Drosophila expresses blood cells that are most akin to the myeloid blood in vertebrates and has been a useful animal model system for hematopoiesis (Gold and Bruckner, 2015; Letourneau et al., 2016; Shim, 2015; Waltzer et al., 2010). Hematopoiesis of Drosophila takes place in two distinct phases: first in the embryonic head mesoderm and later in the larval lymph gland. Blood cells originating from the embryonic head become blood cells which travel the embryo. These cells later form the circulating and sessile blood population that collectively function in innate immune responses during larval stages (Evans et al., 2003). In Drosophila, there are at least three terminally differentiated mature blood cell types: plasmatocytes, crystal cells and lamellocytes (Jung et al., 2005). The plasmatocyte functions similar to macrophages and constitutes around 95% of total blood cells. Hemolectin (Hml) and Peroxidasin (Pxn) are plasmatocyte markers which are useful in Drosophila blood analysis (Jung et al., 2005). The crystal cell is a non-phagocytic cell that plays a role in melanization and wound healing, constituting about 5% of total blood population. Notch is essential for the crystal cell development together with a Runx family transcription factor, Lozenge (Lz) (Lebestky et al., 2003; Mukherjee et al., 2011). The last cell type, the lamellocyte, is a rare cell type that is produced upon immune challenges and participates in encapsulation (Lanot et al., 2001; Rizki and Rizki, 1978; Rizki et al., 1985; Sorrentino et al., 2002). Differentiation of lamellocytes requires activation of Toll, JAK/STAT or JNK pathways, a signaling component triggered by stress responses (Agaisse and Perrimon, 2004; Qiu et al., 1998; Zettervall et al., 2004).

The second wave of hematopoiesis occurs in a specialized organ, called the lymph gland, where blood progenitors proliferate and differentiate into mature blood cells. The lymph gland is composed of three zones: the posterior signaling center (PSC), the medullary zone (MZ) and the cortical zone (CZ) (Fig. 1A). The MZ is located at the inner core of the lymph gland and harbors undifferentiated progenitor blood cells. Domeless (Dome) is a JAK/STAT receptor and a hallmark of the progenitor cells diminished during differentiation (Morin-Poulard et al., 2013). Progenitors in the MZ give rise to mature blood cells that comprise the CZ at the periphery of lymph gland. The PSC is a signaling center for the control of MZ and CZ, and also known to control immune responses (Khadilkar et al., 2017; Mandal et al., 2007; Sinenko et al., 2011). Under normal growing conditions, blood progenitors in the MZ develop into either plasmatocytes or crystal cells at constant rates and do not leave the lymph gland until pupariation (Grigorian et al., 2011). However, upon parasitization or stress conditions, progenitors aberrantly change their fate and precociously differentiate into mature blood cells (Sorrentino et al., 2002). In addition, normal ratios of plasmatocytes, crystal cells and lamellocytes are disrupted by stress signals impinging on pathways involved in cell fate determination of the blood. Both local and systemic factors modify blood cell development during stress responses. Drosophila genetics allows precise investigations into direct genetic interactions of several local or systemic factors linking blood differentiation and stress response. Hypoxia accelerates differentiation of crystal cells by stabilizing Notch through Hif1α (Sima in Drosophila) (Mukherjee et al., 2011). Excessive reactive oxygen species (ROS) leads to precocious differentiation of blood progenitors into mature blood cells (Owusu-Ansah and Banerjee, 2009). Nutrition maintains blood progenitors through the systemic insulin Dilp2 (Benmimoun et al., 2012; Dragojlovic-Munther and Martinez-Agosto, 2012; Shim et al., 2012). Olfactory sensory cue is also important for blood progenitor maintenance through the systemic factor, GABA (Shim et al., 2013). Given that myeloid blood cells are sentinels for stress reactions, further investigation is necessary for unraveling novel molecular mechanisms underlying systemic stress responses of blood cells.

Iron is one of the most abundant transition metal on earth and is an essential element of life. Iron is commonly found in heme-containing enzymes or in iron-sulfur clusters, or as mono- or di-nuclear irons (Ponka, 1997; Rouault and Tong, 2005; Sheftel et al., 2012). Since iron is highly reactive to oxygen, iron has become a key element in aerobic life through evolution. Intracellular and extracellular iron levels are tightly regulated to ensure availability and protection from cellular damages due to high iron concentration (De Domenico et al., 2008). In Drosophila, iron is acquired from diet and its absorption takes place at a specific region of the intestine called the iron region (Fig. 1B). Molecular components involved in iron absorption, storage and allocation are conserved in Drosophila and known to respond well to changes in iron levels (Mandilaras et al., 2013). Ferritin comprises a heteropolymeric complex that is composed of 12 Heavy (H) and 12 Light (L) chain subunits where iron molecules are stored as a primary storage site (Santambrogio et al., 1996). Expression of ferritin genes are controlled during translation by iron regulatory protein (IRP) that binds to an iron responsive element (IRE) at the 5′ UTR of ferritin mRNAs (Georgieva et al., 1999). Drosophila genome encodes three ferritin genes: Ferritin 1 heavy chain homolog (Fer1HCH), Ferritin 2 light chain homolog (Fer2LCH) and Ferritin 3 heavy chain homolog (Fer3HCH). Fer1HCH and Fer2LCH together generate the major ferritin complex while Fer3HCH belongs to the mitochondrial ferritin family abundant in testes (Mandilaras et al., 2013; Missirlis et al., 2006). Loss of Fer1HCH or Fer2LCH results in a wide range of developmental defects in multiple tissues associated with cuticle deposition, dorsal closure, head involution and nervous system development in embryo (Gonzalez-Morales et al., 2015). In larvae, expression of RNAi against Fer1HCH or Fer2LCH in the intestine leads to growth retardation from early instars. In addition, eye-specific RNAi of ferritin genes causes an abnormal compound eye phenotype, and with neuron-specific drivers, neurodegenerative vacuoles are generated due to high levels of iron (Tang and Zhou, 2013). Sequestration of iron during innate immune responses is critical for mounting immune reactions and is evolutionarily conserved from Drosophila to mammals (Drakesmith and Prentice, 2012). However, it is unclear whether iron levels are associated with blood cell development or whether blood cells actively participate in iron homeostasis by expressing iron regulators or both.

In this study, we used genetic approaches to decipher possible roles for iron homeostasis in Drosophila blood cell development. Disrupting iron homeostasis by supplementing iron chelator BPS or loss of Fer1HCH in the intestine causes excessive differentiation of crystal cells in the lymph gland. We discovered that blood cells also express Fer1HCH in addition to the previously identified organs. Correlated to these findings, loss of Fer1HCH in mature blood cells is sufficient to induce crystal cell as well as plasmatocyte differentiation. Thus, this work establishes iron homeostasis as a crucial factor in Drosophila blood development and this key role is modulated by ferritin which is expressed significantly in the blood or in the intestine.

Fly stocks and genetics

The following fly stock lines were used in this study: Oregon R, NP1-gal4 (DGRC112001), Mvl^NP2375-gal4 (DGRC104178), Hml-gal4 (S. Sinenko), Pxn-gal4 (J. Fessler), Dome-Meso-gal4 (U. Banerjee), nSyb-gal4 (BL51635), Hml^LT-gal4 (U. Banerjee), Ppl-gal4 (BL58768), UAS-Hid,Rpr (Nambu JR), Fer1HCH^RNAi (BL60000), Fer2LCH^RNAi (BL44067), IRP-1A^RNAi (BL58117), IRP-1B^RNAi (BL61246), mfrn^RNAi (BL34038), mvl^RNAi (BL55316), Tsf1^RNAi (BL55936), Tsf3^RNAi (BL56015) and Fer1HCH^G188 (DGRC110620). Crosses were kept at 25°C for efficient mating and moved to 29°C to maximize expressions. FAC/BPS feeding experiments were done at 25°C.

Circulating blood cell RT-qPCR

RNA from blood cells were extracted from 30 third instar larvae. First strand cDNA was synthesized using ReverTra ACE^® qPCR RT Kit (Toyobo). Relative quantitative PCR was performed by comparative C_T method using SYBR Green^® Realtime PCR Master Mix (Toyobo) and StepOne Real-time PCR detection thermal cycler (Applied Biosystems). Two sets of primers were used to detect Fer1HCH mRNA. Primer set 1: 5′-CTGCTCCTGTTGGCCGTGGT-3′ (Forward) and 5′-TCCTT CATGTCCACCCAGTCCT-3′ (Reverse). Primer set 2: 5′-ATGGT GAAACTAATTGCTAGC-3′ (Forward) and 5′-TCAGATCGCTG ACTCCCTC-3′ (Reverse). Levels of RpL32 were used to normalize total cDNA.

Prussian blue staining

Guts from wandering third instar larvae were dissected in PBS, transferred to 3.7% formaldehyde/PBS and fixed for 30 min. Samples were then treated with 1% Tween20 in PBS for 15 min. To detect the ferric iron in the gut, tissues were stained with freshly-made Prussian blue staining solution (2% HCl and 3% potassium hexacyanoferrate (II) trihydrate; Sigma P3289) for 45 min in dark condition. After staining, samples were washed in PBS 3 times for 2 min each. Lastly, guts were mounted in VECTASHIELD^® solution and imaging was done with Nikon microscope SMZ18.

FAC, BPS feeding

Parent flies were raised and crossed in standard medium. 3 days later, crossed flies were transferred into conditioned medium supplemented with 5 mM FAC (Ferric Ammonium Citrate; Sigma F5879), 10 mM FAC, 100 μM BPS (Batho-phenanthrolinedisulfonic acid disodium salt; Sigma B1375) or 200 μM BPS. Eggs were reared in iron-supplemented or iron-chelated medium at 25°C and wandering third instar larvae were dissected.

Immunohistochemistry

Wandering third instar larvae were dissected in 1X PBS solution at room temperature. All collected organs were fixed in 3.7% paraformaldehyde for 30 min at room temperature. Washing was done in 0.4% 1X PBS TritonX100 for 10 min and repeated 3 times. Samples were blocked in 10% NGS for 30 min and then treated with primary antibody overnight at 4°C. After another wash, samples were treated with secondary antibody for 3 h. Samples were washed again as described earlier and rewashed in 1x PBS before kept in VECTASHIELD^® (with DAPI). The following primary antibodies were used in this study: Rabbit αPxn (1:2000), Mouse αHnt (1:10 DSHB), Rabbit αcleaved Caspase3 (1:300 Cell signaling #9661) and Rabbit αRab5 (1:1000 AbCam #ab31261). The secondary antibodies used are: FITC or Cy3 conjugated antibodies using a 1:250 dilution (Jackson ImmunoResearch Laboratories).

To generate antisera specific for the Pxn protein, a peptide corresponding to amino acids 326–342 of Pxn sequence (NGGNHPLDSPIDARSNQ) was used as antigen in rabbits (Ab Frontier, Korea).

Quantification of the lymph gland phenotype

Lymph gland staining was quantified as previously described (Shim et al., 2012). In brief: for Hml, Pxn or Dome images, middle one third stacks were compressed and analyzed using ImageJ. For Hnt-positive cells, 3D object counter plug-in was used. Statistical analysis was done using Prism5. Statistical significance was analyzed by Mann-Whitney test for the lymph gland phenotypes and one-way ANOVA, for qRT-PCR. Given natural variations in the blood phenotype, we consider p < 0.01 as significant. Specific genotypes and raw data are indicated in Supplementary Table S1.

Feeding BPS induces crystal cell differentiation in the lymph gland

To understand the importance of iron levels in blood differentiation, we reared larvae on food containing additional iron, ferric ammonium citrate (FAC) or on food containing an iron chelator, bathophenanthroline disulfate (BPS). In normal growing conditions, the lymph gland expresses about 50 crystal cells on average at 25°C (Fig. 1C). Interestingly, feeding 100 μM BPS significantly increases the number of crystal cells while differentiation of plasmatocytes remains unchanged (Fig. 1C). Supplementation of food with 200 μM BPS delays larval growth and reduces size of the lymph gland, which contributes to the decreased number of crystal cells (Fig. 1C). Differentiation of plasmatocytes or crystal cells is not altered by feeding FAC-supplemented food (Fig. 1D). Changes in iron concentrations by supplementing FAC or BPS were confirmed by performing Prussian blue staining in the intestine. Supplementation of FAC enhances Prussian blue staining in the intestine iron region, while feeding BPS reduces the color (Figs. 1E and 1F). In accordance with this, previous study has shown that dietary supplementation of FAC or BPS desirably alters whole-body iron contents (Tang and Zhou, 2013). Collectively, these data suggest that decreased iron levels during larval development disrupts differentiation of blood cells and increases the number of crystal cells in the lymph gland.

Loss of Fer1HCH in the intestine activates crystal cell differentiation

We next investigated whether alterations in iron homeostasis in the intestine give rise to a similar blood phenotype. To address this, we conducted a mini-screen by knocking-down genes involved in iron regulation including Mvl, mfrn, IRP-1A, IRP-1B, Fer1HCH, Fer2LCH, Tsf1 and Tsf3 in the intestine with the use of intestine-specific Gal4, NP1-gal4 (Cronin et al., 2009), and analyzed blood phenotypes in the lymph gland (Figs. 2A and 2B). Expression of RNAi against Fer1HCH significantly induces the differentiation of crystal cells, similar to the phenotype derived by supplementation of BPS (Fig. 2B). However, this genetic background does not modify differentiation of plasmatocytes that are Pxn-positive (Fig. 2C). Knock-down of Fer1HCH in the midgut including iron specific region using Mvl-gal4 reproduces this phenotype (Fig. 2D). Except for Fer1HCH, none of the candidate genes causes abnormal blood differentiation in the lymph gland (Fig. 2B). These data suggest that disruption of iron levels induced by expression of Fer1HCH RNAi in the midgut generates the disproportionate number of crystal cells.

Fer1HCH is expressed in the blood

Drosophila Fer1HCH is one of the major ferritin complex that associates with Fer2LCH and is known to be expressed in various tissues from embryo to adults (Gonzalez-Morales et al., 2015). Although expressions of both subunits of ferritins are predominant in tissues including the intestine, ferritins are also one of the most abundant proteins isolated in the hemolymph (Georgieva et al., 2002; Handke et al., 2013). To assess detailed expression patterns of Fer1HCH in the blood during larval stages, we analyzed expressions of GFP-ferritin trap line, Fer1HCH^G188 (Missirlis et al., 2007). At the wandering third instar, the brain hemispheres except for optic lobes, the ventral nerve cord, the fat body, pericardial cells, Garland cells, intestine and Malpighian tubules express Fer1HCH^G188 in normal growing conditions (Figs. 3A-3B). Interestingly, GFP expression of Fer1HCH^G188 is also found in the lymph gland blood cells (Fig. 3B). Fer1HCH^G188 expression is not restricted to any zone of the lymph gland but shows a sporadic pattern in the entire organ (Fig. 3B). There are two types of expression patterns: few cells exhibit high levels of GFP in the cytosol whereas low GFP⁺ puncta are seen throughout the lymph gland regardless of the zone, similar to the pattern observed in the intestine (Missirlis et al., 2007) (Fig. 3B). These puncta are not Rab5 positive, indicating that Fer1HCH protein is not a part of the endocytic pathway (Fig. 3B). To analyze expression patterns of Fer1HCH^G188 in the lymph gland in detail, we co-stained Fer1HCH^G188 with Dome, Pxn or Hnt. Few Fer1HCH^G188-GFP are expressed in Dome⁺ progenitor cells, but not all lymph gland exhibits Dome⁺Fer1HCH^G188-GFP⁺ blood cells (Fig. 3B). This is also true for crystal cells: some Hnt⁺ crystal cells express low Fer1HCH ^G188-GFP while majority of crystal cells do not show Fer1HCH ^G188-GFP. Nearly all Pxn⁺ mature blood cells display either high or low Fer1HCH^G188-GFP expression (Fig. 3B). Circulating blood cells also exhibit Fer1HCH^G188 expression, again, most of which are co-stained with Pxn (Fig. 3C).

Expression of ferritin in the intestine is a consequence of an iron availability. Feeding FAC supplemented food raises Fer1HCH^G188 expressions as well as Fer1HCH protein levels in the intestine (Missirlis et al., 2007). Yet, it is unclear whether Fer1HCH^G188 in the lymph gland reacts to different levels of systemic iron as in the intestine. We found that supplementation of 5 mM FAC upregulates Fer1HCH^G188 GFP in the lymph gland while feeding 100 μM BPS does not alter the expression (Fig. 3D). Moreover, endogenous expression of Fer1HCH in the blood is verified by qRT-PCR which showed that Fer1HCH RNAi in the blood dramatically reduces normal mRNA levels of Fer1HCH (Fig. 3E). Taken together, our findings indicate that Fer1HCH is expressed in the blood population in addition to previously identified organs and readily responds to systemic iron concentrations.

Fer1HCH plays a key role in blood development

Novel function of Fer1HCH in the blood is further assessed by knocking-down its expression in the lymph gland. Based on its sporadic pattern in the lymph gland, we analyzed functions of Fer1HCH in each zone by expressing RNAi against Fer1HCH with zone-specific drivers. Mature blood cell-specific knock-down of Fer1HCH using Hml-gal4 or Pxn-gal4 leads to a dramatic increase in the number of crystal cells when compared to wild types (Figs. 4A and 4B). Different from the intestine-specific knock-down, these genetic backgrounds also aggravate plasmatocyte differentiation and cause a marked increase of the CZ (Figs. 4A and 4B). These phenotypes are recapitulated by expressing RNAi against Fer1HCH in Hml⁺ lineage-traced population using Hml-gal4 lineage traced line (Fig. 4C). However, crystal cell differentiation is reduced by knocking-down the expression of Fer1HCH in progenitor blood cells with the use of Dome-Meso-gal4 (Fig. 4D). Brain-specific knock-down or fat body-specific RNAi against Fer1HCH does not impinge on blood development in the lymph gland (Fig. 4E).

Previous studies indicated that expression of pro-apoptotic genes, Hid and Rpr in the CZ causes loss of blood progenitors and subsequent differentiation of mature blood cells (Mondal et al., 2011). Similar phenotype is observed when Fer1HCH^RNAi is expressed in mature blood cells (Figs. 4A-4C). We reasoned that this phenotype generated by Fer1HCH^RNAi is due to programmed cell death. However, Fer1HCH RNAi in mature blood cells do not accumulate Caspase3 expression in the lymph gland (Fig. 4F), implying that inordinate expansion of the CZ is not a consequence of regular cell death. Thus, these findings establish that Fer1HCH in the blood maintains iron homeostasis and loss of iron control gives rise to abnormal blood cell differentiation mimicking stress responses.

This research identified iron homeostasis as one of the key factors that govern blood differentiation of Drosophila larva through expressions of ferritins in the intestine and the blood. Feeding 100 μM BPS is sufficient to cause extensive differentiation of the crystal cell in the lymph gland without altering larval development. BPS-mediated blood response is repeated when Fer1HCH RNAi is expressed in the intestine iron region while RNAi against other iron-related genes does not modify the crystal cell number. Consistent with these findings, previous study has shown that expression of RNAi against Fer1HCH in the intestine suppresses whole-body iron levels (Tang and Zhou, 2013). This indicates that both dietary BPS and Fer1HCH RNAi diminishes systemic iron levels leading to the abnormal crystal cell differentiation. Crystal cells exhibit expression of Prophenoloxidase (ProPO) essential for activation of melanization and immune responses (Rizki et al., 1985). Therefore, understanding physiological roles of crystal cells in iron-mediated immunity will be interesting.

Fer1HCH interacts with Fer2LCH in the generation of iron-storage cluster (Mandilaras et al., 2013) and expression of these genes are strictly regulated by iron levels through IRPs (Georgieva et al., 1999). However, we could not detect any blood phenotype by inhibiting Fer2LCH in the intestine. Similarly, recent study has shown that loss of Fer2HCH in the intestine does not suppress the whole-body iron levels (Tang and Zhou, 2013). In the brain, Fer1HCH RNAi in specific neurons do not alter the circadian rhythms while Fer2LCH does (Mandilaras and Missirlis, 2012). These notions together imply that Fer1HCH and Fer2LCH could take part in distinctive pathways in addition to the core function of iron storage.

Iron sequestration is a critical response in innate immunity as acquisition of iron is an important element for pathogenic proliferation (Drakesmith and Prentice, 2012). In Drosophila, expressions of iron regulatory proteins are increased upon bacterial or fungal infections (Sulieman et al., 2004; Yoshiga et al., 1999). A similar response is observed when flies are infected by Wolbachia (Brownlie et al., 2009; Kremer et al., 2009). Our results identified a novel role for iron in blood development that is closely related to the innate immune reaction. As mentioned above, crystal cells react to iron chelation. However, it is unclear how the whole-body iron levels are connected to the differentiation of crystal cells. It is possible that loss of iron directly affects signaling pathways including Serrate-Notch or Lozenge, or modifies unknown systemic factors that merge onto the crystal cell differentiation mechanism.

Our study first demonstrates that functional ferritin is expressed in the lymph gland and loss of Fer1HCH leads to abnormal blood phenotypes. Although Fer1HCH RNAi in the blood or intestine alters blood development, these two phenotypes are not precisely identical. Loss of Fer1HCH in the intestine only gives rise to excessive crystal cells in the lymph gland without disrupting the ratio of the CZ. Fer1HCH RNAi in the CZ promotes differentiation of both crystal cells and plasmatocytes and consequently reduces the MZ. When Fer1HCH is knocked-down in the MZ, moderate decrease in the MZ ratio and the number of crystal cells are observed. These phenotypes suggest that Fer1HCH in the blood may play different roles in each zone. It is interesting to note that Fer1HCH RNAi in the blood substantially increases the CZ without enhancing active Caspase3 expression, implying an alternative pathway that aggravates the CZ via the function of Fer1HCH. Recent studies have shown that loss of Fer1HCH elevates cellular iron concentration leading to ferroptosis in larval discs. This phenomenon is separable from apoptosis, necrosis or autophagy (Dixon and Stockwell, 2014; Dixon et al., 2012; Tang and Zhou, 2013). Ferroptosis is reported recently in mammalian cancer cells and fibroblasts (Dixon and Stockwell, 2014). Precise roles and detailed mechanisms underlying this phenomenon is unclear. Drosophila blood system will be a useful tool to uncover mechanistic details and for identification of novel factors involved in ferroptosis.

Ferritins in the midgut iron region are responsible for the control of systemic iron levels (Tang and Zhou, 2013). However, it is uncertain whether ferritins expressed in other tissues are equally accountable to the iron homeostasis. Different from mammals, significant amounts of ferritin proteins are deposited into the circulation of Drosophila (Handke et al., 2013). However, origins of these circulating ferritins are still elusive. Surprisingly, we observed that larvae expressing Fer1HCH RNAi in the blood experience growth retardation when fed BPS (data not shown). This observation implies a considerable contribution of blood ferritins in the systemic iron homeostasis eliciting future studies on the role of blood in iron regulation. Thus, our study uncovers iron homeostasis as a crucial player in blood development, and suggests an interesting possibility that ferritin in the blood may take part in the systemic iron control.