008 1.2 0.14 0.39 1.56 1.56
M. tuberculosis INHR
912253
M. tuberculosis RIFR
19000
0.5 0.2 >4.8
1.2 0.14 0.39 0.78 0.78
M. tuberculosis EMBR
17003
Sputum Isolate
Sputum Isolate
0.5 0.2 0.008 >19.5 0.14 0.39 1.56 1.56
M. tuberculosis OFLR
0.5 0.2 0.008 1.2
>11 0.78 3.12 1.56

12119
STR: streptomycin; INH: Isoniazid; RIF: Rifampicin; EMB: Ethambutol; OFL: Ofloxacin
Spirochromenes being structurally very similar to spiroindenes were found to follow SAR and SPR
trends similar to spiroindenes. Hence, detailed SAR studies for the same are not presented here.
Spiroindolones differ in structure from spiroindenes and spirochromenes as these lack the
aryloxadiazole substituents at R2 position. Moreover, spiroindolone has more polar amide functionality
in the core structure which in principle, allow obtaining more polar compounds with better
physicochemical properties. Preliminary SAR exploration at R1 position of spiroindolones (Table 4)
indicated that bulky, lipophilic substituent like 4-tert-butylbenzyl group was needed to obtain potent
MICs (compounds 4 and 28) whereas, N-1 in indolone core (R2 position) could be substituted with small
alkyl groups like cyclopropylmethyl with some improvement in Mtb MIC (compounds 30). Compounds
32 and 33 with N-cyclopropylmethyl substituents, but with more polar side-chains like pyridylmethyl
and tetrahydrofurylmethyl at R1 had much weaker Mtb MICs. Changing R2 to 4-methoxybenzyl as in
compounds 39 and 40 did not improve Mtb MICs further. But, these compounds 32 to 40 showed good
physicochemical properties like logD <3.5, solubility >100 µM, improved plasma free fraction.
Compound 39, which is a spiroindolone analogue of spiroindene 53, showed improved physicochemical
properties such as lower logD, high solubility compared to compound 53 while retaining Mtb MIC
(Table 5). Thus, there is a potential to obtain lead compounds with improved physicochemical properties
using spiroindolone core and the antitubercular activity of these compounds may be optimized by
adding substituted oxadiazoles or other heterocycles at R2 position of indolone ring similar to
spiroindene class.

Table 4. Preliminary SAR for spiroindolones
Comp R1
ound
R2
Mtb
Hypoxia MMICa ClogP LogD Human Solubilityd
MIC
IC80
(µM)
_PPBb
(%
(µM)
(µM) (µM )
free)
2
H
H
3.2
200
>50
4.34
4.2
3.6
4
>100
0.8
-
>50
33
2.51
5.61
2.8
4.2
29
<1
>1000
<1
28
30
>200
50
NDc
NDc
>200
>50
>50
>50
2.29
2.26
3.23
3.2
2.5
2.9
NDc
37
673
32
33
39
200
50
>1000
598
16
50
>200
>50
3.76
2.7
16
550
40

b
aMMIC, IC50 against A549 mammalian cells; plasma protein binding; cNot determined; d high-
throughput aqueous solubility generated at pH7.4 using solids from dried DMSO stocks.
Both spiroindene and spiroindolone compounds showed moderate to high rate of metabolism in vitro
as measured in the presence of human liver microsomes and rat hepatocytes (Table 5). This could be due
to high lipophilicity and metabolically labile soft spots in these compounds which would need further
optimization. Compounds, in general, showed good permeability in a Caco-2 assay indicating suitability
for oral absorption.
Table 5. In vitro DMPK properties
Compound
1
57
59
4
30
39
Human microsomal
Clint (µL/min/mg)
Rat hepatocyte Clint
(µL/min/1E6)
22
37
37
83
28
65
22 (78)
100
37
34
100
65
Compounds 57 and 59 were selected for PK studies as MMIC was >100 µM for both compounds.
Both the compounds showed rapid metabolism in presence of human microsomes as well as rat
hepatocytes, which is indicative of cytochrome P450 mediated oxidative metabolism (Table 5). Hence,
these were profiled for plasma exposure via oral route at dose of 100 mg/Kg in mice in the presence of
1-aminobenzotriazole (ABT) (Figure 5, Table 6). ABT, which is a known non-specific cytochrome P450
inhibitor,20 was used here to reduce the CYP mediated metabolism of the compounds so as to achieve
plasma concentration above MIC for a longer duration. Both the compounds had a similar Cmax and
AUC. Total plasma concentrations of 57 and 59 were above the MIC for 6 and 20 h respectively. But

free plasma concentrations for both the compounds were below MIC for the entire dose duration.
Further scaffold optimization could result in compounds with better free levels above MIC, which would
be ideal for in vivo efficacy studies.
(compound 57)
(compound 59)
Figure 5. Mouse oral PK at 100 mg/Kg in the presence of ABT. ꢀ Compound 57, ꢁ Compound 59, ꢃ
Compound 57, ꢂCompound 59. Filled symbols are total concentrations and open symbols are free
plasma concentrations. Error bars indicated standard deviations from three mice.
Table 6. Oral PK in the presence of ABT for Spiroindenes in mice
Compound 57
Compound 59
PK Parameter
Mean
10.1
1.5
SD
1.5
0.9
Mean
9.1
2
SD
2.6
0
Cmax (µM)
Tmax (h)

Tlast (h)
AUCinf (h*µM)
t1/2 (h)
24
69.5
3.5
0
24
99.7
6.2
0
3.2
1.2
20.7
1.7
Isolation and identification of mutants resistant to spiroindenes: Spontaneous resistant mutants to
compound 1 were isolated by plating M. tuberculosis H37Rv cultures on 7H10 agar plates containing
different concentrations of compounds (5 to 10 fold MIC). Spontaneous mutants appeared at a
frequency of 5*10-9 for compound 1.
Two H37Rv mutants were selected from different plates containing compound 1 at 10X MIC (6 µM).
These two mutants, TB333 and TB334, showed 16 fold or higher MIC compared to WT H37Rv by
RBMA. Both these mutants were analysed by whole genome sequencing followed by bioinformatic
analysis to identify polymorphism that could be responsible for the observed resistance phenotype. A
common single point mutation was identified in mmpL3 gene at nucleotide position 874 (a874c),
corresponding to the codon 292 in MmpL3 (I292L), in both TB333 and TB334 mutants. This mutation
was confirmed by Sanger sequencing. Since both TB333 and TB334 were shown to have the same
mutation, only one of the mutants (TB333) was used for further analysis of the mutation.
In order to further confirm that the a874c mutation in mmpl3 was able to confer resistance to
compound 1, this mutation was introduced in Mtb H37Rv by recombineering. After electroporation with
the primer containing the mutation, bacteria were plated on compound 1. Several colonies were isolated
and the presence of the SNP was confirmed by sequencing. Subsequently, their resistance profile was
evaluated by RBMA. All of them showed the same level of resistance as of TB333 (3 µM). Taken
together, these results provide evidence that the mutation in mmpL3 has a key role in conferring
resistance to compound 1.

The mmpL3 gene encodes the inner membrane transporter MmpL3, part of the RND (Resistance-
Nodulation-Cell Division) superfamily, and is conserved in all sequenced mycobacterial genomes. Out
of 12 MmpL-type proteins present in the H37Rv strain, MmpL3 was the only one predicted to be
essential by insertional inactivation21 and by transposon mutagenesis,22 suggesting a possible
fundamental role for this protein in Mtb. Moreover, in the last years, a plethora of papers showed that
MmpL3 is the target of a series of small molecules with different chemical structures.11,12,13,14,15,16,23,24
The mutation (I292L) identified in the resistant mutants isolated by us maps to the beginning of the
predicted fifth membrane spanning domain and is a conservative mutation (a hydrophobic amino acid is
substituted by another hydrophobic amino acid). This is not uncommon; indeed some SNPs observed in
mutants resistant to BM21214 or to pyrazolopyrimidine15 are also conservative mutations. Remuinan MJ
and collaborators15 showed that mutants resistant to similar “spiro” compound have a point mutation at
codon 292 leading to an amino acid change from I to S or T, providing additional evidence that the I at
position 292 is an important residue in MmpL3 structure/function. These mutations could interfere with
the function of MmpL3, avoiding the uptake or efflux of essential components or metabolites of the cell.
It was demonstrated that MmpL3 is involved in the efflux of TMM, donor of mycolates to either
arabinogalactan or another TMM molecule, yielding Trehalose Di Mycolate (TDM).25 Moreover, it was
shown that MmpL3 can bind heme in vitro, suggesting that it can act simultaneously as exporter and
importer.26
Spontaneous mutants arose from compound 57 and compound 59 selection plates at frequency of
5*10-9 for compound 57 and 3.3*10-9 for compound 59 respectively. These isolates, named TB365 and
TB366 (compound 57) and TB375, TB376 and TB377 (compound 59) showed an MIC value of >30 µM
(>16X parental strain MIC) and 15.5 µM (33X parental strain MIC), respectively. All the mutant
isolates showed a cross-resistance with compound 1 (MIC=3 µM), as well as TB333 showed cross-
resistance with both compound 57 and compound 59 (Table 7). Sequencing of mmpL3 in the mutants

TB365, TB366 and TB375 identified the same mutation as TB333 and TB334 (a874c). This suggests
that a common mechanism of resistance and possibly a common target exists for all three compounds
from this series. It is possible that MmpL3 is the target for this series, but further experiments to
demonstrate the inhibition of MmpL3 function is required to confirm this hypothesis.
Spiroindene resistant mutant TB333 was found to be cross-resistant to spirochromene 3 (Table 7),
suggesting similar mode of resistance for the series; whereas TB333 retained sensitivity towards
spiroindolones 4 and 30. This suggests that these spiroindolone compounds, due to large structural
differences at the R2 position have either a different mode of binding to MmpL3 or might interact with a
different target in Mtb.
Table 7. MIC modulation against spiroindene resistant mutants
Compound Compound Compound Compound Compound Compound
Streptomycin
1
57
59
3
4
30
3.1
WT
0.2
3
1.5
>25
>31
>31
>31
>31
15.5
0.5
3.1
0.8
1
1
12.5
TB333
TB365
TB366
TB375
TB376
TB377
12.5
15.5
15.5
15.5
15.5
15.5
1.6
0.4
3
-
-
-
-
-
-
-
-
-
-
0.5
0.5
1
3
3
3
1
1.5
1
Since the mutants have been isolated under aerobic conditions, it is possible that an additional target
exists that may be more relevant for the activity under hypoxia. We did not pursue target identification
under hypoxic conditions, since most compounds had moderate potency on NRP Mtb along with poor
solubility. Further optimization is needed to obtain better compounds which would be ideal for target
identification studies in NRP Mtb.

In conclusion, spiropiperidines were identified as a novel series with potent antitubercular properties
based on a whole cell screen on replicating Mtb followed by further evaluation carried out under
multiple physiological conditions. The spiropiperidine series comprised of three closely related classes
of compounds namely spiroindenes, spiroindolones and spirochromenes. Since spiroindenes were the
largest subseries with good potency spread in the spiropiperidine series, a more detailed analysis of this
class was carried out. The growth inhibitory effect of spiroindenes was restricted to mycobacterial
species only, making them ideal for their specific use in long term therapy. In addition, these
compounds retained their potency on different clinical isolates including single drug resistant strains,
thereby demonstrating their value in a clinical setting. The spiroindenes were also bactericidal under
replicating, non-replicating and intracellular conditions, suggesting that this class warrants further lead
optimization efforts.
Attempts to yield lead-like compounds with reduced lipophilicity, basicity, cytotoxicity and hERG
liabilities, have indicated routes useful for further improvisation. Preliminary SAR studies on
spiroindolones indicated opportunity for scaffold-hopping to make more novel lead-like compounds.
ACKNOWLEDGEMENTS
The research leading to these results received funding from the European Community’s Seventh
Framework Programme (grant 260872).
We thank Matt Bridgland-Taylor and Ann Woods for their contributions towards the hERG
evaluation. We acknowledge the members of the MM4TB from the EU consortium for constructive
discussions during the meetings.

Supporting Information Available. Biology and DMPK methods, HERG method, Chemistry
experimental procedures and analytical data. This material is available free of charge via the Internet at
http://pubs.acs.org.
REFERENCES
1. Gandhi, N. R.; Nunn, P.; Dheda, K.; Schaaf, H. S.; Zignol, M.; van Soolingen, D.; Jensen, P.;
Bayona, J. Lancet, 2010, 375, 1830.
2. Nagaraja, C.; Shashibhushan, B. L.; Sagar, C., Asif, M.; Manjunath, P. H. J. Postgrad. Med.
2011, 57, 181.
3. Fattorini, L.; Piccaro, G.; Mustazzolu, A.; Giannoni, F. Mediterr. J. Hematol. Infect. Dis. 2011,
5, e2013072.
4. World Health Organization. 2009. Treatment of tuberculosis: guidelines. .4th ed.
(WHO/HTM/TB/2009.420).Geneva, Switzerland: WHO.
5. Lienhardt, C.; Raviglione, M.; Spigelman, M.; Hafner, R.; Jaramillo, E.; Hoelscher, M., Zumla,
A.; Gheuens, J. J. Infect. Dis. 2012, 205, S241.
6. Zumla, A.; Nahid, P.; Cole, S. T. Nat. Rev. Drug Discov. 2013, 12, 388-404.
7. Wayne, L. G.; Sohaskey, C. D. Annu. Rev. Microbiol. 2001, 55, 139.
8. Wayne, L. G.; Hayes, L. G. Infection and immunity 1996, 64, 2062.
9. Sala, C.; Dhar, N.; Hartkoorn, R. C.; Zhang, M.; Ha, Y. H; Schneider, P.; Cole S. T.
Antimicrob. Agents Chemother. 2010, 54, 4150.
10. Zhang, M.; Sala, C.; Hartkoorn, R. C.; Dhar, N.; Mendoza-Losana, A.; Cole, S. T.
Antimicrob. Agents Chemother. 2012, 56, 5782.