GQDs are considered as a novel type of QDs with good biocompatibility and provide a bright prospect in the neuroscience application, but the biosafety of them in the CNS has not yet been confirmed. Apart from reported studies showing the blood-brain barrier permeability of GQDs [4], we also observed the fluorescence signal of GQDs in the brain of mice injected with GQDs (Fig. S8). Microglia is an important set of cells to defence foreign chemicals, including carbon-based nanoparticles, in the CNS but to be damaged by them [27]. One study suggested that carbon nanotubes (CNTs) caused cytotoxic effects in microglia-containing cultures instead of neuronal cultures [28].
There have been reported several types of cell death, such as apoptosis, necroptosis and autophagy, are caused by graphene-family nanoparticles, we tried to assess a newly-found PCD called ferroptosis could be induced by GQDs because nanoparticles always cause mixed forms of cell death [13, 29]. Ferroptosis has been identified a form of iron-dependent cell death with lipid peroxidation, and it is mainly occurred in the brain [14, 30]. The microarray data hinted key genes enriched in ferroptosis associated pathways in BV2 cells treated with 100 μg/mL N-GQDs for 24 h.
Firstly, we used two widely reported ferroptosis inhibitors, i.e. a LPO inhibitor Fer-1 and an iron chelating agent DFO, to investigate N-GQDs whether caused ferroptosis in BV2 cells because some researchers have indicated it is necessary to use both lipophilic antioxidants and iron chelators to determine the ferroptosis [23]. The findings showed that pre-treatments of Fer-1 and DFO both effectively prevented decreased cell viability and increased necrotic cells induced by N-GQDs in BV2 cells, which indicated N-GQDs exposure could induce ferroptosis in microglia. It should be noted that N-GQDs exposure simultaneously caused apoptosis in BV2 cells, which was alleviated by Fer-1 but not DFO. The findings indicated the forms of cell death induced by GQDs are mixed and the intracellular oxidative stress could contribute to apoptosis as well, which was confirmed by other antioxidants Trolox and MitoTEMPO.
Meanwhile, the characteristically biological alternations associated with ferroptosis, including iron overload and redox imbalance, were observed in BV2 cells treated with N-GQDs as well. Apart from iron overburden evidenced by a specific fluorescent probe, the disorder of redox homeostasis caused by N-GQDs were indicated by multiple indexes, including GSH depletion, decreased NADPH activity, excessive ROS production and LPO. Similar to other types of graphene-based nanoparticles or QDs, N-GQDs also has been found to be strongly capable of perturbing the redox-sensitive system by inhibiting specific antioxidant enzyme activities in model animals [9, 31].
Additionally, the protection from Fer-1 and DFO against N-GQDs-induced iron accumulation and redox disequilibrium not only authenticated the ferroptosis in BV2 cells caused by N-GQDs but also indicated the association between iron level and redox balance. Pre-treatment of a LPO inhibitor Fer-1 could effectively reverse the iron overload in response to N-GQDs in BV2 cells, because some studies indicated that the LPO is directly attributed to the iron overburden [23, 24]. Furthermore, when BV2 cells pre-treated with an iron chelator DFO, the GSH depletion, increased MDA level and LPO in responses to N-GQDs were also effectively alleviated, which suggested that cytosolic iron burden could contribute to persistent oxidation. Moreover, excessive ROS production caused by N-GQDs inhibited by Fer-1 and DFO could be explained by the facts of iron-mediated ROS generation and LPO via Fenton reaction [32].
After confirming the alternation of gene expression associated with ferroptosis in BV2 cells treated with N-GQDs based on the microarray data, four widely-reported biomarkers of ferroptosis were also used to assess the potential mechanisms of N-GQDs inducing ferroptosis. SLC7A11 is a key component of system X−c that is responsible for maintaining redox homeostasis by participating in synthesizing GSH and one of earliest found key regulators in ferroptosis [24, 33]. GPx4 is at the downstream of SLC7A11 and serves a key factor in reducing LPO by converting reduced major antioxidant GSH to GSSG, which is also a confirmed biomarker of ferroptosis [34, 35]. Therefore, the inhibition of expressions of SLC711 and GPx4 could be one reason of GSH depletion in BV2 cells exposed to N-GQDs.
Acyl-CoA synthetase long-chain family member 4 (ACSL4) is a key enzyme of regulating lipid composition, which has been confirmed to contribute to ferroptosis execution [36, 37]. Cyclooxygenase2 (COX2), also known as prostaglandin-endoperoxide synthase (PTGS), has been reported as a monitor of ferroptosis [38]. In this study, the expressions of ACSL4 and COX2 both up-regulated in BV2 cells treated with N-GQDs. In summary, these four proteins were all indexes for redox equilibrium and the alternation of their expression pattern suggested that N-GQDs might instigate ferroptosis in BV2 cells through overwhelming anti-oxidative system that has been reported a key modulator of LPO and resulting in this novel form of cell death [39].
Mitochondria is the major organelle in regulation of iron metabolism and fatty-acid metabolism [40]. Researchers have found some dramatic morphological changes in mitochondria distinction from other forms of PCD, such as mitochondrial shrinkage, in ferroptotic cells [14, 41], which was also observed in mitochondria of BV2 cells treated with N-GQDs. Therefore, the potential involvement of mitochondria in ferroptosis is highly probability, but whether mitochondria play a central role in N-GQDs-induced ferroptosis in microglia remains unclear [26]. In this study, we found that the internalized N-GQDs in BV2 cells was observed in mitochondria, following with obviously mitochondrial impairments, such as broken ridge and collapsed membrane. Meanwhile, the enhancements in iron level, ROS production and oxidative lipid in mitochondria of BV2 cells exposed to N-GQDs were all highlighted by mitochondria targeted indicators. Furthermore, the dissipation of the MMP indicating mitochondrial dysfunction could be associated with the direct damage from N-GQDs or the excessive ROS production in mitochondria [25].
There have been studies suggesting that mitochondrial iron level increased in cells treated with several ferroptosis inducers, such as eratin, doxorubicin and RSL3 [25, 42, 43]. Taken Fenton reaction into consideration, increased mitochondrial iron inevitably leaded to the mitochondrial ROS accumulation, which has been reported to play momentous role in regulating various forms of PCD [44, 45]. ROS could be able to induce LPO by reacting with the polyunsaturated fatty acids of lipid membranes, because mitochondrial fatty acid metabolism provides the specific lipid precursor required for lipid oxidation [26]. Taken all together, the findings suggested that the increased mitochondrial ROS generation caused by N-GQDs might act as an important contributing factor to ferroptosis in BV2 cells.
The intracellular excessive ROS generation, an indicator of oxidative stress, has been reported being essential for the occurrence of ferroptosis [46]. In order to assess the causal relationship of mitochondrial ROS and ferroptosis, BV2 cells were treated with a total antioxidant Trolox and a mitochondrial targeted ROS scavenger MitoTEMPO before they were exposed to N-GQDs. When both Trolox and MitoTEMPO blocked cytosolic and mitochondrial ROS production caused by N-GQDs, they also rescued cell viability in BV2 cells. Moreover, the iron overload and LPO in mitochondria as well as GSH depletion, LPO and alternation of expression pattern of ferroptosis biomarkers induced by N-GQDs were all impeded by Trolox and MitoTEMPO, and the protective capacity of Trolox and MitoTEMPO presented no obvious difference.
Herein, the increased mitochondrial antioxidant activity benefiting from MitoTEMPO effectively blocked ferroptosis induced by N-GQDs in microglia, which indicated that mitochondrial oxidative stress could be a critical step in the N-GQDs-caused execution of ferroptotic cell death. Similar to two ferroptosis inhibitors, i.e. Fer-1 and DFO, a total and a mitochondrial targeted ROS scavenger confer the protective effects against N-GQDs-induced iron overload and redox imbalance not only in the whole BV2 cells but also in the mitochondria. Since only increasing the mitochondrial antioxidant capacity was capable of effectively alleviating the ferroptosis caused by N-GQDs in BV2 cells, the underlying mode of action to explain the N-GQDs-induced ferroptosis could be through the mitochondrial oxidative stress.
As we known, Chemical modification with amino group is a way to improve the biocompatibility of grapheme-based nanoparticles [47], but it is still not certain this better biocompatibility results in lower toxicity. In this study, since the PLQYs of A-GQDs, a kind of amino-functionalized GQDs and N-GQDs are similar, it is meaningful to evaluate the ferroptotic effects of them at the same administration concentration and time. Even though A-GQDs exposure at a high dose was capable of inducing ferroptosis in BV2 cells, the A-GQDs-induced enhancements in ferroptotic cell, cytosolic iron level and LPO were weaker than that induced by N-GQDs.
Based on the results of milder generations of ROS and LPO as well as MMP hyperpolarization in mitochondria of BV2 cells treated with A-GQDs than N-GQDs, it seems that the amino functionalized group might alleviate mitochondrial oxidative stress and dysfunction caused by GQDs and result in mild ferroptosis. According to available studies, amino group functionalization is capable of promoting the linkage between graphene materials and organic molecules, which improve graphene-based nanoparticles more suitable for biomedical applications [47,48,49]. Some researchers also found that amino group could protect cells from oxidative stress [50]. Even though our findings showed that A-GQDs presented lower toxicity associated with ferroptotic damages and mitochondrial oxidative stress in microglia than N-GQDs, the slight toxicity of A-GQDs still could limit their application in the field of biomedicine. Additionally, N-GQDs seem to be superior on the fluorescent stability because it is recently reported that high concentrations of amino groups tended to exhibit shorter lifetimes [6].
The surface chemical modifications could be associated with other physical and chemical properties, such as surface charge and hydrophobic or hydrophilic nanomaterials, to influencing the physiology of cells. For example, the cellular uptake of positively charged nanoparticles results in higher uptake rates and efficiency in various cell types [51]. Otherwise, whether nanoparticles are hydrophobic or hydrophilic that is mainly determined by their surface ligands are critical to the bio-availability and cytotoxicity of nanoparticles [52]. Although there is not enough data to draw a conclusion in the effects of certain physicochemical parameters of GQDs on the physiological interactions of cells, the researching experience and findings could provide valuable references for further study.
However, there are still some unsolved problems about the molecular pathways of N-GQDs inducing mitochondrial oxidative stress and the specific targets of N-GQDs causing ferroptosis, which should be addressed and clarified before allowing them in biomedical applications. Even though the induction of ferroptosis by GQDs could be expected in cancer therapy via inhibiting cancer cell growth like other nanoparticles [53], it is still a probably serious consequence in response to GQDs in normal tissues, especially the brain that is vulnerable to LPO. The findings advance our understanding of cytotoxicity of GQDs in microglia and provide valuable information for risk assessment of this unique nanoparticles.