Combination therapy for tuberculosis treatment: pulmonary administration of ethionamide and booster co-loaded nanoparticles

Tuberculosis (TB) is a leading infectious cause of death worldwide. The use of ethionamide (ETH), a main second line anti-TB drug, is hampered by its severe side effects. Recently discovered “booster” molecules strongly increase the ETH efficacy, opening new perspectives to improve the current clinical outcome of drug-resistant TB. To investigate the simultaneous delivery of ETH and its booster BDM41906 in the lungs, we co-encapsulated these compounds in biodegradable polymeric nanoparticles (NPs), overcoming the bottlenecks inherent to the strong tendency of ETH to crystallize and the limited water solubility of this Booster. The efficacy of the designed formulations was evaluated in TB infected macrophages using an automated confocal high-content screening platform, showing that the drugs maintained their activity after incorporation in NPs. Among tested formulations, “green” β-cyclodextrin (pCD) based NPs displayed the best physico-chemical characteristics and were selected for in vivo studies. The NPs suspension, administered directly into mouse lungs using a Microsprayer®, was proved to be well-tolerated and led to a 3-log decrease of the pulmonary mycobacterial load after 6 administrations as compared to untreated mice. This study paves the way for a future use of pCD NPs for the pulmonary delivery of the [ETH:Booster] pair in TB chemotherapy.

formulated NPs were remarkably stable upon storage up to 5 months, with less than 5% size variation. The size variation over a period of 1 month is shown in the Supplementary Formulations up to an ETH concentration of 0.43 mg/mL were completely stable. In these conditions the highest achieved drug loading (DL) and encapsulation efficiency (EE) were 11 wt % and 36 wt %, respectively. Unfortunately, raising the amount of ETH to 0.71 mg/mL in the attempt to further increase DL and EE, the DL resulted unchanged and the encapsulation efficiencies decreased to 22 % ( Supplementary Fig. S2). Moreover, increasing ETH amounts in the NP preparation procedure lead to the formation of drug crystals which progressively grew in the suspension media until being seen with the naked eye, few days after NP preparation ( Supplementary Fig. S3).
Supplementary Fig. S3 shows the decreasing of DL and EE (P1 and ETH concentration of 0.71 mg/mL) during the first 8 days after NPs formulation. The formation of crystals was observed after 5 days and corresponds to a further decrease in NPs DL and EE.
ETH crystallisation in P1 NPs was avoided only at ETH concentrations below 0.71 mg/mL, which is insufficient for in vivo applications. Attempts to concentrate the NPs suspensions failed because of a strong aggregation in the absence of protective surfactants.  showing a higher affinity of the Booster for the polymeric matrix.

PLA NPs -Nanoemulsion method
To address the challenge of coencapsulating ETH and Booster, a second method based on the formation of an emulsion, stabilized with polyvinyl alcohol (PVA), a surfactant approved for pulmonary administration, was set up using P4 (see material and methods section).
An extensive study on mean diameter of PLA NPs obtained by nanoemulsion has been performed using the 3 independent methods: PCS, NTA and cryo-TEM (see main text NPs formulated with the nanoprecipitation method were remarkably stable upon storage more than 5 months, with less than 10 % size variation and a PdI still lower than 0.

pCD NPs -Stability and release properties
The drug-loaded particles were stable upon storage and drugs were released only upon dilution of the NPs suspensions. Indeed, the drug molecules inside pCD NPs are in equilibrium with the ones in the suspension medium. Both ETH and Booster, poorly soluble in aqueous solution, strongly bound to pCD NPs. For instance, if the NPs are diluted ten times, one can calculate that around 90% of Booster will still be associated to the NPs.
The mechanism of drug release from pCD NPs, based on a partition mechanisms, has been thoroughly described in our previous publications [1][2][3] .

Section II -Phenotypic extracellular and intracellular assays and in vivo results
To investigate the efficacy of drug-loaded NPs against M. tuberculosis, we used two phenotypic assays that are disease-relevant and amenable to high-throughput. The first one is a simple assay relying on the monitoring of fluorescence from a strain of M. tuberculosis that expresses the green fluorescence protein using a microplate photometer. The second is an image-based model that allows the multi-parametric quantification of M. tuberculosis replication inside its favourite niche, the macrophage. The dedicated protocol used for image analysis with the Columbus software is given in Supplementary S5 below.

Find Spots
Output Population : Spots

Calculate Intensity Properties
Output Properties : Intensity Spot Exp1Cam2

Population : Nuclei 2 Selected
Supplementary Figure S5-Columbus script used for image analysis. Two colour-images were analysed. The red recorded on the channel "Exp1Cam3" was used for the nucleus and cytoplasm detection. The green recorded on the channel "Exp1Cam2" was used to detect the bacteria. ROI: Region of interest.
In Supplementary Fig. S6 we point out that the % of inhibition in the presence of empty PLA and empty pCD is negligible (close to zero) independently from the concentration of the NPs, both in the intra-and extracellular assay. These results show that the antibacterial effect observed is due to the presence of ETH and Booster released from the nanoparticles since the unloaded nanoparticles were not able to impair bacterial replication. In Supplementary Table S2 we present the results of 3 independent in vivo experiments after 2 weeks of treatment as described in the manuscript. In the 3 experiments we observed a significant decrease in the pulmonary bacterial load after treatment with ETH and [ETH:Booster] encapsulated in pCD NPs.

Size measurement by PCS and Zeta Potential measurement
The average hydrodynamic diameter of the NPs was determined at 25°C, in triplicate, with

Drug sample preparation for drug(s) dosage
To determine the amounts of ETH and Booster incorporated in PLGA NPs, 500 µL of NP In all cases, R 2 values for both ETH and Booster were higher than 0.99. From the obtained data the DL and EE were calculated as shown in equation 1 and 2, respectively. The DL can be defined as the mass fraction of a NP that is composed of drug, while the EE can be considered as the fraction of drug effectively encapsulated into the NPs compared with the amount that was initially added during the particles preparation 8,9 . (1)

Extracellular in vitro assay
Bacteria were diluted at 2×10 6 bacteria/mL using complete 7H9 medium and 45 µL/well of bacterial suspension were added in 384-well assay plates. After 5 days incubation at 37°C, 5% CO2, extracellular plates were read using a fluorescence reader (Victor X3, Perkin Elmer) at excitation/emission of 485/535 nm for 0.1 seconds/well with a small emission aperture and CW-lamp energy of 50,000. The read-out, relative fluorescence units (RFU), versus the ETH concentration was then plotted using GraphPad Prism 5.0 software and the concentration required to inhibit 50% of the bacterial replication (IC50) was calculated by nonlinear regression analysis using the equation for a sigmoidal dose-response curve with variable slope.

Intramacrophage in vitro assay
For intracellular assay, bacteria were mixed with RAW 264.7 macrophages to prepare a suspension at 5×10 5 cells/mL and 1×10 6 bacteria/mL (multiplicity of infection 2) in RPMI-1640 + glutamax containing 10% FBS. After 2 hours of infection at 37°C with shaking (120 rpm), infected cells were washed with RPMI-1640 + glutamax containing 10% FBS by centrifugation at 1,100 rpm for 5 minutes. The remaining extracellular bacilli from the infected cell suspension were killed by a 1 hour 50 g/mL amikacin (A2324-5G, Sigma) treatment and then washed twice with RPMI-1640 + glutamax containing 10% FBS. Finally, 45 µL/well of cellular suspension was added in the 384-well assay plate and incubated during 5 days at 37°C, 5% CO2. Macrophages were then stained with 5 µM Syto 60 (S11342, Molecular probes) dye for 1 hour, followed by plate sealing, imaging acquisition and data analysis.

Image acquisition
Confocal images were recorded on an automated fluorescent ultra-high-throughput microscope Opera (Perkin Elmer). This microscope is based on an inverted microscope architecture that allows imaging of cells cultivated in 96-or 384-well microplates. Images were acquired with a 20X-water immersion objective (NA 0.70). A double laser excitation (488-nm and 640-nm) and dedicated dichroic mirrors were used to record green fluorescence of mycobacteria and red fluorescence of the macrophages on two different cameras, respectively. A series of 6 pictures at the center of each well were taken and each image was then analyzed using Columbus system version 2.5.1 as previously described 10 to extract the bacterial area and the number of cells.

Intramacrophage assay data analysis
The intracellular bacterial area was normalized with the negative control DMSO (0% inhibition) and the positive control INH at a concentration of 1 µg/mL (100 % inhibition) by converting it into a percentage of bacterial replication inhibition (% inhibition). % inhibition was calculated as shown in the equation 3.

Lung histology
Infected or uninfected 8-week-old mice were divided in groups and endotracheally administered with water (vehicle) or pCD NPs (Fig 5a, Fig.6).
At the determined end-point mice were euthanized, lungs were harvested and soaked in 4% formaldehyde (10% formalin solution, neutral buffered, HT501128, Sigma-Aldrich) for 24 hours, before being embedded in paraffin. Tissues were sliced with microtome and 5 µm sections were stained with Hematoxylin-Eosin (H-E) for light microscopy examination for anatomopathology.

Flow cytometry
Harvested lungs were cut into small pieces and incubated for 1 hour at 37°C with a mix of For the 1-week treatment and the 2-week treatment, administrations were on day 7, 9, 11 and on day 7, 9, 11, 14, 16, 18 respectively. Body weight were recorded after each treatment ( Fig. 5b).. To assess the reproducibility of NPs administration through the MicroSprayer®, the delivered doses of NP suspensions were collected in glass vials after each spray and were accurately weighed. Then the amount of the delivered drug was quantified by HPLC as already described.
The protocol to administer the NPs in mice was adapted from a previously reported one 11 .
Briefly, mice were placed in isoflurane chamber (Aerrane®, Baxter SAS, France). Then, one mouse was placed on the back on a platform (Mouse Intubation Platform -Model MIP, Penn Century Inc., Wyndmoor, PA) with isoflurane mask and hanging on its teeth. The tongue was pulled out by a tweezer and a laryngoscope (Small Animal Laryngoscope for mouse -Model LS-2-M, Penn Century Inc., Wyndmoor, PA) was used to see the trachea and 50µL of suspensions were delivered inside the lung allowing to aerosolize 50 µL suspensions inside the lungs.
At day 14 or 21, mice were euthanized and lungs were homogenized with MM300 bead beater (Retsch) and eight ten-fold serial dilutions were plated onto 7H11 agar plates supplemented with 10% OADC. CFUs were determined after a three-week growth.
Represented data are mean values ± standard deviation of one representative experiment from three independent experiments. Results for each independent experiment were summarized in Supplementary Table S2. Statistics were performed using Student's t-test and one-way ANOVA analysis. Same p-values for in vivo experiments were obtained with the two tests. *: p<0.1, **: p<0.01,***: p<0.001.