NOX4 is required for cell survival of starved PTC cells.
Previously we reported that NOX4 as a sensitive protein plays an important role in hypoxic PTC cells [10]. Since serum starvation is also a common feature for solid tumors as similar as hypoxia [19], we then wonder whether NOX4 is responded to serum starvation. The results of immunoblot demonstrated that once the serum concentration in medium was decreased from 10–2.5% or 0.5%, NOX4 expressions in BCPAP cells (harboring BRAF mutant) and TPC-1 cells (harboring RET/PTC1 rearrangement) were significantly increased in a varying degree (Figure 1A), suggesting NOX4 is also sensitive to starvation in PTC cells.
Serum starvation is a widely used condition in cell biology to modulate cell viability and apoptosis [20, 21]. Thus, we investigated the role of NOX4 in cell viability and apoptosis under serum-starved condition. First, BCPAP and TPC-1 cells were infected with lentivirus expressing shRNA against NOX4 (shNOX4) or control shRNA (shCtrl) (Figure S1A). Then a CCK8 assay showed that NOX4 knockdown significantly repress cell viability in starved PTC cells (Figure 1B). Subsequently, cell death assays detected by Cell Death Detection ELISAPLUS kit showed significant increments of the apoptotic levels in NOX4-deficient PTC cells compared with the control cells under serum starvation (Figure 1C). In addition, cell cycle was examined by flow cytometry showing that NOX4 knockdown induced a significantly extended G1 phase in starved PTC cells (Figure 1D). Consistently, the increase of apoptotic sub G1 fraction confirmed our previous findings in cell apoptosis (Figure 1C). Altogether, these results underscore the importance of NOX4 in determining cell survival in starved PTC cells.
NOX4 deficiency modulates the levels of energy-associated metabolites of BCPAP cells.
To get metabolic insights in the role of NOX4 in starved PTC cells, the targeted metabolomics analyses involved in energy metabolism by mass spectrometry were performed upon serum-starvation or not in BCPAP cells. Principal component analysis (PCA) of targeted metabolomic analysis demonstrated a distinct separation between shCtrl and shNOX4 cells under 10% serum and 0.5% serum conditions (Figure S2A), which indicates the critical role of NOX4 in the levels of these metabolites. Under either 10%- or 0.5%- serum condition, the level of each metabolite was inconsistently altered by NOX4 deficiency (Figure S2B). In detail, the lactate level was reduced significantly in NOX4-dificient BCPAP cells under serum-starved condition (Figure S2C), which gives us a clue that NOX4 may involve in glycolysis of starved PTC cells. Furthermore, the ration of ADP/ATP and AMP/ATP, were significantly increased in NOX4-deficient cells under either 10%- or 0.5%- serum condition (Figure S2D), showing that NOX4 contributes to the energy demand of PTC cells. Besides, the ratio of NAD+/NADH was also upregulated by NOX4 deficiency under 0.5% serum condition as well (Figure S2E), reflecting a potentially relevant role of NOX4 in redox regulation. In all, these findings indicate a dispensable role of NOX4 in the maintenance of bioenergetic and redox metabolites in starved PTC cells.
NOX4 depends on ROS to oppose chemotherapeutic drugs-induced apoptosis.
Given that NOX4 functions as an energetic sensor coupling cancer metabolic reprogramming to drug resistance [12], and that serum starvation can render drug resistance of cancer cells [22], we then test the hypothesis that NOX4 contributes to the acquisition of drug resistance in serum-starved PTC cells. After BCPAP cells or TPC-1 cells stably expressing shRNA against NOX4 or control vector were treated with chemotherapeutic drugs (etoposide and doxorubicin (DOX)), we found that the cells exposed to etoposide or DOX reveal a normal enhancement of cell apoptosis compared to the control (buffer alone), whereas cells cultured with 0.5% serum showed a modest increase of the drugs-induced cell death (Figure 2A). Importantly, there was a significant increment of apoptosis in cells harboring NOX4 dificiency compared with previously drugs or serum-starvation-induced apoptosis (Figure 2A), which supports an inhibited role of NOX4 in the chemotherapeutic drugs-elicited cell death in either starved or normal PTC cells.
ROS generated by NOX4 was indispensable for metabolism and secretion of PTC cells [10, 23], we then asked whether the effect of NOX4 on drug-induced cell death is mediated by ROS. BCPAP and TPC-1 cells were treated with NAC (a ROS scavenger) and etoposide respectively or combinedly in completed medium. The results showed that NAC can not elicit additional apoptosis but does enhance etoposide-induced apoptosis (Figure 2B), indicating a protective role of ROS in response to etoposide. In starved PTC cells, the results showed that NAC combining with etoposide under serum-starved condition leads to more rates of apoptosis than etoposide-treated alone, almost equaling to that caused by NOX4 knockdown plus DMSO, whereas NOX4 knockdown plus NAC is disabled to aggravate etoposide-induced apoptosis compared with that caused by shCtrl plus NAC (Figure 2C). This indicates that NAC alone in the presence of etoposide increases apoptosis in starved cells, and that NOX4 knockdown does not change NAC effect. It can be further concluded that cell survival under starvation and etoposide treatment depends on ROS. Altogether these results suggest that the survival of PTC cells treated with etoposide depends on intracellular ROS specifically derived from NOX4.
Furthermore, these findings can be confirmed reversely by glucose oxidase (GOD) (a stable enzyme that oxidizes glucose into glucolactone converting oxygen into ROS) addition, showing a significantly inhibited apoptosis when comparing GOD with DMSO upon NOX4 knockdown (Figure 2C). Taken together, the data supports the critical roles of NOX4 and NOX4-derived ROS in the regulation of drug-induced apoptosis of starved PTC cells.
NOX4 is required for PTC cell survival in response to lenvatinib depending on NOX4-derived ROS
Next, we sought to use lenvatinib, a multi-targeted anticancer agent approved by The US Food and Drug Administration (FDA) for differentiated thyroid cancer that got worse even after they received radioactive iodine therapy [13], to investigate the influence of NOX4 in the process. A CCK8 assay demonstrated that lenvatinib does obviously decrease cell viability under both serum-full and serum-starved conditions, but NOX4 knockdown is capable of significantly reducing the cell viability caused by lenvatinib in starved PTC cells (Figure 3A). Correspondingly, in vitro studies on apoptosis showed that the lenvatinib-treated cells expressing shRNA against NOX4 in starved PTC cells has a higher apoptotic level than the cells expressing shCtrl (Figure 3B). Interestingly, it seems that the level of cell death induced by lenvatinib with 10% serum does not depend on NOX4 expression, different from the effects of etoposide and doxorubicin. This may be due to the fact that lenvatinib at 10% serum triggers apoptotic signal pathways different from chemotherapeutic drugs and independent of NOX4 expression.
Likewise, we used NAC and GOD to determine whether the role of NOX4 in Lenvatinib-induced cell death in starved PTC cells is mediated by NOX4-derived ROS. The results revealed that neither NAC nor GOD shows a significance between NOX4 knockdown and its control, whereas NOX4 knockdown did increase significant induction of apoptosis without NAC or GOD (DMSO) (Figure 3C). The data confirmed that NOX4 depends on ROS to regulate lenvatinib-induced apoptosis in starved PTC cells.
The survival of LRBCs requires NOX4 and NOX4-derived ROS.
We afterward characterized lenvatinib-resistant BCPAP cells (LRBCs) that had been exposed to doses of lenvatinib increased by 0.125 µM per week and until reaching a final concentration of 32 µM. IC50 of parent cells treated with lenvatinib in 72h was 2.91 µM, while those of LRBCs were 10.74 µM (Figure S3A), suggesting that LRBCs successfully acquired lenvatinib resistance. To determine the role of NOX4 in LRBCs, we still knocked down NOX4 expression by lentivirus-mediated shRNA. Immunoblots showed that the level of NOX4 protein is overexpressed in LRBCs compared with parent cells, but the level was dropped down when using shRNA agaist NOX4 (Figure S3B). The CCK8 and apoptosis assay together showed that nolenvatinb NOX4 knockdown significantly inhibits LRBCs survival under lenvatinib treatment independent of starvation in PTC cells (Figure 4A and 4B). Of note, starvation was helpful to break the resistance of LRBCs to lenvatinib, but NOX4 deficiency further aggravates the break of resistance.
Resistance to lenvatinib increases glycolysis [16], thus we want to assess the effect of NOX4 on glycolysis in LRBCs. The measurements of the extracellular acidification rate (ECAR) showed that LRBCs have higher glycolytic capacity than the parent cells, but the increment can be dropped close to the level of the parent cells by NOX4 knockdown (Figure 4C). Since the start and end port during glycolysis are represented by glucose uptake and lactate secretion, we investigated the influence of NOX4 on glucose uptake and lactate secretion in LRBCs. As expected, serum starvation decreased glucose levels and increased lactate levels in lenvatinib-treated LRBCs, whereas these differences were disappeared in NOX4-dificient LBRCs cells (Figure 4D), supporting an indispensable role of NOX4 in the start and end port of glycolysis under serum-starved condition in lenvatinib-resistant PTC cells.
Next, we performed modified assay to identify whether ROS mediates the role of NOX4 in the glycolysis of LRBCs by sequential injection with GOD and NAC post glucose. The results demonstrated that NOX4 knockdown leads to the downregulation of the basal glycolysis until the injection of GOD that rescues the effect of NOX4 deficiency, and both increments of ECAR were blocked by the addition of NAC (Figure 4E). Collectively, these findings suggest that NOX4 is required for the survival of lenvatinib-resistant PTC cells in which NOX4-derived ROS have a critical role in glycolytic activity.
GLX351322 leads to decreased cell survival of PTC cells and LRBCs.
Since NOX4-derived ROS is vital for cell survival in starved PTC cells, we next used a chemical inhibitor targeting NOX4, GLX351322 [24], to inhibit NOX4-derived ROS. The results showed that GLX351322 significantly reduced cell viability in starved BCPAP and TPC-1 cells no matter lenvatinib was added or not (Figure 5A), which suggests that the mechanism of the inhibition of GLX351322 and lenvatinib to cell survival of starved PTC cells is not consistent, and the combination of both drugs is expected to have a superimposable antitumor effect in starved PTC tumors in vivo. This effect was also confirmed by the detection of apoptosis showing that GLX351322 enlarged lenvatinib-induced apoptosis in starved PTC cells (Figure 5B).
In LRBCs, GLX351322 significantly decreased the survival rates at a serum concentration of 10% in the presence of lenvatinib; GLX351322 could also significantly inhibit cell survival at a serum concentration of 0.5% independent of the presence of lenvatinib, but the combinatory usage of lenvatinib and GLX351322 is capable of maximizing the inhibitory effect (Figure 5C). A similar trend was confirmed in the detection of apoptosis in LRBCs as well (Figure 5D). These findings suggests that GLX351322 can effectively break the resistance of PTC cells to lenvatinib, and this effect is independent of serum concentration.
Combination of GLX351322 and lenvatinib completely suppresses PTC tumor growth even in LRBCs.
We further inoculated BCPAP cells or LRBCs into the dorsal tissue of NOD/SCID mice for 10 days. Thereafter, we used lenvatinib to target PTC tumors, and combined GLX351322 to break the resistance of PTC tumors to lenvatinib (Figure 6A). Lenvatinib or GLX351322 in these mice partially affected the progression of BCPAP-derived tumors, the combined usage of both drugs significantly reduced tumor volumes at each measurement time point from day 14, eventually eliminating the tumors in vivo (Figure 6B and 6C). Analogously, we obtained the same conclusion that the combination of the two drugs completely suppresses xenograft growth when using LRBCs-derived tumors as a model system,but the difference was that lenvatinib has no significantly inhibitory effect on LRBCs-derived tumors (Figure 6D and 6E). A Kaplan–Meier survival curve and log-rank test also showed the significantly better survival of the mice treated with the combined drug than that of single drug-treated mice (Figure 6F and 6G). Together, these findings signify the importance of NOX4 and the clinical significance of the combination against resistance to lenvatinib in PTC tumors. Of note, we also found that NOX4 expression positively correlated with resistant marker HIF1α, glycolytic markers Glut1 and LDHA, and proliferative marker Ki67, and negatively correlated with apoptotic marker BCL2 in tumor tissues in patients with PTC (Figure S4A). Taken together, these data demonstrate that GLX351322 exerts a suppressive effect on PTC tumor growth, and combination of GLX351322 and lenvatinib can completely suppress PTC tumor growth even in LRBCs.