Characterization of GS-CDs
A series of GS-CDs at 170 °C using GS as the single reactant at different reaction times was synthesized, from 1 to 10 h, via normal hydrothermal synthesis. High-resolution transmission electron microscopy (HRTEM) was used to characterize the size and surface morphologies of GS-CDs@3 h, GS-CDs@5 h, GS-CDs@6 h, and GS-CDs@10 h at different hydrothermal reaction times of 3, 5, 6, and 10 h, respectively. All of the GS-CDs showed good dispersity with even sizes, where the diameter of GS-CDs@3 h was concentrated at 2.95 ± 0.72 nm, the diameter of GS-CDs@5 h was concentrated at 2.75 ± 0.66 nm, the diameter of GS-CDs@6 h was concentrated at 3.00 ± 0.64 nm, and the diameter of GS-CDs@10 h was concentrated at 3.24 ± 0.62 nm (Fig. 1a). The HRTEM images shown in the upper right corner of Fig. 1a presented the crystal lattices of the four prepared GS-CDs. GS-CDs@3 h had a mere crystal lattice, possibly due to the short reaction time and incomplete nanostructure. GS-CDs@5 h had a certain lattice structure with a lattice spacing of 0.219 nm. GS-CDs@6 h and GS-CDs@10 h showed obvious lattices, with lattice spacings of 0.218 and 0.207 nm, respectively. This implied that with prolonged reaction time, the self-assembly of the GS molecules became more regular, and the structures of the GS-CDs became more complete. In UV absorption spectra, the GS aqueous solution had weak UV absorption, while the GS-CDs showed strong absorption at a wavelength of 280 nm (Fig. 1b and Additional file 1: Figure S1-1). The longer the reaction time, the higher the UV absorption peak. Combined with IR spectroscopy analysis, the absorption peak at 280 nm was mainly derived from the vibration and rotation of C = O (Additional file 1: Figure S1-2). The inset (Fig. 1b) is the photos of the four prepared GS-CDs under sunlight (left) and UV light at 365 nm (right), respectively. The GS-CDs with different reaction times had strong blue-green fluorescence, and the brightness increased slightly with increased reaction time.
In the contour map of the three-dimensional fluorescence spectra of the four prepared GS-CDs with different reaction times, there was only one substance with fluorescent emission in each system, indicating a single product (Fig. 1c). As shown in the isometric projection of the three-dimensional fluorescence spectra (Fig. 1d), all the four GS-CDs had excitation dependence. Under the optimal excitation wavelength of ~ 360 nm, the optimal emission wavelength was ~ 440 nm for the four GS-CDs. The longer the reaction time, the higher the obtained fluorescence intensity of the GS-CDs. In the fluorescence attenuation curves of the four GS-CDs (Additional file 1: Figure S1-3), three fluorescence lifetimes (τ) were obtained, after fitted according to a third-order decay exponential function. It indicated that the CDs had three fluorescence centers. The quantum yields of GS-CDs@3 h, GS-CDs@5 h, GS-CDs@6 h, and GS-CDs@10 h were ~ 0.19%, ~ 0.14%, ~ 0.19%, and ~ 0.22%, respectively. X-ray photoelectron spectroscopy (XPS) was used to further characterize and analyze the elemental compositions and functional groups of the four GS-CDs. In the XPS spectra, all the four GS-CDs had two main peaks at 285.1 and 531.6 eV, corresponding to the C 1s and O 1s elements, respectively (Additional file 1: Figure S1-4). The C 1s band had three peaks near 288.5, 285.6, and 284.5 eV, which was attributed to the carbon peaks related to C = O, C-O, and C-C. The O 1s band had two peaks near 534.3 and 532.6 eV, which were attributed to the oxygen peaks of C = O and C-O.
Studies have shown that the content in typical conjugated fluorophore groups, such as carbon-carbon double bonds, carbon-carbon triple bonds, or carbon-based π electrons in GS molecules is very low . The main sources of UV absorption and fluorescence emission of the prepared GS-CDs system did not consist of these classic conjugated structures. The appearance and increase in UV and fluorescence could be attributed to the cross-linking enhanced emission (CEE) effect . Under a hydrothermal condition of 170 °C, and with rapid collisions between the particles, the free GS were cross-linked through a buckling reaction, forming a carbon nucleus. During the reaction, the main tetracyclic triterpene structure of GS molecule showed few changes, according to the mass spectra analysis of GS, GS-CDs@3 h, GS-CDs@5 h, GS-CDs@6 h, and GS-CDs@10 h (Additional file 1: Figure S1-5 and Table S1-1). The active site beside the main structure reacted with H2O, and the number of C = O groups could increase in the system (Additional file 1: Figure S1-2). With prolonged reaction time (3, 5, 6, and 10 h), the buckle reaction, as well as the degree of carbonization increased. The structures of the CDs became denser and more orderly (Fig. 1a), and more C = O and C-O groups in the system became fixed. Their vibrations and rotations were limited, resulting in increased radiation transition. Thus, the UV absorption and fluorescence emissions of the GS-CDs generated by non-classical conjugation were enhanced [38, 39].
Inhibitory effect of GS-CDs on cells
The above results demonstrated that under hydrothermal conditions, and within a certain reaction time, the GS molecules could self-assemble to form CDs. When reaction temperature (170 °C) was far below the threshold temperature for the ring-opening reaction, the main tetracyclic triterpene structure of GS could not be destroyed during the self-assembly process. Thus, we proposed the self-assembled GS-CDs could retain biomedical activity of GS to a certain extent. On this basis, cell counting kit-8 (CCK-8) method was used to estimate cell viabilities of human neuroblastoma cells (SH-SY5Y), human cervical cancer (HeLa), mouse microglial (BV2), rat adrenal pheochromocytoma (PC12), and human hepatoma (HepG2) cells. They were incubated with different GS-CDs (GS-CDs@1 h, GS-CDs@2 h, GS-CDs@3 h, GS-CDs@4 h, GS-CDs@5 h, GS-CDs@6 h, GS-CDs@7 h, GS-CDs@8 h, GS-CDs@9 h, GS-CDs@10 h) for 48 h. The cell viabilities of HeLa, BV2, PC12, and HepG2 cells did not change significantly after incubation with GS-CDs under the same conditions (Additional file 1: Figure S2-1, S2-2, S2-3, and S2-4).
Note that GS-CDs had few inhibitory effects on human neuroblastoma cells (SH-SY5Y), when reaction time was 1–4 h (Fig. 2a). At a GS-CDs dosage of 50 µg·mL− 1, some cytostatic effects appeared. When reaction time increased to 5 h, GS-CDs started to display obvious inhibitory effect on SH-SY5Y cells. At a reaction time of 6 h, GS-CDs had a stronger inhibitory effect on SH-SY5Y cells. The cell viability rates were ~ 87.56%, ~ 73.26% and ~ 55.05% at GS-CDs dosing concentrations of 20, 30 and 50 µg·mL− 1, respectively. The higher the dose, the lower the cell viability. This suggested that the toxic effects of GS-CDs on SH-SY5Y cells were concentration-dependent. When reaction time extended to 7–10 h, the inhibitory effects of GS-CDs on SH-SY5Y cells were similar to GS-CDs@6 h. The two other repeated experiments were mostly identical (Additional file 1: Figure S2-5). The experimental results showed the prepared GS-CDs had a very effective inhibitory effect on SH-SY5Y cells. A low concentration of GS-CDs (50 µg·mL− 1), their inhibition rate of SH-SY5Y cells achieved ~ 45.00%. Since reaction temperature (170 °C) was far below the threshold temperature of the ring-opening reaction, the main tetracyclic triterpenoid structure of GS could not be destroyed. Therefore, we proposed that GS-CDs retained the pharmaceutical activity of the carbon-sourced GS, and introduced largest number of active substances into SH-SY5Y cells through endocytosis, with best efficacy. In addition, since the substance of GS-CDs was natural plant essence, they showed good non-cytotoxicity to ordinary cells, such as human renal epithelial (293T) cells and human normal liver (LO2) cells (Fig. 2b).
Considering the obviously higher inhibitory of GS-CDs on SH-SY5Y than other cells (HeLa, BV2, PC12, and HepG2 cells), there were two speculations: (i) There are differences of active receptor in membrane surface among different tumor cells. GS-CDs may bind to certain specific receptors on the surface of SH-SY5Y cells, activating the tumor apoptosis pathway and thus exerting inhibitory effects. (ii) Gene level effects: The occurrence of tumors is the result of multiple genes, multiple steps, and multiple mutations. Mutations in different genes with different intensities lead to the formation of different tumors. The addition of GS-CDs may change the structure of genetic DNA of SH-SY5Y cells, reverse the genetic characteristics of SH-SY5Y cells, and inhibit the tumor effectively.
Different inhibitory effect on SH-SY5Y cells
Furthermore, under the same conditions we compared the inhibition efficiency of GS molecules and CDs with different structure/composition on SH-SY5Y cells, via in vitro cytotoxicity experiments (Fig. 2c and d, and 2e). During the same concentration range, GS had no inhibitory effects on SH-SY5Y cells. It was inferred that GS in molecular state could not be effectively taken up via cell transport and provided effective drug effect. On contrary, due to the nano-size and nanostructure, GS-CDs could be taken up by cells in large numbers through endocytosis. Thus, GS-CDs showed a much higher inhibitory efficiency on SH-SY5Y cells than GS molecules. Second, molecule glucose of little medicinal activity was used to prepare common glucose carbon nanodots (Glu-CDs) via a similar hydrothermal reaction [14, 34, 40, 41]. When concentration of Glu-CDs was 50 µg·mL− 1, viability of SH-SY5Y cells was weakly increased. It indicated common Glu-CDs synthesized by non-drug molecules had little drug activity. Lastly, GS was loaded onto the carrier Glu-CDs through supramolecular forces. Then drug molecule GS composite carbon nanodots (GS@Glu-CDs) were obtained (Additional file 1: Fig. S2-6). At the same concentration of 50 µg·mL− 1, inhibitory effect (~ 75.69%) of GS@Glu-CDs on SH-SY5Y cells became obviously but much less than that of GS-CDs. It can be explained that after loaded on Glu-CDs, drug molecule GS could enter cell interior through endocytosis and had a damaging effect on SH-SY5Y cells. However, because of the nano-size of Glu-CDs, the loaded quantity of bioactive GS was limited, as well as their inhibitory effect on SH-SY5Y cells. Thus, the construction of nanodrug CDs composed of drug activity molecules could open up new research ideas and application directions for nanomedicine.
Intracellular distribution of GS-CDs in SH-SY5Y cells with time
Because GS-CDs prepared for 6–10 h showed similar inhibitory effects on SH-SY5Y cells, GS-CDs@6 h were chosen as the representative nano-drug, and were denoted as GS-CDs in the follow-up experiments. To further explore the inhibitory effect of GS-CDs on SH-SY5Y cells, entry of GS-CDs into cells at different incubation times was monitored by a fluorescent inverted microscope (Fig. 3a). Accompanied by its effective medicinal effects, the fluorescent GS-CDs system did not need to introduce other fluorescent substances for tracking and labeling. With prolonged incubation time, SH-SY5Y cells gradually shrank and became round. The SH-SY5Y cells gradually became apoptotic with increased time by the inhibition of GS-CDs. In the fluorescence field, GS-CDs entered cells and were mainly concentrated in the cytoplasm. Between 30 min and 6 h, fluorescence intensity of the system gradually increased, indicating the internalized number of GS-CDs increased with time. Between 6 and 12 h, fluorescence intensity of the system decreased. This was possibly due to the reactions and structure destruction of GS-CDs within the cells. Fluorescence intensity of the system was very weak after incubating for 24 h. By 48 h, little fluorescence was detected in the system. While GS-CDs played a medicinal role in the cells, they could also be metabolized out through exocytosis. Fluorescence intensity changes in the incubation system indicated that inhibitory effect of GS-CDs on SH-SY5Y cells was time-dependent.
Cytotoxicity of GS-CDs in SH-SY5Y cells with time
The exact viability changes of SH-SY5Y cell after incubating with GS-CDs for different times were investigated (Fig. 3b). When incubated for 6 and 12 h, the GS-CDs had a minimal inhibitory effect on the SH-SY5Y cells. When incubated for 24 h, at higher GS-CDs concentrations (50 µg·mL− 1), the inhibitory effect appeared. After 48 h of incubation, GS-CDs showed a strong inhibitory effect on the SH-SY5Y cells, which was essentially consistent with the results above (Fig. 2a).
Bio-TEM analysis was conducted to compare the cellular uptake and intracellular distribution of GS-CDs in SH-SY5Y, 293T and LO2 cells, respectively (Additional file 1: Figure S3-1). Many GS-CDs entered all the cells and mainly existed in cytoplasm from 6 to 24 h. At 24 h, GS-CDs were still abundant in SH-SY5Y cells. In that case, some cells shrank, some cell membrane was broken, and some cell shape was destroyed. It proved that GS-CDs inhibited SH-SY5Y cells effectively. Within 6-12 h, GS-CDs also entered into 293T and LO2 cells, but little GS-CDs could be found at 24 h. It implied GS-CDs were basically metabolized within 24 h in 293T and LO2 cells. What was more, 293T and LO2 cells had normal morphology and good growth status at 24 h. It suggested that GS-CDs had little toxic effect on normal cells. Via the bio-TEM analysis, it can be inferred GS-CDs with hydrophilic surface functional groups were able to enter all these cells. GS-CDs possessed a high specific inhibitory effect on SH-SY5Y cells and little inhibitory effect on other cells.
Apoptosis and cell cycle of SH-SY5Y cells induced by GS-CDs
Flow cytometry was used to detect apoptosis and cell cycle of SH-SY5Y cells after incubation at different GS-CDs concentrations after 48 h (Fig. 4a). When GS-CDs concentrations was 20, 30, and 50 µg·mL− 1, apoptosis rate of SH-SY5Y cells was ~ 18.59%, ~ 25.06%, and ~ 59.66%, respectively. The higher the administration concentration, the more significant the apoptosis rate. It demonstrated when GS-CDs were administered at high concentrations (for example, 50 µg·mL− 1), and at a sufficient incubation time (48 h), an inhibition effect on SH-SY5Y cells could be exerted.
When concentrations of GS-CDs were 10 and 20 µg·mL− 1, the cell cycle of SH-SY5Y cells changed a little (Fig. 4b). When concentration of GS-CDs was increased to 30 and 50 µg·mL− 1, SH-SY5Y cells in G0/G1 phase decreased to ~ 30.53% and ~ 28.78%, respectively. While in G2/M phase, SH-SY5Y cells increased to ~ 40.42% and ~ 38.10%, respectively. It indicated that GS-CDs could induce G2/M phase arrest of SH-SY5Y cell cycle.
Effect of GS-CDs on the expression of apoptosis-related proteins in SH-SY5Y cells
The above apoptosis and cell cycle of SH-SY5Y cells showed that GS-CDs may inhibit tumor growth by inducing tumor cell apoptosis. Thus, we chose to detect the expression of apoptosis-related proteins in SH-SY5Y cells (Fig. 4c). Bax and Bcl-2 are homologous water-soluble related proteins, which promote apoptosis in cells. Bax can antagonize the protective effect of Bcl-2. Research showed that when the ratio of Bcl-2/Bax decreased, it can increase the permeability of mitochondrial membrane, activate caspase apoptosis pathway and induce cell apoptosis . Among them, caspase-10 is the downstream signal molecule of mitochondria that mediates apoptosis, and caspase-3 is the executor of apoptosis. Cell death is inevitable after caspase-3 is activated . Western-blot results showed that compared with the control group, when the drug concentration was 30 and 50 µg·mL− 1, the protein contents of Bax, caspase-3 and caspase-10 increased significantly (P < 0.05, P < 0.01, P < 0.001). The content of Bcl-2 protein decreased significantly (P < 0.05, P < 0.001). The above results suggested that apoptosis of SH-SY5Y cells induced by GS-CDs may be through inhibiting the up-regulation of Bcl-2, activating the caspase apoptosis pathway, activating and up-regulating the expression of apoptosis-related proteins caspase-3 and caspase-10. Finally, it caused SH-SY5Y cell death.
High inhibitory effect of GS-CDs on neuroblastoma and toxicity evaluation in vivo
To evaluate the inhibited ability of GS-CDs on human neuroblastoma in vivo, we employed BALB/c nude mice bearing human neuroblastoma tumors as the animal model. The antineoplastic cisplatin for the clinical treatment of neuroblastoma was administered to tumor-bearing mice as the positive group. The real-time photos of mice growth in the model, cisplatin, and GS-CDs group had been recorded carefully (Additional file 1: Figure S5-1, S5-2, and S5-3). Using a small animal CT imaging system, CT scans were performed to observe the growth of the mice tumors on the 1st, 5th, 9th, and 13th days during one experiment cycle (Fig. 5a). The statistic relative tumor volume of the model, cisplatin, and GS-CDs group, was calculated by the CT scans of the largest tumor sections (Fig. 5b). With time, the tumor growth trend of model group was obvious, and tumor growth in cisplatin group was slower than in the model group. The tumor growth rate of mice in GS-CDs group was the slowest and the tumor volume was the smallest. At the end of 13th day of experimental period, mice in each group were dissected and tumor comparisons were made (Fig. 5c). The tumor volume and statistic tumor weights of the GS-CDs group were significantly smaller than the other two groups (Fig. 5d). The above data showed that the prepared GS-CDs could effectively inhibit the growth of human neuroblastoma and produce substantial curative effects in vivo.
Meanwhile, statistical diagrams for weight changes of mice in the control, control + GS-CDs, model, cisplatin, and GS-CDs group were also recorded to assess the toxicity of GS-CDs (Fig. 5e). Compared to the weight of mice in control group, weight of mice in control + GS-CDs group was slightly lower at the initial stage. With more time, weight of mice in control + GS-CDs group exceeded that of control group. This proved that the prepared GS-CDs had almost no toxic effects on organism. Compared to the weight of mice in model group, weight of mice in GS-CDs group was very significant. This also showed that GS-CDs did not produce toxic effects during neuroblastoma treatment. Attractively, because it could effectively inhibit tumor growth and improve health of mice, the weight of mice in GS-CDs group was close to the control group on the 13th day. The weight loss of mice in the cisplatin group was the most. The H&E staining of the main organs of mice in each group results showed that few obvious organ abnormalities were observed in the pathological images of mice in cisplatin group (Fig. 6a). Experimental observations found that the mice injected with cisplatin were in a depressed state compared to mice in other groups. They were reluctant to eat for a period of time. This was possibly due to the severe irritation of cisplatin on the intestines and stomachs of the mice, resulting in significant weight loss. The above results indicated that the GS-CDs had high biocompatibility in the treatment process and could be used safely and effectively to treat neuroblastoma. It should be emphasized that human neuroblastoma is mainly harmful to infants and young children with low immunity. For those children who need to be carefully protected during treatment, the side effects caused by the nano-drug GS-CDs based on natural herbal extracts would be greatly reduced. Thus, the acceptance rates for infants and young children would be much higher. It would be more conducive to the treatment of the disease.
To estimate pathological damage of GS-CDs on tumors, the dissected neuroblastoma of mice in the model, cisplatin, GS-CDs group were stained and analyzed (Fig. 6b-d). Because human neuroblastoma tumor cells were malignant and grew quickly, sizes of tumors in the control group were very large at the end of the test period. The provided blood and nutrients were insufficient for the tumor, causing necrosis of the cells inside tumor. There was abundant blood supply outside tumor, and the tumor could maintain rapid growth. As observed in the H&E (hematoxylin-eosin staining) image, tumor cells in each group had a certain degree of damage, including in the model group. Neovascularization is a necessary condition for tumor growth, and antigen CD31 (endothelial cell adhesion molecule) was employed to label the vascular endothelial cells in tumor. The expression of CD31 in GS-CDs group was clearly lower than the other two groups. In addition, antigen neuron-specific enolase (NSE) is a marker of neuroblastoma. Compared to the other groups, expression level of NSE in GS-CDs group was the lowest.
It was worthy to note that since we did not modify the surfaces of GS-CDs with brilliant commercial fluorescent agent or contrast agent, it was hard to study the exact biodistribution of GS-CDs in animals. We have checked the literatures within last ten years to speculate the fates of GS-CDs in animals (Additional file 1: Table S6-1) [16, 44,45,46,47,48].