Gossypol‐Crosslinked Nanoclusters of Ultrasmall Iron Oxide Nanoparticles for Ultrasound‐Enhanced Precision Tumor Theranostics

Development of tumor microenvironment (TME)‐responsive nanomedicines with simple components for precision tumor theranostics still maintains a great challenge. Here, the design of an intelligent nanocluster (NC) system assembled from gossypol‐mediated crosslinking of phenylboronic acid‐modified ultrasmall iron oxide nanoparticles (USIO NPs) is presented. The gossypol functions as both a chemotherapy (CT) drug and a crosslinker through phenylborate ester bonds that are sensitive to both reactive oxygen species (ROS) and pH. The developed gossypol‐USIO NCs (for short, G‐USIO NCs), having a size of 34.2 nm, possess stability under physiological conditions, enable intracellular ROS generation and glutathione depletion to modulate TME, promote apoptosis of cancer cells in vitro, and inhibit tumor/lung metastasis in vivo through gossypol‐mediated CT and USIO‐mediated chemodynamic therapy owing to the ROS‐ and pH‐triggered dissociation of the NCs to release gossypol and Fe at the tumor site. Likewise, the dissociated USIO NPs from the NCs afford TME‐facilitated T1‐weighted magnetic resonance (MR) imaging. Furthermore, the effects of the tumor T1 MR imaging and the combination therapy can be elevated by ultrasound‐targeted microbubble destruction‐induced cavitation and sonoporation. The designed G‐USIO NCs with simple ingredients are likely developed to be a promising theranostic nanomedicine formulation for ultrasound‐facilitated precision theranostics of various tumor types.


Introduction
Currently, chemotherapy (CT) remains one of the main clinical treatments for cancer. The limitations of smallmolecule chemotherapeutic drugs, such as insufficient water solubility, low bioavailability, and toxic effects on normal tissues, lead to drug resistance, poor efficacy, metastasis, and undesirable side effects. [1] Various nanoplatforms have been developed for efficient drug delivery to the tumor site to enhance its therapeutic efficacy and reduce its toxic side effects including inorganic nanosystems, [2] metal-phenolic networks, [3] micelles, [4] liposomes, [5] dendrimers, [6] and nanogels. [7] However, few nanoplatforms are available for clinical investigation mainly due to the defects including poor drug loading efficacy, potential systematic toxicity and immunogenicity, multi-step synthetic processes, and complex components and structure. [8] Moreover, owing to the heterogeneity and complicated nature of tumors, a single-mode treatment appears inadequate. Therefore, it is of vital www.advancedsciencenews.com www.advsensorres.com significance to develop tumor microenvironment (TME)responsive drug delivery systems with simple components for efficient tumor combination therapy and reduction of systemic toxicity. [9] Unlike normal tissues, many abnormal biochemical indicators in TME have been found such as acidity, excess glutathione (GSH), high-content reactive oxygen species (ROS), over-expressed enzymes, and hypoxia. [10] On the one hand, these abnormalities contribute to the heterogeneity and complexity of tumors; on the other hand, these can be regarded as tools to improve tumor treatment efficacy and reduce undesirable side effects. [11] Therefore, various TME-responsive nanoplatforms have been designed to achieve precise diagnosis and efficient treatment of tumors. [12] Interestingly, the phenylborate ester (PBAE) bond, a dual pH-and ROS-responsive chemical bond, has been used to design intelligent nanoplatforms for the responsive release of drugs and imaging agents in the tumor site. [13,14] For example, our group designed and synthesized a metal-phenolic-network-coated dendrimer-drug conjugate to achieve dual pH-and ROS-stimulated toyocamycin release at the tumor site through the PBAE-bonded conjugation of acetylated generation 5 poly(amidoamine) dendrimers with toyocamycin containing catechol. [6] The design of nanoplatforms with both imaging and therapy functions has shown considerable advantages for the precision theranostics of tumors. [15] Among them, the citric acidstabilized ultrasmall iron oxide nanoparticles (USIO NPs) exhibit easy surface functionalization features, good colloidal stability, homogeneity, biocompatibility, T 1 relaxivity, and ROSgeneration ability. [16,17] Meanwhile, nanoclusters (NCs) formed by crosslinking of USIO NPs can be endowed with T 2 magnetic resonance (MR) imaging potential, and be further dissociated into single USIO NPs with T 1 MR imaging property after TMEbased stimuli triggering to achieve dynamic T 2 /T 1 MR imaging or/and precision therapy of tumors. [18] Previously, our group developed cystamine-crosslinked USIO NCs to achieve T 2 /T 1switchable MR imaging of tumors. [19] Unfortunately, the USIO NP-mediated chemodynamic therapy (CDT) and the disulfide bond-induced GSH depletion were not addressed in that work. In very recent work, through the assistance of microfluidics, intelligent stimuli-responsive cancer cell membrane (CCM)-and polydopamine-coated USIO NCs loaded with cisplatin (FDPC NCs) was developed in our group to improve tumor diagnostic accuracy and therapeutic efficacy. [20] The FDPC NCs enabled dynamic T 2 /T 1 -switchable MR imaging-guided combination photothermo-chemo-CDT of a xenografted 4T1 tumor by virtue of the GSH-and pH-induced disassembly of the FDPC NCs to release both cisplatin and Fe. Unfortunately, the complex components of FDPC NCs may be a major obstacle to clinical applications.
As is known, drug accumulation at the tumor site and treatment efficacy can be largely limited by the dense extracellular matrix and high interstitial fluid pressure of tumor tissues. [21] Ultrasound-targeted microbubble destruction (UTMD), a noninvasive ultrasound stimulation technology, can improve capillary and cell permeability for efficient tumor drug delivery through the created cavitation and sonoporation effect. [22] In brief, the high amplitude oscillation of the pre-injected microbubbles in vivo under ultrasonic irradiation can increase the interstitial space of endothelial cells and the permeability of CCM. [23] Furthermore, high-speed jets caused instantly by the fragmentation of microbubbles can lead to the appearance of large pores in the cell membrane, facilitating drugs to break through CCM. [24] Thus, UTMD has been considered as a promising technology to enhance the delivery of chemotherapeutic drugs, [25] genes, [26] and imaging agents to tumors. [27] For instance, our earlier work showed that under the assistance of UTMD, the delivery of doxorubicin could be improved via the developed CCM-coated pH-responsive NCs for precision tumor theranostics. [28] Here, we aimed to develop dual-responsive USIO NCs with simple components as an intelligent nanoformulation for ultrasound-enhanced tumor precision theranostics (Figure 1). The citric acid-passivated USIO NPs were solvothermally prepared, surface-aminated by ethylenediamine (EDA) to form USIO-NH 2 NPs, modified with 4-(bromomethyl)phenylboronic acid (PBA) to form USIO-PBA NPs, and then crosslinked by gossypol to create the intelligent gossypol-USIO NCs (G-USIO NCs) through PBAE bonds with both ROS-and pHresponsiveness. The obtained G-USIO NCs were thoroughly investigated to disclose their composition, structure, morphology, ROS-and pH-responsiveness, ROS generation capacity, and T 1 MR relaxometry. Furthermore, the G-USIO NCs were explored for T 1 MR imaging and combination CT/CDT of a subcutaneous 4T1 tumor (a murine breast cancer model) in vivo under the UTMD. The major superiorities of the G-USIO NCs lie in the following aspects: 1) Gossypol acts as both a chemotherapeutic drug and a crosslinker, 2) the PBAE bonding adopted to create the G-USIO NCs affords them with dual ROS-and pH-responsive release of gossypol and Fe for enhanced precision tumor theranostics, and 3) the G-USIO NCs formed using simple ingredients enable ultrasound-enhanced T 1 MR imaging and combination CT/CDT of tumors.

Synthesis and Characterization of G-USIO NCs
The citric acid-passivated USIO NPs with a mean diameter of 2.8 nm were first solvothermally prepared and characterized by transmission electron microscopy (TEM) (Figure 2A). [17] After modification with EDA, the aminated USIO NPs were further reacted with 4-(bromomethyl)PBA through a substitution reaction. Then, through forming the PBAE bond between phenylborate acid moiety on the surface of the USIO NPs and the catechol of gossypol, [29] the dual ROS-and pH-responsive G-USIO NCs were formed. To optimize the preparation of G-USIO NCs, the molar ratio between the PBA of USIO-PBA NPs and gossypol was optimized to be 1:15, which could attain the desired gossypol encapsulation efficiency (79.7%) and loading content (24.1%) (Table S1, Supporting Information). As shown in Figure 2B and Figure S1, Supporting Information, the irregularly shaped G-USIO NCs via gossypol crosslinking can be visualized by TEM. These NPs and NCs were also characterized via other techniques. Dynamic light scattering data reveal that the USIO NPs, USIO-NH 2 NPs, USIO-PBA NPs, and G-USIO NCs have hydrodynamic diameters of 122.5 ± 2.7, 150 ± 3.7, 198.3 ± 1.4, and 229.3 ± 1.4 nm, respectively (Table S2, Supporting Information). Obviously, after modification of EDA and PBA, and gossypol crosslinking, the hydrodynamic size of G-USIO NCs is quite larger than those of the NPs. Meanwhile, these NPs and NCs show a quite uniform size distribution ( Figure 2C). Additionally, the USIO NPs display a surface potential of −44.6 ± 6.2 mV owing to the critic acid surface passivation, and EDA modification affords the USIO-NH 2 NPs with an elevated surface potential of −17.4 ± 0.8 mV ascribing to the occupation of the surface carboxyl groups and increment of the surface amine groups. After PBA modification, the surface potential of USIO-PBA NPs decreases to −26.5 ± 0.7 mV attributed to the depletion of the surface amine groups. Further gossypol crosslinking via PBAE bonding makes the G-USIO NCs have an elevated surface potential (−15.8 ± 0.4 mV), likely due to the clustered structure with limited amine exposure.
UV-vis absorption spectra were collected to confirm the characteristic absorption features associated with gossypol at 266 and 380 nm, implying the successful inclusion of gossypol (Figure 2D). Fourier transform infrared (FTIR) spectra reveal that the peak at 588 cm −1 is ascribed to the Fe─O bond of USIO NPs, and the peaks at 800 and 1578 cm −1 are attributed to the benzene ring of gossypol, hinting the successful preparation of G-USIO NCs ( Figure 2E). Meanwhile, X-ray diffraction (XRD) was performed to verify the crystal structure of G-USIO NCs, which is similar to that of USIO NPs ( Figure 2F). Further, the inorganic/organic compositions of USIO NPs, USIO-NH 2 NPs, USIO-PBA NPs, and G-USIO NCs were determined by thermogravimetric analysis (TGA) ( Figure 2G). In comparison with the weight loss of USIO NPs (29.5%), the USIO-NH 2 NPs have a weight loss of 33.6% at 800°C, implying that there is around 4.1% EDA modified onto the surface of the USIO-NH 2 NPs. By comparing the weight losses of USIO-PBA NPs (35.5%) and G-USIO NCs (41.4%) at 800°C, the modification degree of PBA and the loading percentage of gossypol were calculated to be 1.9% and 5.9%, respectively.

Release Kinetics of Gossypol and Fe
The pH-and ROS-responsive release of gossypol and Fe from the G-USIO NCs were studied ( Figure S2, Supporting Information). As shown in Figure S2A, Supporting Information, under pH 7.4 + H 2 O 2 and pH 6.5 + H 2 O 2 , there are about 48.4% and 56.6% of gossypol release from the G-USIO NCs at 72 h, respectively, which are much higher than under pH 7.4 (3.3%) and pH 6.5 (4.6%) in the absence of H 2 O 2 . This should be due to the oxidative breakage of the PBAE bond by H 2 O 2 . [30] Meanwhile, the acidic condition can further promote the release of gossypol, mainly due to the dissociation of the stable PBAE bonds under an acidic pH condition. [13] Meanwhile, the amount of Fe release also increases from 1.3% at pH 7.4 + H 2 O 2 to 4.2% at pH 6.5 + H 2 O 2 , presumably due to the dissociation of Fe 3 O 4 to form Fe 2+ /Fe 3+ ( Figure S2B, Supporting Information). Further, more Fe release can be achieved at pH 6.5 + H 2 O 2 than at pH 6.5 (2.4%) without H 2 O 2 . This should be ascribed to the H 2 O 2 -induced dissociation of the G-USIO NCs containing PBAE bonds.

Detection of ROS Generation
Under an acidic environment, Fe 2+ /Fe 3+ could be released from USIO NPs to efficiently induce hydroxyl radicals (·OH) generation through Fenton/Fenton-like reactions. [31] Meanwhile, the reaction between G-USIO NCs and H 2 O 2 could also allow for efficient ·OH production based on the peroxidase-like activity of USIO NPs under acidic conditions. [32] Methylene blue (MB) degradation was utilized to verify the ·OH generation ability of the G-USIO NCs. As shown in Figure 2H,I and Figure S3, Supporting Information, the degradation of MB is most probably ascribed to the slight dissociation of USIO NPs to form Fe 2+

Cytotoxicity and Cellular Uptake Assays In Vitro
The cytotoxicity of USIO NPs, G-USIO NCs, G-USIO NCs under UTMD (G-USIO NCs + UTMD), and free gossypol were examined using 4T1 cells to evaluate the anticancer effect of the G-USIO NCs. The half maximal inhibitory concentrations (IC 50 s) of G-USIO NCs, G-USIO NCs + UTMD, and free gossypol were  In (A,B), * is for p < 0.05, ** is for p < 0.01, and *** is for p < 0.001, respectively. calculated to be 29.8, 27.2, and 35.4 μg mL −1 , respectively (Table S3, Supporting Information). Obviously, as shown in Figure 3A, USIO NPs alone exhibit little toxicity to 4T1 cells due to the limited CDT effect of USIO NPs. However, 4T1 cells treated with G-USIO NCs, G-USIO NCs + UTMD, and free gossypol display a Fe or gossypol concentration-dependent decrease of cell viability, and the G-USIO NCs + UTMD group has the most noticeable therapeutic effect (p < 0.001) due to both the combined CT (gossypol) and CDT (USIO NCs) effects, and the cavitation and sonoporation effect of UTMD ( Figure 3A and Figure S4A, Supporting Information). Certainly, compared to the G-USIO NCs, free gossypol exhibits a relatively low anticancer activity mainly due to its low bioavailability and single CT effect.
To check the cellular uptake of USIO NPs, G-USIO NCs, and G-USIO NCs + UTMD, inductively coupled plasma-optical emission spectroscopy was used to quantify the amount of Fe taken up by 4T1 cells ( Figure 3B). Clearly, cells incubated with USIO NPs, G-USIO NCs, or G-USIO NCs + UTMD all reveal a concentration-dependent Fe uptake. Notably, the G-USIO NCs display higher Fe uptake than the USIO NPs (p < 0.01) likely attributing to the larger size of NCs than that of USIO NPs, thus facilitating enhanced cellular uptake. [33] Moreover, compared to the G-USIO NCs alone, the addition of UTMD further promotes the cellular uptake of the G-USIO NCs (p < 0.01) through the cavitation and sonoporation effect produced. [28] To investigate the safety of the G-USIO NCs, we also examined the viability of normal cells (L929 cells) under the same treatments ( Figure 3C). Apparently, L929 cells treated with all materials display no significant viability changes under the studied Fe concentrations (0-40 μg mL −1 ). This should be because the G-USIO NCs have a low release of Fe and gossypol under regular physiological conditions. Additionally, L929 cells treated with free gossypol at a gossypol concentration of 35.5 μg mL −1 still have a viability up to 84.8% ( Figure S4B, Supporting Information), implying less cytotoxicity of free gossypol to normal cells.

ROS Generation In Vitro
The ROS generation ability was tested in vitro after 4T1 cells were treated with USIO NPs, free gossypol, G-USIO NCs, or G-USIO NCs + UTMD, respectively. Under the same Fe concentration  fluorescence intensity (p < 0.05), likely ascribed to the generation of ⋅OH through Fenton/Fenton-like reactions in cancer cells having superabundant H 2 O 2 . The G-USIO NCs group exhibits much stronger green fluorescence intensity than the USIO NPs group (p < 0.01), likely owing to the better cellular uptake of G-USIO NCs than USIO NPs and the existence of gossypol, facilitating enhanced ROS production. Furthermore, under the assistance of UTMD, the G-USIO NCs show the best ROS generation performance among all groups, mainly attributed to the further enhanced cellular uptake triggered by the cavitation and sonoporation effect.

GSH Depletion Capacity In Vitro
Next, the cellular GSH depletion ability of USIO NPs, free gossypol, G-USIO NCs, or G-USIO NCs + UTMD was tested (Figure 4A). Different from the PBS control, all other groups show apparently decreased GSH levels, likely due to the ROS generation ability of USIO NPs and gossypol. Furthermore, as opposed to the PBS control, the USIO NPs can deplete GSH (6%) through the release of Fe 3+ /Fe 2+ from USIO NPs to mediate Fenton/Fentonlike reaction for ROS generation. Obviously, the G-USIO NCs deplete more GSH than the USIO NPs and free gossypol (p < 0.001) due to the enhanced GSH consumption by the combination of them together with a larger size-facilitated enhanced cellular uptake of the NCs. Meanwhile, compared to the G-USIO NCs group (30.7%), the GSH depletion level of the NCs under UTMD increases to 41.8% for the G-USIO NCs + UTMD group, which should be due to the further enhanced cell uptake induced by UTMD for improved consumption of GSH. Overall, the G-USIO NCs + UTMD treatment can best modulate TME for ROS generation and GSH depletion in comparison with other materials.

Intracellular LPO Accumulation and Cell Apoptosis
Encouraged by the ROS generation and GSH depletion capacity of the G-USIO NCs, we next explored the LPO accumulation . In (C), * is for p < 0.05, ** is for p < 0.01, and *** is for p < 0.001, respectively (n = 3). and apoptosis of 4T1 cells treated with the USIO NPs, free gossypol, G-USIO NCs, or G-USIO NCs + UTMD. Compared to the PBS control, other groups all show green fluorescence increase and red fluorescence decrease in cells, validating the intracellular LPO accumulation after treatments of cells with USIO NPs, gossypol, or G-USIO NCs ( Figure 4B). In comparison with the USIO NPs and gossypol alone, cells treated with the G-USIO NCs display much higher green fluorescence intensity and much lower red fluorescence intensity. Similarly, the UTMD assistance renders the G-USIO NC-treated cells with higher LPO accumulation than the NC-treated cells without UTMD due to the UTMDpromoted cavitation and sonoporation effect.
The cell apoptosis effect was then examined after different treatments ( Figure 4C). The 4T1 cells treated with USIO NPs display a total apoptotic and necrotic percentage of 44.6%, which is much higher than those treated with PBS (4.3%), hitting the potential CDT performance of the USIO NPs. Cells treated with free gossypol display a total apoptotic and necrotic rate of 64.6%, proving the gossypol-induced cell apoptosis, in consistence with that reported in the literature. [34] The groups of G-USIO NCs and G-USIO NCs + UTMD display a sum of the apoptotic and necrotic rate of 4T1 cells at 68.1% and 73.3%, respectively, which are higher that the USIO NPs and gossypol groups, due to the combined effect of USIO NPs and gossypol and the further ultra-sound enhancing effect. Furthermore, the apoptosis-related proteins (p53, PTEN, Bax, and Bcl-2) in 4T1 cells after various treatments were analyzed by Western blotting to illustrate the therapeutic mechanism of G-USIO NCs. As shown in Figure 4D, tumor suppressor proteins of p53/PTEN and proapoptotic protein of Bax are significantly up-regulated, and the antiapoptotic protein of Bcl-2 is down-regulated, suggesting the activation of the apoptosis pathway after the NC treatment. Obviously, the G-USIO NCs + UTMD group displays the highest Bax/Bcl-2 ratio (p < 0.001) than the other groups ( Figure S6, Supporting Information) due to the UTMD-enhanced combination CT/CDT effect.

T 1 -Weighted MR Imaging
The T 1 -weighted MR imaging performance of the G-USIO NCs was next explored at pH 6.5 with and without H 2 O 2 . In the absence of H 2 O 2 , the MR signal intensity of USIO NPs and G-USIO NCs increases with the Fe concentration ( Figure 5A and Figure S7, Supporting Information), and the r 1 relaxivity of G-USIO NCs (0.7 mm −1 s −1 ) is more or less similar to that of USIO NPs (0.6 mm −1 s −1 ), likely due to the integrated structure of the G-USIO NCs with no significant impact on water proton relaxation. In contrast, in the presence of H 2 O 2 , G-USIO NCs exhibit a higher r 1 value of 1.1 mm −1 s −1 than those in the absence of H 2 O 2 , possibly due to the disassembly of G-USIO NCs to create single particles with different aggregation states than the pristine USIO NPs.
Subsequently, the G-USIO NCs were used for T 1 MR imaging of a xenografted 4T1 tumor model under the assistance of UTMD. The pristine USIO NPs and G-USIO NCs without UTMD were also tested for comparison ( Figure 5B). The tumor T 1 MR signal intensities at 15 and 30 min post-injection of G-USIO NCs are higher than those of USIO NPs at the corresponding time points, possibly due to the larger size-induced amplified passive EPR effect of the NCs and also the dissociation of the G-USIO NCs at the acidic TME containing H 2 O 2 to release single USIO NPs. Moreover, the G-USIO NCs + UTMD group shows the highest T 1 MR signal intensity due to the UTMD-rendered sonoporation effect, in consistence with the literature. [28] With the time extension after injection (at 30 min), the dissociated G-USIO NCs and USIO NPs are gradually metabolized, resulting in attenuated tumor T 1 MR signal intensity. To quantitatively verify the T 1 MR imaging capability of the G-USIO NCs, the timedependent tumor MR signal-to-noise ratio (SNR) was quantified in different treatment groups ( Figure 5C) and the results are in line with the qualitative imaging results. Collectively, under UTMD assistance, the G-USIO NCs enabled highly efficient tumor T 1 MR imaging in vivo.

Combination Tumor Therapy and Biosafety Evaluation
Inspired by the G-USIO NC-mediated combination anticancer performance under UTMD in vitro, we further investigated the therapeutic efficiency of G-USIO NCs + UTMD in vivo through the combination of CT/CDT using a xenografted 4T1 tumor model ( Figure 6A). As can be seen in Figure 6B,D, the tumors in the groups of PBS and USIO NPs grow rapidly with time post-injection. In contrast, the treatment of G-USIO NCs leads to significant tumor inhibition efficacy due to the combination of CT/CDT. Certainly, the most significant antitumor effect is observed in the G-USIO NCs + UTMD group due to the UTMDenhanced tumor uptake of G-USIO NCs. In addition, the body weights of tumor-bearing mice display no appreciable changes in different groups during the treatment period, verifying the good biosafety of all materials/treatments ( Figure 6C).
Subsequently, we assessed the combinational CT/CDT effect of G-USIO NCs + UTMD via the hematoxylin & eosin (H&E) and TdT-mediated dUTP Nick-End Labeling(TUNEL) staining of www.advancedsciencenews.com www.advsensorres.com tumor tissues ( Figure 6E). Except for the PBS group, all other groups show the necrotic and apoptotic cell populations in the tumor slices, and the G-USIO NCs + UTMD group exhibits the most significant necrotic and apoptotic effects on tumors due to the best efficacy of the UTMD-enhanced combination therapy. In addition, lung metastasis can be observed in the groups of PBS, USIO NPs, and gossypol from the H&E staining and photograph of the lung in tumor-bearing mice ( Figure 6F and Figure S8, Supporting Information). In contrast to the groups of PBS, USIO NPs, and gossypol, the number of lung metastasis nodules in the G-USIO NCs + UTMD group is significantly lower (p < 0.05). Conclusively, the UTMD-assisted treatment of G-USIO NCs leads to the most effective combination therapy of both primary tumor and tumor metastasis.
Major organs (heart, liver, spleen, and kidney) of tumorbearing mice were H&E stained to check the biosafety of the materials. As can be seen in Figure S9, Supporting Information, no significant pathological damage and toxic side effects in all organs can be seen for all groups. Moreover, the blood routines and serum biochemistry assay were carried out at 12 days postinjection of all materials. All measured parameters are inside the normal range in all groups, implying the good biosafety of USIO NPs and G-USIO NCs with or without UTMD ( Figure S10, Supporting Information). In order to further comprehensively verify the biosafety of G-USIO NCs, the Fe contents in vital organs and tumors of mice were measured at different time points postinjection of G-USIO NCs with or without UTMD. As shown in Figure S11, Supporting Information, a remarkable amount of Fe was taken up by the spleen and kidney at 1 h post-injection, and the highest level of Fe in the lung was found at 45 min postinjection in both groups. Compared to the G-USIO NCs, the higher level of Fe in the tumor was observed at different time points post-injection in the G-USIO NCs + UTMD group than in the G-USIO NCs group due to the UTMD-enhanced tumor uptake of the G-USIO NCs. With time extension, the uptake of Fe in all organs and tumors gradually decreases. These results suggest that the G-USIO NCs can be accumulated at the tumor site via UTMD-enhanced passive EPR effect, and be gradually metabolized and cleared out of the body, validating the good biosafety of the developed G-USIO NCs.

Conclusions
In summary, we developed gossypol-loaded USIO NCs through gossypol-mediated crosslinking of PBA-modified USIO NPs for TME-responsive T 1 -weighted MR imaging-guided combinational CT/CDT of tumors. We show that USIO NPs can be surface modified with PBA and crosslinked by gossypol-containing catechol to form dual ROS-and pH-responsive G-USIO NCs. Gossypol acts as both a chemotherapeutic drug and a crosslinker. Under acidic TME containing H 2 O 2 , the G-USIO NCs can release both gossypol and Fe, lead to intracellular ROS production and GSH depletion, and exert combinational tumor CT/CDT due to the responsive breaking of the PBAE bond. The tumor T 1 MR imaging and the combinational CT/CDT effects can be further improved through the UTMD-mediated cavitation and sonoporation effect. The developed G-USIO NCs may be regarded as an intelligent theranostic nanosystem for T 1 MR imaging-guided combination therapy of tumors. Moreover, the design of NCs using phenolic drugs as crosslinkers may be extended to prepare other nanomedicines through the boronic ester bonding for different biomedical applications.

Experimental Section
All animal experiments were carried out after approval by the Ethical Committee on Experimental Animal Care and Use of Donghua University and also according to the policy of the National Ministry of Health (China). For statistical analysis, all experimental data are displayed as the mean ± SD (n ≥ 3). One-way analysis of variance statistical method was used to analyze the experimental results through IBM SPSS Statistic 25 software (IBM, Armonk, NY). In all cases, a p-value of 0.05 was chosen as the significance level, and the data were marked with (*) for p < 0.05, (**) for p < 0.01, and (***) for p < 0.001, respectively. Full experimental details are described in the Supporting Information.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.