Drospondin, a glial glycoprotein with similarities to human Reelin/F-spondin, contributes to Drosophila brain development and function

Significance

We recently unveiled a signaling cascade in Drosophila neurons that is activated by an important vertebrate signaling molecule, Reelin. This was rather unexpected since, until recently, it was considered that Reelin is a glycoprotein that only exists in vertebrates. One of the open questions is to identify the endogenous molecule that would be responsible for the Reelin-induced effects in Drosophila neurons. Here, we describe that the protein product of the uncharacterized Drosophila CG17739 gene, which we named Drospondin, shares sequence homology with vertebrate F-spondin and Reelin. By using different approaches, we show that Drospondin contributes to brain development and function, as Reelin does in vertebrates. Therefore, Drospondin is a previously overlooked Drosophila protein with homologous functions to vertebrate Reelin/F-spondin.

Abstract

Reelin is a secreted glycoprotein with roles in the development of the mammalian neocortex, hippocampus, and cerebellum. This vertebrate signaling molecule also contributes to adult brain function. Mammalian Reelin increases the complexity of Drosophila Mushroom Body (MB) neurites, an effect mediated by LpR1 and LpR2, the orthologs of mammalian Reelin receptors. Paradoxically, to date, no Reelin ortholog has been described in Drosophila. Here, we report that the protein product of the uncharacterized Drosophila CG17739 gene, which we named Drospondin, shares sequence homology with vertebrate F-spondin and Reelin. We show that Drospondin is expressed in glial cells and is crucial for MB development. Our results also show that Drospondin genetically interacts with LpRs and that human Reelin rescues neuronal and brain structural defects in Drospondin-deficient flies. Furthermore, Drospondin-deficient flies exhibit altered sleep, locomotion, and social behaviors. Our results reveal that Drospondin is a Drosophila protein with similar functions to mammalian Reelin/F-spondin, that has an essential role in brain development and function, the impairment of which has profound functional consequences for the animal.

Diverse signaling molecules play critical roles in several stages of neuronal development. Importantly, mutations in these molecules or the cellular components associated with their signaling contribute to anatomical or functional alterations that might underlie neurodevelopmental or neurodegenerative disorders.

Reelin is an extracellular glycoprotein with crucial functions in the vertebrate central nervous system (CNS), controlling the radial migration and differentiation of neurons in laminated structures, including the hippocampus, cerebellum, and cerebral cortex (1–3), as well as the migration of mesencephalic dopaminergic neurons (4). In adulthood, Reelin regulates neuronal communication and plasticity by modulating the activity of postsynaptic glutamatergic receptors, and neurotransmitter release (5, 6). Reelin functions depend on the activation of its receptors, apolipoprotein E receptor 2 (ApoER2) and very-low-density-lipoprotein receptor (VLDLR). Mice deficient in Reelin, ApoER2, or VLDLR exhibit defects in long-term potentiation and memory formation (7, 8). Mutations in the Reelin gene, RELN, are associated with microcephaly and lissencephaly (9), and also psychotic and mood disorders, autism spectrum disorder (ASD), schizophrenia, and Alzheimer’s disease (10–14). Reelin is expressed by Cajal-Retzius cells located in the Marginal Zone in the developing CNS (3) and in GABAergic interneurons in the adult brain (3).

We previously demonstrated that cultured Drosophila Mushroom Body (MB) neurons respond to mammalian Reelin by increasing the complexity of their neurites (15), similar to what is observed in neurons from the mammalian hippocampus and cerebral cortex when exposed to this signaling molecule (16). The Reelin-induced responses in cultured Drosophila neurons depend on Lipophorin receptor 1 (LpR1) and Lipophorin receptor 2 (LpR2) (15). Drosophila LpR1 and LpR2 are proteins with high similarity in amino acid sequence and domain organization to Reelin receptors, VLDLR and ApoER2 (17, 18). Furthermore, Reelin-induced effects on Drosophila neurons depend on disabled (Dab) (15), the fly ortholog of Dab1, which is the intracellular adaptor activated by ApoER2 and VLDLR in the Reelin signaling cascade in vertebrates (3).

The MB is a key structure in the fly brain, implicated in olfactory learning and memory, locomotor activity, and sleep regulation, among other functions (19–21). We showed that LpR-deficient flies exhibit deficits in axonal guidance, with consequences for adult MB structure (15). Furthermore, these anatomical deficits are consistent with phenotypes in sleep dynamics, and olfactory learning and memory (15). To date, no Reelin-like protein has been described in Drosophila. Moreover, it has been historically thought that Reelin signaling is only present in vertebrates (22). Interestingly, a genetic analysis proposed that proteins with similar functions to vertebrate Reelin appeared in evolution with the phylum Arthropoda (23). Given that Drosophila melanogaster belongs to this phylum, and that it has orthologs for proteins involved in the vertebrate Reelin signaling pathway, we hypothesized that a Reelin-like signaling pathway must exist in Drosophila (15).

Here, we identified a previously uncharacterized protein, encoded by the CG17739 gene, with the highest amino acid similarity to Reelin. Because the predicted protein exhibits even higher similarity to mammalian F-spondin, we named the CG11739 gene product, Drospondin. Our data show that Drospondin is an extracellular matrix (ECM) protein secreted by glial cells. When added to cell cultures, Drospondin stimulated the arborization of MB neurons in a LpRs-dependent way. In vivo, the downregulation of Drospondin caused MB developmental defects through a genetic interaction with LpRs. Moreover, flies deficient in Drospondin exhibited alterations in sleep, locomotion, and social behavior. These flies also showed a significant decrease in brain size, somewhat resembling the microcephaly observed in individuals with Reelin deficiency. The structural defects detected in the brain and in cultured MB neurons in Drospondin-deficient flies, were rescued when human Reelin was expressed under the control of the endogenous Drospondin promoter. These findings are consistent with Drospondin being a functional ortholog of vertebrate Reelin that controls the development and function of the Drosophila nervous system.

Results

Drospondin Is an Extracellular Glycoprotein with Sequence Similarity to F-spondin and Reelin.

To identify the most similar gene to human RELN in the Drosophila genome, we used the DIOPT-Ortholog prediction tool (24). We also searched for Drosophila proteins containing the highest sequence homology to each of the Reelin domains (Fig. 1). An uncharacterized protein encoded by the CG17739 gene emerged as the main candidate. This protein exhibits 29% amino acid sequence similarity to human Reelin but is also similar to human F-spondin (47% sequence similarity) (SI Appendix, Table S1). Like Reelin, F-spondin (also known as Spon1), is an ECM protein that binds to ApoER2 and VLDLR, regulating neuronal migration across developing cortical layers such as the floor plate of the neural tube, the hippocampus, cerebral cortex, and the olfactory bulb (25–27).

Fig. 1.

CG17739 protein is a glycoprotein with homology to mammalian F-spondin and Reelin. (A) Schematic representation of CG17739-encoded protein, F-spondin, and Reelin and their different functional domains. N-glycosylation sites indicated by yellow arrowheads. (B–E) Drosophila CG17739-encoded protein structure predicted via the RoseTTAfold method (Robetta server). Representations in cartoon (B and D), and in molecular surface (C and E) formats. In (B) and (C), colors show the secondary structures (yellow: β-sheets, purple: α-helix, blue: coils, white: no defined structure); in (D and E) different colors show domains (red: reeler domain; green: spondin domain; blue: thombospondin type I domain; light blue: Kunitz domain). (F) Western blots of Drospondin (CG17739-encoded protein) expression in S2 cells lysates and their supernatants. Two bands are detected in the supernatants; n = 2 independent experiments. (G) Western Blot of S2 cells expressing CG17739 treated with Tunicamycin and their supernatants. Only after overexposure of the membrane it is possible to detect the band at a lower molecular weight. “C” and “T” indicate control and tunicamycin treatment, respectively; n = 2 independent experiments.

The CG17739 gene is predicted to generate only one transcript of 2,830 nucleotides (flybase.org, FBgn0033710) and an ECM protein of 873 amino acids with a molecular weight of 98,282 kDa (uniprot.org, Q7K3Y9_DROME). We generated a model for the CG17739-encoded protein based on data available in the UniProt server, and compared it to the domain organization of human Reelin and F-spondin (Fig. 1A). As expected for secreted proteins, all contain an N-terminal signal peptide that allows their translation and maturation at the rough endoplasmic reticulum. Additionally, the CG17739-encoded protein contains one Reeler domain, followed by a Spondin domain and Thrombospondin type I domain repeats (TSRs). In the case of Reelin, after the Reeler domain, there is a unique region followed by Reelin repeats. The CG17739-encoded protein also comprises a Kunitz domain, absent in F-spondin and Reelin. We generated a three-dimensional model of the protein (28) (Fig. 1 B–E), which shows that the Reeler and Spondin domains exhibit β-sandwich conformations and are physically close to each other (Fig. 1 B and D). The TSRs exhibit antiparallel β-strand conformations, while the Kunitz domain has the characteristic combination of α helix and β strand (alpha-beta fold) (Fig. 1 B and D). Given all the similarities between the CG17739-encoded protein and the vertebrate Reelin and F-spondin, we named this protein Drospondin.

To corroborate that the identified Drospondin open reading frame (ORF) encodes a secreted protein, we cloned and expressed it in Drosophila S2 cells with a N-terminal Myc epitope (Fig. 1F and SI Appendix, Fig. S1 A–C). Immunoblot analysis for the Myc epitope detected two bands of ~85 and ~120 kDa in the culture medium. Interestingly, only the ~120 kDa band was detected in the cell lysates (Fig. 1F and SI Appendix, Fig. S1 B and C). This supports that the higher band is the full-length protein while the lower band may arise from cleavage of the larger species right before or after secretion into the extracellular medium. The 120 kDa band corresponds to a protein with a molecular mass higher than that predicted from the amino acid sequence (Fig. 1F). A search for glycosylation sites revealed the presence of both N-glycosylation and O-glycosylation motifs in the Drospondin sequence (29). We treated S2 cells stably expressing Myc-Drospondin with 5 μg/mL tunicamycin to inhibit N-glycosylation. In these cells, the amount of Drospondin in both the medium and cell lysates was reduced (Fig. 1G and SI Appendix, Fig. S1D). The tunicamycin treatment also resulted in a band of lower molecular weight, which is only evident after membrane overexposure, consistent with the proposition that Drospondin is N-glycosylated (Fig. 1G). These findings suggest that Drospondin N-glycosylation is needed for its correct folding, maturation, stability, and secretion, as reported for other glycoproteins (30, 31).

Drospondin Is Expressed in Glial Cells.

The analysis of published RNA-seq data (32) of several tissues in adult Drosophila shows that the transcript of the CG17739 gene is present in the head and brain (SI Appendix, Fig. S2A). Additionally, single-cell RNA-seq data (33, 34) support that Drospondin is mostly expressed by several glial types (SI Appendix, Fig. S2 B and C). To verify this information, we used a fly strain in which the last two exons of Drospondin are replaced with a CRIMIC cassette (35, 36) (hereafter called Drosp^CR70269). The CRIMIC cassette includes a T2A-GAL4 element followed by a stop codon. This produces a severe loss of function allele while expressing the GAL4 transcription factor in the expression pattern of endogenous Drospondin. We used this genetic tool to identify Drospondin-expressing cells by nuclear localization of mCherry fluorescent protein (UAS-mCherry-NLS) (Fig. 2A). Drospondin was found in some Repo-positive glial cells at the larval 3 stage (L3), and in almost all Repo-positive glial cells in pupae and adult flies (Fig. 2A). The expression pattern of Drospondin suggests that it is mainly found in the ensheathing and cortex glia (37, 38). Further studies revealed no colocalization of Drospondin and the neuronal marker Elav at any of the analyzed developmental stages, suggesting that the protein is not expressed by neuronal cells (Fig. 2B).

Fig. 2.

Drospondin is expressed in glial cells. (A) Immunostaining against Repo, in brains expressing the nuclear fluorescent mCherry protein, directed by the Drospondin promoter (Drosp^CR70269;+; UAS-mCherry.NLS). (B) Immunostaining against Elav in brains expressing nuclear fluorescent protein mCherry directed by the Drospondin promoter (Drosp^CR70269;+; UAS-mCherry.NLS). Experiments carried out in L3 larvae, pupae (stage 9 to 10), and adult flies (0 to 3 d old); panels show staining results and merged images. Right panels show amplifications of areas indicated in panels at Left. (Scale bar, 20 µm.)

Drospondin Increases the Complexity of the MB Neuronal Arbor.

We previously showed that mammalian Reelin increases the complexity of the neuritic arbor in cultured Drosophila MB neurons (15), similar to the effects induced by Reelin or F-spondin in mammalian neuronal cells (16, 39–41). To assess whether Drospondin induces similar effects, MB neurons in culture were treated with this protein (0.5 to 50 nM) resulting in a higher complexity in the neuritic tree than control neurons (Fig. 3 A and B). The maximum effect in the most extended neurite was found at Drospondin 15 and 50 nM (Fig. 3C). The Branching index, a parameter that relates to the complexity of the neuritic arbor, indicates that the Drospondin effects were stronger in neuronal processes arising from second-order neurites rather than on primary neurites (Fig. 3D). Thus, similar to Reelin, Drospondin increases the complexity of the neuritic tree of MB neurons in culture.

Fig. 3.

Drospondin promotes neuritic growth in cultured neurons. (A) Representative black and white images of primary neuronal cultures prepared from c309,CD8.ChRFP pupal brains, after treatment with Drospondin (0.5, 5, 15, and 50 nM) and their corresponding mock treatments. MB neurons identified by RFP expression. (Scale bar, 10 µm.) (B) Sholl profile of MB neurons under experimental conditions. (C) Results for maximum length, a parameter defined as the larger distance reached by a neurite with respect to its soma. Kruskal–Wallis test; **, ****, “ns” indicate P < 0.01, P < 0.0001 and not significant; n = 3 independent experiments, 15 to 20 cells in each experiment. (D) Data for branching index. The Kruskal–Wallis test shows ***, ****, ns, meaning P < 0.001, P < 0.0001, not significant. (E) Description of Reelin treatment. Primary brain cultures from Drospondin mutants were treated with Reelin for 2 or 5 d, and fixed at the seventh day in vitro, regardless of the treatment. (F) Sholl profile of Drospondin mutants that express GFP only in MB neurons after Reelin treatment. Only when mutant neurons are treated with Reelin for 5 d, their Sholl profile is similar to control cultures. (G) Neurite maximum length is decreased in Drospondin mutants; this phenotype is only reverted after a 5-d treatment with Drospondin. In E–G, all Drospondin mutant cells were from haploinsufficient deficient animals. Two-way ANOVA, Tukey posttest; *, **, ***, ****, means P < 0.05, P < 0.01, P < 0.001 and P < 0.0001, respectively; n = 3 independent experiments, 15 to 20 cells in each experiment. Data expressed as mean ± SEM.

We then hypothesized that neurons from Drospondin-deficient animals might show a basal impairment in the development of their neuritic arbor that could be rescued by Reelin. We prepared cultures from Drospondin heterozygous deficient animals containing an insertion in the 5’UTR of the gene (identified from here onward as Drosp^EY18336; SI Appendix, Fig. S3 A and B) and studied the effect of Reelin on cultured MB neurons (Fig. 3 E–G). Primary cultured neurons were treated with recombinant Reelin for the last 2 or 5 d (Fig. 3E). MB neurons from Drosp^EY18336 heterozygous mutants exhibited a significant basal deficit in neurite length and complexity compared to control neurons, as hypothesized (Fig. 3 F and G). Importantly, these defects were not detected when the mutant neurons were treated with Reelin for a total of 5 d, as neurite arborization levels were comparable to those observed in the mock-treated control group. Reducing the Reelin treatment to the last 2 d of culture showed no restoration of neurite growth (Fig. 3 F and G). Therefore, an early Reelin treatment rescued the deficit in neurite growth observed in MB neurons from Drospondin-haploinsufficient flies. These results suggest a critical time window for the role of Drospondin in neuronal differentiation.

Altogether, these findings support that Drospondin shares conserved functions with Reelin and F-spondin in facilitating neuritic outgrowth.

Drospondin Is Required for MB Axonal Development.

The MB is a structure in the Drosophila brain exhibiting a particular anatomical organization where soma, dendritic, and axonal segments are separated from each other; the axons form the so-called MB lobes (19, 42, 43). Our previous work showed that flies with reduced expression of LpRs exhibit defects in MB development (15). Since Drospondin could be a putative ligand for LpRs, we evaluated whether it localizes near the MB region. This was assessed using the Drosp^CR70269 allele to direct the expression of HA-tagged Drospondin under the control of the endogenous promoter (Fig. 4A and SI Appendix, Fig. S4). FasII immunostaining was used to reveal the MB peduncles and lobes (44, 45). Drospondin was mainly present on the surface of the larval brain without any evident localization in the MB. In pupae and adult animals, Drospondin was also present on the brain surface, as well as around the MB calyx, peduncles and lobes. This staining pattern is consistent with a glial expression of Drospondin (Fig. 4A) (37, 38, 46).

Fig. 4.

Drospondin contributes to MB adult structure. (A) Microphotographs showing the MB in larval, pupal, and adult brains. The lobes and calyx regions in pupa and adult animals are shown; peduncles, α and β lobes are identified by FasII immunostaining (green hue). HA-tagged Drospondin (magenta) was expressed under the control of endogenous promoter (Drosp^CR70269/+;UAS-Drospondin.3xHA). Data from n = 5, 7, and 17 L3 larvae, pupae, and adult brains, respectively. (Scale bar, 20 µm.) Bottom, scheme of larval and pupal/adult brains showing the MB. (B–G) Phenotypes observed with different tools used to reduce Drospondin expression. (B) Typical MB structure in control conditions. (C) Axons from one β lobe that reach the opposite β lobe (Repo-GAL4/UAS-RNAi Drosp). (D) Merge of β lobes (Drosp^CR70269/Drosp^EY18336). (E) Thin β lobe (Drosp^CR70269/Drosp^EY18336). (F) α and β are thinner than normal (Drosp^CR70269/+). (G) Short α lobe (Drosp^CR70269/Drosp^EY18336). The arrowhead indicates the abnormality. (Scale bar, 50 µm.) (H) Percentage of fly brains from each genotype exhibiting identified phenotypes. Data presented per strain. Please see SI Appendix, Tables S2 and S3 for additional information.

The α’ and β’ lobes of the MB develop toward the end of the L3 stage, while α and β lobes finish their maturation at the pupal stage (43). Thus, by the beginning of adulthood the mature MB structure has been established (43). Since our studies suggest that Drospondin is expressed in glial cells surrounding the MB, we asked whether a reduction in glial Drospondin expression has any consequences for MB development (Fig. 4 B–H and SI Appendix, Tables S2 and S3). We used the Gal4-UAS system to express a RNAi for Drospondin transcripts under the control of the pan-glial driver Repo-Gal4 (SI Appendix, Fig. S3C), and FasII immunostaining to study MB architecture in adult flies. In contrast to the typical MB structure in control flies (Fig. 4B), in animals with Drospondin knockdown (KD) we observed axons from one β lobe in one side of the brain crossing the midline toward the opposite side (Fig. 4C), while in other brains, there was a complete merge of β lobes (Fig. 4D). Other phenotypes include gross alterations of the MB structure, like the presence of one (Fig. 4E) or two (Fig. 4F) thinner than normal lobes, missing or shorter lobes (Fig. 4G). Thus, about 6% of flies with glial Drospondin KD exhibit some degree of β lobes fusion and 20% of these brains lack parts of the MB lobes (Fig. 4H and SI Appendix, Table S3).

As mentioned, our data and the information from several DNA-seq studies support that Drospondin is mainly expressed in glial cells. To study whether one glial subtype is a major contributor to Drospondin synthesis and MB development, we directed the expression of Drospondin RNAi to specific glial subtypes (SI Appendix, Fig. S5 and Tables S4–S6). Our results show that knocking down Drospondin in cortex glia resulted in 30% of the MB with anatomical defects (SI Appendix, Fig. S5 and Table S6), which phenocopied what we reported when the RNAi for Drospondin was directed by the pan-glial driver Repo-Gal4 (Fig. 4H). Since the expression of RNAi for Drospondin under the control of the other glial drivers did not result in evident phenotypes (SI Appendix, Fig. S5 and Tables S4 and S5), our data support that the cortex glia is the main source for Drospondin.

To corroborate the contribution of Drospondin to the MB architecture we used the CRIMIC strain. Finding homozygous Drosp^CR70269 mutants was challenging (only 2.5% and 1.6% of male and female animals, respectively), as they died prematurely with a median survival of 3 d (SI Appendix, Fig. S6). On the other hand, the transheterozygous mutants Drosp^CR70269/Drosp^EY18336 were more frequently found, although their median survival was 4 d (SI Appendix, Fig. S6). As expected, the heterozygous and transheterozygous mutants expressed Drospondin at a significantly reduced level (SI Appendix, Fig. S3 D and E). Thus, the MB architecture was evaluated in the two mutants in haploinsufficiency (Drosp^CR70269/+ and Drosp^EY18336/+) and transheterozygosity (Drosp^CR70269/Drosp^EY18336). Flies mutant for Drospondin in one allele exhibited β lobes merged in 11% (Droso^EY18336/+) and 10% (Drosp^CR70269/+) of the analyzed brains (Fig. 4). This is not different from what we observed in the transheterozygous mutants (Drosp^CR70269/Drosp^EY18336), with 8.8% of the brains showing β lobes fusion. When assessing gross anatomy of the MB region including missing lobes, about 7% of Drosp^EY18336/+ and Drosp^CR70269/+ brains exhibited this phenotype. Of note, this phenotype increased in prevalence to 17.6% of brains in transheterozygous mutants.

Expressing a HA-tagged Drospondin (Drosp^CR70269/+;UAS-Drosp.3xHA/+) in the haploinsufficient mutant genetic background can correct the MB defective phenotypes (Fig. 4H and SI Appendix, Table S2). Finally, since Reelin and Drospondin exert similar effects on MB neurons in primary culture, we also performed experiments where human Reelin was expressed in the Drospondin-deficient background (Drosp^CR70269/+;UAS-Reelin.3xHA/+). Remarkably, these flies displayed a normal MB architecture (Fig. 4H and SI Appendix, Table S2), demonstrating that human Reelin can functionally replace fly Drospondin.

Overall, these results demonstrated that Drospondin plays a role in MB development, and it is functionally orthologous to mammalian Reelin.

Drospondin Interacts Genetically with LpRs In Vivo and In Vitro.

Flies haploinsufficient for LpRs exhibit impairment in MB organization (15), which is similar to what is observed in Drospondin-deficient flies. To assess whether there is a genetic interaction between Drospondin and LpR1 or LpR2, we used flies heterozygous for a CRIMIC element insertion in LpRs, LpR1^CR70219 and LpR2^CR70220. These mutants showed 16.6% and 10.4% of brains with β lobes fusion, respectively, similar to the proportion of Drospondin heterozygous mutant flies exhibiting this phenotype (Fig. 5 A–G and SI Appendix, Fig. S7 and Tables S7 and S8). Remarkably, in adult flies bearing one mutant allele for Drospondin and one for the LpR1 gene (Drosp^CR70269/+; LpR1^CR70219/+), the frequency of flies with fusion of β lobes increased dramatically to 40.7%. Heterozygous mutant flies for Drospondin and LpR2 genes (Drosp^CR70269/+; LpR2^CR70220/+) only showed a modest increase in the incidence of this phenotype (Fig. 5G and SI Appendix, Tables S7 and S8). We next assessed axonal growth defects in MB. About 5% and 8% of brains showed this phenotype in flies lacking one copy of LpR1 or LpR2, respectively (Fig. 5G). The brains of double mutants Drosp^CR70269/+; LpR1^CR70219/+ flies showed almost no change in the frequency of defects, while in Drosp^CR70269/+; LpR2^CR70220/+, this phenotype was increased to 25% of the brains studied (Fig. 5G).

Fig. 5.

Drospondin and LpRs genetic interaction. Immunostaining of MB using FasII in haploinsufficient flies for Drospondin and LpR1 or LpR2. (A–F) Examples of phenotypes observed. (A) MB in control animals (w¹¹¹⁸); (B) Axons from one β lobe that reach the opposite β lobe (LpR2^CR70220/+). (C) merge of β lobes (LpR1^CR70219/+). (D) Thin β lobe (Drosp^CR70269/+; LpR1^CR70219/+). (E) Short α lobe (Drosp^CR70269/+; LpR2^CR70220/+). (F) Without α lobe (LpR2^CR70220/+). The arrowhead indicates the abnormality. (Scale bar, 50 µm.) (G) Distribution (in percentage) of each phenotype in animals of genotypes indicated. Please see SI Appendix, Tables S7 and S8 for statistical information. (H) Sholl analysis of control MB neurons or with downregulation of LpR1 expression, exposed to Drospondin. (I) Maximum length parameter in MB neurons. Two-way ANOVA, Tukey posttest; ****, “ns” indicate P < 0.0001, not significant. (J) Sholl analysis of control MB neurons or with downregulation of LpR2 expression, after exposure to Drospondin. (K) Maximum length parameter. Two-way ANOVA, Tukey posttest; ****, “”ns” mean P < 0.0001, not significant, respectively. Data (in H–K) from n = 3 independent experiments, 13 to 20 cells in each experiment. Data in graphs expressed as mean ± SEM.

Vertebrate Reelin induces neuritic tree outgrowth in cultured MB neurons, an effect that depends on LpRs (15). We asked whether the Drospondin-induced response in cultured MB neurons depends on LpRs. MB neurons KD for LpR1 (c309,CD8.chRFP/+; RNAi LpR1/+) or for LpR2 (c309,CD8.chRFP/+; RNAi LpR2/+) were unable to respond to 15 nM of Drospondin by incrementing neither the complexity of their neuritic arborization nor the maximum length their neurites can reach (Fig. 5 H–K).

Altogether, these results show that LpRs genetically interact with Drospondin to regulate MB development and neuritic tree complexity, supporting their participation in a common signaling pathway

Drospondin-Deficient Flies Exhibit Smaller Brain Size.

Mutations in RELN have been linked to disorders like lissencephaly with microcephaly and cerebellar hypoplasia (9, 47, 48), which are thought to arise from decreased Reelin signaling during brain development (47, 48). The expression pattern of Drospondin in glial cells throughout the brain, supports that several cell types and brain regions could receive and be influenced by this signaling molecule.

We measured the middle brain area (Fig. 6A) of animals with Drospondin downregulation in glial cells (Repo-GAL4/UAS-RNAi Drosp). Remarkably, these flies exhibited a significant smaller brain area (Fig. 6 D and E) as compared to genetic controls (Fig. 6 B and C) (Repo-GAL4/+ and UAS-RNAi Drosp/+), and 19% of the brains in the KD experimental group were smaller than the smallest brains recorded in any of the two genetic controls (red bar in Fig. 6E). This is reminiscent of severe microcephaly, a human condition identified at birth, defined by a head circumference three or more SD below the mean average size in the entire population (49–51). Our data show that 33% of the brains of flies with Drospondin KD in glial cells fulfill the criterion for microcephaly.

Fig. 6.

Drospondin misexpression results in a microcephaly-like phenotype in flies. (A) Scheme of a fly brain; the discontinuous red line represents the measured area (middle brain). (B–D) Representative brain images obtained in control strains, (B) Repo-GAL4/+ and (C) UAS-RNAi Drosp/+, and (D) flies with KD for Drospondin (Repo-GAL4/UAS-RNAi Drosp). (Scale bar, 100 µm.) (E) Quantification of middle brain area in flies from each genotype. One-way ANOVA, Tukey posteriori test show ***, “ns” which indicate P < 0.001, not significant, respectively. The right y-axis shows percentage of brains with a smaller area than the smallest brain area recorded in any of the two controls (in red bar). (F–L) Representative images of the middle brain area from (F) control animals (w¹¹¹⁸); (G and H) heterozygous mutant flies for Drospondin (Drosp^EY18336/+ and Drosp^CR70269/+); (I) transheterozygous mutant flies for Drospondin (Drosp^CR70269/Drosp^EY18336); (J) control flies bearing the undriven genetic element to express Drospondin (UAS-Drosp.3xHA/+), (K) animals expressing Drospondin under the control of the endogenous promoter, in the Drospondin CRIMIC genetic background (UAS-Drosp.3xHA/+; Drosp^CR70269/+), (L) flies expressing Reelin directed by Drospondin promoter in the Drospondin CRIMIC background (UAS-Reelin.3xHA/+; Drosp^CR70269/+). (Scale bar, 100 µm.) (M) Middle brain area per genotype. Kruskal–Wallis test followed by Dunn posttest, **, ****, “ns” mean P < 0.01, P < 0.0001, not significant. The right y-axis shows percentage of brains with a smaller area than the smallest brain area recorded in either of the controls, w¹¹¹⁸ or UAS-Drosp.3xHA/+ (in red bars). In graphs, data presented as mean ± SEM from at least 3 independent experiments. Additional information in SI Appendix, Table S9.

We also studied this phenotype in Drospondin mutant animals (Fig. 6 F–M). Although about 21% of the Droso^EY18336/+ mutants (red bars in Fig. 6M) exhibited a brain area smaller than the smallest area recorded in control animals, the mean brain area measured in Droso^EY18336/+ mutant flies was not significantly different than that recorded in the control strain (Fig. 6 F, G, and M). In contrast, a significantly smaller mean brain area was measured in Drosp^CR70269/+ as compared to controls (Fig. 6 F, H, and M and SI Appendix, Table S9). In the heterozygous CRIMIC mutants, 83% of the brains had a brain area smaller than that recorded in control flies (red bar, Fig. 6M), and 59% percent of them fulfilled the criterion for microcephaly. The differential effects between the two mutants could be explained by the level at which each genetic allele decreases Drospondin expression, being the CRIMIC mutant more potent (SI Appendix, Fig. S3 A, B, and E). The transheterozygous mutant (Drosp^CR70269/Drosp^EY18336) also showed a significantly smaller brain area as compared to control flies (Fig. 6 F, I, and M), which was not different from the effect detected in the heterozygous CRIMIC mutant (Fig. 6M and SI Appendix, Table S9). Importantly, the expression of Drospondin driven by the endogenous CRIMIC allele (Drosp^CR70269/+;UAS-Drosp.3xHA/+) fully rescued the brain size (Fig. 6 F, J, K, and M). The expression of Reelin in the Drosp^CR70269 mutant background (Drosp^CR70269/+;UAS-Reelin.3xHA/+) was also efficient at rescuing the brain size phenotype (Fig. 6 F, L, and M), although it was still possible to detect about 35% of smaller brains (red bar, Fig. 6M). Remarkably, the smaller size brain phenotype seems not to be accompanied by a reduced body size (SI Appendix, Fig. S8). These results support the notion that Drospondin is a relevant glial signaling molecule in Drosophila whole-brain development.

Drospondin Deficiency Impacts Sleep, Locomotion, and Social Distance in Adult Animals.

As mentioned, the MB region is associated with the regulation of several behaviors in flies. Given the structural phenotypes observed in the MB of Drospondin mutants (Fig. 4), we asked whether Drosp^CR70269/+ animals exhibited any behavioral alteration. First, we assessed circadian motor activity (Fig. 7 A–D). The Drosp^CR70269/+ mutant exhibited an overall longer sleeping time primarily due to a defect during the light phase, as there were no differences during the dark phase (Fig. 7B). In contrast, during the light phase, there was a shortened latency to sleep in mutant animals as compared to controls, but a longer latency during the dark phase (Fig. 7C). Drospondin heterozygous mutants also showed an increased number of sleep events (Fig. 7D). Thus, Drosp^CR70269/+ mutants exhibited a dysregulation in the sleep pattern that is more evident during the day phase.

Fig. 7.

Impairment in sleep pattern, locomotor activity, and social distance in mutants for Drospondin. (A) Sleep profile of haploinsufficient Drospondin mutant flies (Drosp^CR70269/+) as compared to control flies (w¹¹¹⁸). The gray shadowed background represents the hours when flies were at dark. (B) Fraction of sleeping time. (C) Average latency per day. (D) Average number of sleep bouts per day. Data (in A–D) is presented per strain (number of animals): w¹¹¹⁸ control animals (42 files); Drosp^CR70269/+ (36 files). Results analyzed by two-way ANOVA, Bonferroni post-test; *, **, ***, **** and “ns” mean P < 0.05, P < 0.01, P < 0.005, P < 0.001 and not significant, respectively. (E–G) Locomotor activity recorded at different time windows. (E) Total distance, (F) Average speed, and (G) Activity time. Two-way ANOVA, Bonferroni’s multiple comparison test show *, **, ***, ****, “ns” which means P < 0.05, P < 0.01, P < 0.005, P < 0.0001, not significant, respectively. n = 26 animals per genotype at each time window. (H and I) Social distance recorded in the two strains. (H) Representative images of social experiments. (I) Averaged distance to the closest neighbor. Unpaired t test P < 0.01, n = 11 independent experiments, 30 to 45 flies per genotype in each experiment. Data in all graphs as mean ± SEM.

Adult flies were also assessed in their basal locomotion at different ages (Fig. 7 E–G). One- to three-day-old Drospondin mutants showed reduced walking distance (Fig. 7E), speed (Fig. 7F), and activity time (Fig. 7G), as compared to controls, while in the 7- to 10-d-old window, the mutants showed reduced total distance traveled and speed of movement, only (Fig. 7 E–G). Older flies showed a reduction in all motor parameters as compared to control flies.

Finally, as some Reelin mutations are linked to neurodevelopmental disorders associated with social deficits (11, 52), and since the MB modulates fly interactions (53), we evaluated social behaviors in animals with reduced Drospondin expression (Fig. 7 H and I). Drospondin-deficient animals maintained a higher distance to the closest neighbor (Fig. 7 H and I), a proxy for asocial behavior in flies (54, 55).

Overall, these results support that reduced expression of Drospondin throughout development has significant consequences for different behaviors associated with the MB.

Discussion

In the search for a Drosophila Reelin ortholog that explains the role of LpRs in MB formation and function, as well as the LpR-dependent Reelin-induced effect in cultured MB neurons (15), we identified the previously uncharacterized CG17739 gene. This gene encodes for an extracellular glycoprotein secreted by Drosophila glial cells in the brain, which we named Drospondin. Similar to Reelin effects on vertebrate (16, 39, 56) and invertebrate (15) neurons, cultured MB neurons respond to Drospondin by increasing the complexity of their neuritic tree in a LpRs-dependent manner. In addition, flies with reduced expression of Drospondin exhibit MB developmental defects, like those found in LpR deficient animals. Furthermore, Drospondin and LpRs genetically interact to influence MB development. Flies deficient in Drospondin throughout development also exhibit impaired sleep regulation, locomotion, and social distance. We also report that Drospondin expression is relevant for normal brain size in flies. Remarkably, vertebrate Reelin was able to rescue the neurite defects in MB neurons from Drospondin-deficient flies in primary cell culture, as well as the MB structure and brain size in adult flies. Therefore, Drospondin is a protein that was previously overlooked but has an essential role in brain development, the impairment of which has profound functional consequences for the animal. Overall, these findings are consistent with the notion that Drospondin might signal through a pathway analogous to that described for Reelin in vertebrates.

Our bioinformatics analysis indicates that Drospondin is more similar to F-spondin than to Reelin (57, 58). There is at least one other ortholog for vertebrate F-spondin in the Drosophila genome, fat-spondin, encoded by CG6953 (26). Regarding their structures, vertebrate F-spondin and Reelin, as well as Drospondin and fat-spondin, contain a Reeler domain following the signal peptide. In addition to the Reeler domain, Drospondin, fat-spondin, and F-spondin share the spondin (N-terminal) domain followed by TSR domains. All these domains are in the N-terminal region, which exhibit the highest sequence homology among them (41). The common structural features suggest that these proteins could contribute to similar functions in vertebrates and Drosophila. Importantly, there are no previous reports on the contribution of fat-spondin or Drospondin to any biological function in Drosophila.

Drospondin Contributes to Brain Formation.

In vertebrates, Reelin and F-spondin have similar roles but with some differences. For instance, Reelin is not only expressed in neurons but also secreted by Schwann cells and contribute to axonal outgrowth in the periphery (56, 59). As Reelin, F-spondin mRNA is also present in embryonic Schwann cells when sensory and motor axons project to their peripheral targets (60). To accomplish its actions, F-spondin binds to ApoER2 and VLDLR through their TSR domains (61–63), which Drospondin also contains. On the other hand, Reelin binds its receptors through their central tandem repeats (3, 64). Regardless of the mechanism, both Reelin and F-spondin activate Dab1, the first mediator of the signaling cascade responsible for the effects induced by these molecules in mammalian cells (3, 65–68). Thus, we proposed that Reelin, F-spondin, and Drospondin contribute to neurite formation, axon pathfinding, and brain development and asked whether Drospondin contributes to fly brain development, focusing on a well-known structure, the Drosophila MB.

Drospondin-deficient animals exhibit a variety of MB defects. Some of the phenotypes were similar to those reported in flies deficient in LpRs including the merge of β lobes (15). This phenotype can be explained by axons crossing the midline of the brain if Drospondin acts as a stop signal for axon extension. The demonstration of a genetic interaction between Drospondin and LpRs in modifying β lobe fusion, further supports that Drospondin would act as the ligand and LpRs as the receptors. This is consistent with a study in C. elegans reporting that the central TSR1-4 of F-spondin binds a Lipoprotein Receptor Related Protein (LRP), resulting in axon repulsion (69). Likewise, Reelin promotes detachment of neurons from the radial glia once they reach layer V of the cerebral cortex and stop their migration in rodents (3, 70). In addition, high concentrations of F-spondin induce the inhibition of neurite outgrowth (41).

Drospondin deficiency results also in thin or missing MB lobes, which could be attributed to either a reduction in the number of axons protruding from MB neurons or to axons failing to reach their targets. Primary neuronal cultures deficient in Drospondin exhibit poor neurite development, like the phenotypes observed in LpR-deficient animals (15). Schubert et al. (41) previously reported that F-spondin promotes neurite outgrowth of rat nerve precursor cells. Likewise, Reelin induces neurite development in hippocampal and cortical vertebrate neurons (39, 40, 56). Importantly, our work shows that human Reelin can replace fly Drospondin during MB development, demonstrating a conserved function of the protein, and supporting that fly Drospondin, mammalian Reelin and F-spondin, all similarly promote neuronal outgrowth and differentiation.

Our study focused on the role of Drospondin in MB anatomy and functions. However, other brain areas could be affected by the lack of this protein since it is broadly expressed by glial cells throughout the entire brain. Indeed, we found a smaller brain size (microcephaly) in Drospondin-KD flies that could be explained by a smaller number of brain cells because of a delay in brain development, cell cycle defects, increased apoptosis in neuronal progenitors, defects in pattern formation, or lack of cellular differentiation (71, 72). Further studies are needed to explore these possibilities. Importantly, RELN and VLDLR mutations have been linked to lissencephaly with cerebellar hypoplasia in humans (9, 47, 48, 73). Thus, the Drospondin mutant provides a genetic tool to study cellular mechanisms of microcephaly.

Drospondin Contributes to Drosophila Behavior.

Reelin deficiency produces significant morphologic changes in the vertebrate brain and is associated with different neurodevelopmental and psychiatric disorders (11, 13, 74–76). Many of these conditions have been associated with alterations in sleep regulation, motor output, and social interaction, among other features (77–80). However, little is known on the direct contribution of Reelin deficiency to any of these behavioral features. One of the few reports available on this issue shows that patients with lissencephaly linked to RELN mutations exhibit abnormal neuromuscular connectivity (81) and hypotonia (47), which could underlie motor deficits. Less is known about a potential contribution of F-spondin deficits to behavioral alterations in human disorders. In this regard, F-spondin is detected in the neocortex and hippocampus in adult animals, and it might play a role in AD etiology by binding to ApoER2 and the amyloid precursor protein (APP) (58). Drosophila constitutes a tool to gain a better understanding of the effects induced by a deficit in these signaling molecules, how they are linked to human disorders, and what are the molecular mechanisms involved in them.

The homozygous Reeler mice exhibit dystonia, ataxia, and tremor, while mice with one mutant copy of Reln are normal for these phenotypes (82). On the other hand, mice lacking F-spondin fail to maintain intrinsic circadian rhythmicity (83). Accordingly, flies deficient in Drospondin showed alterations in sleep, the execution of motor programs, and the regulation of social space. These are behavioral phenotypes previously reported in Drosophila models for neurodevelopmental and/or mental disorders supporting that fly models can inform on pathological mechanisms underlying related human conditions, as reported elsewhere (54). Future studies should evaluate whether and to what extent specific brain areas and neuronal circuits underlying circadian or social behaviors are affected in Drospondin mutant flies.

In conclusion, our study shows that Drospondin is a signaling molecule in Drosophila that plays a key role in the development of the MB, uncovering a pathway relevant to brain formation and function. Drospondin plays similar functions in Drosophila to those described for mammalian F-spondin and Reelin. Given the similarities between Drospondin and vertebrate Reelin and F-spondin, this work supports the use of Drosophila as an animal model to gain a better understanding of how the brain is affected by deficiencies in these signaling molecules in neurodevelopmental and neurodegenerative disorders.

Materials and Methods

Immunostaining of Drosophila brains were performed as reported (15). Sholl analysis of primary neuronal cultures treated with different Reelin or Drospondin concentrations was carried out as described (15). Behavioral studies were carried out to assess circadian, motor, and social parameters (15, 54, 84). Results obtained were compared in Drospondin-deficient and control flies. Please refer to SI Appendix for a detailed description of procedures and materials.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Study data are included in the article and/or SI Appendix.