Heterodimeric Rifampicin–Tobramycin conjugates break intrinsic resistance of Pseudomonas aeruginosa to doxycycline and chloramphenicol in vitro and in a Galleria mellonella in vivo model

Article abstract of DOI:10.1016/j.ejmech.2019.04.034

Intrinsic resistance in Pseudomonas aeruginosa, defined by chromosomally encoded low outer membrane permeability and constitutively over-expressed efflux pumps, is a major reason why the pathogen is refractory to many antibiotics. Herein, we report that h

Full text of DOI:10.1016/j.ejmech.2019.04.034

Accepted Manuscript
Heterodimeric Rifampicin–Tobramycin conjugates break intrinsic resistance of
Pseudomonas aeruginosa to doxycycline and chloramphenicol in vitro and in a
Galleria mellonella in vivo model
Temilolu Idowu, Gilbert Arthur, George G. Zhanel, Frank Schweizer
PII:
S0223-5234(19)30349-6
DOI:
https://doi.org/10.1016/j.ejmech.2019.04.034
EJMECH 11269
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 15 January 2019
Revised Date: 1 April 2019
Accepted Date: 13 April 2019
Please cite this article as: T. Idowu, G. Arthur, G.G. Zhanel, F. Schweizer, Heterodimeric Rifampicin–
Tobramycin conjugates break intrinsic resistance of Pseudomonas aeruginosa to doxycycline and
chloramphenicol in vitro and in a Galleria mellonella in vivo model, European Journal of Medicinal
Chemistry (2019), doi: https://doi.org/10.1016/j.ejmech.2019.04.034.
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.

ACCEPTED MANUSCRIPT
Graphical Abstract

ACCEPTED MANUSCRIPT
Heterodimeric Rifampicin–Tobramycin conjugates break intrinsic resistance of
Pseudomonas aeruginosa to Doxycycline and Chloramphenicol in Vitro and in a
Galleria mellonella in Vivo model.
a
b
c
,a,c
Temilolu Idowu, Gilbert Arthur, George G. Zhanel, and Frank Schweizer*
a
Department of Chemistry, University of Manitoba, Winnipeg, MB R3T 2N2, Canada
b
Department of Biochemistry and Medical Genetics, University of Manitoba, Winnipeg, MB
R3E 0J9, Canada
c
Department of Medical Microbiology/Infectious Diseases, University of Manitoba, Winnipeg,
MB R3T 1R9, Canada
*
Corresponding Author
F.S.: e-mail, Frank.Schweizer@umanitoba.ca; phone, (+1) 204-474-7012; fax, (+1) 204-474-
608
7
Abbreviations used
CAM, chloramphenicol; CFU, Colony-forming Unit; CLSI, Clinical and Laboratory Standard
Institute; DOX, doxycycline; FICI, Fractional Inhibitory Concentration Index; MDR, multidrug-
resistant; OM, outer membrane; RND, Resistance Nodulation Division; XDR, extensively drug-
resistant.
1

ACCEPTED MANUSCRIPT
Abstract
Intrinsic resistance in Pseudomonas aeruginosa, defined by chromosomally encoded low outer
membrane permeability and constitutively over-expressed efflux pumps, is a major reason why
the pathogen is refractory to many antibiotics. Herein, we report that heterodimeric rifampicin-
tobramycin conjugates break this intrinsic resistance and sensitize multidrug and extensively
drug-resistant P. aeruginosa to doxycycline and chloramphenicol in vitro and in vivo.
Tetracyclines and chloramphenicol are model compounds for bacteriostatic effects, but when
combined with rifampicin-tobramycin adjuvants, their effects became bactericidal at sub MIC
levels. Potentiation of tetracyclines correlates with the SAR of this class of drugs and is
consistent with outer membrane permeabilization and efflux pump inhibition. Overall, this
strategy finds new uses for old drugs and presents an avenue to expand the therapeutic utility
of legacy antibiotics to recalcitrant pathogens such as P. aeruginosa.
Keywords:
adjuvant,
amphiphilic
aminoglycoside,
antibacterial,
antimicrobial,
antipseudomonal, Galleria mellonella, hybrid, rifampicin, synergy.
2

ACCEPTED MANUSCRIPT
1.
Introduction
The problem of antimicrobial resistance is a global phenomenon that is threatening to
reverse the gains and advances in modern medicine.[1–4] Microorganisms are developing
resistance to our antibiotic pipeline at a rate that is practically impossible to keep up with and
there are some phenotypes that are completely resistant to all available treatment options.[5]
Regrettably, no new class of antibiotics with a novel mode of action has been successfully
developed in the last few decades.[6,7] It is not surprising that for the first time ever, the WHO
in 2017 published a list of pathogens for which effective treatment options are becoming
elusive.[8] With a projection that antimicrobial resistance will be a leading cause of death
worldwide by 2050,[9] it is expedient to, in the meantime, develop strategies that can expand
and/or preserve the therapeutic usefulness of our current armamentarium.
Gram-negative bacteria are more difficult to inhibit or kill because of their outer
membrane (OM) that serves as an additional layer of protection.[10] The OM is an asymmetric
bilayer that limits or prevents the uptake of potentially noxious compounds to the bacterial cell,
especially molecules that are hydrophobic and/or have high molecular weight.[11] Thus, most
antibiotics that are active against Gram-positive bacteria are inactive against Gram-negative
bacteria because of their inability to traverse the OM. Furthermore, the physicochemical
requirements for a molecule to traverse the OM and inner membrane of Gram-negative
bacteria are orthogonal to each another.[12] To cross the OM, a molecule needs to be small
and polar (for porin-mediated uptake) or polybasic (for a self-promoted uptake), whereas it
needs to be hydrophobic to traverse the inner membrane.[11] Medicinal chemists often
struggle with developing agents that can tightrope this dual-membrane topology, accumulate in
3

ACCEPTED MANUSCRIPT
the cell, and still be able to bind to a distinct target in the cytosol.[12,13] The frustrations
expressed by the WHO is therefore apparent and it is clear that the priority one pathogens,
which are all carbapenem-resistant (CR) Gram-negative ESKAPE[14] pathogens (i.e. CR
Acinetobacter baumannii, CR Pseudomonas aeruginosa, and various CR Enterobacteriaceae),
are major threats to public health. Of these pathogens, P. aeruginosa is even more difficult to
inhibit or kill because of its sophisticated and high level of intrinsic resistance, such as its
extremely low OM permeability and constitutively over-expressed efflux pumps of broad
substrate specificities that actively extrude drug molecules from its periplasm/cytoplasm.[15]
These, coupled with acquired and adaptive resistance mechanisms,[15] make P. aeruginosa
resistant to many antibiotics that are active against other Gram-negative bacteria. As an
opportunistic pathogen, P. aeruginosa is the leading cause of nosocomial infections such as
respiratory infections in debilitated patients and patients with cystic fibrosis, and it is well
known for its ability to evade antibiotic actions.[11,15]
Current thinking in the development of clinically relevant antipseudomonal agents
include, but not limited to, developing agents that could potentially weaken or destabilize the
OM, such that molecules that are otherwise unable to traverse this barrier can now gain
access into the periplasm and/or cytoplasm.[16–19] Even for drugs with porin-mediated uptake
mechanisms, OM destabilization ensures that the rate of drug influx overwhelms the rate of
efflux thereby enhancing antibiotic accumulation. For instance, SPR741, a polymyxin-based
adjuvant is being developed by Spero Therapeutics as an OM permeabilizer that can facilitate
the entry of OM-impermeable antibiotics into Gram-negative bacteria such as A. baumannii
and Enterobactericeae.[20] Unfortunately, SPR741 is not effective against P. aeruginosa.[20]
On the other hand, tobramycin-based heterodimeric scaffolds have been reported to potentiate
4

ACCEPTED MANUSCRIPT
several legacy antibiotics against P. aeruginosa than in other Gram-negative bacteria.[11,21]
These two scenarios present intriguing dynamics about the physicochemical requirements for
activity and antibiotic potentiation in P. aeruginosa. Hybridization of a second bioactive
molecule to tobramycin via a tether resulted in conjugates that: i) preserve the original mode of
action of the bioactive molecule, ii) abolish the ribosomal effect of tobramycin by itself, and iii)
confer adjuvant properties on the resulting conjugate.[11,21] Tobramycin, an aminoglycoside
antibiotic that interferes with the fidelity of ribosomal protein translation, propagates its own
uptake into P. aeruginosa by displacing the stabilizing divalent cations that cross-bridge
adjacent lipopolysaccharides on the outer leaflet of the OM.[22] This process is known as the
‘
self-promoted uptake mechanism’ and the cationic nature of aminoglycosides is critical to their
ability to induce this process.
To put all of these into perspective, we sought to expand the therapeutic usefulness of
some OM-impermeable antibiotics by leveraging the self-promoted uptake mechanism of
tobramycin to deliver them into Gram-negative bacteria, in a “Trojan-horse” fashion. For
w
example, rifampicin, a large (M = 822.9 g/mol) hydrophobic molecule that binds to RNA
polymerase and inhibit RNA synthesis, is active against Gram-positive bacteria and
mycobacteria but inactive against most Gram-negative bacteria because of its inability to cross
the OM. Unlike most classes of antibiotics, rifamycins are unique in that they are active against
pathogens in slow growing, stationary, and non-replicating metabolic states.[23] Unfortunately,
the greatest limitations of this class of drug is the rapid development of a single mutation in the
β-subunit of bacterial RNA polymerase (rpoB),[24] and the lack of activity against Gram-
negative pathogens. Surprisingly, some tobramycin-based conjugates that strongly potentiate
the activity of rifampicin against P. aeruginosa were also shown to prevent the development of
5

ACCEPTED MANUSCRIPT
resistance after 25 serial passages at sub MIC levels.[25] Hence, by directly conjugating
rifampicin to a tobramycin scaffold using various tether lengths, we wanted to investigate the
possibility of: 1) using tobramycin as a vector to shuttle rifampicin into P. aeruginosa, 2)
preserving the original mode of action of rifampicin, i.e. RNA polymerase inhibition, in P.
aeruginosa, 3) modulating the overall physicochemical property of the resulting conjugate such
that it preserves or amplifies the known adjuvant properties of tobramycin-derived conjugates.
The different tether lengths between the rifampicin and tobramycin domains were meant to
investigate the optimal spatial separation between these two drug moieties. Herein, we report
the synthesis and evaluation of a series of covalently-attached rifampicin–tobramycin
molecules 1 – 3 (Figure 1). When hybridized as a single non-cleavable entity, the potency of
rifampicin was preserved in some Gram-positive bacteria but was attenuated in P. aeruginosa
(Table 1). However, the resulting conjugates 1 – 3 break the chromosomally-encoded intrinsic
resistance of P. aeruginosa to doxycycline and chloramphenicol and significantly potentiate
their activities against wild-type, multidrug- (MDR) and extensively drug-resistant (XDR)
phenotypes in vitro and in vivo. An isolate is MDR if it is non-susceptible to at least one agent
in at least three antimicrobial categories, while an XDR isolate is non-susceptible to at least
one agent in all but two or fewer antimicrobial categories.[26] Typically, P. aeruginosa is
clinically-resistant to doxycycline and chloramphenicol, but in the presence of < 10 µM of
compounds 1 – 3, the Clinical Laboratory Standard Institute (CLSI) susceptibility breakpoints
(extrapolated: doxycycline ≤ 4 µg/ml for Acinetobacter spp, chloramphenicol ≤ 8 µg/ml for
Enterobacteriaceae) were reached in 9 out of 10 phenotypes for doxycycline, and 7 out of 10
for chloramphenicol. This strategy expands the chemotherapeutic utility of doxycycline and
6

ACCEPTED MANUSCRIPT
chloramphenicol to P. aeruginosa and makes the pathogen susceptible to and treatable by
these agents.
7

ACCEPTED MANUSCRIPT
2.
Results
2.1
Design and Synthesis
To design a covalently hybridized conjugate of two bioactive molecules, the point of
attachment on both molecules must be carefully identified in order not to interfere with their
biological activities. Tobramycin was conceptualized as a vector to deliver rifampicin into the
periplasm, due to its self-promoted uptake mechanism, hence, its cationicity must be
preserved. Rifampicin was intended to mediate its antibiotic activity in the cell, hence, the point
of attachment must not be directly involved with its target. Therefore, the design of rifampicin-
tobramycin conjugates was established from the known structure-activity relationships of the
parent molecules. Amphiphilic tobramycins with lipophilic groups at the C-5 position have been
shown to retain its self-promoted uptake mechanism.[27,28] Solved crystal structure of RNA
polymerase complexed with rifampicin revealed that the terminal piperazine ring of rifampicin is
not involved in the binding of the drug to its target.[29] The drugs were therefore linked at
these identified positions under a chemically benign reaction condition using aliphatic
hydrocarbon of different lengths, i.e. C-4, C-6, and C-12. Rifampicin is hydrolyzed in both
acidic and basic conditions, due to its imine and ester groups, respectively, thus, a copper(1)-
catalyzed azide-alkyne cycloaddition reaction (“Click Chemistry”)[30] was employed to join
both compounds together under neutral conditions. This afforded regioselective 1,4-
disubstituted 1,2,3-triazole products 1 – 3 (Figure 1). To achieve this, an azide moiety was
incorporated on the tether length of an amphiphilic tobramycin, while an alkyne group was
installed on the piperazine ring of rifampicin. The tether length was meant to investigate the
8

ACCEPTED MANUSCRIPT
optimal spatial separation between the two domains. The synthetic strategy for preparing
compounds 1 – 3 is outlined in Schemes 1, 2, and 3.
2.2
Chemical Synthesis of Rifampicin–Tobramycin Conjugates 1 – 3.
To install a propargyl group on rifampicin, a 1-amino-4-propagylpiperazine 7 precursor
was synthesized (Scheme 1) and coupled to commercially available 3-formyl rifamycin SV
Scheme 3) following established procedure.[31] Briefly, 7 was prepared by converting one of
(
the secondary amines in piperazine 4 to a nitroso- group under acidic conditions to give
compound 5, followed by a nucleophilic conjugation of the second secondary amine to
propargyl bromide under basic conditions to give 1-nitroso-4-propargylpiperazine 6. The
nitroso group was then reduced to a primary amine in the presence of a strong reducing agent,
LiAlH , to afford precursor 7 (Scheme 1). The amphiphilic tobramycin domain was prepared
4
following previously reported protocol.[27] Tobramycin 8 was purchased from a commercial
source and the amino groups were first protected using di-tert-butyl dicarbonate (Boc
anhydride), followed by silylation of the N-Boc-tobramycin intermediate with excess TBDMSCl
to afford a partially protected derivative 9 with free OH at the C-5 position of the cyclitol ring.
Alkylation of 9 in toluene with 1,n-dibromoalkane (n = 4, 6, 12) in the presence of a phase-
transfer catalyst (TBAHS) afforded bromoalkylated TBDMS-Boc-protected tobramycin
intermediates. The terminal bromo-groups of these intermediates were then displaced by an
azido nucleophile under anhydrous condition to give compounds 10a-c. The protecting groups
were finally removed in a stepwise manner, first by removing TBDMS groups using TBAF, and
then Boc-protecting groups using TFA, to afford compounds 11a-c (Scheme 2). Rifampicin-
tobramycin conjugates 1 – 3 were ligated by coupling 7 to 3-formyl rifamycin SV to give 12,
9

ACCEPTED MANUSCRIPT
followed by a “click chemistry” conjugation to 11a-c under neutral conditions to afford the final
compounds (Scheme 3). The final compounds were retained on C18 reverse-phase silica in a
column and washed with copious amount of deionized water to remove residual copper ions.
2.3
Antimicrobial Susceptibility Screening
The antibacterial activities of the newly synthesized conjugates 1 – 3 and reference
compounds, rifampicin and tobramycin, were assessed against a panel of Gram-positive and
Gram-negative bacteria following the CLSI guidelines. These results are presented as the
minimum inhibitory concentrations (MICs) in table 1. As expected, rifampicin by itself was
potent against several Gram-positive bacteria (MIC = <0.25 – 1 µg/ml) but not against Gram-
negative bacteria (MIC ≥ 8 µg/ml) while tobramycin exhibits potency against most Gram-
positive and Gram-negative bacteria. By design, the rifampicin-tobramycin conjugates 1 – 3
are expected to lose the protein translation inhibitory effect of tobramycin but retain the RNA
polymerase inhibitory properties of rifampicin, as previously reported.[23] Tobramycin
conjugated in this fashion are generally non-ribosomal,[21,25,28] hence, it was designed to
only shuttle rifampicin into Gram-negative bacteria. Table 1 shows that compounds 1 – 3
exhibit slightly potent activity against some Gram-positive bacteria (MIC of 1 – 8 µg/ml) but
were generally not potent against Gram-negative bacteria (MIC > 32 µg/ml). This suggests that
whereas the conjugates might have retained the RNA polymerase effect of rifampicin against
Gram-positive bacteria, they are unable to mediate the same effect in Gram-negative bacteria.
The antipseudomonal activities of these conjugates were further assessed against a panel of
MDR/XDR phenotypes and the results were consistent with the poor activity against Gram-
negative bacteria (Table S1). Against Gram-positive bacteria, compound 1 with the shortest
10

ACCEPTED MANUSCRIPT
tether length (C-4 aliphatic hydrocarbon chain) appears to be more potent than compound 3
with the longest tether length (C12 chain).
2.4
Combination Studies of Rifampicin–Tobramycin conjugates with different classes
of antibiotics against wild-type P. aeruginosa
To investigate the impact of hybridizing a highly hydrophobic rifampicin moiety to a
richly hydrophilic and polybasic tobramycin domain, the adjuvant properties of the resulting
compounds were evaluated against P. aeruginosa. Physicochemical modulation of tobramycin-
based conjugates is believed to be crucial to the nature, type, and degree of potentiation of
different antibiotics.[11] Thus, we assessed the interactions between compounds 1 – 3 and
twenty-one different antibiotics (representing all major classes) against wild-type P. aeruginosa
PAO1 using checkerboard assay (Table S2). Data from this study were interpreted as a
function of the fractional inhibitory concentration index (FICI), a numerical quantification of the
interactions between antimicrobial agents. FICI of < 0.5, 0.5 – 4, and > 4 indicate synergy,
additive or no interaction, and antagonism, respectively.[11,21] Surprisingly, rifampicin-
tobramycin conjugates 1 – 3 were able to potentiate the effects of several antibiotics against
WT PAO1, including rifampicin, but not tobramycin or β-lactams (Figure 2a, Table S2). Unlike
in susceptibility screening against Gram-positive bacteria, conjugates 2 and 3 with longer
tether lengths were better potentiators than compound 1 with four carbon chain length (Figure
2b). Compounds 2 and 3, at ≤ 10 µM, potentiated rifampicin (FICI = 0.15 – 0.28), tetracyclines
(FICI = 0.06 – 0.50), chloramphenicol (FICI = 0.06 – 0.09), fosfomycin (FICI = 0.15 – 0.28),
linezolid (FICI = 0.15 – 0.28), novobiocin (FICI = 0.28), erythromycin (FICI = 0.06 – 0.15), etc.
but not tobramycin (FICI = 1.03), β-lactams (FICI = 1.03), moxifloxacin (FICI = 0.53 – 1.03),
11

ACCEPTED MANUSCRIPT
and nitrofurantoin (FICI = 0.53) (Figure 2a, Table S2). Doxycycline (FICI = 0.06 – 0.09) and
chloramphenicol (FICI = 0.06 – 0.09) were exceptionally potentiated by compounds 2 and 3
against PAO1, hence, these combinations were further investigated against highly resistant
phenotypes of P. aeruginosa. It should be noted that doxycycline and chloramphenicol are
typically not used to treat P. aeruginosa infections because of poor susceptibility due to
intrinsic resistance.
2.5
Rifampicin-Tobramycin conjugates strongly potentiate doxycycline and
chloramphenicol against multidrug and extensively drug resistant P. aeruginosa
A combination of doxycycline or chloramphenicol with compounds 2 or 3 were further
assessed against a panel of highly drug-resistant P. aeruginosa phenotypes. Most of these
isolates are neither susceptible to doxycycline nor chloramphenicol, and are mostly resistant to
carbapenems, the drugs of last resort (Table S3). Remarkably, compounds 2 and 3
significantly potentiated the effects of doxycycline and chloramphenicol against these clinical
isolates (Table 2). For instance, in the presence of ≤ 10 µM of compounds 2 and 3,
susceptibility of P. aeruginosa clinical isolates to doxycycline was increased by 4- to 512-fold
while susceptibility to chloramphenicol was increased by 4- to 256-fold (Figure S1). These
effects were generally dose-dependent against WT and clinical isolates (Figure 3), suggesting
an involvement of a common mechanism in both phenotypes. On the contrary, the degree of
potentiation of doxycycline and chloramphenicol were not as pronounced (nil to 8-fold) in other
Gram-negative bacteria (Figure S2), suggesting a strong antipseudomonal potentiating effect
for 2 and 3.
12

ACCEPTED MANUSCRIPT
To put these findings into context, extrapolated CLSI susceptibility breakpoints were
used as interpretive standards for doxycycline and chloramphenicol since they are not
conventional antipseudomonal agents. Breakpoints are discriminatory antimicrobial
concentrations used in the interpretation of susceptibility testing to define isolates as
susceptible, intermediately-resistant, or resistant. If the MIC of an antibiotic against an
organism is less than or equal to its susceptibility breakpoint, the bacterial strain is considered
to be susceptible to the antibiotic. The CLSI clinical breakpoint of doxycycline for Acinetobacter
spp. is 4 µg/ml while that of chloramphenicol for Enterobacteriaceae is 8 µg/ml.[32]
Susceptibility equal or below these breakpoints were reached for doxycycline in wild-type and
eight out of nine clinical isolates, and for chloramphenicol in wild-type and six out of nine
clinical isolates (Table 2). CLSI clinical breakpoints could not be reached for both antibiotics in
other resistant Gram-negative pathogens such as Acinetobacter baumannii, Klebsiella
pneumoniae and Enterobacter cloacae in the presence of 10 µM of compounds 2 and 3 (Table
S4). It is noteworthy that tobramycin alone, rifampicin alone, or a combination of both do not
potentiate doxycycline or chloramphenicol against any of the P. aeruginosa phenotype used in
this study (Table S5).
2
.6
dependent
A major contributor to the intrinsic resistance of P. aeruginosa to doxycycline and
Potentiation of doxycycline and chloramphenicol in P. aeruginosa is efflux pump-
chloramphenicol is their highly efficient efflux mechanisms, as does the low permeability of the
OM. To accumulate in cells, compounds must traverse the OM (or porins) faster than they are
pumped out. The MexAB/MexXY efflux systems in P. aeruginosa contribute significantly to the
13

ACCEPTED MANUSCRIPT
extrusion of tetracyclines out of the cell while resistance to chloramphenicol is in part due to
the MexAB-OprM efflux system.[33] To investigate the role of these pumps in the ability of
compound 2 to increase the susceptibility of P. aeruginosa to doxycycline and chloramphenicol
in such magnitude, we assessed the synergistic relationships of these combinations in two
mutant strains, PAO200 and PAO750. PAO200 is a mexA-mexB-oprM deletion strain while
PAO750 lacks five clinically important RND pumps (MexAB–OprM, MexCD–OprJ, MexEF–
OprN, Mex JK, and MexXY) and the OM protein OpmH.[34,35] As expected, the efflux-mutant
strains were hypersusceptible to both doxycycline and chloramphenicol (Table 3), consistent
with known contributions of RND efflux pumps to the intrinsic resistance to these agents.[36]
However, at a fixed concentration of compound 2, its ability to potentiate doxycycline and
chloramphenicol was diminished in efflux-deficient mutants compared to the wild-type strain
(Table 3). For instance, compound 2 (at 9.8 µM) increased the susceptibility of WT PAO1 to
doxycycline by 32-fold while an increase in susceptibility of 8- and 4-fold was observed in
PAO200 and PAO750, respectively. The same phenomenon was observed for the potentiation
of chloramphenicol (Table 3 and Figure S3). Even between the two mutants, the test
antibiotics were more potentiated in PAO200 which is deficient in only one of its pumps,
compared to PAO750 which lacks five different pumps. This indicates that interaction with
and/or inhibition of RND efflux pumps play an important role in the degree of potentiation seen
in PAO1 but, expectedly, not in efflux-deficient mutants. However, inhibition of efflux pumps
does not explain the partial increase in susceptibilities of the efflux mutants, suggesting that
other mechanisms such as OM permeabilization might concomitantly be at play. OM
permeabilization is indeed consistent with the ability of these conjugates to potentiate some
OM-impermeable antibiotics such as rifampicin, erythromycin, etc. Overall, interaction with
14

ACCEPTED MANUSCRIPT
RND efflux pumps seems to play a role in the ability of compound 2 to potentiate doxycycline
and chloramphenicol, and strains with higher levels of expression will likely be more
susceptible to these combinations.
2.7
Rifampicin–Tobramycin conjugate potentiates rifampicin and other members of
tetracycline against MDR/XDR P. aeruginosa
To ascertain whether the ability of rifampicin-tobramycin conjugates to potentiate
doxycycline is conserved across the entire class of tetracyclines, synergistic relationships
between compound 2 and minocycline or tigecycline were evaluated against wild-type and
MDR/XDR P. aeruginosa. Results of this study show that minocycline and tigecycline were
similarly potentiated as doxycycline but to a lesser degree (Figure 4, Table S6). The degree of
potentiation of tetracyclines is: doxycycline > minocycline > tigecycline (Figure 4), an
observation that seems to correlate with the structure-activity relationships of tetracyclines.
This variation is perhaps due to the fact that tigecycline evades acquired efflux and target-
mediated resistance to classical tetracyclines[37] but remains vulnerable to the
chromosomally-encoded multidrug efflux pumps of P. aeruginosa.[38] Thus, inhibition of
acquired efflux pumps by compound 2 will have lesser consequential effects on tigecycline
than it will on doxycycline. Rifampicin, an OM-impermeable antibiotic, was also potentiated
against wild-type and MDR/XDR P. aeruginosa isolates (Figure 4, Table S6), suggesting a role
for OM permeabilization as part of the mechanism of potentiation by compound 2. Rifampicin
is not a substrate for RND efflux pumps in P. aeruginosa (Table S1).
2.8
Time-kill Assay
15

ACCEPTED MANUSCRIPT
Time-kill curves can be used to monitor bacterial growth and death over a wide range of
antimicrobial concentrations to evaluate the effect of antimicrobials over time. An antibiotic
could either be bactericidal or bacteriostatic under specific growth conditions. Bactericidal
activity is defined as a ≥ 3-log reduction in the total CFU/mL from the original inoculum over 24
hours while bacteriostatic activity is defined as maintenance of < 3-log reduction in the total
CFU/mL from the original inoculum.[32] Thus, the time-kill kinetics of wild-type P. aeruginosa
PAO1 when incubated with doxycycline or chloramphenicol alone, or in combination with
compound 2, was assessed. Preliminary growth curve was performed to ensure that PAO1
6
could be grown from a starting inoculum of about 10 CFU/ml under our assay conditions, and
that pre-incubation in antimicrobial-free LB medium will reach a stable early- to mid-log phase
after 4 h. The time-kill curves of doxycycline (at 32 µg/ml, i.e. 2× MIC) and chloramphenicol (at
32 µg/ml, i.e. 1× MIC) looked similar, with almost no killing of bacteria within the assay period
(Figure 5). Bacterial growth was completely inhibited by both drugs at these concentrations, an
effect that is consistent with known bacteriostatic properties of tetracyclines (doxycycline) and
chloramphenicol.[39] When PAO1 was incubated with sub MICs of 4 µg/ml doxycycline alone
or 8 µg/ml chloramphenicol alone, i.e. their respective extrapolated CLSI breakpoints, there
was bacterial growth as early as 3 h, similar to the control without antibiotics (Figure 5).
However, in the presence of 4.9 µM or 9.8 µM of compound 2, doxycycline became strongly
bactericidal and chloramphenicol became weakly bactericidal at their respective CLSI
breakpoints. Indeed, a combination of 4 µg/ml of doxycycline and 9.8 µM of compound 2
reduced the viability of P. aeruginosa PAO1 below the limits of detection (< 10 CFU/mL) after 9
h of incubation (Figure 5). Synergistic effect is defined as ≥ 2-log decrease in the number of
CFU/mL between the combination and the most active component of the combination after 24
16

ACCEPTED MANUSCRIPT
h (at least one of the drugs must be present at a concentration that does not affect the growth
curve of the test organism).[40] It is clear that compound 2 exhibited synergistic relationships
with both doxycycline and chloramphenicol against P. aeruginosa PAO1 in a dose-dependent
manner and there was no re-growth after 24 h of incubation.
2.9
In Vivo Efficacy Studies using Galleria mellonella Infection Model
In vitro synergy of many antibiotic/antibiotic or antibiotic/adjuvant combinations against
Gram-negative bacteria have been shown to be often incongruent with in vivo synergy.[41,42]
To investigate whether the in vitro synergy observed with compound 2 and doxycycline or
chloramphenicol is translated in vivo, we examined the ability of different combination
concentrations to offer therapeutic protection against MDR P. aeruginosa-infected Galleria
mellonella wax moths. The capability of this infection model to determine virulence of P.
aeruginosa strains as well as efficacy and pharmacokinetics of antipseudomonal agents have
been widely demonstrated.[11,21,25,42,43] The maximum tolerable dose was first determined
by injecting high concentrations of doxycycline, chloramphenicol, and compound 2 (100 mg/kg
and 200 mg/kg each) into the larvae and survivability was scored for 96 h (4 days). 100 %
survival was recorded in groups injected with 100 mg/kg doxycycline, 100 and 200 mg/kg
chloramphenicol, and 100 mg/kg of compound 2, while groups injected with 200 mg/kg each of
doxycycline and compound 2 recorded 85 % and 97 % survival, respectively, after 4 days
(Figure 6a). This indicates that the compounds were relatively non-toxic to the larvae by
themselves, an effect that is consistent with other in vitro toxicity studies (vide infra). Also, an
inoculum size of 5 CFU of PA260, an extensively drug-resistant P. aeruginosa clinical isolate,
achieved 100% lethality in the larvae after 18 h. Next, the ability of the drugs, alone and in
17

ACCEPTED MANUSCRIPT
combination, to protect the larvae from XDR PA260 3 h post infection was determined using
single doses of 100 mg/kg each as monotherapy, or 75 + 75 mg/kg or 100 + 100 mg/kg as
combination therapy. PA260 is expected to have reached a stable early- to mid-log phase in
the larvae after 3 h of infection. Like the untreated control, 100 % mortality was observed in the
monotherapy treatments with all drugs (Figure 6b). However, a single dose combination of
doxycycline and compound 2 using 75 + 75 mg/kg and 100 + 100 mg/kg resulted in 75 % and
8
7 % survival, respectively after 18 h, and 55 % and 75 % survival, respectively after 24 h
Figure 6b). Similarly, a single dose combination of chloramphenicol and compound 2 using 75
75 mg/kg and 100 + 100 mg/kg resulted in 55 % and 72 % survival, respectively after 18 h,
(
+
and 10 % and 45 %, respectively after 24 h (Figure 6b). This demonstrates the ability of
compound 2 to synergize with doxycycline and chloramphenicol in vivo and offer therapeutic
protection to PA260-infected larvae in a dose-dependent manner at tolerable concentrations.
The higher level of survival seen with doxycycline + compound 2 relative to chloramphenicol +
compound 2 is perhaps attributable to the susceptibility of PA260 to these agents. The
absolute MIC (in the presence of compound 2) of doxycycline against PA260 is 0.125 µg/ml
while that of chloramphenicol is 2 µg/ml (Table 2). Colistin, which is the only antibiotic that
PA260 is susceptible to besides amikacin (Table S3), served as the positive control for this
experiment while treatment with PBS only served as negative control.
2.10 Toxicity Studies
To evaluate in vitro toxicity and margin of therapeutic safety of compounds 1 – 3 against
eukaryotic cells, they were screened against human liver (HepG2) and kidney (HEK293) cells,
as well as porcine erythrocytes. The results of these studies were compared to the approved
18

ACCEPTED MANUSCRIPT
parent antibiotics, rifampicin and tobramycin. For cytotoxicity against human cells, doxorubicin,
a very potent anticancer drug, was used as a positive control for this experiment. The results of
the study showed that none of the rifampicin-tobramycin conjugates 1 – 3, rifampicin, or
tobramycin was toxic to HepG2 and HEK293 (Figure 7a). At the effective adjuvant
concentration of < 10 µM (< 16 µg/mL) used for this study, cell viability was greater than 90 %
for both cells, and greater than 70 % at the highest concentration tested (50 µM, i.e. > 100
µg/mL). Expectedly, doxorubicin reduced the viability of HepG2 and HEK293 cells to less than
10% at about 12 µM (6.5 µg/mL), consistent with its cytotoxic properties (Figure 7a). We also
investigated the toxicity of the combination therapies against HepG2 and HEK293 cells (i.e.
compound 2 + doxycycline and compound 2 + chloramphenicol) and found that the
combination regimen did not elevate toxicological profiles of the primary antibiotics (Figure 7b).
For toxicity against freshly collected porcine erythrocytes, 0.1% Triton X-100 served as the
positive control and was used to calculate percent hemolysis. All the newly synthesized
conjugates 1 – 3 exhibited insignificant hemolytic effects (< 2%) at very high concentrations of
1024 µg/mL (Figure 7c), a 64-fold higher dose than the maximum synergistic concentration
used in the study.
19

ACCEPTED MANUSCRIPT
3.
Discussion
The sustained and emerging resistance of recalcitrant pathogens to our antibiotic
armamentarium constitutes an acute threat to public health and a cause for alarm. The
growing gap between clinical needs and drug innovation is a direct consequence of the
difficulty in identifying and bringing new drugs to market, especially against Gram-negative
pathogenic bacteria. Intrinsic resistance in Gram-negative bacteria, such as their protective
outer membrane and constitutively overexpressed efflux pumps, is a major survival weapon
that renders them untreatable by most antibiotics.[11] To fill the current void in antibacterial
drug discovery against Gram-negative bacteria, a plausible strategy that has gained wide
attention is the development of adjuvants that could either rescue/preserve the efficacy of
available treatment options,[17] or expand the therapeutic usefulness of antibiotics that are not
clinically indicated for such infections.[11,20,25] For example, tazobactam, avibactam, and
varborbactam are adjuvants that have all been approved by FDA to rescue/preserve the
efficacies of ceftolozane, ceftazidime, and meropenem, respectively, while membrane
permeabilizing agents (e.g. SPR741) have been shown to be capable of expanding the
therapeutic usefulness of OM-impermeable drugs (e.g. rifampicin) to Gram-negative
bacteria.[11,25,44]
To cross the OM, a molecule must either pass through the porin (usually very small and
polar, e.g. fluoroquinolones, β-lactams),[45] have an active transport mechanism, or induce a
self-promoted uptake mechanism (could be big, e.g. colistin, aminoglycosides). Cationicity is
critical for a self-promoted uptake mechanism across the OM and hydrophobicity facilitates
penetration of membrane-active compounds across bacterial membrane. Consequently,
20

ACCEPTED MANUSCRIPT
amphiphilicity have often been employed to achieve the dual physicochemical requirement of
electrostatic interactions and transmembrane navigation in cells.[13,46,47] Rifampicin–
Tobramycin conjugates 1 – 3 were designed such that rifampicin could be shuttled into the
periplasm of Gram-negative bacteria, especially P. aeruginosa, using tobramycin as a vector
due its self-promoted uptake mechanism. The hydrophobic rifampicin domain may also
facilitate uptake across the inner membrane. Based on earlier studies, tobramycin was
expected to lose its ability to inhibit protein translation,[21,47] but rifampicin was expected to
retain its activity.[29] A different group had previously used similar ‘Trojan-horse’ approach to
deliver OM-impermeable erythromycin as an effective treatment of Klebsiella pneumonia in
vivo.[31] The resulting conjugates 1 – 3 retained activity as standalone agents against some
Gram-positive bacteria but not against Gram-negatives (Table 1). This observation is similar to
that of a dual-acting rifampicin-ciprofloxacin hybrid, TNP-2092, a former drug candidate that
was only active against Gram-positive bacteria.[23] Since conjugated tobramycins modified at
the C-5 position are completely inactive against Gram-positive and Gram-negative bacteria,[25]
the activity of compounds 1 – 3 against some Gram-positive pathogens is most likely driven by
the rifampicin domain, suggesting that ligation to its piperazine moiety do not significantly
affect its biological activities. Previous SAR study of tobramycin-based compounds have
shown that the second domain of this scaffold is amenable, and that the spectrum of activity,
degree of potentiation, and type of antibiotics it potentiates can be altered by changing the
second domain.[11] Thus, we investigated the physicochemical impact of hybridizing a
hydrophobic rifampicin with a cationic hydrophilic tobramycin domain, as it relates to the ability
to permeate bacterial membrane and potentiate legacy antibiotics. The conjugates (Figure 1)
were found to potentiate different classes of Gram-positive-only antibiotics against P.
21

ACCEPTED MANUSCRIPT
aeruginosa, most especially doxycycline and chloramphenicol (Figure 2), while the individual
parent molecules alone or in combination could not (Table S5). Compound 1 with the shortest
tether was more active than 2 and 3 as standalone antibacterial agents, while compounds 2
and 3 were more potent than 1 as adjuvants. Indeed, compounds 1 – 3 potentiated antibiotics
that are clinically not indicated for the treatment of Pseudomonas infections but did not
potentiate typical antipseudomonal agents (Figure 2). This suggests a mechanism that
involves the reversal of intrinsic resistance of P. aeruginosa to these agents, especially
resistance to doxycycline and chloramphenicol.
P. aeruginosa is clinically resistant to tetracyclines and chloramphenicol due to its
intrinsic and adaptive resistance mechanisms. Tetracyclines enter Gram-negative bacteria via
a porin-mediated process and they inhibit the elongation phase of protein synthesis by
blocking the association of aminoacyl tRNA with the bacterial ribosome.[48] On the other hand,
chloramphenicol, which also prevents protein chain elongation by inhibiting the peptidyl
transferase activity of bacteria ribosome,[49] is believed to be taken up into the cell via an
energy-dependent process.[50] A major mechanism of resistance to these agents by P.
aeruginosa is by actively extruding them out of the cell using its abundantly expressed efflux
pumps.[48] The ability of compounds 2 and 3 to potentiate the effects of doxycycline and
chloramphenicol were found to be dependent on efflux pumps (Table 3), suggesting that
mitigating the effects of these pumps, in addition to outer membrane permeabilization, might
be responsible for the sensitization of P. aeruginosa to these agents. Outer membrane
permeabilization is a known property of all tobramycin conjugates synthesized to
date.[11,21,25] Agents that alter transmembrane protein environment (such as membrane
charge, fluidity, and thickness) and/or steric hindrance of membrane-embedded proteins can
22

ACCEPTED MANUSCRIPT
prevent the relay of signaling cascades required to elicit conformational changes necessary to
extrude substrate molecules by efflux pumps.[11] The perturbation of transmembrane efflux
protein domains by compounds 2 and 3 via alteration of lipid composition surrounding the
protein may, therefore, contribute to their ability to potentiate doxycycline and chloramphenicol
in P. aeruginosa. Indeed, rapid influx and reduced efflux are simultaneous complementary
processes necessary for enhanced drug bioaccumulation in bacteria cell.
Tetracyclines and chloramphenicol often act as model compounds for bacteriostatic
effects,[50] and the association of tetracycline with the ribosome is reversible,[48] providing an
explanation for the bacteriostatic nature of this class of drugs. The concentration-independent
bacteriostatic effects of doxycycline and chloramphenicol against wild-type PAO1 were
confirmed by growth curves that do not indicate any decrease in the number of CFU/mL from
the initial inoculum at MIC and supra MIC levels (Figure 5). However, in the presence of
compound 2, P. aeruginosa was sensitized to doxycycline and chloramphenicol (Table 2) and
their effects became bactericidal (Figure 5). The ability to switch the bacteriostatic effects of
these drugs to bactericidal effects (in combination with compound 2) in the same organism,
under the same growth conditions, is quite remarkable. Given the lack of antimicrobial activity
of compounds 2 and 3 against Gram-negative bacteria (Tables 1 and S1), the mechanism of
synergistic relationships with doxycycline and chloramphenicol is not immediately obvious. It is
possible that compounds 2 and 3 disrupt the OM of P. aeruginosa, thus, enhancing
bioaccumulation of doxycycline and chloramphenicol in the cytosol. However, this does not
fully explain the bactericidal effects of these combinations, given that very high concentrations
of tetracyclines and chloramphenicol are often bacteriostatic against P. aeruginosa.[25,50] It is
therefore conceivable that the bactericidal effect of these combinations emanates from the
23

ACCEPTED MANUSCRIPT
antimicrobial/biochemical potentiation of both participating molecules. For example, ribosome-
targeting antibiotics (tetracyclines, chloramphenicol, aminoglycosides) are known to
preferentially inhibit the biosynthesis of envelop proteins,[48–50] thus, impacting the OM and
facilitating the entry of some chemical entities.[51] We speculate that interference with the
fidelity of ribosomal proteins by doxycycline and chloramphenicol enhances the uptake of
compounds 2 and 3 across a structurally-weakened OM, thereby promoting the antibiotic
efficacy of the rifampicin domain. Rifampicin is bactericidal against WT P. aeruginosa.[25] This
scenario is consistent with the activity of compounds 2 and 3 against some Gram-positive
bacteria that lack an OM (Table 1), an effect that is believed to be mediated by the rifampicin
domain. The membrane-weakening effects of doxycycline and chloramphenicol possibly
augment the OM-permeabilizing properties of the amphiphilic tobramycin domain in
compounds 2 and 3, thus facilitating their uptake across the OM, while the rifampicin domain
possibly anchors the conjugate into the periplasm to mediate it antibacterial effect. The lack of
meaningful synergistic relationship between a double and triple combination of doxycycline or
chloramphenicol and tobramycin or rifampicin (Table S5) underscores the importance of linking
the two domains in compounds 2 and 3. Aminoglycosides (tobramycin) are taken up across
the cytoplasmic membrane via an energy-dependent process[22] while rifampicin lacks the
requisite physicochemical properties to navigate the OM.
It is gratifying to note that, consistent with in vitro potentiation, synergy was also
observed in vivo between doxycycline or chloramphenicol and compound 2 in G. mellonella in
vivo infection model (Figure 6). A 100 mg/kg monotherapy each of doxycycline,
chloramphenicol, or compound 2 resulted in 100 % mortality of PA260-challenged G.
mellonella larvae after 18 h, while a single dose administration of 100 + 100 mg/kg
24

ACCEPTED MANUSCRIPT
combination therapy each of doxycycline + compound 2 or chloramphenicol + compound 2
resulted in 87 % and 72 % survival, respectively, after 18 h in a dose-dependent manner
(Figure 6). This indicates the therapeutic potential of rifampicin-tobramycin conjugates in
combination with doxycycline or chloramphenicol to treat infections caused by MDR/XDR P.
aeruginosa. A 100 mg/kg of meropenem alone resulted in 100% mortality of the larvae after 24
h while colistin, the only antibiotic (besides amikacin) that the strain is susceptible to, require
up to 75 mg/kg to confer ≥ 50% survivability on the larvae after 24 h. Despite the obvious
limitation of this model such as the simple body plan and lack of many defined organs/tissues
in the larvae, the similarities between the innate immune responses in G. mellonella larvae and
humans, the demonstrable virulence of pathogenic bacteria and efficacy of therapies in both G.
mellonella larvae and humans, and the logistical and ethical relief of using G. mellonella make
it a suitable preliminary model to determine efficacy of therapeutic agents.[11,21,25,42,43]
Moreover, unlike other invertebrate models such as Caenorhabditis elegans and Drosophilia
melanogaster, G. mellonella larvae can survive at 37 °C and therefore allow the investigation
of temperature-dependent microbial virulence and antibiotic action.[52] However, further
studies of this combination is needed in mice infection model.
The newly synthesized compounds 1 – 3, alone and in combination with antibiotics,
were found to be non-cytotoxic to human kidney (HEK293) and liver (HepG2) cells in vitro
(Figures 7a and b), non-hemolytic to porcine erythrocytes in vitro (Figure 7c), and non-toxic to
G. mellonella larvae in vivo (Figure 6a) at relatively high concentrations. As a result, it rules out
the suspicion of a non-specific mode of action and demonstrates the relative safety of these
compounds against eukaryotic cells. Amphiphilic aminoglycosides have previously been
shown to target bacterial membranes more selectively than eukaryotic membranes.[53]
25

ACCEPTED MANUSCRIPT
26

ACCEPTED MANUSCRIPT
4. CONCLUSIONS
The lack of a robust pipeline of new agents, particularly against resistant Gram-negative
bacteria, emphasizes the importance of optimizing our current antimicrobials. Rifampicin–
Tobramycin conjugates have been shown to break the intrinsic resistance of P. aeruginosa to
doxycycline and chloramphenicol in vitro and in vivo, thus expanding the therapeutic
usefulness of these agents. In the presence of <10 µM of compounds 2 or 3, extrapolated
CLSI susceptibility breakpoints were reached in nine out of ten resistant P. aeruginosa
phenotypes for doxycycline (≤ 4 µg/ml), and seven out of ten for chloramphenicol (≤ 8 µg/ml).
A single dose combination of compound 2 with doxycycline or chloramphenicol also confer
survivability on MDR P. aeruginosa-challenged G. mellonella larvae while a monotherapy of
compound 2, doxycycline, or chloramphenicol does not. Whereas doxycycline alone exhibits
bacteriostatic effects against P. aeruginosa at supra MIC levels, when combined with
compound 2, its effect became bactericidal at sub MIC levels. In developed countries, usage of
chloramphenicol as broad-spectrum antibiotics has diminished over the years due to
increasing resistance and its well-described adverse effects, but the ability to achieve potent
bactericidal effects at lower concentrations could rejuvenate interest in this drug, especially in
this era of unabating resistance development. The current study has also demonstrated that
chemical entities with no intrinsic activity as standalone agents against Gram-negative bacteria
does not necessarily mean they are biologically irrelevant. Indeed, the attainment of specific
physicochemical threshold in a molecule is central to permeating the outer membrane of P.
aeruginosa. Further studies with tobramycin-based conjugates in general are warranted to
carefully elucidate and situate their role in clinical practice.
27

ACCEPTED MANUSCRIPT
5. Experimental Section
5.1
Chemistry. All chemicals and reagents were purchased from Sigma-Aldrich (Oakville,
ON, Canada) except 3-formyl rifamycin SV and tobramycin that were purchased from AK
Scientific Inc. (CA, USA). The chemicals were all used without further purification. Air and
moisture-sensitive reactions were performed with dry solvents under nitrogen atmosphere.
Thin-layer chromatography (TLC) was carried out on aluminum-backed silica gel 60 F254 GF
plates (0.25 mm) and/or aluminum-backed reverse phase silica gel 60 RP-18 F254S plates
(Merck KGaA, Germany) with the indicated solvents, and visualized under ultraviolet light
and/or by staining within ninhydrin solution in n-butanol. Compounds were purified by flash
chromatography on silica gel 60 (230-400 ASTM mesh) and/or reverse-phase C18 silica gel
(Silicyle, USA). Yields refer to chromatography-purified homogenous materials, except
1
13
otherwise stated. H and C NMR spectra were recorded on Bruker AMX-300 and AMX-500
spectrometers (Germany) as solutions and reported in the order of chemical shifts (δ) in ppm
relative to the indicated solvent, multiplicity (s, singlet; d, doublet; t, triplet and m, multiplet),
1
13
number of protons, and coupling constants (J) in hertz (Hz). H and C of compounds were
assigned using 1D and 2D NMR experiments such as Proton, COSY, Carbon-13, DEPT-135,
HSQC, and HMBC. ESI-MS and MALDI-TOF MS analyses were performed on Varian 500-MS
ion trap mass spectrometer (USA) and Bruker Daltonics Ultraflextreme MALDI TOF/TOF mass
spectrometer (Germany), respectively. Analytical HPLC was performed on Breeze HPLC
Waters equipped with W2998 PDA detector (1.2 nm resolution) coupled to Phenomenex
Synergi Polar (50 × 4.6 mm) 4 µm reverse-phase column with phenyl ether-linked stationary
28