# Sequential drug release via chemical diffusion and physical barriers enabled by hollow multishelled structures

Sep 7, 2020

### Characterization of MIT–TiO2–HoMS

TiO2–HoMS samples with different shell numbers were fabricated through STA20,23 by adjusting the adsorption conditions of metal ions, and then MIT was loaded by a typical drug-loading process24. Transmission electron microscopy (TEM) images (Supplementary Fig. 2a, b and Fig. 2a) show that various samples with different shell numbers were fabricated. Based on the statistical analysis of more than 100 TiO2–HoMS particles for each sample, the average size of TiO2 hollow spheres and double-shelled (2s-) and triple-shelled (3s-) TiO2–HoMS is estimated to be 726 ± 47 nm, 642 ± 30 nm, and 583 ± 35 nm with a narrow size distribution, respectively (Supplementary Fig. 3). The average shell thickness and shell spacing are also given (Supplementary Table 1). The high-resolution TEM images demonstrate the high crystallinity and random distribution of the anatase and rutile TiO2 nanocrystals in the shells (Supplementary Fig. 2d)12. Furthermore, X-ray diffraction confirms that TiO2–HoMS is a composite of the anatase phase (JCPDS card No. 21-1272) and rutile phase (JCPDS card No. 21-1276) (Supplementary Fig. 2e)25. TEM-EDX mapping images of MIT-3s-TiO2–HoMS show that S is evenly distributed on the shell of TiO2–HoMS, indicating uniform drug loading (Fig. 2c, d). The cryo-TEM image shows a uniform contrast in HoMS after MIT loading, further indicating the successful loading of MIT (Supplementary Fig. 2f).

### MIT molecule absorption in HoMS

Fourier transform infrared (FTIR) spectra (Fig. 2e) were collected after different MIT loading periods to study the interaction between MIT and TiO2–HoMS. The peaks in the FTIR spectrum of MIT-3s-TiO2–HoMS at 714, 1418, 1618, and 1640 cm−1 could be ascribed to C–S (stretching), CH3 (bending), C=C, and C=O (stretching) in MIT26, respectively (Fig. 2e). Interestingly, a redshift is observed for the C=O stretching band with extension of the drug-loading period, indicating different adsorption modes during MIT loading (Fig. 2f).

The thermal release of MIT through TiO2 hollow spheres and 2s- and 3s-TiO2–HoMS occurred in three stages, as exhibited in the thermogravimetric analysis (TGA) and differential thermal analysis (DTA) results (Fig. 3a–e). The first stage occurred at ~50 °C, with a weight loss of ~5% and accompanied by a endothermic process, which corresponded to the evaporation of water in this MIT–TiO2 system27,28. The second stage was in the range of 70–150 °C, which corresponded to the desorption of MIT loaded on the outside of the outer shells of HoMS and the MIT in cavities between shells. The third stage took place after 150 °C, and was associated with the disintegration of a small amount of MIT strongly bound to the shells of TiO2–HoMS by hydrogen bonds29. Notably, the second peak of the DTA curve of pure MIT was at 150 °C, while the peak temperature decreased to 126, 127, and 130 °C after MIT was loaded into the TiO2 hollow spheres and 2s- and 3s-TiO2–HoMS, respectively (Fig. 3f). Furthermore, according to the Speil theory30, the relative value of the endothermic heat can be calculated by integrating the DTA curve and then normalizing to the mass of MIT (Fig. 3g). The relative values are 65.28, 43.01, 43.38, and 43.57 kJ mg−1 for pure MIT, MIT–TiO2 hollow spheres and MIT-2s- and MIT-3s-TiO2–HoMS, respectively. It can be noted that after loading MIT into HoMS, a smaller driving force is needed for molecule release. The MIT loading capacity can be calculated as 0.2274, 0.3000, and 0.3292 (normalized to the weight of TiO2) for TiO2 hollow spheres and 2s- and 3s-TiO2–HoMS, respectively. This result indicates that building up more surface in a single hollow particle is helpful for drug adsorption. In this case, an increased number of shells improved the drug-loading capacity, proving that HoMS is a good candidate for mass loading. Notably, when we used SBA-15 as the carrier, the MIT loading amount was calculated as 0.3888 after normalization to the weight of SiO2, which is higher than the values obtained using TiO2 as a carrier (Fig. 3h). However, the specific surface area loading capacity (Fig. 3i) of 3s-TiO2–HoMS is surprisingly 46.5 times higher than that of SBA-15 owing to the much larger effective surface area of HoMS (Supplementary Table 2). To investigate the type of absorption between MIT and TiO2–HoMS, the Langmuir (first equation) and Freundlich isothermal adsorption models (second equation) were used to fit the experimental data31:

$$frac{1}{{q_e}} = frac{1}{{Q_0}} + frac{1}{{Q_0K_Lc_e}},$$

(1)

where ce = equilibrium concentration of adsorbate (mg L−1), qe = amount of MIT adsorbed per gram of carrier at equilibrium (mg g−1), Q0 = maximum monolayer coverage capacity (mg g−1), and KL = Langmuir isotherm constant (L mg−1):

$$lg,q_e = lg,K_f + frac{1}{n}lg,c_e,$$

(2)

where Kf = Freundlich isotherm constant (mg g−1), n = adsorption intensity, ce = equilibrium concentration of adsorbate (mg L−1), and qe = amount of MIT adsorbed per gram of carrier at equilibrium (mg g−1). The R2 values of the Langmuir model and the Freundlich model for MIT adsorption on TiO2–HoMS are 0.3685 and 0.9931, respectively (Supplementary Fig. 4). Therefore, the adsorption of MIT on TiO2–HoMS can be considered as multimolecular layer adsorption with Freundlich model.

### Sequential drug release

Taking the release of MIT, a model antibacterial compound, for characterization purposes, the mass release performance was tracked by UV–Vis spectrophotometer (Supplementary Fig. 5). The MIT release process shows roughly three stages (Fig. 4a). After the MIT-carrier composite entered the solution, a quick increase in drug concentration was observed at the first stage (~4 h), which is called burst release. Afterwards, the mass release speed slowed, and the MIT concentration remained stable at the second stage, which is named sustained release. 2s- and 3s-TiO2–HoMS and SBA-15 all have the capacity to maintain the concentration in the second stage (Fig. 4a). Notably, when carrying the same drug amount, 3s-TiO2–HoMS was found to be the most effective in inhibiting the growth of Escherichia coli (E. coli) when we continually introduced bacteria into the system and could inhibit bacterial growth even after 432 h (Fig. 4b). The presented fluorescence images32 show that on the 10th day, the bacterial viability was 66%, 58%, 33%, 0%, and 11% for MIT, MIT-loaded TiO2 hollow spheres, MIT-2s-TiO2–HoMS, MIT-3s-TiO2–HoMS, and MIT-SBA-15, respectively (Fig. 4c and Supplementary Fig. 6). Moreover, all the TiO2 carriers show good stability, which did not present any morphological changes even after 30 days (Supplementary Fig. 7).

Interestingly, this stimulus-responsive release, the third stage of sequential release, is a unique feature of the HoMS system. After the concentration of MIT stabilized at ~60 ppm (slightly higher than the minimum inhibitory concentration (MIC)) with the same total drug amount for various samples (Supplementary Fig. 8), bacteria were added to the solution to investigate the responsive release performance. 2s- and 3s-TiO2–HoMS and SBA-15 all presented responsive release performance, i.e., after the rapid decrease in the concentration of MIT, equilibrium was gradually recovered. 3s-TiO2–HoMS showed the best recycling performance among all the samples and maintained the concentration over the MIC after 14 rounds of addition of bacteria. Impressively, the recovery process was different for various samples. 3s-HoMS showed a smaller drop (9.8%) than 2s-HoMS (17.3%) with the same amount of bacteria added (Fig. 4f), indicating that the MIT molecules reserved in 3s-HoMS are easier to release under the stimulating conditions. In comparison, the MIT-loaded TiO2 hollow spheres and pure MIT do not show the ability of responsive release (Fig. 4e). MIT-loaded SBA-15 has excellent sustained-release performance; however, its responsive release is not as good as that of 3s-TiO2–HoMS. SBA-15 shows the largest drop of 18.1% after adding bacteria (Fig. 4f), and it cannot reach the initial concentration during the recovery stage, even after 20 h.

In addition to responding to bacteria, MIT–HoMS can also respond to pH changes (Supplementary Fig. 9a). A low pH value can change the zeta potential of TiO2 from −19.96 to 0.42 mV, weaken TiO2–MIT interactions and further affect MIT–MIT interactions; thus, more MIT molecules can be easily released (Supplementary Fig. 9b).