Compound design guidelines for evading the efflux and permeation barriers of Escherichia coli with the oxazolidinone class of antibacterials: Test case for a general approach to improving whole cell Gram-negative activity

Article abstract of DOI:10.1016/j.bmcl.2017.10.018

Previously we reported the results from an effort to improve Gram-negative antibacterial activity in the oxazolidinone class of antibiotics via a systematic medicinal chemistry campaign focused entirely on C-ring modifications. In that series we set about testing if the efflux and permeation barriers intrinsic to the outer membrane of Escherichia coli could be rationally overcome by designing analogs to reside in specific property limits associated with Gram-negative activity: i) low MW (<400), ii) high polarity (clogD_7.4 <1), and iii) zwitterionic character at pH 7.4. Indeed, we observed that only analogs residing within these limits were able to overcome these barriers. Herein we report the results from a parallel effort where we explored structural changes throughout all three rings in the scaffold for the same purpose. Compounds were tested against a diagnostic MIC panel of Escherichia coli and Staphylococcus aureus strains to determine the impact of combining structural modifications in overcoming the OM barriers and in bridging the potency gap between the species. The results demonstrated that distributing the charge-carrying moieties across two rings was also beneficial for avoidance of the outer membrane barriers. Importantly, analysis of the structure-permeation relationship (SPR) obtained from this and the prior study indicated that in addition to MW, polarity, and zwitterionic character, having ≤4 rotatable bonds is also associated with evasion of the OM barriers. These combined results provide the medicinal chemist with a framework and strategy for overcoming the OM barriers in GNB in antibacterial drug discovery efforts.

Full text of DOI:10.1016/j.bmcl.2017.10.018

Accepted Manuscript
Compound design guidelines for evading the efflux and permeation barriers of
Escherichia coli with the oxazolidinone class of antibacterials: test case for a
general approach to improving whole cell Gram-negative activity.
Andrew Spaulding, Khuloud Takrouri, Pornachandran Mahalingam, Dillon C.
Cleary, Harold D. Cooper, Paola Zucchi, Westley Tear, Bilyana Koleva, Penny
J. Beuning, Elizabeth B. Hirsch, James B. Aggen
PII:
S0960-894X(17)30990-3
https://doi.org/10.1016/j.bmcl.2017.10.018
BMCL 25347
DOI:
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
10 July 2017
26 September 2017
7 October 2017
Please cite this article as: Spaulding, A., Takrouri, K., Mahalingam, P., Cleary, D.C., Cooper, H.D., Zucchi, P., Tear,
W., Koleva, B., Beuning, P.J., Hirsch, E.B., Aggen, J.B., Compound design guidelines for evading the efflux and
permeation barriers of Escherichia coli with the oxazolidinone class of antibacterials: test case for a general approach
to improving whole cell Gram-negative activity., Bioorganic & Medicinal Chemistry Letters (2017), doi: https://
doi.org/10.1016/j.bmcl.2017.10.018
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.

Graphical Abstract
To create your abstract, type over the instructions in the template box below.
Fonts or abstract dimensions should not be changed or altered.
Compound design guidelines for evading the efflux and permeation
barriers of Escherichia coli with the oxazolidinone class of
antibacterials: test case for a general approach to improving whole
cell Gram-negative activity
Leave this area blank for abstract info.
Andrew Spauldinga , Khuloud Takrouria ∗, Pornachandran Mahalingama, Dillon C. Clearya, Harold D. Cooperb, Paola Zucchic,
Westley Teara, Bilyana Kolevaa, Penny J. Beuninga, Elizabeth B. Hirschc, and James B. Aggend
F
F
Iterative Medicinal
Chemistry Campaign
O
O
O
R1
N
O
N
N
O
R2
NHAc
Linezolid
(LZD)
Properties associated with minimal

Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com
Compound design guidelines for evading the efflux and permeation barriers of Escherichia coli with the oxazolidinone class of
antibacterials: test case for a general approach to improving whole cell Gram-negative activity.
Andrew Spauldinga , Khuloud Takrouria ∗, Pornachandran Mahalingama, Dillon C. Clearya, Harold D. Cooperb, Paola Zucchic, Westley
Teara, Bilyana Kolevaa, Penny J. Beuninga, Elizabeth B. Hirschc, and James B. Aggend
aDepartment of Chemistry and Chemical Biology, Northeastern University, 360 Huntington Ave, Boston, MA 02115, USA
b9709 Bedder Stone Place, Bristow, VA 20136
cDepartment of Pharmacy and Health Systems Sciences, School of Pharmacy, Northeastern University, 360 Huntington Ave, Boston,
MA 02115, USA
dRevolution Medicines, 700 Saginaw Drive, Redwood City, CA 94063, USA
ARTIC LE INFO
ABSTRACT
Previously we reported the results from an effort to improve Gram-negative
Article history:
Received
antibacterial activity in the oxazolidinone class of antibiotics via a systematic
medicinal chemistry campaign focused entirely on C-ring modifications. In that series
we set about testing if the efflux and permeation barriers intrinsic to the outer
membrane of Escherichia coli could be rationally overcome by designing analogs to
reside in specific property limits associated with Gram-negative activity: i) low MW
(< 400), ii) high polarity (clogD7.4 < 1), and iii) zwitterionic character at pH 7.4.
Indeed, we observed that only analogs residing within these limits were able to
overcome these barriers. Herein we report the results from a parallel effort where we
explored structural changes throughout all three rings in the scaffold for the same
purpose. Compounds were tested against a diagnostic MIC panel of Escherichia coli
and Staphylococcus aureus strains to determine the impact of combining structural
modifications in overcoming the OM barriers and in bridging the potency gap between
the species. The results demonstrated that distributing the charge-carrying moieties
across two rings was also beneficial for avoidance of the outer membrane barriers.
Importantly, analysis of the structure-permeation relationship (SPR) obtained from
this and the prior study indicated that in addition to MW, polarity, and zwitterionic
character, having ≤ 4 rotatable bonds is also associated with evasion of the OM
barriers. These combined results provide the medicinal chemist with a framework and
strategy for overcoming the OM barriers in GNB in antibacterial drug discovery
efforts.
Revised
Accepted
Available online
Keywords:
Gram-negative
Outer membrane permeability
Oxazolidinones
Efflux pumps
Porins
2009 Elsevier Ltd. All rights reserved.
———
∗ Corresponding author. Tel.: +0-000-000-0000; fax: +0-000-000-0000; e-mail: author@university.edu

In spite of the widespread and increasing prevalence of
multidrug-resistant Gram-negative and Gram-positive bacteria
(GNB and GPB), the number of new drugs being developed to
treat bacterial infections is at its lowest point since the dawn of
the antibiotic era.1,2 Hence, there is an urgent need for new
discovery platforms and approaches to develop new antibacterial
drugs, and to stop the dangerous trends of increasing multidrug-
resistance. The conventional target based high throughput assays
have been demonstrated earlier as successful approaches to
identify hits with micro- and nano-molar scales of activities in
cell free assays, however poor membrane permeability and high
susceptibility to active efflux have consistently frustrated efforts
to evolve these “hits” into “leads” with promising whole cell
activity.3,4 The outer membrane of Gram-negative bacteria
restricts penetration of hydrophobic compounds, and efflux
pumps extrude any hydrophobic molecules that leak in through
that barrier.5,6 The net result is that hydrophobic compounds with
potential antibacterial activity, as suggested by potent inhibition
of their intended targets, are not active against these bacteria due
to an inability to achieve sufficient concentrations at their sites of
action. Extensive reviews of physicochemical properties of
compounds that have shown good activity against GNB provided
a range of structural features that do not necessarily fall into the
Lipinski’s rule of five.7 ,8
The oxazolidinone class of antibacterials covers the
important Gram-positive pathogens including methicillin-
resistant Staphylococcus aureus (MRSA), vancomycin-resistant
Enterococci (VRE), and penicillin-resistant Streptococcus
pneumoniae.9 Clinical resistance has emerged, albeit slowly, to
this class of drugs, however structural modifications able to
evade the mechanisms responsible have been reported.10
Linezolid was the first oxazolidinone antibiotic to be approved
by the US Food and Drug Administration (FDA) and has been
used for the treatment of serious infections caused by Gram-
positive strains such as MRSA and VRE since 2000 (Figure 2).
Favorable clinical results shown by linezolid prompted many
pharmaceutical industries and academic institutions to explore
the possibilities of expansion of antibacterial spectrum of this
class leading to a broad SAR information available covering this
scaffold (Figure 1).11,12,13,14 Many hundreds of oxazolidinone
derivatives have been reported over the years since linezolid was
discovered, but these compounds do not have structures that
adhere to the guidelines we believe necessary for GNB activity.
From our perspective, this explains why no derivative has yet
been identified having a fully broadened spectrum of activity. In
fact, linezolid is known to have much improved potency against
E. coli and other common GNB when the efflux system is
compromised.15
Figure 1. Summary of known SAR in the oxazolidinone class of GPB antibacterials.
We recently reported the results from a focused
medicinal chemistry campaign aimed at improving whole cell
GNB activity in the oxazolidinone class of antibacterials.16 Our
approach was to design C-ring analogs with physicochemical and
structural properties we believed to be in the correct range for
GNB activity: i) low MW (preferably <400); ii) relatively high
polarity (clogD7.4 < 1); and iii) charged character with bias to
zwitterionic character at physiological pH.16 From this effort we
found that zwitterionic character in particular, when imparted
onto
the C-ring, resulted in significantly reducing the impact of efflux
and the permeation barriers on activity against E. coli. This is
best demonstrated by the lead compound, DP-326, identified in
that effort (Figure 2). Unfortunately, the modifications that led to
minimization of the impact of the OM barriers were accompanied
by a reduction in protein synthesis inhibitory potency compared
to linezolid, as determined with a cell-free transcription-
translation assay. This reduced biochemical potency appears
reflected in the higher than desired antibacterial MIC values for
DP-326.

Figure 2. Physicochemical properties and structures of Linezolid and DP-326.
In the present work we continued to leverage the known
SAR in the oxazolidinone class of GPB antibacterials (Figure 1)
but expand our medicinal chemistry efforts to explore other
regions on the oxazolidinone scaffold beyond the C-ring.
Specifically, we investigated A- and B-ring regions that may be
tolerant of modification, or where the available data suggest
certain modifications have led to potency enhancements. Our
goal was to identify analogs with distributed polar groups and
multi-charge character that demonstrated reduced liability
OM barriers, but that avoided the undesired activity loss seen in
our C-ring series. Each compound prepared was tested against a
diagnostic MIC panel of wild-type and outer membrane
compromised E. coli strains and a Staphylococcus aureus control
strain to determine the impact of structural modifications on OM
permeability and efflux liability, and to monitor our progress
towards bridging the potency gap between the species caused by
these barriers. We also analyzed our current overall SPR data
gathered from our earlier and current studies in an effort to
address the question of whether emphasizing appropriate
towards
the
physicochemical properties represents
a useful medicinal
chemistry approach for optimizing Gram-negative antibacterial
activity in the oxazolidinones class of antibacterials.
In our previous study16 we reported a rational and
systematic medicinal chemistry approach, led by guidelines
derived from data gathered on earlier GNB potent antibiotics,
toward adding Gram-negative antibacterial activity in the
currently narrow spectrum class of oxazolidinone antibiotics. We
found that C-ring modified oxazolidinone analogs that carry
physicochemical properties correlating with GNB-activity, i.e.
(i) low MW (preferably <400), (ii) relatively high polarity
(clogD7.4 ≤ 1), and (iii) charged character with bias to
zwitterionic character at physiological pH, led to moderate
improvements in Gram-negative activity associated with
overcoming the efflux and permeability barriers.
Moving forward, in this study we investigated other
directions in the structural design while leveraging reported SAR.
We began by investigating which regions in the scaffold beyond
the C-ring tolerated modifications that increased polarity and
were expected to carry significant charge character at
physiological pH. Our novel oxazolidinone analogs were
designed to incorporate polar and/or charged groups to generate a
focused library of analogs with optimal physicochemical
properties for penetration based on the leading guidance (MW <
400, clogD7.4 ≤ 1, ionic character at physiologic pH), Figure 3.
Once polar and charged groups are identified that were beneficial
at different scaffold positions, these will be combined in hybrid
analogs to further increase polarity and charge character. In light
of our results in the C-ring series, we anticipated that combining
multiple beneficial changes together in hybrid compounds will
result in additive, or multiplicative, improvements in potency
against
GNB.

Figure 3: Rational design of GNB penetrating oxazolidinone derivatives.
A general synthetic approach using the palladium
catalyzed Suzuki bi-aryl coupling reaction was utilized for the
synthesis of the designed oxazolidinone analogs as shown in
Figure 4. These reactions were carried out using the synthesized
intermediates in Scheme 1 in either the halide or boronic ester
form together with the intended C-ring that is substituted with
boronic acid in case of using halide substituted B-ring
intermediates, or halide in case of boronic ester substituted B-
ring intermediates, to form the C-ring/ B-ring bi-aryl system of
the planned analogs. C-ring moieties were either used as
purchased from commercial sources or further synthetically
customized either before or after the Suzuki coupling reaction to
give the desired functionalities. (See Experimental section).
Figure 4: Synthesis of A- and/or B- and/or C-Rings modified oxazolidinone
analogs designed for GNB penetration.
The assembly of the chiral oxazolidinone ring system,
with various A-rings and C-rings while B-ring intact is shown in
Scheme 1. The general intermediates with 4-halo or 4-boronic
ester and 3-fluoro substitutions on B-ring, i.e 3-11, were
synthesized in multi-step reactions from commercially available
anilines using classical methods as described in Scheme 1 and
detailed in the experimental section.
Scheme 1: Synthetic route for the general intermediates used in the synthesis of A- and C-ring modified oxazolidinone analogs. a
aReagents and reaction conditions: i) CBz-Cl, K2CO3 or NaHCO3, THF, 0 oC to RT, 1-3 hrs. ii) n-BuLi or LHMDS, -78 oC to RT, 12 hrs. iii) NIS, TFA, RT, 2
hrs. iv) Bis(pinacolato)diboron, potassium acetate, Pd(dppf)2Cl2, DMSO, 80 oC, 12 hrs. v) MsCl, TEA, DCM, 0-5 oC, 30 min. vi) Potassium phthalimide, DMF,
70 °C, 4 hrs. vii) Hydrazine monohydrate, ethanol, reflux, 2 hrs. viii) Desired acetic anhydride, pyridine, 50 oC, 2 hrs. ix) Bis(pinacolato)diboron, potassium
acetate, Pd(dppf)2Cl2, DMSO, 80 oC, 12 hrs. x) NaN3, DMSO, 90 oC, 5 hrs. xi) CuSO4, sodium ascorbate, t-BuOH, water, 50 °C, 16 hrs. xii) Cp*RuCl(PPh3)2,
1,4-dioxne, 50 °C, 12 hrs. xiii) BBr3, DCM, 0 oC- RT, 12 hrs.

For the assembly of the chiral oxazolidinone ring
system, with various B-ring and C-ring systems while the A-ring
intact, we used the general synthetic approach that is shown in
Scheme 2. The general intermediates with 4-halo or 4-boronic
ester substitutions on various B-ring systems, i.e 13a-m, were
synthesized in multi-step reactions from commercially available
anilines using classical methods as described in Scheme 2 and
detailed in the experimental section. Some compounds needed
future customized routes such as DP-337 (Scheme 3).
Scheme 2: Synthetic route for the general intermediates used in the synthesis of B-ring modified oxazolidinone analogs. a
aReagents and reaction conditions: i) CBz-Cl, K2CO3 or NaHCO3, THF, 0 oC to RT, 1-3 hrs. ii) n-BuLi or LHMDS, -78 oC to RT, 12 hrs. iii) BBr3, DCM, 0 oC-
o
RT, 12 hrs. iv) TBAF, THF, 0 C, 3 hrs. v) Bis(pinacolato)diboron, potassium acetate, Pd(dppf)2Cl2, DMSO, 80 oC, 12 hrs.

Scheme 3: Synthesis of DP-337.a
aReagents and reaction conditions: i) (4-(((tert-butoxycarbonyl)amino)methyl)phenyl)boronic acid, potassium carbonate, Pd(dppf)2Cl2, Dioxane–ethanol-water,
90 °C. ii) CuSO4, sodium ascorbate, t-BuOH, water, 50 °C, 16 hrs. iii) BBr3, DCM, 0 oC-RT, 12 hrs.
Penetration of drug into the periplasm is affected by
outer membrane permeation and efflux to varying degrees, so we
felt it critical to determine the SAR for these variables separately.
We assessed the impact of efflux on each compound by
comparing its MIC for an efflux-competent E. coli strain to its
MIC for an efflux-deficient mutant of that strain (∆acrAB mutant
possessing an intact OM). The ratio of these two MICs is
reported in Tables 1-4 as the “efflux ratio.” Similarly, the effect
of the OM permeation barrier on antibacterial activity of each
compound (permeability ratio, "Perm. ratio") was assessed by
comparing its MIC for the parent strain of E. coli to its MIC for
that strain in the presence of the OM-permeabilizing agent
polymyxin-b nonapeptide (PMBN). We aimed to test compounds
up to 256 µg/ml in hopes of detecting MIC values for all test
strains, but
water solubility limits with some compounds, as evident by
obvious cloudiness or precipitation forming following dilution
from DMSO into media for the MIC testing, prevented testing to
that concentration (ie. DP-57, -297; Table 2).
For this series we initially explored analogs having range of
B-ring modifications to determine what degree of change is
tolerated, Table 1. In contrast to prior findings,17,18 our analogs
without a C-ring i.e., DP-25, DP-26, DP-30, DP-33, DP-340 and
DP-341, were not tolerated as demonstrated by a complete loss in
antibacterial activity (i.e. S. aureus MIC). Pyridyl and pyridyl-N-
oxide B-rings were also not well tolerated, i.e. DP-293 and DP-
263. Des-fluoro phenyl B-rings were generally well tolerated, i.e.
DP-296, DP-318, DP-329, and DP-332. Replacing the fluorine
on the B-ring with hydroxyl or methoxy was not tolerated, i.e.
DP-327, and DP-328.
Table 1: MIC data and calculated physicochemical properties of B-ring modified oxazolidinone analogs (using ChemAxon).
MIC (µg/ml)a
Ratios
Physicochemical properties
E. coli
MG1655
(WT)d
S. aureus
ATCC
Rotatable
Bonds
(#)
Cmpd
Structure
E. coli
E. coli
Perm.
Ratiob
Efflux
Ratioc
clogD
% ionized
(pH7.4)
MW
cpKa
+PMBNe
∆acrABf
(pH7.4)
29213g
Linezolid (40)
130
36
11
3.1
3.6
12
337.3
>14
0.64
0
4
DP-25
DP-26
DP-30
DP-33
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>256
>64
>64
>64
>64
ND
ND
ND
ND
ND
ND
ND
ND
255.2
227.2
239.2
227.2
3.34
8.42
9.57
8.18
-2.72
0.69
0.43
0.66
99.99
8.80
3
2
3
2
0.27
14.26
DP-340
>256
>256
>256
>64
ND
ND
278.3
9.79
-1.38
99.60
5

DP-341
DP-293
DP-296
>128
>256
>64
>128
>256
>64
>128
>256
8
>64
64
4
ND
ND
ND
ND
ND
>8
295.3
320.3
287.3
5.23
7.88
14.6
0.45
0.21
2.68
66.90
25.03
4
3
0.00
3
1.79
3
DP-318
>256
32
16
1
>8
>16
303.3
9.14
2.37
99.14
4
DP-327
>256
>256
>256
>256
>256
64
>256
>256
32
>64
>64
4
ND
ND
>4
ND
ND
>8
328.4
314.3
321.3
9.46
9.57
8.06
-0.51
-0.50
2.43
DP-328
99.17
4
19. 17
3
DP-329
50.23
4
DP-332
DP-333
DP-263
>256
>64
32
8
2
>8
ND
ND
>32
ND
ND
327.3
299.3
311.3
5.01
14.6
7.87
1.45
2.38
1.34
>64
>256
>64
>256
64
0.00
4
3
>256
>64
25.36
aMICs reported are mean values from triplicate runs. bEfflux ratio = MIC vs. E. coli MG1655 / MIC vs. E. coli ∆acrAB. cPermeability ratio = MIC vs. E. coli
MG1655 / MIC vs. E. coli MG1655 + PMBN. dE. coli MG1655 (WT), wild-type Gram-negative bacterium E. coli. eE. coli +PMBN, a membrane permeabilized
mutant. fE. coli ∆acrAB, an efflux pump deficient strain. gS. aureus ATCC 29213, Gram-positive bacterium S. aureus. ND not detected.
to impart significant anionic character at this position at pH 7.4.
N-Acetyl-aminomethyl substituent of A5 position,
We combined these modifications in analogs with moderately
found in linezolid and radezolid, and small heterocycles often
acidic or basic groups on the C-ring to determine whether
provides a 4 to 8-fold potency enhancement over the smaller
hydroxymethyl group.19 However such modifications (Table 2)
placement of charged groups at both ends of our compounds is as
beneficial as clustering them on one end, based on our prior
did not enhance the antibacterial activity in our series compared
report. Unfortunately, in this small set of analogs we did not
to our prior findings with the parent hydroxymethyl analogs. i.e.
observe a clear benefit towards minimizing the impact of the OM
barriers when dual or zwitterionic character was introduced in a
distributed fashion, i.e. DP-55, DP-57, DP-297, DP-298, DP-
320, DP-322 and DP-338. .
DP-51, DP-52, DP-53, DP-54, and DP-381.
We then explored the impact of increasing the acidity of
the A5 N-acetyl amide by the addition of 2 and 3 adjacent
fluorines

Table 2. MIC data and calculated physicochemical properties of 5-methyl acetamido substituted A-ring oxazolidinone analogs (using ChemAxon).
MICa
Ratios
Physicochemical properties
(µg/ml)
Cmpd
DP-51
Structure
E. coli
MG1655
(WT)d
S. aureus
ATCC
Rotatable
Bonds
(#)
E. coli
∆acrABf
E. coli
+PMBNe
Perm.
Ratioc
Efflux
Ratiob
clogD
(pH7.4)
% ionized
(pH7.4)
MW
cpKa
29213g
128
32
8
2
2
4
16
344.3
6.2
0.91
1.81
5.90
4
DP-52
>256
32
32
>8
>8
369.3
7.83
8.17
27.03
14.51
4
DP-53
DP-54
>256
128
16
64
8
8
2
4
>16
>32
16
362.3
373.4
2.16
4
5
99.76
(18.13
zwitterion)
8.05;
10.3
2
-0.39
99.32
(22.06
zwitterion)
7.93;
9.47
128
32
16
4
8
2
4
4
32
393.4
398.3
0.50
1.12
5
4
DP-55
99.26
(7.13
zwitterion)
>32
(solubility)
5.17;
6.33
≥4
≥8
DP-57
64
32
8
8
4
2
1
2
8
357.4
398.3
9.45
-0.49
2.48
99.12
5
4
DP-294
DP-297
34.33
(2.91
dianion)
>32
(solubility)
7.72;
8.43
>4
>8
99.68
(98.33
zwitterion)
5.32;
9.46
DP-298
(2)
64
11
16
2
2
6
32
>256
>2
411.4
440.3
416.3
1.77
1.52
2.38
5
5
4
99.53
(40.80
zwitterion)
5.30;
7.25
>256
>256
<2
256
<0.5
32
>16
>2
DP-320
DP-322
99.33
(12.63
dianion)
5.29;
8.24
256
5.23;
6.07;
8.03
99.42
(18.20
dianion)
>256
>256
>256
>256
128
>64
ND
ND
ND
>2
456.3
370.3
1.08
0.72
5
4
DP-338
DP-381
>256
32
7.21
60.96
b
aMICs reported are mean values from triplicate runs. Efflux ratio = MIC vs. E. coli MG1655 / MIC vs. E. coli ∆acrAB. cPermeability ratio = MIC vs. E. coli
MG1655 / MIC vs. E. coli MG1655 + PMBN. dE. coli MG1655 (WT), wild-type Gram-negative bacterium E. coli. eE. coli +PMBN, a membrane permeabilized
mutant. fE. coli ∆acrAB, an efflux pump deficient strain. gS. aureus ATCC 29213, Gram-positive bacterium S. aureus. ND not detected.

However, the addition of fluorine atoms to acidify the
A5 N-H proton has the negative impact of increasing overall
hydrophobicity (compare DP-294 to -55 and -298; Table 2),
which we believe would tend to increase liability towards efflux.
these A5 heterocyclic variants lost antibacterial activity against
the GPB, and showed no clear indication of improved relative
GNB activity. Four we were particularly interested in making,
two hydroxytriazole variants DP-471 and DP-472, and two other
heterocycles calculated to have significant anionic character at
pH 7.4 (DP-473 and DP-477) were essentially inactive.
Thus we investigated
a small set of functionalized A5
heterocyclic substituents, including a few variations expected to
carry significant charge character at pH 7.4 (Table 3). In general,
Table 3: MIC data and calculated physicochemical properties of oxazolidinone analogs with various heterocycle substituted A-ring (using ChemAxon).
MICa
(µg/ml)
Ratios
Physicochemical properties
Cmpd
Structure
E. coli
MG1655
(WT) d
S. aureus
ATCC
29213g
%
ionized
(pH7.4)
Rotatable
Bonds
(#)
E. coli
∆acrABf
E. coli
+PMBNe
Perm.
Ratioc
Efflux
Ratiob
clogD
(pH7.4)
MW
cpKa
256
128
64
32
16
16
2
16
397.4
9.45
9.45
-0.35
99.12
99.12
6
DP-334
DP-337
>64
>64
ND
>4
411.4
1.12
7
99.65
(99.12
zwitterio
n)
3.02;
9.45
>256
>256
>256
128
>256
128
>64
ND
>2
ND
>2
411.4
411.4
0.12
1.12
6
6
DP-339
DP-373
99.89
(86.70
dication)
8.22;
9.45
8
>256
>64
256
>64
>64
128
32
32
>1
ND
ND
>2
>2
411.4
383.4
383.4
9.45
0.62
2.27
1.98
99.12
7
5
5
DP-374
99.82
(78.62
zwitterio
n)
6.82;
9.45
8
DP-471
DP-472
99.13
(98.65
zwitterio
n)
5.08;
9.45
>64
>64
>64
ND
99.12
(zwitteri
on)
3.33;
9.45
>128
>256
>256
>128
128
64
>128
16
ND
ND
>4
>4
411.4
379.3
379.3
-0.34
2.41
2.70
6
4
4
DP-475
DP-468
DP-469
64
8
>2
8.17
8.17
14.64
14.64
64
32
>4
99.20
(8.14
zwitterio
8.41;
9.49
>64
>64
>64
256
32
16
ND
ND
ND
383.4
385.4
0.15
0.18
5
5
DP-473
DP-476
n)
>256
>256
≥2
9.48
99.12

99.60
(53.65
zwitterio
n)
7.32;
9.48
>256
>256
>256
>64
ND
ND
398.4
-0.91
5
DP-477
>256
>256
>256
>64
ND
ND
397.4
9.45
-0.51
99.12
6
DP-478
b
aMICs reported are mean values from triplicate runs. Efflux ratio = MIC vs. E. coli MG1655 / MIC vs. E. coli ∆acrAB. cPermeability ratio = MIC vs. E. coli
MG1655 / MIC vs. E. coli MG1655 + PMBN. dE. coli MG1655 (WT), wild-type Gram-negative bacterium E. coli. eE. coli +PMBN, a membrane permeabilized
mutant. fE. coli ∆acrAB, an efflux pump deficient strain. gS. aureus ATCC 29213, Gram-positive bacterium S. aureus. ND not detected.
Other A-ring modifications that incorporated methyl
amines, carboxylic acid the reverse amide which are more
compact in size and has increased polarity were not well
tolerated, i.e. DP-46, DP-47, and DP-60 (Table 4). However,
despite its unimpressive activity, we did note that the zwitterionic
analog DP-58 was not impacted by efflux or the permeation
barrier. Of additional interest we noted that DP-58 demonstrated
merely a four-fold difference in potency between the species, in
contrast to the >40-fold difference seen with linezolid (Table 1).
Unfortunately, we did not detect MICs vs. E. coli strains using
the close analog DP-59. Additional medicinal chemistry
optimization on Ring-C such as DP-565, DP-479 and DP-295
did not provide potency enhancement.
Table 4: MIC data and calculated physicochemical properties of oxazolidinone analogs with various substituents on A-ring or C-ring (using ChemAxon).
MICa
(µg/ml)
Ratios
Physicochemical properties
Cmpd
Structure
E. coli
S. aureus
ATCC
29213g
%
ionized
(pH7.4)
Rotatable
Bonds
(#)
E. coli
+PMBNe
E. coli
∆acrABf
Perm.
Ratioc
Efflux
Ratiob
clogD
(pH7.4)
MG1655
(WT) d
MW
cpKa
>256
>256
>256
>64
>64
ND
ND
ND
329.3
9.45
-0.69
99.12
4
4
DP-46
99.12
(zwitteri
on)
3.09;
9.45
>256
>256
>256
ND
330.3
-0.14
DP-47
97.50
(14.09
zwitteri
on)
8.10;
8.99
256
256
256
64
64
1
1
320.3
327.3
0.94
0.82
3
3
DP-58
DP-59
97.87
(26.23
zwitteri
on)
7.80;
8.95
>256
>256
>256
ND
ND
>256
>256
>256
>256
>256
>64
ND
ND
295.3
346.3
357.3
8.92
4.69
-0.98
0.77
97.07
3
4
DP-60
O
F
H2N
N
O
N
DP-565
16
16
1
>16
>16
0.31
OH
HO
N
64
8
1
>32
ND
8
8.31
1.5
2.84
0.97
10.89
0.00
3
5
DP-479
DP-295
>128
>128
>128
32
ND
b
aMICs reported are mean values from triplicate runs. Efflux ratio = MIC vs. E. coli MG1655 / MIC vs. E. coli ∆acrAB. cPermeability ratio = MIC vs. E. coli
MG1655 / MIC vs. E. coli MG1655 + PMBN. dE. coli MG1655 (WT), wild-type Gram-negative bacterium E. coli. eE. coli +PMBN, a membrane permeabilized
mutant. fE. coli ∆acrAB, an efflux pump deficient strain. gS. aureus ATCC 29213, Gram-positive bacterium S. aureus. ND not detected.
activity is a retrospective analysis by a group from AstraZeneca’s
The most comprehensive report currently available
examining the impact of physicochemical properties on GNB
antiinfectives division. From this analysis they found that high
polarity (clogD <0) was associated with GNB activity, low serum

MIC shifts, and a reduced potential for non-specific cytotoxicity,
although there were notable exceptions.20 They partitioned their
antibacterial set based on high or low efflux susceptibility in both
P. aeruginosa and E. coli, and observed that the high efflux bins
for both species demonstrated normal bell-curve distributions
centered on the average clogD value, while the low efflux bins
for both species were bi-modal with the second mode at the more
polar end of their range (clogD <0). Similarly, molecular weight
gave a normal bell-curve distribution centered on the average
value in the high efflux bins, while in the low efflux bins there
was a fairly even spread with no clear trend. However, the
examples included in this retrospective work appear to be
dominated by β-lactams, agents acting in the periplasm, so it is
difficult for us to draw many conclusions from this.
The effort we are reporting on here was an attempt to
apply our recently improved understanding of the properties
associated with GNB activity in the forward direction. In hopes
of confirming that this is a reliable route to achieving useful
GNB
activity we have followed an analog design strategy aimed at
producing compounds that fall within the physicochemical
property space linked to this activity: (i) low MW (preferably
<400), (ii) relatively high polarity (clogD ≤1) and (iii) charged
character with bias to zwitterionic character at physiological pH.
Many of the compounds we prepared and tested do not satisfy all
these criteria, which allowed us to challenge these putative
correlations.
In an effort to address the question of whether
emphasizing appropriate physicochemical properties represents a
useful medicinal chemistry approach for optimizing Gram-
negative antibacterial activity in the oxazolidinones class of
antibacterials, we analyzed our current overall structure activity
relationship and structure property relationship data gathered
from our earlier and current studies. We relied on the efflux
ratios and the permeability ratios as indicators for improvement
against E. coli, because these values normalized the differences
in potency. These ratios for each of our compounds were plotted
against a variety of physicochemical properties in hopes of
identifying correlations (Figures 5-7). Plotting of a variety of
single parameters, including clogD7.4, MW, # rotatable bonds,
and polar surface area, against the efflux ratios or the
permeability ratios from both studies combined failed to show
clear correlations between these properties and OM barrier
susceptibility. This is perhaps not surprising given the complex
and multivariate nature of OM barriers and translocation. In
addition, in our test set the range of PSA values was narrow (88
to 109Å2), as was the MW range (287-418), so this may also
have contributed to a lack of correlation (not shown). To
exemplify the results, we included the plot of clogD vs. the
permeability and efflux ratios (Figure 5). One key observation
from this plot is that efflux tends to be the more significant
barrier in this series than permeation. While there is no obvious
trend line in the broad scatter, it seemed apparent that those
compounds in our set with the lowest efflux and permeability
ratios had values of clogD < 1.5.
While our focus for these early studies were on whole
cell activity, it was important to confirm that inhibitory activity
against protein synthesis was retained. Our previous work details
an in vitro transcription/translation (TT) assay, which confirmed
that our compound activity was consistent with improved entry
into the cell.16 The TT assay also showed that the potency loss
from the WT E. coli to WT S. aureus was biochemically driven
rather than by changes in ability to cross the cytoplasmic
membrane. This further confirmed that the S. aureus control was
a sufficient surrogate for protein synthesis inhibition.16 For
several key compounds in this program for which MICs against
the whole panel were obtained, the ratios of MIC values from E.
coli and S. aureus were essentially equivalent to the multiple of
the efflux and permeability ratios, further lending support that the
potency differences between these species was predominantly
driven by the OM entry barriers. While other compounds were
not tested in the TT assay at this stage, it would be informative to
include this assay broadly in the next stage of the program when
focusing on increasing biochemical potency while maintaining
structural features that facilitate entry into GNB.
3
2.5
2
1.5
1
0.5
0
0
10
20
30
40
50
60
70
-0.5
-1
Efflux Ratio
-1.5
-2
Ratio
.

Figure 5. Plot of calculated logD values at pH 7.4 (clogD; ChemAxon calculator) vs. Efflux and Permeability ratios for our oxazolidinone analogs.
Our view that charge character was important to entry
and efflux evasion prompted us to include many ionizable groups
in our analogs. Based on the concerns around traversing both the
outer and the cytoplasmic membranes to reach the antibacterial
target of this series, we biased our selection towards substituents
that have some neutral character at pH 7.4, such as moderately
basic and acidic groups, rather than biasing towards strongly
basic
(ie. guanidine) and acidic (ie. aryl carboxylic acids) groups.
While we viewed this as important primarily for passing through
the cytoplasmic membrane, it was interesting to see that a plot of
calculated pKa values verses the efflux and permeability ratios
(Figure 6) shows that those compounds with the lowest ratios
clustered in the pKa range of 6.2 to 10.2.
16
14
12
10
8
Efflux Ratio
6
4
2
0
0
10
20
30
40
50
60
70
Ratio
Figure 6. Plot of calculated pKa values (ChemAxon calculator) vs. Efflux and Permeability ratios for our oxazolidinone analogs.
Finally, in an attempt to determine if a particular charge state
was associated with best improvements against the GNB outer
membrane barriers, we plotted the fully ionized charge state vs.
the efflux and permeability ratios in that portion of our set with
ionic character at pH 7.4 (Fig. 7). The three ionic states included
in this plot were -1 (single anions), +1 (single cations), and net 0
(zwitterions). There were multiple examples in each category that
demonstrated low ratios of 2-4 as well as higher ratios. However,
only zwitterions were represented in those few examples having
both efflux and permeability ratios of 1-2.
70
60
50
40
30
20
10
0
Permiability
Ratio
-2.00
-1.00
0.00
1.00
2.00
Charge: acidic (-1), basic (1), zwitterionic (0)
Figure 7. Plot of fully ionized charge state vs. Efflux and Permeability ratios for our oxazolidinone analogs.
Our results from these analyses are consistent with prior
reports [20], and indicate that no single compound parameter yet
identified strongly correlates with evasion or susceptibility to the
OM efflux and permeation barriers in common Gram-negative
bacteria. However, in our data we do see a strong correlation to
the composite of properties that include MW, polarity, charge
character, and the # of rotatable bonds. As summarized in Table
5, all five compounds found to be minimally impacted by the OM
barriers fell within these criteria: i) MW < 400; ii) clogD ≤ 1; iii)

ionized at pH 7.4 with a bias towards zwitterionic character; iv) #
rotatable bonds ≤ 4. Indeed, four of these five are calculated to
have significant zwitterionic character at pH 7.4, hence our
thoughts that this is a key characteristic are supported. Only one
compound in the full set of analogs, DP-19, met all these criteria
but remained significantly impacted by the OM barriers. Several
close analogs were present in the set that fell just outside of these
criteria, such as DP-54 and DP-55 (Table 5), and in all cases they
remained significantly impacted by the OM barriers, most
notably efflux.
Table 5. MIC data and calculated physicochemical properties of oxazolidinone analogs with key analogs from this and our prior report [16] (using ChemAxon).
Ratios
Physicochemical properties
Cmpd
Structure
Rotatable
Bonds
(#)
Perm.
Ratioa
Efflux
Ratiob
clogD
(pH7.4)
% ionized
(pH7.4)
MW
cpKa
DP-28116
1
2
1
1
350.3
7.73
-2.21
1.14
31.83
4
4
28.18
(3.8%
zwitterionic)
DP-32116
361.3
350.3
7.96; 6.10
7.86;10.18
99.69
(25.5%
DP-32516
2
1
0.17
4
zwittionic)
98.95 (56.5%
zwitterionic)
DP-32616
1.6
1.6
333.3
320.3
332.3
7.30; 9.37
8.10; 8.99
8.05; 10.34
-0.30
0.94
4
3
4
97.50
(14.09
zwitterion)
1
1
DP-58
99.76
(18.1%
DP-1916
2.6
10.2
-0.11
zwitterionic)
99.76
(18.1
zwitterion)
2
16
373.4
393.4
8.05; 10.3
-0.39
5
DP-54
99.32
(22.1%
4
32
7.93; 9.47
0.50
5
DP-55
zwitterion)
aEfflux ratio = MIC vs. E. coli MG1655 / MIC vs. E. coli ∆acrAB. bPermeability ratio = MIC vs. E. coli MG1655 / MIC vs. E. coli MG1655 + PMBN.
strategy for rationally overcoming the intrinsic OM barriers of
The results reported herein include observations from
our continued effort to test whether conforming the
physicochemical properties of the oxazolidinones class of
antibiotics, an important class of FDA approved antibiotics that
are currently active only against GPB, to those deemed vital for
GNB activity is a viable strategy for evading the OM barriers in
the common Gram-negative bacterium E. coli. In this work we
explored the effectiveness of simultaneously combining
structural modifications in the oxazolidinones A-, B- and C-ring
regions to complement the structural variations reported
previously.16 Analysis of the structural properties relationship
(SPR) data from our entire focused library of oxazolidinones
including those reported here and in the prior study provided
further support to our hypothesis that OM barriers in GNB are
surmountable when the appropriate physicochemical properties
are incorporated in the structural design of the oxazolidinone
analogs.16 Importantly, we observed a strong correlation between
surmounting the OM barriers and adherence to a set of specific
property criteria. Indeed, from our data we have derived a set of
rough but simple guidelines to arm medicinal chemists with a
efflux and permeation. These guidelines are, that in order to
evade the OM barriers in common Gram-negative bacteria, one
should bias compound design towards analogs having: i) a MW
<400; ii) a clogD7.4 ≤ 1; iii) charged or preferably zwitterionic
character at pH 7.4; and iv) # rotatable bonds ≤ 4.
Our effort has been focused on oxazolidinones, but our
approach was scaffold-agnostic, therefore we anticipate that the
results are likely to carry over into other antibacterial classes.
While the GNB MIC values achieved in this work were not
impressive, we view this demonstration in overcoming the OM
barriers as a significant advance. With the guidelines we propose
in hand, we believe it will be possible to further enhance
antibacterial potency while maintaining OM barrier evasion.
Structural rigidification is a classic medicinal chemistry approach
to enhancing potency by freezing out bioactive conformations,
and this approach, if applied to the oxazolidinone antibacterial
leads we have identified, may enable researchers to finally
identify compounds in this class with potent activity against
GNB.

Activity Against the Fastidious Gram-Negative Organisms
Haemophilus influenzae and Moraxella catarrhalis. J. Med.
Chem. 43, 953-970.
Acknowledgements
We thank Dr. Vincent H. Tam from the University of Houston
for the gift of MG1655 and MG1655 ∆acrAB used in the
experiments. As well as the financial support of the Defense
Threat Reduction Agency (#HDTRA1-14-1-0019), the American
Cancer Society (RSG-12-161-01-DMC), and the aid of the
Northeastern University Office the Provost Dissertation Grant.
15. Schumacher, A.; Trittler, R.; Bohnert, J.A.; Kummerer, K.; Pages,
J.-M.; Kern, W.V. (2007) Intracellular accumulation of linezolid
in Escherichia coli, Citrobacter freundii and Enterobacter
aerogenes: role of enhanced efflux pump activity and inactivation.
J. Antimicrob. Chemother. 59, 1261-1264.
16. Takrouri, K.; Cooper, H. D.; Spaulding, A.; Zucchi, P.; Koleva,
B.; Cleary, D.C.; Tear, W.; Beuning, P. J.; Hirsch, E.B.; Aggen, J.
B. (2016) Progress against Escherichia coli with the
oxazolidinone class of antibacterials: test case for a general
approach to improving whole cell Gram-negative activity. ACS
Infect. Dis. (DOI: 10.3410/f.726380671.793518957).
References
1. Michael S. Kinch et al., “An Analysis of FDA-Approved Drugs
17. Gregory, W.A.; Brittelli, D.R.; Wang, C.-L.J.; Wuonola, M.A.;
McRipley, R.J.; Eustice, D.C.; Eberly, V.S.; Bartholomew, P.T.;
Slee, A.M.; Forbes, M. (1989). Antibacterials. Synthesis and
Structure-Activity Studies of 3-Aryl-2-oxooxazolidines. 1. The
"B" Group. J. Med. Chem. 32, 1673-81.
for Infectious Disease: Antibacterial Agents,” Drug Discovery
Today
19,
no.
9
(2014):
1283–87,
doi:10.1016/j.drudis.2014.07.005.
2. The Pew Charitable Trusts, Infectious Diseases Society of
America, Pharmaceutical Research and Manufacturers of
America, “Reviving the Pipeline of Life-Saving Antibiotics:
Exploring Solutions to Spur Innovation,” conference proceedings,
18. Park, C.-H.; Brittelli, D.R.; Wang, C.-L.J.; Marsh, F.D.; Gregory,
W.A.; Wuonola, M.A.; McRipley, R.J.; Eberly, V.S.; Slee, A.W.;
Forbes, M. (1992). Antibacterials. Synthesis and Structure-
Activity Studies of 3-Aryl-toxooxazolidines. 4. Multiply-
Substituted Aryl Derivatives. J. Med. Chem. 35, 1156-65.
Sept.
22,
2011,
Washington,
http://www.idsociety.org/uploadedFiles/IDSA/Policy_and_Advoc
acy/Current_Topics_and_Issues/Advanci
g_Product_Research_and_
Development/Bad_Bugs_No_Drugs/Press_Releases/Conference%
20Proceedings_FINAL_4.11.12.pdf.
19. Renslo, A. R.; Luehr, G. W.; Gordeev, M. F. (2006). Recent
developments in the identification of novel oxazolidinone
antibacterial agents. Bioorg. Med. Chem. 14, 4227–4240.
3. Tommasi, R., Brown, D.G., Walkup, G.K., Manchester, J.I.,
Miller, A.A. (2015) ESKAPEing the labyrinth of antibacterial
20. Brown, D. G.; May-Dracka, T. L.; Gagnon, M. M.; Tommasi, R.
(2014). Trends and exceptions of physical properties on
antibacterial activity for Gram-positive and Gram-negative
pathogens. J. Med. Chem. 57, 10144-10161.
discovery. Nat. Revs. Drug Discov. [online] July
3 2015,
doi:10.1038/nrd4572.
4. Payne, D. J.; Gwynn, M. N.; Holmes, D. J.; Pompliano, D. L.
(2007). Drugs for bad bugs: confronting the challenges of
antibacterial drug discovery. Nature Rev. Drug Discov. 6, 29-40.
5. Nikaido, H. (2003) Molecular basis of bacterial outer membrane
permeability revisited. Microbiol. Mol. Biol. Rev. 67(4), 593-656.
6. Zgurskaya, H.I.; Lopez, C.A.; Gnanakaran, S. (2015).
Permeability Barrier of Gram-Negative Cell Envelopes and
Approaches To Bypass It. ACS Infect. Dis. 1(11), 512-522.
7. O’Shea, R., Moser, H.E. (2008) Physicochemical properties of
antibacterial compounds: implications for drug discovery. J. Med.
Chem. 51(10), 2871-2878.
8. Brown, D. G.; May-Dracka, T. L.; Gagnon, M. M.; Tommasi, R.
(2014). Trends and exceptions of physical properties on
antibacterial activity for Gram-positive and Gram-negative
pathogens. J. Med. Chem. 57, 10144-10161.
9. Barbachyne, M.R.; Ford, C.W. (2003) Ozaxolidinone structure-
activity relationships leading to linezolid. Angew. Chem. Int. Ed.
42, 2010-2023.
.
10. Locke, J.B.; Finn, J.; Hilgers, M.; Morales, G.; Rahawi, S.; Kedar,
G.C.; Picazo, J.J.; Im, W.; Shaw, K.J.; Stein, J.L. (2010).
Structure-Activity Relationships of Diverse Oxazolidinones for
Linezolid-Resistant Staphylococcus aureus Strains Possessing the
cfr Methyltransferase Gene or Ribosomal MutationsЀ
׋ՐլկՏկՍըխԯЀ ׋ֈЀ ២יթՏխլיթըЀ ՜Ւ ᝺РПԺϿ
՜ՑՑֽ֤ՒՑЀ ֺ
11. Pandit, N.; Singla, R.K.; Shrivastava, B. (2012) Current updates
on oxazolidinone and its significance. Int. J. Med. Chem.
(DOI:10.1155/2012/159285).
12. Michalska, K.; Karpiuk, I.; Krol, M.; Tyski, S. (2013) Recent
development of potent analogues of oxazolidinone antibacterial
agents. Bioorg. Med. Chem. 21, 577-591.
13. Gordeev, M.F.; Yuan, Z.Y. (2014) New Potent Antibacterial
Oxazolidinone (MRX-I) with an Improved Class Safety Profile. J.
Med. Chem. 57, 4487-4497.
14. Genin, M.J.; Allwine, D.A.; Anderson, D.J.; et al. (2000)
Substituent Effects on the Antibacterial Activity of Nitrogen-
Carbon-Linked (Azolylphenyl)oxazolidinones with Expanded