All chemicals and reagents were purchased from Sigma Chemical (St. Louis, Missouri, United States) unless otherwise stated. Hydroxyapatite (HA) powder was purchased from Chemical Reagent Co., Ltd. (Shanghai, China). Rifapentine (RPT) (content: 98%) was obtained from Sichuan Med-Shine Pharmaceutical Co., Ltd. (Sichuan, China). PLGA-PEG-COOH (Mw: 12,000; lactide:glycolide = 50:50) was purchased from Jinan Daigang Biomaterial Co., Ltd. (Shandong, China). Dicyclohexylcarbodiimide (DCC) and N-Hydroxysuccinimide (NHS) were obtained from Xiya Chemical Technology Co., Ltd. (Shandong, China). TC was purchased from Shanghai Yuanye Biological Technology Co., Ltd. (Shanghai, China). 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide (MTT) cytotoxicity assay kit was purchased from Bejing Solarbio Science and Technology Co., Ltd. (Bejing, China). Dubelcco’s Modified Eagle’s Medium (DMEM), trypsineEDTA, fetal bovine serum (FBS) and penicillin-streptomycin solution were obtained from Thermo Fisher Scientific (Waltham, MA, United States). 1,1′-Dioctadecyl-3,3,3′,3′-tetramethylindotricarbocyanine iodide (DiR) was obtained from Abcam (Cambridge, United Kingdom). Shirasu porous glass (SPG) membranes were purchased from SPG Technology Co. Ltd. (Miyazaki, Japan).

A DCC/NHS technique was used to synthesize the TC-PLGA-PEG conjugates (Figure 1). Briefly, 500 mg of PLGA-PEG-COOH, 25.8 mg of DCC, and 14.5 mg of NHS were dissolved in 30 mL anhydrous dimethyl sulfoxide (DMSO). This solution was stirred for 24 h to activate the carboxylic acid under the protection of nitrogen. The by-product was removed using a syringe filter with 0.45 μm pore size. Then, 18.5 mg of TC was added to the activated PLGA-PEG, the mixture was sequentially stirred under the protection of nitrogen for 24 h at 60°C. The resultant solution was dialyzed against deionized water using a dialysis membrane (MWCO 3.5 kDa) for another 48 h, and then centrifuged at 18,000 rpm for 15 min. The resultant solution washed three times with deionized water. The final product was lyophilized and stored.

To evaluate molecular state of the product, UV-2550 spectrophotometer spectra (Shimadzu, Japan) of TC, PLGA-PEG and TC-PLGA-PEG were recorded with DMSO as the solvent. Then, the product was further confirmed by Fourier transform infrared (FTIR) spectra (Shimadzu, Japan). The spectra of TC, PLGA-PEG and PLGA-PEG-TC between the wavelengths of 500 and 4,000 cm⁻¹ were recorded. In addition, those conjugates were further confirmed by ¹ H NMR spectra on Bruker spectrometer.

Drug-loaded bone-targeted NPs (rifapentine/TC-PLGA-PEG NPs) were prepared by premix membrane emulsification technique (Figure 2). At first, aqueous solution was mixed with DCM containing rifapentine and TC-PLGA-PEG, and the mixture was sonicated using a probe sonicator (UP400S, Hielscher Ultrasonics, Germany) at 50% amplitude for 1 min to form primary emulsion. It was then coarsely emulsified through SPG membrane (1 μm pore size, SPG Technology Co., Ltd., Sadowaracho, Japan) under magnetic stirring for 5 min to form coarse emulsion. Secondly, this coarse emulsion was subjected to five homogenization cycles by the usage of SPG membrane (0.1 μm pore size) with a high pressure. The obtained emulsion was poured quickly into 100 mL of a 1% (w/v) PVA solution and stirred overnight to evaporate organic solvent. Afterwards, the solutions were filtered through the 0.45-μm microporous membrane. The separated NPs were collected by centrifugation and washed for three times followed by lyophilization. The non-targeted NPs (rifapentine/PLGA-PEG NPs) were obtained using the same method. To make fluorescent-labelled NPs, coumarin 6 or DIR was added to the DCM, as the only change to the procedure described above.

The zeta potential (ZP), polydispersity index (PDI) and particle size (PS) were determined using a ZetaSizer Nano-ZS (Malvern Instruments Ltd., Malvern, United Kingdom). The morphology of rifapentine/TC-PLGA-PEG NPs was confirmed using a transmission electron microscopy (JEM-1230, Jeol Ltd., Japan) after staining with sodium phosphotungstate solution (2%, w/v). The drug entrapment efficiency (EE%) and drug loading (DL%) were assayed by ultraviolet spectrophotometry method at a wavelength of 475 nm (Wu et al., 2015). To assess the stability of NPs, samples were stored in a stable chamber at 25°C for 90 days. Periodically, a sample was collected to measure PDI, EE% and particle size.

In vitro rifapentine release study was evaluated by dialysis method. Briefly, free rifapentine, rifapentine/PLGA-PEG and rifapentine/TC-PLGA-PEG NPs were placed in a dialysis bag (MWCO 3.0 kDa), followed by incubated in PBS medium (pH 7.4). The PBS medium was maintained at 37°C and stirred at 100 rpm. At pre-determined intervals, samples were withdrawn and replaced by the same volume of fresh medium. The content of rifapentine was measured using ultraviolet spectrophotometry method as described above.

To evaluate the bone mineral binding ability, HAp powder was used as previously reported (Xie et al., 2018). The coumarin 6-labelled rifapentine/PLGA-PEG and rifapentine/TC–PLGA–PEG NPs (5 mL, 0.1 wt%) were detected via fluorospectrophotometry before reaction. Then, the solutions were separately added into PBS containing HAp (10 mg/mL) and shaken in the dark for 2 h. After filtrated with 0.22 μm microporous membrane, the intensity of the supernatant was detected. The binding ratio (%) in relation to initial intensity was calculated using the following equation: B i n d i n g   r a t i o   % = I i n t e n s i t y   b e f o r e   H a p   a d s o r p t i o n − I i n t e n s i t y   a f t e r   H a p   a d s o r p t i o n / I i n t e n s i t y   b e f o r e   H a p   a d s o r p t i o n × 100

Bone marrow mesenchyml stem cell (BMSCs) were obtained from 12-week-old New Zealand white rabbits according to our previous report (Wang et al., 2015). The cytotoxicity of NPs was evaluated by MTT assay. Rifapentine solution, rifapentine/PLGA-PEG NPs and rifapentine/TC-PLGA-PEG NPs were incubated with BMSCs at 37°C for 24 h at final RPT concentrations of 0, 5, 25 and 50 μg/mL. Afterwards, 15 μL of MTT was added to each well and incubated further for another 4 h. After dissolving and shaking, the absorbance of formazan at 570 nm was measured with a microplate reader. Furthermore, different concentrations of blank NPs (ranging from 0 μg/mL to 1,200 μg/mL) were also incubated with BMSCs. The cell viability (%) in relation to negative control group was calculated using the equation: v i a b i l i t y   % = A   c e l l s   i n c u b a t e d   w i t h   f o r m u l a t i o n A   c e l l s   i n c u b a t e d   o n l y   w i t h   c u l t u r e   m e d i u m × 100

Haemolytic toxicity was conducted to evaluate the compatibility of rifapentine loaded NPs with erythrocytes. In this study, the blood was collected from 12-week-old New Zealand white rabbits. Briefly, the erythrocytes were separated and resuspended in 0.9% w/v normal saline. Various concentrations (0.5–3.5 mg/mL) of NPs were mixed with normal saline for interaction with erythrocytes suspension. Then, a UV–visible spectrophotometer at 545 nm was used for absorbance measurements. The haemolytic rate of the sample was calculated by using the equation: H a e m o l y t i c   r a t e   % = A a b s o r b a n c e   o f   s a m p l e − A a b s o r b a n c e   o f   s a l i n e /   A a b s o r b a n c e   o f   w a t e r − A a b s o r b a n c e   o f   s a l i n e × 100

The antimycobacterial activity of NPs against M. tuberculosis was evaluated by microplate Alamar blue assay following a previously reported protocol (Que et al., 2022). Firstly, 200 μL of sterile deionized water was added in outer peripheral wells. Secondly, 100 μL of 7H9GC-OADC broth and 100 μL of rifapentine was added separately. Thirdly, serial dilutions (1:2) were performed in columns 3 to 10. The final drug concentration range of free rifapentine was 1.0 μg/mL to 0.004 μg/mL, and the final drug concentration range of rifapentine loaded NPs was 0.75 μg/mL to 0.003 μg/mL. Finally, M. tuberculosis inocula were added in columns 2 to 11. Column 11 was considered as a positive control without drug. After 5 days of incubation, 50 μL of Alamar Blue reagent and 10% Tween 80 was added and re-incubated for another 24 h. The results were analysed by means of colour changing in each well. The concentration corresponding to no observable bacterial growth is the minimum inhibitory concentration of formulation.

Kunming male mice (body weight 27 ± 2 g) were obtained from the Animal Center of Xinjiang Medical University. Animal experiments and welfare were complied with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (No. 8023). The experiments were approved by the Ethics Committee of The Xinjiang Medical University (No 20180223–13). The mice were divided into three groups: rifapentine, rifapentine/PLGA-PEG NPs and rifapentine/TC-PLGA-PEG NPs. The mice were given a single dose of 10 mg/kg (rifapentine) intravenously. Then, blood samples were taken from the retro-orbital plexus of each animal per time point. Acetonitrile was used to precipitate the separated plasma. Then, the supernatants were removed after vortex and centrifugation. The final sample was injected on the HPLC column to estimate the amount of rifapentine. The HPLC analysis was performed using a C18 column (250 mm × 4.6 mm, 5 µm particle size; Waters, United States) at 25°C. The mobile phase consisted of methanol-0.0367 mol/L KH₂PO₄, flow rate of 1 mL/min, and detection at a wavelength of 254 nm. The maximum concentration (C _max ), the time to reach maximum plasma concentration (T _max ), the peak concentration (C _max ), the half-life (t _1/2 ), the mean residence time (MRT), and the area under the curve (AUC_0–∞) were evaluated.

For a single-dose rifapentine disposition study, the mice were given a single dose of 10 mg/kg rifapentine intravenously in each group. The mice were divided into four groups: control group (normal saline), rifapentine/PLGA-PEG NPs and rifapentine/TC-PLGA-PEG NPs. The animals were sacrificed at a specified time interval. After the plasma was obtained, their tissues (liver, kidney and bones) were excised and washed quickly with normal saline, then stored at −20°C. The drug concentration was estimated in 20% (w/v) tissue homogenates. The content of rifapentine was measured using HPLC method as described above. The results were expressed as μg/g (microgram per Gram of tissue).

To further confirm the bone-targeting efficacy of rifapentine/TC-PLGA-PEG NPs, Dir-loaded NPs were injected into the Kunming mice through the tail vein after anesthetization. The Kunming mice were randomly divided into two groups of DIR loaded rifapentine/TC–PLGA–PEG NPs, and DIR loaded rifapentine/PLGA–PEG NPs. Animals in each group were observed 48 h post injection using IVIS^® Spectrum In vivo Imaging System.

To observe the side effects of NPs on mouse organs, the animals were sacrificed at the fourth day after the injection. The heart, kidney, liver, lung and spleen were harvested, then fixed in 10% neutral formalin for 72 h. The samples were paraffin embedded for haematoxylin and eosin (H&E) staining. In addition, the blood samples were also obtained for blood biochemistry testing before the mice were euthanized.

Statistical analysis was performed using SPSS version 17.0. Results are expressed as mean ± SD, and differences between continuous variables were assessed via One-way ANOVA with Tukey’s test. A statistically significant difference was defined as p < 0.05.

The UV spectra of TC, PLGA–PEG and TC–PLGA–PEG were shown in Figure 3. The spectrum of TC showed two characteristic absorption peaks at 266 nm and 367 nm. In contrast, no obvious UV absorption peak was observed at 200 nm–800 nm before the reaction, indicating that the mixed solution of PLGA–PEG copolymer had no obvious UV absorption peak. Compared with the spectrum of TC before the reaction, spectrum of TC-PLGA-PEG showed no significant UV absorption peak near 367 nm instead of a weak absorption peak at 328.5 nm, and the peak at 266 nm still existed. The results showed that the synthesis of TC makes PLGA–PEG solution without characteristic peaks appear obvious characteristic peaks, and these characteristic peaks are similar to the characteristic peaks of TC.

The infrared spectra of TC, PLGA–PEG and TC–PLGA–PEG were shown in Figure 4A. In the spectrum of PLGA–PEG, the absorption band at 3,593 cm⁻¹ was assigned to terminal hydroxy groups in the copolymer, and the band at 2,956 cm⁻¹ was C-H stretch. A strong band at 1764 cm⁻¹ was attributed to C=O stretch, and the absorption at 1,169–1,092 cm⁻¹ was assigned to C-O and C-O-C groups, which are consistent with the data in the previous reports. (Ge et al., 2018). The spectrum of TC showed two particular bands at 3,310 cm⁻¹ and 1,650–1,450 cm⁻¹, corresponding to the N-H stretch and benzene ring, respectively. While all above mentioned peaks of PLGA–PEG and TC were presented in the spectrum of TC–PLGA–PEG conjugates as well, there were some minor changes and shifts by which it was possible to conclude the successful incorporation with each other. In the spectrum of TC–PLGA–PEG, the absorption band around 3,330 cm⁻¹ was assigned to the mixed band of OH in the copolymer and NH in amido bond; The absorption band around 1760 cm⁻¹ was assigned to C=O stretch in the copolymer; The absorption band around 1,630 cm⁻¹ was assigned to C=O in amido bond; The absorption band around 1,530 cm⁻¹ was assigned to C-N-H in amido bond; The absorption band around 1,300 cm⁻¹ was mainly assigned to the mixed band of C-N and N-H in amido bond.

The ¹ H NMR spectrum of TC, PLGA–PEG and TC–PLGA–PEG were shown in Figure 4B. The peaks at 7.0–8.0 ppm belonged to the hydrogens of the benzene ring of TC, peaks at 4.0–6.0 ppm belonged to the methylene and methylidyne hydrogens of PLGA, and peaks at 3.6 ppm belonged to the–CH2CH2O–units of PEG, respectively. These peaks are consistent with the previous reports (Wang et al., 2015; Xie et al., 2018). Based on ¹ H NMR, UV and infrared spectrum, it was evidence that the TC-PEG-PLGA conjugates were synthesized successfully.

The particle size of RPT/TC–PLGA–PEG NPs appears to be slightly larger than RPT/PLGA–PEG NPs, but the difference is not significant. Drug encapsulating efficiency of both RPT–loaded NPs was greater than 80% with high DL%. With regard to applying single emulsifier, the NPs prepared by solvent evaporation showed noticeably sustained release behavior when 1% PVA were used, and the concentration of PVA produced NPs with high EE% (>80%). The PDI of both NPs was found to be around 0.20, suggesting that the NPs had good homogenetity (Table 1). Transmission electron microscope showed that rifapentine loaded TC-PLGA-PEG NPs, rifapentine loaded PLGA-PEG NPs, TC-PLGA-PEG NPs and PLGA-PEG NPs were dispersed as individual particles with a spherical shape (Figure 5). There were no significant changes on the surface or interior of NPs, which is consistent with the previous findings reported in the literature (Wang et al., 2015; Que et al., 2022). In addition, there were no significant differences in stability studies, indicating that both NPs were stable for at least 3 months (Table 2).

The release mode of rifapentine from TC–PLGA–PEG and PLGA–PEG NPs was shown in Figure 6A. The release pattern was characterized by an initial burst and subsequent controlled release. By contrast, free rifapentine was released in a burst manner within 2 h, and approximately 90% in 6 h. The release kinetics of both NPs were evaluated using the zero-order equation, first-order equation, and Higuchi equation models. In the TC-PEG-PLGA group, the R square of the zero-order equation, first-order equation and Higuchi equation was 0.82084, 0.95765 and 0.93803, respectively. Similarly, in the PEG-PLGA group, the R square of the zero-order equation, first-order equation and Higuchi equation was 0.81452, 0.96099 and 0.93233, respectively. Therefore, it can be concluded that the release kinetics of both NPs follow the first-order equation model.

Figure 6B showed the binding ratio of rifapentine/PLGA-PEG and rifapentine/TC-PLGA-PEG NPs. Most (68.0%) of the rifapentine/TC-PLGA-PEG NPs bound to the HAp powder, whereas only small amounts of coumarin 6 (4.6%) and rifapentine/PLGA–PEG NPs (7.2%) were non-specifically adsorbed.

Cytotoxicity was evaluated on BMSCs. The results exhibited in Figure 6C showed that both NPs were nontoxic for BMSCs. Both drug-loaded NPs did not present a significant difference (p > 0.05) compared with the negative control. However, free rifapentine presented a cell viability of less than 70% at concentrations of 50 μg/mL, demonstrating that the nano-formulation may reduce the toxicity of rifapentine. To further evaluate the effect of blank NPs on cell viability, the cells were exposed to NPs at different concentrations. The results revealed that cell viability was still above 90% even at a concentration of 0.6 mg/mL. However, it must be emphasized that high concentration (1.2 mg/mL) of NPs can still affect cell activity, but in practical applications, the dose is much smaller than this concentration (Figure 6D). Furthermore, both NPs were found to possess negligible haemolysis of erythrocytes even at the highest concentrations (Figure 6E). Therefore, rifapentine loaded NPs were potentially feasible for intravenously administration.

As shown in Figure 7, the MIC of free PRT was 0.25 μg/mL, whereas this of rifapentine loaded NPs was 0.094 μg/mL (approximately 2.7 lower), indicating that both NPs were more effective than free drug. Furthermore, the MIC of PEG–PLGA NPs was also 0.094 μg/mL, which demonstrated that the conjugated TC has no effects on minimum inhibitory concentration.

As shown in Figure 8, compared with free rifapentine, both rifapentine loaded NPs showed a prolonged drug release profile for up to nearly 100 h. Table 3 showed the major pharmacokinetic parameters. To further confirm the bone-targeting ability of TC–PEG–PLGA NPs, Dir-loaded NPs were intravenously injected in mice. At 48 h after administration, their bones were collected after in vivo imaging acquisition. As shown in Figure 9, bones in the mouse receiving Dir loaded TC-PLGA-PEG NPs exhibited strong fluorescence in comparison to that receiving Dir loaded PLGA-PEG NPs. However, even in the bone-targeting group (TC-PLGA-PEG NPs), NPs were inevitably accumulated in the liver. The results of tissue drug distribution (Figure 10) revealed that the drug concentration of bone tissue in TC–PLGA–PEG group was significantly higher than that in PLGA–PEG group and RPT group at all time (p < 0.05). In addition, the drug concentration of TC–PLGA–PEG group in bone tissue was significantly higher than that in blood 24 and 48 h after administration (p < 0.05), while the drug concentration of PLGA–PEG group in bone tissue was significantly lower than that in blood 24 and 48 h after administration (p < 0.05). The drug concentration of TC–PLGA–PEG group in both liver and kidney was significantly lower than that in blood 1 h after administration (p < 0.05). Furthermore, as shown in Figure 11A, there were no obvious abnormalities in the hearth, spleen, liver, lung or kidney. Blood biochemistry analysis was also confirmed that both NPs had no obvious organ toxicity compared with the control group (Figure 11B).