GPX4 deficiency results in HSPCs ferroptosis in vitro
HSCs rapidly proliferate and differentiate to generate single-cell colonies when they are cultured in vitro. For determination of whether GPX4 is essential for HSC function, individual long-term HSCs (LT-HSCs) from wild-type mice were sorted and treated with RSL3, a widely used GPX4 inhibitor. Strikingly, RSL3 completely blocked the colony formation of LT-HSCs (Fig. 1a). To exclude the off-target effects of RSL3, we generated Gpx4flox/flox Vav-Cre mice in which Gpx4 was specifically knocked out in the hematopoietic system and Gpx4flox/flox Mx-Cre mice in which Gpx4 was depleted in the hematopoietic system by pIpC treatment . Then, the LT-HSCs isolated from the Gpx4flox/flox Vav-Cre mice and the pIpC-treated Gpx4flox/flox Mx-Cre mice were evaluated by a single-cell colony-forming assay. Similar to the RSL3-treated LT-HSCs, the Gpx4-depleted (Gpx4−/− LT-HSCs lost the ability to produce single-cell colonies (Fig. 1b, c). Next, we assessed the susceptibility of different types of HSPCs to RSL3. We observed that LT-HSCs and short-term HSCs (ST-HSCs) were more resistant to RSL3 than multipotent progenitors (MPPs) and granulocyte-macrophage progenitors (GMPs) (Supplementary Fig. 1a). Consistently, the expression levels of GPX4 in the LT-HSCs and ST-HSCs were higher than those in the MPPs and GMPs (Supplementary Fig. 1b). These results suggest that GPX4 is indispensable for HSC function in vitro.
Given that GPX4 is a vital suppressor of ferroptosis, we wondered whether HSPCs undergo ferroptosis in the absence of GPX4. Hence, lineage−sca1+c-kit+ (LSK) cells isolated from the wild-type (Fig. 1d, e) or Gpx4-depleted mice (Fig. 1f, g) were cultured in vitro and treated with different drugs. Both the RSL3-treated LSK cells and the Gpx4-depleted LSK cells accumulated lipid peroxides and underwent cell death within two days (Fig. 1d−g). Moreover, neither Nec-1s nor Z-VAD affected cell viability, excluding the involvement of necroptosis and apoptosis. In contrast, the lipophilic antioxidant ferrostatin 1 (Fer-1) and the iron chelator deferiprone (DFO) reduced lipid peroxide levels and suppressed cell death (Fig. 1d−g). We noticed that although lipid peroxidation was diminished, nearly half of the RSL3-treated LSK cells failed to be rescued by DFO (Fig. 1d). Considering the essential biological function of iron in cells, we speculated that DFO may be toxic to HSPCs as they are highly dependent on iron. Indeed, after treatment with DFO alone, both LSK cells and GMPs died in vitro (Supplementary Fig. 1c, d). These results demonstrate that GPX4 deficiency leads to LSK cell ferroptosis in vitro. Furthermore, a similar scenario was observed in GMPs: both RSL3-treated and Gpx4-depleted GMPs accumulated lipid peroxides and succumbed to ferroptosis (Supplementary Fig. 1e–h). In summary, the above results suggest that HSPCs undergo ferroptosis in vitro when GPX4 is inhibited or depleted.
Gpx4 deletion does not affect homeostasis of the hematopoietic system
It has been reported that GPX4 is indispensable in embryonic development and the integrity of multiple tissues [9,10,11, 13]. To further explore the function of GPX4 in HSPCs in vivo, we studied the phenotype of HSPCs in mice in which Gpx4 was knocked out in the hematopoietic system. Gpx4 depletion in the hematopoietic system of the Gpx4flox/flox Vav-Cre mice was confirmed by Western blots (Fig. 2a). Surprisingly, no significant difference was observed between the Gpx4flox/flox Vav-Cre mice (Gpx4−/−) and the Gpx4flox/flox mice (Gpx4f/f) in the numbers of total bone marrow (BM) cells (Fig. 2b) and HSPCs (Fig. 2c, d) including LT-HSCs, ST-HSCs, MPPs, LSK cells, GMPs, and common lymphoid progenitors (CLPs). Moreover, Gpx4 ablation had no significant impact on either the ROS levels of LT-HSCs and lineage−c-kit+ (LK) cells or the lipid ROS levels of BM cells (Fig. 2e, f). Thus, the HSPC phenotype in the Gpx4flox/flox Vav-Cre mice at homeostasis was not affected by Gpx4 depletion. Given that Gpx4 deletion is detrimental in numerous tissues and Gpx4 was deleted at the embryonic stage in the Gpx4flox/flox Vav-Cre mice, it is possible that some compensatory mechanism conferred HSPC resistance to the Gpx4 deficiency. To rule out this possibility, we constructed Gpx4flox/flox Mx-Cre mice, in which Gpx4 deletion in the hematopoietic system was induced by pIpC treatment (Supplementary Fig. 2a). Similar to that in the Gpx4flox/flox Vav-Cre mice, the HSPC phenotype was not impaired in the Gpx4flox/flox Mx-Cre mice treated with pIpC (Supplementary Fig. 2b–g). These results demonstrate that Gpx4 deficiency does not affect homeostasis of the hematopoietic system in mice.
Gpx4 deficiency does not affect HSC function in vivo
During homeostasis, most HSCs remain quiescent, with low metabolic activity and low ROS levels . We wondered whether GPX4 is essential for HSCs under stress in vivo. The Gpx4-deleted mice were treated with 5-fluorouracil (5-FU) to remove cycling HSCs, which are susceptible to 5-FU. The remaining quiescent HSCs then proliferated and differentiated to restore the hematopoietic system. Nine days post 5-FU treatment, we found no significant difference in the number of HSPCs in the Gpx4-deleted mice compared with the wild-type mice (Fig. 3a). Neither the levels of ROS nor the proliferation rates of LSK cells and LT-HSCs were affected by Gpx4 depletion (Fig. 3b, c). Similar results were found in mice 22 days post 5-FU treatment (Supplementary Fig. 3a, b). These results suggest that the hematopoietic system recovered normally in the Gpx4-deleted mice.
To further investigate the function of HSCs in vivo, we conducted competitive HSC transplantation to evaluate the long-term rebuilding capacity of HSCs in the Gpx4-depleted mice. LT-HSCs obtained from the Gpx4f/f mice or Gpx4−/− mice (CD45.2) together with BM cells obtained from competitor mice (CD45.1) were transplanted into recipient mice (CD45.1.2), whose hematopoietic system was destroyed by a lethal dose of X-ray irradiation (Fig. 3d). Once transplanted, donor-derived HSCs home to bone marrow and replenish the hematopoietic system in recipient mice. However, no obvious difference in the chimerism rates of peripheral blood cells within four months (Fig. 3e) was observed even though the T cell chimerism rates were lower in the mice that received Gpx4-depleted HSCs (Supplementary Fig. 3c). Furthermore, the chimerism rates of various HSPCs especially CLPs in the primary recipient at the fourth month were not affected by Gpx4 depletion (Fig. 3f), indicating that the difference in T cell chimerism rates was not due to the difference in HSC differentiation. Indeed, it has been reported that Gpx4-deficient T cells have intrinsic defects in homeostatic balance maintenance and cell expansion upon stress . After secondary transplantation, there was still no significant difference in the chimerism rates of the HSPCs between the recipients of the Gpx4-deficient and Gpx4-sufficient LT-HSCs (Fig. 3g). Likewise, transplantation with LT-HSCs derived from the pIpC-treated Gpx4flox/flox Mx-Cre mice resulted in no significant change in the chimerism rates of the Gpx4-deficient HSPCs during either primary or secondary transplantation (Fig. 3h−j). Thus, our results show that Gpx4 deficiency does not affect HSC self-renewal and differentiation in vivo, which is quite distinct from the scenario in vitro.
α-Toc rescues HSPCs from ferroptosis ex vivo
What leads to the entirely different destiny of the Gpx4-deficient HSPCs in vivo and ex vivo? The levels of oxygen are much higher in the culture medium than in the bone marrow niche, which may inflict more ROS stress on HSPCs . We then tested whether ferroptosis is mediated by higher oxygen levels in HSPCs in vitro. The results revealed that RSL3-induced ferroptosis in both LSK cells and GMPs, was only partially relieved when they were cultured at low oxygen levels. In addition, the oxygen levels did not affect the viability of either the Gpx4-deleted LSK cells or the Gpx4-deleted GMPs (Supplementary Fig. 4a, b). Consistently, the lipid ROS levels of HSPCs were not reduced under low levels of oxygen (Supplementary Fig. 4c, d). These results suggest that oxygen levels are not the main reason for ferroptosis in the Gpx4-deficient HSPCs.
It has been reported that dietary vitamin E alleviates phenotypes resulting from lipid peroxidation in T cells and reticulocytes [16, 17]. α-Toc, the prominent component of vitamin E, is a lipophilic antioxidant and has been proven to reduce lipid peroxidation and block ferroptosis in vitro [4, 12] (Fig. 4a). We suspected that α-Toc can inhibit HSPC ferroptosis in vitro. Certainly, we observed that α-Toc, instead of N-acetyl-L-cysteine (NAC), a hydrophilic antioxidant, significantly reduced lipid ROS and inhibited ferroptosis in both LSK cells and GMPs (Fig. 4b−e). Furthermore, α-Toc endowed Gpx4-deficient LT-HSCs with the ability to generate colonies (Fig. 4f). Thus, α-Toc protects HSPCs from ferroptosis ex vivo.
Deficiency of both GPX4 and vitamin E induces HSPC ferroptosis and impairs homeostasis of the hematopoietic system in vivo
Vitamin E is absent in the cell culturing medium but is abundantly supplied from experimental animal food. Given that α-Toc rescues HSPCs in vitro, we hypothesized that dietary vitamin E is critical to protect the Gpx4-deficient HSPCs from ferroptosis in vivo. In previous experiments, mice were fed natural ingredient diets containing ≥120 IU/kg vitamin E. To verify our hypothesis, we then fed mice a fixed formulation diet (containing 75 IU/kg vitamin E) or a vitamin E-depleted diet for 3 weeks. Strikingly, we found that combined depletion of Gpx4 and vitamin E led to the loss of body weight, splenomegaly, and reduced BM cell number (Fig. 5a−c). Moreover, in peripheral blood, the proportion of both T cells and B cells decreased significantly while the proportion of M cells increased (Fig. 5d). More importantly, the numbers of LT-HSCs, CMPs, GMPs, and CLPs decreased significantly upon the removal of Gpx4 and vitamin E together (Fig. 5e). However, Gpx4 deletion alone did not impair the HSPC phenotype, consistent with previous results (Fig. 5a−e). Similarly, the depletion of vitamin E alone did not affect the HSPC phenotype (Fig. 5a−e). These data suggest that vitamin E and GPX4 cooperate to maintain hematopoietic system homeostasis.
To determine, how combined depletion of Gpx4 and vitamin E impaired hematopoietic system homeostasis, we first detected the cell cycle of HSPCs. In the absence of Gpx4 and vitamin E, no significant change in the percentage of LT-HSCs and LSK cells in G0 phase was found (Fig. 6a), indicating that the proliferation of these cells was not blocked. In contrast, significant death of HSPCs was found in the mice deficient in both Gpx4 and vitamin E (Fig. 6b), which was responsible for the decreased number of these HSPCs. Next, we observed enhanced lipid ROS levels in c-kit+ cells (Fig. 6c). Moreover, the ROS levels dramatically increased in LK cells, LSK cells, and LT-HSCs of the Gpx4-deficient mice fed vitamin E depleted chow (Fig. 6d). Taken together, our results reveal that GPX4 and vitamin E cooperatively protect HSPCs from ferroptosis in vivo.