The chemicals used include tetraethoxysilane (TEOS), cetyltrimethylammonium bromide (CTAB), methanol, sodium hydroxide (NaOH), manganese chloride tetrahydrate, H₂O₂, sodium oleate, n-hexane, chloroform, ethyl acetate, 1-octadecene, HA (M.Wt: 10,000–18,000 Da), (3-aminopropyl)triethoxysilane (APTS), and O-(benzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate (HBTU). All chemicals were purchased from Sigma Chemical Co.

The synthesis of MnO nanoparticles is described as follows: 1.24 g of the Mn–oleate complex (2 mmol) was dissolved in 10 g of 1-octadecene. The resulting solution was degassed at 70°C for 1 h under vacuum and heated to 300°C with vigorous stirring. The reaction mixture was maintained at this temperature for 1 h to induce sufficient growth. The solution was then cooled to room temperature, and 20 ml of hexane was added to improve the dispersibility of the nanoparticles, followed by adding 80 ml of acetone to precipitate the nanoparticles. The precipitate was obtained by centrifugation. The above purification procedure was repeated two more times to remove the excess surfactant and solvent.

MnO@MSN was prepared using the following procedure. The MnO nanoparticles stabilized with oleic amine were dispersed in chloroform at a concentration of 12.8 mg Mn/ml. Next, typical mesoporous silica coating onto MnO nanoparticles was performed using a sol–gel reaction of TEOS in an aqueous solution containing CTAB and MnO nanoparticles stabilized with the oleic amine. First, 1 ml of MnO nanoparticles in chloroform was poured into 5 ml of 0.05-M aqueous CTAB solution, and the resulting solution was stirred for 1 h, forming an oil-in-water microemulsion. The mixture was then heated to 70°C to evaporate chloroform. Next, the resulting transparent solution of MnO/CTAB was added to a mixture of 40 ml of water and 1.4 ml of 2-M NaOH solution, and the mixture was heated to 60°C. Then, 0.25 ml of TEOS and 1.8 ml of ethylacetate were added to the reaction solution in sequence, and the reaction was continued for 2 h. The washing steps for MnO@MSN nanoparticles with ethanol were performed to remove unreacted species, and then, the nanoparticles were redispersed in 5 ml of ethanol.

For HA coating, we modified the outer surface of MnO@MSN with primary amine groups (MnO@MSN-NH₂) using APTS. Then, 20 mg of HA and 30 mg of HBTU were added to 1 mg of MnO@MSN-NH₂ in the phosphate buffered saline solution. The mixture was stirred at room temperature for 4 h, and then, the resulting HA-MnO@MSN was collected and washed with ethanol and water using a centrifuge (12,000 rpm × 3).

The morphology of the samples was characterized using a transmission electron microscope (TEM) (Hitachi, H-7650), operating at an accelerated voltage of 80 kV. Fourier transform infrared spectroscopy (FTIR) was recorded on a Nicolet 550 spectrometer using KBr pellets (approximately 1 mg of the sample was pressed with 300 mg KBr). ZetaSizer Nano was used to measure the hydrodynamic size of the nanoparticles. MRI acquisitions were performed on a 9.4 T magnet (Bruker-Biospin, Billerica, MA, United States) using a 35-mm volume quad-coil (Bruker-Biospin, Billerica, MA, United States). Mn concentrations were based on the molar concentration of manganese atoms measured using ICP-MS (Perkin Elmer Elan 6100).

T₁ relaxation times were calculated using a RAREVTR inversion recovery sequence. Ten experiments were performed with inversion times (TR) ranging from 150 to 10,000 ms, TE = 9.8 ms, matrix = 128 × 128, FOV = 0.30 mm, slice thickness = 0.30 mm, and NEX = 2. The specific relaxivity (r ₁) of the MnO nanoparticles was measured as follows. Each sample was prepared in five different concentrations, and T₁ values were measured for each concentration, which was then used for r ₁ calculations. Relaxivity was determined from the slope of concentration-dependent T₁ changes.

In vitro US imaging of MnO@MSN and HA-MnO@MSN was performed in 200-µM H₂O₂ solution. An agarose gel (3%, w/v) phantom was prepared using a 500 μl Eppendorf tube, and the tube was removed after the phantom gel had cooled. Nanoparticles in 200-µM H₂O₂ solutions (1 mg/ml) were prepared and placed in the agarose phantom, and the change in US intensity for each sample was measured up to 30 min using a homemade 128-channel high-frequency US platform (Vantage 128, Verasonics Inc., Washington, DC, United States). The entire US system was controlled using a custom-developed graphical user interface (GUI) based on MATLAB^® (R2007a, MathWorks Inc., Natick, MA, United States). The US signals were acquired using a high-frequency 18.5-MHz US transducer (L22-14v, Verasonics Inc., Washington, DC, United States).

To investigate the ability of oxygen-evolving property, 1 mg of MnO@MSN and HA-MnO@MSN were incubated with 200 μM of H₂O₂, and then, the O₂ concentration was measured using a dissolved oxygen meter (OX10, Unisense Instruments, Denmark). For the quenching experiments, HA-MnO@MSN (1 mg/ml) was added with H₂O₂ (10 mM) to initiate the reaction. The residual concentration of H₂O₂ was determined over time by measuring the absorbance of H₂O₂ at 210 nm. The continuous catalytic effect was verified by repetitive addition of 10 mM of H₂O₂ to HA-MnO@MSN (1 mg/ml) solution, followed by determining the concentration by measuring the absorbance.

In vivo tumor oxygenation saturation (sO₂) measurements were performed using a PA imaging system. For this purpose, three HCT116 tumor–bearing mice (size ranging from 600 to 750 mm³) from the National Laboratory Animal Center, Taiwan, were deeply anesthetized with isoflurane (1–4%) using an inhalation device and placed on a heating pad. Note that these three mice were used for proof-of-concept experiments in this study. Then, 500 µg of HA-MnO@MSN was injected intratumorally into the subcutaneous HCT116 tumor–bearing mice. The baseline image was acquired preinjection and following measurements at various time points postinjection. A 128-channel Verasonics high-frequency US platform (Vantage 128, Verasonics Inc., Washington, DC, United States) was employed for dual-modality imaging (both PA imaging and US imaging). The entire PA system was controlled using a custom-developed GUI based on MATLAB^® (R2007a, MathWorks Inc., Natick, MA, United States). In order to operate the system in the PA mode, laser excitation and data acquisition were synchronized using triggering. The excitation laser was a compact Nd:YAG-laser system with an integrated tunable optical parametric oscillator (OPO, SpitLight 600 OPO, InnoLas Laser GmbH, Krailling, Germany). The OPO generates approximately 7-ns duration pulses at a 20-Hz repetition rate with tunable wavelengths from 680 to 2,400 nm. The PA signals were acquired using a high-frequency 18.5-MHz US transducer (L22-14v, Verasonics Inc., Washington, DC, United States). This transducer has a −6-dB fractional bandwidth of 67% and 128 active elements. The acoustic waves were received, reconstructed, and displayed on a computer screen at a frame rate of 20 frames per second. The American National Standards Institute safety limit is 20 mJ/cm², and the incident energy density on the sample surface during PA imaging was estimated to be approximately 12 mJ/cm², which is within the safety limit. The sO₂ around the tumor was measured via the differential optical absorption of oxygenated and deoxygenated hemoglobin at different wavelengths of 850 and 750 nm, respectively. To facilitate the comparison of sO₂ patterns in different groups, the regions of interest in the tumor were employed in the proximity of the HA-MnO@MSN injection site and identified using US imaging. PA B-scans of mice generating averagely oxygenated and deoxygenated hemoglobin signals were analyzed using custom-developed software based on MATLAB^® (R2007a, The MathWorks, United States), and sO₂ is defined as sO₂ = [HbO₂]/[HbO₂] + [Hb]. A customized, precision 3D translation stage with motorized x-, y-, and z-axes was used to control the transducer to obtain A-scan, B-scan (i.e., two-dimensional; one axis is the lateral scanning distance, and the other axis is the imaging depth), and C-scan (i.e., three-dimensional) images. For in vivo imaging, the PA probe was immersed in an acrylic water tank with a rectangular cutout at the bottom serving as an imaging window. The cutout was sealed with a thin polyethylene film of 15-μm thickness. US gel (POC Medical, Inc., Zhongli City, Taiwan) or an agarose pad was then used to provide a coupling interface between the imaging window and the animal.

For in vivo MR imaging studies, cultured HCT116 cancer cells (2 × 10⁶ cells) were injected into the right thigh regions of mice (n = 3) to establish the tumors. Note that these three mice were used for proof-of-concept experiments in this study. After the tumors developed up to a size of approximately 200 mm³, mice were deeply anesthetized with isoflurane (1–4%) using an inhalation device and placed on a heating pad. Then, 500 µg of HA-MnO@MSN was injected intratumorally. MRI acquisitions were performed on a 9.4 T magnet (Bruker-Biospin, Billerica, MA, United States) using a 35-mm volume quad-coil (Bruker-Biospin, Billerica, MA, United States). T1 relaxation times were calculated using a RAREVTR inversion recovery sequence. The baseline image was acquired preinjection and following measurements at various time points postinjection. The MR parameters were set up as TR = 600 ms, TE = 10.5 ms, FOV = 4 × 4 cm, slice thickness = 1 mm, 20 slices, NEX = 8, and matrix size = 256 × 128 recovered to 256 × 256.

Oleic acid–stabilized MnO nanoparticles with an average diameter of 15 nm were synthesized via the thermal decomposition of the manganese–oleate complex (Na et al., 2007). To prepare the MSN shell, a hydrophobic oleic acid–capped MnO nanoparticle was transferred into an aqueous solution using CTAB. Then, a sol–gel-type condensation reaction with TEOS resulted in the formation of an MSN shell. Next, in order to prepare enzyme-responsive nanoparticles, we modified the outer surface of MSN with primary amine using APTS. This is because the repeating unit of HA contains carboxylic acid, which was chemically conjugated to primary amine-functionalized MSN in the presence of HBTU via amide bond formation. At last, CTAB was removed by extraction with acidic ethanol to generate a mesoporous silica shell. Figure 2 shows the TEM images of nanoparticles. MnO nanoparticles were highly monodisperse and spherical in shape, with an average diameter of 15 nm (Figure 2A). The TEM image (Figure 2B) of MnO@MSN shows the MSN coating of MnO with a distinct core–shell morphology, and HA-MnO@MSN particles exhibited a spherical or quasispherical shape with an average diameter of 50 nm (Figure 2C).

The surface coverage of HA on MnO@MSN was supported by the zeta potential results (Figure 3A), which increased from −18 (MnO@MSN) to 5 mV. When treated with the enzyme HAdase, the zeta potential of HA-MnO@MSN increased further to 37 mV. This might be due to the digestion of HA by HAdase revealing primary amine groups on the MSN outer surface. Figure 3B shows the FTIR spectra of MnO@MSN and HA-MnO@MSN. From the MnO@MSN spectrum, we can observe the characteristic peak of the silica structure at 760 cm ⁻¹ (Si–O stretching), 960 cm ⁻¹ (Si–OH stretching), and 1,200 cm ⁻¹ (Si–O–Si stretching). After conjugation with HA, the resulting HA-MnO@MSN exhibited similar peaks of silica and a notable sharp peak at 1,620 cm⁻¹, which corresponds to the COOH asymmetric stretching of HA, which confirms successful chemical conjugation. Western blot analysis was utilized to study the expression of the HAdase enzyme in HCT116 cancer cells. The analysis showed that HCT116 cells expressed both hyaluronoglucosaminidase 1 (HYAL1) and hyaluronoglucosaminidase 2 (HYAL2) (Figure 3C). HYALl and HYAL2 were present in cell lysate, which indicates that HCT116 cells express and secrete both HAdases.

The primary strategy to increase the r₁ relaxivity of MnO NP is to increase the availability of water molecules in closer proximity to the magnetic core. Because the major relaxation mechanism constitutes the dipole-dipole coupling between water protons and the manganese ions (Hsu et al., 2016). In the case of the core-shell-structured HA-MnO@MSN in this study, we achieved the r1 relaxivity increase via the structural modifications of the coating to enhance its water permeability. To test the use of MnO@MSN as a T₁ contrast agent, its longitudinal relaxivity was characterized using a 9.4 T MRI scanner in aqueous suspension. Figures 4A,B show a significant decrease in relaxation time for the incremental MnO concentrations. The molar relaxivity of CTAB-extracted MnO@MSN was determined to be 1.29 mM⁻¹s⁻¹, which was calculated by measuring the relaxation rate with increasing MnO concentrations. Relaxivity is defined as the change in the relaxation rate of water protons in the presence of contrast agents. The r ₁ value of CTAB-extracted MnO@MSN was significantly greater than the values of bare MnO (∼0.28 mM⁻¹ s⁻¹) and non-CTAB–removed MnO@MSN (∼0.108 mM⁻¹ s⁻¹). It should be noted that the increase in r ₁ value is mainly due to the MSN shell, which is consistent with previous reports (Kim et al., 2007; Kim et al., 2011). MSN allowed optimal access of water molecules through its nanochannels to the MnO core, thus effectively relaxing the nearby water protons. For non-CTAB–removed MnO@MSN with similar MnO concentrations, the presence of CTAB in the MSN nanochannel significantly affected the interaction between the MnO core and water, thus indicating poor T₁ relaxation.

In general, CTAB used as a structural directing agent in the synthesis of MSN must be completely removed from nanochannels for successful loading of guest molecules, such as anticancer drugs (Wu et al., 2013). In this study, CTAB was removed by acidic extraction with ethanol for 3 h. The presence of CTAB in MSN nanochannels could act as a barrier to the diffusion of water molecules to the MnO core, which will significantly affect MRI properties. In addition, CTAB is known to induce dose-dependent cytotoxicity (Wang et al., 2008; He et al., 2011; Schachter, 2013), and thus, removal of CTAB is essential for clinical applications. Hence, we evaluated relaxivity at various time intervals during CTAB extraction. As shown in Figure 5A, during CTAB extraction from 0 to 60 min, a steady decrease in the relaxivity of MnO@MSN was observed, followed by little or no significant decrease in the relaxivity up to 9 h. This indicates that CTAB was partially removed in 30 min, followed by the complete removal at 60 min. This result suggests that CTAB removal is crucial for the molar relaxivity of MnO@MSN.

Likewise, the molar relaxivity of HA-MnO@MSN was also investigated in the presence of HAdase (Figure 5B), an enzyme degrading the backbone of HA. Upon incubation with HAdase, the relaxivity of HA-MnO@MSN decreased remarkably, which might be due to the degradation of the HA backbone. This result indicates that HA-MnO@MSN can be used as an enzyme-responsive contrast agent for MR imaging. It is well known that HAdase is abundant in the cytosol of cancer cells. Hence, this unique behavior of HA-MnO@MSN may allow the development of site-specific drug delivery systems for cancer therapy.

Owing to their tunable structures and unique physiochemical properties, manganese oxide nanomaterials (MON) have drawn attention in various biomedical applications such as bioimaging, biosensing, and drug/gene delivery (Ding et al., 2020). Of late, the catalytic activity of these nanomaterials found significant applications as tumor microenvironment–responsive biomaterials. It is well known that the tumor microenvironment is characterized by hypoxia, mild acidity, and elevated production of H₂O₂ (Brown and Wilson, 2004). MON has the potential to alter the tumor microenvironment by catalyzing H₂O₂ to O₂, which can be utilized for bioimaging and anticancer therapies.

We investigated the generation of oxygen bubbles by HA-MnO@MSN using US imaging. Gas bubbles are excellent contrast agents for US imaging. However, gas- or air-filled bubbles suffer from poor stability in vivo, and their inability to target disease constitutes major challenges. In addition, tumor vasculature permeation is difficult because of their size (3–10 µm) (Paefgen et al., 2015). The formation of oxygen bubbles through the catalytic conversion of H₂O₂ to O₂ by HA-MnO@MSN could serve as an excellent US contrast agent due to the ability of gas bubbles to reflect US by generating strong signals. Figure 6A shows the US imaging of H₂O₂ alone and NP treated with H₂O₂. In the H₂O₂-only group, there was no US signal throughout the study because H₂O₂ alone could not generate bubbles. Although both HA-MnO@MSN and MnO@MSN showed strong US signal intensities, this was due to the catalytic conversion of H₂O₂ to O₂. The formation of oxygen as nanobubbles or microbubbles was able to strongly reflect US by generating strong signals. Quantitative analysis showed that the US signal generated from HA-MnO@MSN was slightly weaker than that from MnO@MSN, and this might be due to the presence of HA coating delaying the access of H₂O₂ to reach the MnO core to start the catalytic reaction (Figure 6B). We observed the direct formation of oxygen bubbles through optical imaging under the same conditions to support US imaging. The formation and growth of oxygen bubbles can be clearly seen in both MnO@MSN and HA-MnO@MSN treated with H₂O₂ (Figure 6C).

We further tested whether our NP can generate a sufficient amount of oxygen at a low H₂O₂ concentration using an oxygen electrode. As anticipated, a significant amount of oxygen was generated by both MnO@MSN and HA-MnO@MSN (Figure 6D). Then, we evaluated the catalytic effect of HA-MnO@MSN by measuring residual H₂O₂ after the addition of HA-MnO@MSN, and we found that H₂O₂ was quenched by HA-MnO@MSN (Figure 7A). One of the important features of HA-MnO@MSN is its capability to continuously generate O₂. The continuous catalytic activity of the nanoparticles was maintained, even after the repetitive addition of H₂O₂ (Figure 7B).

Therapeutic approaches such as radiotherapy, photodynamic therapy, and sonodynamic therapy exert their therapeutic effects via copious ROS generation to kill cancer cells (Yang et al., 2019; Wang et al., 2021a). One of the important pitfalls of these therapeutic approaches is the presence of low oxygen tension in the tumor that results in low or partial therapeutic effects, which encourages the residual tumor mass to migrate to various organs, resulting in metastasis. As a consequence, supplemental tumor oxygenation is imperative for better therapeutic outcomes. The rationale for supplemental tumor oxygenation is that the resulting increase in arterial pO₂ will enhance the diffusion of soluble oxygen into tissues. For instance, carbogen breathing has been shown to improve the oxygenation of both experimental and human tumors. Carbogen is a normobaric high-oxygen-content gas mixture (95% O₂ with 5% CO₂ or 98% O₂ with 2% CO₂) that increases intravascular oxygen availability, resulting in greater oxygen uptake by tumors (Alonzi et al., 2009). Moreover, various nanomaterials have been recently used to supply oxygen to tumors. These nanomaterials catalytically generate oxygen by consuming hydrogen peroxide in the tumor microenvironment (Zhang et al., 2019; Liu et al., 2020).

Herein, synthesized MnO@MSN will serve as an oxygen generator by catalase-like activity. From the in vitro results, which demonstrated the ability of HA-MnO@MSN to decompose H₂O₂, as a consequence, continuous oxygen generation was proved. These results encouraged us to test the capability of our nanoplatform to modulate oxygen saturation in blood inside tumors by PA imaging based on the differential absorption characteristics of oxygenated and deoxygenated hemoglobin. Figure 8A shows the representative sO₂ map of HCT116 tumors before and after i. t. injection of HA-MnO@MSN 3. The sO₂ map of the tumor before NP treatment shows the presence of a hypoxia core. After NP injection, the hypoxia level in the tumor core was reduced gradually with an increase in sO₂ over time due to the catalytic activity of HA-MnO@MSN-3 by converting H₂O₂ to O₂. The observed pattern might be due to the continuous catalytic conversion of H₂O₂ by HA-MnO@MSN-3 to generate O₂ up to 150 min and decreased because of the insufficient levels of H₂O₂ for O₂ generation.

Figure 8B shows the quantitative analysis of relative tumor sO₂ values from three different experiments after NP injection. HA-MnO@MSN 1 and 2 showed similar tumor sO₂ trends. For both HA-MnO@MSN 1 and 2, sO₂ steadily increased from 9% at 10 min to reach a maximum of 20% (actual sO₂, 69%) (HA-MnO@MSN 1) and 15% (actual sO₂, 68%) (HA-MnO@MSN 2) at 30 min and then decreased to 0.5 and 8% at 90 min. Next, HA-MnO@MSN 3 demonstrated a unique trend in sO₂. After i. t. injection, sO₂ increased to 2% (actual sO₂, 42%) at 10 min, which is significantly less compared to the sO₂ values of HA-MnO@MSN 1 and 2, and slowly reached a maximum of 15% (actual sO₂, 57%) at 150 min, followed by a slight decrease. This relative sO₂ pattern is inconsistent with actual sO₂. As previously explained, the presence of a hypoxic core influences the sO₂ pattern, and tumor hypoxia is characterized by abnormal vasculature, which hinders the supply of nutrients (Brown and Wilson, 2004). Here, HA-MnO@MSN 3 after injection was not able to diffuse in all regions of the tumor because of defective vasculature. Hence, only a few NP had a chance to interact and be digested by HAdase, followed by H₂O₂ diffusion to the NP core to generate O₂ slowly.

The relaxivity of traditional MnO-based core–shell structures depends on the interaction of MnO with water molecules based on the porosity of the coating material. Herein, HA as a coating material in HA-MnO@MSN would be digested by tumor-specific hyaluronidase to allow the optimal accessibility of the water molecules to the MnO core through nanochannels of MSN. Our in vitro results encouraged us to test the potential of HA-MnO@MSN for MR imaging of a tumor, and we administered an intratumoral injection of HA-MnO@MSN and MnO@MSN dispersion to HCT116 human colon tumor-xenograft-bearing nude mice and monitored MR images as a function of time (Figures 9A,B). For HA-MnO@MSN, immediately after injection, we did not observe any contrast enhancement at the tumor site. However, enhancement in T₁ weighted images was observed at 5 h postinjection. In addition, this contrast enhancement lasted up to 48 h. The T₁ contrast enhancement was consistently supported by in vitro data (Figure 5B), which is due to the time-dependent enzymatic digestion of the HA backbone by HAdase. For MnO@MSN, T₁ contrast enhancement was observed at 2 h postinjection (data not shown) and slowly decreased after 5 h. This might be a result of the rapid interactions between water molecules and the MnO core due to the lack of HA coating. Thus, MnO@MSN exhibited a rapid but transient T₁ signal.

To further confirm that HA-MnO@MSN indeed accumulated in the tumor, we used TEM to examine tissue samples obtained from animal models after imaging at various time intervals. As shown in Figures 10A–D, many HA-MnO@MSN (black dots) were clearly seen in tumor samples from 10 min followed by up to 2 h. Then, the slow clearance of nanoparticles from tumor was observed at 24 h. This trend is consistent with ICP-MS data (Figure 10E).