H2O2 self-providing synergistic chemodynamic/photothermal therapy using graphene oxide supported zero valence iron nanoparticles

Chemodynamic therapy (CDT) represents an emerging modality that treats cancer and other malignant diseases by using Fenton or Fenton-like catalysts to decompose hydrogen peroxide (H2O2) into toxic hydroxyl radicals (·OH). Despite its great promise, chemodynamic therapy is still limited by low endogenous H2O2 levels and lack of highly efficient nanocatalysts. In this study, we have developed multi-functional therapeutic nanocomposites GO–ZVI–GOx (GO = graphene oxide, ZVI = zero valence iron nanoparticles and GOx = glucose oxidase), where the GOx can catalyze the intracellular glucose and self-produce H2O2 for enhanced CDT therapy, and the GO is used as a template to avoid the aggregation of ZVI nanoparticles and also as an excellent photo-thermal converter for photothermal therapy under near-infrared (NIR) light. Our results show that this H2O2 self-generating nanoplatform can produce substantial amounts of reactive radicals under 808 nm NIR light due to the combinational effect of dual chemodynamic and photothermal therapy, which eventually leads to a significant decrease in cancer cell viability. It is believed that the methodology developed in this study enables conventional chemodynamic therapy to be efficiently improved, and holds great potential for overcoming challenges in many other H2O2-dependent cancer therapies.


Introduction
Chemodynamic therapy (CDT) refers to the cancer treatment that utilizes chemical agents to catalyze endogenous hydrogen peroxide (H 2 O 2 ) into highly reactive oxygen species (ROS) via the Fenton or Fenton-like reactions in order to induce cell death. [1][2][3] Considering that the tumor microenvironment (TME) is characterized by mild acidity, Fenton-based CDT therapy is advantageous in its high tumor selectivity and specicity. 4,5 Part of the reason is that the Fenton reaction is signicantly limited by the slightly basic microenvironment in the normal tissue region. 6,7 Therefore, CDT therapy has emerged as a promising strategy for selective intervention in cancer and other malignant diseases. 8,9 To date, many types of metal ions (e.g., Fe, Co, Ni, and Mn) have demonstrated their outstanding catalytic capability of decomposing H 2 O 2 into toxic hydroxyl radicals ($OH). 1 Excessive amounts of these reactive radicals can cause deleterious oxidative stress that can severely alter many types of bimolecular structures, such as cell membranes, proteins, lipids, deoxyribonucleic acid (DNA), etc.10 Among these Fenton reagents, zero-valence iron (ZVI) represents a class of iron-based nanoparticles which have been long applied in the degradation of organic pollutants, 11 removal of heavy metal ions, 12 and in situ remediation of soil, 13 due to their exceptional properties including large surface-area-to-volume ratio, low standard potential (E 0 ¼ À0.44 V), and high reactivity. [14][15][16] In addition, recent results have also shown promising prospective of ZVI for cancer therapy as an effective antitumor reagent. For example, Wu et al. reported that ZVI-based nanoparticles could selectively inhibit cancer cell but spare normal healthy ones, 17,18 which was attributed to the cancer-specic cytotoxicity of the non-oxidized ZVI core in these nanoparticles. 19,20 Later, amorphous ZVI nanoparticles were reported and used for cancer therapy by triggering the local Fenton reaction in the tumor region. 2 In a more recent study, PVP (polyvinyl pyrrolidone) modied ZVI nanoparticles were demonstrated to have great potentials to serve as both the contrast agents for magnetic resonance imaging and the therapeutic reagents for synergetic cancer therapy. 21 Despite the great progress, the application of ZVI-based Fenton therapy is still limited by the low endogenous H 2 O 2 level, insufficient generation of ROS and easy aggregation of ZVI nanoparticles. Restricted by the low catalytic efficiency of Fenton reactions, it is still rather challenging to completely eliminate tumors relying solely on the ZVI-based CDT treatment. In this context, a synergistic therapy combining CDT and other treatment modality complimentary to CDT, would signicantly amplify the therapeutic efficiency. Herein, we reported a self-supplying H 2 O 2 nanoplatform that can simultaneously achieve both chemodynamic and photothermal (PTT) therapy. The specially designed nanoagent is composed of graphene oxide (GO) deposited by ZVI and GOx (glucose oxidase), where the GOx can catalyze the intracellular glucose into H 2 O 2 and continuously provide H 2 O 2 source for CDT treatment, thus removing the dependence of ZVI-based Fenton reaction on endogenous H 2 O 2 . Besides, GO is used as a template to prevent the ZVI from aggregating and also as a good photo-thermal converter for PTT therapy under nearinfrared light (Fig. 1). The results show that substantial amount of $OH radicals is generated upon excitation with the 808 nm laser light, as a result of the combinational effect of chemodynamic and photothermal therapy. Consequently, the growth of cancer cells is signicantly prohibited. It is believed that the therapeutic nanoplatform developed in this work represents an efficient modality to resolve issues in the ZVIbased CDT treatment (e.g., insufficient endogenous H 2 O 2 level and easy aggregation of ZVI nanoparticles) and hold great potential for improved cancer therapy. purchased from Sigma-Aldrich. Graphite, sodium borohydride (NaBH 4 ), 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (EDC), N-hydroxysuccinimide (NHS), potassium iodide (KI) and benzoic acid were purchased from Adamas. Glucose oxidase (GOx) was purchased from Aladdin. Cell culture medium (DMEM), fetal bovine serum (FBS) and penicillin-streptomycin were purchased from Gbico. Cell Counting Kit-8 (CCK-8) was purchased from Dojindo.

Synthesis of GO-ZVI
A modied Hummers method used to synthesize GO can be found elsewhere, 22,23 while ZVI was prepared by a liquid-phase reduction method. 24,25 In brief, the synthesis of GO-ZVI nanocomposite is as follows: 10 mL GO with a concentration of 5 mg mL À1 was added to 100 mL three-necked ask, and different amounts of FeSO 4 $7H 2 O was weighed and dissolved in the deionized water. Then, the dissolved FeSO 4 $7H 2 O was added to the above GO solution and stirred overnight to ensure the adsorption of Fe 2+ ions on the GO surface. Next, the mixed solution was aerated with Ar gas for 30 minutes to remove the O 2 from the ask, and then different amounts of NaBH 4 were dropwise added into the mixed solution of GO and FeSO 4 $7H 2 O through the separator funnel. Then, the reaction was kept for another 30 minutes to ensure the sufficient production of ZVI nanoparticles on the GO surface. Next, the obtained black precipitate was centrifuged at 13 000 rpm for 15 minutes, and later dissolved in deoxygenated deionized water. The solution was again centrifuged at 4000 rpm for 5 minutes, the supernatant was retained. The process was repeated for 3 times. Finally, the samples of GO-ZVI with different GO/Fe ratios were sealed and stored. The procedure of preparation of ZVI was the same as that of GO-ZVI, but without the addition of GO.

Synthesis of GO-ZVI-GOx
The loading of GOx on GO-ZVI was performed by the assistance of EDC and NHS. Specically, 10 mL EDC, 0.6 mg NHS and 0.2 mg GOx were rst dissolved in PBS at a pH of 6.5, and constantly stirred for 2 hours. Then, 10 mL GO-ZVI was added to the above solution and stirred overnight. Next, the obtained GO-ZVI-GOx sample was centrifuged and washed three times with acid deoxygenated water and sealed for later use.
2.4 Synthesis of GO-Fe 2 O 3 0.5 g Fe (NO 3 ) 3 $9H 2 O was weighed and dissolved in 10 mL ultrapure water. Then, GO solution with a concentration of 5 mg mL À1 was added dropwise to the above solution. Aer stirring, the solution was transferred to the drying oven at 60 C for aging about 8 hours. Next, the products were washed with ultra-pure water for three times. Aer centrifugation at 13 000 rpm for 15 minutes, the precipitates were dried in the drying oven at 70 C for later use.

In vitro detection of H 2 O 2 produced by GO-ZVI-GOx
The generation of H 2 O 2 was qualitatively detected by KI. This is because H 2 O 2 can react with KI and produce I 3À ions, which has a strong absorption peak at around 350 nm. In brief, KI was rst added to the aqueous solution with a glucose content of 20 mM, and then GO-ZVI-GOx was added to the above solution. Aer 10 minutes of reaction, the supernatant was centrifuged and its absorption intensity in the spectral region from 250 nm to 500 nm was measured.

In vitro detection of $OH produced by GO-ZVI-GOx
Since the $OH radicals produced by the Fenton reaction could react with benzoic acid (C 6 H 5 COOH) and form a uorescent product, the benzoic acid was used as a probe to detect the generation of $OH by GO-ZVI-GOx. Specically, 100 mL glucose solution (2 mol mL À1 ) was added into the 10 mL benzoic acid solution. The pH of the mixture was adjusted to different values (i.e., 7.5 and 6.4), followed by the addition of 500 mL GO-ZVI-GOx with the concentration of 2 mg mL À1 . Aer 10 minutes, the mixture was centrifuged and the uorescence of the supernatant was measured by the spectrouorometer.

Evaluation of photothermal performance of GO-ZVI-GOx
The in vitro photothermal capability of as-prepared GO-ZVI-GOx under near infrared laser light was assessed as follows: 1 mL GO-ZVI-GOx solution with different concentrations (100, 200 and 400 mg mL À1 ) was placed 2.3 cm away from an 808 nm laser light and irradiated for six minutes at different power densities (1 W cm À2 , 1.5 W cm À2 , and W cm À2 ). The temperature of the solution upon irradiation was measure every 30 seconds using a thermocouple probe. The photothermal efficiency (h) was calculated by the following equations: 26 t ¼ Às s ln q (2) h ¼ hSðT max À T surr Þ À Q Dis I À 1 À 10 ÀA 808 Á (4) in which T max is the maximum temperature (i.e., 52.2 C), T surr is the surrounding temperature (i.e., 23.3 C). Then, the relation of t and -ln q can be obtained according to eqn (2). The time constant of this system, s s , is determined to be 207 s through the linear tting. In eqn (3), h is heat transfer coefficient, S is the surface area of the cuvette, m D and C D are the mass and heat capacity of ultrapure water used as the solvent, which are 1 g and 4.2 J g À1 . The hS in eqn (3) is 0.0203 mW per C. In eqn (4), Q Dis is the baseline energy induced by the sample cell, which is negligible because of its small value. I is the laser power (1.5 mW) and A 808 is the absorption intensity of nanomaterials at 808 nm, which is measured to be 0.011. Finally, the photothermal efficiency (h) of GO-ZVI-GOx under 808 nm irradiation is 15.64% according to eqn (4).

Cytotoxicity of GO-ZVI-GOx
HeLa cell line was used to evaluate the toxicity of GO-ZVI-GOx. HeLa cells were cultured in DMEM medium containing 10% FBS, 100 units mL À1 of penicillin, and 100 mg mL À1 streptomycin and incubated in a 37 C incubator with a moderate concentration of 95% and 5% carbon dioxide. The CCK-8 was used to detect the cell viability. HeLa cells were rst seeded in 96-well plates with 8000 cells per well and the pH of cell culture medium DMEM was adjusted to be 7.4 and 6.5, respectively. Aer 24 hours of incubation, 100 mL no-glucose DMEM containing different concentrations of GO-ZVI-GOx (0, 12.5, 25, 50 100 and 200 mg mL À1 ) was added, and ve parallel wells were prepared for each concentration. Aer continuous incubation for another 24 hours, the DMEM containing nanomaterials was discarded and the cells were rinsed with PBS for twice. Then, 100 mL DMEM containing 10% CCK-8 was added into each well to selectively stain the viable cell, and the cells were incubated for another 3 h. The absorption of each well at 450 nm was monitored by microplate reader.

Chemodynamic/photothermal therapy of GO-ZVI-GOx
HeLa cells were seeded in 96-well plates with 8000 cells per well and incubated for 24 hours. In order to assess the chemodynamic therapy alone, the high-glucose DMEM (pH ¼ 7.4 and pH ¼ 6.5) containing GO or GO-ZVI-GOx with different concentration (0, 12.5, 25, 50 100 and 200 mg mL À1 ) was added. Then, HeLa cells were incubated in 37 C for another 24 hours and DMEM containing GO or GO-ZVI-GOx was replaced by fresh DMEM containing CCK-8. For the evaluation of synergistic chemodynamic and photothermal therapy, the culture medium was replaced with fresh DMEM (pH ¼ 6.5) containing GO-ZVI-GOx with different concentration. Then 808 nm laser was irradiated for 5 minutes at a power density of 1.5 W cm À2 . Aer incubation for 24 hours, the medium containing GO-ZVI-GOx was discarded and a fresh medium containing CCK-8 was added. The absorbance at 450 nm was tested by microplate reader.

Intracellular detection of ROS
A 2,7-dichlorodi-hydrouorescein diacetate (DCFH-DA) probe was used to detect intracellular ROS generated by Fenton reaction because the DCFH-DA probe can be converted to DCF and emit strong green signals by the 488 nm light excitation. HeLa cells were rst seeded in confocal dishes with 1.5 Â 10 5 per dish and incubated with PBS,GO or GO-ZVI-GOx for 24 hours. Then, 2 mL high-glucose DMEM containing 1 mL DCFH-DA solution was substituted for the previous DMEM and continued to be incubated for a total of 30 minutes. For the group of chemodynamic therapy, cells were washed with PBS for three times and then the uorescence images under 488 nm light excitation were observed under confocal laser scanning microscope. For the group of dual chemodynamic/photothermal therapy, aer washing cells with PBS for three times, the cells were irradiated with 808 nm at a power density of 1.5 W cm À2 for 5 minutes, and then the uorescence imaging was observed under confocal laser scanning microscope.

Calcein-AM/PI staining
Calcein-AM (green uorescence) and propidium iodide (PI, red uorescence) was used to stain the live and dead HeLa cells, respectively. Briey, HeLa cells were rst seeded on 12 well plates at a density of 1.5 Â 10 5 and incubated with PBS,GO and GO-ZVI-GOx for 24 hours. Next, for the laser-treated group, cells were irradiated with 808 nm laser light (1.5 W cm À2 ) for 5 min. Aer incubation for another 4 hours, buffer solution containing Calcein-AM and PI was added and then the uorescence imaging was observed under confocal laser scanning microscope.

Animal model and in vivo treatment
All animal procedures were performed in accordance with the guidelines for care and use of laboratory animals of Shanghai University and experimental protocols were approved by the Animal Ethics Committee of Shanghai University. Female BALB/c mice of 4 weeks were used for in vivo tumor therapy, which were purchased from Shanghai Laboratory Animals Center (SLAC, shanghai). The mouse tumor model was established by subcutaneously injecting 5 Â 10 6 HeLa cells into the right oxter of each mouse. When the tumor volume reached about 100 mm 3 , the in vivo therapy was conducted. Tumorbearing mice were randomly divided into 6 groups, which were (1) PBS, (2) PBS + 808 nm light, (3) GO, (4) GO + 808 nm light, (5) GO-ZVI-GOx, (6) GO-ZVI-GOx + 808 nm light. Each group was injected with 100 mL of the corresponding samples every three days. For the groups of PBS + 808 nm light, GO + 808 nm light and GO-ZVI-GOx + 808 nm light, the 808 nm laser light (1.5 W cm À2 ) was applied for 5 minutes aer each injection. Besides, body weight and tumor size were measured every two days. The normalized body weight was calculated as m/m 0 , where m and m 0 represents the weight of each weighing and initial weight, respectively. The normalized tumor volume also calculated by the volume of each measuring to the initial volume, and the formula used to calculate the tumor volume is V ¼ (LW 2 )/2, where L and W represents tumor length and width, respectively. Aer 14 days of treatment, the mice were sacri-ced. Their vital organs, including heart, liver, spleen, lung, kidney and tumor, were removed and sectioned, stained with H&E, followed by histological analysis under microscope.

Synthesis and characterization of GO-ZVI-GOx nanocomposites
The synthesis procedure of GO-ZVI-GOx nanocomposites mainly consists of three steps, as illustrated in Fig. 1a. Firstly, GO (graphene oxide) was obtained by a modied Hummers method, 22,23 and then used as a template to grow ZVI (zero valence iron) nanoparticles and obtain the GO-ZVI nanostructures. Lastly, through EDC/NHS surface modication, the glucose oxidase enzyme (GOx) was conjugated with GO-ZVI due to the electrostatic forces. In Fig. 2a, the TEM (transmission electron microscope) image of GO shows that the as-prepared GO forms the shape of gauze with a large amount of wrinkles, suggesting that the GO exhibits a large specic surface area and can provide favorable nucleation sites for the growth of ZVI nanoparticles. Fig. 2b and c show the SEM (scanning electron microscopy) and TEM image of the as-synthesized ZVI nanoparticle, respectively, demonstrating that the ZVI nanoparticles are aggregated and particularly organized in the chain-like structure due to the magnetic dipole interactions. 27,28 In order to solve the aggregation issue, the ZVI nanoparticles were in situ anchored on the GO surface, which process was mainly based on the redox reaction between the NaBH 4 and Fe 2+ ions adsorbed on GO surface. The corresponding chemical equation is as follow: Besides, the water solubility of the GO-ZVI nanocomposites should also be taken into consideration to meet the requirements for biological applications. Therefore, the optimum ratio of GO to ZVI is determined by comparing GO-ZVI nanocomposites with different Fe : GO ratios of 0.1 : 1, 0.5 : 1, 1 : 1 and 5 : 1. Fig. 2d presents the TEM image of the GO-ZVI sample with the Fe : GO ratio of 0.5 : 1 and Fig. S1 † shows the TEM images of the GO-ZVI sample with the Fe : GO ratio of 0.1 : 1, 1 : 1, and 5 : 1. Clearly, with the increase of Fe : GO ratio, there is an increased amount of nanoparticles observed on the GO surface. For the GO-ZVI sample with the Fe : GO ratio of 0.5 : 1, the ZVI nanoparticles are uniformly distributed on the GO surface (Fig. 2d). Moreover, the higher magnication TEM image (inset in Fig. 2d) also conrms that the ZVI nanocrystals about 5-10 nm in size are deposited on the GO sheet. However, with the Fe : GO ratio higher than 0.5 : 1, the agglomeration of ZVI nanoparticles also becomes more pronounced, as shown in Fig. S1. † Water dispersity of the GO-ZVI nanocomposites are also compared among different Fe : GO ratios before and aer 24 hour standing, see Fig. S2. † For the GO-ZVI samples with Fe : GO ratio of 0.1 : 1 and 0.5 : 1, no precipitation is observed at the bottom of the cuvette, while a large amount of precipitation appears for the GO-ZVI samples with Fe : GO ratio of 1 : 1 and 5 : 1. In view of the aggregation and water dispersity, the Fe : GO ratio of 0.5 : 1 is selected as the optimal value for preparing the GO-ZVI nanocomposites. The phase composition of the asprepared GO-ZVI nanocomposites was further investigated by the X-ray diffraction (XRD) technique. In Fig. S3, † XRD patterns are compared for GO, ZVI and GO-ZVI samples, where GO-ZVI sample exhibits two major diffraction peaks at 2q ¼ 10.5 and 44.5 , corresponding to the (001) crystal plane of GO and (110) crystal plane of ZVI nanoparticles. This is another conrmation of successful deposition of the ZVI nanoparticles onto the GO surface. Following the synthesis of GO-ZVI, the EDC/NHS modied GOx is further anchored onto the GO-ZVI nanocomposites. Fig. 2e presents the TEM image of the obtained GO-ZVI-GOx nanocompounds with ZVI nanoparticles still remain uniformly distributed throughout the GO surface, suggesting that the GOx decoration does not interrupt the distribution of ZVI nanoparticles in the system. In Fig. 2f, X-ray photoelectron spectra (XPS) of the GO-ZVI-GOx nanocomposites are obtained in which the peaks at around 706.9 and 720.2 eV correspond to the Fe 0 and peaks at 711.1 and 725.1 eV indicate the existence of Fe 2 O 3 . Since XPS technique is very sensitive to the surface structures, these results suggest that the surface of ZVI nanoparticles are slightly oxidized to iron oxide. 15 Notably, previous studies have also reported that the non-oxidized ZVI core still exhibits the cancer-specic cytotoxicity regardless of surface oxidization. 19,20 In order to verify the successful loading of GOx on the GO-ZVI, the zeta potential and UV-vis absorption spectra of GO-ZVI-GOx were then recorded. As shown in Fig. 3a, the zeta potential of GO-ZVI and GOx is measured to be À30.37 mV and À21.7 mV, respectively. However, aer surface modication by EDC/NHS, the zeta potential of GOx-EDC/NHS changes to be +4.38 mV due to the introduction of positively charged amine group. The reason why GOx was rst modied with EDC/NHS was to increase the stability and conjugation yield. 29 In addition, the electrostatic attractions between the oppositely charged surface of GO-ZVI and GOx-EDC/NHS would lead to the formation of GO-ZVI-GOx nanocomposite upon mixing of the two solutions. Fig. 3b shows the UV-vis absorption spectra of GOx, GO-ZVI and GO-ZVI-GOx, which clearly demonstrates the GO-ZVI-GOx and pure GOx samples have similar features in the spectral region from 200 nm to 300 nm. However, the GOx-free GO-ZVI sample exhibits no evident absorption signals within this range. More importantly, the infrared spectra in Fig. 3c also indicates that both GO-ZVI-GOx and GOx display evident features around 1650 cm À1 and 1540 cm À1 (indicated by the black arrows) corresponding to the amide I and amide II band of the amide group of GOx, respectively. 30 Besides, we have also studied the stability of GO-ZVI-GOx in different solutions before and aer 24 hour standing (Fig. S4 †). As shown in Fig. S4, † no precipitates are observed for the four solutions (i.e., H 2 O, PBS, ethanol, and DMEM) aer standing for 24 hours, indicating the good stability of as-prepared GO-ZVI-GOx nanocomposites. Particularly, the good solubility and stability of GO-ZVI-GOx nanocomposites in DMEM also allows for the subsequent in vitro experiments.

Evaluation of H 2 O 2 productivity by GO-ZVI-GOx nanocomposites
The capability of the as-prepared GO-ZVI-GOx to produce H 2 O 2 was evaluated using the KI reagent. With the existence of H 2 O 2 , the KI can be reduced to I 3À , which has a prominent absorption  around 270-400 nm and can thus be used as a probe to detect H 2 O 2 generation. 31 As shown in Fig. 4a, there is no evident absorption in the spectral range from 250 nm to 400 nm for the sample groups of KI (black), KI + glucose (red), and KI + GO-ZVI-GOx (blue). However, for the group of KI + glucose + GO-ZVI-GOx (purple), obvious absorption peaks appeared around 290 nm and 350 nm, suggesting the generation of H 2 O 2 . In addition, only the solution treated with the group of KI + glucose + GO-ZVI-GOx turns to light-yellow among all the four tested samples (inset in Fig. 4a). This is a clear indication that the capability of GOx to catalyze glucose to H 2 O 2 is still remained aer being loaded onto the GO-ZVI surface.

Investigation of Fenton activity by GO-ZVI-GOx nanocomposites
The Fenton activity of the GO-ZVI-GOx nanocomposites to generate reactive radicals (e.g., $OH) was further evaluated by the benzoic acid. Notably, the benzoic acid (BZA) can be readily oxidized by the reactive $OH radicals and produce hydroxybenzoic acid characterized with high uorescence in the spectral range from 350 nm to 500 nm. 32 As a result, the generation of reactive radicals during the Fenton reaction can be effectively monitored by measuring the variations of uorescence intensity of benzoic acid. Fig. 4b compares the integrated uorescence intensity of benzoic acid solution containing GO-ZVI in the presence or absence of H 2 O 2 at different pH values (i.e., 5.4 and 6.5). It can be seen that there are no signicant uorescence signals without the existence of H 2 O 2 , independent of the pH of the solutions. In contrast, an evident increase in the uorescence is observed in the presence of H 2 O 2 . Particularly, the uorescence intensity at pH of 5.4 is much stronger than that at pH of 6.5, suggesting the pH-dependent performance of Fenton reactions. 33,34 Fig. 4c compares the integrated uorescence intensity of benzoic acid solution containing GO-ZVI-GOx and glucose solution at pH 5.4 (dark red) and pH 6.5 (blue), both of which show adequate uorescence signals even without the presence of H 2 O 2 , indicating good capability of GOx to catalyze glucose and produce H 2 O 2 for Fenton reactions. Besides, with lower pH, much higher uorescence intensity is observed. This pH-related Fenton behavior is crucial for the efficient and precise chemodynamic therapy, as the microenvironments of the cancer cells are characterized with lower pH than normal cells. Based on these observations, the proposed pathway of Fenton reactions for GO-ZVI-GOx nanocomposite is as follows (Fig. 4d): (1) Glucose in the solution is rstly catalyzed to H 2 O 2 by GOx anchored on the GO-ZVI surface.
(2) The highly active ZVI nanoparticles are oxidized to Fe 2+ ions by the H 2 O 2 in the acidic environment.

Photothermal evaluation by GO-ZVI-GOx nanocomposites under NIR light
It has long been recognized that the two major components, i.e., GO and ZVI, of the as-prepared GO-ZVI-GOx nanocomposite are promising nanoagents for photothermal therapy, owing to their high absorbance at the NIR window, outstanding photothermal conversion efficiency and strong photostability. Therefore, the in vitro photothermal capability of our GO-ZVI-GOx composite was evaluated through exposure to the 808 nm laser light. As shown in Fig. 5a, the temperature of the test solution is positively related to the concentration of GO-ZVI-GOx nanocomposites. For example, the temperature was elevated by 22 C in six minutes when the concentration of GO-ZVI-GOx was 100 mg mL À1 , and by 27 C when the concentration of GO-ZVI-GOx was 200 mg mL À1 . Fig. 5b presents the temperature variation via irradiation by 808 nm laser light at different power densities (1.0 W cm À2 , 1.5 W cm À2 , and 2.0 W cm À2 ) with the same concentration of GO-ZVI-GOx (200 mg mL À1 ). As expected, the temperature rises with an increase in the laser power density. The maximum temperature increase reaches around 35 C aer exposure to the 808 nm laser light for six minutes at the power density of 2.0 W cm À2 . In order to further evaluate the photothermal conversion efficiency of GO-ZVI-GOx, the GO-ZVI-GOx solution was exposed to the 808 nm laser light for 400 seconds. The temperature change of the GO-ZVI-GOx solution was then recorded while the laser light was off (Fig. 5c). The tted time constant for the heat transfer from the system is determined to be 207 s and the photothermal conversion efficiency(h) is calculated to be 15.64% (Fig. 5d), which is comparable to some conventional photothermal agents such as semiconducting polymer nanoparticles (15.8%) 35 and higher than gold nanoshells (13%). 36 This high photothermal conversion efficiency under 808 nm NIR light is believed to endow the as-prepared GO-ZVI-GOx nanocomposites with good potential for PTT therapy of tumor cells. Notably, the h for the GO-ZVI sample is calculated to be 16.62% (Fig. S5 †), also indicating that the GOx loading on GO-ZVI has neglectable effect on the photothermal conversion efficiency.

Comparison of GO-ZVI and GO-Fe 2 O 3
In addition, we have also synthesized GO-Fe 2 O 3 nanocomposites for a better comparison with GO-ZVI in the aspects of Fenton activity and photothermal capabilities. Fig. 6a presents the TEM image of as-prepared GO-Fe 2 O 3 nanocomposites, which demonstrates that the Fe 2 O 3 nanoparticles are uniformly distributed on the GO surface. In Fig. 6b, the XRD patterns of GO-Fe 2 O 3 are compared with GO and Fe 2 O 3 , where the GO-Fe 2 O 3 sample exhibits several diffraction peaks that correspond to different crystallographic planes of GO and Fe 2 O 3 . This conrms the successful deposition of the Fe 2 O 3 nanoparticles onto the GO surface. Next, the Fenton catalytic activity of GO-ZVI and GO-Fe 2 O 3 is compared. Similar to Fig. 4, benzoic acid (BZA) is also used to evaluate the $OH radicals generated during the Fenton reaction. Fig. 6c compares the integrated uorescence intensity of benzoic acid solution containing GO-ZVI, GO-Fe 2 O 3 and H 2 O in the presence of H 2 O 2 . Clearly, it can be obtained that the uorescence intensity of GO-ZVI is about one third time stronger than that of GO-Fe 2 O 3 . This is because the oxidization of ZVI can produce large amount of Fe 2+ , which promotes more generation of hydroxyl radicals 37 . Fig. 6d presents the temperature variations of GO-ZVI solution (200 mg mL À1 ), GO-Fe 2 O 3 solution (200 mg mL À1 ) and H 2 O illuminated by 808 nm light (2W cm À2 ), which also demonstrates that the temperature rises higher in the GO-ZVI solution than the GO-Fe 2 O 3 solution. Therefore, it can be obtained that the GO-ZVI nanocomposite exhibits better performance than GO-Fe 2 O 3 in the aspects of both Fenton activity and photothermal capability.

In vitro synergistic chemodynamic/photothermal therapy by GO-ZVI-GOx
The in vitro synergistic chemodynamic/photothermal therapy of GO-ZVI-GOx nanocomposites is assessed using the HeLa human cell line, as illustrated in Fig. 7a.
On one hand, Fenton-based chemodynamic therapy is initiated by decomposing the H 2 O 2 into $OH radicals once the GO-ZVI-GOx enters the mildly acidic microenvironment of the cancer cells. The H 2 O 2 can also be additionally provided by GO-ZVI-GOx nanocomposites via the catalysis of glucose into H 2 O 2 . On the other hand, upon the illumination by the 808 NIR laser light, the high photo-thermal conversion efficiency of GO-ZVI-GOx enables the local hyperpyrexia and further increases the amount of cytotoxic radicals. Notably, it is of signicance to evaluate the cytotoxicity of GO-ZVI-GOx nanocomposites prior to performing the in vitro treatment. In this study, the standard CCK-8 assay was conducted to assess cell viabilities. Fig. 7b presents the viability of HeLa cells co-incubated with different concentrations of GO-ZVI-GOx (12.5, 25, 50 100 and 200 mg mL À1 ) at two pH values (7.4 and 6.5). Apparently, the cytotoxicity of GO-ZVI-GOx is rather low at concentrations below 100 mg mL À1 , regardless of pH values. Over 90% cell viability was achieved with GO-ZVI-GOx concentration up to 200 mg mL À1 , indicating low cytotoxicity of the GO-ZVI-GOx to HeLa cells even at high doses. Following the cytotoxicity test, the chemodynamic/photothermal therapeutic effect of GO-ZVI-GOx in the presence of glucose is evaluated at the cellular level. Fig. 7c shows the viability results of HeLa cells incubated with GO and GO-ZVI-GOx under different treatments. As shown in Fig. 7c, no evident cell damages are observed in the control groups including, H 2 O, PBS, PBS + 808 nm and GO alone without light illumination. However, when the HeLa cells are treated with GO solution (200 mg mL À1 ) and 808 nm light irradiation, the cell viability is reduced to 50%, indicating the photothermal (PTT) effect of GO under 808 nm light irradiation. Similarly, when the HeLa cells are incubated with GO-ZVI-GOx (200 mg mL À1 ) in the presence of glucose, the cell viability is about 50% at pH 7.4 and 30% at pH 6.5, which explicitly shows the chemodynamic (CDT) effect of GO-ZVI-GOx. In a comparison with the cell viability in Fig. 7b, it indicates the important roles that acid condition and glucose play in the Fenton reaction by GO-ZVI-GOx. Since Fenton reaction is pH dependent, lower pH promotes the reaction and thus generates more cytotoxic $ radicals. Meanwhile, the added glucose can be catalyzed by GO-ZVI-GOx into H 2 O 2 which is one of the important prerequisites for the Fenton reaction. The additional exposure of 808 nm laser light leads to further drop of the cell viability to below 5%, in the presence of GO-ZVI-GOx (200 mg mL À1 ), demonstrating its high inhibition efficiency for the HeLa cells. This can be explained by the fact that the laser exposure increases the temperature and further boosts the generation of $OH radicals, 21 leading to the substantial decrease in the cell viability.
In order to visualize the in vitro production of ROS signals by the GO-ZVI-GOx, the 2 0 ,7 0 -dichloruorescein-diacetate (DCFH-DA) probe was used to detect the intracellular ROS as it could be oxidized to DCF and released the strong green emissions upon 488 nm excitation. 38,39 As shown in Fig. 8, for the group treated with PBS, PBS + 808 nm and GO, the green uorescence signals is rather weak. However, for HeLa cells treated with GO and 808 nm light illumination (GO + 808 nm), the green uorescence is identiable, which is mainly attributed to the good photothermal effect of GO. For the group treated with GO-ZVI-GOx nanocomposites in the presence of glucose at pH of 7.4, adequate uorescence signals are observed, while more ROS signals can be identied at a lower pH of 6.5. Besides, much stronger uorescence signals are generated when the cells are additionally exposed to irradiation with the 808 nm light. Particularly, the strongest uorescence signals can be observed for the cells treated with GO-ZVI-GOx nanocomposites under 808 nm light illuminations at pH of 6.5 (i.e., GO-ZVI-GOx + glucose (pH 6.5) + 808 nm light). These results are consistent with the signicant decrease in the cell viability under light exposure (Fig. 7c), conrming the increased generation of ROS through the synergistic effect of dual chemodynamic and photothermal therapy based on GO-ZVI-GOx nanocomposites.
Additionally, we have also used Calcein-AM (green uorescence) and propidium iodide (PI, red uorescence) to stain the live and dead HeLa cells, respectively. As shown in Fig. 9, in the absence of GO, the 808 nm light irradiation does not cause any evident damage to cells. Similarly, HeLa cells only treated with GO solution (200 mg mL À1 ) exhibit strong green uorescence, indicating the GO has no negative inuence on cell viability. However, when the HeLa cells are treated with GO solution (200 mg mL À1 ) and 808 nm light irradiation, red uorescence signals (dead cells) appear, which is due to the photothermal effect of GO under 808 nm light irradiation. Besides, for the HeLa cells only exposed to GO-ZVI-GOx solution (200 mg mL À1 ), dead cells can already be observed. Particularly, higher number of HeLa cells are killed at pH 6.5 than 7.4, suggesting that more hydroxyl radicals are produced in the more acidic environment. These results are also consistent with the in vitro results of CCK-8 (Fig. 7c) and ROS detection (Fig. 8). Most importantly, when the cells were treated with GO-ZVI-GOx solution (200 mg mL À1 ) and 808 nm light irradiation, nearly all the cells exhibited strong red uorescence, indicating the synergistic photothermal/chemodynamic therapy of as-prepared GO-ZVI-GOx nanocomposites at 808 nm light irradiations.

In vivo treatment
The small mouse model was established by intratumorally injecting HeLa cells into mice to verify the therapeutic effect of GO-ZVI-GOx in vivo. Firstly, in order to investigate the photothermal effect of the as-prepared materials in vivo, the temperature change of tumor sites in three groups of tumor-bearing mice was recorded by infrared thermal imager under different irradiation time of 808 nm laser (1.5 W cm À2 , 5 minutes). As shown in Fig. 10, when only PBS solution was injected into the tumor, the temperature of the tumor site did not signicantly increase. However, when the solution of GO or GO-ZVI-GOx sample was injected, the temperature of the tumor site evidently increased over irradiation time. Specically, aer 5 minutes of irradiation under 808 nm light, the temperature could reach over 50 C and 45 C, respectively. Suggesting a good photothermal conversion capability of as-prepared GO-ZVI-GOx samples.
Next, the therapeutic effect of GO-ZVI-GOx samples are systematically investigated. When the tumor reached about 100 mm 3 , tumor-bearing mice were randomly divided into six groups,: (1) only injection of PBS, (2) injection of PBS and irradiated with 808 nm laser light, (3) only injection of GO, (4) injection of GO and irradiated with 808 nm laser light, (5) only injection of GO-ZVI-GOx, (6) injection of GO-ZVI-GOx and irradiated with 808 nm laser light. Notably, it would be be ideal to detect the production of $OH radicals during the treatment, which, however, still requires reliable and artefact-free methods to quantify the amount of $OH radicals in vivo. 40 Therefore, only the body weight and tumor volume of the mice were measured every two days. Fig. 11a and Fig. 11b shows the trend of changes in body weight and tumor volume, respectively. It can be obtained that the body weight of the tumor-bearing mice in all groups increased slightly compared with the initial body weight, indicating that the injection of GO and GO-ZVI-GOx had neglectable effects on the health of the tumor-bearing mice. However, regarding the change of tumor volume, for the group treated with PBS alone (PBS), the tumor size was about 4.5 times larger than the initial value. Similarly, for the group treated with PBS and 808 nm irradiation (PBS + 808 nm), the tumor size also grew rapidly, indicating that 808 nm irradiation alone did not have effective prohibition ability on tumor growth. For the mice Fig. 9 Confocal imaging of living and dead cells stained by Calcein-AM/PI probes after different treatments, including PBS, PBS + 808 nm, GO, GO + 808 nm, GO-ZVI-GOx + glucose (pH 7.4), GO-ZVI-GOx + glucose (pH 6.5), GO-ZVI-GOx + glucose (pH 7.4) + 808 nm light, and GO-ZVI-GOx + glucose (pH 6.5) + 808 nm light.   11 The trend of changes in body weight (a) and tumor volume (b) for treated mice over the 14 day treatment (n ¼ 5, mean AE s.d., ***P < 0.001). treated with GO + 808 nm light and GO-ZVI-GOx, the size of tumor aer treatment was slightly reduced. However, for the group treated with GO-ZVI-GOx under 808 nm light irradiation, the tumor size was signicantly smaller and the tumor size was about only half of the initial size aer 14 days of treatment. In addition, Fig. S6 † shows the photographs of the excised tumors from mice, which also indicates that the treatment with GO-ZVI-GOx under 808 nm irradiation substantially prohibited the tumor growth with the smallest tumor size. All these results suggest that GO-ZVI-GOx can effectively inhibit the tumor growth under 808 nm light, due to the synergistic therapeutic effects of chemodynamic and photothermal therapy.
In order to further conrm the therapeutic effect, histological analysis was carried out on tumor tissus (Fig. 12a). Compared with control groups (i.e., PBS, PBS + 808 nm, GO), the groups with treatment (i.e., GO + 808 nm, GO-ZVI-GOx and GO-ZVI-GOx + 808 nm) showed evident tumor tissue damages. Particularly, the group treated with GO-ZVI-GOx + 808 nm showed most severe apoptosis and cell shrinking. Fig. 12b presents the H&E staining of several major organs (heart, liver, spleen, lung, and kidney), which all maintain normal physiological states without obvious damages, indicating that GO-ZVI-GOx has neglectable short-term toxicity in mice.

Conclusions
In this study, a novel H 2 O 2 self-providing dual chemodynamic/ photothermal therapeutic nanoplatform was reported, in which zero valence iron nanoparticles (ZVI) and glucose oxidase (GOx) were deposited on graphene oxide (GO). On one hand, GO functions as a template to anchor ZVI nanoparticles for chemodynamic therapy and GOx for self-generating H 2 O 2 via the catalysis of glucose to further enhance chemodynamic therapy. On the other hand, GO and ZVI could also act as efficient photothermal conversion agents for photothermal therapy under 808 nm laser light. Due to such synergistic effect of chemodynamic and photothermal therapies, a substantial amount of $OH signals can be generated upon 808 nm laser light irradiation, which eventually resulted in the signicant decrease in viability of cancer cells. It is believed that the strategically designed synergistic nanoplatform based on ZVI and GO could be potentially utilized for cancer therapy in the future.

Conflicts of interest
The authors declare no competing nancial interest.