Ferritin overexpression in Drosophila glia leads to iron deposition in the optic lobes and late-onset behavioural defects

Abstract

Cellular and organismal iron storage depends on the function of the ferritin protein complex in insects and mammals alike. In the central nervous system of insects, the distribution and relevance of ferritin remains unclear, though ferritin has been implicated in Drosophila models of Alzheimers’ and Parkinsons’ disease and in Aluminum-induced neurodegeneration. Here we show that transgene-derived expression of ferritin subunits in glial cells of Drosophila melanogaster causes a late-onset behavioural decline, characterized by loss of circadian rhythms in constant darkness and impairment of elicited locomotor responses. Anatomical analysis of the affected brains revealed crystalline inclusions of iron-loaded ferritin in a subpopulation of glial cells but not significant neurodegeneration. Although transgene-induced glial ferritin expression was well tolerated throughout development and in young flies, it turned disadvantageous at older age. The flies we characterize in this report contribute to the study of ferritin in the Drosophila brain and can be used to assess the contribution of glial iron metabolism in neurodegenerative models of disease.

Introduction

Iron is implicated in the pathologies of disorders such as Alzheimer’s and Parkinson’s disease potentially via the function of numerous iron-containing enzymes, but also in its labile form by catalyzing the formation of reactive oxygen species and the propagation of lipid peroxidation reactions (Zecca et al., 2004). Furthermore, defects in several human iron metabolism genes lead to heritable neurodegenerative pathologies (Gregory et al., 2009; McNeill et al., 2008; Rouault and Tong, 2008). Mutations in ferritin light chain gene cause a dominant adult-onset basal ganglia disease, termed neuroferritinopathy (Curtis et al., 2001; Levi et al., 2005; Vidal et al., 2008). In addition, mutation in the iron responsive element (Hentze et al., 1987) of the ferritin light chain gene lead to crystalline inclusions of ferritin in the human lens causing cataracts (Beaumont et al., 1995). Neuronal ferritin overexpression has been associated with beneficial effects attributed to either protection from metal-induced oxidative stress (Zhu et al., 2009), or participation in intracellular trafficking by modulating endogenous opioid signalling (Sengupta et al., 2009), but also with detrimental phenotypes in mammalian models of neurodegeneration (Kaur et al., 2009; LaVaute et al., 2001). Maintenance of metal homeostasis in the brain is essential for neuronal health and a critical component in the development of therapeutic strategies against neurodegenerative disease (Crouch et al., 2009; Ghosh et al., 2008; Madsen and Gitlin, 2007; Molina-Holgado et al., 2007).

Numerous Drosophila models of human neurodegenerative disorders were developed in recent years because they can be subjected to genetic (Clark et al., 2006; Fernandez-Funez et al., 2000; Mollereau, 2009; Park et al., 2006) and pharmacological analysis (Apostol et al., 2003; Outeiro et al., 2007; Rana et al.; Tain et al., 2009). Oxidative stress was shown to mediate toxicity in fly models of Parkinson’s disease (Faust et al., 2009; Meulener et al., 2006; Wang et al., 2006; Whitworth et al., 2005; Yang et al., 2005), Alzheimer’s disease (Dias-Santagata et al., 2007; Rival et al., 2009), lysosomal storage disease (Sweeney, 2008; Venkatachalam et al., 2008), and in the fly model of Friedreich’s ataxia, a condition associated with mitochondrial iron overload (Anderson et al., 2008; Llorens et al., 2007; Navarro et al., 2010; Runko et al., 2008). The fly model of Pantothenate kinase associated neurodegeneration (Bosveld et al., 2008; Rana et al., 2010; Wu et al., 2009), a human disorder resulting in iron accumulation in the globus pallidus (Zhou et al., 2001), has not yet been used to address why and how iron accumulates predominantly in specific brain regions, but was shown instead to accumulate high amounts of zinc (Gutierrez et al., 2010).

Indeed, the effect of transition metals in Drosophila models of neurodegeneration has received little attention, despite numerous indications implicating ferritin in neuronal pathology and oxidative stress response. For example, proteomic analysis of adult heads from a Drosophila model of Parkinson’s disease identified Ferritin Light Chain Homologue-2 (Fer2LCH) as 1 of 5 proteins upregulated relative to controls (Xun et al., 2008). Fer2LCH was recently shown to accumulate with age in flies raised at 28 C (Robinson et al., 2010). Similarly, the Fer2LCH gene was induced in photoreceptors under endoplasmic reticulum stress and protected them from retinal degeneration (Mendes et al., 2009). Furthermore, when flies were exposed to conditions of hyperoxia that caused marked degeneration in dopaminergic neurons, both Ferritin Heavy Chain Homologue-1 (Fer1HCH) and Fer2LCH were induced (Girardot et al., 2004; Gruenewald et al., 2009). In the only functional studies to date, ferritin suppressed β-amyloid toxicity in Drosophila models of Alzheimer’s disease (Rival et al., 2009) and it also suppressed Aluminum toxicity (Wu et al., 2010). Specifically, transgene-derived neuronal overexpression of either Fer1HCH or Fer2LCH alone was sufficient to rescue the reduced survival of flies expressing a mutant form of the β-amyloid peptide (Arctic Aβ_1–42; Rival et al., 2009). In contrast, transgene-derived ubiquitous overexpression of either Fer1HCH or Fer2LCH was not sufficient to rescue from Aluminum-based toxicity, but simultaneous overexpression of both subunits did (Wu et al., 2010).

Recent findings highlight the role of glia in a fly model of Friedreich’s ataxia (Navarro et al., 2010) and in the control of circadian behaviour (Jackson, 2010). Null mutants of the ebony gene exhibit arrhythmic patterns of locomotor activity because of elimination of a rhythmic glia-specific enzyme (Suh and Jackson, 2007). The proximity of Ebony-containing glial cells to aminergic neurons of several types (histamine, serotonin, dopamine) within the fly optic lobe and central nervous system suggests that the rhythm phenotype of mutants may be due to defective amine recycling. Consistent with a role for dopamine in the phenotype, the ebony¹ mutation suppresses the hyperactivity associated with a Drosophila dopamine transporter allele (Jackson, 2010; Suh and Jackson, 2007). To evaluate if Drosophila ferritin has a role in glia and to assess a potential role of iron in normal circadian behaviour, we interfered with normal ferritin expression in this cell type. Since ferritin is a heteropolymer of two subunits and resides in the secretory pathway of cells and in the hemolymph (Georgieva et al., 2002; Missirlis et al., 2007), we used subunit-specific antibodies to monitor each ferritin subunit individually in the adult brain. We tested flies overexpressing ferritin subunits for behavioural alterations in elicited locomotion at different times during their lifespan and checked their ability to maintain circadian rhythms in the absence of environmental inputs. We describe alterations in brain iron homeostasis, accompanied by a late-onset behavioural syndrome in these flies.

Materials and Methods

Fly maintenance and Western blotting

The pan-glial drivers Repo-Gal4/TM3, Sb and Nrv2-Gal4 were obtained from the Bloomington stock centre. Transgenic stocks UAS-Fer1HCH and UAS-Fer2LCH have been described and characterized in detail elsewhere (Missirlis et al., 2007). SDS PAGE followed by Western blotting was performed under non-reducing conditions to detect assembled holoferritin complexes or in the presence of β-Mercaptoethanol to detect individual ferritin subunits as described previously (Missirlis et al., 2006; Missirlis et al., 2007).

Iron staining

For detection of ferric iron, paraffin embedded sections were incubated for 45 minutes in the dark with Prussian blue stain (10% K₄Fe(CN)_6: 20% HCl, 1:1) prior to de-paraffinization. Three washes with water followed and staining was enhanced using DAB following standard procedures. Then sections were de-paraffinized using ascending gradients of ethanol to xylene and preparations were mounted in Glycergel mounting medium (DAKO) and imaged on a Nikon microscope.

Immunohistochemistry and confocal imaging

Immunohistochemistry on paraffin sections was performed as previously described (Kosmidis et al., 2010). Rabbit anti-Fer1HCH (1:100) or rabbit anti-Fer2LCH (1:50) primary antibody incubations were followed by fluorescent secondary antibodies at 1:1000 (Invitrogen) and mounted in DAKO Fluorescent Mounting medium (antibodies described in Missirlis et al., 2006; Missirlis et al., 2007). Sections incubated without the primary antibodies were used to control for autofluorescence. Imaging was performed on a Zeiss confocal microscope.

Electron Microscopy

Adult brains were prepared, cut, and stained as described (Kretzschmar et al., 1997). Ultrathin Epon plastic sections were post stained with 2% uranyl acetate, followed by Reynolds' lead citrate and stabilized for transmission electron microscopy by carbon coating. Examination was done with a Zeiss EM10C/VR (Oberkochen, Germany) electron microscope at 40–80 kV.

Behavioural assays

Male adult flies collected 0–2 days after eclosion were kept throughout their lifetime at 25 C in groups of 20 individuals per vial. For elicited-escape response tests flies were adapted for at least 10 min in an environment of 25 C and 70–80% humidity illuminated by red light. They were then placed in 14mL polystyrene Falcon tubes individually and each fly was vortexed at the highest speed for 2–3 sec and was tested twice in negative geotaxis and horizontal escape assays (Wittmann et al., 2001). A minimum 20 flies per genotype were tested for each experimental datapoint and the data analyzed parametrically for statistical significance using the JMP software. Locomotor activity rhythms were recorded automatically using the Drosophila Activity Monitoring system (Trikinetics, Waltham, MA). Flies were raised in light-dark cycles (12 hr: 12 hr) at a constant temperature of 25 C and flipped to fresh food every 3–4 days during ageing to avoid microbial infections. They were then released to constant darkness for the measurements. To quantify the phenotypes we used the rhythmicity statistic value of 1 as the threshold to classify flies as rhythmic (Levine et al., 2002) and present the average period and the average R.S. value for the pool of flies deemed rhythmic. Life span determinations have been described previously (Missirlis et al., 2003).

Results and Discussion

Altered brain iron homeostasis causes a late-onset behavioural decline

To investigate whether increasing levels of ferritin in Drosophila resulted in functional compromise of the CNS revealed as aberrant behaviours, we assessed the consequence of ferritin overexpression in behavioural tests throughout the life span of Drosophila melanogaster (Figure 1). Ferritin transgenes were driven in all adult brain glia with the Repo-Gal4 driver. Two different methods were used to quantify the deficits in the fly’s innate escape responses. We measured the elicited escape responses of control and transgenic flies in the horizontal (walking) and vertical (climbing) axes. Statistically significant differences in escape response between ferritin overexpression and control flies (Figure 1A, B) were not uncovered in the first four weeks. This indicates that experimental manipulation of ferritin levels in glia does not detectably alter this core behaviour for a significant portion of the life span. However, flies simultaneously accumulating both ferritin subunits in glia exhibited an abrupt and pronounced deficit in escape responses in both horizontal and vertical directions at five and six weeks of age (Figure 1A, B). Ferritin overexpression by means of the Gal4/UAS binary expression system was shown previously to require induction of both ferritin subunits to achieve quantifiable accumulation of assembled holoferritin polymers in adult flies (Missirlis et al., 2007), which could explain why the phenotype strongly manifested only when both subunits were induced.

Figure 1.

Age dependent behavioural deficits caused by ferritin overexpression in glia. A, B) Elicited escape response of flies was assessed on a weekly basis. The first bar on each graph corresponds to the first day of the first week and so on for six weeks. Control flies (carrying the driver transgene only, i.e. Repo-Gal4/+) and flies expressing either the heavy or the light chain alone did not exhibit statistical differences in their escape response until the fifth week. However at 6 weeks post-eclosion their response was significantly different than that of all younger flies (Tukey-Kramer α=0.01). In contrast, flies overexpressing both ferritin subunits in glia showed a highly significant impairment at 5 and 6 weeks of age (Tukey-Kramer α=0.001). The delay in the vertical direction had an even earlier onset at 4 weeks (Tukey-Kramer α=0.01). Planned comparisons of the vertical and horzontal escape response between Repo-Gal4/+ and flies expressing ferritin subunits at 6 weeks reveled a highly significant difference (p<0.001) only with flies expressing simultaneously both subunits. C) Age dependent loss of circadian rhythmicity caused by ferritin overexpressin in glia. Locomotor activity was monitored automatically during the third and sixth weeks of ageing male flies. Five representative outputs spanning six days were averaged and plotted (output days 2, 3, 4, 5 are duplicated in plots to help visualize the rhythmic pattern). A significant proportion of transgenic flies with simultaneous overexpression of both ferritin subunits in glia resulted in a arrhythmic behavioural phenotype during the sixth week of adult life span. This phenotype is less common during the third week of life span in sibling flies with the same genotype and in control flies carrying the transgenes in absence of a driver (UAS-Fer1HCH, UAS-Fer2LCH/TM3, Sb). The quantification of results is presented in Table 1.

As glia have been suggested to function in the fly circadian clock by interacting with neuro-aminergic pathways (Jackson, 2010), we tested whether overexpression of ferritin could impact brain iron homeostasis causing circadian disturbances in Drosophila. We monitored activity patterns in six-week old flies in complete darkness and quantified individual outputs (Figure 1C and Table 1). Quantification of rhythmic patterns (Levine et al., 2002) showed that 57 % of six-week old individuals overexpressing both ferritin subunits in glia remained rhythmic in constant darkness (Table 1), albeit with a reduced value for rhythmicity statistic (1.7 ± 0.7), suggesting behavioural disturbances even in the flies categorized as rhythmic. 85% of control flies carrying the transgenes without the driver were rhythmic with typical R.S. values greater than 2.5 (R.S.=1 indicates rhythmic behaviour at the 95% confidence interval; see Table 1). Three-week old flies overexpressing both chains in glia showed robust rhythms in 69% of the individuals tested. We repeated these experiments with an independent Gal4-driver specific for glia cells, Nrv2-Gal4 (Pereanu et al., 2005). Consistently, 63% of 6-week old Nrv2-Gal4/UAS-Fer1HCH, UAS-Fer2LCH flies were rhythmic compared to 91% for similar aged Nrv2-Gal4/+ control flies (Table 1). Flies overexpressing Fer1HCH in glia were less affected (75% and 80% rhythmicity when expression was driven with Repo-Gal4 or Nrv2-Gal4, respectively) but flies overexpressing Fer2LCH with Repo-Gal4 showed a circadian phenotype (54% rhythmic and R.S. = 1.9 ± 1.2). Overall, the results were consistent with the milder defects described in the elicited escape responses of individuals from these genotypes. The age-dependence of the phenotype may be related to the gradual decline in cellular degradation pathways, shown in Drosophila for both the proteasomal and the autophagic/lysosomal degradative pathways (Girardot et al., 2006; Pandey et al., 2007; Tonoki et al., 2009; Vernace et al., 2007), to a reduced output of mitochondrial activity due to oxidative stress (Hao et al., 2010), or to both physiological functions being affected in parallel. Disturbance of circadian rhythmicity suggests that these flies can be used to study the relevance of brain iron trafficking to the function of the circadian clock.

Table 1.

Effects of ferritin overexpression on rhythmic behaviour of flies kept in constant darkness. Longer than 24h periods observed in the genotypes are attributed to the per^SLIH allele (Hamblen et al., 1998), present in both transgenic Gal4 stocks used.

Genotype	Age	N	% Rhythmic	Period	R.S.
TM3, Sb/ UAS-Fer1HCH, UAS-Fer2LCH	3 wk	16	93	26.2 ± 1.4	2.7 ± 0.7
Repo-Gal4/UAS-Fer1HCH, UAS-Fer2LCH	3 wk	16	69	26.3 ± 1.9	2.6 ± 1.1
TM3, Sb/ UAS-Fer1HCH, UAS-Fer2LCH	6 wk	13	84	26.2 ± 2.5	2.4 ± 1.0
Repo-Gal4/UAS-Fer1HCH, UAS-Fer2LCH	6 wk	28	57	25.7 ± 2.6	1.7 ± 0.7
Nrv2-Gal4/UAS-Fer1HCH, UAS-Fer2LCH	6 wk	16	63	26.1 ± 2.6	2.7 ± 1.1
Nrv2-Gal4/+	6 wk	11	91	26.9 ± 2.1	2.5 ± 1.0
TM3, Sb/UAS-Fer1HCH	6 wk	14	85	26.3 ± 0.9	2.6 ± 0.9
Repo-Gal4/UAS-Fer1HCH	6 wk	32	75	26.4 ± 1.9	2.3 ± 1.3
Nrv2-Gal4/UAS-Fer1HCH	6 wk	15	80	26.8 ± 2.1	2.1 ± 0.8
TM3, Sb/UAS-Fer2LCH	6 wk	14	85	26.7 ± 0.4	3.1 ± 1.1
Repo-Gal4/UAS-Fer2LCH	6 wk	24	54	24.9 ± 2.7	1.9 ± 1.2
Nrv2-Gal4/UAS-Fer2LCH	6 wk	15	80	26.1 ± 2.7	2.1 ± 0.8

Changes in the distribution of ferritin subunits upon formation of assembled ferritin

We used antibodies capable of distinguishing individual ferritin subunits (Missirlis et al., 2007) to follow ferritin accumulation in glial cells when expression was induced with the pan-glial Repo-Gal4 from the respective transgenes. Adult brains were dissected and Western blots performed following SDS-PAGE (Figure 2A,B). Under reducing conditions, ferritin subunits run in monomeric form, and our blots indicated that each subunit accumulated when transgenic expression was driven either alone or in combination with the alternate subunit (Figure 2A). Non-reducing SDS-PAGE was used to visualize assembled ferritin using an antibody against Fer2LCH (Figure 2B). This assay confirmed the requirement of both chains to form stable assembled ferritin as previously proposed (Missirlis et al., 2007). Induction of assembled ferritin was only observed when both Fer1HCH and Fer2LCH transgenes were simultaneously activated by Repo-Gal4, but smaller aggregates of Fer2LCH were also identifiable in dissected brains from Repo-Gal4/UAS-Fer2LCH flies. These results suggest that overexpression of a single ferritin subunit in Drosophila glia does not lead to a significant increase in the formation of additional functional ferritin molecules.

Figure 2.

Ferritin accumulation differs when single subunits are expressed in all glia, or if both subunits are expressed simultaneously. Ferritin was overexpressed by means of the UAS/Gal4 system. A) Western blots following reducing SDS-PAGE (ferritin chains dissociate) probed with antibodies against both subunits and syntaxin as a loading control. Note that Fer2LCH (which has higher molecular weight in Drosophila (Georgieva et al., 2002)) accumulates when overexpressed either singly or in combination with Fer1HCH. Similarly, Fer1HCH accumulates when overexpressed either singly or in combination with Fer2LCH. B) Gel electrophoresis under non-reducing conditions maintained the integrity of ferritin complexes. The Western blot was probed with an antibody against the Fer2LCH subunit. Multiple Fer2LCH complexes were detected in brains overexpressing the single gene. Co-expression of both ferritin genes resulted in significant accumulation of assembled ferritin and absence of low molecular weight bands. C–J) Ferritin immunohistochemistry was performed in sections of the adult optic lobes of flies overexpressing Fer1HCH and Fer2LCH in glia. Flies were sectioned in sagittal orientation; optic retina, lamina and medulla are shown. Note large accumulations of Fer1HCH in glial cells in UAS-Fer1HCH/+; Repo-Gal4/+ flies (E) but not of Fer2LCH in the same genotype (F) and, conversely, Fer2LCH (but not Fer1HCH) accumulations in UAS-Fer2LCH/+; Repo-Gal4/+ flies (G, H). Dramatic overexpression of both subunits was seen when both subunits are co-expressed (UAS-Fer1HCH, UAS-Fer2LCH/+; Repo-Gal4/+) but holoferritin was primarily concentrated in the lamina (I, J).

We then used the antibodies in immunohistochemical experiments to monitor changes in ferritin accumulation within the central brain of the same flies. We present images from sagittal sections through the optic lobes, where the most striking staining patterns emerged (Figure 2C–J). Repo-Gal4 induction of single ferritin subunits resulted in independent accumulation of the respective subunits in cell bodies of all glial subtypes present in the optic lobes (Figure 5K for Fer1HCH and 5N for Fer2LCH). Upon simultaneous expression of both subunits with Repo-Gal4, Fer1HCH and Fer2LCH accumulation was greatly concentrated in satellite, epithelial, marginal and medulla glia but not in neuropil medulla glia or glia from the lobula complex (Figure 2I, J; for identification of glial subtypes see Chotard and Salecker, 2007). These results indicated that glia in the lamina and the frontal region of the medulla could accumulate large amounts of assembled holoferritin, whereas glia in other parts of the optic system and the central brain did not accumulate assembled holoferritin to the same degree. They suggest that some qualifications may be required when interpreting highly interesting and robust phenotypes from experiments using single subunit overexpression (Rival et al., 2009).

Ferritin-iron inclusions in the brain optic lobes

We next asked whether ferritin overexpression in glia was associated with morphological abnormalities of the adult central nervous system or alterations in brain iron accumulation. For this purpose, paraffin-embedded sections of transgenic adult brains from ferritin overexpressors were stained for iron (Figure 3A–H). Strong staining for iron was specifically detected in the lamina of optic lobes from Repo-Gal4 driven UAS-Fer1HCH, UAS-Fer2LCH (Figure 3G, H). In contrast, iron accumulation was neither detected in control flies (Figure 3A, B), nor in flies with single subunit overexpression (Figure 3C–F). We therefore suggest that this increased iron staining is the result of iron-loaded ferritin accumulation in the lamina. Plastic embedded sections of adult brains were also prepared to assess how increased ferritin and iron may impact brain morphology (Figure 3I–L). Interestingly, inclusions were clearly visible with toluidine staining, but the overall brain neuroanatomy of the affected flies was well preserved even at six weeks of age (Figure 3L).

Figure 3.

Ferritin iron inclusions accumulate specifically in glial cells of the lamina following overexpression of both ferritin subunits in all glia. A–H) Iron stainings were performed on paraffin-embedded sections of 40-day old flies. Close up views from laminas are juxtaposed to the entire optic lobes. Note that iron accumulation is seen only in laminas of flies overexpressing both ferritin subunits using the Repo-Gal4 driver line. Iron is detected in a punctate pattern, but larger inclusions are also visible (arrows in G, H). I–L) Plastic embedded sections from 38-day old fly heads stained with toluidine blue. Abnormal inclusions (arrow in L) are seen only in the lamina and only in flies overexpressing both ferritin subunits in glia (Repo-Gal4/UAS-Fer1HCH, UAS-Fer2LCH).

The inclusions were also identifiable at higher resolution using electron microscopy due to their electron dense, rodular shaped nature (Figure 4). They only appeared in the lamina from flies simultaneously overexpressing both ferritin subunits in all glia (Figure 4A, B). Closer examination revealed that the ferritin-iron inclusions have a crystal-like (i.e. symmetric) morphology (Figure 4C, inset) and are surrounded by an intracellular membrane, indicating that they are likely trapped in the secretory system of the lamina glia (Figure 4C, black arrows). Electron density of these inclusions is also suggestive of iron accumulation and indeed similar structures have been observed before in iron-loaded cells of the Calpodes ethlius intestine (Locke and Leung, 1984). The inclusions described here are different from those found in patients with hyperferritinemia-cataract syndrome, in that they are intracellular, iron-loaded and composed from both ferritin subunits. In contrast, ferritin crystals in patients are secreted in the eye lens, are iron-poor and form from ferritin light chains only (Beaumont et al., 1995). Nevertheless, a screen for genetic or pharmacological suppressors of crystal formation in flies could also test if the underlying biology leading to ferritin inclusions would be relevant to the human condition.

Figure 4.

Electron micrographs of the iron-ferritin inclusions (white arrows). Inclusions were only found in the lamina of optic lobes. Transverse (A) and longitudinal (B) views of the rodlike shaped inclusions at 16,000X and 1,000X magnifications are shown, respectively. The electron-dense material showed symmetric properties consistent with a putative crystalline structure and inclusions are found within the endomembrane system (black arrows point to membranes in C; image at 80,000X). D) No significant differences were noted in total or mean life span of flies with targeted overexpression of ferritin subunit(s) in all glia.

Overexpression of both subunits in glia lead to enhanced assembly of ferritin evident both in western blots (Figure 2A, B), immunohistochemical sections (Figure 2I, J) iron and toluidine blue staining of brain sections (Figure 3G, H, L), and electron micrographs (Figure 4). Lack of apparent neurodegeneration in these flies was corroborated by lifespan determinations showing no significant differences between the different genotypes (Figure 4D). Further confirmation of the behavioural and anatomical phenotypes presented so far was obtained by repeating experiments with a different glial Gal4 driver, Nrv2-Gal4, and also by using independent UAS-Fer1HCH and UAS-Fer2LCH transgenes and finding similar results as presented above (data not shown). Strikingly, using either glial Gal4 driver, increased ferritin concentrations were not distributed evenly among the entire glial population, but accumulated in the optic lobes and more precisely in the outermost layer of the medulla and in the lamina, where they formed intracellular inclusions. We interpret this finding to mean either that degradative pathways operate with different efficiencies in the various glial subtypes or that secretion of ferritin is somehow affected in the lamina glia. It is interesting to note that iron trafficking has been described in mammalian retina (Hadziahmetovic et al.; Hahn et al., 2004), whereas the pathways securing an adequate supply of iron to the Drosophila retina, and whether ferritin has a role to play (Mehta et al., 2009), have not been elucidated to date.

A link between brain iron homeostasis and the circadian clock in the Drosophila model

The accumulation of iron-loaded ferritin in glia when both ferritin subunits were overexpressed simultaneously suggests that mis-regulation of brain iron homeostasis may result in failure to maintain circadian rhythms and an overall behavioural decline at late age. The anatomical location of the ferritin inclusions raises the intriguing possibility that the function of the Hofbauer-Buchner eyelet, previously shown to participate in the synchronization and entrainment of natural rhythms (Veleri et al., 2007), might be affected as a consequence of ferritin aggregation. Indeed, the eyelet sends projections through the chiasma between the lamina and medulla (Helfrich-Förster et al., 2002) and might be physically or metabolically constrained by the presence of crystalline formations of iron-loaded ferritin in surrounding glia. Experiments in this study support a previously suggested role of glia in the maintenance of the circadian clock (Jackson, 2010). We propose that Drosophila is a useful model to elucidate interactions between iron homeostasis and the endogenous clock.

A poorly understood connection between brain iron metabolism and the dopaminergic neural network is sometimes invoked in explanations of the Restless legs syndrome (RLS) pathology (Clemens et al., 2006; Dowling et al., 2011; Earley et al., 2006; Patrick, 2007; Salas et al., 2010). RLS is a condition characterized by circadian timing of movement disorder. Some patients with RLS respond well to dietary iron supplementation (Patrick, 2007; Wang et al., 2009). Tyrosine hydroxylase, an iron-dependent enzyme in the dopamine biosynthetic pathway, may be induced despite (or as a consequence of) systemic iron deficiency (Connor et al., 2009), a finding that remains poorly understood. As iron deficiency is a serious risk factor for developing RLS (Mizuno et al., 2005, Salas et al., 2010), it will be interesting to investigate whether the iron accumulation in glial ferritin inclusions, upon ferritin overexpression, may deplete bioavailable iron for other cellular functions. Future experiments aimed to uncover links between ferritin, the circadian clock and neuroaminergic outputs in the fly brain will hopefully enhance our understanding of the role of iron in the numerous human neurological disorders it has been already implicated.