A potent antibiotic-loaded bone-cement implant against staphylococcal bone infections

Abstract

New antibiotics should ideally exhibit activity against drug-resistant bacteria, delay the development of bacterial resistance to them, and be suitable for local delivery at desired sites of infection. Here, we report the rational design, via molecular-docking simulations, of a library of 17 candidate antibiotics against bone infection by wild-type and mutated bacterial targets, and the identification, via screening of the library for activity against multidrug-resistant clinical isolates, of an antibiotic that exhibits potent activity against resistant strains and the formation of biofilms, that decreases the chances of bacterial resistance, and that is compatible with local delivery via a bone-cement matrix. The antibiotic-loaded bone cement exhibited greater efficacy than currently used antibiotic-loaded bone cements against staphylococcal bone infections in rats. Potent and locally delivered antibiotic-eluting polymers may help address antimicrobial resistance.

Antibiotics underpin most of the modern medicine^1,2. For example, antibiotics are essential for surgical procedures and are critical for immunocompromised patients³. However, the emergence of resistant bacterial strains is rendering the existing arsenal of antibiotics ineffective^1,4. Antibiotic resistance kills approximately 700,000 people every year and is estimated to kill 10 million people by 2050 unless resistance is blocked through antibiotic stewardship or new antibiotics are continuously developed^5,4,6. Paradoxically, the new antibiotics pipeline has dried up as commercial incentives for the industry to develop new antibiotics are diminishing⁵. We propose a shift towards developing drug-device combinations that integrate the design of next-generation antibiotics with delivery tailored for the site of infection in the host. Such specialized drug-device strategies, targeting specific medical conditions, can potentially offer commercially viable paths towards developing next-generation antibiotics, and meet the requirements of antibiotics stewardship by limiting systemic exposure while ensuring that the target organ gets the required concentration of drug⁶. In this study, we describe the engineering of a drug-device combination for the treatment of orthopaedic infections.

Orthopaedic procedures are increasing, driven by the aging global population and trauma⁷. For example, the demand for total hip and total knee arthroplasties is predicted to grow to over 4 million procedures in US alone by 2030⁸. However, despite pre-operative and prophylactic antibiotics therapy, infections associated with invasive orthopaedic procedures are common, and, for example, can exceed 25% in complex open fractures or revision surgeries^9–11. Post-operative orthopaedic infections, not only from Staphylococcus sp., but also from Enterococcus sp., Cutibacterium sp., Pseudomonas sp., present a serious challenge even in the most advanced settings^8,12. Sections of dead bone are formed from the infection, which, together with limited vasculature, create a niche that is characterized by acidic pH, local immune suppression, and low antibiotics penetrance^13,14. In most cases, persistent bacterial biofilms develop¹², leading to chronic infections that require surgical interventions. While systemic antibiotics are widely used, this requires a sufficiently high dose of antibiotics to achieve efficacy at the necrotic region. Such systemic exposures of antibiotics in turn can alter the normal microbiome. To address these challenges, antibiotics-eluting acrylic bone cement spacers are increasingly being used¹⁵. Currently, vancomycin-, gentamicin-, tobramycin- or clindamycin-loaded bone cements are approved by the FDA. However, none of these antibiotics were originally developed for local delivery from bone cement. Furthermore, resistance has emerged against these antibiotics^16,17. Thus, there is a need for next-generation antibiotics that are optimized to meet this emerging need. Here, we describe the engineering of a new antibiotic for local delivery from polymethylmethacrylate (PMMA) bone cement matrix, using rational design and in silico screening against wild-type and mutated targets, as well as in vitro screening using clinical panels of susceptible- and drug-resistant bacteria. We compare the efficacy of this antibiotic-loaded PMMA cement against clinically used PMMA bone cements in a prophylactic and an established Staphylococcus sp.-infected tibial injury models, in vivo. Such drug-device combinations, tailored for specific medical applications, can emerge as the new paradigm in engineering next-generation antibiotics.

Results

Rationally designed antibiotics library.

As the first step, we designed a library of novel antibiotic molecules compatible with orthopaedic use. Fluoroquinolones (FQ) are reported to penetrate and attain high concentrations in the bone¹⁸, and bind to DNA gyrase and exert a potent bactericidal effect¹⁹. We therefore used a fluoroquinolone backbone as the template for designing our library. However, bacteria can develop resistance to fluoroquinolones²⁰. To address this, we rationalized that next-generation antibiotics should be designed with two active pharmacophores. While there is a high probability of a bacteria mutating a single target to gain resistance, mutating two distinct targets becomes probabilistically impossible, which can retard the development of new resistance. We designed and synthesized a library of molecules by incorporating various functionalized bio-active heterocyclic pharmacophores like imidazole, furan, thiazole, benzimidazole, and benzofuran moieties with optimum linkages to the piperazine unit at the C-7 position of a parent 1-cyclopropyl-6-fluoro-8-methoxy-4-oxo-7-(piperazin-1-yl)-1,4-dihydroquinoline-3-carboxylic acid nucleus (FQ nucleus) (Fig. 1a). Based on four designed structures (Formula I to IV), five molecules VCD-068, VCD-076, VCD-077, VCD-367, and VCD-381 were initially synthesized with different nitro-heterocyclic motifs having varying reduction potential (Fig. 1a). All the structures maintained a distance of ~6.1 to 6.8 Å between piperazine nitrogen to C5 in the case of the five-membered heterocyclic motif (Formula I to III) and C6 for the nine-membered fused heterocyclic unit (Formula IV). Furthermore, extensive target-based structure-activity relationship studies were performed by replacing the nitro group with hydrogen, methyl, bromo, amino, or other chemically or functionally equivalent groups and bio-isosteres of a nitro functional moiety of the screened molecules. Seventeen FQ molecules were designed satisfying the above design characteristics and subjected to sequential computational and in vitro screening (Fig. 1a, b).

Fig. 1. Design and in silico screening a library of next-generation antibiotics.

(a) Schematic showing the design of a library of next-generation antibiotics starting with a fluoroquinolone (FQ) backbone and functionalized bio-active heterocyclic pharmacophores with general formula I, II, III and IV. IA and IIA belong to the sub-family of I and II respectively. Distance between piperazine nitrogen to C5 in case of five membered heterocyclic motif (Formula I to III) and C6 for 9-memebered fused heterocyclic unit (Formula IV) was maintained at 6.1 to 6.8 A in all the energy minimized docked poses of ligands. (b) Schematic showing the screening of lead molecule from a library of 17 FQ derivatives. (c-d) Surface diagram of QBP of gyrase-DNA complex in (c), wild type (PDB ID: 5CDQ) and (d), mutant variant of the protein. In the mutant variant, 2 critical residues Ser84 and Glu88 were changed to Leu84 and Lys88 to reflect the mutations in nature. The surface area of QBP has been altered to 4181.9 sq. Å compared to 4383.6 sq. Å in wild type. (e-h) Docking poses of selected VCD molecules to S. aureus DNA gyrase in complex with DNA. Binding pose of (e), VCD-068 and (f) VCD-077 to wild type S. aureus DNA gyrase-DNA complex. The crystal structure has been taken from PDB ID: 5CDQ. VCD-068 and VCD-077 both maintained stable water-Mg²⁺ bridge interaction while interacting with at GyrA amino acid residues and additional interactions with GyrB residues. Binding pose of (g) VCD-068 and (h) VCD-077 to mutated S. aureus DNA gyrase-DNA complex with mutations Ser84Leu and Glu88Lys. Mutation generally causes perturbation in water-Mg²⁺ bridge interaction and steric clashes between bulky Leu residues and VCDs. However, in both cases, docking pose is marginally affected and interactions with GyrB are maintained. The residue numbers mentioned in the structures are based on Staphylococcus aureus DNA numbering of the PDB file. Dotted lines denote the H-bond interactions between ligand (cyan) and amino acid residues (green). The mutated residues are labeled in red and drawn in yellow. The DNA residues (orange) are coded as DA: deoxyadenosine; DG: deoxyguanosine; DT: deoxythiamine and DC: deoxycytosine. Figures were drawn using PyMol. (i) The docking score (kcal/mol) of seventeen VCD compounds against mutated S. aureus gyrase protein compared to moxifloxacin (number 18; closed triangle). Based on interactions and docking scores, five VCDs, VCD-068, VCD-077, VCD-176, VCD-367 and VCD-380 were found to maintain favourable docking pose with mutated gyrase protein as compared with moxifloxacin, a clinically-approved fluoroquinolone, which was used as control.

In silico modelling of the binding to the bacterial target.

Quinolone pharmacophores induce bacterial cell death by interfering with DNA supercoiling in DNA gyrase-DNA complex²¹. We, therefore, performed in silico molecular docking to study the binding of the designed molecules with both wild type S. aureus DNA-gyrase complex and its mutated protein variant with two critical residues of gyrase A subunit altered. Specifically, we mutated the polar, non-charged Ser84 to non-polar Leu residue and the acidic Glu88 to basic Lys. The choice of replacing Ser84Leu and Glu88Lys was based on reports of genomic analysis of clinical isolates conferring high quinolone resistance^22,23,24. Mutation of Ser84 to leucine or tryptophan confers high resistance levels against quinolone, whereas mutations that change Ser84 to alanine result in lower resistance levels²⁵. The surface area of the quinolone binding pocket (QBP) region was altered due to the introduction of the two mutations in gyrase A (Fig. 1c, d).

As shown in Fig. 1e, f and Supplementary Fig. 1, the binding poses of 5 molecules (VCD-068, VCD-076, VCD-077, VCD-367, and VCD-381) bound to the QBP of the wild-type protein-DNA complex revealed that the binding space of all these molecules at the active site overlapped with the conventional fluoroquinolone-binding site. All the molecules maintained stable water-Mg²⁺ bridge interactions between C3, C4- di-keto of the ligand and Ser84 and Glu88 residues of gyrase A protein, similar to the co-crystal structure of gyrase-moxifloxacin complex²⁶. Additionally, all the VCDs molecules interacted with Arg122 of gyrase A subunit through the C-3 carbonyl group of the FQ nucleus. All these interactions with gyrase A are responsible for stabilizing the stacking interactions of the aromatic quinolone ring with guanine and adenine bases of DNA.

As per our design strategy, VCD molecules interacted with gyrase B residues directly, exhibiting better binding than conventional fluoroquinolones (Supplementary Table 1a), except VCD-076, where the bulky benzimidazole group posed steric clashes with the protein residues (Supplementary Fig. 1a). VCD-367, containing a nine-membered benzofuran (Supplementary Fig. 1b), and VCD-381, containing nitroimidazole (Supplementary Fig. 1c), showed weaker interactions with gyrase B, unlike VCD-068 (Fig. 1e) and VCD-077 (Fig. 1f). VCD-068 and VCD-077 displayed favorable binding poses compared to the rest of the molecules, maintaining maximum interactions through N atoms of piperazine moiety and nitro group with gyrase B residues. The crucial interactions of VCD-077 with DNA gyrase B subunit, where furan ring creates pi-cation interactions with Arg458, salt bridge interaction of nitro group with both Arg458 and Lys417, and H-bonding interactions with the carbonyl group of amide linkage with Asn476, provided a unique advantage of tight tethering of the molecule at the binding site. Similarly, nitrogen-atom of nitrothiazole and the nitro group linked to the nitrothiazole in VCD-068 made additional contacts with Asn476 and Asn474, resulting in a stable VCD-068-gyrase-DNA complex. Both VCD-068 and VCD-077 showed van der Waals interactions with DNA gyrase residues, Arg458 and Glu477.

The Ser84Leu mutation can perturb the water-ion bridge formation because of steric constraints due to the decrease in hydrophilic surface area and the bulky sidechain of Leu protruding in the QBP (Fig. 1g, h). This generates positive van der Waals energy, which makes a classical fluoroquinolone-gyrase-DNA complex unstable. Since VCD-077 had very weak interactions with Ser84 (distance between the two electronegative groups Ser hydroxyl and the C3 carboxylate group of VCD-077 being 3.6 Å) (Fig. 1h), binding pose with the mutant variant was only marginally affected (−10.6 Kcal/mol to −9.3 Kcal/mol) (Supplementary Table 1b). The distance between the C3 keto acid was higher than the other molecules to accommodate the bulky Leu producing less steric hindrance (Fig. 1h). Further, the loss in interactions with gyrase A residues was compensated by strong H-bonding and van der Waals interactions of nitrofuran ring with amino acid residues at the gyrase B site. Similarly, VCD-068 (nitro-thiazole tethered) and VCD-367 (nitro-benzofuran tethered) also maintained the binding poses by interacting with gyrase B residues (Fig. 1g, Supplementary Fig. 1k). However, nitro-imidazole containing VCD-381 binding was greatly hampered due to a lack of strong interactions with the gyrase B. The constrained binding mode of nitro-benzimidazole containing VCD-076 was not improved in the quinolone resistance-determining region of gyrase protein (Supplementary Fig. 1j), resulting in a less stable ligand-protein complex among all other nitro-functionalized molecules.

Further modifications of VCD-068 and VCD-077 were carried out by replacing their nitro group of the nitroheterocycle moiety with other functional groups, and docking interactions were evaluated with the enzyme-DNA complex (Supplementary Fig. 1, Fig. 1i). Substitution of the nitro group with hydrogen, methyl, and bromo functionality seriously affected the interactions of VCD-365, VCD-370, and VCD-369 molecules, respectively, towards the gyrase B site, resulting in an overall weak ligand-DNA-gyrase complex than VCD-077. Functionally equivalent cyano (VCD-375), hydroxyimidamide (VCD-376), and tetrahydrofuran-2-one (VCD-135) moieties maintained strong H-bonding and van der Waals interaction with the wild type target protein but failed to interact with mutated protein effectively due to conformational constraints of the functional group. The distance between the carboxyl group of the FQ scaffold and the Leu84 sidechain created steric clashes in the binding of these molecules, thereby making the van der Waals interactions unfavorable. Furthermore, all the analogues exhibited weak interactions with gyrase B residues, Asn476, Asn474, and Glu477, unlike the nitrofuran-containing group of VCD-077 molecule (Supplementary Fig. 1; Supplementary Table 1b). Similarly, in the case of VCD-68 analogues, VCD-087 (methyl thiazole), VCD-373 (thiazole), VCD-082 (methyl 1,3,4-thiadiazole), and VCD-211 (hydroxyimidamido thiazole), docking analysis against mutated target protein revealed that nitro-thiazole of VCD-068 was able to maintain optimum binding interaction compared to other functionalized thiazole scaffolds (Supplementary Fig. 1, Supplementary Table 1b). VCD-176 (cyano thiazole) and VCD-380 (amino thiazole) showed marginally better binding with the mutated enzyme-DNA complex.

Based on interactions and docking scores, five VCDs, VCD-068, VCD-077, VCD-176, VCD-367, and VCD-380 were found to maintain a favourable docking pose with mutated gyrase protein as compared with moxifloxacin, a clinically-approved fluoroquinolone, which was used as control (Fig. 1i). These docking analyses were further verified by in vitro studies testing all the compounds against quinolone-resistant S. aureus strain.

In vitro activity of VCD molecules.

We next synthesized the drug library and chemically characterized the molecules using NMR and mass spectrometry as described in materials and methods (Supplementary Fig. 2–4). We then evaluated the molecules for in vitro bioactivity against different target pathogens. As Staphylococcus is the most common bacteria underlying bone infections, we first screened the candidate molecules against an antibiotic-sensitive strain (ATCC25923), a methicillin-resistant Staphylococcus aureus (MRSA) strain (ATCC 43300) and a fluoroquinolone-resistant S. aureus (QRSA) strain (CCARM3505). Our goal was to identify the rank-order of the VCD molecules based on the lowest minimum inhibitory concentrations (MIC) values across both antibiotic-susceptible as well as drug-resistant strains of S. aureus. We used gentamicin as a positive control as it is the most common drug in FDA-approved antibiotic-loaded bone cement. We additionally used ciprofloxacin (second generation fluoroquinolone) and moxifloxacin (fourth generation fluoroquinolone) as additional controls to directly compare our molecules to other approved antibiotics of the same class. MIC of VCD molecules against S. aureus strains showed consistent results with in silico findings, i.e., the molecules that showed low docking scores due to constrained binding pose in gyrase-DNA complex, so called negative controls, were found to exhibit high MIC values corroborating with theoretical predictions. For example, VCD-135 maintained strong H-bonding and van der Waals interaction with the wild-type target protein but failed to interact with mutated protein effectively due to conformational constraints of the functional groups, resulting high MIC values. Similarly, the binding of nitroimidazole containing VCD-381 and nitro-benzimidazole containing VCD-076 was not improved in the QRDR domain of gyrase protein, resulting in less stable ligand-protein complex among all other nitro- functionalized molecules as reflected from their high docking energy scores (Supplementary Table 1a and 1b) and high MIC values (Table 1). Indeed, MIC of VCD-076, VCD-381 and VCD-135 against FQ resistant S. aureus strain (CCARM 3505) was found to be greater than 16 μg/ml.

Table 1 |.

Minimum inhibitory concentrations of VCD molecules in S. aureus strains.

Antibiotics	Minimum Inhibitory Concentrations μg/ml)
	ATCC 25923*	ATCC 43300**	CCARM 3505#
VCD-068	0.02	0.01	0.50
VCD-076	>16.00	>16.00	>16.00
VCD-077	0.03	0.02	0.50
VCD-082	0.06	0.06	4.00
VCD-087	0.03	0.06	16.00
VCD-135	0.50	0.13	16.00
VCD-176	0.03	0.03	2.00
VCD-211	0.13	0.06	4.00
VCD-365	0.13	0.13	4.00
VCD-367	0.03	0.03	4.00
VCD-369	0.06	0.06	4.00
VCD-370	0.13	0.13	8.00
VCD-373	0.03	<0.03	2.00
VCD-380	1.00	1.00	>16.00
VCD-381	0.25	0.25	16.00
Moxifloxacin	0.03	0.06	8.00
Ciprofloxacin	0.50	0.50	64.00
Gentamicin	0.50	>16.00	0.50

^*ATCC 25923: drug susceptible strain;

^**ATCC 43300: MRSA strain;

^#CCARM 3505: quinolone resistant strain.

Based on the above screen, the nitrofuran- and nitrothiazole-based fluoroquinolones, VCD-077 and VCD-068, respectively, exhibited low MIC values against both quinolone-sensitive and -resistant Staphylococcus strains as well as MRSA. In contrast, the classical ciprofloxacin exhibited a significant increase in MIC values in the resistant strain (Table 1). Molecules showing MIC higher than 2 μg/ml in the quinolone resistant CCARM 3505 strain were not selected for subsequent studies. We next tested VCD-068 and VCD-077 against an array of clinical isolates of QRSA that were methicillin-sensitive (MSSA) or additionally methicillin-resistant (MRSA). The MIC of VCD-068 and VCD-077 were found to be 16-fold lower than ciprofloxacin and 8-fold lower than moxifloxacin against the clinical isolates (Table 2). Gentamicin also exhibited high MIC values against these S. aureus strains, confirming the prevalence of resistance to gentamicin. We additionally tested the drugs for efficacy against Staph. epidermidis (ATCC 35984 and S12-1), a facultative anaerobe Cutibacterium acne (strain MTCC 1951) and a Gram -ve E. coli, which are all implicated to different degrees in bone infections. As shown in Table 2, both the VCD molecules were effective against these strains, confirming their broad-spectrum activity.

Table 2 |.

Minimum inhibitory concentrations of VCD-077 and VCD-068 in comparison to other known antibiotics in different bacterial strains.

Organism	Minimum Inhibitory Concentrations μg/ml)				1041
	Ciprofloxacin	Moxifloxacin	Gentamicin	VCD-077	VCD-068
S. epidermidis ATCC 35984	0.13	0.06	>16.00	<0.03	<0.03
S. epidermidis S12-1	8.00	1.00	1.00	0.13	0.06
C. acnes MTCC 1951	0.50	0.25	1.00	<0.03	0.06
MSSA-1, QRSA	16.00	8.00	2.00	1.00	0.50
MSSA-2, QRSA	16.00	8.00	2.00	1.00	0.50
MRSA-1, QRSA	8.00	4.00	31.50	<0.50	0.50
MRSA-2, QRSA	8.00	4.00	63.00	<0.50	0.50
MRSA-3, QRSA	8.00	4.00	15.80	<0.50	0.50
MRSA-4, QRSA	8.00	4.00	31.50	<0.50	0.50
MRSA-5, QRSA	8.00	4.00	63.00	<0.50	0.50
E. coli MTCC 1687	0.02	0.06	4.00	1.00	1.00

^*S12-1: clinical isolate; MSSA: methicillin susceptible S. aureus strain; MRSA: methicillin resistant S. aureus strain; QRSA: quinolone resistant S. aureus strain.

Preliminary safety screen for VCD-077.

We next tested VCD-077 and VCD-068 for toxicity on human osteoblasts. VCD-077 had lesser inhibitory effect on the cells compared with VCD-068, with a translating into a wide therapeutic index of ~5107. In comparison, the therapeutic index of VCD-068 was ~4505 (Supplementary Fig. 5). As a result, we selected VCD-077 for further studies. To test for any potential toxicity of VCD-077 on kidney or the liver, we next treated mice with a high dose of VCD-077 (150 mg/kg/mouse ~ 750 mg human equivalent dose, once daily for 5 days), administered orally. No significant changes in blood urea or alanine aminotransferase (ALT), alkaline phosphatase (ALP), aspartate aminotransferase (AST), globulin, albumin, total protein, triglycerides and cholesterol levels between control and treatment group was observed at this dose. Additionally, no change in body weight or behaviour was observed over the treatment period (Supplementary Fig. 6). Furthermore, a computational in silico model to predict the potential for hERG K+-channel blockage by VCD-077 revealed an IC₅₀ of about 6.68 mM, suggesting a wide safety margin. We next evaluated whether VCD-077 induced cytochrome p450 (CYP) 1A2, 2B6, and 3A4 isozymes or inhibited CYP 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1, and 3A4 isozymes in a human liver microsome assay. The induction of each isozyme by VCD-077 was observed by following the metabolite production of the FDA-recommended substrate of the specific isozyme. As shown in Supplementary Fig. 7, VCD-077 is not an inducer or inhibitor of CYP enzymes.

In vitro efficacy against clinical isolates of sensitive and resistant bacterial strains.

We next validated the efficacy of VCD-077 in 10 vancomycin-intermediate MRSA (VISA) and 10 VRSA clinical strains from diverse infections, which were obtained from the National Antibiotic Resistant S. aureus (NARSA) collection (JMI, Iowa, USA). The MIC₅₀ and MIC₉₀ of VCD-077 against these strains were comparatively lower than antibiotics presently used with bone cements, as well as other potent fluoroquinolones, including levofloxacin and moxifloxacin (Table 3, Supplementary Table 2). Finally, we tested VCD-077 against 50 different S. aureus strains (where 80% strains were resistant to levofloxacin) collected from US medical centers within the SENTRY Antimicrobial Surveillance Program in 2017. As shown in Fig. 2, VCD-077 exhibited MIC values mostly in the range of 1 μg/ml or less against these strains, whereas the moxifloxacin MIC values were typically ~2 μg/ml or more.

Table 3.

In vitro efficacy against clinical isolates of sensitive and resistant bacterial strains from the National Antibiotic Resistant S. aureus (NARSA) collection.

Clinical isolates	Antibiotics	MIC50 μg/ml)	MIC90 μg/ml)	Number of isolates
VISA	Vancomycin	4.00	4.00	10
	Gentamicin	0.50	> 32.00
	Moxifloxacin	4.00	> 4.00
	Levofloxacin	> 8.00	> 8.00
	VCD-077	0.50	1.00
VRSA	Vancomycin	> 32.00	> 32.00	10
	Gentamicin	32.00	> 32.00
	Moxifloxacin	> 4.00	> 4.00
	Levofloxacin	> 8.00	> 8.00
	VCD-077	0.50	2.00

Fig. 2. In vitro efficacy against clinical isolates of sensitive and resistant bacterial strains.

MIC distribution of VCD-077 compared to moxifloxacin when screened against 50 clinical isolates of Staphylococcus aureus collected from US medical centers within the SENTRY Antimicrobial Surveillance Program. VCD-077 exhibited an MIC value of less than 1μg/ml in most strains.

VCD-077 retards the development of resistance.

A critical parameter for next-generation antibiotics is to be able to retard the development of resistance. The propensity of VCD-077 to generate resistant strains was checked using a serial passage assay against the S. aureus standard strain (ATCC 25923), as well as MRSA (ATCC 43300). We used ciprofloxacin as a positive control as it is a widely used antibiotic of the same mechanism class and because S. aureus is known to develop resistance against ciprofloxacin²⁷. Indeed, the MIC of ciprofloxacin increased stepwise to 64.0 μg/ml by the 20th passage. In contrast, the bacteria remained sensitive to VCD-077 (MIC reached a maximum of 2.0 μg/ml) even at the at the 28th passage (Fig. 3a), suggesting that VCD-077 has a low propensity to develop resistance even after prolonged exposure. The difference was more pronounced in MRSA strain where the ciprofloxacin reached a MIC of 512 μg/ml as early as the 15th passage, whereas VCD-077 showed an MIC of 1 μg/ml at that passage (Fig. 3b). Mutant prevention concentration (MPC) indicates the lowest concentration of the drug above which the selective growth of resistant mutants is only rarely expected to occur³³. VCD-077 displayed an MPC of 4 μg/ml, which is ~64 × MIC (Supplementary Table 3). When we sub-cultured the residual S. aureus colonies (observed on plates with 2 μg/ml concentration of VCD-077) in drug-free media for two passages and then treated with VCD-077, the bacteria was found to be susceptible at the original MIC (0.06 μg/ml), indicating the surviving bacteria were not true resistants but ‘persisters’²⁸. Persisters are the bacteria that tolerate acute high concentrations of an antibiotic yet are sensitive to the same drug after a drug washout²⁸. In contrast, the MPC of ciprofloxacin was 4 – 8 × MIC, and the sub-MPC colonies maintained the high MIC even after subculturing, which could explain the early emergence of resistance.

Fig. 3. VCD-077 retards resistance development.

Generation of resistant mutants by VCD-077 using serial passage assay showing changes in MIC through 28 passages against (a) S. aureus (ATCC 25923) and (b) MRSA (ATCC 43300). VCD-077 has low propensity to develop resistance even after prolonged exposure (c) Effect of VCD-077 on DNA supercoiling using S. aureus DNA gyrase. The data is represented as percentage of DNA supercoiling (mean ± S.D., n=3). The percentage of supercoiling of the relaxed plasmid DNA in presence of S. aureus DNA gyrase was measured by estimating the intensity of the supercoiled plasmid DNA band. Ciprofloxacin was used as a positive control. The intensity of the supercoiled DNA band in absence of any drug was taken as 100%. Statistical analysis was performed using unpaired t-test with Welch correction. VCD-077 inhibited S. aureus DNA gyrase-mediated DNA supercoiling in a dose-dependent manner and to a greater degree than equimolar concentration of ciprofloxacin. (d) Cell viability of E. histolytica (which is responsive to the effect of nitro-heterocyclic but not quinolone motifs) upon the treatment of VCD-077. Moxifloxacin was used as a negative control and nitazoxanide, a nitroheterocyclic compound, was used as a positive control. Data is represented as percentage of cell viability (mean ± S.D., n=6). Statistical analysis was performed using two-tailed t-tests (*p<0.0001 vs vehicle control).

Dual mechanism of action of VCD-077.

To validate the in-silico prediction that VCD-077 binds with bacterial DNA gyrase, we explored the action of VCD-077 on S. aureus DNA gyrase using a cell free assay system. As shown in Fig. 3c, VCD-077 inhibited S. aureus DNA gyrase-mediated DNA supercoiling dose dependently, with a lower IC₅₀ than ciprofloxacin. Additionally, VCD-077 has a nitroheterocycle-based second pharmacophore, which can exert its own bactericidal activity via an independent DNA-damaging effect. To test if the second pharmacophore is active, we used a functional assay in Entamoeba histolytica (HM-1: IMSS), where fluoroquinolone moieties have minimal activity whereas the nitroheterocycle compounds are active. Indeed, VCD-077 induced potent cell kill, while the negative control, moxifloxacin, exhibited no efficacy in this model system. A nitro-heterocyclic compound, nitazoxanide, used as a positive control, was highly effective in this system (Fig. 3d). This dual pharmacophore functionality could underpin the efficacy of VCD-077 in quinolone-resistant bacteria seen earlier as well as the ability to retard the development of resistance.

Anti-biofilm activity of VCD-077.

Biofilm formation is of particular concern in bone infections¹². Thus, it is an essential prerequisite for antibiotics targeted for bone infection therapy to have biofilm eradication property. We used two bacterial strains, S. aureus (ATCC 6598P) and S. epidermidis (ATCC35984), that are well known to form biofilms²⁹ to test the anti-biofilm activity of VCD-077. Confocal microscopy by labelling the biofilms of S. aureus with SYTO 9 as well as scanning electron microscopy revealed that VCD-077 treatment disrupts the biofilm to a greater degree than either gentamicin or vancomycin. VCD-077 treatment reduced biofilm biomass by more than 8-fold and biofilm thickness by more than 5-fold from growth control samples at 72 h growth control (Fig.4, Supplementary Table 4). A similar efficacy was observed with VCD-077 against the S. epidermidis biofilm (Supplementary Fig. 8a). This can potentially be explained by the structural features of VCD-077 having a nitro-heteroaromatic substituent attached at the C7 position of piperazine moiety. VCD-077 is a weak acid with a single pKa of 5.8, i.e., it remains neutral at an acidic pH (Supplementary Fig. 8b–c). Indeed, consistent with this observation, VCD-077 was found to retain its potency in the pH range 5.5 to 7.4, unlike gentamicin, which lost some efficacy at an acidic pH 5.5 (Supplementary Fig. 8d). The translocation of fluoroquinolone type of hydrophobic molecules through the membrane is reported to take place via passive diffusion, both in Gram-negative and Gram-positive bacteria, with the net charge of the molecule being the key determining factor for controlling the diffusion rate of any given molecule through the thick peptidoglycan-based membrane. The translocation free energy barrier is found to be less with neutral form of the molecule than the zwitterionic form of the same molecule³⁰. Therefore, the neutral form of VCD-077 can allow higher penetration through matured biofilms, where an acidic environment prevails³¹. Vancomycin exhibited poor antibiofilm activity despite maintaining antibacterial activity at acidic pH of the biofilm, consistent with published reports³². VCD-077, therefore, offers a unique advantage in the acidic niche prevalent in the infected bone environment.

Fig. 4. Antibiofilm activity of VCD-077.

Confocal images showing effect of drug (16 μg/ml) on SYTO^® 9 stained biofilms of S. aureus ATCC 6538P grown for 72 h. (b) SEM images showing effect of drug (16 μg/ml) on matured biofilms (72 h) of S. aureus ATCC 6538P. The magnification of the image was 5000×. VCD-077 has potent anti-biofilm activity against S. aureus unlike vancomycin or gentamicin.

VCD-077 is physiochemically compatible with bone cement.

PMMA is commonly used as bone cement in orthopaedic surgeries³³. We rationalized that the lipophilic VCD-077 could be efficiently loaded into the moderately hydrophobic PMMA bone cement. VCD-077 and PMMA were mixed into a homogenous powder and then processed into beads or cylinders (Fig. 5a). Polymerization of PMMA is an exothermic reaction, with temperatures reaching almost 50°C at the bone surface and over 100°C in the mixing bowl³⁴. We therefore first tested the effect of temperature on VCD-077 activity. As shown in Supplementary Table 5, there was no change in the MIC of VCD-077 even after being heated to 130°C. Furthermore, FT-IR analysis indicated that the functional groups of the drug molecules and PMMA remains intact after mixing them (Extended Data Fig. 1).

Figure 5: Physico-chemical characterization of PMMA loaded beads and cylinders.

(a) Pictorial representation of VCD-077 loaded beads with 4.6 mm height and 4.2 mm diameter and cylinders of 12 mm height and 6 mm diameter. (b) Drug release kinetics from 3 different bone cements. Data is represented as mean ± SD (n=3). The best fit curves for the release kinetics of VCD-077 from Palacos (R² = 0.9971), Simplex (R² = 0.9909) and Smartset HV^® (R² = 0.9849) bone cements. A quick onset and sustained release were observed from Smartset HV^® versus the other two variants. (c) Comparative drug release rate from different bone cement beads at 40:3 polymer drug ratio calculated from the release kinetics. The data is represented as mean ± S.D. (n=3). Statistical analysis was performed using two-tailed t-tests (*p<0.005 compared to other two cements). Smartset HV^® shows higher drug release rates (3.6682 μg/ml/√hr) compared to Palacos R^® (0.8309 μg/ml/√hr) and Simplex P^® (0.7301 μg/mL/√hr) PMMA bone cements. (d) Comparative drug release rate from VCD-077 impregnated Smartset HV^® bone cement beads at different polymer:Drug ratio. Data is represented as mean ± S.D. (n=2). Statistical analysis was performed using two-tailed t-tests (*p<0.005 compared to other two cement:drug combinations). Higher drug loading enhances drug release rates (3.277 μg/ml/√hr) by 10X compared to drug release rate (0.315 μg/ml/√hr) at low drug : polymer ratios (1 : 40) in Smartset HV^® PMMA bone cement. (e) Surface morphology (SEM images) of Smartset HV^® beads with or without drug (at 40:1 polymer-drug ratio) at day 0 and at 14^th day of drug release stage. Solvent channels observed after drug release on day 14 are denoted with yellow arrows. The conspicuous micro channels in the cement mantle can lead to widening of the open spaces in the drug polymer network, compared to one without drug. (f) Compression strength as per ASTM F451 of Smartset HV^® (PMMA) cylinders alone, VCD-077 loaded PMMA cylinders with drug:polymer ratio 3:40 on Days 0, 7 and 14 during drug release. Data is represented as mean ± SD (n=5). Statistical analysis was performed using two-tailed t-tests (ns: nonsignificant; p>0.5 w.r.t Smartset HV^® without drug).

We next tested the impact of different variables, such as drug: polymer ratios, the particle size of the drug, and the nature of bone cement matrices on the release kinetics of the entrapped drug from the bone matrix. We used three commercial bone cements, Smartset HV (high viscosity), Simplex P (medium viscosity), and Palacos R (high viscosity). Quick onset and sustained release of VCD-077 were observed from Smartset HV versus the other two variants (Fig. 5b). Smartset HV exhibited higher drug release rates (3.6682 μg/ml/√hr) compared to Palacos R (0.8309 μg/ml/√hr) and Simplex P (0.7301 μg/mL/√hr) PMMA bone cements (Fig 5c). We, therefore, selected the Smart set HV bone cement for further studies. We next studied the impact of increased drug loading on release from Smart set HV bone cement. We used drug-entrapped beads of size ~4.2 mm × 4.6 mm at room temperature, and measured cumulative release concentration of VCD-077, incorporated at 1:40, 2:40, and 3:40 drug: polymer ratio, over 14 days. Release kinetics of all drug: polymer ratios were dependent on drug loading and were characterized by an initial burst, followed by a sustained release profile, which is desirable in vivo. (Extended Data Fig. 1d). Higher drug loading enhanced drug release rates (3.277 μg/ml/√h), almost ten times compared to the drug release rate (0.315 μg/ml/√h) at low drug: polymer ratio (1:40) in Smartset HV PMMA bone cement (Fig. 5d). Analysing surface morphology of PMMA beads and drug-loaded PMMA beads using scanning electron microscopy (SEM) revealed a continuous, fused, spherical polymeric chain structure in the case of PMMA beads (without antibiotic) (Fig. 5e). In contrast, surface non-uniformity was noted when PMMA was loaded with VCD-077. Further, surface morphology examination of the beads after 14 days of drug release revealed the appearance of conspicuous micro channels in the cement mantle leading to widening of the open spaces in the drug-polymer network, compared to one without drug. These micro channels could potentially facilitate sustained drug release from the polymeric matrix. In a recent study, micro or nanoclusters of drugs were found to result in a more optimal release from bone cement beads than free drug³⁵. We, therefore, tested the impact of the size of drug particles on release studies from the PMMA beads. The beads were loaded with either larger particle of VCD-077 (mean particle size ~134 μm or micronized VCD-077 (mean particle size ~13 μm) at a 1:40 drug: polymer ratio. Beads with micronized VCD-077 were found to marginally improve the drug release rate (Extended Data Fig. 1e). No significant difference was observed in the release of drug from the bone cement between 32°C and 37°C (Extended Data Fig. 1f).

A concern with the appearance of microchannels is the potential loss of mechanical strength of the cement structure. We next performed mechanical characterization of VCD-077-impregnated bone cement using cylindrical specimens 6 mm × 12 mm of Smartset HV versus Smartset HV with VCD-077 at 3:40 drug: polymer ratios as per ASTM-F451 protocol³⁶. The compressive strength of Smartset HV and VCD-077 loaded Smartset HV were around 86.65 ± 4.40 MPa and 87.38 ± 5.16 MPa respectively. Similarly, Young’s modulus for Smartset HV and VCD-077-loaded Smartset HV were found to be 1311.00 ± 99.35 MPa and 1335.00 ± 100.00 MPa, respectively. Thus, VCD-077 loading into Smartset HV did not alter the inherent mechanical property of the bone cement matrix. Further mechanical properties of these cylinders were checked at day 7 and day 14 of drug release and compared with drug-loaded beads at day 0. Mechanical strength remained unchanged despite drug release (Fig. 5f).

In vitro efficacy of antibiotic-loaded bone cement against Staphylococcus sp.

We next evaluated the efficacy of VCD-077-impregnated PMMA beads against MRSA (ATCC 43300) in vitro by measuring the zone of inhibition (ZOI) over 14 days (Supplementary Table 6). We used a clinically used gentamicin-loaded beads as a positive control, which exhibited no activity against the MRSA strain. In contrast, we observed a rapid and sustained clearing of the resistant bacteria with VCD-077-impregnated PMMA beads, consistent with the release kinetics seen previously.

VCD-077-impregnated bone cement decreases Staphylococcus infection load in a tibial injury model.

We used two distinct models to test the efficacy of VCD-077-loaded bone cement as in vivo proof of concept. In the first model, we tested whether VCD-077-loaded bone cement prevented bacterial colonization. We drilled a hole in the rat tibia, placed a drug loaded PMMA bead in the hole of the tibia after inoculation with the bacteria (Fig. 6a). Given that gentamicin-loaded PMMA cements are the most widely used antibiotic in FDA-approved bone cements, we included it as a comparator control. We also included two additional negative control arms: (1) surgical site-infected animals without any treatment (infection control) and (2) surgical site-infected animals that received PMMA beads without any antibiotic (PMMA control). At different time points post-inoculation, animals were sacrificed, and the infected bone was isolated, and the bacterial load was estimated for using a colony formation assay. As shown in Fig. 6b, while both gentamicin- and VCD-077-loaded bone cements inhibited infection by Day 7, the latter resulted in significantly greater antibacterial efficacy compared to vehicle control (p<0.0001) as well as gentamicin-loaded beads (p<0.005) at longer time points. This is consistent with our in vitro observations that confirm that we achieve a steady state release of VCD-077 over a long period of time. Histopathology analysis after H & E staining of infected bone samples revealed increased reactive new bone formation, presence of dead bone fragments, as well as acute inflammatory cells, suggesting signs of osteomyelitis in infection and vehicle control arms. Treatment with VCD-077-loaded bone cement resulted in improvement in the above pathological parameters (Fig. 6c, Supplementary Table 7). In parallel, we collected blood at different time points over 24 h (and the bone tissue at the last time point) after implanting the beads in the bone and used liquid chromatography-tandem mass spectrometry to measure the drug concentration. The level of VCD-077 was found to be below detection limits (49.4 ng/ml) in blood, consistent with the minimal drug applied locally at the bone site. In contrast, the drug concentration in the bone at 24 h post-implantation was found to be 1.06 ± 0.83 μg/g at the site of application (Supplementary Fig. 9a), which is higher than the MIC value for VCD-077. Taken together, the efficacy data and the pharmacokinetic measurements indicate that we achieve the desirable localized concentrations with minimal systemic exposure, which was the goal of delivering the antibiotic from the bone cement. Additionally, we evaluated the skeletal muscles adjoining the site of application for any toxicological effect of VCD-077 treatment. As shown in Supplementary Fig. 9b–c, histopathology analysis of muscle tissue after 28 days treatment demonstrated that VCD-077 did not show any adverse effect on the muscle tissue.

Figure 6: VCD-077-embedded bone cement beads are effective against bone infections.

(a) Procedure and experimental design for in vivo bone infection model to test the prophylactic activity of antibiotics-loaded bone cement. (b) Graph shows bacterial load (log₁₀CFU/ml) in bone tissue of the S. epidermidis-infected tibia, determined at designated time points during the efficacy experiment. Drug-loaded PMMA beads were implanted at the site of infection on day 0 prophylactically (except in infection control group). Smartset PMMA bone cement served as the negative control. The data is represented as mean ± SD (n = 4) (*p=0.004; **p<0.0001). (c) H&E-stained images of bone sections at Day 28 from animal efficacy model. (d) Schematic shows experimental design for testing VCD-077-impregnated PMMA bone cement in an experimental model of established osteomyelitis. S. epidermidis (ATCC 35984, 25 μl of bacterial innoculum at 2 ×10⁷ CFU/ml) was injected through a hole drilled into the medullary cavity of the tibia, and the wound was closed. Twenty-one days post-infection, the animals underwent limited surgical debridement of the infection site, and a SmartSet HV PMMA loaded with VCD-077 (3 mm bead) placed by boring a ~3.2 mm hole through the cortical bone. SmartSet HV PMMA (Veh) served as negative control while SmartSet HV PMMA loaded with either gentamicin or rifampicin served as positive controls. Graph shows the efficacy of different bone cements in reducing bacterial load at day 10 post-application. Data shown are mean ± SEM (n=3–4, ANOVA followed by Tukey’s multiple comparison test).

We next tested the efficacy of the VCD-077-impregnated bone cement in an established model of osteomyelitis (Fig. 6d). We injected S. epidermidis (2 × 10⁷ CFU/ml) into the hole drilled into the tibia and closed the surgical site using a non-absorbable sterile surgical ligature. Twenty-one days later, we performed limited surgical debridement of the infected site by creating a ~3.2 mm hole through the cortical bone to the medullary cavity and implanted the SmartSet HV PMMA bead loaded with VCD-077 into the hole. As active comparator, we used the FDA-approved gentamicin-loaded SmartSet HV PMMA bead. Although, rifampin-loaded bone cement is not approved as with PMMA, we included it as a second positive control given that gentamicin exhibited a high MIC value against bacterial strain. VCD-077-impregnated bone cement resulted in a significantly greater antibacterial effect as compared with either gentamicin or rifampin-loaded bone cement. All the active treatments improved pathological parameters of infection over the infected control (Supplementary Table 8). In both these studies, the total dose of VCD-077 applied in the bone was ~13.5 mg/kg per 100g rat, which translates to ~130 mg human equivalent dose, which is significantly lower than the dose used in our chronic toxicity studies.

Discussion

Infection of the bone is increasingly becoming a prevalent medical problem with implications on quality of life³⁷. Current management strategies require prolonged systemic antibiotic interventions, and, in many cases, revision surgeries that include bone debridement or replacement of implants, which are all sub-optimal solutions³⁸. Here, we report the rational design of a novel antibiotic, VCD-077, that is physicochemically compatible with PMMA bone cement, exhibits desired drug release kinetics without loss of mechanical characteristics of the bone cement, and exerts potent efficacy against a broad range of clinically isolated sensitive as well as drug-resistant strains of bacteria, while retarding the development of resistance. Additionally, as compared with antibiotics commonly used with bone cements in the clinical setting, VCD-077 was able to exert a greater anti-biofilm activity. Biofilms are often overlooked while screening antibiotics and pose a major challenge in bone infections by limiting the efficacy of antibiotics. Taken together, the local application of VCD-077-loaded bone cement could emerge as a next-generation drug-device combination for the treatment of orthopaedic infections.

In their guidance for developing such innovative combination products, the FDA recommends that consideration should be given to the potential interactions between the device and the drug³⁹. Our studies indicate that the drug does not interact with the PMMA matrix, which could explain the efficient drug release from the matrix, and additionally, the drug retains efficacy after being loaded into the cement. The compression strength of cement beads was also comparable to the predicate device, i.e., commercial bone cement matrix (Smartset HV) alone. Maintaining the physicochemical characteristics of the bone cement similar to a predicate device after drug incorporation can facilitate regulatory development for clinical translation. Here, we used FDA-approved PMMA bone cement as it is the gold standard for use in orthopaedic procedures⁴⁰.

The second component of the drug-device combination is the drug itself. Here, the critical element in increasing efficacy is the use of rational antibiotic design, which together with in silico screening allowed us to rapidly shortlist lead molecules. Indeed, the application of computational tools holds the potential for revolutionizing antibiotics discovery⁴¹. We adopted a new design strategy that involved engineering a library of antibiotics by tethering the FQ nucleus with bioactive functionalized heterocyclic scaffolds. While FQs are known to penetrate and achieve desirable concentrations in the bone, in silico modelling revealed that the functionalization with certain heterocyclic scaffolds confers improved binding with the bacterial target even when critical residues are mutated, which allowed prioritization of further development of these molecules. Additionally, as seen with E. histolytica, which is not susceptible to the FQ pharmacophore activity, the nitroheterocycle pharmacophore could potentially confer an additional mechanism of cell kill. Indeed, we observed that VCD-077 exhibited a broad antimicrobial spectrum, including against QRSA, suggesting that, mechanistically, such dual-action antibiotics can be effective against bacteria that are already resistant to existing antibiotics or carry a mutation that can limit the impact of one of the two pharmacophores, and thereby reduce the probability of the bacteria developing resistance⁴¹. Additionally, we observed that the bone cement-based delivery resulted in a local concentration of drug that exceeded the MIC values, which is desirable to kill the bacteria as well as retard the development of resistant strains, while the systemic concentration was below the detectable limits, which can limit the adverse impact that oral or systemic antibiotics have on the microbiome. This is a significant goal of antibiotic stewardship efforts to minimize the development of resistance.

According to the FDA, innovative technologies combining drug and device hold great promise for making treatments safer, more effective, or more convenient and acceptable for patients. In a recent regulatory guideline, the FDA has defined a drug-led development pathway for drug-device products that combine a novel drug, for example, VCD-077, with an established device such as the PMMA bone cement⁴². This pathway relies on the demonstration of both safety and effectiveness of the novel drug. Indeed, our proof-of-concept toxicity studies with VCD-077 supported a wide therapeutic index, minimal interaction with cytochrome p450 enzymes and minimal changes in liver and kidney function tests. Additionally, in two distinct models of osteomyelitis, the first to test a prophylactic treatment, and the second to treat chronic osteomyelitis, VCD-077-loaded bone cement compared well against existing treatment modalities. Taken together with the safety data, this activity of VCD-077-loaded bone cement in different bone infections supports further development of this drug-device combination. Indeed, the need for new drugs that can be used to treat infectious diseases has never been more acute due to the emergence of resistant pathogens⁴⁷. Our proof-of-concept study suggest it may be possible to design new antibiotics that are active against the resistant strains and additionally retard the development of new resistance and integrating with organ (bone)-specialized delivery could result in requisite drug concentration at the site of action with minimal systemic exposure, consistent with the emerging antibiotics stewardship guidelines. We recognize the limitations of this study: (1) We have used rodents in our in vivo studies and there could be differences between the behaviour of antibiotics-loaded bone cement in rodent and human bones; (2) Significant GLP toxicity studies are required prior to clinical translation; and (3) Future antibiotic development will need to focus on molecules or combinations of them hat can target both resistant and persistent bacterial populations. Integration of such novel antibiotics and combinations into innovative disease-site-tailored delivery devices might transform the landscape of infectious diseases.

Methods

General materials.

All solvents and chemicals were of ACS grade. Starting materials for the chemical synthesis like 1-cyclopropyl-6,7-difluoro-8-methoxy-4-oxo-1,4-dihydroquinoline-3-carboxylic acid, 5-methylfuran-2-carboxylic acid, 5-bromofuran-2-carboxylic acid and 1-(3-chloro-2-hydroxypropyl)-2-methyl-5-nitroimidazole were procured from Labex Corporation, India. Other starting materials and reagents were procured from different sources like Alfa Aesar, MA, USA, Spectrochem Pvt. Ltd,India, Sigma Aldrich, MO, USA and GLR Innovations, India. Thin-layer chromatography was carried out on aluminum sheets pre-coated with silica gel 60 GF 254 (Merck, Darmstadt, Germany) and checked in UV-254 light or developed with phosphomolybdic acid (PMA) or iodine vapors. Anhydrous sodium sulfate was procured from Merck Life Sciences Pvt. Ltd. and used as a drying agent after extractions. Unless mentioned, all the compounds were purified using CombiFlash Rf 200 flash column chromatography system (Teledyne Isco, NE, USA) with silica gel (230 – 400 mesh). ¹H (500 MHz) and ¹³C (125 MHz) NMR of the various intermediates and final compound were recorded on a Jeol 500 MHz NMR spectrometer (Jeol, MA, USA) using deuterated solvents like CDCl₃, d6-dimethyl sulphoxide (DMSO) calibrated to an internal standard TMS (tetramethylsilane) resonances (δ_H 0.0). Chemical shifts (δ) are quoted in parts per million (ppm) and referenced to the residual solvent peak. ESI-MS of the various intermediates and final compound was recorded on a Shimadzu, LC-201EV (Shimadzu, Kyoto,Japan) or a Water CHA, SQD-2 LC-MS (Waters Corp,MA,USA) instrument. Melting points were obtained using a Polmon apparatus. The purity of the final compound was analysed by reverse-phase high-performance liquid chromatography (RP-HPLC) on a Kromasil-C18 column (5 μm, 4.6 × 250 mm, 1.5 ml min⁻¹) coupled to a UV detector operating at 292 nm using an isocratic 70:30 (solvent A:solvent B) composition where solvent A is buffer (NH₄H₂PO₄, pH = 3) and solvent B is acetonitrile. S. aureus ATCC 25923, ATCC 43300, ATCC 6538P, and S. epidermidis ATCC 35984 were procured from LGC Promochem (Bangalore, India); C. acnes MTCC 1951 and E. coli MTCC 1687 were obtained from MTCC (IMTECH, Chandigarh, India). S. aureus CCARM 3505 was procured from CCARM (Seoul, Korea). Studies with 50 different S. aureus strains, VRSA, and VISA were performed at JMI Labs (IA, USA). The human osteoblast cell line Saos-2 was procured from LGC Promochem (Bangalore, India).

Synthesis of antibiotics.

Detailed description of synthesis and characterization of the antibiotics library is included in the supplementary information.

Molecular docking studies.

All the molecular docking calculations were done using the software Glide program (Schrödinger Release 2017-2: Glide, Schrödinger, LLC, New York, NY, 2017). Docking was performed using the SP (Standard Precision Mode) docking protocol.

Preparation of wild type protein structures and mutated protein structures:

The crystal structure of catalytically active heterotetrameric DNA gyrase protein (PDB:5CDQ) having resolution (2.95 Å) was obtained from the RCSB databank. Since crystal structure of S. aureus mutant DNA gyrase complexed with DNA was not available in the databank, point mutations were introduced in the crystal structure through Schrodinger’s protein preparation wizard in Maestro 11 (Schrödinger, LLC, New York, NY) at positions 84 (Ser to Leu) and 88 (Glu to Lys) (numbering as per S. aureus amino acid sequence). The protein was prepared through a multistep process by using the protein preparation wizard in Maestro to obtain an optimized structure. Specifically, we first assigned the bond order in the protein, added hydrogen atoms and considered the water molecules which participated in interactions with Mg²⁺ ion in this step. Then we determined the protonation and tautomer/flip states for the protein part in the system. Finally, we subjected the protein to energy minimization by utilizing the OPLS3 force field with implicit solvation.

Grid generation:

Glide (Grid-based ligand docking with energetics) examines favorable interactions between single or multiple typically small-sized ligand molecules and typically larger-sized target molecules. The binding site of the receptor is represented by a grid that defines the position and size of the active site. For protein structure, a grid box of 24.78 × 28.78 × 24.78 ų with an inner box (10 × 14×10 ų) was centered on the corresponding co-crystallized ligand, moxifloxacin. The receptor grid was specified as an enclosing box at the centroid of the co-crystallized ligand and included the cofactor and substrate binding sites.

Ligand preparation:

The ligands were prepared using the LigPrep module of the Schrodinger suite using the OPLS3 force field for minimization of the structure. The selection of the ionization step also helps to determine the ionization state of the ligand. These ionization states were derived by removing or adding protons on the ligand. All possible conformers, tautomers, and stereoisomers with significant populations for each input structure were generated.

Ligand Docking:

In the initial Glide docking stage, the van der Waals radius scaling factor for the non-polar atoms in the ligand was assigned as 0.9 Å, and the corresponding scaling factor for the target atoms was set to 0.8 Å(< 1) to simulate the flexibility of the receptor. The partial-charge cutoff values were 0.15 and 0.25 for the ligand and the protein receptor, respectively. Finally, the ligand was docked at the binding site using the “Standard Precision (SP)” protocol in Schrodinger. All the molecules were docked into the target protein active sites and final scoring was performed in terms of Glide-score multi-ligand scoring function. Initially, the root mean squares deviation (RMSD) was calculated through self-docking of the co-crystal ligand i. e. moxifloxacin in the active site of the protein. This is the validation technique to justify the specific poses of docking/scoring combinations and a score of the ligand in the protein site. Using the above protocol, we initially docked moxifloxacin, into the crystal structure of DNA gyrase to verify the reproducibility of the binding pose. Further, all other molecules were docked separately into the QBP of the optimized protein. All figures were prepared using PYMOL1.7⁴⁹. The residue numbering of the protein in docking studies was done as per the PDB numbering.

Determination of minimum inhibitory concentrations in vitro.

The MICs of molecules were determined using the micro broth dilution method following the Clinical and Laboratory Standards Institute (CLSI) guidelines, 2012, M100S-S22. Bacterial strains were cultured in Brain Heart Infusion Agar (BHIA) (Himedia) at 35 ± 2°C for 18 – 24 h for aerobic bacterial strains and 36 – 48 h for C. acnes strain under anaerobic conditions. Initial stock solutions of drugs were prepared in DMSO or autoclaved Milli-Q water depending on the drug solubility, and the working stock was made in BHI broth. For MIC determination, 100 μl of sterile Brain Heart Infusion (BHI) broth was added into each well of 96 well plates. Next, 100 μl of drug-containing BHI broth was added to the first well and a serial dilution was performed. For bacterial inoculum, the turbidity of bacterial culture was set to 0.5 McFarland standard (approx. 1.5 × 10⁸ CFU/ml) by adjusting absorbance to 0.1 at 600 nm and further diluted by 100 times with sterilized BHI broth.100 μl of S. aureus suspension was added to each well except for the sterility control wells. The plates were incubated at 35 ± 2°C for 18 – 24 h for aerobes and facultative anaerobes. C. acnes strain was incubated under anaerobic conditions for 36 – 48 h. The assays were performed in triplicates. The MICs of the test compounds were determined by observing the lowest concentration of the test compound that prevented visual bacterial growth.

Emergence of resistance assay.

The in vitro emergence of resistance assay was performed by passaging S. aureus ATCC 25923 and S. aureus ATCC 43300 culture exposed to sub MIC (minimal inhibitory concentration) of the test compounds for 30 cycles as described by Entenza et al.⁵⁰ with some modifications. Initially, the test compound’s MIC was determined at 35 ± 2°C in BHI broth using the micro broth dilution method. For subsequent passages, the bacterial inoculum was prepared from the previous passage in the BHI agar plate, showing visual growth (sub MIC concentration). The serial passage experiment was carried up to 28 subsequent cycles. The concentration ranges of the test compound were modified over the study as and when the compound’s MIC was showing increased value. The assay was done in triplicates in each cycle. MPC of antibiotics were determined against S. aureus strains as per Metzler et al⁵¹.

DNA supercoiling assays.

DNA supercoiling activity was assayed using a cell free assay system. The reactions were carried out in a final volume of 30 μl containing 0.5 μg/ml relaxed pBR322 DNA, and 1 unit of S. aureus DNA gyrase using appropriate buffers as per the manufacturer’s protocol (Inspiralis, Norwich, UK). Effect of the drug (each at 5 different concentrations) on DNA supercoiling activity by the gyrase was checked their action. The reaction mixture was incubated at 37°C for 1 h with the antibiotics and then terminated with STEB (sucrose-Tris-EDTA and bromophenol blue) and extracted with chloroform: isoamyl alcohol (24:1). DNA samples were collected and loaded on 1% agarose gel (without ethidium bromide) and were run at 60 V for 2 h and stained with ethidium bromide (1 μg/ml). The gel showed the positions of relaxed DNA as well as gyrase reaction products (supercoiled DNA). Since quinolones are known to interact with DNA-enzyme complex and inhibit supercoiling (by yielding linear DNA cleavage product), the degree of supercoiling was assessed by checking the intensity of supercoiled DNA band and quantitated using Image J. The data are represented as the percentage of the degree of supercoiling taking the degree of supercoiling in the presence of enzyme without drug as 100% (n=3). Further, to check the effect of antibiotics on DNA cleavage by topoisomerase IV reactions were carried out containing 0.5 μg/ml supercoiled pBR322 DNA, and 1 unit of S. aureus Topo IV enzyme using appropriate buffers (final volume 30 μl) as per the manufacturer’s protocol (Inspiralis, Norwich, UK). The effect of the drug (each at 5 different concentrations) on supercoiled DNA in presence of topoisomerase IV was checked. The reaction mixture was incubated at 37°C for 1 h with the antibacterials followed by incubation at 37°C for 30 min with 1% SDS, digested with proteinase K, and then terminated with STEB followed by extraction with chloroform: isoamyl alcohol (24:1). DNA samples were collected and loaded on 1% agarose gel (without ethidium bromide) and were run at 40 V for 4 h and stained with ethidium bromide (1μg/ml). The gel showed the positions of supercoiled DNA as well as cleavage complex products (linear DNA). Since quinolones are known to interact with DNA-enzyme complex and formed cleavage complex was assessed by checking the intensity of linear DNA band and quantitated using Image J.

Biofilm disruption assay.

S. aureus and S. epidermidis were grown in Brain Heart Infusion Agar (BHIA) at 35±2°C for 24 h. The loop full of bacterial culture was suspended in sterile water and the turbidity adjusted to the absorbance of 0.1 at 600 nm (about 1.5 × 10⁸ cells) and further diluted by 100 times with sterilized BHI broth. 1ml of diluted culture suspension was added into a 12-well plate and plates were further incubated at 35 ± 2°C for 48 h for biofilm formation. The biofilm was carefully washed twice with sterile water to get rid of planktonic cells. Thereafter biofilm was treated with 1 ml of BHI broth suspended with various concentrations of test molecule VCD-077 along with the comparators and further incubated at 35 ± 2°C for another 24 h followed by washing with sterile water twice. The biofilm was resuspended with 1 ml of 1X trypsin EDTA solution (0.25% Trypsin and 0.02% EDTA in Dulbecco’s Phosphate Buffered Saline) then serially diluted, plated on BHI agar, and incubated for 18 – 24 h. Plates were observed for colony forming units (CFU) and log reduction was calculated by comparing bacterial count (CFU/ml) with respect to 48 h growth control.

For confocal microscopy, Staphylococcus culture was grown in BHI agar at 35 ± 2°C for 18 – 24 h. The loop full of bacterial culture was suspended in sterile water and adjusted the turbidity to 0.1 OD at 600 nm in a UV-visual spectrophotometer (about 1.5 × 10⁸) and further diluted by 100 times with sterilized BHI broth. The culture suspension was seeded in 8-well chamber slides and incubated at 35 ± 2°C for 36 – 48 h for biofilm formation. The biofilm was carefully washed twice with sterile water to get rid of planktonic cells. Then biofilm was treated with BHI broth containing various concentrations of test compounds and further incubated at 35 ± 2°C for 18 – 24 h. After treatment, the chamber slides were washed with PBS and stained with SYTO 9 green, and incubated at room temperature for 20 – 30 min in dark. After incubation, the excess stain was aspirated and washed twice gently with sterile water. Further, the biofilm was fixed by adding neutral buffered formalin to each well and incubated for 30 min at room temperature. After fixation, the slide was washed twice with sterile water and observed under the confocal microscope (LSM 880, Carl Zeiss, Oberkochen, Germany). The excitation/emission maxima for SYTO 9 dyes were 480/500 nm. Confocal images were captured in different magnifications.

The biofilm inhibition property of VCD-077 was further checked using scanning electron microscopy (SEM). Culture suspension of S. aureus ATCC 6538P was seeded on coverslip placed in 12-well plates and incubated at 35 ± 2°C for 48 h for biofilm formation. The biofilm was carefully washed twice with sterile water to get rid of planktonic cells. Thereafter, biofilm was treated with test compounds dissolved in BHI broth and further incubated at 35 ± 2°C for 24 h. After treatment, the coverslips were washed with PBS and fixed in 2.5% glutaraldehyde solution. The coverslips were then washed with 0.1 M PBS buffer for two times for 15 min and dehydrated by replacing the buffer with increasing concentrations of ethanol (30%, 50%, 70%, 80%, 90%, 95%, and 100%) for 10 min for each condition. After critical-point drying and gold sputter coating (BU015331-T, Baltec, Switzerland), the samples were examined under the scanning electron microscope (EVO18 Zeiss, Thornwood, USA).

Cell viability assay with human osteoblast Saos-2 cells.

Saos-2 cells were seeded in a 96-well plate (20,000 cells per well), without the test agent and allowed to grow for 12 h. Appropriate volumes of VCD-077 (in DMSO) were added to the wells to achieve final concentrations of 12.5, 25, 50, 100, and 200 μg/ml. Appropriate assay controls including medium control group (medium without cells), negative control group (medium with cells but without including the experimental compound), positive control group (medium with cells), and vehicle control group (DMSO) were included. The experiment was performed in triplicates and data plotted as the viability of cells in percentage compared to that of vehicle control. The IC₅₀ of the drug was calculated using the software Graph Pad Prism 7.

In silico prediction of cardiotoxicity.

The method for predicting the risk of cardiac toxicity is based on the inhibitory effect of the human ether-to-go-go (hERG) gene that encodes the hERG potassium (K⁺) channel, which transmits the rapid delayed rectifier repolarizing current in human cardiac myocytes. We have checked the hERG inhibition property of VCD-077 using Schrödinger’s Qikprop method (Schrödinger Release 2015: QikProp 4.6, Schrödinger, LLC, New York, NY, 2015). The software calculates log IC₅₀ for hERG K⁺-channel blockage based on the molecular structure and descriptor of the molecule. A calculated value of −5 for a drug corresponds to an IC₅₀ of 10,000 nM for hERG binding which is insignificant. A value less than −5 for log(hERG) corresponding to a IC₅₀ less than 10,000 nM is a concern.

Cytochrome P450 induction and inhibition by VCD-077.

The in vitro induction of cytochrome P450 (CYP) 1A2, 2B6, and 3A4 isozymes by VCD-077 was evaluated in human plateable hepatocytes. Primary cryopreserved hepatocytes (procured from Sekisui XenoTech LLC) from one donor were plated and exposed to VCD-077 (10, 30, and 100 μM) for 3 days. A specific positive control inducer was incubated separately in hepatocytes for 3 days and the formation of a selective metabolite of the probe substrate was monitored using LC/MS-MS technique. The induction effect of increasing the concentration of VCD-077 from 10 to 30 and 100 μM on the production of the substrate metabolite was determined. CYP induction was assessed by measuring the activity of CYP1A2, CYP2B6, and CYP3A4. Each condition was performed in a singlet. The formation of metabolites is presented in the form of the fold of induction. The in vitro inhibition of cytochrome P450 (CYP) 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1, and 3A4 isozymes by VCD-077 was evaluated in human liver microsomes. For each isozyme in the CYP inhibition assay, a CYP-specific probe substrate was incubated together with human liver microsomes and cofactors to observe the formation of a selective probe substrate metabolite. The inhibitory effect of increasing the concentration of VCD-077 up to 50 μM on the metabolite production was determined, and the inhibitor concentration required for reducing 50% of the measured enzyme activity was estimated. Standard positive control inhibitors were tested for each isozyme concurrently with VCD-077 in this assay.

Repeat dose toxicity study.

BALB/c mouse (8–10 weeks, Charles River Laboratories) were administered a homogenous suspension of the drug (VCD077) in 0.5% hydroxylpropyl methylcellulose, 0.1% tween 80 by oral gavage. Each mouse received either vehicle (150 uL) or drug (150 mg/kg) every day for 5 days. Mice were euthanized on day 5 after 2 hours of drug or vehicle administration, and after 24h of drug administration. Blood collection was done by cardiac puncture using 23g needle. Whole blood was collected in serum separator tubes (BD365967) and centrifuged to separate the serum. Serum was analysed utilizing standard liver and kidney panels on a Roche COBAS 6000_c501 with ISE. All animal work has been performed following approved institutional animal care guidelines.

Preparation of antibiotic loaded bone cement beads and cylinders.

Typically, 40 gm of Smartset HV powder was homogeneously mixed by geometric mixing with 1 g or 3 g of VCD-077 under sterile conditions to get the uniform mixture as described in manufacturer’s protocol. Then, 20 ml monomeric solution was added into the drug-PMMA mixture and mixing was continued for 30 sec to form a homogeneous soft doughy mixture at room temperature. The above mixture was then allowed to set for 1 to 2 min to obtain the doughy homogenous mixture with the required viscosity of rubbery material. It was further spread either into a silicone mold to obtain small beads of size 4.2 mm × 4.6 mm or large cylindrical specimens with 6 mm in diameter and 12 mm in height. The curing time at room temperature (25°C) for small beads and cylinder is 2–3 min and 8–9 min respectively. The hardened cement beads or cylinders were then pulled out from the mold and stored under the sterile condition at room temperature.

Fourier transform infrared (FT-IR) spectroscopy.

Powdered samples of VCD-077, commercial bone cement (Smartset HV), and a mixture of drug and Smartset HV (2.5% and 7.5%) of approximately (2 – 3 mg) were individually mixed with 100–200 mg of potassium bromide powder (KBr) and finely pulverized and placed into a pellet-forming die. About 10 tons of force was applied to prepare the pellets and such solid rigid discs were placed in the equipment (JASCO FT/IR-4100). The final spectra were obtained after averaging sixteen scans within the range of 400 – 4000 cm⁻¹.

Drug release assay from bone cement beads.

Drug release study was carried out in duplicate or triplicate (n = 2 or 3) with bone cement beads immersed in PBS buffer (pH 7.4). The sample was kept in an incubator rotator at 32°C or 37°C at 35 rpm for the study duration. At each time point (0 h, 2 h, 4 h, 6 h, day 1, day 2, day 7, and day 14), and replaced with an equal volume of fresh PBS buffer to maintain sink condition throughout the study. The measurement of drug (VCD-077) content in each aliquot was determined by reverse-phase high-performance liquid chromatography (RP-HPLC) on a Kromasil-C18 column (5 μm, 4.6 × 150 mm, 1.0 ml/min) coupled to a UV detector operating at 292 nm using an isocratic composition 60:40 (solvent A:solvent B) where solvent A is buffer (NH₄H₂PO₄, pH 3.0) and solvent B is acetonitrile. Alternatively, we also used a Shimadzu UV spectrophotometer at 292 nm. The unknown concentration of the drug was quantified against a known concentration of the reference standard. The cumulative release of the drug (in μg/ml) was plotted over time. Drug release rates were calculated from the linear regression⁵¹ of straight lines obtained by plotting cumulative drug concentration as a function of the square root of time, K (rate of drug release) = ΔC/Δt, where ΔC is the change in concentration and Δt is the time interval as per Guy & Hadgraft⁴³.

Scanning electron microscopy.

Surface morphological characteristics of the PMMA alone (Smartset HV) and drug-loaded bone cement beads were studied using a scanning electron microscope (EVO 18, Carl Zeiss, Oberkochen, Germany). The bead samples were gold coated using an agar sputter coater before SEM analysis.

Compressive strength analysis of bone cement cylinders.

Compressive strength of cylindrical specimens (6 mm × 12 mm) was measured by Instron, model 3345 at room temperature 24°C and 63% humidity as per ASTM-F451 protocol³⁸. The cylindrical samples were uniaxially compressed at a speed of 20 mm/min with a maximum load of 5000 N. Compressive strength is calculated from the ratio of peak load (Newton) to cross-sectional area (27.70 – 27.98 square mm) of the sample. The results were expressed in megapascal (MPa) and repeated 8 times for each set of samples with Smartset HV and Smartset HV with 3 g VCD-077. Youngs Modulus (expressed in MPa) was calculated from the slope of the initial linear portion of the stress-strain curve for each sample and averaged out to obtain the mean value.

Sequential zone of inhibition assay using antibiotic loaded PMMA beads.

BHA plates were spread with 100 μl of bacterial suspension (0.5 McFarland equivalent). Drug-loaded beads in triplicates were placed above the spread plates. Plates are incubated at 35 ± 2°C for 24 h. Thereafter the ZOI was measured and further, the bead was transferred to fresh BHA plates spread with bacterial suspension and kept for incubation. This process was repeated till 14 days.

Quantification of VCD-077 in bone and plasma, in vivo.

In vivo release of VCD-077 into the bone from drug-loaded beads was determined in healthy male Sprague Dawley rats. Animals were anesthetized and the proximal third of the tibia in each rat was surgically exposed. A ~3.2 mm hole was drilled into the medullary cavity. PMMA bead (3 mm size) loaded with VCD-077 (in 3:40 drug to cement ratio, ~13.5mg/kg for 100g rat) was placed immediately into the drilled hole, followed by the closing of the hole using sterile bone wax and suturing the surgical opening. Post implantation, animals were sacrificed to obtain bone samples. Blood was collected from the retro-orbital plexus to check the release kinetics in the plasma as well. The plasma samples were analysed at 0, 2, 6, 12, and 24 h, whereas the bone samples were extracted at 24 h after the bead implantation and compared with bone samples at 0 h as our release kinetics study had shown that the drug released from bone cement exhibited highest concentration in the first 24 hour followed by sustained release. Based on this logic we have evaluated drug concentration in the bone at 24h post-implantation. The bone tissue was homogenized by adding weight equivalent deionized water and beads and homogenized in Bullet Blender tissue homogenizer (Next Advance Inc., NY, USA). The plasma/homogenate (50μl) was spiked with 10μl of internal standard, and the drug was extracted from plasma and bone homogenate using acetonitrile (250 μl). The samples were centrifuged at 14000 rpm for 5 min at 4°C, and 150μl of supernatant was isolated for analysis. VCD-077 was quantified using LC-MS/MS (API 4000, Sciex, Framingham, MA, USA) with electro spray ionization (ESI) and multiple reaction monitoring in positive ionization mode. Isocratic elution was used comprising of mobile phase (A) (0.1% formic acid in 5mM ammonium formate solution) and (B) (0.1% formic acid in acetonitrile) in 40:60 v/v ratio, and run on a Phenomenex Luna 5μm HILIC 200A LC column (150 mm × 4.6 mm). Verapamil was used as an internal standard. Percentage recovery as compared with spiked samples was greater than 80% for bone homogenate and > 70% for plasma. Animal studies were performed at TheraIndx Lifesciences Pvt. Ltd., (Bangalore, India) approved by TheraIndx Lifesciences Animal Studies Ethics Committee in compliance with the CPCSEA guidelines for animal care.

Prevention of osteomyelitis in vivo.

Animal studies were performed at TheraIndx Lifesciences Pvt. Ltd., (Bangalore, India) approved by TheraIndx Lifesciences Animal Studies Ethics Committee in compliance with the CPCSEA guidelines for animal care (similar to AAALAC guidelines) as reported earlier^44,45 with some modifications. The animal studies had been performed by certified personnel after taking Institute ethical clearances. Male Sprague Dawley rats (6–8 weeks old, ~100g) were procured from Vivo Biotech Ltd (Hyderabad, India). Animals were anesthetized with 3% – 5% isoflurane in an oxygen flow set at approximately 0.5 l/min and anaesthesia was maintained by exposing the animals to 1% – 2% isoflurane in an oxygen flow set at approximately 0.5 l/min. The proximal 3rd of the tibia was surgically exposed. Osteomyelitis was induced in rats in the upper metaphysis of the tibia by injecting arachidonic acid (250 ng/tibia) through a ~3.2 mm hole in the proximal third of the tibia, followed by subsequent injection of 25 μl of bacterial inoculum (2 × 10⁷ CFU/ml). PMMA beads (SmartSet HV PMMA loaded with drugs) of 3 mm size (18.7–19.2 mg) were prepared in sterile conditions. A single PMMA bead loaded with VCD-077 (in drug to cement ratio of 3:40 wt/wt, i.e. ~13.5mg/kg for a 100g rat) was placed into the hole in the tibia after inoculation. A commercially available gentamicin-loaded PMMA (Palaco R+G preformulated as 0.5:40 wt/wt ratio, ~2.5 mg/kg for 100g rat) bead was used as a comparator. Animals receiving no treatment and those with PMMA without drug were used as infection control and vehicle control respectively.. The infection site was sutured and all animals were treated with ketoprofen (5 mg/kg) to manage pain. Animals were sacrificed on days 0, 7, 14, 21, and 28 and bone tissue samples were collected, and homogenized as described above and tested for bacterial load using a colony forming assay. Bone tissues were harvested and evaluated for gross pathology and histopathology on the 28th day. Histopathology analysis was done at Dabur Research Foundation, India.

In vivo efficacy in an established osteomyelitis rat model.

Experimental osteomyelitis was established in the proximal left tibia in Sprague Dawley rats as previously described in the earlier experiment using S. epidermidis ATCC 35984. Animals were anesthetized with 3% – 5% isoflurane in an oxygen flow set at approximately 0.5 l/min and the proximal third of the left tibia surgically exposed followed by arachidonic acid (250 ng/tibia) injection through a hole drilled into the medullary cavity. Thereafter, 25 μl of bacterial inoculum (2 × 10⁷ CFU/ml) was injected into the hole, following which the skin and muscle were sutured separately using a non-absorbable sterile surgical ligature. After twenty-one days, the animals underwent limited surgical debridement procedure at the infected site by creating a ~3.2 mm hole through the cortical bone to the medullary cavity, and treatment was initiated. A 3 mm SmartSet HV PMMA bead loaded with VCD-077 (in a 3:40 drug to cement ratio, ~13.5mg/Kg for a 100g rat) prepared in sterile conditions was placed into the hole of the rat tibia. Positive control animals received SmartSet HV PMMA loaded with gentamicin (commercial Palacos R+G, preformulated as 0.5:40 drug-polymer ratio, ~2.5 mg/kg for 100g rat) and rifampicin (formulated in 3:40 drug to polymer ratio, ~13.5mg/Kg for a 100g rat). Animals receiving no treatment were used as infection control. Animals were routinely monitored for pain and distress and were given ketoprofen (5 mg/kg) for pain management. Animals were sacrificed on days 0, 5, and 10, and the bone samples were homogenized as described above and tested for bacterial load using a colony forming assay. Gross pathological examination of the infected site was performed on Day 21 post-infection and on Day 10 post-treatment to evaluate the wound healing. Animal studies were performed at TheraIndx Lifesciences Pvt. Ltd., (Bangalore, India) approved by TheraIndx Lifesciences Animal Studies Ethics Committee in compliance with the CPCSEA guidelines for animal care.

Statistical analyses.

All the results are expressed as the means ± SD (or median with interquartile ranges) for the number of separate experiments indicated in each case. The student’s t-test was used for comparison between two groups while multiple comparisons were tested using ANOVA with appropriate post hoc tests.

Reporting summary.

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Extended Data

Extended Data Fig. 1. Physico-chemical characterization of VCD-077 impregnated PMMA beads.

(a-c) FT-IR spectrum of different groups (a) VCD-077, (b) PMMA, (c) VCD-077 impregnated PMMA at (40:1) with VCD-077 peaks 1 (3352.75 cm⁻¹), 2 (3114.27 cm⁻¹), 3 (1643.82 cm⁻¹), 4 (1615.48 cm⁻¹), 5 (1595.79 cm⁻¹), 6 (1550.68 cm⁻¹), 7 (1352.31 cm⁻¹), 8 (1313.02 cm⁻¹). (d) Release of VCD-077 from Smartset HV^® (PMMA) bead at different drug:polymer ratio (1:40, 2:40 and 3:40) in pH 7.4 buffer. Data is represented as mean ± SD (n=3). (e) Release of VCD-077 from Smartset HV^® (PMMA) bead at different particle size from drug:polymer ratio (1:40) or (f) at different temperatures, in pH 7.4 buffer. Data is represented as mean ± SD (n=3).

Data availability.

The main data supporting the results in this study are available within the paper and its Supplementary Information. Source data for the figures is provided with this paper. The raw and analysed datasets generated during the study are available for research purposes from the corresponding authors on reasonable request.