Multifunctional Magnetic Nanoplatform Eliminates Cancer Stem Cells via Inhibiting the Secretion of Extracellular Heat Shock Protein 90

Yajing Liu, Xiaomin Suo, Haotong Peng, Weixiao Yan, Hongjuan Li, Xinjian Yang, Zhenhua Li, Jinchao Zhang,* and Dandan Liu*

Abstract

Cancer stem cells (CSCs) are responsible for malignant tumor initiation, recurrences, and metastasis. Therefore, targeting CSCs is a promising strategy for the development of cancer therapies. A big challenge for CSCbased cancer therapy is the overexpression of therapeutic stress protein, heat shock protein 90 (Hsp90), which protects CSCs from further therapeuticinduced damage, leading to the failure of treatment. Thus, efficient strategies to target CSCs are urgently needed for cancer therapy. To this end, a multifunctional nanoparticle (MNP) for CSC-based combined thermotherapy and chemotherapy is reported. This strategy dramatically suppresses tumor growth in breast CSC xenograft-bearing mice. Furthermore, a new mechanism is present that the MNP exerts its striking effects on CSCs by inhibiting the secretion of extracellular Hsp90 (eHsp90), resulting in the interruption of several key signaling pathways. These findings open new perspectives on the use of an MNP for effective CSC-based cancer treatment by inhibiting the function of eHsp90. their DNA damage repair ability, which enable CSCs to proliferate the population of tumor cells along with chemo- or radiotherapy.[3,4] Since conventional chemotherapy is insufficient to eradicate CSCs, efficient strategies to target CSCs are needed urgently for cancer treatment. Nanomedicine has demon strated great potential for developing safe and targeted strategies against solid tumors.[5,6] Nanoparticles (NPs) are vital tools in the field of biology and nanomedicine, and provide novel ideas for life medical science application, including drug delivery in cancer treatment and gene therapy.[7] Cancer nanotechnology, as an integrated platform, provides great opportunities for improving drug efficacy and pharmacokinetics to reduce side effects. Various smart multi-

1. Introduction

Cancer stem cells (CSCs) have been evident to have selfrenewal, cancer initiating, and multi-potent differentiation, and metastases properties.[1,2] Thus, developing the CSC-targeted therapies is a major clinical challenge. The present existential threat of CSCs is their infinite proliferation and drug resistance ability. CSCs are highly resistant to chemo- and radiotherapy due to their dormancy, overexpression of ABC-transporter, and functional NPs, such as polymeric NPs, mesoporous silica NPs, and micelles, hold bright prospects for the delivery of anticancer drugs.[8–12] The combinational strategy that integrates two or more therapeutic agents to form a multifunctional NP system for targeting, drug delivery and cancer therapy exhibits an enormous potential to revolutionize the way of the future cancer treatment.[13–15] However, the currently used combinational strategies, such as different drugs synergy, combined chemotherapy and photothermal therapy, are far from a perfect regimen for cancer patients.[16–19] The combination of different anticancer drugs may cause the multidrug resistance, resulting a counterproductive therapeutic efficiency.[20,21] In the case of nanoparticlebased combined chemotherapy and photothermal therapy, near infrared reflection (NIR) irradiation can result in a highly localized temperature (over 43 °C, which can cause damage to surrounding normal tissues) increase at the tumor site. However, inappropriate combined chemo- and thermo-therapy may markedly induce the high expression of heat shock proteins (HSPs), which can protect cancer cells, making them more resistant to therapies. The unkilled cancer cells may acquire the stemness, which may have profound implications for tumor progression and clinical outcome.[22] Considering this, precisely controlled thermal and combinatorial therapies are needed. The magnetic nanoparticle-based localized thermotherapy can easily achieved the temperatures above 43 °C and be well-controlled at tumor site by applying an alternative magnetic field (AMF).[23] Such a strategy can achieve synergistic effects (thermal enhance sensitivity of chemotherapy) and further benefit cancer therapy without noticeable side effect on surrounding normal tissues. However, a big challenge for thermotherapy is the overexperssion of heat shock protein 90 (Hsp90), which is a highly conserved molecular chaperone and frequently upregulated in tumors, and involved in maintaining the stability and function of variety of client proteins protecting them from environment (or therapeutic) stress-induced damage.[24,25] Therefore, inhibiting the expression of Hsp90 is the key for improving the efficiency of combined thermotherapy and chemotherapy.
To this end, we designed a simple, but smart multifunctional nanoparticle (MNP) for CSC-based cancer therapies through combined thermotherapy and chemotherapy. We developed a silica-based MNP (CD44-HSPI/Fe3O4@SiNP) that encapsulated magnetic cores (Fe3O4NPs) and a chemotherapeutic agent (Hsp90 inhibitor (HSPI), 17-DMAG (phase II/III)) and coated with a specific fluorescent dye conjugated-antibody (CD44) against surface markers of breast CSCs (BCSCs). The MNP has several functions, including: 1) antibody against CD44 (anti-CD44) was conjugated to NPs to achieve BCSC targeting; 2) controlling the combination of thermotherapy and chemotherapy by applying an AMF; 3) locating in the endosomes and releasing the 17-DMAG to inhibit the expression of Hsp90; 4) the fluorescent dye (such as Cy7, PE, and FITC) can be used as imaging agent for in vitro or in vivo monitoring the internalization and localization of the MNPs. A hypothesis was proposed that the MNPs could inhibit the expression of Hsp90, and then eliminate the population of CSCs and improve cancer therapies. We used BCSCs and BCSC-xenograft nude mice as an in vitro and an in vivo model, respectively, to test our hypothesis by applying the MNPs under an AMF. A significant inhibition of BCSC survival and tumor xenograft growth was achieved after administration of the MNPs and AMF treatment without overt side effects during the process of treatment. More specifically, the method present in this work was extended to target cervical CSCs and liver CSCs for combined thermotherapy and chemotherapy, resulting significant inhibition of ovarian CSCs and liver CSCs survival rate.
Furthermore, the molecular mechanisms of combined thermotherapy and chemotherapy eradiated CSCs and the effects localized thermotherapy may have on tissues are largely unknown and worth investigating. In this work, we present a new mechanism that MNPs exerts its striking effects on CSCs by inhibiting the secretion of extracellular Hsp90 (eHsp90), resulting the interruption of several key signal transduction pathways. We believe that the optimal antitumor efficacy and high biocompatibility of MNPs hold great potential for further development in CSC-targeted cancer treatments.

2. Results and Discussion

2.1. Designing MNPs for BCSCs Targeted Therapy

In this work, we synthesized and characterized biocompatible multifunctional silica-based NPs encapsulating magnetic cores (Fe3O4 NPs) and chemotherapeutic agents (including heatshock protein inhibitors) and coated with a specific fluorescent dye conjugated-antibody (CD44) against surface markers of BCSCs for CSC-targeted and combined thermotherapy and chemotherapy under an AMF. It has been reported that the receptor CD44 was strongly expressed by the CSCs,[26,27] therefore, we selected anti-CD44 as BCSCs targeting agent, which facilities the endocytosis of NPs. The design idea of nanosystem and schematic were present in Figure 1A. For the application of the nanosystem, we focused on the treatment of breast cancer. The CSCs were sorted from MCF-7 based on the established model.[28] We hypothesized that efficiently eliminating BCSCs through a targeted therapy may be the key to successful treatments for breast cancer. Furthermore, the method present in this work was extended to target cervical CSCs (derived from Hela cells) and liver CSCs (derived from HepG2 cells) for combined thermotherapy and chemotherapy, resulting significant inhibition of ovarian CSCs and liver CSCs survival rate.

2.2. Characterization of MNPs

As shown in Figure 1B,C, the scanning electron microscopy (SEM) and the transmission electron microscopy (TEM) images demonstrated that the Fe3O4@SiNPs and CD44 conjugatedFe3O4@SiNPs (CD44-Fe3O4@SiNPs) were mainly 50 nm and was narrowly distributed, and the silica thickness was finely controlled from 15 to 20 nm, and the diameter of the Fe3O4 NP core (dark color) was ≈25 nm (Figure 1B,C). The Zeta potential results (Figure S4C, Supporting Information) showed that the surface charges of the Fe3O4@SiNPs and CD44-Fe3O4@SiNPs were −28.4 and −27.5 mV, respectively. After AMF treatment, the whole structure of CD44-Fe3O4@SiNPs collapsed, as well as the size changes, indicating the property of heat-triggered drug release (Figure S4D, Supporting Information). As seen in Figure 1D, the fluorescent signal of the Cy7-CD44-conjugated NPs was located at a maximum emission wavelength of 790 nm, which identical to a solution of free Cy7-CD44 antibody (Figure S4A Spectrofluorometer, Supporting Information), indicating the successful conjugation of Cy7-CD44 antibody on the surface of HSPI/Fe3O4@SiNPs. To further demonstrate the successful conjugation of CD44 on the HSPI/ Fe3O4@SiNPs, X-ray photoelectron spectroscopy (XPS) of CD44-HSPI/Fe3O4@SiNPs and Fe3O4@SiNPs were analyzed. As shown in Figure S16 (Supporting Information), the full XPS scan of Fe3O4@SiNPs clearly indicated the Fe, O, and Si atoms, and CD44-HSPI/Fe3O4@SiNPs indicated the Fe, O, Si, and N atoms. Anti-CD44 is a protein, which is the source of N atom, therefore, the anti-CD44 was successfully conjugated on the surface of HSPI/Fe3O4@SiNPs. The high-resolution Fe 2p spectrum revealed two distinct peaks of Fe 2p3/2 and Fe 2p1/2 at 710.8 and 724.4 eV, respectively, the appearance of Fe3+cations (711 eV) and Fe2+cations (709 eV) indicate the exist of the Fe3O4 phase.

2.3. Magnetic Hyperthermia Study and Controlled Drug Release

The magnetization saturation (Ms) for the Fe3O4 NPs and Fe3O4@SiNPs was evaluated by testing the hysteresis curves through a vibrating sample magnetometer (VSM). The curve passed through the origin, indicating of Fe3O4 NPs and CD44-Fe3O4@SiNPs were 73.08 and that both the Fe3O4 NPs and Fe3O4@SiNPs were super- 8.14 emu g−1, respectively. The Fe3O4@SiNPs have a weaker paramagnetic. As shown in Figure 1E, the Ms values magnetization than the naked Fe3O4 NPs under the applied magnetic field because the strength of magnetization is related to the amount of magnetic material in the sample (Figure S5, Supporting Information).
The heating rate of MNPs under an AMF is closely based on the high Ms value. The comparative temperature rise of the MNP suspensions versus exposure time is shown in Figure 1F. The highest temperature achieved by the Fe3O4@SiNPs and CD44-HSPI/Fe3O4@SiNPs suspension was 53.7 and 53.3 °C, respectively, which is much higher than that observed for the SiNP suspension in PBS solution (27.6 °C). Thus, CD44-HSPI/Fe3O4@SiNPs even dispersed in a neutral medium have attractive potential for magnetic hyperthermia, as well as other biomedical applications such as heat-triggered drug delivery systems.
The critical features of a drug delivery system are controlled and sustained drug release. As shown in Figure 1G, the accumulative release profile of HSPI (Figure S6, Supporting Information) from the Fe3O4@SiNPs (1 mg mL−1) was obtained at different time points under different pH conditions. An in vitro release study showed that the Fe3O4@SiNPs exhibited a sustained release rate of HSPI was up to 81% and 75% after 100 h under an AMF with a PH of 5.4 (the pH value of endosome was about 5.0–6.0) and pH of 7.4 (normal PH value in biosytem), respectively. However, drug release rates of only 25.3% and 25% were observed for up to 100 h without the AMF trigger with pH values of 5.4 and 7.4, which allows for controlled release in the animal body by applying AMF.

2.4. In Vitro Cellular Uptake and In Vivo Tumor-Targeted Accumulation

First, we established a BCSC model from breast cancer cell line (MCF-7, which represents hormone receptor-positive breast cancer) by serum-starved sparse cell culture. The data were shown in supplementary information (Figures S1–S3, Supporting Information). As shown in Figure 2A, the uptake of CD44-HSPI/Fe3O4@SiNPs by BCSCs was higher than that of HSPI/Fe3O4@SiNPs after 1 h incubation, indicating high uptake efficiency with the targeting agent. Besides, FITCtransferrin (TfR, a marker for endosome staining) was used to colocalize with MNPs, indicating that endosome internalization of CD44-Fe3O4@SiNPs. Moreover, the uptake mechanism of CD44-HSPI/Fe3O4@SiNPs in BCSCs showed that anti-CD44-conjugated HSPI/Fe3O4@SiNPs entered the cells more quickly than free HSPI/Fe3O4@SiNPs, which might be due to the receptor-mediated endocytosis pathway.[22]
In vivo tumor-targeting efficacy and whole body distribution of CD44-HSPI/Fe3O4@SiNPs was shown in Figure 2B. A fluorescent signal in the CD44-HSPI/Fe3O4@SiNPtreated mice was clearly present at the tumor site as time elapsed. Specifically, the fluorescence signals of MNPs were mainly accumulated in tumor site after 4 h post-tail vein injection, whereas little fluorescence was observed in the liver. However, no detectable signal was recorded from the HSPI/Fe3O4@SiNPs in the tumor. This finding was confirmed by the ex vivo imaging that no obvious fluorescence signals in the other organs, but a low amount was observed in the liver (the principal organ for metabolizing).
To further clarify the in vivo imaging results, Fe average content in organ tissues of the control group and treated groups were shown in Figure S7 (Supporting Information). As can be seen, much lower Fe concentrations, but always well above baseline levels, were observed in liver tissue (dependent on its detoxification function). In contrast, the heart, spleen, lung, and kidneys were hardly increased in comparison with those found in the control group.

2.5. Efficacy of Combined Thermotherapy and Chemotherapy on BCSCs

After confirming the noncytotoxicity of the nanocarrier (SiNPs) (Figure S8, Supporting Information), the efficiency of the combinatorial thermotherapeutic and chemotherapeutic (HSPI as anticancer drug) CD44-HSPI/Fe3O4@SiNPs was tested by incubating BCSCs with MNPs under an AMF for 15 min. Notably, compared with the control (medium only), there was a significant decrease in the survival rate of BCSCs in the presence of CD44-HSPI/Fe3O4@SiNPs for 1 (Figure 3A), 24, and 48 h (Figure S15, Supporting Information). Moreover, the cell survival rate decreased to 68% in the presence of Fe3O4@SiNPs upon application of an AMF, indicating the important role of HSPI during the therapy. Although the separated thermotherapy (treated with Fe3O4@SiNPs) and chemotherapy (treated with HSPI@SiNPs) reduced the survival rate of CSCs, there were still around 40% cells survive compared to the survival rate of 2.68% after 48 h CD44-HSPI/Fe3O4@SiNPs treatment. These results indicated the significant therapeutic efficiency of combined thermotherapy and chemotherapy. The temperature of BCSCs treated with CD44-HSPI/Fe3O4@SiNPs reached to 52 °C (Figure 3B,C), which is sufficient to kill the BCSCs combined with the inhibition of Hsp90. Furthermore, IC50 value was also calculated to evaluate the therapeutic effects of a pharmaceutical formulation. As shown in Table S3 (Supporting Information), the IC50 values of Fe3O4@SiNPs, HSPI@SiNPs, and HSPI/Fe3O4@SiNPs was significant higher than CD44-HSPI/Fe3O4@SiNPs after 24 h and 48 h incubation under AMF. These results suggest that the therapeutic effect of CD44-HSPI/Fe3O4@SiNPs was better than other groups due to the specific CSCs targeting combined thermotherapy and chemotherapy.
To figure out the mechanism of cell death caused by MNPmediated thermotherapy and chemotherapy, the apoptosis and necrosis of cells were evaluated through the double staining of annexin-v-FITC (a marker of apoptosis) and PI (a marker of cell death). Consistent with the above findings, the percentage of BCSCs that were positive for CD44-HSPI/Fe3O4@SiNPs reached to 59.98% under AMF treatment (Figure 3D). However, apoptotic cells (annexin-v-positive, PI-negative) were not observed after AMF treatment, indicating that necrosis was the primary form of cell death observed in the BCSCs.

2.6. High Biocompatibility of MNPs In Vivo

The biocompatibility of CD44-HSPI/Fe3O4@SiNPs was evaluated through analyzing the hemolysis, and the CD44-HSPI/Fe3O4@SiNPs showed a dramatic tumor growth inhibition, with almost no apparent growth when compared with control group. However, the group treated with freeHSPI and HSPI-unloaded CD44-Fe3O4@SiNPs did not significantly affect tumor growth. The body weight of each group increased proportionately during the observation period (Figure S10A, Supporting Information). The ex vivo tumor volume (Figure S10B, Supporting Information) confirmed the in vivo results that CD44-HSPI/Fe3O4@SiNP-mediated combined thermotherapy and chemotherapy efficiently inhibit the tumor growth.
Furthermore, the ex vivo tumor histological analysis was carried out to further evaluated the efficiency of combined therapy of MNPs. In comparison to control group (tumor tissue was well maintained with cancer nests), the significant apoptosis and necrosis occurred in the MNP-treated tumor region were observed. The apoptotic and necrotic cells appeared round with dark eosinophilic cytoplasm and a dense purple nucleus (Figure 4D). In addition, the fluorescent immunohistochemical staining showed that the tumors treated with CD44HSPI/Fe3O4@SiNPs depleted BCSCs, as shown by a dramatic decrease in the expression of CD44 and CD133 relative to that in untreated tumors (Figure 4E). For the group that received CD44-Fe3O4@SiNPs, there were still residual BCSCs in tumor tissue, with expression of CD44 and CD133.
To demonstrate the universal application of the combinatorial thermotherapy and chemotheraoy, the nanosystem present in this work was applied in treatment of cervical CSCs and liver CSCs. The results were shown in Figure S11–S13 (Supporting Information). With the great versatility and flexibility of MNPs, as well as their proven safety and CSC-targeting advantage, this nanodelivery system has the potential for clinical translation and to become a universal platform for CSC-targeted simultaneous thermotherapy and chemotherapy in cancer treatment.

2.8. CD44-HSPI/Fe3O4@SiNPs Inhibit the Secretion of Extracellular Hsp90

Hsp90 is a heat stress protein that has been an attractive molecular target for thermal-based cancer therapy.[29] Despite the clinical attraction, a potential role for Hsp90 as a regulator of CSCs has not been well researched. In this study, we showed that CSCs overexpressed the HSP90 and heat shock factors (HSFs), indicating a key factor to regulate the survival of CSCs. Therefore, specific inhibiting of Hsp90 may elicit the efficiency therapeutics for CSCs. In this work, we proposed that MNPs significantly inhibit the expression of Hsp90 by delivering the HSPI into the BCSCs to improve the efficacy of the combined thermotherapy and chemotherapy. However, we have observed some interesting phenomena, but did not get the expected experimental result. As shown in Figure 5A,C, both of MCF-7 and BCSCs expressed the intracellular Hsp90 after CD44Fe3O4@SiNP-induced hyperthermia. Even the cells treated with CD44-HSPI/Fe3O4@SiNPs, high expression of Hsp90 was also observed in MCF-7 and BCSCs. This finding was contrary to expectation that the MNPs cannot inhibit the expression of Hsp90 in the cells.
Interestingly, the expression of extracellular Hsp90 (eHsp90) was significant different between MCF-7 and BCSCs. Except its intracellular form, Hsp90 has also been found in the extracellular space, a tendency frequently observed in tumor tissues. The eHsp90 may exist either in a secreted form, released form or cell membrane surfacebound form, and plays a vital role in tumor progression.[30] During the treatment process (especially, the hyperthermiaassistant anticancer drug sensitivity methods), a pathophysiologic secreted or leaked eHsp90 from cancer cells could make a small population of cancer cells survive and spread to distant lesions.[30,31] Over the past decades, a growing body of evidence proves the key role of intracellular Hsp90 in the malignant tumor progression.[32,33] However, whether the secreted or leaked eHsp90 could facilitate the cancer cell to survive, especially in a cell-autonomous manner, is not known. In addition, since the Hsp90 inhibitors cannot distinguish the intracellular Hsp90 and eHsp90, the potential function of eHsp90 as a regulator of CSCs has not been well investigated.[30] Therefore, we further studied whether the eHsp90 is a key factor to maintain the survival of BCSCs during the treatment process, and how the MNPs inhibit the expression of eHsp90. It has been reported that the hyperthermia induced Hsp90 (will be secreted outside) to go into the nanovesicles called exosomes and then secreted outside the cells by activating the stemness-related signaling pathways.[30] All the proteins that have been identified in exsosomes are located in the endosomal compartments, but never in the mitochondria, ER, Golgi apparatus or nucleus.[34] Interestingly, the MNPs can enter cells by clathrin-modulated endocytosis, and intracellular located in the endosomal compartments (Figure 2 and Figure S14 enlargement images, Supporting Information). Since the escape of NPs from endosomes is extremely difficult, thus take advantage of the same location of MNPs and hyperthermia induced Hsp90 (both in endosomes), the designed MNPs could release the HSPI to inhibit the secretion of Hsp90 outside the cells. As indicated in Figure 5A,B, the level of Hsp90 in the conditional medium and cell membranes was significantly downregulated after CD44-HSPI/Fe3O4@SiNPs treatment under AMF. However, no significant difference was observed for the level of intracellular Hsp90 (from listed cells) after AMF treatment. This result confirmed our hypothesis that endosome-located NPs could inhibit the secretion of endosomelocated Hsp90 outside the cells, thus, the eHsp90 could be downregulated.

2.9. Inhibition of eHsp90 Attenuates the Downstream Signaling Pathways

We identified eHsp90 is a key factor to regulated the survival of CSCs. CSCs are considered to be a cause of tumor initiation, relapse, chemo and radioresistance. Hence, complete understanding the complex mechanisms of eHsp90 on maintaining the survival and stemness in solid tumors could possibly be an encouraging direction for future cancer therapy. To receive mechanistic insights into how eHsp90-inhibited CD44-HSPI/ Fe3O4@SiNPs suppressed the growth of BCSCs, the expression of 84 genes were analyzed, which involved in several signaling pathways. As shown in Figure 6A, the heat maps indicated the mRNA expression profiles of BCSCs after MNPs treatment for 24 h. In comparison with CD44-Fe3O4@SiNPs treatment, the CD44-HSPI/Fe3O4@SiNPs treatment remarkably reduced the level of 51 genes related to PI3K pathway, G protein signaling, and RAS signaling pathways. A dramatic downregulation of HSP90 was obtained after the CD44-HSPI/Fe3O4@SiNPs treatment (Figure 6B). The detailed information of 84 genes was displayed in supplementary Table S2 (Supporting Information).
Cell proliferation maintains the tissue formation both in normal status and cancer. Unlike the normal cells, in cancer, cell cycle is out of control, resulting the infinite cell dividing and proliferation. In normal cells, when the growth is needed, the secreted growth factors sent the proliferation instruction through specifically binding to receptor proteins with extracellular binding surfaces and intracellular signaling domains.[35] Therefore, the transmembrane receptors received the growth signal and pass it inside of cells through a network of conductive proteins, resulting ultimately cell proliferation.[36] Deregulation of the signal transduction at each step can lead to the occurrence of tumor development due to uncontrolled proliferation.
Most of the receptors, growth factors, hormones, enzymes, and proteins that involved in the pathways are tumorigenic when activated or overexpressed through mutation. Interestingly, most of these molecular proteins (such as MMP2, ErbB2, CD91, IGF, and EGFR) are clients of eHsp90, thus magnification of eHsp90 indulges the unrestrained proliferation of cells.[37] These signaling pathways can be interdicted by the Hsp90 inhibitors at each individual Hsp90 chaperoned step, leading to multitargeted cancer therapy. The function of eHsp90 is distinct from but perhaps overlapping with its intracellular function. As shown in Figure 6C, the expression of PIK3C3, PIK3CA, PIK3C2A, AKT1, and AKT2 was significantly downregulated by CD44-HSPI/Fe3O4@SiNPs treatment, inactivating the transcription and translation of downstream genes, including mTOR, NFkB, BCL2, and BIRC5 genes, by which the PI3K/Akt signaling pathway was ultimately blocked. Mammalian target of rapamycin (mTOR) is a serine/threonine kinase that plays a critical role in mediating cells cycle progression.[38] The inhibition of mTOR arrested the cell cycle of BCSCs. It is reported that Akt activates the NFκB to maintained the stemness and promote tumor growth and survival in CSCs.[39] The suppression of eHsp90 also caused a reduction in stemness and tumor growth in BCSCs (Figure 6E). BCL2 (B-cell lymphoma 2) and BIRC5 (also called survivin) are specifically considered an important antiapoptotic proteins that is highly upregulated in many cancers as compared to normal cells.[40] The inhibition of eHsp90 suppressed the expression of BCL2 and BIRC5, leading to the apoptosis of BCSCs. Protein 53 (p53) is a factor closely related to the cancer development that arrests cell proliferation and promotes apoptosis in response to inhibition of eHsp90.[41] In addition, Rho is demonstrated to be a potential client protein of the Hsp90 that has the ability to bind GTP. The capability of Rho binds to GTP was regulated by the eHsp90 and HSP90 ATPase cycle.[42] By using the Hsp90 inhibitor, the binding of Rho to GTP was suppressed, which led to the attenuation of the G protein signaling pathway, resulting the inhibition of DNA damage repair (Figure 6D). There is crosstalk between the different signaling pathways in cancer progression; therefore, inhibition of one could lead to the attenuation of the others. Ultimately, the signaling pathways transcriptionally control the cancer cell activity by suppressed the expression of related-genes, including drug metabolism, cell cycle, apoptosis, ability of histone deacetylases, and expression of protein kinases (Figure 6F,G). The schematic diagram of key pathways by which CD44-HSPI/Fe3O4@ SiNPs inhibits tumor growth was demonstrated in Figure 6H. Through the above analysis, eHsp90 may serve as a broad and effective CSCs inhibitor for clinic cancer therapy.

3. Conclusion

In this study, we have developed an MNP (CD44-HSPI/ Fe3O4@SiNP), composed of Fe3O4 NPs and HSPI, simultaneously delivering both thermotherapeutic and chemotherapeutics agent to tumor region. This work first evaluated the CSCs targeting and therapeutic efficacy of CD44-HSPI/Fe3O4@ SiNPs systematically both in vitro and in vivo. BCSC-targeting ability of the MNPs was dramatically enhanced by CD44 mediated active targeting. The internalization of CD44-HSPI/ Fe3O4@SiNPs induced the necrosis of BCSCs due to a synergistic effect of thermotherapy and chemotherapy in AMF. Mechanistically, we present here a unique link between the eHsp90 and the survival of CSCs. Due to the endocytosis, the MNPs were located in the endosome, where also is the transfer location of eHsp90, thus blocking the Hsp90 secretion in the endosome site inhibit the expression of eHsp90 instead of intracellular Hsp90. Therefore, the CD44-HSPI/ Fe3O4@SiNPs can both inhibit the secretion of Hsp90, leading the necrosis of BCSCs due to a synergistic effect of thermotherapy and chemotherapy in AMF. We confirmed that CD44HSPI/Fe3O4@SiNPs eliminated the BCSCs by disturbing the function of eHsp90 (distinct from but perhaps overlapping with its intracellular function) and thereby attenuating several key signaling pathways (PI3K/Akt, G protein, and Ras signaling). Furthermore, the combined thermotherapy and chemotherapy presented in this work was extended to cervical CSCs and liver CSCs for combined thermotherapy and chemotherapy. Taken together, we developed a great versatility and flexibility of MNPs for CSC-targeted therapy and highlighted the novel mechanism of MNPs effect on the BCSCs, which is an exciting and promising strategy for CSC-based cancer therapy.

4. Experimental Section

Materials: The reagents, cell lines, chemicals, and animals used in this work are shown in the Supporting Information. All the experiments were performed following the ethical guidelines of administration office committee of Hebei University.
Isolation of Cancer Stem Cells from Cell Lines. Human breast adenocarcinoma cell line (MCF-7), human cervical carcinoma cell line (Hela), and human hepatoma (HepG2) cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin under 37 °C and 5% CO2. BCSCs, cervical CSCs, and liver CSCs were sorted from the cell lines and incubated in a DMEM/F12 medium supplemented with 10 ng mL−1 bFGF, 2% B27, 1% N2, and 20 ng mL−1 EGF.
Western Blotting: The proteins from MCF-7 and BCSCs were collected after 4 h treatment with Fe3O4@SiNPs or MNPs by using RIPA-lysis buffer. The intracellular proteins and proteins on membrane were separated through extracting the cell membrane. Medium from Fe3O4@SiNPs or MNPs treated MCF-7 and BCSCs were also collected and concentrated through freeze drying and analyzed for protein concentration. Then 15–30 µg of total proteins were loaded into a 12% SDS-PAGE and separated under 90 V. Then the gels were transferred onto a nitrocellulose membrane, and then blocking in 10% bovine serum albumin (BSA). Primary antibody of Hsp90 or β-actin was added for overnight incubation at 4 °C. Nitrocellulose membranes were washed with TBST for 3 times, and then added the second antibody for 2 h incubation at room temperature. Nitrocellulose membranes were then washed 3 times and imaged with Bio-rad imaging system (GelDoc XR+).
Quantitative RT-PCR: Real-time PCR array was employed to study the stemness and related signaling pathways of MCF-7 and BCSCs. Trizol Plus RNA purification kit was used to extract the total RNA from cells which were treated with or without HSPI, or CD44-Fe3O4@SiNPs, or CD44-HSPI/Fe3O4@SiNPs (100 µg mL−1). The first-strand cDNA was obtained from RNA according to the TaKaRa protocol (TaKaRa, Tokyo). An RT2ProfilerPCR Array Profile (Cat #: PAHS-507Z and Cat #: PAHS176Z, Qiagen) containing 84 relevant genes was carried out using ABI StepOnePlus System (Applied Biosystems, USA). The PCR profile was performed according to the manual of the RT2ProfilerPCR Array. The relative amount of mRNA was expressed as fold change, which was calculated by the comparative CT (2−ΔΔCT) relative to control group as a reference: 2−ΔΔCT = 1.
In Vitro Tumor Spheroid Formation and Passage Assay: For the tumor spheres formation, cells with density of 3000 cells pre well were seeded and cultured in ultralow-attachment 12-well plates. To evaluate the proliferation of BCSCs, the tumor spheres were concentrated by gentle centrifugation, and then digested into single cells by trypsin-EDTA. Then the separated single cells were washed with PBS and reseeded in ultralow-attachment 12-well plates at a density of 3000 cells per well and cultured in a DMEM/F12 medium supplemented with 20 ng mL−1 EGF, 1% N2, 10 ng mL−1 bFGF, and 2% B27 for 10–12 d to form tumor spheres.
Expression of Stemness Markers of CSCs: The tumor spheres were digested to obtain single cell, and then cells were harvested by centrifugation. According to the manufacture, single CSC suspension was rinsed with PBS and stained with human FITC-conjugated anti-CD44, PE-conjugated anti-CD133, FITC-conjugated anti-Oct4, PE-conjugated anti-Hsp90. The images were captured under
In Vivo Tumorigenic Ability of BCSCs: MCF-7 and BCSCs (1000 cells per 100 µL) were suspended in 100 µL sterile PBS and xenografted on the back of the BALB/c nude mice (four weeks old) with subcutaneous injections. The tumor-bearing mice were monitored for up to 40 d to observe the tumorigenesis of BCSCs compared to MCF-7 cells. All animal experiments performed in this work were in compliance with the guideline of Medical Comprehensive Experimental Center of Hebei University (Approval No. IACUC-2018006).
Synthesis of Antibody-Modified HSPI/Fe3O4@SiNPs: First, FeCl2 (0.5 mL, 2 mol L−1) and FeCl3 (2 mL, 1 mol L−1) were dissolved in HCl (2 mol L−1) were mixed and added into diluted NH3 solution (25 mL, 0.7 mol L−1) and stirred for 30 min, followed 17-DMGA solution (prepared in PBS) was added in the mixture. Then, the mixture (in the presence of 350 µL Fe3O4 NPs (2.0 mg mL−1) and 50 µg mL−1 17-DMAG) was dispersed into 7.3 mL cyclohexane, 2 mL igepal co-520, and 1.6 mL hexanol and stirred for 1 h to form the microemulsion system, and then added 35 µL tetraethoxy orthosilicate (TEOS) into the mixture, followed by the addition of 75 µL aqueous ammonia for the TEOS hydrolysis under stirring for 24 h. After that, 50 µL TEOS and 50 µL 3-aminopropyl triethoxysilane (APTS)/50 µL carboxyethylsilane triol sodium salt (CETS) were added to the mixture for another 24 h stirring. Finally, acetone was added to destabilize the microemulsion system. The HSPI-loaded Fe3O4@SiNPs were isolated via centrifugation and washed in sequence with ethanol and D.I. water for purification.
To conjugate the specific antibody on the surface of HSPIloaded Fe3O4@SiNPs, the NPs were functionalized with carboxyl groups, and then activated by 1-ethyl-3-(3-dimethyllaminopropyl) carbodiimide hydrochloride (EDC), and N-hydroxysuccinimide (NHS) in 2-(N-morpholino)ethanesulfonic acid (MES) buffer for 30 min. The MES buffer was then replaced by PBS (pH 7.4), and then PE-labeled anti-CD44 (50 µL, 1 ×10−6 m) was added into the HSPI-loaded Fe3O4@SiNPs suspension and stirred for 3 h in the dark. The resulting multifunctional NPs with magnetic, fluorescence, and targeting properties were purified by washing with D.I. water and centrifugation, following by diluted in 500 µL PBS buffer and then were utilized for in vitro and in vivo experiments.
Characterization of Antibody-Conjugated HSPI/Fe3O4@SiNPs: TEM (Philips Tecnai G2 F20 S-TWIN) and SEM (Philips XL30) were used to study the morphology of the multifunctional NPs. XPS (Thermo Scoemtific K-Alpha) was used to conform the conjugation of anti-CD44 and HSPI/Fe3O4@SiNPs. The size and zeta potentials were examined by a Malvern Zetasizer NanoZS instrument (Malvern, NanoZS). Magnetic property of NPs was tested with a VSM (Model 1600, Digital Measurement System, Newton, MA). The heat generation capability of the multifunctional NPs was examined in a 5 mm diameter 10-turn induction coil powered by a 8 kW AMF generator (SPG-10A-I, Shenzhen Shuangping Power Supply Technologies Co. Ltd.). Three differentiation samples (PBS, SiNPs, and Fe3O4@SiNPs) were exposed to the AMF for 15 min. The frequency was kept constant at 250 kHz and temperature was monitored by using a thermometer immersed in a test tube containing 1 mL of solution.
In Vitro SiNPs Cytotoxicity Evaluation: The acute cytotoxicity of MNP carrier was tested by 4,5-(dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay. Cells were incubated with SiNP solutions at the concentrations of 25, 50, 100, and 200 µg mL−1 for 1, 2, and 3 d, and then 10 µL of MTT solution was added in the medium. The absorbance of the medium was obtained at 570 nm by using a microplate reader (Bio-Tek).
In Vitro Analysis of HSPI (17-DMAG) Release from Fe3O4@SiNPs: The HSPI-loaded Fe3O4@SiNPs (1 mL, 1 mg mL−1) were loaded in a dialysis bag with a molecular weight cut-off of 10 kDa. The dialysis bag was then immersed in 9 mL PBS and kept in a horizontal laboratory shaker maintaining a constant temperature under the AMF and stirring. The amount of 17-DMAG (300 mL) were collected and analyzed at defined time points via UV–vis spectrophotometry (PerkinElmer, PE Lamda 750, USA). The drug release rate was calculated through a concentrationabsorbance standard equation.
In Vitro Uptake and Internalization of MNPs by BCSCs: BCSCs were seeded on coverslip in 24-well plate at the density of 1 × 104 cells per well, and incubated for 24 h, then incubated with 20 µg mL−1 CD44conjugated HSPI/Fe3O4@SiNPs (CD44-HSPI/Fe3O4@SiNPs) for 1 h at 37 °C. The nuclear and endosomes were stained with 1 mg mL−1 Hochest33342 and 5 mg mL−1 FITC-transferrin (TfR) for 5 and 30 min, respectively. Then cells were rinsed three times with PBS, and fixed and mounted with cover slip. The fluorescent images were observed with a confocal microscope (LSCM880, ZEISS).
In Vitro Cell Viability: BCSCs (5 × 104 cells per well) were incubated with 100 µg mL−1 CD44-Fe3O4@SiNPs, CD44-HSPI/Fe3O4@SiNPs, SiNPs, and HSPI for 1, 24, and 48 h and then treated with AMF for 15 min by a 5 cm diameter 10-turn induction coil powered by a 3 kW. Cells without treatment were used as control. While frequency was kept constant at 250 kHz and temperature was monitored by using a thermometer immersed in a test tube containing 1 mL of solution. Cell survival was assessed by MTT assay, and the value of IC50 was calculated according to the reported method.[43–45]
Analysis of Apoptosis and Necrosis: BCSCs (5 × 105 cells per well) were treated with 100 mg mL−1 CD44-Fe3O4@SiNPs, CD44-HSPI/Fe3O4@ SiNPs, SiNPs, and HSPI under AMF for 24 h, and then collected to detect the apoptosis and necrosis of CSCs. CSCs were washed with PBS and tested by Apoptosis Detection Kits (Annexin V/PI, or YO-PRO-1/7-AAD, Life technologies) according to the protocol in the manufacturer. In brief, treated cells were stained with Annexin V and PI solution in the dark for 30 min, and then analyzed by flow cytometry (BD FACSCalibur, BD Biosciences).
In Vivo Therapeutic Efficacy of CD44-HSPI/Fe3O4@SiNPs: 4–6 week BALB/c nude mice (15–20 g) were obtained from Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China). To setup the tumor model, BCSCs at the density of 3 × 104 cells/100 µL were injected 1900160
into the subcutaneous space of back region of the nude mouse. The tumor-bearing mice (the tumor volume reached about 100 mm3) were administered with either PBS, or HSPI, or CD44-Fe3O4@SiNPs, or CD44HSPI/Fe3O4@SiNPs (2 mg kg−1) via tail vein (each group, n = 10), and then treated with thermotherapy under AMF (10 cm diameter 15-turn induction coil powered by an 8 kW). Tumor volume was measured every 4 d by using a caliper. After 30 d treatment, the nude mice were executed and tumors were weighed. The tumor volume can be calculated from the formula: length × width × depth × π/6.
In Vivo Distribution of MNPs in Nude Mouse Body: The BCSC-tumor bearing nude mice were injected with CD44-HSPI/Fe3O4@SiNPs or Fe3O4@SiNPs (the fluorescent dye was encapsulated in silica shell) via the tail vein. Images were captured under the in vivo imaging system (Xenogen IVIS Spectrum, PerkinElmer) at 0, 1, 2, 3, and 4 h after injection. The nude mice were killed at 24 h, and the ex vivo image of the organs (lung, heat, liver, kidneys, spleen, and tumor) was taken by the in vivo imaging system. The content of the Fe element in organs was analyzed by the inductively coupled plasma mass spectrometry (ICP-MS, Thermo Scientific ELEMENT 2).
Hemolysis Assay: To assess the hemolytic effect of the SiNPs, 500 µL of diluted Red blood cells (RBCs) suspension was incubated with CD44-HSPI/Fe3O4@SiNPs (final concentration 12.5, 25, 50, 100, and 200 mg mL−1) at 37 °C with gentle shaking for 1 h, followed added 500 µL PBS to final volume of 1 mL. The absorbance of the supernatant was tested by microplate reader at 540 nm. RBC suspension incubated with 1 mL H2O and 500 µL PBS as the positive and negative control, respectively.
Biochemical Analysis and Blood-Element Test: The blood of BCSCtumor bearing mice (after 30 d treatment) was collected, and then the serum biochemical markers tested according to the previous work.[28]
Immunohistochemical and Haematoxylin and Eosin Staining of Tumor Xenograft and Organ Iissues: After 30 d treatment, the tumor and other organs of nude mouse were fixed overnight in 10% neutral buffered formalin and embedded in paraffin blocks, including spleen, liver, heart, kidneys, and lung. The blocked organs were sliced into 5–8 µm sections, and then stained by using hematoxylin-eosin (H&E). The sections were then observed by a Digital Imaging System (Axioplan2, Zeiss).
Immunohistochemical staining of tumor xenograft sections was performed to confirm the significant therapeutic efficacy of multifunctional NPs to BCSCs. The tissue sections were blocked in serum, and then incubated with FITC-anti-CD44 and PE-anticonjugated CD133 at 37 °C for 1 h. The stained Alvespimycin tissues were imaged under a confocal laser scanning microscope.
Statistical Analysis: All data were presented as mean ± standard deviation (SD). Significant differences were calculated using the Student’s t-test or one way ANOVA, and p < 0.05 present the significant differences.

References

[1] T. Reya, S. J. Morrison, M. F. Clarke, I. L. Weissman, Nature 2001, 414, 105.
[2] C. J. Eaves, Nature 2008, 456, 581.
[3] H. Dianat-Moghadam, M. Heydarifard, R. Janhanban-Esfahlan, Y. Panahi, H. Hamishehkar, F. Pouremamali, R. Rahbarghazi, M. Nouri, J. Controlled Release 2018, 288, 62.
[4] T. J. Desai, D. G. Brownfield, M. A. Krasnow, Nature 2014, 507, 190.
[5] V. Balasubramanian, Z. H. Liu, J. Hirvonen, H. A. Santos, Adv. Healthcare Mater. 2018, 7, 1700432.
[6] J. J. Shi, P. W. Kantoff, R. Wooster, O. C. Farokhzad, Nat. Rev. Cancer 2017, 17, 20.
[7] X. W. Zeng, W. Tao, L. Mei, L. Q. Huang, C. Y. Tan, S.-S. Feng, Biomaterials 2013, 34, 6058.
[8] W. Cheng, C. Y. Liang, L. Xu, G. Liu, N. S. Gao, W. Tao, L. Y. Luo, Y. X. Zuo, X. S. Wang, X. D. Zhang, X. W. Zeng, L. Mei, Small 2017, 13, 1700623.
[9] X. W. Zeng, G. Liu, W. Tao, Y. Ma, X. D. Zhang, F. He, J. M. Pan, L. Mei, G. Q. Pan, Adv. Funct. Mater. 2017, 27, 1605985.
[10] X. W. Zeng, M. M. Luo, G. Liu, X. S. Wang, W. Tao, Y. X. Lin, X. Y. Ji, L. Nie, L. Mei, Adv. Sci. 2018, 5, 1800510.
[11] Y. M. Peng, J. P. Nie, W. Cheng, G. Liu, D. W. Zhu, L. H. Zhang, C. Y. Liang, L. Mei, L. Q. Huang, X. W. Zeng, Biomater. Sci. 2018, 6, 1084.
[12] W. Cheng, J. P. Nie, N. S. Gao, G. Liu, W. Tao, X. J. Xiao, L. J. Jiang, Z. G. Liu, X. W. Zeng, L. Mei. Adv. Funct. Mater. 2017, 27, 1704135.
[13] Y. Yong, X. J. Chen, T. Bao, M. Zu, L. Yan, W. Y. Yin, C. C. Cui, D. L. Wang, Z. J. Gu, Y. L. Zhao, ACS Nano 2015, 9, 12451.
[14] Z. J. Gu, S. Zhu, L. Yan, F. Zhao, Y. L. Zhao, Adv. Mater. 2018, 8, 1800662.
[15] S. Behzadi, G. A. Luther, M. B. Harris, O. C. Farokhzad, M. Mahmoudi, Biomaterials 2017, 146, 168.
[16] S. Y. Wang, G. Kim, Y. E. K. Lee, H. J. Hah, M. Ethirajan, R. K. Pandey, R. Kopelman, ACS Nano 2012, 6, 6843.
[17] H. Y. Chen, X. Zhang, S. H. Dai, Y. X. Ma, S. S. Cui, S. Achilefu, Y. Q. Gu, Theranostics 2013, 3, 633.1900160
[18] G. Tian, X. Zhang, Z. Gu, Y. Zhao, Adv. Mater. 2015, 27, 7692.
[19] M. Z. Lei, M. Ma, X. J. Pang, F. P. Tan, N. Li, Nanoscale 2015, 7, 15999.
[20] L. Zhou, H. Wang, Y. P. Li, Theranostics 2018, 8, 1059.
[21] Y. M. Peng, J. P. Nie, W. Cheng, G. Liu, D. W. Zhu, L. H. Zhang, C. Y. Liang, L. Mei, L. W. Huang, X. W. Zeng, Biomater. Sci. 2018, 6, 1084.
[22] J. E. Visvader, G. J. Lindeman, Cell Stem Cell 2012, 10, 717.
[23] Z. R. Stephen, F. M. Kievit, O. Veiseh, P. A. Chiarelli, C. Fang, K. Wang, S. J. Hatzinger, R. G. Ellenbogen, J. R. Silber, M. Zhang, ACS Nano 2014, 8, 10383.
[24] L. Whitesell, S. L. Lindquist, Nat. Rev. Cancer 2005, 5, 761.
[25] B. K. Eustace, T. Sakurai, J. K. Stewart, D. Yimlamai, C. Unger, C. Zehetmeier, L. Blanca, T. Claudia, H. W. Stefan, B. Gerald, S. T. Bradley, N. Len, L. I. Leodevico, G. J. Daniel, Nat. Cell Biol. 2004, 6, 507.
[26] K. Lee, V. P. Drachev, J. Irudayaraj, ACS Nano 2011, 5, 2109.
[27] X. C. Zheng, X. Wang, H. Mao, W. Wu, B. R. Liu, X. Q. Jiang, Nat. Commun. 2015, 6, 5834.
[28] J. D. Sun, Y. J. Liu, M. Ge, G. Q. Zhou, W. T. Sun, D. D. Liu, X.-J. Liang, J. C. Zhang, Sci. Rep. 2017, 7, 16236.
[29] A. L. Joly, A. Deepti, A. Seignez, A. Goloudina, S. Hebrard, E. Schmitt, S. Richaud, E. Fourmaux, A. Hammann, A. Collura, M. Svrcek, G. Jego, E. Robinet, E. Solary, O. Demidov, E. Kohli, C. Garrido, Oncogene 2016, 35, 2842.
[30] X. F. Wang, X. M. Song, W. Zhuo, Y. Fu, H. B. Shi, Y. Liang, M. M. Tong, G. D. Chang, Y. Z. Luo, Proc. Natl. Acad. Sci. USA 2009, 106, 21288.
[31] W. Li, D. Sahu, F. Fred Tsen, Biochim. Biophys. Acta, Mol. Cell Res. 2012, 1823, 730.
[32] J. Trepel, M. Mollapour, G. Giaccone, L. Neckers, Nat. Rev. Cancer 2010, 10, 537.
[33] H. Fuhrmannstroissnigg, Y. Y. Ling, J. Zhao, S. J. Mcgowan, Y. Zhu, R. W. Brooks, Nat. Commun. 2017, 8, 422.
[34] C. Thery, L. Zitvogel, S. Amigorena, Nat. Rev. Immunol. 2002, 2, 569.
[35] J. E. Visvader, G. J. Lindeman, Nat. Rev. Cancer 2008, 8, 755.
[36] Y. Morikawa, T. Heallen, J. Leach, Y. Xiao, J. F. Martin, Nature 2017, 547, 227.
[37] G. H. Zhang, Z. Liu, H. Ding, Y. Zhou, H. A. Doan, K. W. Sin, Z. R. J. Zhu, R. Flore, Y. F. Wen, X. Gong, Q. Y. Liu, Y. P. Li, Nat. Commun. 2017, 8, 589.
[38] G. F. Ball, Proc. Natl. Acad. Sci. USA 2017, 114, 9240.
[39] Y. Cai, J. Wang, R. Li, G. Ayala, M. Ittmann, M. Liu, Cancer Res. 2009, 69, 819.
[40] S. Dias, S. V. Shmelkov, G. Lam, S. Rafii, Blood 2002, 99, 2532.
[41] E. Pick, Y. Kluger, J. M. Giltnane, C. Moeder, R. L. Camp, D. L. Rimm, H. M. Kluger, Cancer Res. 2007, 67, 2932.
[42] P. Workman, M. V. Powers, Nat. Chem. Biol. 2007, 3, 455.
[43] C. Y. Liang, H. P. Wang, M. Zhang, W. Cheng, Z. H. Li, J. P. Nie, G. Liu, D. Z. Lian, Z. H. Xie, L. Q. Huang, X. W. Zeng, J. Colloid Interface Sci. 2018, 525, 1.
[44] D. W. Zhu, W. Tao, H. L. Zhang, G. Liu, T. Wang, L. H. Zhang, X. W. Zeng, L. Mei, Acta Biomater. 2016, 30, 144.
[45] X. W. Zeng, W. Tao, Z. Y. Wang, X. D. Zhang, H. J. Zhu, Y. P. Wu, Y. F. Gao, K. W. Liu, Y. Y. Jiang, L. Q. Huang, L. Mei, Part. Part. Syst. Charact. 2015, 32, 112.