Tetrabutyl titanate (TBT) was purchased from Kermel (China). HAc was purchased from Xilong chemicals (China). DOX was purchased from Sigma-Aldrich (China). Carboxyfluorescein and Quercetin were purchased from Aladdin (China). Congo red was purchased from BASF (China). Sudan Ⅳ and Rhodamine B were purchased from Guoyao Reagents (China). RPMI-1640 culture medium and fetal bovine serum (FBS) was purchased from Gibco (United States). All other reagents, including ethanol, HCl, K₂HPO₄, NaH₂PO_4, and NaOH were purchased from Xilong chemicals (China). DOX, Carboxyfluorescein, Quercetin, Congo Red, Sudan Ⅳ, and Rhodamine B stock solutions were freshly prepared and stored in the dark at 4°C. The ultrapure water was used in all experiments.

The morphology of TiO₂ has an important influence on its drug loading and release. The TiO₂ particles were produced with relatively uniform size as shown in Figure 1A. After zooming in, they are pompon-like spheres with a diameter of ∼2.0 µm. The highly porous structure of TiO₂ spheres can be observed from the scanning electron microscope and transmission electron microscopy images with higher magnification (Figures 1B–D). Chemical composition analysis by Energy dispersive spectroscopy (Figure 2) illustrates that the porous spheres are composed of titanium and oxygen elements. The same synthesis method can sometimes form different morphology of TiO₂, which is mainly related to some influencing factors. The common influencing factors are temperature, pH value, adding certain ions, reaction time, and so on. At present, the morphologies of TiO₂ are spherical, microsphere, hollow sphere, flower, nanofiber, nanotube, flake, rod, etc. At this research, the pompon like TiO₂ microspheres were synthesized by using ethanol as solvent at 200°C and the structure has not been reported before. T i ( O C 4 H 9 ) 4 + x C H 3 COOH → (C H 3 COO) x Ti ( O C 4 H 9 ) 4 − x + x C 4 H 9 OH (1) C 4 H 9 OH+C H 3 COOH → C H 3 COO C 4 H 9 + H 2 O (2) ≡ Ti − OR + H 2 O → ≡ Ti − OH + ROH ( R = C 4 H 9   or   CH 3 CO ) (3) ≡ Ti − OR ′ + ≡ Ti − OR " → ≡ Ti − O − Ti ≡ R ′ OR " ( R ′ = H   or   CH 3 CO ; R " H   or   C 4 H 9 ) (4)

A tentative mechanism for the formation of anatase TiO₂ mesocrystals was proposed as illustrated in Route 1. In the current solvothermal synthesis at 200°C, the reaction between Ti(OC₄H₉)₄ and HAc first resulted in the coordination of HAc to titanium centers to form unstable titanium acetate complexes (CH₃COO)xTi(OC₄H₉)₄-x by ligand exchange/substitution, concomitant with the release of C₄H₉OH (Eq. 1). The produced C₄H₉OH could then react with the solvent HAc to form water by a slow esterification reaction (Eq. 2). Subsequently, Ti-O-Ti bonds would form by both hydrolysis-condensation and nonhydrolytic condensation processes (Eqs 3,4) leading to the formation of crystallized pompon-like TiO₂. In the current situation, the nascent anatase nanocrystals with high surface energies might be temporarily stabilized by acetic acid, which could then orient each other to form a crystallographic oriented mesocrystalline architecture with some butyl acetate enwrapped in the interparticle pores. Finally, nanoporous anatase TiO₂ mesocrystals with preserved morphology will result from removal of the organic residuals by subsequent calcination, accompanied by a moderate increase in the size of the nanocrystal subunits. During the mesoscale assembly of the pompon-like-shaped, anatase mesocrystals with a single-crystal-like structure, the solvent acetic acid played multiple roles. First, it acted as the chemical modifier of TBT to lower its reactivity. Second, it reacted with TBT to form transient or metastable precursors for slow release of soluble titanium-containing species for the continuous formation of nascent anatase nanocrystals. Third, it acted as the stabilizing solvent to temporarily stabilize the tiny anatase nanocrystals against immediate single crystal formation or uncontrolled aggregation, but the binding to the {001} facets was relative weak, so that the subsequent oriented attachment along the [001] direction could take place readily. Fourth, it reacted with TBT to produce butyl acetate that played the role of porogen or template during the oriented aggregation of anatase nanocrystals to form mesocrystals (Ye et al., 2010).

It can be seen from the Figure 3 that both have the characteristic diffraction absorption peak of Ti: 2θ = 35.2°, 38.6°, 40.3°, 53.1°, 63.0°, this is caused by the titanium substrate. Different from Figure 3A, new peaks appears at 2θ = 25.2°, 37.9°, 47.8°, 53.9°, 54.9° in Figure 3B. This is the characteristic absorption peak of TiO₂ anatase, which may be due to the transformation of TiO₂ from amorphous to anatase after hydrothermal treatment.

In order to clarify the relationship between the carrier and the drug, we used FT-IR spectroscopy to detect the sample. The FT-IR spectra of TiO₂, DOX–TiO_2, and pure DOX are presented in Figure 4. The bands at 3,425 is attributed to the stretching mode and the “scissor”-bending mode of −OH, respectively, which come from H₂O absorbed on the surface of the products and −OH on the surface of TiO₂ (Chen et al., 2009). The absorption band at 690 cm⁻¹ is attributed to Ti–O stretching vibration (Chen et al., 2009; Cui M. et al., 2012). The bands at 1641 and 1500 cm⁻¹ may attributed to C = O and C-O-C of by products of the reaction as route 1 showed (Ye et al., 2010). Two characteristic absorption peaks at 1619 and 1580 cm⁻¹ are ascribed to the stretching vibration of C = O from the anthraquinone ring of DOX. The peaks at 993 and 2930 cm⁻¹ are assigned to the C–O–C and CH₃ stretching vibration of DOX, respectively. The couple of the peaks are also observed in DOX–TiO₂, and no new bands are observed. According to the results, it can be speculated that DOX is loaded onto TiO₂ surface mainly through noncovalent bond.

It is accepted that the DOX molecules are entrapped within the pores by immersing process and liberated via a diffusion-controlled manner. In addition, the hydroxyl, carboxyl, and carbonyl groups in drug molecules can form hydrogen bonding with Ti–OH groups on the porous TiO₂ surface. At the drug release stage, PBS solution carry the drugs out of TiO₂ pores by dissolving the drugs. Before the dynamic concentration equilibrium is achieved, the continuous drug release is conducted by diffusion-controlled manner which maily depends on the dynamic equilibrium between the drug concentration inside the pores and the drug concentration in the bulk solution in PBS. Thus, a sustained drug release stage is reached.

Figure 6A shows the drug-release result of DOX–TiO₂ in simulated body fluid and in cells. As for drug-release research in vitro, the process of releasing DOX molecules seen from Figure 6A can be divided into two stages: fast release stage for the first 3 h and continuing release stage in the following 21 h. Certain amount of drug molecules can be absorbed onto the surface of TiO₂ spheres via physical effects. However, due to the weak absorption of the drug molecules, the drugs will quickly dissolve into buffer solution and initiate the fast release stage once contacting with buffer solution. After that, drugs absorbed into pores or on the surface of TiO₂ spheres by strong hydrogen bonds continue slowly releasing, which induces the second stage and lasts for 21 h. Obviously, TiO₂ spheres has the ability of drug storage and slow release. Besides, the initial fast release stage of DOX molecules is important for the treatment of cancer since the amount of released drugs is larger, which can rapidly inhibit multiplication of cancer cells. The rest of cancer cells will gradually die during the process of continual drug releasing. Therefore, the system of DOX–TiO₂ is promising in cancer treatment. This release behavior is similar to the drug release behavior in TiO₂ nanotubes, both of which are rapid release at the beginning and slow release afterwards. However, compared with titanium dioxide nanotubes, because the surface area of nanotubes is larger and the more drug is adsorbed (Figure 5), its rapid release period and slow release period are longer (Moseke et al., 2012). We fitted the drug release data and found that the drug release was suitable for the first-order kinetic equation, the fitted equation is:

ln (1-M_t/M_∞) = −0.3369t−0.4369, the correlation coefficient R = 0.99

This indicates that the drug release is completed by DOX dissolution and desorption, and that the drug loading is completed by adsorption without chemical reaction, which is consistent with the previous analysis; The amount of drug release per unit time was less and less with the extension of time.

To visualize the cellular uptake of TiO₂ spheres and the intracellular distribution of DOX, the samples are readily characterized by optical microscopy. After incubation for 24 h at 37°C, observations reveal that nearly all cells contain DOX–TiO₂ spheres, which shows that microspheres can be taken in by HeLa cell, similar as previous reports (Cui J. et al., 2012; Chen et al., 2014). Accumulation of red fluorescence in the cell nuclei is apparent to observe, which shows that nearly all of the nuclei contains DOX. DOX-associated fluorescence images are also presented in the cytoplasm (Figures 6C–E), and this suggests that DOX has been released from DOX–TiO₂ samples.

MTT is applied to analyze and study the cell activity, and the result is showed in Figure 6B. Besides that, pure DOX at the same concentration is also added into the medium to culture cells, and the cell viability is detected by MTT. The results indicate that within the limits of consistency the activity of cells incubated by TiO₂ spheres culture medium maintains 100%, which manifests that pure TiO₂ spheres has great biocompatibility. That is to say that this kind of material is nontoxic, and can be applied to the biomedical field. For the cells incubated in DOX-TiO₂ medium, at low concentration cell viability of the system is higher than that of cells incubated in pure DOX medium, while at high concentration cell viability of the system is higher than that of cells incubated in pure DOX medium, which may be due to the different modes to enter cells. The mode of DOX to enter cells is the diffusion, and the mode of DOX–TiO₂ to enter cells is endocytosis. So the speed of DOX’s entering cells is faster than that of DOX–TiO₂, which accounts for the higher cytotoxicity of DOX at low concentration. At high concentration, DOX–TiO₂ is transferred into cancer by endocytosis, and then DOX is released gradually to kill the cancer cells.