To identify new niche factors involved in the balance between NB self-renewal and differentiation, we used repo-GAL4 to perform an RNAi-mediated screen in glial cells. Fer1HCH and Fer2LCH, two ferritin subunit encoding genes, were identified to be essential for NB maintenance and brain development. Specifically, knocking down Fer1HCH or Fer2LCH in glial cells significantly reduced the number of NBs in the CB and VNC at the third-instar larval stage, which resulted in decreased size of the CNS (Figure 1A and B and Figure 1—figure supplement 1A). To further assess the impact on NB proliferation, we employed the mitotic marker phosphohistone H3 (PH3) and the thymidine analogue 5-ethynyl-2’-deoxyuridine (EdU), revealing a substantial decrease in the proliferation rate of NBs upon glial ferritin knockdown (Figure 1A and C and Figure 1—figure supplement 1B and C). These results indicated that glial ferritin is required for maintaining NBs and their proliferation.

Figure 1 with 3 supplements see all

Download asset Open asset

To identify the glial subpopulation in which ferritin functions, we employed subtype-restricted GAL4 to perform ferritin knockdowns (Table 1). We observed that knocking down ferritin genes in cortex glia led to a similar yet milder phenotype compared to that observed in the pan-glia knockdown (Figure 1—figure supplement 1D). This milder effect may be attributed to potential compensatory mechanisms from other glial subpopulations. Interestingly, knocking down ferritin in other glial subtypes did not yield any discernible phenotype (Figure 1—figure supplement 2). These findings suggest that ferritin functions collectively across all glial populations to regulate the activity of NBs.

We confirmed the phenotype with multiple independent RNAi lines for Fer1HCH or Fer2LCH, and observed similar abnormalities of larval NBs and brain development (Figure 1—figure supplement 3A). Hereafter, UAS-Fer1HCH-RNAi (THU5585) and UAS-Fer2LCH-RNAi (TH01861.N) were used for most studies, and we only quantified NB number and proliferation rate in the CB for convenience in this research.

To evaluate the efficiency of RNAi for Fer1HCH or Fer2LCH, RT-PCR was performed with total RNA isolated from the whole brain dissected from third-instar larvae. The abundance of Fer1HCH or Fer2LCH mRNA was significantly reduced in glial Fer1HCH or Fer2LCH knockdown, respectively (Figure 1D and E). Western blot analysis of larval brain protein extract using polyclonal antibodies against Fer1HCH or Fer2LCH further confirmed that both subunits were down-regulated in glial knockdown of either subunit (Figure 1F–I), consistent with previous research in Drosophila (Missirlis et al., 2007). To exclude the possibility of RNAi off-target effects, we used repo-GAL4 to overexpress RNAi-resistant Fer1HCH (or Fer2LCH) containing silent mutations in the region targeted by shRNA, which resulted in a significant increase in the number and proliferation rate of NBs compared to glial ferritin knockdown alone (Figure 1J–L and Figure 1—figure supplement 3B and C). Additionally, similar but weaker phenotypes were observed when Fer1HCH^DN, a ferroxidase-inactive form of Fer1HCH (Missirlis et al., 2007), was overexpressed in glial cells (Figure 1—figure supplement 3D). These results support the conclusion that ferritin in glial cells is essential for the maintenance and proliferation of NBs, suggesting that iron homeostasis is involved in this process.

To investigate the role of iron homeostasis in Drosophila CNS, we tried to knock down genes involved in iron transport, storage, and regulation in NBs, neurons, and glial cells, respectively. Interestingly, only glial knockdown of ferritin-related genes led to NB loss and defective brain development. It appears that ferritin generated in glial cells is the only crucial protein for regulating the behavior of NBs and, ultimately, the larval brain development. To recapitulate the endogenous ferritin expression pattern, we took advantage of a Fer1HCH/Fer2LCH enhancer trapped GAL4 line to drive a variant of DsRed protein with a nuclear localization signal (DsRed Stinger). Surprisingly, our immunostaining experiments revealed that most, if not all, DsRed was co-localized with the glia nuclear protein Repo, instead of the NB nuclear protein Dpn (Figure 2A and A’). We also used another Fer2LCH enhancer trapped GAL4 line to verify this result, which was consistent with the above result (Figure 2—figure supplement 1). These results further confirm that glial cells are the main producers of ferritin in the larval brain.

Figure 2 with 1 supplement see all

Download asset Open asset

Fer1HCH^G188 encodes a GFP-tagged Fer1HCH subunit (Missirlis et al., 2007), which labels the distribution of Fer1HCH, representing the pattern of ferritin. As shown in the Figure 2B and B’, ferritin was detected in NBs besides glial cells, despite that ferritin is specifically produced in glial cells. To explore the possibility of ferritin transfer from glial cells to NBs, we observed the GFP signal of Fer1HCH^G188 with glial ferritin knockdown, and found that GFP in the brain was almost completely depleted (Figure 2B). Furthermore, we used repo-GAL4 to drive the expression of GFP-tagged Fer1HCH and mCherry-tagged Fer2LCH in glial cells. We observed that both Fer1HCH::GFP and Fer2LCH::mCherry were present in NBs, while the control GFP and mCherry were restricted to glial cells (Figure 2C). Altogether, we conclude that ferritin is generated in glial cells and then transferred to NBs in the larval brain.

In Drosophila, both ferritin subunits contain secretion signal peptides that direct protein to the endoplasmic reticulum during translation, which allows the ferritin to be transported out of the cell. To investigate whether the ferritin was transported from glia to NBs via the vesicular transport, we expressed a temperature-sensitive dominant-negative Shibire dynamin protein (shi^ts) to block vesicle trafficking from glial cells, and found that Fer1HCH::GFP accumulated in glia, accompanying with significantly diminished Fer1HCH signal in NBs (Figure 2D, D’ and E), suggesting glial ferritin secretes into NBs through vesicle system.

Given that ferritin is required for iron storage as well as delivery, glial ferritin knockdown might increase the free iron level, leading to glial iron overload. Alternatively, it will block the iron transport from glia to NBs, resulting in NB iron deficiency. To determine the underlying cause of the brain development defects observed upon glial ferritin knockdown, we used iron chelator or iron salt ferric ammonium citrate (FAC) to manipulate iron intake and detect how it will influence NBs. We observed that depriving iron intake by adding iron chelator bathophenanthrolinedisulfonic acid disodium (BPS) to the food led to a dramatic decrease in the number and proliferation rate of NBs, compared to control larvae on normal food (Figure 3A). To investigate the effect of feeding behavior in different foods, we tested the feeding behavior of flies by incorporating 1% Brilliant Blue (sigma, 861146) into different foods (Tanimura et al., 1982) and found the amount of dye in the fly body was similar between normal group and BPS group (Figure 3—figure supplement 1A), which suggested that BPS almost did not affect the feeding behavior. In line with this result, blocking iron absorption by knocking down intestinal ferritin (Tang and Zhou, 2013) resulted in the similar defects of the brain (Figure 3—figure supplement 1B–D). However, supplementing the diet with FAC did not induce any observable defects (Figure 3A). We also tested another iron chelator deferiprone (DFP) (Soriano et al., 2013), and found that DFP alone did not cause any brain defects in wild-type (WT) larvae, but it significantly exacerbated the effect of glial ferritin knockdown on defective NB maintenance (Figure 3B and C). Furthermore, we attempted to rescue the brain defects observed in glial ferritin knockdown by providing dietary FAC, but the effect was negligible (Figure 3—figure supplement 1E and F). This may be due to the possibility that the knockdown of glial ferritin largely, if not completely, prevented the transport of iron from glia to NBs, which hindered the additional iron from entering NBs. These results indicate that the NB loss phenotype induced by glial ferritin knockdown is primarily due to iron deficiency in NBs.

Figure 3 with 3 supplements see all

Download asset Open asset

Zip13, generally known as a zinc transporter, functions as an iron transporter, cooperating with ferritin on iron transport in the Drosophila intestine (Xiao et al., 2014). Specifically, the knockdown of Zip13 leads to failure of iron loading in secretory vesicles and consequently defective iron efflux (Xiao et al., 2014). To investigate whether the NB loss phenotype of glial ferritin knockdown is caused by defective iron transport, we next sought to identify the role of Zip13 in the larval brain. First, we examined the expression pattern of Zip13 and found that, similar to ferritin, the DsRed stinger driven by Zip13-GAL4 was co-localized with Repo in glial cells rather than Dpn in NBs (Figure 3D and D’). Subsequently, we knocked down Zip13 in glial cells and observed a similar but weaker phenotype with a slightly reduced NB number and proliferation rate (Figure 3E–G). Consistently, the defective brain development was confirmed in the Zip13 mutant (Figure 3—figure supplement 1G–I). Moreover, the effect of double knockdown of both Zip13 and Fer2LCH was much more severe than that of either single knockdown, with a significant reduction in NB number and proliferation rate (Figure 3E–G). Furthermore, we found that the reduced proliferation rate caused by Zip13 knockdown could be partially rescued by iron supplementation in the food (Figure 3H and I). Overall, these results suggest that glial ferritin collaborates with Zip13 to deliver iron from glia to NBs.

Since previous studies have demonstrated the essential role of glial cells in NB survival and proliferation (Bailey et al., 2015; Read, 2018), it is possible that iron overload in glial cells up-regulated the level of reactive oxygen species (ROS), leading to apoptosis of glial cells, which ultimately resulted in the loss of NBs. To rule out this possibility, firstly, we examined the ROS level of glial ferritin knockdown via a transgenic reporter GstD-GFP for oxidative stress signaling (Sykiotis and Bohmann, 2008). GstD1 is a prototypical oxidative stress response gene (Sawicki et al., 2003). We found an obvious increase as expected (Figure 3—figure supplement 2A and B). However, overexpressing anti-oxidant genes in glial cells or NBs did not restore the NB number and proliferation (Figure 3—figure supplement 2C and D). ROS accumulation leads to lipid peroxidation, which in turn results in ferroptosis (Hirschhorn and Stockwell, 2019). Mitochondria in cells undergoing ferroptosis typically exhibit a characteristic shrinkage and damage, while the nuclei remain structurally intact (Mumbauer et al., 2019). In order to investigate whether the NB loss is due to ferroptosis, we examined the mitochondrial morphology using Transmission Electron Microscopy (TEM). However, our result showed that the mitochondrial double membrane and cristae were clearly visible whether in the control group or glial ferritin knockdown group, which suggested that ferroptosis was not the main cause of NB loss upon glial ferritin knockdown (Figure 3—figure supplement 2E and F). In addition, we investigated the level of 4-HNE that indicates lipid peroxidation, the other hallmark of ferroptosis. This result showed that the 4-HNE level did not change significantly after ferritin knockdown (Figure 3—figure supplement 2G and H), suggesting that lipid peroxidation was stable, which supported the exclusion of the ferroptosis in glial ferritin knockdown. Taken with the findings from iron chelator feeding together, we conclude that NB loss resulting from glial ferritin knockdown is not a consequence of ferroptosis. Second, we examined the apoptosis in larval brain cells via Caspase-3 or TUNEL staining, and found the apoptotic signal remained unchanged after glial ferritin knockdown (Figure 3—figure supplement 3A–D). Furthermore, we overexpressed the baculovirus protein P35 to inhibit apoptosis in glial cells, but failed to restore the lost NBs (Figure 3—figure supplement 3E). Given that glial morphology is important for NB lineages (Spéder and Brand, 2018), we used GFP driven by repo-GAL4 to denote glia cells and found that the glial chamber was almost intact after glial ferritin knockdown (Figure 3—figure supplement 3F). Together, we conclude that it is iron deficiency in NBs, rather than iron overload in glial cells, that leads to NB loss induced by ferritin defects.

To further prove that defective iron transport leads to the iron deficiency in NBs, we examined the iron availability by measuring aconitase activity in the brain. The activity of aconitase is sensitive to the level of Fe-S synthesis, so it can serve as an indicator of the availability of iron in the cell (Haile et al., 1992; Tong and Rouault, 2006). Aconitase activity is determined in a coupled enzyme reaction in which citrate is converted to isocitrate by aconitase (Aconitase Activity Assay Kit, sigma, MAK051). As shown in the Figure 4A, aconitase activity of ferritin knockdown diminished significantly, indicating that the availability of iron in the brain was reduced. This result also implies the Fe-S clusters in NBs are probably deficient when knocking down ferritin in glia. Fe-S clusters are essential to activate the aconitase that converts citrate into isocitrate in the tricarboxylic acid cycle (TCA cycle) in mitochondria, and are one type of critical electron carrier for electron transport chain (ETC), especially for the Complex I, II and III function. Considering that the TCA cycle and ETC are critical biological processes for ATP production, we proposed that the knockdown of glial ferritin might lead to ATP deficiency due to the block of the TCA cycle and ETC, which resulted in the failure of maintaining NBs and their proliferation. To verify this hypothesis, we first assessed the function of Fe-S clusters in NBs by knocking down mitochondrial Fe-S protein Nfs1 and the components of the cytosolic iron-sulfur protein assembly (CIA) complex. After knocking down Nfs1 and the CIA component CG17683, galla2, and ciao1 in NBs respectively, the proliferation level of NBs decreased significantly and brain size became smaller (Figure 4B and C), which phenocopied the knockdown of Fer1HCH or Fer2LCH in glia. This result implies that Fe-S clusters are indispensable for NBs to maintain their proliferation.

Figure 4 with 1 supplement see all

Download asset Open asset

To identify the biological processes related to Fe-S clusters in NB lineages, we performed transcriptome analysis of sorted NB lineages from glial ferritin knockdown as well as the control groups. We used NB driver Insc-GAL4 to express UAS-DsRed, labeling NB lineages, and then sorted these cells via fluorescence-activated cell sorting (FACS) (Figure 4D; Harzer et al., 2013). We found that thousands of genes were significantly altered in the Fer1HCH and Fer2LCH knockdown groups compared to the control group. Specifically, 880 genes were simultaneously downregulated among the significantly changed genes in both the Fer1HCH and Fer2LCH knockdown groups contrasted to the control group (Figure 4E). Given that iron deficiency in NBs probably leads to the defects of energy production and biosynthesis, we performed gene enrichment analysis in these genes based on Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG). GO term analysis showed that mitochondrial organization, translation, transport, oxidative phosphorylation (OxPhos), and ETC were enriched in the Biological Processes (Figure 4F), indicating NB mitochondrial functions including OxPhos and ETC were abnormal after glial ferritin knockdown. Moreover, NADH dehydrogenase and ribosome-related terms were also enriched whether in the Cellular Component or Molecular Function (Figure 4F), suggesting that NADH dehydrogenase and biosynthesis were defective. Further, we validated the enriched term associated with mitochondrial function by qRT-PCR (Figure 4—figure supplement 1A). And the enriched KEGG pathways included OxPhos and ribosome (Figure 4G), aligning with the GO enrichment results and supporting the hypothesis that glial ferritin knockdown disrupted mitochondrial function including OxPhos and ETC as well as further biosynthesis.

To confirm whether the function of NADH dehydrogenase was disrupted, we used the SoNar that could bind to NAD⁺ with a specific conformation and excitatory wavelength to monitor the NAD⁺ level (Bonnay et al., 2020; Zhao et al., 2016). SoNar was driven by the Insc-GAL4 to indicate the NAD⁺ level in NBs. Considering the possibility of protein accumulation in NBs due to the lower proliferation rate in the glial ferritin knockdown group, the signal of SoNar was normalized to DsRed Stinger driven by Insc-GAL4. We observed significantly declined SoNar/DsRed after knocking down Fer2LCH (Figure 4H and I), suggesting the normalized level of NAD⁺ in NBs was decreased, which supports the conclusion that the activity of NADH dehydrogenase was declined. To further verify the change in energy production, we utilized the ATP Determination Kit which offers a convenient bioluminescence assay for the quantitative determination of ATP with recombinant firefly luciferase and its substrate D-luciferin. We found that ATP level decreased significantly after glial ferritin knockdown when compared to the control (Figure 4J). Therefore, we tried to supply the energy for NBs by restoring bioenergy-dependent NAD⁺ metabolism. To validate this, we first determined the ATP production after overexpressing Ndi1 in NBs upon glial Fer2LCH knockdown. The data suggested that expression of Ndi1 can restore ATP production (Figure 4—figure supplement 1B). Furthermore, the NB number could be partially rescued by overexpressing proton-pump-independent yeast mitochondrial NADH dehydrogenase Ndi1 in NBs under the background of glial ferritin knockdown (Figure 4K and L; Bonnay et al., 2020). In summary, the iron deficiency leads to NB loss due to the insufficient energy required for essential biological processes.

To investigate how glial ferritin knockdown leads to NB loss, we first examined whether NB origin was affected by altering the timing of ferritin knockdown. We used the temperature-sensitive GAL4 inhibitor GAL80^ts to suppress ferritin RNAi at a permissive temperature (18℃) during the embryonic stage, and then shifted to a restrictive temperature (29℃), or vice versa until we dissected them at the third-instar larval stage. Ferritin knockdown during the larval stage led to severe brain development defects (Figure 5A and B), while the knockdown only during the embryonic stage showed no apparent abnormality (Figure 5—figure supplement 1A and B). Together, these results indicate that glial ferritin is required for NB maintenance at the larval stage. Consistent with this result, we found no obvious defects of the brain at the first-instar larval stage when knocking down glial ferritin (Figure 5—figure supplement 1C and D). Since newborn NBs are delaminated from the neuroectoderm during the embryonic stage (stage 9–11), and ferritin RNAi only during the embryonic stage did not show any phenotype, suggesting that NB loss after glial ferritin knockdown is not ascribed to defective NB origin.

Figure 5 with 3 supplements see all

Download asset Open asset

Furthermore, we explored the earliest stage when the phenotype (NB number and proliferation) existed after glial ferritin knockdown. The result showed that NB proliferation decreased significantly, but NB number declined slightly at the second-instar larval stage (Figure 5—figure supplement 1E and F), suggesting that brain defect of glial ferritin knockdown manifests at the second-instar larval stage.

As mentioned before, we did not detect any apoptotic signal in the larval brain after glial ferritin knockdown, suggesting that there is no association between apoptosis and NB loss. The double binary systems, lexA/lexAop, and GAL4/UAS, were employed to further exclude the possibility of apoptosis. We overexpressed P35 in NBs by Insc-GAL4 and knocked down ferritin by repo-lexA, which failed to restore the decreased number of NBs (Figure 5—figure supplement 1G). These results reinforce the notion that NB loss occurs independently of the apoptotic pathway.

Next, we hypothesized that the NB loss might be attributed to nuclear-Prospero (Pros)-dependent premature differentiation of NBs (Cabernard and Doe, 2009; Choksi et al., 2006). Pros in NB nuclear was hardly detected in third-instar WT larvae, but was increased significantly with glial ferritin knockdown (Figure 5C and D). To further verify that the NB loss caused by glial ferritin knockdown was due to nuclear-Pros-dependent premature differentiation, we used the Insc-GAL4 to drive UAS-pros-RNAi in order to prevent the NB loss in glial ferritin knockdown background. We found that the NB number was substantially restored (Figure 5E and F). Interestingly, despite of dramatic NB number increase, the decreased proliferation rate was not rescued. We proposed that pros knockdown in NBs blocked premature terminal differentiation, but it could not replenish the reduced energy caused by glial ferritin knockdown. As NB self-renewal and proliferation are both energy-consuming processes, the self-renewal was rescued via inhibiting differentiation by Insc-GAL4 driving UAS-pros-RNAi, but at the expense of energy consumption for proliferation, leaving the proliferation rate unchanged, or even lower. Together, we conclude that glial ferritin maintains the number of NBs by preventing the nuclear-Pros-dependent premature differentiation.

Furthermore, we also investigated whether overexpression of Ndi1 could restore Pros localization in NBs. This result showed that overexpressing Ndi1 could significantly restore Pros localization in NBs (Figure 5—figure supplement 2), which supported that energy shortage induced the entry of Pros into nuclei, leading to the premature differentiation of NBs.

Considering that asymmetric division defects in NBs may lead to premature differentiation, we also explored the asymmetric division by staining the classical asymmetric marker aPKC and found it displayed a crescent at the apical cortex based on the daughter cell position whether in control or glial ferritin knockdown (Figure 5—figure supplement 3A). This result indicated that there was no obvious asymmetric defect after glial ferritin knockdown.

Given that reduced proliferation is a distinctive characteristic of quiescent NBs, we tried to determine whether NBs were in a quiescent state after glial ferritin knockdown by examining the other two key features associated with quiescent NBs: decreased cell size and the formation of the extended cellular protrusions (Chell and Brand, 2010; Ly et al., 2019). However, we found that the NB size remained at the same level as the control, and the extended cellular protrusion was not observed after glial ferritin knockdown (Figure 2B and Figure 5—figure supplement 3B). Furthermore, since insulin signaling is necessary and sufficient to reactivate NBs from quiescence (Sousa-Nunes et al., 2011), we tried to overexpress a constitutively active form of InR (InR^act) in NBs to promote reactivation at the background of glial ferritin knockdown. However, InR^act overexpression failed to ameliorate the NB abnormalities including the decreased proliferation (Figure 5—figure supplement 3C and D). In conclusion, NBs exhibited a low proliferation rate after glial ferritin knockdown, but were not in a quiescent state.

We have demonstrated that glial ferritin is essential for NB self-renewal and proliferation by providing glia-to-NB iron transport to support energy production in NBs. We next tried to determine whether manipulating this biological function could suppress tumor development. We induced tumor development by driving UAS-brat-RNAi or UAS-numb-RNAi under the control of Insc-GAL4, and knocked down ferritin genes under the control of repo-lexA. We observed that the size of tumors was strikingly decreased in the larval brains lacking glial ferritin (Figure 6A and B and Figure 6—figure supplement 1A and B). This result suggests that glial ferritin is required for tumor development, highlighting a potential target for tumor treatment.

Figure 6 with 1 supplement see all

Download asset Open asset

To verify whether iron is involved in the tumor suppression induced by blocking glial ferritin, we tried to decrease iron levels by adding the iron chelator BPS in the Drosophila food. Tumor formation in the brain at the third instar larval stage was induced with the same method as mentioned above. The addition of BPS into dietary food significantly reduced the tumor size (Figure 6C and D and Figure 6—figure supplement 1C and D), suggesting that iron is required for tumor development in Drosophila larval brain. Considering the conservation of iron’s fundamental function as an essential nutrient for cell growth and proliferation, we thus tried to suppress brain tumors in mice via iron deficiency. Initially, we constructed an orthotopic glioma model in C57BL/6 J mice. One week after inoculation of glioma cells (GL261-luc), the iron chelator DFP was intraperitoneally injected every two days based on the murine weight until being dissected on days 7, 14, and 21, respectively. Histological examination with HE staining revealed that the size of glioma in mice, injected with DFP, decreased significantly on days 14 and 21 (Figure 6—figure supplement 1E–G). Since glioma cells GL261-luc carry the luciferase gene, we injected the substrate luciferin potassium salt into the tumor-bearing mice and detected the bioluminescence signal of mouse gliomas using IVIS Spectrum. This result showed that the bioluminescence signal of glioma in mice with the injection of 10 mg/ml DFP was significantly reduced (Figure 6E and F). The above results suggested that iron deficiency induced by DFP could suppress glioma in mice effectively. Furthermore, we assessed the survival time of the tumor-bearing mice with DFP or saline injection and found that 10 mg/ml DFP injection could significantly prolong the survival time of tumor-bearing mice (Figure 6G). Altogether, blocking iron in the tumor is a potential therapy for tumor suppression.

The aforementioned results have highlighted the significant role of glial ferritin in controlling NB number and proliferation. We next wondered whether NBs could provide feedback to regulate glial ferritin levels. To answer this question, we analyzed ferritin levels after altering NB proliferation. Previous studies have corroborated that NB proliferation was diminished by blocking the Tor signaling pathway or enhanced by knocking down tumor suppressor genes, such as brat or numb (Bowman et al., 2008; Reichardt et al., 2018; Sousa-Nunes et al., 2011). When NBs were in low proliferative status, ferritin declined slightly with no significance (Figure 7A–D). Conversely, highly proliferative NBs, able to form tumors, induced significantly increased levels of the ferritin (Figure 7A–D). In summary, the production of ferritin in glial cells depends on the level of NB proliferation.

Figure 7

Download asset Open asset

Previous literatures have reported how the glial niche regulates NB reactivation, proliferation, and survival. Blood-brain barrier (BBB) glia secretes dILP2 and dILP6 that bind on the InR of NBs, which switches on Insulin signaling to reactivate NBs (Chell and Brand, 2010; Spéder and Brand, 2014). In addition, Pvr (PDGFR ortholog) signaling in cortex glia sustains NB survival and proliferation in a PI3K-DE-cadherin-dependent manner (Read, 2018). However, it is unknown whether the glial niche is required for maintaining the self-renewal of NBs. In this study, we demonstrate that the glial niche maintains the self-renewal and proliferation of NBs through supplying the necessary microelement iron into NBs for enough energy production, which implies that mammalian glia likewise provide critical microelement for maintaining NSC function. Oligodendrocytes have been reported to secret ferritin to protect neurons (Mukherjee et al., 2020), however, it remains unknown whether the ferritin secretion is employed by oligodendrocytes to regulate NSCs. Astrocytes regulate NSC proliferation via secreting growth factors (Shetty et al., 2005), but it is unclear whether ferritin from astrocytes regulates NSC proliferation and other NSC behaviors, such as self-renewal and differentiation. Based on our findings, it is feasible to regulate the self-renewal and differentiation of NSCs indirectly through niche cells, which would help to reduce the risk of directly manipulating NSCs to treat neural diseases.

Our results show that ferritin generated outside the brain was not detected in the CNS and vice versa (data not shown), suggesting ferritin can hardly cross the BBB glia, possibly because ferritin is too large to permeate. In this way, BBB glia separates the labile systemic iron from local iron inside the brain to guarantee the normal brain development, as the neural cells, including NBs, require a stable iron level for continuous energy generation during neurogenesis.

Although the dominant function of ferritin is iron storage in mammals, increasing evidences suggest that ferritin also can transport iron. H-ferritin expression in astrocytes is necessary for proper oligodendrocyte development and myelination (Cheli et al., 2021), which implies H-ferritin probably transports iron from astrocytes to oligodendrocytes for myelination. In mice, ferritin heavy chain is secreted by oligodendrocytes via extracellular vesicles (Mukherjee et al., 2020), which possibly endows ferritin with the function of iron transport. Consistently, previous results showed that ferritin can deliver iron across the cultured BBB (Chiou et al., 2019; Fisher et al., 2007). Our results emphasize the critical role of ferritin in iron transport in the CNS. Future studies would be interesting to investigate whether ferritin is able to transport iron in the mammalian brain.

Ferritin specifically targets on the tumor cells, which has been applied to tumor visualization and targeted drug delivery (Fan et al., 2012; Fan et al., 2018). Our study shows that the block of ferritin or iron can inhibit tumors effectively, which offers an explanation that ferritin targets on the tumor is to transport iron for maintaining the cell proliferation, highlighting a potential target for tumor suppression. In consideration of the role of ferritin in normal tissue, such as antioxidant defense function for neurons (Mukherjee et al., 2020), the entire block of ferritin for suppressing tumor probably brings neurological disorders. Therefore, we should target on ferritin specifically in tumors to suppress tumor growth.

Iron is an essential component for Fe-S clusters which participate in TCA and OxPhos, providing energy required for self-renewal and proliferation of NBs. Previous studies claimed that OxPhos is required for NB proliferation (van den Ameele and Brand, 2019). However, another research observed that OxPhos is dispensable in type II NBs (Bonnay et al., 2020). Our data imply that energy deficiency in NBs caused by iron shortage leads to NB loss and low proliferation possibly through simultaneous disruption of TCA and OxPhos. Consistently, iron-loaded ferritin nanoparticles can be used to improve the self-renewal ability and differentiation potential of human NSCs in vitro (Lee et al., 2018). Additionally, previous studies have shown that iron-loaded ferritin can promote the proliferation of cultured cell from Drosophila (Li, 2010). Furthermore, iron depletion was reported to suppress human cancer cells growth at G1/S by inducing cyclin D1 proteolysis (Nurtjahja-Tjendraputra et al., 2007). These studies suggest that iron, as a necessary microelement for energy production, is essential to NSC maintenance and proliferation.

Iron homeostasis can be regulated systemically by the liver-derived hormone hepcidin and the iron exporter ferroportin, and it is also regulated cellularly by the IRP/IRE system (Hentze et al., 2010). Here, we propose a bicellular iron homeostasis between NB and glial niche cell, which is distinct from classic cellular and systemic iron homeostasis. The bicellular iron homeostasis requires an elaborate mechanism to guarantee normal neurogenesis.

In a proliferative state, NBs receive iron from glial cells via ferritin to fulfill the energy requirements for proliferation and self-renewal. Therefore, the iron required by NBs is mainly stored in the ferritin of glial cells and allocated as needed. The bicellular iron homeostasis mechanism makes glial cells the main pool for ferritin production, relieving NBs from the need to produce ferritin and transferring the potential risk of iron overload-induced oxidative stress to glial cells. Furthermore, glial cells can subsequently clear oxidative stress through lipid droplets (Bailey et al., 2015), ensuring normal NB function during neural development. It is intriguing to explore whether bicellular iron homeostasis exists in other stem cells and their niches.

When normal NBs are converted into immortal tumor NBs, becoming highly proliferative, they need more nutrients including iron (Bonnay et al., 2020; van den Ameele and Brand, 2019). In response, more ferritin is generated to transport iron from the glial cells to NBs. However, it is unclear how NBs communicate with glial cells to determine the level of ferritin production. It’s interesting to identify the specific cues from NBs that regulate the level of ferritin in future studies. In addition, iron distribution in the Drosophila brain is not observed due to low iron concentration and technical limitations in iron detection, which results in a lack of direct evidence for iron trafficking via ferritin.

Fly strains were bred on yeast-containing medium at a constant of temperature 25℃ in 12 hr light/dark cycle unless otherwise stated. UAS-Fer1HCH-RNAi (THU5585) and UAS-Fer2LCH-RNAi (TH01861.N) were used for most experiments except Figure 1—figure supplement 2A (BDSC: 60000, VDRC: 14491, and VDRC: 12925) and Figure 3—figure supplement 3E (VDRC: 14491). UAS-Fer1HCH, UAS-Fer2LCH, lexAop-Fer1HCH-RNAi and lexAop-Fer2LCH-RNAi were generated in this paper. GstD1-GFP was from Fisun Hamaratoglu. repo-lexA was from Margaret S Ho. UAS-Ndi1 and UAS-SoNar were from Juergen A. Knoblich. Insc-GAL4 (BDSC:8751) was from Yan Song. UAS-RedStinger was BDSC:8546. UAS-Fer1HCH::GFP; UAS-Fer2LCH::mCherry was from Fanis Missirlis. Fer1HCH^G188 was from Bing Zhou. Other fly lines used in this study were bought from Bloomington Drosophila Stock Center, Tsinghua Fly Center, Vienna Drosophila Resource Center, and Drosophila Genetic Resource Center (Supplementary file 1 and Supplementary file 2).

200 RNAi lines were screened in this study. These genes were related to classical signaling pathways, high expression in glial cells, or secretory protein. UAS-RNAi lines were crossed with repo-Gal4, and then the third-instar larvae of F1 were dissected to harvest brains. Brains were performed immunostaining with Dpn and PH3. Finally, brains were observed in Confocal Microscope.

All mouse experiments were performed under the guidelines of the Institutional Animal Care and Use Committee at Southeast University (Approval number: 20211104002). Mice were maintained in a barrier facility, at 25℃, on a regular 12 hr light and 12 hr dark cycle.

GL261-luc murine glioma cell line was purchased from Shanghai Zhong Qiao Xin Zhou Biotechnology Co., Ltd. These cells were identified by STR and tested negative for mycoplasma contamination. Cells were cultured in DMEM medium (Gibco) supplemented with 10% FBS (Gibco) and 1% penicillin/streptomycin (Gibco) in an incubator with 5% CO₂ at 37 °C.

For generating RNAi-resistant UAS-Fer1HCH and UAS-Fer2LCH constructs used for transgenic fly lines, Fer1HCH cDNA (PA type) and Fer2LCH cDNA (PA type) with silent mutations in the region targeted by siRNA were inserted into pUAST plasmids, respectively. The cDNA with silent mutations was synthesized by Sangon Biotech (Shanghai) Co., Ltd. To generate transgenic fly lines of lexAop-Fer1HCH-RNAi and lexAop-Fer2LCH-RNAi, the shRNA sequences targeted on Fer1HCH and Fer2LCH were generated by PCR amplification of UAS-Fer1HCH-RNAi (TH01861.N) and UAS-Fer2LCH-RNAi (THU5585) DNA using primers (5’- ACGGAGCGACAATTCAATTCA-3’/5’-TGATGCCTACCTGATGCCAA-3’), respectively. The fragments were digested by EcoRI and XbaI, and then ligated into lexAop vectors. The constructs were verified by sequencing.

Larval brains were dissected in PBS (1x), fixed in 4% paraformaldehyde/PBS for 20 min, and washed three times with 0.3% PBST (Triton X-100). Then the brains were blocked in 1% BSA for 1 hr at room temperature. Finally, these brains were incubated with antibodies overnight at 4℃. Primary antibodies: Guinea pig anti-Dpn (1:1000, gift from Juergen A. Knoblich), rabbit anti-PH3 (1/100, CST, 9701), rat anti-Mira (1/1000, abcam, ab197788), mouse anti-Repo (1/100, DSHB, 8D12), rabbit anti-GFP (1/1000, Torrey Pines Biologies, TP401), rabbit anti-Caspase3 (1/100, CST, 9661), mouse anti-Pros (1/20, DSHB), mouse anti-aPKC (1/50, Santa Cruz, Sc-17781).

To detect primary antibodies, the following Alexa-Flour conjugated second antibodies from Invitrogen were used: goat anti-rat Alexa 555 (1/100, A21434), goat anti-guinea pig Alexa 555 (1/100, A21435), donkey anti-rabbit Alexa 488 (1/100, A21206), donkey anti-rat Alexa 488 (1/100, A21208), donkey anti-mouse Alexa 488 (1/100, A21202), goat anti-rat Alexa 633 (1/100, A21094), donkey anti-mouse Alexa 647 (1/100, A31571). All immunostaining images were captured using an LSM700 or an LSM900 (Zeiss) confocal microscope.

Dissected freshly brains were incubated in EdU solution for about 30 min to incorporate EdU, then fixed and permeabilized referring to the steps from immunostaining. EdU was detected by Click-iT EdU Alexa Fluor 555 Imaging Kit according to the instructions. TUNEL detection was determined using a kit based on the manufacturer’s instructions.

Total RNA was extracted from larval brains by a TRIzol reagent. cDNA was reverse-transcribed from total RNA using a kit. Semiquantitative RT-PCR was performed using the specific primers to amplify the targeted gene segment. Quantitative Real-time PCR was performed on LightCycler 96 (Bio-Rad) using qPCR SuperMix (TransGen Biotech). Following primers were used:

• rp49 F: GCACCAAGCACTTCATCC

• rp49 R: CGATCTCGCCGCAGTAAA

• Fer1HCH F: ATGGTGAAACTAATTGCTAGC

• Fer1HCH R: TCAGATCGCTGACTCCCTC

• Fer2LCH F: GCATGCTCTACGTCAGCCT

• Fer2LCH R: TTACTGCTTCTGCAGATACTC

• ND15 F: ATGTCGCTTACCCCCTTTCTACG

• ND15 R: CTAGTAGGCATCAACGCGTGG

• CG15715 F: ATGGCACGTGGACACCAG

• CG15715 R: TCAGACCTCCTTCAGCTCCTCG

Brains were dissected in PBS, and then homogenized in the RIPA buffer with SDS. Homogenate was centrifuged at 12,000 rpm for 5 min at 4℃ three times and boiled for 10 min. The protein was separated on 15% SDS-PAGE and transferred to nitrocellulose membranes. The membranes were blocked in 5% milk and incubated overnight at 4℃ with antibodies. The blots were detected using Tanon 5200. Primary antibodies used: rabbit anti-Fer1HCH (1/500, this study), rabbit anti-Fer2LCH (1/500, this study), mouse anti-tubulin (1/1000, abcam, ab7291), mouse anti-Repo (1/200, DSHB, 8D12), rabbit anti-4-HNE (1/1000, abcam, ab46545). Rabbit polyclonal antibodies were generated against recombinant proteins His-Fer1HCH or His-Fer2LCH. In brief, the aa 23–200 of Fer1HCH (PA type) or aa 41–227 of Fer2LCH (PA type), contained by different Fer1HCH or Fer2LCH cDNA isoforms, was synthesized and cloned into pET28a vector, respectively. The recombinant proteins were expressed in E. coli, extracted, and injected into rabbits for generating the antibodies (Abclonal Biotech.). Then the antibodies were affinity-purified and applied in western blot.

GAL80^ts was introduced into this system to control the function of GAL4 according to the temperature for determining the temporal window during which ferritin functions to maintain NBs. The eggs of tubulin-GAL80^ts; repo-gal4 crossing with UAS-Fer1HCH RNAi or UAS-Fer2LCH RNAi were collected, and cultured at 18℃ (permissive temperature for GAL80^ts) before hatching, then transferred to 29℃ (restrictive temperature for GAL80^ts) after hatching, or vice versa until third-instar larval stage. Then brains were observed and analyzed.

Supplemented iron salt and iron chelators (1x) in Drosophila food were used as follows: 5 mM ferric ammonium citrate (FAC) (sigma), 0.1 mM bathophenanthrolinedisulfonic acid disodium (BPS) (sigma), 2.5 mM deferiprone (DFP) (sigma).

NB lineages were labeled with DsRed drive by Insc-GAL4. These cells were dissociated and sorted according to the previous protocol (Harzer et al., 2013). Briefly, brains were dissected in Schneider’s medium and transferred into Rinaldini’s solution. Then Rinaldini’s solution was removed, and dissociation solution was added to the brains to incubate at 30℃ for 1 hr. After washing these brains twice in Rinaldini’s solution, the brains were disrupted in PBS. Finally, NB lineages were sorted by flow cytometer (Beckman) based on the red signal.

Protein was extracted with assay buffer from Aconitase Activity Assay Kit (sigma, MAK051). The concentration of protein was determined by Pierce Rapid Gold BCA Protein Assay Kit (Thermo Fisher, A53225). Aconitase was measured according to the Aconitase Activity Assay Kit.

TEM samples were prepared by standard procedures (Guangming et al., 2020). In brief, brains from third instar larvae upon WT and glial ferritin knockdown were dissected in fresh PBS and fixed at 4 °C overnight in a mixture of 2% glutaraldehyde and 2% formaldehyde in 0.1 M sodium cacodylate buffer (pH 7.4), followed by several rinses with cacodylate buffer. The samples were then postfixed for 2 hr with 1% OsO₄ in cacodylate buffer and rinsed twice with distilled water. The preparations were stained for 2 hr with 2% saturated uranyl acetate in distilled water and rinsed twice with distilled water. The specimens were dehydrated in an ethanol series, passed through propylene oxide two times, and embedded into a sheet in Epon812 (SPI Science). Each slice was 90 nm thick, and 30–40 slices were gathered into a group and attached to a grid. Samples were analyzed with a Hitachi H-7650 electron microscope operated at 80kV.

To establish an orthotopic glioma model (Wang et al., 2022), C57BL/6 mice (4–6 weeks) were anesthetized using 2% isoflurane and positioned in a stereotactic instrument. GL261-luc cell suspension (1×10⁵ cells in 3 μL PBS) was injected into the striatum. Specifically, the injection site is at 0.5 mm anterior, 2 mm lateral to the bregma, and 3.5 mm below the skull. The injection was done slowly about 10 min.

The mice with glioma were injected intraperitoneally with DFP every 48 hr (Eybl et al., 2002). DFP was dissolved in saline (0.9% NaCl) to 10 mg/ml. The doses of injection in mice were 150 mg/kg using 10 mg/ml DFP. The injection volume was determined based on murine weight.

For tumor imaging, mice were injected with luciferin (150 μg/g) in PBS (Wang et al., 2019). After 10 min, the mice were imaged using the IVIS Spectrum (PerkinElmer).

At the time of sacrifice, mouse brains were removed and fixed in 4% paraformaldehyde for 24 hr. After fixation, brains were immersed in 30% sucrose and embedded in OCT for cryosectioning. Sections were stained by Hematoxylin and Eosin (HE). Tumor volumes were estimated using Image J.

For quantification of NSCs, Dpn- or Mira-positive NSC of CB or the thoracic VNC at the indicated stage were counted. Mitotic index is the number of PH3-positive cells among Dpn-positive cells. For tumor size in Drosophila, the area of CNS was measured by NIH ImageJ. For quantifications of NB size, fluorescence, or western blot intensity, NIH ImageJ was used to measure.

All experiments were repeated at least three times. The figure legends showed the number of samples, statistical parameters, and significance. GraphPad Prism 6 was used to perform statistical analyses. For comparisons between two groups, an unpaired two-sided Student’s t-test was employed to assess statistical significance. For comparisons between three of more groups, statistical significance was determined using one-way ANOVA with a Bonferroni test. The log-rank test was used to compare the survival curves of mice. The sample distributions were assumed to be normal and no particular statistical tests were used to check for normality within the samples. Statistical results were presented as means ± SD and p-values of less than 0.05 were considered statistically significant. Asterisks indicate critical levels of significance (*p<0.05; **p<0.01; ***p<0.001; ****p<0.0001).

The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

We are grateful to Drs. Bing Zhou, Margaret S Ho, Juergen A Knoblich, Fanis Missirlis, Fisun Hamaratoglu, Yan Song, Andrea H Brand; Bloomington Drosophila Stock Center (Indiana University, Bloomington, Indiana), the Drosophila Genetic Resource Center (Kyoto Institute of Technology, Kyoto, Japan), the Vienna Drosophila RNAi Center (Vienna, Austria), and Tsinghua Fly Center (Tsinghua University) for fly stocks and reagents. We thank Dr. Bing Zhou for the useful advice on this study. We thank Drs. Junhua Geng, and Mingdao Mu for comments on the manuscript. National Natural Science Foundation of China (grant no. 32100784, grant recipient, Menglong Rui; grant no. 31970675, grant recipient, Su Wang).

All mice experiments were performed under the guidelines of the Institutional Animal Care and Use Committee at Southeast University (Approval number: 20211104002).

1. Sent for peer review: November 9, 2023
2. Preprint posted: November 12, 2023
3. Reviewed Preprint version 1: March 4, 2024
4. Reviewed Preprint version 2: August 19, 2024
5. Version of Record published: September 10, 2024

You can cite all versions using the DOI https://doi.org/10.7554/eLife.93604. This DOI represents all versions, and will always resolve to the latest one.

© 2024, Ma et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.