Antimonene with two-orders-of-magnitude improved stability for high-performance cancer theranostics

Antimonene was engineered with significantly improved stability, superior tumor targeting capacity and increased photothermal efficacy for high-performance cancer theranostics.


Introduction
Exploring new materials with special functions has provided efficient ways for cancer diagnosis and treatment, 1-4 such as photoacoustic (PA) imaging, 5,6 magnetic resonance imaging (MRI), 7,8 photothermal therapy (PTT), 9,10 and photodynamic therapy (PDT). 11,12 Recently, antimonene (AM) has emerged as a new candidate for cancer theranostics due to its extremely strong light absorption in the NIR region. [13][14][15] This property endows it with the capacity of high photothermal conversion efficiency (AM quantum dots: 45.5%; AM nanosheets: 41.8%), 16,17 which was much higher than those of many other nanomaterial-based photothermal agents such as graphene oxide, Au nanorods, Au nanoshells and BP quantum dots. The high PTCE also resulted in sensitive PA signals in vivo. 17 Considering that most photothermal materials show a capacity for 1 O 2 generation, 18,19 the possibility of utilizing AM as a PDT agent thus exists, and this feature could further augment the anticancer performance of AM but remains unexploited.
Unfortunately, the direct utilization of AM for cancer theranostics is rare so far. One of the most critical issues is its rapid degradation in physiological medium. 20,21 Such poor stability not only closes the time window to play a therapeutic role in tumors, but also raises safety concerns. 22,23 As AM degradation highly depends on oxygen, 21 one efficient strategy to overcome this impediment is likely to decrease the specic surface area, which can be realized by regulating the shape and size. 24,25 The other feasible way may be surface modication, which can shield nanomaterials from the medium. [26][27][28][29][30] Due to the lack of reactive groups on the AM substance surface, traditional chemical reaction-based approaches seem impossible. To date, the only reported method for AM modication is coating with DSPE-PEG via van der Waals forces and hydrophobic interactions. 16,17 Although effective, this approach has not yet provided the silver bullet for full utilization in vivo. For example, the gradual exfoliation of these physically bound DSPE-PEG molecules is inevitable, which undoubtedly compromises the improvement in stability. 31 Additionally, an unexpected immunogenic response known as "accelerated blood clearance (ABC)" has been observed with PEGylated nanocarriers, resulting in increased clearance during repeated administration. 32,33 In addition, considering the stochastic nature of the enhanced permeability and retention (EPR) effect, an active targeting capacity is further needed to improve AM accumulation at tumor sites. [34][35][36] To improve the stability and thus maximize the theranostic performance in vivo, we herein engineered AM by using a synergic approach, which included dimension optimization, size control, and surface camouage (Scheme 1). Typically, distinct ultrasonic probes with high-power liquid exfoliation were utilized to prepare AM nanoparticles (NPs) and nanosheets (NSs), which could be separated with uniform size by differential centrifugation. Upon comparative evaluations, AM NPs with a size of $55 nm outperformed other candidates in terms of suitable size to utilize the EPR effect and superior stability for phototherapy. Subsequent camouaging of the cancer cell membrane (CM) would not only further improve stability but also endow mNPs with good tumor targeting capacity. Aer intravenous injection, the resulting CmNPs could accumulate at tumor sites due to their homotypic tumor self-recognition ability. Once exposed to NIR light, photoacoustic/photothermal multimodal imaging enabled noninvasive guidance. Additionally, efficient heat and 1 O 2 were generated for PTT and PDT, respectively. As a result, complete tumor inhibition could be achieved with few abnormalities, showing great promise for utilizing CmNPs as a new modality to ght against cancer.

Results and discussion
Synthesis, characterization and stability comparison of AM NSs and NPs AM NSs can be simply exfoliated from bulk powder through a modied two-step liquid exfoliation strategy, which involves ultrasound probe sonication together with ice-bath sonication in N-methyl-2-pyrrolidone. Unfortunately, the high specic surface area sourced from two dimensions (2D) facilitates the exposure of AM NSs to oxygen in the medium, thus accelerating the degradation process. To decrease the specic surface area, we utilized an ultrasound probe with a distinct jagged head and obtained AM NPs. For comparative evaluation, the obtained AM NSs and NPs were further separated with uniform size ($55 nm, Fig. 1a) by differential centrifugation. Compared with traditional 2D AM NSs, 3D AM NPs halved the dissolution rate (Fig. 1b), indicating their superior stability in medium. Correspondingly, we observed signicant increases in the absorbance at 808 nm and temperature aer irradiation.
Having demonstrated the benet of 3D, we next examined the effect of particle size, which was another important factor correlated with the specic surface area of NPs. Taking the aforementioned AM NPs as a middle reference (denoted as mNPs), we also prepared small AM NPs (sNPs, $2 nm) and large AM NPs (lNPs, $295 nm) by differential centrifugation for further comparative evaluation (Fig. 1c, S1 and S2 †). Compared with mNPs, sNPs with a 24-fold increase in specic surface area exhibited a signicantly accelerated dissolution rate, while an evidently decreased absorbance at 808 nm and temperature aer irradiation was observed ( Fig. 1d-f and S3 †), which excluded the feasibility of efficient phototherapy. In contrast, the 5-fold decrease in the specic surface area of lNPs resulted in moderately improved stability and photothermal effects. Based on this in vitro performance alone, lNPs seemed to be slightly superior to mNPs. However, the oversize of lNPs might, in turn, remarkably compromise the EPR effect due to the limited fenestration space of newly formed tumor vessels. 34 Taking these aspects together into consideration, mNPs with a suitable size for the EPR effect and good stability for phototherapy thus shed a brighter light on their in vivo performance.

Characterization and stability evaluation of CmNPs
Although improved, the stability of mNPs was still unsatisfactory. Even at a high concentration of 300 mg mL À1 , only 47.9% mNPs remained undispersed in PBS within 24 h, leading to a signicantly decreased photothermal efficacy (Fig. S4 †). To go a step further, we continued to consider shielding mNPs from the medium by surface modication. Instead of DSPE-PEG adsorption, we herein utilized the cancer cell membrane to physically enwrap mNPs, which not only mitigated the Scheme 1 Schematic illustration of the preparation of cancer cell membrane-camouflaged AM nanoparticles (CmNPs) and their application in multimodality imaging-guided combinatorial cancer therapy. exfoliation during AM dissolution but also provided tumor targeting capacity via homotypic tumor self-recognition. As shown in Fig. 2a and S5 †, the particle size, surface charge and protein of the CM coated nanoparticles, the colocalization of mNPs/CM signals in the confocal laser scanning microscopy (CLSM) images and the membrane layer around mNPs in the TEM image together demonstrated the successful CM camou-age. Especially, the outer lipid shell thickness ($8 nm) in the TEM image and the increase in the hydrodynamic diameter ($20 nm) aer CM coating indicated the single layer membrane coating on the surface of AM NPs. 37,38 CmNPs remained well dispersed with an average hydrodynamic diameter of approximately 106 nm (Fig. 2b), which was favorable for intravenous injection and the EPR effect. More importantly, the CmNPs exhibited a much improved particle stability ( Fig. S4 and S6 †). Taking the concentration of 100 mg mL À1 as an example, almost all sNPs and mNPs had dispersed within 1 h and 3 h at 37 C, respectively (Fig. 2c). In contrast, more than 80% CmNPs remained stable even over 24 h. The degradation half-life of CmNPs increased over 27.7-fold compared with that of mNPs, and this value could jump to 319.3-fold compared with that of sNPs (Fig. 2d). Correspondingly, the particle size, surface zeta potential, and polydispersity index (PDI) of CmNPs showed little change in one-week storage in PBS and 10% FBS medium (Fig. 2e, S7 †) if stored at 4 C. Such superior stability further endowed CmNPs with outstanding photoproperties.
As shown in Fig. 2f and S8, † the photothermal heating curves showed strong concentration-and time-dependent photothermal effects for CmNPs, and the temperature increment (DT) was up to 44.5 C at a concentration of 200 mg mL À1 aer 8 min of irradiation with a low power laser (0.8 W cm À2 ). Accordingly, the high photothermal effect resulted in a strong PA signal that showed a good linear relationship with the concentration (Fig. 2g). Meanwhile, typical 1 : 1 : 1 triplet signals in electron spin resonance (ESR) spectra (Fig. 2h) as well as the decreased signal of ABDA and DPBF ( Fig. S9 and S10 †) revealed the capacity of efficient 1 O 2 production. Moreover, we found that CmNPs had outstanding biocompatibility. They could be rapidly degraded via NIR irradiation as usual (Fig. S11 †), and more than 90% of the particles could be gradually dispersed in 7 days at 37 C even without NIR-irradiation (Fig. S12 †). As a result, no obvious in vitro and in vivo toxicity was observed ( Fig. S13 and S14 †). The aforementioned performance of CmNPs together paved the way for their in vivo application.

Specic targeting capability of CmNPs
In addition to the improved stability, another benet sourced from the camouage of the cancer CM could be the tumor targeting capacity. To explore this idea, we next systematically investigated the effect of CM cloaking on the uptake proles. As shown in Fig. 3a, camouaging mNPs with the 4T1 breast cancer CM (4T1/CmNPs, also denoted as CmNPs) signicantly increased the cellular uptake into 4T1 cells. Taking the dose of 100 mg mL À1 as an example, the increment of intracellular AM (detected by ICP) yielded 6.1-fold compared to that of naked mNPs. Such plentiful uptake could be attributed to the multiple surface adhesion molecules (such as E-cadherin and epithelial cell adhesion molecules) on the membrane with the homotypic tumor self-recognition ability. 39 For further verication, we comparatively investigated the uptake of CmNPs (labeled with Cy5, Fig. S15 †) into different cells, including 4T1, CT26, J774A.1, and HUVEC. As expected, a large number of CmNPs with marked uorescence were observed in 4T1 cells rather than other cells (Fig. 3b). Once the mNPs were cloaked with different CMs sourced from the 4T1, CT26, J774A.1 or HUVEC cells, 4T1 cells exhibited much higher uptake of CmNPs compared to their counterparts cloaked with other CMs (Fig. S16 †), again conrming the homotypic targeting performance of cloaked CmNPs.
To evaluate the targeting capacity in vivo, we next intravenously injected the formulations into the 4T1-xenograed mice and determined the intratumoral accumulation via timeelapsed PA signals ( Fig. 3c and S17 †). Although the mNP and CmNP groups shared similar kinetics in the tumor area, a much brighter PA signal was observed in the CmNP group, and the value of the CmNP group was 3-4 times higher than that of the mNP group. This effect should be attributed to the improved stability and targeting capacity sourced from the CM camou-age. For further verication, we simultaneously inoculated the mice with two different types of tumor xenogras at parallel sites (CT26 tumor on the le and 4T1 tumor on the right) and utilized in vivo uorescence imaging to investigate the biodistribution of the intravenously injected formulations (labeled with Cy7). As shown in Fig. 3d, the accumulation of mNPs and CmNPs in both 4T1 and CT26 tumors again shared similar kinetics with the peak at 24 h. However, CmNPs showed much superior accumulation compared to that of mNPs in the 4T1 tumor area, while their signals in the CT26 tumor area were inconspicuous (Fig. 3e). In this case, the indicator ratio (I 4T1/ CT26 ) for recognition specicity in the CmNP group reached up to 5, which was markedly higher than the value of the mNP group. Such a disparity could also be observed in the excised tissues and histological sections (Fig. S18 and S19 †), again demonstrating the specic targeting recognition ability to homologous tumors sourced from CM camouage.

Phototherapy performance of CmNPs in vitro
The excellent targeting properties of CmNPs encouraged us to explore their phototherapy performance. As shown in Fig. 4a, the viability of 4T1 cells aer 808 nm irradiation exhibited a dose-dependent effect. Owing to the improved stability and targeting capacity, the value in the CmNP group kept lower than that in the mNP group. The superior phototherapy performance of CmNPs was further conrmed by using calcein-AM and propidium iodide (PI) as probes to indicate the live and dead 4T1 cells, respectively (Fig. 4b), wherein treatment with CmNPs resulted in more dead cells (indicated by the red color). To gain a deeper insight, we also investigated the dual functions of PDT and PTT at the concentration of 100 mg mL À1 . Aer irradiation, a high level of local hyperthermia occurred in the CmNP group, while the temperature in the PBS group was almost unchanged (Fig. 4c inset). Accordingly, we observed the highest expression of HSP90 in CmNP-treated cells, indicating a superior PTT for hyperthermia treatment (Fig. 4c). With respect to PDT, irradiation of the PBS group exhibited little 1 O 2 (detected by DCFH-DA with a green signal). Notably, CmNPs again outperformed mNPs, since the 1 O 2 generation in the CmNP group was more than 3-fold of that in the mNP group (Fig. 4d). Once the hyperthermia from PTT was cooled by an ice-bath or the 1 O 2 from PDT was absorbed by quenching, the cell viability was signicantly recovered, in turn demonstrating the orchestration of PDT and PTT for potent cytotoxicity (Fig. 4e and S20 †).
In addition to the 2D culture system, we also evaluated the phototherapy performance on a 3D multicellular tumor spheroid (MCTS) model, which could closely mimic the native structure of tumor tissue and therefore provided a more accurate antitumor estimation. 40 As shown by the Live/Dead assay in Fig. 4f, all cells in MCTS remained alive (represented by a green  color) in both the mNP and CmNP groups without irradiation, again suggesting good biocompatibility. When irradiation was added, a few dead cells appeared in the mNP group, while almost all cells in the CmNP group were dead. Such a disparity further resulted in different effects on MCTS growth (Fig. 4g). As expected, the non-irradiation group showed little effect on MCTS development. Upon irradiation, the MCTS volumes decreased in both mNP(L) and CmNP(L) groups. Comparatively, the MCTS in the CmNP(L) group shrank to a much smaller level, again conrming a more potent phototherapy effect.

In vivo antitumor efficacy of CmNPs
Encouraged by the above results, we nally evaluated in vivo therapeutic efficacy. To this end, 4T1 tumor-bearing mice were randomly divided into ve groups: PBS(L), mNP, CmNP, mNP(L), and CmNP(L). Via the hyperthermia effect, we rst imaged the infrared thermographic maps and monitored real-time temperature changes in the tumor area (Fig. 5a).
Compared to the stable temperature in the non-irradiation group, the temperature in the mNP(L) group gradually increased to 48.3 C, while it climbed up to 57.1 C in the CmNP(L) group. To estimate the 1 O 2 production in tumors, DCFH-DA was intratumorally administered for in situ imaging by multiphoton laser confocal scanning microscopy and subsequent quantitative assay by ow cytometry (Fig. 5b and c). In the non-irradiation groups, little 1 O 2 signaling was found either in situ or ex vivo. Once upon irradiation, a much brighter signal could be observed in the CmNP(L) group, and the intensity was 3.3 times greater than that in the mNP(L) group.
The different photothermal effects and 1 O 2 generation cooperatively resulted in distinguishing the inhibition of tumor cell proliferation (detected by the expression of the Ki67 indicator in Fig. 5d). As could be seen, the mNP(L) group and CmNP(L) group showed moderate and signicant inhibition of Ki67 expression, respectively. Correspondingly, mNP(L) treatment slightly suppressed tumor development, and the mice gradually died within 40 days. In contrast, the tumor could be completely eliminated in the CmNP(L) group, and the survival rate was 100% during 100 day observation ( Fig. 5e and f, S21 †). As metastasis to distant lung and bone tissues was always observed for breast cancer in the clinic, the anti-metastasis performance was also evaluated (Fig. 5g, S22 †). Compared with other groups that showed metastatic foci in the lungs and metastasis-induced erosion in the tibia to different extents, no signs of metastasis were observed in the CmNP(L) group, again conrming the superior phototherapy performance of CmNPs. Considering the safety concerns of antimony, 41-43 the biosafety was systematically evaluated again. Few abnormalities were found in all markers in CmNP(L) treated mice (Fig. S23, Table S1 †). Such a good safety should be attributed to the superior targeting capacity of CmNPs, which could decrease the exposure of antimony in normal organs. Therefore, the safety concerns of CmNPs could be eliminated. All results indicated that the developed CmNPs held great promise for cancer theranostics.

Conclusions
In summary, we succeeded in signicantly improving the stability of mNPs by the cooperation of dimension optimization, size control, and CM camouage, which meanwhile endowed the mNPs with homotypic targeting capacity at tumor sites. The resulting CmNPs thus exhibited not only satisfactory PA/photothermal multimodal imaging but also potent PDT/PTT combination anticancer therapy with no notable side effects. The maximized theranostic performance in vivo therefore supported CmNPs as a safe and high-performance modality to ght against cancer.

Animals and Tumor Models
Four-week-old Balb/c mice were purchased from Vital River Laboratories (Beijing, China). Animals received care in accordance with the Guidance Suggestions for the Care and Use of Laboratory Animals. The animal protocol was approved by the Institutional Animal Care and Use Committees at the Institute

Conflicts of interest
There are no conicts to declare.