Novel bacterial topoisomerase inhibitors derived from isomannide

Article abstract of DOI:10.1016/j.ejmech.2020.112324

A series of Novel Bacterial Topoisomerase Inhibitors (NBTIs) employing a linker derived from isomannide were synthesized and evaluated. Reduced hERG inhibition was observed compared to structure-matched analogues with different linkers, and compound 6 showed minimal proarrhythmic potential using an in vitro panel of cardiac ion channels. Compound 6 also displayed excellent activity against fluoroquinolone-resistant MRSA (MIC₉₀ = 2 μg/mL) and other Gram-positive pathogens.

Full text of DOI:10.1016/j.ejmech.2020.112324

Journal Pre-proof
Novel bacterial topoisomerase inhibitors derived from isomannide
Antony Okumu, Yanran Lu, Sheri Dellos-Nolan, Jonathan Papa, Bryan Koci, Nicholas
T. Cockroft, Judith Gallucci, Daniel J. Wozniak, Jack C. Yalowich, Mark J. Mitton-Fry
PII:
S0223-5234(20)30293-2
DOI:
https://doi.org/10.1016/j.ejmech.2020.112324
EJMECH 112324
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 4 February 2020
Revised Date: 3 April 2020
Accepted Date: 7 April 2020
Please cite this article as: A. Okumu, Y. Lu, S. Dellos-Nolan, J. Papa, B. Koci, N.T. Cockroft, J.
Gallucci, D.J. Wozniak, J.C. Yalowich, M.J. Mitton-Fry, Novel bacterial topoisomerase inhibitors
derived from isomannide, European Journal of Medicinal Chemistry (2020), doi: https://doi.org/10.1016/
j.ejmech.2020.112324.
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 Published by Elsevier Masson SAS.

Novel Bacterial Topoisomerase Inhibitors Derived from Isomannide
Antony Okumu,1 Yanran Lu,1 Sheri Dellos-Nolan,2 Jonathan Papa,3 Bryan Koci,4 Nicholas T. Cockroft,1
Judith Gallucci,5 Daniel J. Wozniak,2,6 Jack C. Yalowich,3 Mark J. Mitton-Fry*,1
*Corresponding author. Tel. (614) 292-1903, email address mitton-fry.1@osu.edu
1Division of Medicinal Chemistry and Pharmacognosy. The Ohio State University, Columbus, OH 43210
2
3
USA. Microbial Infection and Immunity. The Ohio State University, Columbus, OH 43210 USA. Division
4
of Pharmaceutics and Pharmacology. The Ohio State University, Columbus, OH 43210 USA. Eurofins
Panlabs. St. Charles, MO 63304 USA. Department of Chemistry and Biochemistry. The Ohio State
University, Columbus, OH 43210 USA. Department of Microbiology. The Ohio State University,
5
6
Columbus, OH 43210 USA.
ABSTRACT
A series of Novel Bacterial Topoisomerase Inhibitors (NBTIs) employing a linker derived from isomannide
were synthesized and evaluated. Reduced hERG inhibition was observed compared to structure-
matched analogues with different linkers, and compound 6 showed minimal proarrhythmic potential
using an in vitro panel of cardiac ion channels. Compound 6 also displayed excellent activity against
fluoroquinolone-resistant MRSA (MIC90 = 2 µg/mL) and other Gram-positive pathogens.
Keywords
Cystic fibrosis; MRSA; DNA gyrase; NBTI

1. INTRODUCTION
Three decades ago, scientists identified alterations to the gene encoding the cystic fibrosis
transmembrane conductance regulator (CFTR) as the cause of cystic fibrosis (CF).1-3 Mutations in this
gene have subsequently been shown to affect the expression, localization, and function of this chloride
ion channel.4 This detailed understanding has led to the discovery of novel therapies for CF, including
combinations of potentiators such as ivacaftor and correctors such as tezacaftor.5 Most recently, the
triple combination of these two drugs alongside the novel corrector elexacaftor demonstrated
revolutionary efficacy in two Phase 3 trials6,7 and was approved by the United States Food and Drug
Administration. Nevertheless, bacterial infection remains highly problematic for individuals living with
CF, contributing to excess mortality and diminished quality of life.
In patients with cystic fibrosis, Staphylococcus aureus, including methicillin-resistant S. aureus
(MRSA), constitutes one of the most important bacterial pathogens.8,9 A recent European study
identified S. aureus as the most common pathogen in CF patients and demonstrated increased
prevalence of S. aureus infection in all age groups from 2011 to 2016.9 In its 2019 report on antibiotic
resistance,10 the United States Centers for Disease Control and Prevention (CDC) classified MRSA as a
serious threat and ranked it second only to Clostridioides difficile as a source of mortality, with >10,000
deaths per year in the U.S. While the past decade has seen numerous new medicines approved for
MRSA infection, morbidity and mortality remain unacceptably high, and new therapeutic options are
badly needed.
Novel bacterial topoisomerase inhibitors (NBTIs)11 have emerged as a promising strategy for the
cure of infections caused by MRSA and other antibiotic-resistant bacteria. These small molecule
therapeutics inhibit antibacterial “golden targets,” the bacterial topoisomerases previously exploited by
the successful fluoroquinolone class. Importantly, the NBTIs bind to a novel site12 and exhibit
mechanistic differences from the fluoroquinolones.12-14 As a result, the NBTIs lack cross-resistance with
fluoroquinolones and thus have been classified as having a novel mechanism of action.15 The recent
advancement of gepotidacin (1, Figure 1)14,16 into Phase 3 clinical trials illustrates the tremendous
promise of the NBTIs. Challenges to the success of the NBTI class remain, however, including
cardiovascular safety concerns associated with hERG inhibition.11,17
Our approach to tackling this issue has focused on the design and synthesis of NBTIs with
reduced basicity. Reck and coworkers previously revealed the potential of such a tactic, identifying
AZD9742 (2, Figure 1) as a clinical candidate through basicity reduction by the electronegative fluoro
substituent.18 Our initial strategy entailed the use of a 1,3-dioxane linker (e.g. in 4), in which basicity was
reduced by the two neighboring electronegative oxygen atoms.19 Through a series of matched pair
comparisons, we demonstrated consistently reduced hERG inhibition as compared, for example, to
NBTIs with cyclohexane linkers (e.g. in 3), which are both more lipophilic and more basic. More recently,
we reported additional analogues in which the weakly basic secondary amine was replaced with a
secondary amide.20 In some cases, this effort led to highly reduced hERG inhibition (for example, 5 with
an IC50 > 100
µM). In this report, we disclose additional results focused on NBTIs with potent anti-MRSA
activity, here employing a linker moiety derived from isomannide (e.g. as in compound 6).

2. RESULTS AND DISCUSSION
2.1 Molecular Design
As described above, our research has focused on attenuating the basicity of the amine that
interacts with an aspartate at the entrance to the NBTI-binding pocket of DNA gyrase.19,20 Analogously to
the dioxane linker, we hypothesized that NBTIs with an isomannide-derived linker (see prototype in

Figure 1) would display reduced basicity by virtue of the electronegative oxygen atoms positioned beta
to the nitrogen. Indeed, the calculated pKa (Chemdraw) of the prototype isomannide-linked NBTI (7.7) is
similar to that of the reported dioxane 4 (8.1). The calculated lipophilicity of the two compounds is also
nearly identical (cLogP of 3.9 and 4.0, respectively), indicating that the physicochemical properties of the
isomannide- and dioxane-linked series are quite similar. Notably, these NBTIs are still quite lipophilic,
and the introduction of additional polarity might be beneficial in further reducing hERG inhibition as well
as improving metabolic stability.21 This consideration and additional efforts directed at optimization of
potency and synthetic efficiency (vide infra) afforded compound 6 (calculated pKa = 6.9, cLogP = 2.5).
In addition to modulating the properties of the NBTIs, the linker domain plays a key role in
appropriately positioning the DNA-binding and enzyme-binding motifs for productive interaction in a
NBTI/DNA/gyrase ternary complex.12 We
reasoned that the isomannide-derived
linker would achieve this objective and
began by superimposing compound 7
(purple, Figure 2) onto the structure of
AM-8191 (green), as reported by Singh and
colleagues in ternary complex with
bacterial DNA and S. aureus gyrase (PDB
code 4PLB).22 The superimposition
procedure was performed iteratively with
Maestro Academic from Schrödinger Suite
2017-2 by selecting reference atoms from
AM-8191, aligning the reference atoms
with analogous atoms in 7, and then
manually modifying the dihedral angles of
7 to match the conformation of AM-8191.
As shown, 7 can adopt a conformation that
positions the DNA-binding fluoronaphthyridine and enzyme-binding pyridooxazinone moieties
appropriately. The linker domain naturally shows variability as compared to the oxabicyclooctane of AM-
8191, but the linker region is tolerant of considerable variability due to the absence of explicit
interactions with either DNA or the NBTI-binding pocket on gyrase.12 Consequently, we undertook the
synthesis of a series of isomannide-derived NBTIs.
2.2 Chemical Synthesis
2.2.1 NBTIs with fused linker derived from isomannide
Isomannide 8 was initially mono-tosylated to provide 9 in a modest yield by reaction with para-
toluenesulfonyl chloride in the presence of pyridine23 (Scheme 1). Nucleophilic displacement of the
tosylate by potassium phthalimide24 introduced the key nitrogen atom in 10.
Oxidation of the remaining secondary alcohol was carried out using the Dess-
Martin periodinane25 to afford ketone 11. Grignard addition of vinylmagnesium
chloride proceeded to afford the tertiary alcohol 12 with high

diastereoselectivity. The stereochemistry of intermediate 12 was established unambiguously through X-
ray crystallography (Figure 3).
The DNA-binding motif was coupled using a Heck reaction26 to provide 15 and 16. Different
conditions were required for the quinoline intermediate 13 and the fluoronaphthyridine intermediate
14. A comprehensive optimization of the Heck reaction was not carried out, and the origins of the
differential reactivity have not been determined. Hydrogenation of the alkene afforded the saturated
intermediates 17 and 18, which were deprotected as previously described19,20,27 using excess
ethanolamine in refluxing ethyl acetate to yield 19 and 20. Finally, diverse enzyme-binding motifs were
introduced by means of reductive amination, affording quinoline analogues 21-23 and
fluoronaphthyridine derivatives 6, 7, and 24-26. Enzyme-binding motifs were chosen based on structure-
activity relationships we observed in dioxane-linked NBTIs19,20 to facilitate comparison between the two
series. In summary, two sub-series of NBTIs have been prepared in eight steps from commercially
available materials.

HO
HO
H
H
pyridine, TsCl, CH2Cl2
0 °C to rt, 24 h, 31-48%
Potassium Phthalimide
DMSO, 130 °C, 39-55%
O
O
O
O
H
H
OH
OTs
8
9
HO
HO
H
H
O
H
O
O
O
Dess-Martin reagent
MgCl, THF
O
O
O
O
O
O
H
H
N
N
H
N
CH2Cl2, 0 °C to rt, 50-73%
0 °C to rt, 36-48%
O
O
O
10
Br
11
12
O
13
O
O
N
Pd(dba)2,Et3N, DMF
130 °C, 67-84%
O
H
HO
O
N
N
H
HO
H
H
O
O
Pd / C, H2
O
O
OR
EtOAc / MeOH (1:1)
50 psi, 24 h
Br
O
X
R
O
X
R
O
N
F
14
N
N
N
Pd(OAc)2, Ag2CO3,
benzene, reflux (100 °C)
31-71%
15
16
17
18
(R = H, X = CH)
(R = F, X = N)
(R = H, X = CH): 49-90% yield
(R = F, X = N): 66-78% yield
O
N
R'
O
NH2
H
H
HO
O
H
H
H
HO
O
NH2
HO
H
R'
O
ZnCl2, NaBH3CN,
MeOH, rt
ethylacetate, 70 °C
16 h
O
X
R
O
X
R
N
N
19
20
(R = H, X = CH): 72-94% yield
(R = F, X = N): 91-92% yield
R = H, X = CH
R = F, X = N
O
O
O
O
6
24
25
( )
(
)
)
21
(
)
R' =
N
O
O
H
N
N
O
(
(
22
23
(
(
)
)
7
( )
O
Cl
Cl
26
)
Scheme 1.

2.3 Biological Evaluation
2.3.1 Antibacterial activity
Minimum inhibitory concentrations (MICs) were determined in triplicate according to CLSI
guidelines28 using one or more strains of S. aureus. Strain ATCC 29213 was used for initial evaluation,
and ciprofloxacin was employed as a positive control in each assay. Compounds were also assayed using
a USA30029 isolate of MRSA as well as a S. aureus strain isolated from a cystic fibrosis patient,30 using
vancomycin as a positive control. Finally, the breadth of antibacterial activity against Gram-positive
pathogens was assessed using one representative compound.
The quinoline analogue 21 bearing a benzodioxine gyrase-binding motif was ~32-fold less potent
than our previously reported dioxane-linked analogue 419 (Table 1). We hypothesized that a subtle
difference in the linker length might underlie this loss of activity through a steric clash with the bottom
of the NBTI-binding pocket on gyrase. MM2 energy minimization of a simplified isomannide linker in
Chem3D revealed the isomannide to be 0.4 Å longer than an analogous cyclohexane. Consequently, we
evaluated the shorter benzyl analogue 22, but it similarly lacked activity. The 4-methylbenzyl 23
derivative likewise failed to demonstrate potent antistaphylococcal activity. An alternative explanation
is that the precise geometry of the isomannide-derived linker leads to a suboptimal positioning of the
DNA- and/or enzyme-binding moieties.
Our previous results20 and those of other authors18,22,31-34 suggest that the antibacterial activity
of NBTIs can be enhanced through the use of a fluoronaphthyridine DNA-binding motif in place of the
quinoline present in 21-23. Indeed, fluoronaphthyridine 24 was ~32-fold more potent than the
analogous quinoline 21 and demonstrated a promising MIC range of 0.5-1 µg/mL against the ATCC strain
(Table 1). Additional analogues 6-7 and 25-26 also delivered similarly potent MICs. Benzodioxine 24 was
nevertheless somewhat less potent (~4- to 8-fold) than the analogous dioxane 27 [compound 37 in ref.
20] that we reported previously.20
Due to their novel binding mode,12 NBTIs are expected to preserve activity against
fluoroquinolone-resistant strains. We validated this assumption for this novel series of isomannide-
derived NBTIs: the MICs against the ciprofloxacin-resistant USA300 strain were essentially identical to
those against the ciprofloxacin-sensitive ATCC strain. The compounds also demonstrated equivalent
MICs against a S. aureus isolate from a CF patient,30 further demonstrating their promising activity.
Table 1. Antistaphylococcal activitya of isomannide-derived NBTIs and reference compounds.
Cmpd
419
21
Linker
Dioxane
Novel
Novel
Novel
Dioxane
Novel
Novel
Novel
Novel
Novel
DNA-binding moiety ATCC 29213 USA300
CF isolate
quinoline
quinoline
quinoline
quinoline
0.25-1
16-32
>8
0.5
16
>8
>8
0.5
16
>8
>8
22
23
>8
2720
6
fluoronaphthyridine ≤0.25
fluoronaphthyridine 1-4
0.125-0.25 0.25
2
2
7
fluoronaphthyridine
2
1-2
1
2
24
fluoronaphthyridine 0.5-1
fluoronaphthyridine 1-2
fluoronaphthyridine ≤0.25-0.5
1
25
2
2
26
0.5
0.5

ciprofloxacin NAb
vancomycin NA
NA
NA
0.25-0.5
NT
16-3219,20
1-2
0.25
1-2
aMICs determined according to CLSI broth microdilution guidelines,28 all values in
bNA, not applicable
µ
g/mL.
Owing to its superior hERG profile (vide infra), 6 was chosen for additional profiling. MIC values
were determined against a variety of relevant Gram-positive pathogens. The data are summarized in
Table 2; our previously published data for 27 and ciprofloxacin are included for comparison. For
staphylococcal and enterococcal isolates, 6 is somewhat less potent than dioxane-linked 27, whereas
streptococcal activity was similar in this small study. Using a panel of fluoroquinolone-resistant MRSA
isolates, the MIC90 of 6 was 16-fold better than that of ciprofloxacin (2 vs. 32
comparison, the MIC90 of gepotidacin against levofloxacin-resistant S. aureus is 0.5
the pubished data are limited, AZD9742 appears to be somewhat more potent, with an MIC range of
0.06 – 0.125
g/mL against four drug-resistant S. aureus isolates.18 In summary, isomannide-linked NBTIs
µ
g/mL, respectively). By
µ
g/mL.35 Although
µ
offer promising whole cell antibacterial activity against representative Gram-positive pathogens,
including fluoroquinolone-resistant isolates.
Table 2. Minimum inhibitory concentrationsa of 6 against representative Gram-positive bacteria.
Strain
Cmpd. 6 MIC
Cmpd.
MIC20
≤0.06b
≤0.06
≤0.06
0.25
27 Ciprofloxacin MIC20
S. aureus ATCC 29213 (n=3)
MRSA DRL-3161
0.25b
1
0.25
>64
0.25
2
Streptococcus pyogenes LSI-1235
Streptococcus pneumoniae LSI-4234c
Streptococcus pneumoniae LSI-4246d
Enterococcus faecium DRL-4193e
Enterococcus faecium LSI-3340g
Enterococcus faecalis ATCC 29212
Fluoroquinolone-resistant S. aureus
MIC90 (11 strains)
aDetermined at Laboratory Specialists, Inc. (Westlake, OH, USA) according to CLSI broth microdilution
guidelines,28 all values in
pinpoint wells. Penicillin-susceptible. Penicillin-resistant. Vancomycin-susceptible. Trailing observed;
no viable growth within trailing wells. gVancomycin-resistant.
≤0.12
≤0.12
≤0.12
2
≤0.06
0.12f
0.12
0.5
0.25
16
0.25
2
0.5
0.5
32
2
NT
b
µ
g/mL. Very small pinpoint growth up to 4 µg/mL, no viable growth from
c
d
e
f
2.3.2 Evaluation of gyrase/TopoIV dual-inhibition in biochemical and whole cell assays
We evaluated the potent isomannide-derived NBTIs for inhibition of both DNA gyrase and TopoIV from
S. aureus, with the goal that balanced inhibition of both enzymes (dual targeting) would reduce the rate
of resistance emergence.36,37 A direct comparison of IC50 values from topoisomerase gel-based assays
that necessarily rely on different endpoints (inhibition of supercoiling by gyrase and of decatenation by
TopoIV) has intrinsic limitations and does not capture cellular sequelae such as the SOS response.
Nevertheless, the utility of these in vitro data in assessing dual-targeting ability by NBTIs has been
previously reported.32-34,37,38 In comparison with dioxane-linked 27, the new NBTIs were moderately less
potent inhibitors of DNA gyrase (Table 3), in-line with their weaker antibacterial activity (vide supra). As
seen previously with other series of NBTIs,19,20,22,31-34,38 the isomannide-derived NBTIs inhibited S. aureus

DNA gyrase more potently than TopoIV. Compounds 6, 24, and 26 did, however, show slightly more
balanced ratios (~2-fold improvement) of TopoIV/gyrase IC50 values as compared to dioxane 27.
Table 3. Inhibition of topoisomerase enzymes in biochemical assays and additional MIC assays.
Cmpd Gyrase
TopoIV
TopoIV/gyrase S.
aureus S. aureus
Fold loss of
IC50
(
µ
M)a,b IC50 M)b
(µ
IC50 ratio
3527
D83N mutant potency from
3527 to D83N
MIC (
2, 4
µ
g/mL)c MIC ( g/mL)c
µ
6
7
0.53
0.54
7.4
No
14
2, 8
Up to 2
8 to 16
Not
1, 2
8, 32
activity
2.6
determined
24
0.2
13
42
15
33
1
2
4
4
25
0.18
0.2
7.5
16
8
26
2720
2.9
0.25-0.5
≤0.25
4, 4
4
8 to 16
≥16
0.03
1.0
aDetermined at Inspiralis, Ltd. (Norwich, UK). See Experimental section for details. cDetermined
b
according to CLSI broth microdilution guidelines.28
A previously reported multidrug-resistant MRSA isolate (3527, previously named 109534,38) was
employed alongside an isogenic NBTI-resistant daughter strain34,38 with a D83N amino acid substitution
in DNA gyrase. The D83N strain used here was originally selected by antibacterial pressure from
cyclobutylaryl-substituted NBTIs.38 This amino acid substitution has been associated with a substantial
loss of gyrase inhibition. For example, a prototype compound within the cyclobutylaryl series
demonstrated a >200
µM IC50 using the D83N mutant gyrase as compared to an IC50 of 1.19 µM with the
wild-type enzyme. Similar reductions in activity were seen at the whole cell level, with an MIC of ≥ 64
µ
g/mL against the mutant strain versus 0.20 µg/mL for the parent strain (≥ 320-fold loss). Similar results
were reported independently for NXL-101 at the enzyme and whole cell levels.39
Consequently, NBTIs preferentially inhibiting gyrase are expected to display reduced
antibacterial activity against this mutant strain. In contrast, it has been observed that potent TopoIV
inhibition can partially restore antibacterial activity through this secondary target.32-34,38 Within the
cyclobutylaryl series, optimization of dual-targeting reduced the loss of whole cell activity from ≥ 320- to
ca. 20- to 40-fold,34 and other authors have also demonstrated improved MICs through enhanced dual
targeting.32,33 Consequently, we employed the NBTI-resistant (D83N mutant) strain, alongside its wild-
type multidrug-resistant progenitor (3527), and measured the loss of potency in the mutant strain.
For compound 6 bearing a pyridodioxine enzyme-binding moiety, only minimal shifts in activity
were observed between the wild type and NBTI-resistant strains: identical MICs (2
and only a two-fold shift (4 to 8 g/mL) in a repeat study (Table 3). Benzodioxine 24 was similar,
showing a four-fold loss of activity in the D83N strain (1 to 4 g/mL). A larger loss of potency (8- to 16-
µg/mL) in one assay
µ
µ
fold) was observed with the remaining analogues (7, 25, and 26). The observed trend of higher MICs in
the D83N mutant strain is consistent with gyrase serving as the primary target for these isomannide-
derived NBTIs. In contrast with the previous reports,34,38 neither the MIC in the D83N strain nor the
magnitude of the shift between the wild type and mutant strains could be directly correlated to the
TopoIV/gyrase IC50 ratio or to the TopoIV IC50. This observation suggests that a deeper understanding of
the underlying pharmacology is required. We hypothesize that antibacterial activity in the D83N mutant
strain may be attributed, at least in part, to inhibition of the mutant gyrase enzyme by these

isomannide-derived NBTIs. In fact, the IC50 for compound 24 against the D83N mutant gyrase (6.5
µM) is
similar to its IC50 for TopoIV (2.6 M, see Table 3). We are actively investigating this issue using a range
µ
of NBTIs with diverse structures; additional results will be reported in due course.
2.3.3 Determination of spontaneous frequencies of resistance
Compounds 6 and 24 were also evaluated for their ability to suppress the emergence of NBTI-
resistant mutants in a spontaneous frequency of resistance (FoR) study. At concentrations of 8X and 16X
MIC, 24 proved superior to 6 but was unable to suppress the emergence of resistance completely. In
contrast, no resistant mutants were obtained either at 8X or 16X MIC concentrations using ciprofloxacin.
Thus, as with other NBTIs, these isomannide-derived NBTIs have not yet achieved the goal of ultra-low
frequencies of spontaneous resistance in S. aureus.32,33,40 These results stand in contrast to those
obtained using NBTI 5463 in E. coli, in which it was not possible to select resistant mutants using
standard methodologies as a consequence of balanced inhibition of DNA gyrase and TopoIV in that
organism.37
Table 4. Spontaneous Frequency of Resistance.a
Cmpd.
S. aureus ATCC 29213 Spontaneous mutation Spontaneous mutation
MICb (
1
µg/mL) (n=2)
frequency (8X MIC)
7.89x10-8 (n=1)
1.35x10-9 (mean, n=2)
<2.25x10-10
frequency (16X MIC)
1.57x10-9 (n=1)
6
24
0.5
0.5
4.49x10-10 (mean, n=2)
ciprofloxacin
<2.25x10-10
aDetermined at Micromyx, LLC (Kalamazoo, MI, USA); see Experimental section for full details. Assays
were conducted according to CLSI agar dilution guidelines.28,41
b
2.4 In vitro safety evaluation
2.4.1 Cardiovascular safety
The design of these isomannide-derived NBTIs focused on enhanced polarity and modulation of
basicity through the two electronegative oxygen atoms in the linker. We hypothesized that this
alteration of physicochemical properties would correlate with reduced hERG inhibition, as we and
others have observed previously.18-20 Consequently, hERG IC50 values for compounds 6-7 and 24-26 were
determined (Table 5). Of these analogues, pyridodioxine 6 had the most favorable IC50 (66
µM), being
ca. 3-fold higher than benzodioxine 24 (20
µ
M). This shift is similar to our observations with dioxane-
linked NBTIs.19,20 Compounds 25 and 26, with monocyclic enzyme-binding motifs, showed greater hERG
inhibition. We expected the pyridooxazinone 7 to inhibit hERG quite potently on the basis of our early
observations in the dioxane series,19 but the IC50 (12
µM) fell between the two other sets of compounds.
Evaluation of the safety hypothesis requires comparison to suitable analogues with different
linkers. Table 5 also shows hERG IC50 values, where available, for matched pairs of dioxane-linked and
isomannide-derived analogues. Compounds 6, 7, and 24 are all superior to the corresponding dioxanes20
(28, 29, and 27, 2.5- to 8.5-fold), perhaps reflecting the lower lipophilicity (~0.7 cLogP units) and basicity
(~0.6 pKa units) of the newer compounds. Reck and colleagues observed a 5.3-fold improvement in
hERG IC50 (from 44 µM to 233 µM) through basicity reduction via fluorination to afford the clinical

candidate AZD9742.18 In contrast, the highly lipophilic isomannide 26 was similar to the corresponding
dioxane 30. A small set of compounds with a hydroxylated cyclohexane linker42,43 (31-33) was also
assayed (for details on synthesis, see Supporting Information). Such analogues are more basic (~1.8 pKa
units) and more lipophilic (~0.7 cLogP units) than the isomannide-derived NBTIs. 31 and 32 are
diastereomers of one another. The relative stereochemistry was not firmly proven, rather tentatively
assigned based on a roughly 16-fold greater potency for 31 compared to 32 in the S. aureus ATCC 29213
MIC assay (0.063 vs. 1 µg/mL). Notably, 31 and 32 had identical hERG IC50 values, and the analogous
isomannide 24 was superior to both (>8-fold higher IC50), further exemplifying the value of this new
linker. This advantage was eroded, however, when comparing the 3,4-dichlorophenyl analogue 26 to 33.
1
13
We tentatively assigned the stereochemistry of 33 based on greater similarity of its H and C NMR
spectra to 32 compared to 31. Intriguingly, 33 still displayed a potent S. aureus ATCC 29213 MIC (0.125
µg/mL).
Table 5. hERG IC50 values for isomannide-derived NBTIs and representative comparator compounds.
hERG IC50 (µ
M)a
Compound Linker
RHS
cLogPb pKa b
6
Novel
pyridodioxine
66
2.5
1.7
3.2
3.8
4.6
3.9
3.2
2.4
5.3
3.9
3.9
5.3
6.9
6.8
7.5
7.6
7.3
8.1
7.4
7.4
7.9
9.3
9.3
9.1
7
Novel
pyridooxazinone 12
24
25
26
27
28
29
30
31
32
33
Novel
benzodioxine
4-methylphenyl
3,4-di-Cl phenyl
benzodioxine
pyridodioxine
20
Novel
4.2
6.1
8.0
15
Novel
Dioxane
Dioxane
Dioxane
Dioxane
pyridooxazinone 1.4
3,4-di-Cl phenyl
7.9
2.2
2.3
3.8
cyclohexane benzodioxine
cyclohexane benzodioxine
cyclohexane 3,4-di-Cl phenyl

aDetermined using IonWorks Barracuda at Charles River (Cleveland, OH, USA) using previously reported
methodology.20 bCalculations carried out using Chemdraw Professional version 17.1.
Early drug discovery programs commonly assess hERG inhibition as the only indicator of
potential cardiovascular safety concerns. Nevertheless, NBTIs such as gepotidacin have demonstrated
QT prolongation in clinical trials despite very weak hERG inhibition (IC50 = 1.3 mM).44 Researchers have
recently recommended a more rigorous evaluation of multiple cardiac ion channels to provide a more
comprehensive cardiovascular risk assessment.45 Given the promising initial results for 6, further
evaluation was undertaken (in duplicate) in a Comprehensive in vitro Proarrhythmia Assay (CiPA) eight
assay panel run on the QPatch HT (Sophion, Denmark) automated patch clamp system at three
concentrations (3, 30, and 100 µM). The data (mean percent inhibition values) are shown in Table 6.
Results demonstrate <50% inhibition at 100 µM for all of the six targets tested except for the Nav1.5 late
current antagonist assay. This functional in vitro cardiovascular profile of the six cardiac targets predicts
minimal preliminary proarrhythmic risk. An integrated assessment would be required to understand
how the individual inhibitions for each target contribute to modifying the action potential shape. In
addition, the results need to be compared to in vivo PK (Cmax free at efficacious exposures in an in vivo
model of infection) to determine adequate safety margins.
Table 6. Inhibition of cardiac ion channels by compound 6 at varying concentration.
Ion Channel
% Inhibition at 3 / 30 / 100
-2.4 / 11.3 / 39.0
-2.9 / 1.5 / 22.1
µM
hERG (tail current)
KCNQ1 (peak current)
Kv4.3 (peak current)
Kv4.3 (end current)
Kir2.1 (peak current)
Kir2.1 (end current)
Cav1.2 (peak current)
Nav1.5 (peak current)
11.0 / 25.1 / 43.6
-4.4 / -19.6 / 7.1
6.9 / 18 / 29.7
8.3 / 19.6 / 31.9
4.7 / 23.3 / 24.9
6.1 / 16.7 / 21.1
Nav1.5 Late (antagonist – inhibition of ATXII EC50 current) 4.3 / 19.1 / 50.9
Nav1.5 Late (agonist – activation of late current)
3.4 / 7.7 / 17.0
aConducted by Eurofins Panlabs (St. Charles, MO, USA), see Experimental section for full details.
2.4.2 Broader safety screening
Potential off-target pharmacology for compound 6 was also evaluated by screening in duplicate
against a panel46 of 44 receptors, enzymes, etc. (for full list, see Supporting Information). Of the 44
potential targets, only five were inhibited by >50% at a concentration of 100
µM (Table 7).
Concentration-response studies were not carried out; consequently, these results should be considered
preliminary but serve to highlight potential risks that might require close monitoring in future studies.
Table 7. Targets inhibited by 6 by >50% at concentration of 100 µM.
Assay
% Inhibition at 100 µM conc.
Histamine H2
Norepinephrine transporter
Nav site 2
51
68
78
95
5-HT2B

Dopamine transporter
96
aConducted by Eurofins Panlabs (St. Charles, MO, USA), see
Experimental section for full details.
2.4.3 Cytotoxicity in mammalian cells
The cytotoxicity of compound 6 was measured using an ATP viability assay in HepaRG cells, with
a five-point concentration-response curve (Table 8). Even at the highest tested concentration (100
only minimal cytotoxicity was observed (14%). Results indicate a substantial safety margin between the
concentrations required to achieve antistaphylococcal activity (MIC for S. aureus ATCC 29213 = 2 - 8
[1 - 4 g/mL]) and the concentrations required to elicit toxicity in this cellular assay.
µM),
µM
µ
Table 8. Cytotoxicity of 6 in HepaRG cells.a
% Cytotoxicity
Test Concentration (µM)
100
10
14
8
1
9
0.1
0.01
4
6
aConducted at RTI, International (Research Triangle
Park, NC, USA), see Experimental section for full details.
2.5 Preliminary ADME assessment
Compound 6 was further evaluated in a series of ADME assays. Protein binding was moderately
high in both mouse and human plasma, with unbound fractions of 4.7% and 13.8%, respectively. Under
kinetic conditions, compound 6 was soluble in buffered aqueous solution (pH 7.4) at 76.2 µM. Further
characterization of solubility under thermodynamic conditions would be important for continued
development.47 Permeability was high in Caco-2 cells48 (A‰B of 10.8x10-6 cm/sec), and the compound
showed higher permeability in the B‰A direction (85. 3x10-6 cm/sec), suggesting the potential for efflux.
Using mouse and human microsomes, compound 6 demonstrated a very short half-life of 3 and 6 min,
respectively, strongly suggesting the need for focused optimization of this key parameter, potentially by
further reductions in lipophilicity21 (cLogP = 2.5). Finally, inhibition of the cytochrome p450 enzymes was
assessed at 10
inhibition was observed for CYP3A4 in the screening assay at 10
concentration-response assay in triplicate showed an IC50 > 10 M (Table 9). In summary, profiling of 6
µM. Less than 50% inhibition was seen for CYP1A2, 2B6, 2C9, 2C19, and 2D6. While 56%
µM concentration, a subsequent
µ
as a representative of the isomannide-derived NBTI series revealed generally favorable drug-like
properties, with the exception of rapid metabolism in microsomal assays.
Table 9. Concentration dependence of CYP3A4 (midazolam hydroxylase) by 6.
%
of control % of control
% of control
Test Concentration (µM)

(Exp. 1)
49.5
(Exp. 2)
82.3
(Exp. 3)
74.5
10
3
74.3
77.2
60.4
1
60.4
89.8
89.9
0.3
0.1
89.9
113.1
108.4
99.2
115.5
97.0
aConducted at RTI, International (Research Triangle Park, NC, USA), see
Experimental section for full details.
3. CONCLUSIONS
We undertook the synthesis and evaluation of a series of NBTIs with a linker derived from
isomannide. These compounds demonstrated potent antibacterial activity against Gram-positive
bacterial pathogens such as MRSA and an improved in vitro safety profile as compared to other linker
moieties. Compound 6 displayed excellent activity against fluoroquinolone-resistant MRSA (MIC90 = 2
µg/mL), a moderate hERG IC50 (66 µM), and generally favorable ADMET properties. Additional
optimization to reduce metabolic clearance is necessary. These isomannide-derived NBTIs contribute
further to the understanding of the structure-activity relationships underpinning antibacterial activity as
well as cardiovascular safety.
4. EXPERIMENTAL
4.1. Chemistry
4.1.1. General materials and methods
Unless noted otherwise, air- and/or moisture- sensitive reactions were carried out using oven-
dried glassware under an atmosphere of either dry nitrogen or dry argon. Dichloromethane (DCM),
toluene, tetrahydrofuran (THF), and N,N-dimethylformamide (DMF) were dried before use by passage
through activated alumina under nitrogen atmosphere. Automated flash chromatography on silica gel
was performed using a Teledyne-ISCO Combiflash-Rf+. 1H NMR spectra were recorded at 400 MHz using
residual protiated solvent as the internal reference: (CD3)2SO (2.50 ppm), CDCl3 (7.26 ppm), or CD3OD
1
(3.31). Assayed compounds had a purity of >90% as determined by H NMR analysis; spectra are
13
provided in the Supporting information. C NMR spectra were recorded at 100 MHz using the solvent
signal as the internal reference: (CD3)2SO (39.52 ppm), CDCl3 (77.16 ppm), or CD3OD (49.00). High-
resolution mass spectrometry was performed using electrospray ionization.
4.1.2. Chemical synthesis

4.1.2.1. (3R,3aS,6R,6aR)-6-hydroxyhexahydrofuro[3,2-b]furan-3-yl 4-methylbenzenesulfonate (9). To a
solution of isomannide (8, 15.0 g, 103 mmol, 1.0 equiv) in DCM (51 mL) at 0 °C under nitrogen was
added pyridine (16.6 mL, 205 mmol, 2.0 equiv) and p-toluenesulfonyl chloride (22.5 g, 118 mmol, 1.15
equiv). The reaction temperature was gradually raised to room temperature and stirring continued for
24 h. The reaction was diluted with DCM, and quenched at 0 °C by addition of 1 N HCl solution. The two
layers were separated, and the aqueous phase extracted with DCM (3 × 50 mL). The combined organic
layers were washed with brine, then dried over anhydrous sodium sulfate and concentrated under
vacuum. The residue was purified by chromatograghy using 0 -10% methanol / DCM to give the title
compound as a white solid (11.69 g, 38.99 mmol, 38% yield). This reaction was run eight times under
similar conditions with an average yield of 39% (range 30-48%). 1H NMR (CDCl3, 400 MHz): δ 7.79 (d, J =
8.3 Hz, 2H), 7.33 (d, J = 8.1 Hz, 2H), 4.91 – 4.83 (m, 1H), 4.45 (t, J = 4.9 Hz, 1H), 4.39 (t, J = 5.0 Hz, 1H),
4.29-4.21 (m, 1H), 3.97 (dd, J = 9.3, 6.5 Hz, 1H), 3.91 (dd, J = 9.3, 6.5 Hz, 1H), 3.75 (dd, J = 9.3, 7.5 Hz, 1H),
3.51 (dd, J = 9.3, 7.3 Hz, 1H), 2.61 (bs, 1H), 2.42 (s, 3H); 13C NMR (CDCl3, 100 MHz): δ 145.3, 133.1, 130.0,
128.0, 81.5, 80.1, 78.5, 73.9, 72.4, 70.1, 21.7; HRMS (ESI) m/z calcd for C13H16SO6Na [M + Na]+:
323.05653; found: 323.05701.
4.1.2.2. 2-((3S,3aR,6R,6aR)-6-hydroxyhexahydrofuro[3,2-b]furan-3-yl)isoindoline-1,3-dione (10). To a
solution of 9 (11.2832 g, 37.6019 mmol, 1.0 equiv) in DMSO (67 mL) at room temperature under argon,
was added potassium phthalimide (9.0540 g, 48.882 mmol, 1.3 equiv). The reaction mixture was stirred
at 130 °C for 16 h, then cooled to room temperature and poured into cold water. The resulting mixture
was extracted with DCM (3 × 50 mL), then the combined organic layers were dried over anhydrous
sodium sulfate, filtered, and concentrated under vacuum. The residue was purified by chromatography
using 30-70% ethyl acetate/ hexanes to give the title compound as a white solid (4.95 g, 18.0 mmol, 48%
yield). This reaction was run five times under similar conditions with an average yield of 47% (range 39-
1
55%); the obtained solid was often yellowish in color. H NMR (CDCl3, 400 MHz): δ 7.87-7.80 (m, 2H),
7.76-7.69 (m, 2H), 5.01 (dd, J = 4.9, 2.1 Hz, 1H), 4.89 (t, J = 5.2 Hz, 1H), 4.87 – 4.80 (m, 1H), 4.31 – 4.24
(m, 2H), 4.05 (dd, J = 9.3, 6.3 Hz, 1H), 3.91 (dd, J = 9.3, 5.9 Hz, 1H), 3.70 (dd, J = 9.3, 6.5 Hz, 1H), 2.60 (s,
13
1H); C NMR (CDCl3, 100 MHz): δ 167.7, 134.4, 131.7, 123.6, 86.1, 82.9, 72.7, 72.1, 72.0, 57.4; HRMS
(ESI) m/z calcd for C14H13NO5Na [M + Na]+: 298.06914; found: 298.06958.

4.1.2.3. 2-((3S,3aR,6aS)-6-oxohexahydrofuro[3,2-b]furan-3-yl)isoindoline-1,3-dione (11). To a stirred
solution 10 (3.23 g, 11.7 mmol, 1.0 equiv) in DCM (58 mL) at 0 °C was added the Dess-Martin
periodinane reagent (9.96 g, 23.5 mmol, 2.0 equiv). After stirring at 0 °C for 15 min, reaction was
continued overnight at room temperature, then quenched at 0 °C by addition of a saturated aqueous
sodium thiosulfate/ sodium bicarbonate solution. The two layers were separated, and the aqueous layer
extracted with ethylacetate (3 × 50 mL). The combined organic layers were washed with brine, then
dried over anhydrous sodium sulfate, filtered, and concentrated under reduced pressure. The residue
was purified by chromatography with 40 -50% ethylacetate/ hexanes to give a white solid product (2.33
g, 8.53 mmol, 73% yield). This reaction was run three times under similar conditions with an average
yield of 64% (range 50-73%). 1H NMR (CDCl3, 400 MHz): δ 7.90-7.83 (m, 2H), 7.79-7.72 (m, 2H), 5.28 (d, J
= 4.6 Hz, 1H), 5.06 – 5.00 (m, 1H), 4.75 (d, J = 4.7 Hz, 1H), 4.35 (dd, J = 9.5, 7.0 Hz, 1H), 4.28 (d, J = 17.4
Hz, 1H), 4.09 (dd, J = 9.5, 4.8 Hz, 1H), 4.02 (d, J = 17.4 Hz, 1H); 13C NMR (CDCl3, 100 MHz): δ 209.8, 167.7,
134.6, 131.6, 123.7, 86.1, 80.7, 73.1, 70.9, 57.7; HRMS (ESI) m/z calcd for C14H12NO5 [M + H]+:
274.07155; found: 274.07366.
4.1.2.4. 2-((3S,3aR,6R,6aS)-6-hydroxy-6-vinylhexahydrofuro[3,2-b]furan-3-yl)isoindoline-1,3-dione (12).
To a cooled solution (0 °C) of 11 (1.46 g, 5.33 mmol, 1.0 equiv) in THF (10.6 mL) was slowly added
vinylmagnesium chloride (4.0 mL, 1.6 M in THF, 6.4 mmol, 1.2 equiv). The reaction was warmed to room
temperature and stirred for 16 h, then quenched at 0 °C with saturated NH4Cl solution. The aqueous
layer was extracted with ethyl acetate (3 × 25 mL), and the combined organic layers were washed with
brine, dried over anhydrous sodium sulfate, filtered, and concentrated under vacuum. The residue was
purified by chromatography using 30 -50% ethylacetate/ hexanes to give the title compound as a white

solid product (775 mg, 2.57 mmol, 48% yield). This reaction was run three times under similar conditions
1
with an average yield of 43% (range 36-48%). H NMR (CDCl3, 400 MHz): δ 7.88-7.81 (m, 2H), 7.77-7.70
(m, 2H), 5.95 (dd, J = 17.3, 10.9 Hz, 1H), 5.49 (dd, J = 17.3, 1.2 Hz, 1H), 5.24 (dd, J = 10.9, 1.2 Hz, 1H), 5.06
(dd, J = 4.9, 2.0 Hz, 1H), 4.90 (ddd, J = 7.5, 5.8, 1.8 Hz, 1H), 4.65 (d, J = 4.9 Hz, 1H), 4.34 (dd, J = 9.3, 7.5
Hz, 1H), 4.07 (dd, J = 9.3, 5.9 Hz, 1H), 3.79, 3.78 (ABq, J = 9.3 Hz, 2H), 2.97 (s, 1H); 13C NMR (CDCl3, 100
MHz): δ 167.7, 138.2, 134.5, 131.8, 123.6, 115.2, 87.8, 86.4, 79.7, 76.3, 72.3, 57.5; HRMS (ESI) m/z calcd
for C16H15NO5Na [M + Na]+: 324.08479; found: 324.08545.
4.1.2.5. 2-((3S,3aR,6R,6aS)-6-hydroxy-6-((E)-2-(6-methoxyquinolin-4-yl)vinyl)hexahydrofuro[3,2-b]furan-
3-yl)isoindoline-1,3-dione (15). 4-bromo-6-methoxyquinoline (13, 0.6792 g, 2.853 mmol, 1.5 equiv) and
12 (0.5730 g, 1.902 mmol, 0.794 mmol, 1.0 equiv) were dissolved in DMF (19 mL) under argon. Pd(dba)2
(0.1094 g, 0.1902 mmol, 0.10 equiv) was then added followed by slow addition of trimethylamine (0.80
mL, 5.7 mmol, 3.0 equiv). The reaction was stirred 30 min at room temperature and 24 h at 120 °C. After
cooling to room temperature, the reaction was quenched by addition of water and the aqueous layer
extracted repeatedly with ethyl acetate. The combined organic layers were washed with brine, dried
over Na2SO4, filtered, and concentrated under vacuum. The residue was purified by chromatography
(20-80% ethyl acetate in hexanes) to afford the title compound as a yellow solid, contaminated with
small impurities (580.3 mg, 1.266 mmol, 67%). This reaction was repeated under similar conditions once
1
with a yield of 84%, although the impurity profile in that case was worse. H NMR (CDCl3, 400 MHz): δ
8.72 (d, J = 4.6 Hz, 1H), 8.04 (d, J = 9.2 Hz, 1H), 7.89-7.83 (m, 2H), 7.79-7.72 (m, 2H), 7.55 (d, J = 15.7 Hz,
1H), 7.42 (d, J = 4.6 Hz, 1H), 7.39 (dd, J = 9.2, 2.7 Hz, 1H), 7.33 (d, J = 2.7 Hz, 1H), 6.50 (d, J = 15.7 Hz, 1H),
5.19 (dd, J = 4.9, 1.9 Hz, 1H), 4.98 (ddd, J = 7.4, 5.8, 1.8 Hz, 1H), 4.83 (d, J = 4.9 Hz, 1H), 4.43 (dd, J = 9.3,
7.5 Hz, 1H), 4.14 (dd, J = 9.4, 5.8 Hz, 1H), 3.96 (s, 3H), 3.94 (s, 2H), 3.31 (s, 1H).

4.1.2.6.
2-((3S,3aR,6R,6aS)-6-((E)-2-(3-fluoro-6-methoxy-1,5-naphthyridin-4-yl)vinyl)-6-
hydroxyhexahydrofuro[3,2-b]furan-3-yl)isoindoline-1,3-dione (16). A solution of 8-bromo-7-fluoro-2-
methoxy-1,5-naphthyridine (14, 200 mg, 0.778 mmol, 1.0 equiv), 12 (240 mg, 0.794 mmol, 1.02 equiv),
Pd(OAc)2 (17.4 mg, 77.8 mmol), and Ag2CO3 (429 mg, 1.56 mmol, 2.0 equiv) in benzene was refluxed at
100 °C under argon for 5 days. The reaction mixture was cooled to room temperature, then diluted with
ethyl acetate, and the insoluble material filtered off over celite. After washing the celite with ethyl
acetate, solvent was removed from the filtrate under vacuum, and the residue purified by
chromatography with 15-50% ethylacetate / hexanes to give the title compound as a white solid product
(145 mg, 0.304 mmol, 39% yield). This reaction was run three times under similar conditions (48 – 120 h
1
at reflux) with an average yield of 47% (range 31-71%). H NMR (CDCl3, 400 MHz): δ 8.63 (d, J = 2.1 Hz,
1H), 8.15 (d, J = 9.0 Hz, 1H), 7.88-7.81 (m, 2H), 7.76-7.70 (m, 2H), 7.67 (d, J = 16.4 Hz, 1H), 7.35 (d, J =
16.4 Hz, 1H), 7.06 (d, J = 9.0 Hz, 1H), 5.19 (dd, J = 4.9, 1.9 Hz, 1H), 4.96 (ddd, J = 7.5, 5.8, 1.8 Hz, 1H), 4.84
(d, J = 4.9 Hz, 1H), 4.40 (dd, J = 9.3, 7.5 Hz, 1H), 4.12 (dd, J = 9.3, 5.9 Hz, 1H), 4.08 (s, 3H), 3.94 (s, 2H),
13
3.29 (s, 1H); C NMR (CDCl3, 100 MHz): δ 167.7, 162.5, 157.6, 155.0, 140.8, 140.7, 140.4, 140.3, 140.2,
139.03, 139.00, 138.6, 138.3, 134.5, 131.7, 125.03, 124.95, 123.6, 118.1, 115.32, 115.29, 88.0, 86.6,
80.4, 76.6, 72.5, 57.5, 54.2; HRMS (ESI) m/z calcd for C25H20FN3O6Na [M + Na]+: 500.12339 ; found:
500.12531.
4.1.2.7. 2-((3S,3aR,6R,6aS)-6-hydroxy-6-(2-(6-methoxyquinolin-4-yl)ethyl)hexahydrofuro[3,2-b]furan-3-
yl)isoindoline-1,3-dione (17). A mixture of 15 (0.3431 g, 0.7484 mmol, 1.0 equiv) and 10% Pd/C (0.080 g)
in 1:1 methanol / ethyl acetate (30 mL) was vigorously stirred under 1 atm H2 (balloon) for 48 h. The
mixture was then filtered through celite, washing several times with ethyl acetate, methanol, and DCM.

The filtrate was concentrated under reduced pressure and the resulting residue purified by
chromatography (2 – 5% methanol in DCM) to afford the title compound as a yellow solid, contaminated
with solvent residues and ca. 5% unreacted starting material (0.2464 g, 0.5351 mmol, 72% yield). This
reaction was run three times under similar conditions with an average yield of 70% yield (range 49-90%),
1
but the product was never completely pure. H NMR (CDCl3, 400 MHz): δ 8.65 (d, J = 4.4 Hz, 1H), 8.04-
7.98 (m, 1H), 7.86-7.80 (m, 2H), 7.76-7.69 (m, 2H), 7.37-7.32 (m, 2H), 7.22 (d, J = 4.5 Hz, 1H), 5.06 (dd, J =
4.9, 2.1 Hz, 1H), 4.90 (ddd, J = 7.7, 6.6, 2.0 Hz, 1H), 4.65 (d, J = 4.9 Hz, 1H), 4.35 (dd, J = 9.2, 7.7 Hz, 1H),
4.09 (dd, J = 9.3, 6.4 Hz, 1H), 3.94 (s, 3H), 3.81, 3.78 (ABq, J = 9.3 Hz, 2H), 3.30-3.15 (m, 3H), 2.12-1.94
(m, 2H); HRMS (ESI) m/z calcd for C16H25N2O6 [M + H]+: 461.17126 ; found: 461.17102.
4.1.2.8.
2-((3S,3aR,6R,6aS)-6-(2-(3-fluoro-6-methoxy-1,5-naphthyridin-4-yl)ethyl)-6-
hydroxyhexahydrofuro[3,2-b]furan-3-yl)isoindoline-1,3-dione (18). Compound 16 (145 mg, 0.304 mmol,
1.0 equiv) was dissolved in a 1:1 mixture of ethyl acetate and methanol (12 mL). To the solution was
added 10% Pd/C (0.0646 mg), and the mixture was subjected to hydrogenation at 50 psi in a Parr
hydrogenator. After 24 h, the reaction mixture was filtered over celite, and then washed with ethyl
acetate and methanol. The filtrate was concentrated under reduced pressure, and the crude product
was purified by chromatography with 20-50% ethyl acetate / hexanes to give the title compound as a
whitish solid (95.4 mg, 0.199 mmol, 66% yield, contaminated with ethyl acetate). This reaction was run
1
three times under similar conditions with an average yield of 70% (range 66-78%). H NMR (CDCl3, 400
MHz): δ 8.61 (s, 1H), 8.16 (d, J = 9.0 Hz, 1H), 7.87-7.81 (m, 2H), 7.77-7.70 (m, 2H), 7.06 (d, J = 9.0 Hz, 1H),
5.04 (dd, J = 5.0, 2.1 Hz, 1H), 4.92 – 4.86 (m, 1H), 4.72 (d, J = 5.0 Hz, 1H), 4.34 (dd, J = 9.2, 7.6 Hz, 1H),
4.11 (s, 3H), 4.07 (dd, J = 9.3, 6.3 Hz, 1H), 3.82 (s, 2H), 3.46 – 3.30 (m, 2H), 3.13 (s, 1H), 2.10 – 1.92 (m,
2H); 13C NMR (CDCl3, 100 MHz): δ 167.7, 162.6, 158.4, 155.8, 141.6, 141.5, 140.3, 138.62, 138.60, 138.2,
137.9, 134.5, 132.1, 131.9, 131.8, 123.6, 115.43, 115.40, 86.9, 86.2, 79.6, 76.2, 72.0, 57.4, 54.1, 35.5,
18.0, 17.9; HRMS (ESI) m/z calcd for C25H22FN3O6Na [M + Na]+: 502.13904 ; found: 502.13925.

4.1.2.9.
(3R,3aS,6S,6aR)-6-amino-3-(2-(6-methoxyquinolin-4-yl)ethyl)hexahydrofuro[3,2-b]furan-3-ol
(19). Compound 17 (0.1530 g, 0.3323 mmol, 1.0 equiv) and ethanolamine (0.30 mL, 5.0 mmol, 15 equiv)
were combined in ethyl acetate and stirred at 70 °C for 16 h. Ethyl acetate was removed under vacuum,
and the residue was dissolved in DCM and washed with brine. The aqueous layer was extracted with
DCM, and the combined organic layers were dried over Na2SO4, filtered, and concentrated under
vacuum. The resulting residue was purified by chromatography (0 – 8% methanol in DCM) to afford the
title compound as a thick, clear oil (0.0867 g, 0.262 mmol, 79%, contaminated with small amount
methanol). This reaction was run two additional times under similar conditions with yields of 72% and
1
94%, but the products were not completely pure. H NMR (CD3OD, 400 MHz): δ 8.55 (d, J = 4.5 Hz, 1H),
7.91 (d, J = 9.2 Hz, 1H), 7.45 (d, J = 2.7 Hz, 1H), 7.39 (dd, J = 9.2, 2.6 Hz, 1H), 7.36 (d, J = 4.6 Hz, 1H), 4.37
(s, 2H), 4.06 (dd, J = 9.3, 4.5 Hz, 1H), 3.96 (s, 3H), 3.79 (dd, J = 9.3, 1.7 Hz, 1H), 3.75 (d, J = 9.2 Hz, 1H),
3.66 (d, J = 9.2 Hz, 1H), 3.50 (dd, J = 4.4, 1.7 Hz, 1H), 3.30-3.17 (m, 2H), 2.02-1.89 (m, 2H); HRMS (ESI)
m/z calcd for C18H23N2O4 [M + H]+: 331.16578 ; found: 331.16589.
4.1.2.10.
(3R,3aS,6S,6aR)-6-amino-3-(2-(3-fluoro-6-methoxy-1,5-naphthyridin-4-
yl)ethyl)hexahydrofuro[3,2-b]furan-3-ol (20). To a solution of 18 (118.4 mg, 0.2469 mmol, 1.0 equiv) in
ethyl acetate (8.3 mL), in a flask fitted with a reflux condenser, was added ethanolamine (220 µL, 3.6
mmol, 15 equiv). The reaction was stirred at 70 °C for 24 h. After cooling to room temperature, ethyl
acetate was removed under vacuum, and then the residue was dissolved in DCM, and washed with
brine. The two layers were separated, and then aqueous layer was extracted with DCM (3×10 mL).
Combined organic layers were dried over anhydrous Na2SO4, and then filtered, and concentrated under
vacuum. The residue was purified by chromatography with 0-10% methanol / DCM to give the title
compound (78.7 mg, 0.225 mmol, 91%) as a whitish solid. This reaction was run three times under

1
similar conditions with an average yield of 91% (range 91-92%). H NMR (CDCl3, 400 MHz): δ 8.58 (s,
1H), 8.14 (d, J = 9.0 Hz, 1H), 7.03 (d, J = 9.0 Hz, 1H), 4.37 (d, J = 4.4 Hz, 1H), 4.25 (d, J = 4.4 Hz, 1H), 4.06
(s, 3H), 3.99 (dd, J = 9.1, 4.2 Hz, 1H), 3.77 (br d, J = 9.1 Hz, 1H), 3.76 (d, J = 9.3 Hz, 1H), 3.67 (d, J = 9.3 Hz,
1H), 3.59 (br d, J = 3.6 Hz, 1H), 1H), 3.41 – 3.25 (m, 2H), 2.06 – 1.83 (m, 2H); 13C NMR (CDCl3, 100 MHz): δ
162.5, 158.3, 155.8, 141.5, 141.4, 140.3, 138.60, 138.58, 138.1, 137.8, 132.0, 131.9, 115.35, 115.32,
89.9, 85.8, 79.9, 77.1, 76.1, 58.6, 53.9, 36.1, 17.99, 17.97; HRMS (ESI) m/z calcd for C17H21FN3O4 [M + H]+:
350.15610 ; found: 350.15143.
4.1.3. General reductive amination procedure
Reductive amination to synthesize compounds 21-23 (from amine 19) and compounds 6, 7, and
24-26 (from amine 20) was carried out using a minor modification of our earlier conditions20 as follows:
The amine (1.0 equiv) and the requisite aldehyde (1.2 equiv) were combined in DCM (0.05 M), and ZnCl2
(0.2 equiv) was added. The reaction was stirred 1h, whereupon NaBH3CN (3.0 equiv) was added
(effervescence observed). The reaction was stirred overnight, methanol was removed in vacuo, and the
residue was dissolved in DCM and washed with brine. The aqueous layer was extracted with DCM, and
the combined organic layers were dried over Na2SO4 and concentrated in vacuo. The crude product was
dissolved in DCM, concentrated onto silica gel, and purified by flash chromatography on silica gel using a
gradient elution (0 to 5 or 10% methanol in DCM). Concentration in vacuo afforded the reported
compounds.
4.1.3.1. (3R,3aS,6S,6aR)-6-(((2,3-dihydro-[1,4]dioxino[2,3-c]pyridin-7-yl)methyl)amino)-3-(2-(3-fluoro-6-
methoxy-1,5-naphthyridin-4-yl)ethyl)hexahydrofuro[3,2-b]furan-3-ol (6). From 20 according to the
general reductive amination procedure with 2,3-dihydro-[1,4]dioxino[2,3-c]pyridine-7-carbaldehyde
1
(Yield 39.4 mg, 79.0 µmol, 64.0%). H NMR (CD3OD, 400 MHz): δ 8.58 (d, J = 0.6 Hz, 1H), 8.14 (d, J = 9.1
Hz, 1H), 7.97 (s, 1H), 7.11 (d, J = 9.1 Hz, 1H), 6.94 (s, 1H), 4.45 (br d, J = 4.6 Hz, 1H), 4.37 – 4.32 (m, 3H),
4.29 – 4.25 (m, 2H), 4.08 (s, 3H), 4.02 (dd, J = 9.4, 4.9 Hz, 1H), 3.84 (dd, J = 9.4, 2.6 Hz, 1H), 3.79 (d, J =
9.0 Hz, 1H), 3.76, 3.75 (ABq, J = 14.5 Hz, 2H), 3.64 (d, J = 9.0 Hz, 1H), 3.39 – 3.24 (m, 3H, overlaps with
13
solvent signal), 1.99 – 1.81 (m, 2H); C NMR (CD3OD, 100 MHz): δ 164.1, 159.6, 157.0, 153.7, 152.5,
142.7, 142.6, 142.1, 140.6, 139.43, 139.41, 139.0, 138.8, 138.5, 134.1, 134.0, 116.8, 116.7, 112.4, 88.6,
87.2, 81.1, 76.4, 74.4, 66.5, 66.0, 65.4, 54.6, 52.9, 37.1, 18.90, 18.88; HRMS (ESI) m/z calcd for
C25H28FN4O6 [M + H]+: 499.19929; Found: 499.19910.

4.1.3.2.
6-((((3S,3aR,6R,6aS)-6-(2-(3-fluoro-6-methoxy-1,5-naphthyridin-4-yl)ethyl)-6-
hydroxyhexahydrofuro[3,2-b]furan-3-yl)amino)methyl)-2H-pyrido[3,2-b][1,4]oxazin-3(4H)-one (7). From
20 according to the general reductive amination procedure with 3-oxo-3,4-dihydro-2H-pyrido[3,2-
1
b][1,4]oxazine-6-carbaldehyde (Yield 33.8 mg, 66.1 µmol, 62.2%). H NMR ([CD3]2SO, 400 MHz): δ 11.16
(s, 1H), 8.76 (s, 1H), 8.27 (d, J = 9.0 Hz, 1H), 7.30 (d, J = 8.1 Hz, 1H), 7.23 (d, J = 9.0 Hz, 1H), 7.00 (d, J = 8.1
Hz, 1H), 4.65 (s, 1H), 4.60 (s, 2H), 4.36 (br d, J = 4.3 Hz, 1H), 4.19 (d, J = 4.4 Hz, 1H), 4.04 (s, 3H), 3.88 (dd,
J = 9.1, 4.8 Hz, 1H), 3.73 (dd, J = 9.1, 2.0 Hz, 1H), 3.71 – 3.64 (m, 3H), 3.49 (d, J = 8.6 Hz, 1H), 3.27 – 3.15
(m, 3H), 1.86 – 1.70 (m, 2H); 13C NMR ([CD3]2SO, 100 MHz): δ 165.8, 162.0, 157.7, 155.1, 140.78, 140.76,
140.7, 140.4, 138.13, 138.11, 138.07, 137.8, 137.6, 131.7, 131.6, 123.3, 116.8, 115.30, 115.28, 86.6,
85.3, 79.4, 74.2, 72.9, 66.7, 64.7, 53.3, 51.6, 35.7, 17.5; HRMS (ESI) m/z calcd for C25H27FN5O6 [M + H]+:
512.19454; Found: 512.19489.
4.1.3.3.
(3R,3aS,6S,6aR)-6-(((2,3-dihydrobenzo[b][1,4]dioxin-6-yl)methyl)amino)-3-(2-(6-
methoxyquinolin-4-yl)ethyl)hexahydrofuro[3,2-b]furan-3-ol (21). From 19 according to the general
reductive amination procedure with 2,3-dihydrobenzo[b][1,4]dioxine-6-carbaldehyde (Yield 28.9 mg,
60.4 mmol, 66.5%). 1H NMR (CD3OD, 400 MHz): δ 8.55 (d, J = 4.5 Hz, 1H), 7.91 (d, J = 9.2 Hz, 1H), 7.46 (d,
J = 2.7 Hz, 1H), 7.39 (dd, J = 9.2, 2.7 Hz, 1H), 7.36 (d, J = 4.6 Hz, 1H), 6.86-6.74 (m, 3H), 4.48 (dd, J = 4.7,
1.0 Hz, 1H), 4.32 (d, J = 4.8 Hz, 1H), 4.22-4.17 (m, 4H), 4.05 (dd, J = 9.4, 5.2 Hz, 1H), 3.96 (s, 3H), 3.81 (dd,
J = 9.4, 3.2 Hz, 1H), 3.74 (d, J = 9.1 Hz, 1H), 3.68, 3.65 (ABq, , J = 13.2 Hz, 2H), 3.66 (d, J = 9.4 Hz, 1H),
3.28-3.19 (m, 2H), 2.03 – 1.88 (m, 2H); 13C NMR (CD3OD, 100 MHz): δ 159.6, 149.7, 148.2, 145.0, 144.7,

144.3, 133.5, 131.2, 129.9, 123.3, 122.6, 122.3, 118.4, 118.1, 102.9, 88.6, 87.1, 81.0, 76.2, 74.4, 66.7,
65.60, 65.58, 56.2, 52.1, 37.9, 27.4; HRMS (ESI) m/z calcd for C27H31N2O6 [M + H]+: 479.21821; Found:
479.21719.
4.1.3.4. (3R,3aS,6S,6aR)-6-(benzylamino)-3-(2-(6-methoxyquinolin-4-yl)ethyl)hexahydrofuro[3,2-b]furan-
3-ol (22). From 19 according to the general reductive amination procedure with benzaldehyde (Yield
1
20.3 mg, 48.3 mmol, 59.1%). H NMR (CD3OD, 400 MHz): δ 8.56 (d, J = 4.5 Hz, 1H), 7.91 (d, J = 9.2 Hz,
1H), 7.47 (d, J = 2.7 Hz, 1H), 7.40 (dd, J = 9.2, 2.7 Hz, 1H), 7.38-7.29 (m, 5H), 7.27 – 7.22 (m, 1H), 4.51 (dd,
J = 4.8, 0.9 Hz, 1H), 4.33 (d, J = 4.8 Hz, 1H), 4.06 (dd, J = 9.4, 5.2 Hz, 1H), 3.96 (s, 3H), 3.84 (dd, J = 9.4, 3.2
Hz, 1H), 3.81, 3.78 (ABq, , J = 13.0 Hz, 2H), 3.75 (d, J = 9.1 Hz, 1H), 3.67 (d, J = 9.1 Hz, 1H), 3.36 – 3.30 (m,
1H, partially obscured by solvent), 3.29 – 3.18 (m, 2H), 2.03 – 1.89 (m, 2H); 13C NMR (CD3OD, 100 MHz):
δ 159.6, 149.7, 148.2, 144.7, 140.5, 131.2, 129.9, 129.7, 129.5, 128.3, 123.3, 122.3, 102.9, 88.6, 87.1,
81.0, 76.2, 74.5, 65.9, 56.2, 52.7, 38.0, 27.4; HRMS (ESI) m/z calcd for C25H29N2O4 [M + H]+: 421.21273;
Found: 421.21207.
4.1.3.5.
(3R,3aS,6S,6aR)-3-(2-(6-methoxyquinolin-4-yl)ethyl)-6-((4-
methylbenzyl)amino)hexahydrofuro[3,2-b]furan-3-ol (23). From 19 according to the general reductive
1
amination procedure with 4-methylbenzaldehyde (Yield 26.7 mg, 61.4 mmol, 70.1%). H NMR (CD3OD,
400 MHz): δ 8.55 (d, J = 4.5 Hz, 1H), 7.91 (d, J = 9.2 Hz, 1H), 7.45 (d, J = 2.7 Hz, 1H), 7.39 (dd, J = 9.2, 2.7
Hz, 1H), 7.35 (d, J = 4.6 Hz, 1H), 7.24-7.19 (m, 2H), 7.15-7.10 (m, 2H), 4.49 (dd, J = 4.8, 1.3 Hz, 1H), 4.32
(d, J = 4.8 Hz, 1H), 4.05 (dd, J = 9.4, 5.2 Hz, 1H), 3.95 (s, 3H), 3.82 (dd, J = 9.4, 3.2 Hz, 1H), 3.76, 3.72 (ABq,
, J = 12.9 Hz, 2H), 3.74 (d, J = 8.9 Hz, 1H), 3.66 (d, J = 9.1 Hz, 1H), 3.34 – 3.30 (m, 1H, partially obscured by

13
solvent), 3.29 – 3.16 (m, 2H), 2.29 (s, 3H), 2.01 – 1.88 (m, 2H); C NMR (CD3OD, 100 MHz): δ 159.6,
149.7, 148.2, 144.7, 138.0, 137.3, 131.2, 130.1, 129.9, 129.6, 123.3, 122.3, 102.9, 88.6, 87.1, 81.0, 76.2,
74.5, 65.8, 56.2, 52.4, 37.9, 27.3, 21.1; HRMS (ESI) m/z calcd for C26H31N2O4 [M + H]+: 435.22838; Found:
435.22787.
4.1.3.6.
(3R,3aS,6S,6aR)-6-(((2,3-dihydrobenzo[b][1,4]dioxin-6-yl)methyl)amino)-3-(2-(3-fluoro-6-
methoxy-1,5-naphthyridin-4-yl)ethyl)hexahydrofuro[3,2-b]furan-3-ol (24). From 20 according to the
general reductive amination procedure with 2,3-dihydrobenzo[b][1,4]dioxine-6-carbaldehyde (Yield 43.7
mg, 87.8 µmol, 84.5%). 1H NMR (CD3OD, 400 MHz): δ 8.57 (d, J = 0.9 Hz, 1H), 8.13 (d, J = 9.1 Hz, 1H), 7.10
(d, J = 9.1 Hz, 1H), 6.82 – 6.71 (m, 3H), 4.43 (dd, J = 4.7, 0.8 Hz, 1H), 4.32 (d, J = 4.8 Hz, 1H), 4.21-4.15 (m,
4H), 4.08 (s, 3H), 4.01 (dd, J = 9.4, 5.1 Hz, 1H), 3.82 – 3.76 (m, 2H), 3.68 – 3.59 (m, 3H), 3.38 – 3.23 (m,
3H, overlaps with solvent signal), 1.98 – 1.80 (m, 2H), small amount DCM present; 13C NMR (CD3OD, 100
MHz): δ 164.1, 159.6, 157.0, 144.9, 144.2, 142.7, 142.6, 140.6, 139.40, 139.38, 138.8, 138.5, 134.1,
134.0, 133.5, 122.5, 118.4, 118.1, 116.8, 116.7, 88.6, 87.2, 81.1, 76.3, 74.3, 65.7, 65.6, 65.5, 54.6, 52.0,
37.1, 18.89, 18.87; HRMS (ESI) m/z calcd for C26H29FN3O6 [M + H]+: 498.20404; Found: 498.20401.
O
NH
H
H
HO
O
O
N
F
N
(3R,3aS,6S,6aR)-3-(2-(3-fluoro-6-methoxy-1,5-naphthyridin-
4-yl)ethyl)-6-((4-methylbenzyl)amino)hexahydrofuro[3,2-
b]furan-3-ol
4.1.3.7.
(3R,3aS,6S,6aR)-3-(2-(3-fluoro-6-methoxy-1,5-naphthyridin-4-yl)ethyl)-6-((4-
methylbenzyl)amino)hexahydrofuro[3,2-b]furan-3-ol (25). From 20 according to the general reductive

amination procedure with 4-methylbenzaldehyde (Yield 32.6 mg, 71.9 µmol, 82.3%). 1H NMR (CDCl3, 400
MHz): δ 8.61 (d, J = 0.5 Hz, 1H), 8.17 (d, J = 9.0 Hz, 1H), 7.19 (d, J = 8.0 Hz, 2H), 7.13 (d, J = 7.9 Hz, 2H),
7.06 (d, J = 9.0 Hz, 1H), 4.43 (br d, J = 4.6 Hz, 1H), 4.38 (d, J = 4.7 Hz, 1H), 4.08 (s, 3H), 4.02 (dd, J = 9.3,
4.7 Hz, 1H), 3.87 (dd, J = 9.2, 2.2 Hz, 1H), 3.82 – 3.74 (m, 2H), 3.78 (d, J = 9.3 Hz, 1H), 3.71 (d, J = 9.3 Hz,
1H), 3.46 – 3.41 (m, 1H), 3.41 -3.27 (m, 2H), 2.32 (s, 3H), 2.05 – 1.87 (m, 2H); 13C NMR (CDCl3, 100 MHz):
δ 162.6, 158.4, 155.8, 141.6, 141.5, 140.3, 138.7, 138.6, 138.2, 137.9, 137.0, 136.6, 132.1, 132.0, 129.4,
128.2, 115.39, 115.36, 87.6, 86.0, 79.8, 77.1, 74.1, 64.7, 54.0, 51.9, 36.1, 21.2, 18.05, 18.03; HRMS (ESI)
m/z calcd for C25H29FN3O4 [M + H]+: 454.21421; Found: 454.21484.
4.1.3.8.
(3R,3aS,6S,6aR)-6-((3,4-dichlorobenzyl)amino)-3-(2-(3-fluoro-6-methoxy-1,5-naphthyridin-4-
yl)ethyl)hexahydrofuro[3,2-b]furan-3-ol (26). From 20 according to the general reductive amination
procedure with 3,4-dichlorobenzaldehyde (Yield 38.2 mg, 75.1 µmol, 92.1%). 1H NMR (CDCl3, 400 MHz):
δ 8.60 (s, 1H), 8.16 (d, J = 9.0 Hz, 1H), 7.42 (d, J = 1.9 Hz, 1H), 7.37 (d, J = 8.2 Hz, 1H), 7.14 (dd, J = 8.2, 2.0
Hz, 1H), 7.06 (d, J = 9.0 Hz, 1H), 4.41 (br d, J = 4.6 Hz, 1H), 4.36 (d, J = 4.7 Hz, 1H), 4.08 (s, 3H), 4.02 (dd, J
= 9.3, 4.7 Hz, 1H), 3.85 (dd, J = 9.3, 2.2 Hz, 1H), 3.82 – 3.74 (m, 2H), 3.78 (d, J = 9.4 Hz, 1H), 3.71 (d, J =
13
9.4 Hz, 1H), 3.42 – 3.37 (m, 1H), 3.37 – 3.27 (m, 2H), 2.06 – 1.85 (m, 2H); C NMR (CDCl3, 100 MHz): δ
162.6, 158.4, 155.8, 141.6, 141.5, 140.4, 140.0, 138.67, 138.65, 138.2, 137.9, 132.7, 132.0, 131.9, 131.3,
130.5, 130.0, 127.4, 115.41, 115.39, 87.5, 86.1, 79.8, 77.1, 73.9, 64.8, 54.0, 50.9, 36.1, 18.03, 18.01;
HRMS (ESI) m/z calcd for C24H25Cl2FN3O4 [M + H]+: 508.12062; Found: 508.12082.
4.1.4. X-ray crystallography of compound 12
4.1.4.1. Sample preparation
The sample for X-ray crystallography was prepared as follows: 23.2 mg 12 was dissolved in toluene in a
test tube. The test tube was placed in a larger chamber containing pentane. The chamber was covered
and allowed to sit undisturbed for three weeks, whereupon crystals were collected.
4.1.4.2. Data collection and analysis
The data collection crystal was a colorless plate cut from the end of a rod. Examination of the
diffraction pattern on a Bruker D8 Venture diffractometer system with a Photon II detector indicated a

monoclinic crystal system. All work was done at 150 K using an Oxford Cryosystems Cryostream Cooler.
The data collection strategy was set up to measure a quadrant of reciprocal space with a redundancy
factor of 6, which means that 90% of the reflections were measured at least 6 times. Omega and phi
scans with a frame width of 0.5˚ and a frame time of 10 seconds were used. The data frames were
collected using the program APEX3 and processed with the SAINT program within APEX3.49 Absorption
and beam corrections were made with the multiscan technique in SADABS.50
The structure was solved by the direct methods procedure in SHELXT51 in P21. Full-matrix least-
squares refinements based on F2 were performed in SHELXL-2014/7,52 as incorporated in the WinGX
package.53 The correct enantiomer was chosen based on a known chiral center in the molecule.
The hydroxyl group hydrogen atom was refined isotropically. The remaining hydrogen atoms
were added at calculated positions using a riding model with U(H) = 1.2 * Ueq(bonded carbon atom).
The final refinement cycle was based on 3289 intensities and 204 variables and resulted in agreement
factors of R1(F) = 0.026 and wR2(F2) = 0.069. For the subset of data with I > 2*sigma(I), the R1(F) value
is 0.026 for 3261 reflections. The final difference electron density map contains maximum and minimum
peak heights of 0.23 and -0.19 e/Ǻ3. Neutral atom scattering factors were used and include terms for
anomalous dispersion.54
4.2. Microbiology
4.2.1. Determination of minimum inhibitory concentrations by broth dilution
MIC values in Tables 1-3 were determined according to the guidelines established by the Clinical
and Laboratory Standards Institute.28
4.2.2. Determination of minimum inhibitory concentrations on agar
In order to determine the spontaneous mutation frequency of investigational agents (Table 4),
the MIC was first determined using the agar dilution method as recommended by the Clinical and
Laboratory Standards Institute.28,41 This is because MIC values generated by agar vs. broth assays may
differ, and the concentration of drug utilized in the spontaneous mutation assay must be as close to the
agar dilution MIC as possible in order to produce meaningful results.
The test agents and ciprofloxacin were assayed using a drug concentration range of
0.008 – 16 µg/mL. Triplicate independent inocula of S. aureus ATCC 29213 were evaluated.
4.2.3. Spontaneous mutation frequency determination
Serial dilutions of the test compounds (6 and 24) were made in DMSO; ciprofloxacin was diluted
in water. Agar plates for selection of mutants were prepared by mixing 0.5 mL of concentrated drug
(40X the agar dilution MIC) with 49.5 mL of molten (52°C) Mueller-Hinton Agar (MHA), and pouring the
mixture into 15-by-150 mm petri dishes. The plates were allowed to solidify and dry at room
temperature prior to inoculation. Duplicate plates were prepared to contain the agents at 8X and 16X
the median agar dilution MIC. In addition, duplicate 10 x 100 mm plates (standard small petri plates)
were also prepared with 8X and 16X drug and were spot-inoculated with the standard inoculum density
used for agar dilution according to CLSI;28 this was to confirm that the drug content of the plates was
inhibitory for a standard inoculum thus validating that the stock drug content of the plates exceeded the
MIC.

The inoculum for the assay was prepared by streaking S. aureus ATCC 29213 onto Mannitol Salt
Agar (MSA; Hardy Diagnostics, Santa Maria, CA; Cat. No. C6231; Lot No. 422932) and incubating at 35°C
overnight. Using a single isolated colonies from the MSA plate, S. aureus was sub-cultured onto 10 TSAB
plates and grown for approximately 20 hr at 35°C. Using a sterile swab, several well-isolated colonies
were removed and resuspended in CAMHB at a heavy concentration equivalent to 1.7 on the
turbidimeter to reach a target inoculum of 109 to 1010 CFU on each spontaneous mutation plate. Each
plate was inoculated with 0.25 mL of the cell suspension; each test concentration was inoculated onto
two plates (except 6 which was inoculated onto a single plate for each test concentration due to
insufficient drug volume) so that each drug was tested at 8X and 16X the MIC in duplicate. In addition, a
portion of the inoculum was enumerated by making serial 10-fold dilutions, tracking them across the
surface of a MHA plate, and counting colonies after a 24 hr incubation at 35°C.
Once the inoculum was absorbed into the agar, the drug-containing plates were incubated at
35°C for 48 hr and the colonies resulting colonies were counted. The spontaneous mutation frequency
was determined by dividing the number of colonies that appeared at a given drug concentration,
averaging the counts from the duplicate plates, and dividing by the number of bacteria applied to the
agar surface. Number of colonies were adjusted based on expected morphology (e.g. alpha hemolysis on
TSAB) and/or confirmatory ID by MALDI.
4.3. Topoisomerase inhibition
4.3.1. Staphylococcus aureus wild type and D83N mutant DNA gyrase supercoiling assays
The wild type DNA gyrase assay was carried out at Inspiralis, Ltd. (Norwich, UK) using the
previously reported method.20 The D83N mutant DNA gyrase supercoiling assay was carried out in-
house utilizing D83N mutant enzyme and an assay kit purchased from Inspiralis, Ltd. according to the
manufacturer’s instructions.
4.3.2. Staphylococcus aureus TopoIV decatenation assay
This assay was carried out in-house using an assay kit purchased from Inspiralis, Ltd. with slight
modifications previously reported as Method B.20
4.4. In vitro safety assessment
4.4.1. Determination of hERG IC50 values
These results reported in Table 5 were determined at Charles River (Cleveland, OH, USA) under
the previously reported conditions.20
4.4.2. Comprehensive in vitro proarrhythmia assays
4.4.2.1. General information
The results determined in Table 6 were determined at Eurofins Panlabs (St. Charles, MO, USA)
using a QPatch HT (Sophion, Denmark) automated patch clamp system at three concentrations (3, 30,
and 100
µM). All data were filtered for seal quality, seal drop, and current amplitude. A stock solution of
the test compound at 30 mM concentration was prepared in DMSO.

4.4.2.2. Nav1.5 Human Sodium Ion Channel Cell Based QPatch CiPA Assay
The parameters measured were the maximum inward current evoked on stepping to -15mV
from the test pulse. The peak current amplitude was calculated before and after compound addition and
the amount of block was assessed by dividing the Test compound current amplitude by the Control
current amplitude. Control is the mean hNav1.5 current amplitude collected 15 seconds at the end of
the control; Test Compound is the mean hNav1.5 current amplitude collected in the presence of test
compound at each concentration. The positive control compound was tetracaine.
4.4.2.3. hERG Human Potassium Ion Channel Cell Based QPatch CiPA Assay
The parameters measured were the maximum tail current evoked on stepping to 40mV and
ramping back to -80mV from the test pulse. The peak current amplitude was calculated before and after
compound addition and the amount of block was assessed by dividing the Test compound current
amplitude by the Control current amplitude. Control is the mean hERG current amplitude collected 15
seconds at the end of the control; Test Compound is the mean hERG current amplitude collected in the
presence of test compound at each concentration. The positive control was E-4031.
4.4.2.4. KCNQ1/hminK Human Potassium Ion Channel Cell Based QPatch CiPA Assay
The parameters measured were the maximum outward current evoked on stepping to 60mV
from the test pulse. The peak current amplitude was calculated before and after compound addition and
the amount of block was assessed by dividing the Test compound current amplitude by the Control
current amplitude. Control is the mean hKCNQ1/hminK current amplitude collected 15 seconds at the
end of the control; Test Compound is the mean hKCNQ1/hminK current amplitude collected in the
presence of test compound at each concentration. The positive control was chromanol 293B.