Amino acid modified [70] fullerene derivatives with high radical scavenging activity as promising bodyguards for chemotherapy protection

Despite the great efforts for tumor therapy in the last decades, currently chemotherapy induced toxicity remains a formidable problem for cancer patients, and it usually prohibits the cancer therapy from successful completion due to severe side effects. In general, the main side effects of chemotherapeutic agents are from the as-produced reactive oxygen species (ROS) that not only harm the tumor cells but also damage the patients’ organs. Here we report the application of amino acid derivatives of fullerene (AADF) in the chemotherapy which strongly scavenge the excess ROS to protect the tested mice against the chemotherapy-induced hepatotoxicity and cardiotoxicity. Two amino acids, i.e., L-lysine and β-alanine were separately employed to chemically modify C70 fullerene, and L-lysine derivative of fullerene (C70-Lys) exhibits superior radical scavenging activity to β-alanine derivative of C70 (C70-Ala). As expected, C70-Lys show much better protective effect than C70-Ala against the chemotherapy injuries in vivo, which is verified by various histopathological, haematological examinations and antioxidative enzyme studies. Moreover, the L-glutathione level is increased and the cytochrome P-450 2E1 expression is inhibited. They are potentially developed as promising bodyguards for chemotherapy protection.

Characterization of water-soluble AADF. Solid AADF were obtained after freeze drying, then characterized by elemental analysis (EA), thermogravimetric analysis (TGA) (Shimadzu DTG-60H, Japan), dynamic light scattering (DLS) and zeta-potential measurements. TGA was conducted under a nitrogen flow (50 mL/min) and the samples were heated from room temperature to 500 °C at a ramp rate of 5 °C/min. DLS and zeta-potential measurements were performed on a NanoZS ZEN3600, Malvern instrument at 25 °C. All of the samples were measured three times in Milli-Q water at pH 7.0 after filtering through a 0.22 μm pore size membrane.
Cryo-TEM characterization. Samples for cryo-TEM were prepared with a Leica EM GP immersion freezer (Leica Microsystems) as follows: 3 μL of each sample was loaded onto a glow discharged carbon-coated holey TEM gird (GIG) and blotted with a filter paper to obtain a thin liquid film on the grid. Then the grid was quickly plunged into liquid ethane (cooled by liquid nitrogen). The vitrified samples were then transferred to a cryogenic sample holder (626, Gatan) and examined with a TEM (JEM-2010, JEOL) at 94 K. Micrographs were captured with a CCD camera (Gantega, Olympus Soft Imaging Solutions).
Cell viability assessment. Cell viability was evaluated by the activity of mitochondrial dehydrogenase according to a protocol previously described 20 . In brief, to detect the pre-protection of AADF against DOX-induced oxidative stress, the HUVECs were cultured with different concentrations (100, 300, 500, 800, 1000 ug/mL) of AADF in 96-well plates for 3 h separately, and then treated with 4 μg/mL DOX for 1 h. Between DOX and AADF, the cells were washed with phosphate buffer saline (PBS) to remove the residue. After that, the medium was replaced with fresh medium and incubated in the dark at 37 °C for 24 h. A WST-8 assay with a Cell Counting Kit-8 (CCK-8; DOJINDO, Kumamoto, Japan) was used to evaluate the cell viability. The absorption value at 450 nm was read using a 96-well plate reader (iMark microplate reader, Bio-RAD, USA). Each treatment was repeated three times. For the post-repair of AADF evaluation, the HUVECs were cultured with DOX for 1 h first and followed by the incubation with AADF for 3 h. Other processing kept consistent with the pre-protection action. Experiments were performed at least three times.
Scavenging of hydroxyl radicals. Interception of hydroxyl radicals by AADF was determined by electron paramagnetic resonance (EPR) (JEOL, JEF FA200, Japan). Hydroxyl radicals were generated by H 2 O 2 with the irradiation of ultraviolet light (UV), and DMPO, as a radical capturer, could immediately transfer them to DMPO-OH which would be detected by EPR. A solution containing 100 μL of DMPO (100 mM) and 50 μL of H 2 O 2 (1 M) was mixed with 50 μL of ultrapure water, C 70 -Lys (100 μg/mL), and C 70 -Ala (100 μg/mL), separately. The solutions were firstly irradiated with a 500 W UV lamp for 4 min, and then recorded the X band EPR spectrum in the dark. The scavenging effect was judged by comparison with the control group lacking AADF.

Animals and treatments.
Six-week-old female BALB/c mice with body weights in range of 18.0 g to 20.0 g were purchased from Center for Experimental Animals, Institute of Process Engineering, Chinese Academy of Sciences (Beijing, China). All animal experiments complied with the protocols approved by the National Regulation of China for Care and Use of Laboratory Animals. Mice were randomly distributed into five groups (n = 6) as follows: 1) control healthy group treated with saline only every day; 2) DOX + saline group; 3) DOX + VC group; 4) DOX + C 70 -Lys group; 5) DOX + C 70 -Ala group. The latter four groups were treated with intraperitoneal (i.p.) injection of saline, VC (10 mg/kg/day) and AADF (10 mg/kg/day) once a day for eleven days, respectively. The choice for effective dosage was investigated by detecting a short-term change of haematological parameters and organ coefficients. And at the 4th day all the mice (except control healthy group) received a single intravenous injection (i.v.) of DOX 30 minutes after the injection of saline, VC or AADF. DOX was injected at a dose of 20 mg/kg, which would induce cardiotoxicity and hepatotoxicity in mice.
Coefficients of liver and heart. The coefficients of liver and heart to body weight were measured by weighing the organs and calculating the ratio of organ tissues (wet weight, mg) to body weight (g).
Hematology examination. By quickly removing the eyeball on the eleventh day, the blood samples were divided into two parts, one for blood routine index examination and the other centrifuged for biochemical analysis. Some blood routine indexes were measured by a hematology analyzer (BC-5000VET, Shenzhen, China), including WBC, RBC, HGB, HCT, PLT, neutrophil and lymphocyte. The second portion was used to evaluate the toxicity towards liver and heart by measuring serum levels of ALT, AST, LDH, CK and α-HBDH, using an automated biochemical analyzer (TBA-2000FR, Tochigi, Japan).
Biomarker assay of liver and heart tissues. Each liver and heart was quickly removed from the sacrificed mice and placed into liquid nitrogen for the analysis of enzymatic activity. They were minced and homogenized for the measurements of MDA and the assays of total protein concentration, LDH, GSH, GSSG, GSH-Px, GR, CAT and SOD. The levels of LDH, GSH and GSSG were measured by an automatic microplate reader (MULTISKAN MK3, Thermo, USA). And the spectrophotometry was used to determine the activities of GSH-Px, GR, CAT and SOD (UV-2000, Jinghua, China).
Histopathological examination. Immediately after surgical removal, the liver and heart tissues were harvest and fixed in a 4% paraformaldehyde solution. Hematoxylin and eosin staining was conducted based on the standard manual. In brief, the tissues were immersed in paraffin blocks, sliced into 5 μm sections, unfolded on the histological slides for staining. Histological observations and photomicrography of normal tissue slides were performed using an optical microscope and the sections of liver/heart were scanned by a scanner (Magscanner, KF-PRO-005, Japan).
Western blotting. Western blotting was performed to determine the protein expression of CYP2E1. A 200 μg portion of each precleared, detergent-solubilized liver and heart tissues was used for total protein preparation. Bicinchoninic Acid (BCA) protein assay (Thermo Fisher Scientific, USA) was used to determine protein concentration. All operations were carried out in accordance with the manual specification.
Statistical Analysis. All data presented are means or means ± standard deviations. According to a Student's t-test the variance in different groups was statistically analyzed to evaluate its significance and P < 0.05 were recognized as statistically significant (*) and P < 0.01 were greatly statistically significant (**).

Results and Discussion
Preparation and Characterization of AADF. Both of C 70 -Lys and C 70 -Ala were simply prepared by a facile one-step synthetic method with the addition of alkali and ethanol under heating condition (Scheme in Fig. 1a). With the higher number of amino group in L-lysine than that of β-alanine, L-lysine is susceptible to react with the carbon cage and needs less reaction time for the synthesis of C 70 -Lys. The obtained aqueous layers after reaction were concentrated and further purified by dialysis. It was well known that the conventional characterization methods for small molecules were not suitable for our fullerene derivatives. Thus a previously established method 25,26 , the elemental analysis along with a study for hydration water by TGA, was conducted to estimate their average structures (Table S1 and Fig. S1). The ratio of nitrogen to carbon (N/C) from the elemental analysis was used to assess the average number of amino acid group, approximately with eight L-lysine and six β-alanine molecules severally attached to the fullerenes. According to the TGA measurements, C 70 -Lys and C 70 -Ala showed 9.9% and 12.4% weight loss before 150 °C, respectively, which was the degradation of physically absorbed and hydration water. The subsequent loss of the samples was ascribed to the detachment of the polyol moieties and amino acid groups from fullerenes. Combined with the hydrogen contents of elemental analysis, we got the average numbers of hydroxyl group, nine for C 70 -Lys and fifteen for C 70 -Ala, respectively. Therefore, the average structures of the two samples were deduced as C 70 (OH) x (NH(CH 2 ) 4 CHNH 2 COOH) y (x ≈ 9, y ≈ 8) and C 70 (OH) m (NHCH 2 CH 2 COOH) n (m ≈ 15, n ≈ 6). Cryo-TEM characterization of AADF has indicated their sizes in aqueous solution. Both of them tend to aggregate into nanoparticles with a wide distribution (20-70 nm), and the diameter of C 70 -Lys is larger than C 70 -Ala (Fig. 1b). The DLS investigation shows that the hydrodynamic diameters of AADF in aqueous solution are 155.2 nm (C 70 -Lys) and 139.8 nm (C 70 -Ala) (Fig. 1c), which accords with the result of Cryo-TEM. Both of them revealed a single peak, which was a stable form driven by the aggregation behavior and would be possible for cellular uptake. They revealed high stability in physiological mediums without forming any aggregation even after centrifugation at 8000 rpm for 10 min (Fig. S2). It is worth mentioning that the zeta potential of C 70 -Lys (−37.0 mV) was more positive than that of C 70 -Ala (−45.4 mV), implying successful introduction of L-lysine onto C 70 and exposing more amino groups outside (Fig. 1d). The distinctions in structure and zeta potential between these two samples would greatly influence their physicochemical properties, especially the radical scavenging ability and the behaviors at cellular level as discussed below.
Hydroxyl radical scavenging capability. Hydroxyl radicals ( • OH), the most potent free radical produced from oxygen, are extremely powerful and active oxidants in the presence of biologic substrates, damaging their nuclear and mitochondria DNA, membrane lipids and carbohydrates 27,28 . We chose hydroxyl radical as a representative to evaluate the radical scavenging capability of AADF by the EPR method. The effects of the two kinds of AADF on the EPR spectra of • OH are shown in (Fig. 1e,f, Fig. S3). The intensities of the DMPO-OH adduct were reduced by addition of C 70 -Lys or C 70 -Ala, indicating both of them possessing the radical scavenging capability.
The radical scavenging properties of lysine and alanine were detected by the EPR method, which didn't show any capability in scavenging free radicals (Fig. S4). At the same concentration (100 μg/mL), C 70 -Lys scavenged about 87% of the generated hydroxyl radicals compared with reduction in signal intensity of 49% for C 70 -Ala. We attributed the different efficiency of scavenging abilities to the distinction of the fullerene cage integrity in accord with the structural properties of AADF. A better conjugation provided more reactive sites to attack the ROS, thus enhanced the efficiency of scavenging reactive species.
Protective effects of AADF against DOX-induced oxidative stress using HUVECs. Before the evaluation of the protective effects, we have investigated the cytotoxicity of the samples toward HUVECs by incubation with AADF for 24 h (Fig. S5). Their negligible toxicity is an absolute and obvious prerequisite for their potential use in biomedicine. The protective effects of the two AADF against DOX-induced oxidization were investigated using HUVECs. For the study of pre-protection, the cells were treated with AADF for 3 h before the incubation with DOX for 1 h. While for the post-repair test, we exchanged the order of AADF and DOX treatments. As shown in Fig. 2, the concentration dependent protection and repair were observed and C 70 -Lys exhibited an apparent protective effect compared with C 70 -Ala. As the concentration increasing from 100 to 1000 μg/ mL, the cell viability treated with C 70 -Lys displayed a steady rise, with respect to the mild increase of that treated with C 70 -Ala (pre-protection ratio: C 70 -Lys 21.5% versus C 70 -Ala 2.5%; post-repair ratio: C 70 -Lys 27.9% versus C 70 -Ala 10.4%). AADF also showed their superioty to C 70 suspensions, which may be ascribed to the poor dispersity of C 70 suspensions (Fig. S6). This result demonstrated a correlation between the cellular protective effects and their radical scavenging capability.

Protective effects of AADF against DOX-induced hepatotoxicity and cardiotoxicity in vivo.
The DOX-induced toxicities towards liver and heart have been widely reported 29,30 , thus we choose DOX as a chemotherapeutic agent to investigate in vivo the potential hepatoprotective and cardioprotective effects of AADF during the chemotherapy. Firstly, we set two groups of mice treated with C 70 -Lys or C 70 -Ala for 7 days. We tested the haematological parameters (neutrophil and lymphocyte, the indicator of inflammation) at 3th and 7th day (Table S2), and the histopathological examinations of the main organs (heart, liver, spleen, lung and kidneys) were conducted by H&E staining (Fig. S7). Neither C 70 -Lys nor C 70 -Ala showed any toxicity towards the mice. For system study, thirty BALB/c mice were randomly distributed into five groups (n = 6): control healthy group, DOX + saline group, DOX + VC group, DOX + C 70 -Lys group and DOX + C 70 -Ala group. The control healthy group was treated with saline only and all the other four groups were received intravenously (i.v.) application of DOX (20 mg/kg of body weight) at the fourth day after the pre-treatments of saline, VC, C 70 -Lys or C 70 -Ala for 3 days respectively. Then they continued to be intraperitoneally (i.p.) injected with the drugs in the following six days (scheme Fig. 3a). The optimal dosage of AADF was investigated ahead by detecting a short-term change of haematological parameters (Table. S3) and organ coefficients (Table. S4). We chose the concentration of 1.0 mg/ mL for system studies in vivo. Allowing for high toxicity usually results in weight loss, the body weight of each mouse was checked every other day during the treatments. For the first three days before the injection of DOX, the body weights of all the mice exhibited a rapid growth, and among them the group treated with C 70 -Lys had the fastest growth rate. Besides, the DOX + C 70 -Lys and DOX + C 70 -Ala groups showed a greater resistance to the toxicity caused by the DOX based on the decay rate of weight (Fig. 3b). By comparison, the group treated with VC, slightly resisted the toxicity induced the drop of body weight. The final body weights of the DOX-treated mice were more than 10% less with respect to the control healthy group. And significant different values of the coefficients for liver and heart were found between the AADF-treated groups and saline-treated group (Fig. 3c).
For the AADF-treated groups, the coefficients were obviously lower compared with the DOX + saline group, implying the AADF had a certain improving effect on the hepatomegaly and enlarged hearts caused by DOX. This macroscopic variation has a good correlation with the symptom in DOX-induced liver and heart injury previously reported 10,31 . The histopathological examinations of the liver and heart were studied using hematoxylin&eosin (H&E) stained tissue slices. Compared with the control group, the DOX induced extensive and severe lesions, classified as hepatocellular ballooning degeneration, vacuolization of cytoplasm and single cell necrosis in the liver tissues (Fig. 4a). It also caused the interstitial myocardial edema, disordered arrangement of myocardial fibers with vesicular degeneration and loose cytoplasm of myocardial cell in the heart tissues (Fig. 4b). These histopathological changes qualitatively agreed with the macroscopic variation of the liver and heart coefficients, which implied the swelling of the organs treated with DOX. Although slight vacuolization of cytoplasm was observed in DOX + C 70 -Lys group and DOX + C 70 -Ala group, the degeneration of hepatocytes and myocytes induced by DOX were significantly suppressed. The protection of AADF from histologic observation reflected its advantage over the traditional antioxidant, VC. It's worth mentioning that there were almost no hepatocellular ballooning degeneration and disordered arrangement of myocardial fibers in the mice treated with AADF.
We collected the blood samples to measure haematological parameters and some markers of cellular injury in serum. From the results of white blood cells (WBC), red blood cells (RBC), hemoglobin (HGB), hematocrit (HCT), blood platelet (PLT), neutrophil and lymphocyte, a positive influence of AADF was found on the abnormality of haematological parameters after treatment with DOX (Table 1). And significant variation in almost all parameters was noted for the C 70 -Lys and saline treated DOX groups, especially the neutrophil and lymphocyte which represented the inflammation. For the healthy mice, about 80% of WBC were lymphocytes and 17% were neutrophils. It was obvious that the DOX treatment caused a decrease of lymphocyte and an increase of neutrophils. The neutrophil, a first responder to the xenobiotic, usually raised number as the acute inflammatory processes 31 . The level of producing neutrophils of AADF-treated groups was lower than that of the DOX + saline group, which indicated the AADF had an improvement for the injury induced by DOX. The variation of RBC and WBC in groups treated with AADF also showed the positive effects on chemotherapy-induced myelosuppression. Compared comprehensively, DOX + C 70 -Lys group had a better chemoprotective effect than the DOX + C 70 -Ala group.
It is well known that creatine kinase (CK), alanine aminotransferase (ALT), aspartate aminotransferase (AST), lactate dehydrogenase (LDH) and α-hydroxybutyrate dehydrogenase (α-HBDH) are usually used as serum enzymatic markers to detect hepatic and cardiac injury 31,32 . ALT and AST are liver-specific enzymes and the ratio of AST to ALT is always used to evaluate the liver status, a high value of the ratio typically indicating the hepatocellular damage 33,34 . According to our results, DOX treatment induced a highly significant increase in serum ALT and AST levels in comparison with the control (Fig. 5a,b). Combined with the ratio of AST/ALT, it could be concluded that the DOX injection had induced marked hepatotoxicity. AST is also a heart-sensitive enzyme and its activity is approximately proportional to the extent of heart injury. In addition, the elevated level of CK, LDH or α-HBDH is commonly the index of myocardial infarction, cardiovascular damage, anaemia or hepatitis 35,36 . With the high level of AST, the increase of CK and LDH activity in the serum of mice treated with DOX is expected (Fig. 5c,d), in accordance with the DOX-induced cardiac damage as previously reported 37 . Administration of AADF markedly lowered the serum ALT and AST levels, as well as the CK and LDH levels, which revealed that the AADF not only had the hepatoprotective but also cardioprotective effects. The α-HBDH levels of the five groups were not statistically different when the mice were sacrificed (Fig. 5e). This phenomenon was also reported by Bjelogrlic 38 and Injac 36 , which verified the occurrence of a high α-HBDH level within 48 h after an administration of chemotherapeutic agents. With recovery processing in our experimental procedure, the α-HBDH activity had probably dropped to the relatively lower level after one week of DOX treatment. Comparing the AST/ALT and LDH/α-HBDH ratios in Fig. 5f, we demonstrated the advantage of C 70 -Lys over C 70 -Ala in the protection of liver and heart, even better than the commercial liposomal formulations and earlier published fullerene derivative 22 (Figs S8 and 9).
To further investigate oxidative status in the liver and heart, we tested various parameters of oxidative stress in the liver and heart tissues. Application of DOX significantly raised the activity of all the examined antioxidative enzymes in the liver with regard to the control group (Table 2). Pre-and post-AADF succeeded in preventing  the oxidative stress by meaningfully lowered the glutathione oxidized (GSSG), glutathione peroxidase (GSH-Px), glutathione reductase (GR), catalase (CAT) and superoxide dismutase (SOD) levels in different degrees. The improvement of AADF in DOX-caused peroxidative alterations was also evident by the depletion of GSH and elevation of malondialdehyde (MDA) production. GSH exists in all kinds of organs, which can adjust the synthesis of protein and ribonucleotide and is closely related to the antioxidant capacity of the body. MDA is one of the well-known secondary products of lipid peroxidation, and it is widely used as an indicator of cell membrane injury 39 . Their changes had the same trend with the antioxidative enzymes, which once again confirmed that AADF were potential hepatoprotectors against DOX-induced hepatotoxicity. Moreover, fullerene nanoparticles have been reported to own antineoplastic activities by some possible mechanisms, such as angiogenesis inhibition 40,41 , cancer stem cell-specific inhibition 42 or the regulation of the tumor microenvironment 43 . AADF and DOX may have a synergistic effect to enhance the anti-tumor efficacy, without the impact on the anticancer properties of DOX (Fig. S10).
To better understand the mechanism relevant with the protection of AADF, the CYP2E1 protein expression was measured by western blotting (WB) analysis. CYP2E1, a membrane protein expressed in high levels in the liver, plays an important role in several metabolic reactions, including the deactivation of endogenous or exogenous harmful substances accompanied by the depletion of GSH 23,44 . When the toxic metabolites exceed the eliminating capacity of GSH, they react with the sulphhydryl groups of protein, resulting in the formation of protein adducts, subsequent mitochondrial dysfunction, oxidant stress, peroxynitrite formation and nuclear DNA fragmentation 45 . As shown in Fig. 6a,b, both the hepatic and cardiac GSH levels of DOX + saline group were greatly higher than that of the control, while the groups treated with AADF were improved in different degree. Associated with the CYP2E1 expression (Fig. 6c,d, Fig. S11), it was ascribed to the modulation of AADF on CYP2E1 expression, thus changing the normal metabolic way of DOX. Notably, AADF were markedly inhibited the hepatic CYP2E1 expression, which can be explained in terms of the high levels of AADF accumulation and CYP2E1 expression in liver.

Conclusions
We have investigated the impact of different surface modifications on the physicochemical property of fullerene derivatives, as well as the radical scavenging capability. In vitro study of cytoprotection against the DOX-induced oxidative stress by using HUVECs demonstrated the superiority of C 70 -Lys compared with C 70 -Ala. We further explored the potential protective effects of AADF in mice with DOX-induced hepatotoxicity and cardiotoxicity, and the distinction between C 70 -Lys and C 70 -Ala in radical scavenging was more distinguishable in the in vivo models. These results clearly supported that the chemical functionalization of fullerene plays a significant role in their properties and specific surface modification is great benefit for the biomedical application of these materials.