Poly(2-oxazoline) Micelles for Co-Delivery of Paclitaxel and Metronidazole Benzoate for Dual Chemotherapeutic and Antibacterial Targeting in the Tumor Microenvironment

Abstract

Tumor-resident pathogenic bacteria can promote cancer progression and reduce chemotherapy efficacy, yet strategies to simultaneously target both tumor cells and intratumoral microbes remain limited. Here, we report a poly(2-oxazoline) micelle (POx) platform co-encapsulating paclitaxel (PTX) and metronidazole benzoate (MB) to achieve concurrent delivery of anticancer and antibacterial agents. The POx/PTX/MB micelles produced monodisperse populations with high drug loading efficiency and capacity and remained stable in physiological conditions. In vitro, the co-loaded formulation retained cytotoxic activity against two triple-negative breast cancer (TNBC) cell lines and bactericidal activity against Fusobacterium nucleatum. POx/PTX/MB micelles were well-tolerated at pharmacologically relevant doses in a murine model. This work provides a feasible strategy to integrate antimicrobial therapy with chemotherapy, highlighting the potential of POx micelles as a versatile platform for targeting both cancer cells and tumor-associated pathogens. These findings support further development of combination chemotherapeutic–antimicrobial strategies for tumors harboring pathogenic bacteria.

Introduction

The tumor microenvironment (TME) is increasingly recognized as a complex ecosystem where cancer cells and resident microorganisms each play a role in shaping disease progression and therapy response^1,2. Tumors across human cancers harbor distinct microbial communities that are now understood to be an integral component of the TME³. Pioneering work using clinical tissue profiling, molecular detection approaches, and functional studies has demonstrated that bacteria can reside intracellularly within both cancer cells and immune infiltrates, and that their composition, abundance, and localization influence cancer progression and therapeutic outcomes⁴. In breast cancer (BC), these tumor-resident bacteria have been linked to key hallmarks of malignancy, including modulation of local immune networks and inflammatory signaling, alterations in metabolic pathways, and correlations with clinical features and prognostic signatures^5–7. Collectively, these findings support the concept of oncobiosis: a dysbiotic tumor-associated microbiome that promotes invasion, metastasis, and therapy resistance through bacterial-derived metabolites, immune modulation, and direct microbe-host signaling. Despite these findings, most cancer treatment strategies focus solely on targeting tumor cells, largely neglecting the contribution of intratumoral microbes.

Tumor microbiome profiling is still in its infancy, and although accumulating evidence indicates that BC-associated microbial communities differ from those of healthy tissues, identifying specific oncogenic microbes using clinical human samples remains challenging. These studies are limited by numerous confounding factors, including patient heterogeneity, differences in diet, medication use (particularly antibiotics), geographic and environmental influences, tumor stage, and treatment history. The difficulty in controlling these variables has likely contributed to the variation in bacterial species identified as differentially abundant within tumors across studies. Nevertheless, multiple studies have implicated the Gram-negative, obligate anaerobe Fusobacterium nucleatum (F. nucleatum) in the progression of various malignancies, with some reports detecting increased abundance of this bacterium in up to one-third of BC patients^8–10. Furthermore, systematic reviews and meta-analyses have identified F. nucleatum as one of the more consistently detected bacterial species associated with BC^11,12. Experimental work has begun to define potential mechanisms by which F. nucleatum may colonize breast tumors and influence tumor progression, including modulation of inflammatory signaling, immune responses, and cancer cell behavior^13–18. Collectively, this provides sufficient justification for using F. nucleatum as a model for the development of therapeutic strategies targeting the BC tumor microbiome. To investigate the complex interactions between F. nucleatum and the TME in response to treatment, we choose to develop formulations for the therapy-resistant triple-negative BC (TNBC) subtype. Due to its lack of expression of ER, PR, and HER2, TNBC is ineligible for BC therapies targeting these markers, limiting patient treatment options. Meanwhile, TNBC has a higher risk of recurrence, earlier metastasis, and poorer overall survival than other BC subtypes. Together, this provides strong motivation to develop novel therapies which are efficacious in TNBC models.

Polymeric micelles, nanoscale self-assembled structures formed by amphiphilic block copolymers, have emerged as versatile carriers capable of solubilizing and delivering poorly water-soluble drugs with high loading capacity and favorable pharmacokinetic profiles¹⁹. When compared with other clinical and preclinical formulation materials, poly(2-oxazoline) (POx) based micelles exhibit high drug loading capacity and stability under physiological conditions, enabling sustained systemic exposure and enhanced tumor accumulation of therapeutic agents. In rodent and non-human primate models, high-capacity POx micelle formulations were used to deliver paclitaxel (PTX), a microtubule-stabilizing taxane chemotherapeutic widely used to treat aggressive BC²⁰. POx/PTX achieved bioavailability comparable to that of clinical formulations, highlighting its translational potential²¹. Furthermore, POx micelles have been successfully employed to co-encapsulate combinations of chemotherapeutic or immunomodulatory agents, including PTX paired with an inhibitor of colony-stimulating factor 1 receptor (CSF1R) in models of TNBC or the combination of PTX with cisplatin in both breast and ovarian cancer models, resulting in improved anti-tumor efficacy and survival outcomes^22,23. Co-delivery of chemotherapeutic and antimicrobial agents within a single POx micelle platform may simultaneously target malignant cells and tumor-resident pathogens. Bacterial elimination could increase exposure to pathogen-associated molecular patterns (PAMPs), while PTX induces immunogenic cell death and the release of damage-associated molecular patterns (DAMPs), providing complementary mechanisms that may promote stronger antitumor immune responses^24,25.

In this study, we report on the initial formulation and characterization of a POx micelle system, its in vitro activity assessment using TNBC and F. nucleatum models, and an in vivo safety screening. The micelles co-encapsulate PTX and metronidazole benzoate (MB), a nitroimidazole prodrug selectively activated through metabolization by anaerobic bacteria. MB is a narrow-spectrum agent capable of selectively eliminating intratumoral F. nucleatum with a low risk of resistance development^26,27. Co-delivery of these agents within a single micellar formulation demonstrates a framework for combined tumor-directed cytotoxic and targeted antimicrobial activity. The micelles exhibit high drug-loading efficiency, maintain stability under physiological conditions, and show no observable systemic toxicity in a murine model. Together, these results provide initial evidence supporting the feasibility of integrating antimicrobial agents with chemotherapy within a POx micelle platform and motivate further investigation of this approach in tumors associated with pathogenic bacteria.

Methods

Polymer synthesis

Amphiphilic triblock copolymer p[MeOx₃₄-b-BuOx₂₇-b-MeOx₄₂] was synthesized by first distilling 2-methyl-2-oxazoline (MeOx; Sigma-Aldrich 137448) and 2-n-butyl-2-oxazoline (BuOx; Sigma-Aldrich 799637) monomer on a Schlenk line under constant argon flow, which were confirmed to have a water content below 200 ppm (C20S Coulometric Karl Fisher titrator; Mettler Toledo). Then, anhydrous acetonitrile was added to a Schlenk flask, which had been dried at 120°C and allowed to cool to room temperature under an argon atmosphere. This was followed by the addition of methyl triflate initiator and MeOx to form the first block. The Schlenk flask was sealed, then the mixture was heated to 80°C and allowed to react overnight. Reaction progress was monitored by ¹HNMR. After full monomer conversion and the desired degree of polymerization was achieved, BuOx and MeOx monomer were sequentially added to form the remaining two blocks of the copolymer. Finally, the reaction was terminated with excess piperidine. To purify, the reaction mixture was dissolved in a 50/50 methanol/chloroform mixture and precipitated in ice-cold ether, followed by three days of dialysis in water. The polymer block lengths and polydispersity were characterized using ¹HNMR and GPC and then labeled with the batch name P2-JR-13. This data is publicly available at https://doi.org/10.15139/S3/AP98GQ.

HPLC analysis

All high-performance liquid chromatography (HPLC) analyses were conducted on an Agilent Technologies 1200 series using a C8 column (Supelco S19170). An isocratic method ran for 15 minutes with the column heated to 40°C using a 40/60 HPLC grade water/acetonitrile + 0.1% trifluoroacetic acid mobile phase. Drug concentrations were determined by measuring the AUC of their eluted peaks and comparing them to a calibration curve.

Micelle preparation and characterization

POx micelles were prepared using a thin-film hydration method. Briefly, polymer, PTX (MedKoo Biosciences 100690), and MB (AmBeed A137827) were first dissolved separately in ethanol to produce 10 mg/mL stock solutions. For in vitro studies, micelles were prepared using a small-batch method (≤ 0.5 mL total volume). Ethanol stock solutions of POx, PTX, and MB were combined in a 1.5 mL microcentrifuge tube at a determined mass ratio, described as each component’s concentration in mg/mL (POx/PTX/MB). The solvent was evaporated at 65°C under a gentle nitrogen stream to form a uniform thin film on the tube walls. The resulting film was hydrated with DPBS (Gibco 14190–144) to achieve a final POx concentration of 10 mg/mL. The sample was maintained at 65°C and intermittently agitated to facilitate complete film detachment and spontaneous micelle self-assembly.

For in vivo studies, micelles were prepared using a large-batch method. Ethanol stock solutions of POx, PTX, and MB were combined in a round-bottom flask at determined ratios. Ethanol was removed using rotary evaporation at 65°C under reduced pressure to form a dry thin film on the flask walls. The film was hydrated with DPBS to a final POx concentration of 40 mg/mL. Hydration was performed at 65°C with rotation to promote complete film dissolution and micelle formation.

For both preparation methods, micelle hydrodynamic diameter (Z-average) and polydispersity index (PDI) were measured by dynamic light scattering (DLS) (Zetasizer Nano-SZ; Malvern Panalytical). To quantify the micelle drug content, a 10 μL aliquot of the micelle solution was removed and diluted in 100% acetonitrile for measurement on HPLC (this concentration represents the drug added to the formulation). The remaining micelle solution was incubated at 4 °C for 30 minutes, followed by centrifugation at 10,000 × g for 10 minutes at 4°C to pellet any unencapsulated material. A 10 μL aliquot of the supernatant was then collected and diluted in 100% acetonitrile and measured on HPLC (this concentration represents the portion of drug fully encapsulated within the formulation). The measured values were used to quantify drug concentration. This was multiplied by the formulation volume to determine drug mass.

The drug mass values were then evaluated for loading efficiency (LE%) and loading capacity (LC%). LE% represents the fraction of the initially added drug that is successfully encapsulated within the micelles after formulation, reflecting the efficiency of drug incorporation (Equation 1). LC% represents the proportion of drug relative to the total mass of the micelle formulation, indicating how much drug is carried per unit mass of POx carrier (Equation 2). This data is publicly available at https://doi.org/10.15139/S3/KZP1CV and https://doi.org/10.15139/S3/CQLQQW.

$$LE\%=\frac{M_{\mathit{\text{drug encapsulated}}}}{M_{\mathit{\text{drug added}}}}\times 100\%$$ (1)

$$LC\%=\frac{M_{\mathit{\text{drug encapsulated}}}}{M_{\mathit{\text{drug encapsulated}}}+M_{\mathit{\text{excipient}}}}\times 100\%$$ (2)

For storage, micelle formulations were aliquoted, flash-frozen, and lyophilized (FreeZone Triad; Labconco). The resulting dry powders were stored under vacuum. For reconstitution, ultrapure water at room temperature was added to each vial at a volume equivalent to the original pre-lyophilization aliquot volume. Formulations were then filtered through a sterile 0.2 μm filter into sterile 1.5 mL tubes.

Micelle stability study

To evaluate micelle stability under storage and physiologically relevant conditions, formulations were prepared in either DPBS or DPBS with 10% fetal bovine serum (FBS; GeminiBio S12450) to achieve a final POx concentration of 9 mg/mL. Samples were sealed with airtight caps and wrapped in parafilm to prevent evaporation over the course of the study. Samples intended to model storage conditions were incubated at 4°C, while those intended to model physiological conditions were incubated at 37°C. All samples were equilibrated to room temperature and resuspended with gentle agitation prior to measurement. Z-average and PDI were measured by DLS at 48-hour intervals to monitor changes in particle size distribution over time. This data is publicly available at https://doi.org/10.15139/S3/ADXHMT.

Drug release kinetics study

Drug release kinetics were evaluated using a dialysis membrane-based release assay. Dialysis devices with a 3.5K molecular weight cut-off (Thermo Scientific 69552) were pre-soaked in deionized water for 10 minutes prior to use. A stock solution of micelles was prepared to a concentration of 100 μg/mL PTX in saline. The release medium consisted of 5% FBS in DPBS, pre-warmed to 37°C. For the release study, 100 μL of micelle stock solution was added to each dialysis device (n = 4 replicates per time point). The devices were immediately placed in floating tube racks and immersed in 50 mL release medium in 500 mL beakers. The beakers were maintained at 37°C with gentle agitation on an orbital shaker. Zero-hour samples were collected directly from the stock solution at the start of the incubation period. At predetermined time points, dialysis devices were removed from the release medium, and the entire contents were collected in 1.5 mL tubes. These samples were stored at −20°C until the end of the incubation period.

Collected samples were then frozen in liquid nitrogen, freeze-dried overnight, and reconstituted in 100 μL 100% acetonitrile. After equilibration at 4°C for 30 minutes, samples were centrifuged at 10,000 × g for 8 minutes at 4°C to remove any precipitated proteins. 70 μL from the supernatants were transferred to HPLC vials for analysis. This data is publicly available at https://doi.org/10.15139/S3/QVCZA2.

In vitro cytotoxicity study

Two murine TNBC cell lines, 4T1 and T11-APOBEC were obtained from cryopreserved stock passages maintained in the Kabanov Laboratory. 4T1 cells were purchased from ATCC (CRL-2539), and T11-APOBEC cells were produced by transfecting T11 parent cells to allow overexpression of the Apobec3 gene, and were provided in-kind by the Charles Perou lab²⁸. 4T1 cells were cultured in RPMI (Gibco 11875–093) media supplemented with 10% FBS and 1% antibiotic solution (100 U/mL penicillin and 100 μg/mL streptomycin; Gibco 15140–122). T11 cells were cultured in RPMI media supplemented with 5% FBS and 1% antibiotic solution. All cells were maintained in a humidified incubator at 37°C with 5% CO₂.

Cells were seeded into flat-bottom tissue culture-treated 96-well plates at a density of 3,000 cells per well and allowed to adhere overnight. The cells were then treated for 72 h with POx/PTX/MB micelles (10/4/1.5 mg/mL) or PTX/MB solubilized in DMSO (4/1.5 mg/mL). Treatment concentrations were made with 10-fold serial dilutions ranging from 20 to 2 × 10⁻⁷ μg/mL, reported in terms of the PTX concentration (n = 18 replicates per concentration). Following the treatment incubation period, 10 μL of Cell Counting Kit-8 (CCK-8; MedChemExpress HY-K0301) reagent was added to each well, and the plates were incubated at 37°C for 2 h. Then, absorbance was measured at 450 nm using a microplate reader (SpectraMax M5; Molecular Devices; SoftMax Pro 7.0 software). Cell viability was expressed as a percentage relative to untreated control cells. The half-maximal inhibitory concentration (IC₅₀) values were calculated by nonlinear regression analysis using a four-parameter dose–response model using GraphPad Prism 11.0.2. Each concentration was tested with 18 technical replicates. This data is publicly available at https://doi.org/10.15139/S3/KSEXPV.

In vitro antimicrobial studies

Fusobacterium nucleatum (ATCC 25586; Microbiologics 89503–582) cultures were inoculated following manufacturer instructions and subcultured once prior to cryopreservation in glycerol stocks, which were stored at −80°C for subsequent experiments. For all antimicrobial studies, bacterial handling, inoculation, and incubation procedures were performed within an anaerobic chamber (Bactron 660; Sheldon Manufacturing). Bacteria were expanded to early log-phase as liquid cultures in sterile pre-reduced Columbia broth (Difco 294420) before all experiments and then plated on sterile pre-reduced Brain Heart Infusion agar (Difco 241830). Bacteria inoculum concentrations, reported as colony-forming units/mL (CFUs/mL) were determined by taking OD600 measurements using optical glass cuvettes and a spectrophotometer (Mettler Toledo; Easy VIS), then finding the corresponding CFU/mL by plating aliquots onto agar and counting colonies after 24 h of growth.

The minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC) of POx vehicle, POx/PTX, and POx/PTX/MB were determined using the broth microdilution method adapted from Clinical and Laboratory Standards Institute (CLSI) guidelines. To determine MIC, serial dilutions of each formulation were prepared in Columbia broth in sterile 96-well microplates. Wells with treatments were inoculated with bacterial suspension to achieve a final inoculum of approximately 2.5 × 10⁵ CFU/mL (n = 6 replicates per concentration). Untreated bacteria and sterile broth served as positive and negative controls, respectively. Following anaerobic incubation at 37°C for 18 h, wells were gently resuspended by pipetting before plates were placed atop a dark, non-reflective background under consistent overhead illumination to assess turbidity. The MIC was defined as the lowest concentration producing no detectable increase in turbidity compared with the sterile broth control.

The MBC was determined immediately following the MIC assessment. First, 20 μL aliquots from all treatment concentrations exhibiting no detectable increase in turbidity were plated onto drug-free agar and incubated at 37°C for 24 h (n = 3 replicates per concentration). Colonies were counted, and the MBC was defined as the lowest concentration resulting in a ≥99.9% (3-log10) reduction in viable bacteria relative to the initial inoculum.

In vivo safety study

Female, 8-week-old BALB/c mice (Jackson Laboratory 000651) received 100 μL of sterile filtered POx micelles in saline, dosed at 75 mg/kg PTX and a variable MB dose (9, 18, or 37.5 mg/kg), administered either intravenously (IV) or by oral gavage (OG), once every four days for a total of four doses (q4dx4). Mice were monitored for weight loss, signs of discomfort or dehydration, and peripheral neuropathy. Three days after their last dose, the mice were sacrificed, and whole blood was collected via cardiac puncture. A 40 μL sample was aliquoted in EDTA anti-coagulation tubes (Microvette 20.1341.102) for complete blood count (CBC) analysis. This data is publicly available at https://doi.org/10.15139/S3/TJJWIR.

Results

Preparation and characterization of drug-loaded polymeric micelles

Polymeric micelles loaded with PTX and MB were prepared using a standard thin-film hydration method (Figure 1)²⁹. For multidrug-loaded micelles, the PTX/MB mass ratio was varied with the POx concentration held constant at 10 mg/mL. Drug incorporation into the micelles was monitored by the absence of visible precipitate and further validated by measuring the LE%. As a result, multidrug polymer micelles with various POx/PTX/MB mass ratios were obtained (Supplemental Table S1), with the highest PTX/MB ratio and loading capacity achieved at 10/4/2 mg/mL of POx/PTX/MB at 34.7% LC and 2:1 drug ratio. These micelles form a narrowly distributed nanoparticle population with a hydrodynamic diameter of 30.80 nm and a PDI of 0.061.

Figure 1.

Graphical schematic of experimental workflow. Upper left: Poly(2-oxazoline) micelle structure. Lower left: POx micelle synthesis. Upper right: In vitro evaluation of cytotoxicity (IC₅₀) and antimicrobial activity (MIC/MBC). Lower right: In vivo safety assessment.

The formulations which could produce monodisperse populations immediately after preparation were then subjected to a longitudinal stability evaluation (Supplemental Tables S2–4). Micelles were kept under storage (4 °C) and physiological (37 °C) conditions in DPBS or a solution buffered to have physiological blood pH, tonicity, and supplemented with 10% FBS, with Z-average and PDI monitored by DLS over time to assess changes in particle size distribution (Supplemental Tables S2–4). Stability was regarded as maintaining a monodisperse population with a PDI below 0.200. Our results indicate stable micelles are dependent on the balance between the PTX/MB ratio and the total drug loading. Micelles encapsulating MB alone were unstable and had very poor loading efficiency (Figure 2A). The stable formulations with the highest drug loading capacity were 10/3/1 and 10/4/1.5, with 29.8% and 33.0% LC, respectively (Figure 2B–C). Higher loading micelles became unstable over time, such as the 10/4/2 formulation (Figure 2D). Attempts to load micelles at PTX/MB ratios higher than this resulted in reductions in LE% (Supplemental Table S1).

Figure 2.

Polymer micelle formulation stability. Micelle formulations at POx/PTX/MB mass ratios of (A) 10/0/1, (B) 10/3/1, (C) 10/4/1.5, and (D) 10/4/2. Formulations were diluted 9:1 in DPBS or physiological buffer for a final concentration of 9 mg/mL POx. Data are presented as mean ± SEM (n = 3 per group).

It is notable that MB alone is poorly solubilized in this system, while co-loading MB with PTX markedly improved both its solubility and micellar stability (Supplemental Table S1). This solubility-rescue effect has been previously reported for other compounds that could not be incorporated into POx micelles on their own, but were successfully co-solubilized in the presence of PTX, rendering it an effective mixing partner²³.

Drug cargo in POx/PTX/MB micelles produce differential release kinetics

A drug release assay was performed using 10/4 mg/mL POx/PTX and 10/4/1 and 10/4/2 mg/mL POx/PTX/MB micelles under sink conditions. The co-encapsulated agents exhibit distinct release profiles, with MB releasing rapidly within the first 4 hours and approximately 96% of the drug released by 12 hours, while PTX is released more slowly, only reaching 92–96% released after 48 hours (Figure 3). The release profile of PTX is conserved in both single- and dual-drug loaded micelles.

Figure 3.

Drug release kinetics of POx/PTX/MB micelles were assessed using a dialysis-based assay in 5% FBS-containing DPBS at 37°C with gentle agitation. Micelle formulations at POx/PTX/MB mass ratios of (A) 10/4, (B) 10/4/1, and (C) 10/4/2 were evaluated at an initial PTX concentration of 100 μg/mL. At designated time points, samples were collected, processed, and quantified by HPLC. Data are presented as mean ± SEM (n = 4 per time point).

POx/PTX/MB micelles are effective cytotoxic and antimicrobial agents against breast cancer and F. nucleatum

The in vitro cytotoxic activity of the 10/4/1.5 POx/PTX/MB micelles was evaluated in two TNBC cell lines, 4T1 and T11-APOBEC (Figure 4). 4T1 is a well-established, highly metastatic basal-like TNBC model, whereas T11-APOBEC is a genetically engineered claudin-low model characterized by Apobec3-driven mutagenesis and increased genomic heterogeneity. Evaluating both models allowed assessment of the formulation across biologically distinct TNBC subtypes, increasing confidence that the observed effects were not limited to a single cell line²⁸. To determine whether the presence of POx impacted cytotoxicity, treatment with 10/4/1.5 POx/PTX/MB micelles was compared with 4/1.5 PTX/MB solubilized on DMSO, a standard solvent used in cytotoxicity studies due to its miscibility in both organic and aqueous phases allowing the solubilization of hydrophobic moieties. IC₅₀ values for both PTX and MB were determined by nonlinear regression analysis using a four-parameter dose–response model, and both the PTX and MB concentrations are reported. In 4T1 cells, treatment with POx/PTX/MB micelles resulted in IC₅₀ values of 10.95 nM for PTX and 20.50 nM for MB (Figure 4A), whereas treatment with the PTX/MB DMSO solution yielded IC₅₀ values of 2.92 nM and 5.46 nM for PTX and MB, respectively (Figure 4B). In T11-APOBEC cells, the IC₅₀ values following treatment with POx/PTX/MB micelles were 9.59 nM for PTX and 17.94 nM for MB (Figure 4C). Treatment of T11-APOBEC cells with the PTX/MB DMSO solution resulted in IC₅₀ values of 2.79 nM for PTX and 5.23 nM for MB, respectively (Figure 4D). These values are consistent with previously reported IC₅₀ values for POx/PTX²². No statistically significant differences in cytotoxicity were observed between the POx/PTX/MB micelles and the PTX/MB DMSO solution, nor between the 4T1 and T11-APOBEC cell lines (Figure 4E). These results indicate that the POx carrier did not measurably affect the cytotoxic activity of the PTX/MB combination in this assay and both cell lines displayed similar responses to treatment.

Figure 4.

In vitro cytotoxicity of POx/PTX/MB against 4T1 (A) and T11-APOBEC (B) cell lines, and of PTX/MB solubilized in DMSO against 4T1 (C) and T11-APOBEC (D) cell lines. Data are presented as mean ± SEM (n = 18 per concentration per condition). (E) PTX and MB IC₅₀ values from all conditions are summarized. (F) MIC and MBC of POx/PTX/MB against F. nucleatum, reported as MB concentration. IC₅₀ values for each drug component were evaluated using two-way ANOVA with multiple comparisons (ns > 0.05 * ≤ 0.05, ** ≤ 0.01, *** ≤ 0.001, **** ≤ 0.0001).

The antimicrobial activity of the POx/PTX/MB micelles was evaluated against F. nucleatum (Figure 4F). POx vehicle, POx/PTX, and POx/PTX/MB were incubated with F. nucleatum for 18 hours at 10/0/0, 10/4/0, and 10/4/1.5 mass ratios, respectively. POx vehicle and POx/PTX produced no growth inhibition at concentrations up to 54 μg/mL and 135 μg/mL, respectively. The MIC for POx/PTX/MB was found to be 5.0 μg/mL while the MBC was 6.3 μg/mL. Together, these results indicate that MB remained active after solubilization in the POx carrier and that it is the primary antimicrobial agent within the formulation.

Drug combination micelles are tolerable in mouse models

POx/PTX micelles have been studied extensively for their safety in both rodent and non-human primate models²¹. A review of the published drug interaction literature, including pharmacokinetic reviews, consensus recommendations, and retrospective clinical evaluations, did not identify any reported interactions between MB and PTX^30–34. However, the combination of MB with PTX into a single micelle may lead to increased toxicity. To further evaluate POx/PTX/MB micelle systems as a potential therapeutic modality, we conducted a safety screening in healthy BALB/c mice (n=4 per group) via two administration routes at three mass ratios of varying MB: 10/4/0.5, 10/4/1, and 10/4/2. Each dose level was administered by IV injection and oral gavage, once every four days for a total of four doses (q4dx4). All mice received 75 mg/kg PTX, which we have previously established as safe, and the MB dose was escalated stepwise from 9 to 18 to 37.5 mg/kg^21,22. The mice were monitored for weight loss and signs of peripheral neuropathy, discomfort, or dehydration (Supplemental Figure S1). Over the course of the study, all mice maintained a body condition score of 3 and a peripheral neuropathy score of 0. No signs of toxicity were seen after four doses of 9 mg/mL and 18 mg/mL MB. No signs of toxicity were seen in mice after receiving two doses of 37.5 mg/mL MB; however, treatment was halted due to the colloidal instability of the micelles at this concentration. This formulation was successfully produced and lyophilized, but precipitated within an hour of reconstitution, limiting its usefulness for further study.

Blood was collected from mice given the highest dose which completed the treatment schedule, 10/4/1, for both the oral gavage and IV groups 72 hours after their last dose to perform a CBC analysis (Figure 5, Supplemental Table S5). Most hematologic parameters, including total white blood cells (WBC), lymphocytes, monocytes, and platelets, did not differ significantly between oral and IV administration. Monocyte % were elevated in both treatment groups relative to published reference ranges, with similar values observed following oral and IV administration. Total neutrophil counts were below the healthy reference range in both groups, with significantly lower levels observed in the IV group compared with oral gavage treatment. Neutrophil % were below the healthy reference range only in the IV group and were significantly reduced relative to the oral group. Notably, several erythroid-related parameters, including red blood cell (RBC) count, hemoglobin, hematocrit %, and reticulocyte %, showed small but statistically significant differences between administration routes. Although values in the oral group remained within normal ranges, the IV group exhibited borderline suppression of RBC and hemoglobin levels alongside reductions below healthy minimums in hematocrit and reticulocyte %. While the magnitude of these changes is small, their consistency suggests a route-dependent effect that warrants further investigation to determine underlying mechanisms and potential clinical impact.

Figure 5.

Hematologic analysis from BALB/c mice 72 hours after final dose of 75 mg/kg PTX and 18 mg/kg MB POx/PTX/MB treatment via IV injection or oral gavage (OG). Gray shaded regions indicate healthy reference ranges from published data. Colored bars represent the median, error bars indicate the min-to-max range, and individual points represent biological replicates (n = 4 per group). Statistical significance was determined using the Mann-Whitney test (ns > 0.05 * ≤ 0.05, ** ≤ 0.01, *** ≤ 0.001, **** ≤ 0.0001).

Discussion

Physicochemical characterization and release profile determinization of the POx/PTX/MB formulation provide preliminary support for its stability and suitability as a delivery platform for solid tumor applications. The multidrug-loaded polymeric micelles formed stable nanoparticles under physiologically relevant conditions, in a size range consistent with passive tumor targeting mechanisms^35,36.

Of note is the improved incorporation of MB into POx micelles when co-loaded with PTX. This effect has been consistently observed in our studies, where multiple drugs co-formulated with PTX demonstrate enhanced incorporation, as documented in both our published work and additional unpublished data that will be presented in future reports²³. The precise molecular basis for this phenomenon remains unclear and warrants further investigation.

In the present study, two non-mutually exclusive mechanisms may contribute to the enhanced loading of MB in the presence of PTX. First, PTX belongs to a class of compounds that are not only efficiently solubilized within POx micelles but also contribute to stabilizing their spherical morphology. Specifically, PTX increases the overall thermodynamic stability of POx micelles and lowers their critical micelle concentration (CMC), indicating strong interactions with the micellar core, which we have previously shown increase with higher PTX loading^37,38. Such stabilization may enable more efficient incorporation and retention of an additional compound, resulting in the formation of more stable three-component assemblies composed of PTX, MB, and POx. We have recently highlighted that PTX, a diterpene compound, possesses several structural features conducive to cooperative co-assembly with POx block copolymers. These include (i) multiple aromatic rings, (ii) hydrogen-bond donor functionalities, (iii) hydrogen-bond acceptors, and (iv) a high number of rotatable bonds. Consequently, PTX can interact with POx micelles through hydrophobic and van der Waals forces as well as hydrogen bonding with the carbonyl (C=O) groups of the polymer backbone³⁹.

Second, direct interactions between PTX and MB may also contribute to their co-solubilization. Although MB lacks hydrogen-bond donor groups, which are often associated with efficient loading in POx micelles, it contains aromatic ring systems capable of π–π interactions with PTX. Such drug–drug interactions could promote cooperative packing within the micellar core and facilitate MB incorporation. Therefore, both PTX-mediated stabilization of the POx micelle and direct PTX–MB interactions may contribute to the enhanced loading of MB observed in PTX-containing formulations.

The in vitro release profiles should be interpreted with caution, as release measured under sink conditions using dialysis does not necessarily reflect drug release or accumulation within the TME. Pharmacokinetic and biodistribution studies will be required to determine the extent and temporal overlap of MB and PTX delivery to tumors and are the subject of ongoing work. Under the conditions evaluated here, MB was released more rapidly than PTX, with >90% release occurring within 6–12 h compared with 24–48 h for PTX. This difference is likely attributable to the strong polymer-PTX interactions described above, but differences in molecular weight between PTX and MB and differences in serum protein affinities may also contribute to the rate disparity. Whether this difference in release kinetics is advantageous or detrimental for therapeutic efficacy remains unknown and will depend on the optimal sequencing of antimicrobial and chemotherapeutic activity in vivo. Consequently, the release profiles reported here should be viewed as an initial physicochemical characterization of the co-formulated micelles rather than a direct predictor of in vivo therapeutic performance.

The cytotoxicity profile of the POx/PTX/MB micelles was evaluated to determine whether the drugs retained their activity after incorporation into the POx. Across two TNBC cell lines, POx/PTX/MB micelles demonstrated comparable cytotoxic activity to a DMSO PTX/MB control, suggesting that micellar incorporation preserved PTX-mediated cytotoxicity while avoiding measurable cytotoxic effects from the POx carrier itself. These findings are consistent with previous reports demonstrating the biocompatibility of POx-based materials and support the use of POx as a drug delivery platform²². However, because POx/MB formulations were not sufficiently stable for the duration of the cytotoxicity studies, evaluation was limited to the co-loaded formulation. As MB primarily targets anaerobic bacteria and protozoa, it is unlikely to substantially contribute to cytotoxicity under normoxic conditions, although hypoxia-dependent anticancer activity has been reported²⁷.

The antimicrobial activity of POx/PTX/MB micelles against F. nucleatum was evaluated to determine whether incorporation of MB into the POx micellar system affected its antibacterial activity. As expected, the POx vehicle and POx/PTX formulations did not inhibit bacterial growth at the concentrations tested. Meanwhile, POx/PTX/MB micelles exhibited potent bactericidal activity, demonstrating that MB retained its antibacterial efficacy following co-loading with PTX. These findings confirm that the POx micelle platform enables incorporation of MB without diminishing its activity, supporting the use of POx/PTX/MB micelles as a multimodal delivery system for simultaneous chemotherapeutic and antimicrobial treatment.

To assess the formulation’s safety, CBC analysis was conducted and identified generally comparable hematologic profiles between oral and IV administration groups, with statistically significant differences observed primarily in erythroid-related parameters. Although RBC count, hemoglobin, hematocrit, and reticulocyte percentage were modestly lower following IV administration, the biological significance of these differences remains unclear. Because CBC measurements represent a single time point of circulating blood cell populations, they cannot distinguish among altered production, peripheral utilization, or redistribution. Additional studies evaluating bone marrow function, reticulocyte production, and longitudinal hematologic changes will be needed to determine the mechanism underlying these observations.

One possible explanation is the well-established myelosuppressive effect of PTX on proliferating bone marrow progenitors, which commonly manifests as neutropenia and anemia at therapeutic systemic exposures⁴⁰. Pharmacokinetic differences between administration routes are also consistent with this interpretation. PTX has low oral bioavailability due to poor aqueous solubility, P-glycoprotein-mediated efflux, and extensive first-pass metabolism, resulting in minimal systemic exposure following oral administration (often <1%)⁴¹. In contrast, IV administration bypasses these barriers and produces systemic drug exposures associated with hematologic toxicity. By comparison, MB exhibits substantially higher oral bioavailability (70–80%) and is not typically associated with bone marrow suppression at clinically relevant exposures^42,43.

However, because no direct assessments of bone marrow or organ function were performed in this study, alternative explanations, including transient systemic effects or organ toxicity, cannot be excluded. The absence of weight loss, behavioral abnormalities, or overt clinical signs of toxicity suggest that any adverse effects were modest under the conditions examined, although these observations alone cannot rule out subclinical organ toxicity. Overall, these findings are consistent with mild route-dependent hematologic effects following IV administration, and are in agreement with previously reported studies of single-drug POx/PTX formulation while highlighting the need for more comprehensive toxicological evaluation, including assessment of liver and kidney function with serum chemistry panels, histopathology, and bone marrow analysis.^20,21.

While oral administration of POx/PTX/MB may reduce systemic PTX exposure and associated hematologic effects, IV delivery provides more predictable and consistent plasma exposure, supporting the achievement of therapeutically relevant tumor concentrations. An additional advantage of IV administration is the potential to spare the gut microbiome by reducing direct intestinal exposure to the antibacterial component, which is increasingly recognized as important for maintaining immune homeostasis and potentially influencing therapeutic outcomes⁴⁴. Together, these considerations support continued development of IV formulations while emphasizing the importance of optimizing the therapeutic window to maximize antitumor efficacy and minimize systemic toxicity.

Future studies will characterize pharmacokinetics and biodistribution following oral and IV administration to define circulation time, tumor accumulation, and intratumoral exposure, alongside longitudinal hematologic analyses to assess IV-associated erythroid effects. The efficacy will then be evaluated to determine impacts on tumor growth, immune modulation, metastasis, survival, and tumor-associated microbiota, as well as effects on the gut microbiome. These findings will guide optimization of polymer design, release kinetics, dosing strategies, and drug combinations, and will be validated across tumor types to support development of nanomedicine approaches targeting both cancer cells and the tumor microbiome.

Conclusion

This study demonstrates the feasibility of a dual-delivery approach using POx micelles to simultaneously target tumor cells and intratumoral bacteria. By co-encapsulating paclitaxel (PTX) and metronidazole benzoate (MB), we established a system capable of delivering a cytotoxic chemotherapeutic to proliferating tumor cells while eliminating Fusobacterium nucleatum, a tumor-resident pathogen implicated in cancer progression and therapy resistance. Compared to single-drug POx/MB micelles, the combination with PTX reduced polydispersity and increased colloidal stability. The combination micelles exhibited high drug-loading efficiency, maintained stability under physiological conditions, and were well-tolerated in a murine model, supporting their suitability for systemic administration. These results provide initial evidence for integrating antimicrobial therapy with conventional chemotherapy and underscore the potential of POx micelles as a versatile platform for combination treatment strategies in tumors harboring pathogenic bacteria.

Data Availability Statement

All data supporting the findings of this study are available within the paper and its Supplementary Information. Data used in this manuscript has been published for public access at https://dataverse.unc.edu/dataverse/TargetingTumorMicrobiome.