The CPEB ortholog Orb2 regulates brain size through the TRIM-NHL RNA-binding protein, Brain tumor

Abstract

Neurodevelopment requires precise translational control, the disruption of which is implicated in various neurological disorders, including developmental delays, intellectual disability, and microcephaly. We report a novel role for the Drosophila CPEB-family protein Orb2, a translational regulator, in controlling brain size in a dose dependent manner. Loss of orb2 results in larval brain hypotrophy, whereas orb2 overexpression causes brain overgrowth. We demonstrate that orb2 is required for neural stem cell development from embryonic stages through larval neurogenesis. Structure-function analysis reveals that Orb2 RNA-binding activity promotes brain growth, while its poly-Q and ZZ domains act to restrain overgrowth. Further genetic and biochemical evidence indicates that orb2 functions upstream of the translational repressor Brain tumor (Brat), modulating Brat protein levels and, consequently, influencing brain size. These findings support a model wherein the antagonistic activities of Orb2 and Brat are critical for balanced brain growth during Drosophila neurodevelopment.

INTRODUCTION

RNA-binding proteins (RBPs) regulate neurodevelopment by controlling neural stem cell (NSC) proliferation, differentiation, and maintenance through post-transcriptional gene regulation (Darnell, 2013; Lennox et al., 2018; Prashad & Gopal, 2021). RBPs are critical for regulating various aspects of stem cell function and homeostasis by directly modulating RNA metabolism, localization, splicing, and transcript turnover, which are essential for proper neuronal development and function (Lennox et al., 2018; Parra & Johnston, 2022). Consequently, mutations in RBPs are associated with intellectual disability, developmental delays, seizures, microcephaly, and other neurodevelopmental disorders (Allanson et al., 2009; Barkovich et al., 2012; Pirozzi et al., 2018).

The conserved cytoplasmic polyadenylation element-binding (CPEB) proteins are RBPs that orchestrate the translation of target transcripts through canonical CPE sites in the 3’ untranslated region (3’UTR; (Hake & Richter, 1994). While CPEB proteins, such as CPEB4, are implicated in neurodevelopmental disorders, such as autism spectrum disorder and Fragile X Syndrome (Udagawa et al., 2013; Garcia-Cabau et al., 2025), the complete repertoire of CPEB protein interactions and RNA targets, particularly across different tissues and developmental stages, remains incompletely understood.

Drosophila encodes two CPEB proteins from unique genes, Oo18 RNA-binding protein (Orb) and Orb2 (Lantz et al., 1992; Hafer et al., 2011). Drosophila orb2 is orthologous to mammalian CPEB2–4 and has known roles in mRNA localization and translation (Huang et al., 2006; Keleman et al., 2007; Hafer et al., 2011; Xu et al., 2014). Orb2 undergoes conformational changes to switch between a translational repressor and an activator dependent upon prion-like oligomerization, during which the N-terminal glutamine-rich (poly-Q) domain promotes self-association (Hake & Richter, 1994; Keleman et al., 2007; Si et al., 2010; Khan et al., 2015; Hervás et al., 2016). CPEB proteins like Orb2 bind CPE motifs in target mRNAs through RNA recognition motifs (RRMs; (Hake et al., 1998; Krüttner et al., 2012; Majumdar et al., 2012; Khan et al., 2015; Stepien et al., 2016). While the C-terminal ZZ-type zinc-finger domain (ZZ-domain) was previously implicated in RNA binding by anchoring CPEB to target transcripts (Hake et al., 1998), evidence also supports an alternative function in protein-protein binding (Legge et al., 2004; Merkel et al., 2013; Oroz et al., 2020). Consistently, recent data support a role for the ZZ domain in the recruitment of translational repressors to its target transcripts (Low et al., 2025).

CPEB proteins like Orb2 influence translation levels by regulating the polyadenylation status of target RNAs. Monomeric Orb2 binds to the deadenylase CG13928 to repress translation, while synaptic activity triggers Orb2 oligomerization and binding to CG4612, a predicted poly(A)-binding protein, to promote polyadenylation and translational activation of target mRNAs (Si et al., 2003a; Si et al., 2003b; Si et al., 2010; Majumdar et al., 2012; Khan et al., 2015). For example, recent work indicates that Orb2 promotes rare-codon mRNA translation in the brain (Stewart et al., 2024).

Cell type- and stage-specific expression of orb2 enables dynamic mRNA stability and translation control required for neuronal function. Orb2 also autoregulates its own translation by binding CPE motifs in its 3’ UTR (Majumdar et al., 2012; Stepien et al., 2016). Despite dynamic changes in orb2 expression and localization throughout neurogenesis (Hafer et al., 2011), Orb2 is best known for its roles in memory and learning. Orb2 has been extensively studied as a key synaptic regulator of long-term memory, wherein activity-induced aggregation facilitates sustained neuronal responses to learning. This molecular process is essential for stabilizing memory traces and supporting lasting behavioral adaptations (Si et al., 2003a; Keleman et al., 2007; Krüttner et al., 2012; Majumdar et al., 2012; Khan et al., 2015; Hervás et al., 2016; Li et al., 2016; Oroz et al., 2020; Kozlov et al., 2021; Kozlov et al., 2023). By comparison, relatively little is known about Orb2 functions during neurodevelopment.

Drosophila neurodevelopment commences during embryogenesis, when NSCs are specified by spontaneous delamination from the neuroepithelium (Hartenstein & Campos-Ortega, 1984; Doe, 1992). After delamination, NSCs undergo repeated rounds of asymmetric cell division to generate a self-renewing stem cell and a ganglion mother cell fated for differentiation into neuronal or glial subtypes (Broadus & Doe, 1997; Cabernard & Doe, 2009). The balance between self-renewal versus differentiation is governed by precise segregation of cell fate determinants and is vital to maintain homeostasis in neurogenesis, where deregulation can lead to tumorigenesis, neurodegeneration, or disorders such as microcephaly (Wang et al., 2016). NSCs enter quiescence in late embryonic stages and resume proliferation during the first larval instar, termed NSC reactivation (Truman & Bate, 1988; Maurange & Gould, 2005). While about 10% of the neurons in the adult brain are born during embryonic stages, the remaining 90% of neurons arise throughout larval development (Prokop & Technau, 1991). Defects in NSC specification, asymmetric division, or maintenance during early stages are associated with neurodevelopmental deficits (Gonzalez et al., 1990; Rujano et al., 2013; Gogendeau et al., 2015; Schoborg et al., 2015; Ramdas Nair et al., 2016; Vargas-Hurtado et al., 2019; Mannino et al., 2023). For example, loss of the translational repressor and tumor suppressor brain tumor (brat) leads to overexpansion of Type II NSCs, resulting in brain hypertrophy (Bello et al., 2006; Lee et al., 2006).

In this study, we examined the role of Orb2 in neurodevelopment and identified an unexpected requirement for Orb2 in the regulation of larval brain size. We find that Orb2 governs brain size in a dose-dependent manner reliant on its RNA-binding activity, with mutants showing smaller brains consistent with microcephaly. In contrast, orb2 overexpression results in enlarged brains. We identify cell autonomous requirements for Orb2 to regulate brain size through the specification and maintenance of NSCs during embryonic and larval development. We further uncover that orb2 functions upstream of brat, demonstrating that the opposing activities of these RBPs support balanced brain growth.

RESULTS

Orb2 is a homeostatic regulator of brain size

To investigate Orb2 contributions to neurodevelopment, we first quantified brain volumes from age-matched third instar larvae (Link et al., 2019; Hailstock et al., 2023). Strikingly, orb2 mutant brains were ~70% smaller than wild-type (WT) brains (Fig. 1A,B and D; p<0.0001 by one-way ANOVA). These data indicate Orb2 is required for brain growth. To test if orb2 overexpression was sufficient to enlarge larval brains, we drove expression of a UAS-orb2-GFP transgene (Krüttner et al., 2012) using actin-GAL4 (act-GAL4; hereafter, orb2^OE) and confirmed overexpression via western blot (Fig. S1A,B; p<0.01 by two-tailed t-test). Indeed, orb2^OE brains showed an overgrowth phenotype with two-fold increased volumes relative to WT (Fig. 1C,D; p<0.0001 by one-way ANOVA). We conclude that Orb2 regulates larval brain size in a dose-dependent manner.

Figure 1. orb2 is necessary and sufficient for brain growth.

Images show the optic lobe region (dashed circles) from age-matched third instar larval brains of the indicated genotypes stained for DAPI. (A) WT, (B) orb2 null (orb2^Δ36) mutants show reduced brain size, and (C) act-GAL4 driven expression of UAS-orb2 (orb2^OE) results in increased brain volumes, as quantified in (D). Graph shows normalized brain volumes, where each dot represents a measurement of a single optic lobe from N=13 WT, 17 orb2, and 15 orb2^OE brains. Data shown are pooled from 2 (WT, orb2^OE ) or 3 (orb2) independent biological replicates. Representative images of brains expressing (E) UAS-orb2^RNAi (no-GAL4 control) alone or (F–L) in combination with the specified GAL4. Reduced brain volume was observed following depletion of orb2 by (F) ubiquitous expression via act-GAL4, (G) within neuroepithelia cells of the outer proliferation center via c855a-GAL4, in neural stem cells via (H) wor-GAL4 or (I) insc-GAL4, and in (J) and neurons via elav-GAL4. In contrast, orb2 depletion in (K) mature neurons via brp-GAL4 increased brain size, while (L) brain volume was unaffected in glia using repo-GAL4. (M) Quantification of brain volume normalized to controls, where each dot represents a single measurement from a single optic lobe from N= 30 control, 20 act>orb2^RNAi, 25 wor> orb2^RNAi, 18 insc>orb2^RNAi, 47 elav>orb2^RNAi, 27 c855a>orb2^RNAi, 12 brp>orb2^RNAi, and 25 repo> orb2^RNAi pooled from 2 (control, wor, insc, c855a, and repo) or 3 (act, elav, brp) independent biological replicates. Significance was determined by one-way ANOVA. n.s., not significant; *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. Scale bars: 50 μm.

Orb2 is required in neuronal lineages to support brain size

The larval brain comprises distinct cellular identities, including NSCs, marked by Miranda (Mira, (Shen et al., 1997)) or Inscuteable (Insc, (Kraut et al., 1996)); glia, marked by the reversed polarity (repo) gene (Xiong et al., 1994); and neurons, marked by embryonic lethal abnormal vision (elav) (Robinow & White, 1988). To characterize the cellular etiology of orb2-dependent microcephaly, we selectively depleted orb2 in defined cellular lineages within the larval brain, including NSCs, neurons, glia, and the neuroepithelium using shRNA (orb2^RNAi) and a panel of GAL4 drivers (Hrdlicka et al., 2002). We then compared brain volumes from age-matched larval brains to controls that carry the RNAi transgene, but not a GAL4 driver; hereafter, ‘no GAL4’ controls.

We first confirmed that uniform expression of orb2^RNAi with act-GAL4 was sufficient to reduce brain volume (Fig. 1E,F, and M). Depletion of orb2 by RNAi reduced brain volume to a lesser degree than the orb2 null mutation, likely due to the partial depletion of orb2 activity, as shown by immunoblot (~50% reduced; Fig. S1C,D). Similar reductions in brain volume were also detected when orb2 was depleted in neurogenic lineages: neuroepithelia cells, which give rise to medulla neuroblasts of the optic lobe (c855a-GAL4 (Egger et al., 2007; Wang et al., 2011; Shard et al., 2020) p<0.0001), NSCs (wor-GAL4 (Ashraf et al., 1999); p<0.05 or insc-GAL4; p<0.01), or neurons (elav-GAL4; p<0.01), underscoring requirements for orb2 in these lineages to promote brain growth (Fig. 1G–J and M; by one-way ANOVA). In contrast, no difference in brain volume was noted upon depletion of orb2 in glia via repo-GAL4 (Fig. 1L, M; n.s. by one-way ANOVA). These data indicate that Orb2 facilitates the establishment of normal brain size in neural progenitors and their neuronal progeny.

orb2 expression becomes progressively restricted to mature neurons

We next examined where Orb2 is expressed during larval neurodevelopment. To assess Orb2 distributions in live larval brains, we imaged third instar larval brains expressing endogenously-tagged Orb2-GFP, previously shown to functionally restore orb2 activity (Krüttner et al., 2012), relative to an mCherry nuclear reporter driven by cell type specific drivers. Through this approach, we found that Orb2 is not detectable within glia, marked by NP2222-GAL4 or repo-GAL4, nor in the optic lobe neuroepithelium or lamina precursor cells, marked by c855a-GAL4 (Fig. 2A). In contrast, Orb2-GFP was coincident within a subset of smaller insc-GAL4 expressing cells (arrows; Fig. 2A and discussed below). However, at this stage, Orb2 was not detected within third instar larval NSCs, characterized by their distinctive, large nuclei (dashed circles; Fig. 2A). Taken together, these data show that Orb2-GFP positive cells are non-glial descendants of the NSC lineage (i.e., neurons).

Figure 2. Orb2 becomes restricted to mature neurons.

(A) Live images of endogenous Orb2-GFP localization relative to a UAS-mCherry_NLS reporter (magenta) driven by the indicated GAL4 lines: NSCs (insc), neurons (elav^c155), mature neurons (brp), glia (NP2222 and repo), and neuroepithlelial cells (c855a). Dashed circles mark NSCs, asterisks mark glia, and arrows highlight cells positive for both Orb2 and the mCherry reporter; NE, neuroepithelium. (B) Representative immunofluorescence images of third larval instar brains stained for Orb2 (magenta) localization relative to neurons (Elav, yellow) and NSCs (Mira, cyan). Boxed area marks Elav and Orb2-positive neurons. (C) Live images of Orb2 (green) distributions relative to NSCs (insc) or neurons (elav; magenta) throughout the first (1h), second (24h), and third (72h) larval instars; time is hours post larval hatching (PLH). Orb2 exclusion from NSCs (yellow circles) becomes apparent at 24h and 72h PLH. Over time, Orb2 enriches to a subset of elav+ neurons (72h). (D) Seraut dot plot comparing the average expression of the indicated genes (x-axis) across all larval stages relative to known cell type biomarkers (y-axis). (E) UMAP plot of scRNA-seq data (Corrales et al., 2022) comparing cells from the first (1h), second (24h), and third (72h) instar larval brains with normalized expression values of orb2 >3.0 (green), insc >0.75 (cyan), and elav >3.5 (magenta). Scale bars: 10 μm.

Prior work indicates that Orb2 is primarily localized to the cytoplasm of Drosophila neurons, including mature glutamatergic neurons (Si et al., 2003a; Keleman et al., 2007; Li et al., 2016; Stewart et al., 2024). Consistently, Orb2-GFP expression was detected within a subgroup of elav^c155-positive cells determined to be mature larval neurons given strong co-expression with bruchpilot (brp)-GAL4 (Wagh et al., 2006) (arrows; Fig. 2A). Immunostaining confirmed that endogenous Orb2, as detected with anti-Orb2 antibodies, labels a subset of Elav+ neurons, but not NSCs, within third instar larval brains (box; Fig. 2B).

Because loss of orb2 activity in NSCs results in impaired brain growth (Figure 1), we next assayed whether Orb2 enriches within NSCs at earlier stages of development. Low levels of Orb2 were detected within embryonic Mira+ dorsoanterior neural progenitors, consistent with a possible early function in neural lineages (Fig. S1E). Upon examining first, second, and third instar stages, we noted that the distribution of Orb2 becomes progressively more restricted over time (Fig. 2C). At the first larval instar (1 hr post-larval hatching, PLH), Orb2-GFP was detected within subsets of insc- or elav-expressing cells. However, the exclusion of Orb2-GFP from NSCs became apparent in second (24h PLH) and third (72h PLH) instar larval brains (dashed circles; Figure 2C).

To further examine orb2 expression, we analyzed published single cell transcriptomics (scRNA-seq) data of the Drosophila larval central nervous system across the first, second, and third instar stages (Corrales et al., 2022). In accordance with our protein reporter assay, orb2 RNA was found to be expressed within brp+ mature larval neurons (MLN), including Kenyon cells of the mushroom body. In contrast, levels of orb2 RNA within mira+ NSCs were lower (Figure 2D). Examination of UMAP clustering patterns likewise indicated that orb2 mRNA distributions are more similar to elav than the NSC marker insc across larval development (Figure 2E). Thus, during larval development, orb2 and elav are expressed within cells with similar transcriptional profiles.

Together, these data suggest that orb2 expression becomes increasingly restricted within neurons as development ensues. To test if orb2 is necessary in mature neurons to promote brain growth, we drove orb2^RNAi expression using a brp-GAL4. Unexpectedly, depletion of orb2 in mature neurons was not sufficient to reduce brain size, but caused overgrowth (Fig. 1K,M; **p<0.01 by one-way ANOVA). In contrast to brp, elav expresses strongly not only in mature neurons, but also in NSCs and their immediate progeny (O’Neill & Rusan, 2022). We conclude that orb2 is required in neurogenic lineages to regulate brain size. Our data further suggest that Orb2 likely functions within neural progenitors and newly born neurons to promote brain growth. These findings underscore a previously undefined role for Orb2 in neurodevelopment.

Multiple Orb2 domains instruct brain size

orb2 encodes two protein isoforms, Orb2A (predicted molecular weight, ~60kDa) and Orb2B (~75kDa), which differ only in the length of an unstructured N-terminal region with unknown function (Hafer et al., 2011). As previously noted, Orb2B is the predominant isoform in third instar larval brains (Fig. S1C; (Mastushita-Sakai et al., 2010; Hafer et al., 2011; Majumdar et al., 2012)). Because multiple domains within Orb2 support its function as an RBP implicated in translational regulation, we performed structure-function analysis to identify which domain(s) regulate larval brain size (Fig. 3A; (Krüttner et al., 2012; Majumdar et al., 2012; Khan et al., 2015; Stepien et al., 2016).

Figure 3. Multiple Orb2 domains instruct brain size.

(A) Schematic of annotated Orb2 domains: poly-Q domain (glutamine-rich, blue), RNA-recognition motif (RRM, green), and ZZ domain (magenta). (B–G) Representative images of age-matched third larval instar brains of the indicated genotypes stained for DAPI. A schematic of Orb2 structure is shown above each image (astericks denote single AA mutations). (H) Quantification of brain volumes, where each dot represents a single measurement from a single optic lobe from N=13 WT, 25 orb2^ΔQ, 32 orb2^RBD1*, 30 orb2^RBD1/2*, and 9 orb2^ΔZZ brains. Data are pooled from 2 (WT, orb2^ΔQ, and orb2^ΔZZ) or 3 (orb2^RBD1* and orb2^RBD1/2*) independent biological replicates. Significance was determined by Brown-Forsythe and Welch’s ANOVA (H). *p<0.05; **p<0.01; ***p<0.001. Scale bars: 40 μm.

The glutamine-rich (poly-Q) domain enables Orb2 to form stable oligomers through prion-like conformational changes that are required for synaptic plasticity, learning, and memory (Si et al., 2003a; Krüttner et al., 2012; Majumdar et al., 2012). Deletion of the poly-Q domain via the orb2^ΔQ allele (Keleman et al., 2007) resulted in brains ~40% bigger than WT (Fig. 3B,C and G; **p<0.01 by one-way ANOVA). We used CRISPR-Cas9 editing to mutate both RNA recognition motifs (RRMs), targeting the same amino acids previously identified as critical for RNA-binding activity: Y492, F494, R601, and F604 (Hake et al., 1998; Krüttner et al., 2012). Mutation of RRM1 (Y492A, F494A) or both RRMs (Y492A, F494A, R601P, and F604S; see Methods) was sufficient to reduce brain size (Figure 3D,E and G; *p<0.05 or **p<0.01, respectively, by one-way ANOVA). Within CPEB proteins, the conserved C-terminal cysteine and histidine-rich ZZ domain contributes to protein-protein interactions (Legge et al., 2004; Merkel et al., 2013; Oroz et al., 2020; Low et al., 2025). We analyzed mutants lacking the ZZ domain (AA634–704), orb2^ΔZZ, generated by CRISPR-Cas9. As in orb2^ΔQ, loss of the ZZ domain resulted in brain overgrowth (Fig. 3F and G; **p<0.01 by one-way ANOVA). The brain growth responses observed in our mutant panel are not due to changes in Orb2 protein expression, as detected by immunoblotting or previously reported (Figure S1F–I) (Xu et al., 2014; Khan et al., 2015). Taken together, these results indicate that the Orb2 RRM1 is necessary to promote brain growth, while the poly-Q and ZZ domains restrict brain overgrowth.

Orb2-mediated translational activation regulates brain size

Orb2 functions as either a translational repressor or activator depending on its oligomerization status and association with specific cofactors (Fig. 4A) (Si et al., 2003a; Si et al., 2003b; Si et al., 2010; Majumdar et al., 2012; Khan et al., 2015). To determine whether Orb2 regulates brain size through its role in translational control, we depleted its translation cofactors by shRNA (Khan et al., 2015; Stewart et al., 2024). Loss of the repressive cofactor, CG13928, did not alter brain volume (Fig. 4B,C; n.s. not significant by two-tailed t-test). However, depletion of the activating cofactor, CG4612, increased brain size by ~30% relative to controls (Fig. 4D,E; ***p<0.001 by two-tailed t-test). Our findings that both the Orb2 poly-Q domain and CG4612 restrict brain growth are consistent with the hypothesis that Orb2 translational activation activity may limit brain size.

Figure 4. Orb2-mediated translational activation regulates brain size.

(A) Schematic of Orb2 repression and activation complexes when bound to cofactors CG13928 or CG4612 (Khan et. al 2015). Representative images of age-matched no-GAL4 control and act-GAL4 driven (B) CG13928^RNAi or (D) CG4612^RNAi third instar larval brains stained for DAPI and associated quantification (C and E), where each dot represents a measurement from a single optic lobe from (C) Control, N= 44 and CG13828= 35; (E) Control= 31 and CG4612= 47. Data shown are pooled across three replicates. (F) Immunoblot shows levels of Orb2 relative to the actin loading control from control and CG4612^RNAi whole brain extracts, as quantified from triplicated experiments in (G). Significance was determined by Welch’s t-test. n.s., not significant; ***p<0.001; ****p<0.0001. Scale bars: 50 μm. Uncropped blots are available online: https://doi.org/10.6084/m9.figshare.29155505.v1.

Because loss of the translation activation cofactor CG4612 and orb2^OE both resulted in enlarged brains, we predicted that CG4612 loss would increase Orb2 protein levels. Unexpectedly, depletion of CG4612 reduced Orb2 protein levels by ~90% (Fig. 4F,G; ****p<0.0001 by two-tailed t-test). These data suggest that CG4612 promotes the translation of Orb2, consistent with Orb2 autoregulation (Majumdar et al., 2012; Stepien et al., 2016). Alternatively, CG4612 may stabilize Orb2. Taken together, we conclude that the Orb2 translational co-activator, CG4612, limits brain growth.

Orb2 is necessary to maintain the number and size of central brain NSCs

How does Orb2 promote brain growth? To begin to address this question, we tabulated total NSCs in orb2 versus WT larval brains. Loss of orb2 or NSC-specific depletion of orb2 via wor-GAL4 resulted in ~60% fewer Mira+ NSCs per lobe relative to WT controls (Fig. 5A,B; p<0.0001 by one-way ANOVA). Moreover, as compared to WT, NSC volume was also significantly reduced in orb2 mutants and following NSC-specific depletion (Fig. 5C; p<0.0001 and p<0.001, respectively, by one-way ANOVA). These findings are consistent with a requirement of Orb2 in NSC specification and/or maintenance and size.

Figure 5. Orb2 enables NSC specification.

(A) Representative images show NSCs (Mira, greyscale) within age-matched third instar larval brains of the indicated genotypes. (B) Quantification of NSC number from N= 33 WT, 31 orb2, and 9 wor>orb2^RNAi brains. (C) Quantification of NSC volume from WT: N=10 brains and n=55 NSCs, orb2: N=10 brains and n=80 NSCs, and wor>orb2^RNAi: N=9 brains and n=90 NSCs. For B and C, each dot represents a measurement from a single optic lobe. (D) Stage 11 embryos stained for Dpn (magenta and greyscale) to label neural progenitors and DAPI (yellow). (E) Quantification of embryonic neural progenitors, where each dot represents a measurement from a single embryo from N=28 WT and orb2 embryos. Data are pooled from 2 (B and C) or 3 (E) biological replicates. Significance was determined by Brown-Forsythe and Welch’s ANOVA (B and C) or Welch’s t-test (E). ****p<0.0001. Scale bars: (A) 50 μm and (D) 20 μm.

NSC lineages can be identified based on the expression of defined markers. While all NSCs express the transcription factor Deadpan (Dpn), only Type I NSCs also express Asense (Ase)(Boone & Doe, 2008). The Dpn+ Ase- Type II NSCs produce neurons and glia via intermediate neural progenitors, which divide multiple times to form diverse neural niches (Bello et al., 2008; Boone & Doe, 2008; Izergina et al., 2009). While ~20% fewer Type II NSCs were noted in orb2 mutants (p<0.01 by one-way ANOVA), no change was detected upon depletion NSC-specific depletion of orb2 via wor-GAL4, which expresses in both Type I and Type II NSCs (Ashraf et al., 1999; Weng & Cohen, 2015) (Fig. S2 A,B n.s. by one-way ANOVA). Thus, Type II NSCs may be less sensitive to loss of orb2 relative to Type I NSCs.

Consistent with a reduction in orb2 activity, depletion of CG4612 also decreased total NSCs (Fig. S2 C-E). Intriguingly, the NSCs within CG4612^RNAi brains were enlarged (Figure S2F, G; ***p<0.001 by two-tailed t-test). These results suggest that brain size may be uncoupled from NSC number and more closely align with cell size. Taken together, these data imply reduced orb2 activity may disrupt NSC growth processes, perhaps through the misexpression of cell growth-related proteins.

Orb2 enables NSC specification

To elucidate mechanisms underlying NSC loss in orb2 mutants, we first tested the hypothesis that NSCs are eliminated by cell death. We quantified the coincidence of pro-apoptotic cleaved caspase 3 (CC3; (Fan & Bergmann, 2010)) in NSCs marked with the Dpn antibody (Bier et al., 1992; Zhu et al., 2012)). Similar rates of apoptosis were observed in WT and orb2 NSCs, indicating NSC loss occurs by other mechanisms (Fig. S3 A-C; n.s. by t-test). Analysis of death caspase protein-1 (Dcp-1) also failed to uncover elevated rates of NSC death in orb2 mutants (Figure S3 D-F). Another mechanism whereby NSCs may be depleted is via premature differentiation (Cabernard & Doe, 2009; Lai & Doe, 2014; Abdel-Salam et al., 2020). Normally, the pro-differentiation marker Prospero (Pros) is confined to the nucleus of the differentiating ganglion mother cell and absent from NSCs (Doe et al., 1991; Vaessin et al., 1991), while retention of Pros in the NSC nucleus promotes premature differentiation (Lai & Doe, 2014). We quantified the coincidence of Pros with Dpn and detected no significant difference relative to WT, indicating premature differentiation does not direct NSC loss in orb2 mutants (Fig. S3 G-I; n.s. by t-test). Finally, we examined whether orb2 affects mitotic progression, reasoning impaired NSC self-renewal might contribute to NSC loss. However, the mitotic indices in WT and orb2 brains were not significantly different (33.6+4.9% per lobe in WT vs. 36.3+10.1% in orb2 brains; Fig. S3 J-L; n.s. by t-test). Taken together, these data suggest that NSC loss is not due to excess apoptosis, premature differentiation, or impaired proliferation.

Both Type I and Type II NSCs arise from proneural clusters during embryogenesis and serve as the primary source of larval NSCs (Walsh & Doe, 2017; Álvarez & Díaz-Benjumea, 2018). We therefore examined whether orb2 is required during embryonic neurogenesis for NSC development. Indeed, orb2 mutant embryos had ~40% fewer Dpn+ neural progenitors compared to age-matched WT embryos (Fig. 5 D,E; p<0.001 by two-tailed t-test). Taken together, these data underscore a role for Orb2 in the specification and/or maintenance of embryonic neural progenitors and suggest that the reduced brain size in orb2 third instar larvae may result, at least in part, from the reduced number of NSC specified in embryos.

Orb2 activity is required throughout neurogenesis to regulate brain size

Having identified a requirement for Orb2 in embryos, we next assessed the developmental requirement of orb2 to promote brain growth. We employed a temperature-sensitive GAL80 mutant (GAL80^TS) to selectively deplete orb2 in embryos versus larvae. For this strategy, raising progeny at permissive temperatures (18 °C) blocks expression of orb2^RNAi, while growth at the restrictive temperature (29 °C) inactivates GAL80^TS, leading to orb2^RNAi expression and depletion of orb2 (Fig. 6 A; (McGuire et al., 2003; Zhou et al., 2016)). To specifically deplete orb2 in early embryos, embryos were reared at the restrictive temperature and then shifted to the permissive temperature. In contrast, the role of orb2 in larval development was examined by rearing embryos at the permissive temperature until larval hatching, at which point larvae were shifted to the restrictive temperature (Figure 6A, B). Depleting orb2 during embryogenesis or larval development resulted in significantly reduced larval brain volumes relative to controls (Fig. 6C,D; p<0.01 and p<0.0001, respectively, by one-way ANOVA). We conclude that Orb2 activity is required throughout neurogenesis to promote brain growth.

Figure 6. orb2 regulates brain size throughout neurogenesis.

(A) Schematic of the temperature-sensitive depletion of orb2 by RNAi. (B) The temperature shift schemes used in this study: (top row) controls (GAL80^TS, UAS-orb2^RNAi) were reared at 18°; (middle row) to deplete orb2 activity in embryos (embryo>orb2^RNAi), embryos expressing act>GAL4 and GAL80^TS, UAS-orb2^RNAi were reared at 29°C, then shifted to 18°C 24h PLH; (bottom row) to deplete orb2 activity in larvae (larva>orb2^RNAi), embryos expressing the same transgenes were reared at 18° until 24h PLH, then shifted to 29°C. (C) Representative images of age-matched third instar larval brains from controls, embryos raised at the restrictive temperature (embryo>orb2^RNAi), and larvae raised at the restrictive temperature (larva>orb2^RNAi), as outlined in (B), and stained for DAPI. (D) Quantification shows normalized third instar brain volume from N=39 control, 17 embryo>orb2^RNAi, and 15 larva>orb2^RNAi brains, where each dot represents a measurement from a single optic lobe. Data shown are pooled from three replicates. (E) Hatch rate analysis of maternal mutant (M−/Z+) versus maternal and zygotic mutant (M−/Z−) orb2 embryos relative to WT, where each dot represents the average across 4 replicates from N=800 WT or orb2 (M−/Z+) and N=820 orb2 (M−/Z−) embryos. Significance was determined by Brown-Forsythe and Welch’s ANOVA. n.s., not significant; *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. Scale bars: 50 μm.

To further examine Orb2 requirements during embryogenesis, we conducted a hatch rate analysis. While most WT embryos hatch into first instar larvae within 24-hr, about half of orb2 null embryos (M⁻/Z⁻) failed to hatch after 72-hr, indicating Orb2 is required for embryonic viability. To distinguish whether orb2 is maternal-effect lethal or required zygotically, after the maternal-to-zygotic transition, we mated orb2 virgin females to WT males (Tadros & Lipshitz, 2009; Vastenhouw et al., 2019). The resulting orb2/+ (M⁻/Z⁺) embryos hatched at WT levels, indicating that maternal orb2 is not required for viability. We conclude that orb2 is necessary zygotically for embryonic survival, and that escaper larvae require orb2 for proper neurodevelopment (Fig. 6E).

Orb2 is required for neurogenesis

Our data show that orb2 loss or depletion results in fewer total NSCs. Because NSCs give rise to the neurons and glia, we also tabulated the number of these cell types within orb2 mutant larvae relative to controls. Loss of orb2 resulted in ~50% more glia, as compared to WT (Fig. S4 A, B; **p<0.01 by two-tailed t-test). In addition, significantly more orb2 glia were proliferating relative to controls, which may contribute to their expansion (Pereanu et al., 2005) (Fig. S4A and C; **p<0.01 by two-tailed t-test). In contrast, neurons were ~30% reduced in mutants (Fig. S4 D, E; **p<0.01 by two-tailed t-test). Together, these data suggest that orb2 normally functions to promote neuronal versus glia cell fate.

orb2 is epistatic to the tumor suppressor, brat

Brat is a translational repressor and tumor suppressor also implicated in brain size determination. brat loss results in massive expansion of Type II NSCs and brain overgrowth (Bello et al., 2006; Betschinger et al., 2006; Lee et al., 2006; Bowman et al., 2008). Considering their opposing phenotypes, we reasoned that Orb2 might instruct brain size by regulating Brat levels. In support of this model, the brat 3’UTR contains several cannonical CPE sites and was previously identified as a direct Orb2 RNA target (Fig. 7A and S5 A,B; (Hake & Richter, 1994; Hake et al., 1998; Mastushita-Sakai et al., 2010; Stepien et al., 2016)).

Figure 7. orb2 is epistatic to brat.

(A) Schematic of conserved CPE sites in the brat 3’UTR. (B) Immunoblot shows Brat protein levels relative to the actin loading control from WT and orb2 whole brain extracts, as quantified from triplicated experiments in (C). (D) Immunoblot showing Orb2 levels are unchanged in brat [fs1/k06028] transheterozygotes relative to WT, as quantified from triplicated experiments in (E). (F–I) Representative images of age-matched third instar larval brains from WT, orb2/+ hemizygote, brat [fs1/k06028], and brat;orb2/+ mutants stained for DAPI. (J) Quantification of brain volume normalized to WT, where each dot represents a single measurement from N=19 WT, 17 orb2/+, 9 brat, and 8 brat;orb2/+ brains. (K–M) Representative images of age-matched third instar larval brains from orb2^RNAi control, act>orb2^RNAi, and brat^OE rescue (act>orb2^RNAi and UAS-brat-myc) stained for DAPI. (N) Quantification of brain volume normalized to control, where each dot represents a single measurement from N= 17 control, 21 orb2^RNAi, and 29 brat^OE brains. Data are pooled from 2 (J) or 3 (N) biological replicates. (O) Proposed model of orb2-regulated brain growth through brat. Loss of orb2 results in smaller brains, while loss of brat causes brain overgrowth. orb2 functions upstream of brat, likely through regulation of Brat translation. Restoring Brat levels is sufficient to rescue improper brain growth in orb2 mutants. Significance was determined by Brown-Forsythe and Welch’s ANOVA or Welch’s t-test (E). n.s., not significant; *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. Scale bars: 50 μm. Uncropped blots are available online: https://doi.org/10.6084/m9.figshare.29155505.v1.

We aligned a region within the brat 3′UTR conserved across multiple brat isoforms to multiple Drosophila species using the conservation insect track on the UCSC genome browser and found that these CPE motifs were conserved across millions of years of evolutionary distance, hinting they may be functionally important (Fig. 7A; (Kent et al., 2002)). To promote translation, CPEB proteins bind CPE sites within 100 nt of polyadenylation signal (PAS) motifs (i.e., the hexanucleotide motif AAUAAA), which recruit the cleavage and polyadenylation specificity factor (CPSF) to cleave pre-mRNA and add the poly(A) tail, aiding mRNA stability, export, and translation (Mendez et al., 2000). We identified PAS motifs (blue boxes) within the brat 3′UTR proximal to multipe CPEs (red boxes), consistent with a role for Orb2 in translational activation (Fig. 7A and S5A–C). We further identified multiple AU-rich elements (AREs; green boxes), which modulate mRNA decay or translation, in this same region (Fig. 7A and S5C). This arrangement of CPE, PAS, and ARE motifs suggests precise regulation of brat expression to enable normal neurodevelopment (Hake et al., 1998; Mendez & Richter, 2001). These findings also suggest that Brat may be subject to Orb2 translational regulation (Mastushita-Sakai et al., 2010; Santana & Casas-Tintó, 2017). Consistent with this idea, several Orb2 reponse elements (OREs), as recently identified in Orb2 translational targets (Low et al., 2025), are also present in the brat 3’UTR, further suggestive of a role in translational regulation.

Orb2 represses the expression of a Brat translation reporter in Drosophila S2 cells (Mastushita-Sakai et al., 2010). To uncover the relationship between Orb2 and Brat within larval brains, we first examined their relative distributions. As previously reported, Brat is ubiquitously distributed throughout the CNS. In mitotic NSCs, however, Brat colocalizes with the adapter protein Mira as a crescent at the basal cortex (arrowheads, Fig. S5A; (Bello et al., 2006; Betschinger et al., 2006; Lee et al., 2006)). In contrast, Orb2 was undetectible in NSCs or within Mira+ basal crescents (Fig. S5A). Intriguingly, while Orb2 and Brat did partially overlap within differentiated neurons, regions of high Orb2 enrichment contain markedly less Brat, consistent with a potential role for Orb2 in modulating Brat levels (dashed region, Fig. S5A).

To assess if Orb2 regulates Brat expression in larval brains, we examined steady-state Brat levels in WT versus orb2 mutants by western blot. Loss of orb2 reduced Brat protein levels from larval brain extracts by ~50% (Fig. 7B,C). By comparison, Orb2 levels are unchanged in brat larval brain lysates (Fig. 7D,E). These data are consistent with a role for Orb2 in the translational control of Brat.

We next assayed whether orb2 and brat genetically interact to regulate brain size. Consistent with a strong synthetic lethal interaction, we were unable to recover brat; orb2 homozygous double-mutant larvae. Nevertheless, we examined whether hemizygosity of orb2, previously shown to reduce orb2 mRNA and protein levels by ~75% and ~60%, respectively (Gilmutdinov et al., 2021), was sufficient to modify the brat brain overgrowth phenotype. orb2/+ hemizygotes showed reduced brain volumes relative to WT, consistent with a dose-dependent effect of orb2 on brain size (Fig. 7J, F and G; **p<0.01 by ANOVA). In contrast, loss of brat led to the expected increase in brain size (Fig. 7H,J; ***p<0.001 by ANOVA). Critically, hemizygosity of orb2 was sufficient to supress the brain overgrowth of brat mutants (brat; orb2/+; Fig. 7F-J; *p<0.05 by ANOVA), indiciating that orb2 is epistatic to brat in the regulation of brain size.

Taken together, our data support a model whereby Orb2 regulates the translation of Brat to modulate brain growth. Such a model predicts that restoring Brat levels in orb2 mutants would rescue orb2-dependent microcephaly. Consistent with this idea, overexpression of a brat transgene in an orb2-depleted background fully restored brain size (orb2^RNAi, brat^OE; Fig. 7K-N; ****p<0.0001). We conclude that Orb2 functions upstream of Brat to balance brain growth (Fig. 7O).

DISCUSSION

CPEB proteins like Orb2 are important for learning and memory (Si et al., 2003a; Keleman et al., 2007; Krüttner et al., 2012; Majumdar et al., 2012; Khan et al., 2015; Hervás et al., 2016; Li et al., 2016; Oroz et al., 2020; Kozlov et al., 2021; Kozlov et al., 2023). Here, we define a novel and dose-dependent requirement of Orb2 in regulating brain size. The dynamic regulation of orb2 expression throughout neurogenesis likely ensures proper orb2 dosage required for neurodevelopment. We find that Orb2 functions in early neural progenitors and elav+ neurons to promote brain growth through a mechanism also requiring its RNA-binding activity. We further uncover Orb2 translational activity regulates brain size, as shown by loss of the Orb2 poly-Q domain or the translation co-activator, CG4612.

The mechanism by which CPEB proteins regulate target transcripts includes modulating mRNA translation, stability, and localization through CPE elements to influence gene expression. Like other CPEB proteins, the Orb2 RRMs are required for RNA-binding activity (Hake et al., 1998; Krüttner et al., 2012; Stepien et al., 2016). However, we unexpectedly uncovered opposing phenotypes in response to deleting different Orb2 functional domains. While the RRMs support brain growth, consistent with the orb2 null phenotype, the poly-Q and ZZ domains apparently restrict overgrowth. Given that Orb2 autoregulates its own translation via CPE motifs in its 3’ UTR, the various Orb2 domains may play conformationally significant and conserved roles in forming complexes with cofactors that influence expression of neurogenic genes or pro-growth factors (Tan et al., 2001).

Given the varying phenotypes resulting from our structure-function analysis, we speculated that Orb2 may promote the translation of a repressor of brain growth. Such a factor would act in opposition to Orb2, which normally promotes brain growth. This model would establish a feedback loop to normalize brain growth, consistent with a role for Orb2 in the homeostatic regulation of brain size.

Our data suggest that Orb2 regulates brain size through translational regulation of brat mRNA (Figure 7O). This model is supported by prior work demonstrating that Orb2 binds brat mRNA directly at defined CPE sites within the brat 3’UTR (Stepien et al., 2016). We further identify hexanucleotide CPSF binding sites proximal to conserved CPE sites necessary for CPEB-dependent translational activation (Mendez et al., 2000). We also show Orb2 regulates Brat protein levels. Finally, we demonstrate that orb2 is epistatic to brat and that restoring Brat levels by expression of a UAS-brat transgene is sufficient to rescue orb2-dependent microcephaly.

In the early embryo, Brat represses hunchback (hb) translation, critical for axial patterning (Sonoda & Wharton, 2001). In neurogenic embryos, Brat segregates asymmetrically with Mira in dividing NSCs to promote neuronal differentiation (Betschinger et al., 2006). In larvae, Brat limits NSC proliferation by regulating various transcription factors, including dMyc, and works alongside Pros to repress NSC renewal and promote differentiation (Betschinger et al., 2006). While our study suggests Orb2 regulates Brat levels to ensure proper NSC specification, growth, and neurogenesis, other factors must cooperate with Orb2 to tune Brat expression. orb2 loss reduced, but did not eliminate, Brat. Likewise, we did not detect changes in NSC proliferation in orb2 mutants. Further study is necessary to identify additional regulators of brat activity and how such factors coordinate with Orb2.

Interestingly, temperature-shift experiments uncovered requirements for orb2 function during both the embryonic and larval neurogenic stages. Indeed, our data suggest that Orb2 may be required for NSC specification in embryos. This early requirement for orb2 likely establishes the NSC pool required at later developmental stages. During normal development, late embryonic NSCs enter quiescence. In early larval stages, however, NSCs reactivate, an early step of which involves the doubling in NSC size (4–5 μm to 10–15 μm; (Truman & Bate, 1988; Prokop & Technau, 1991; Ding et al., 2016)). Consequently, the dividing NSCs normally produce daughter cells with marked cell size asymmetries: a larger, self-renewing stem cell and a smaller daughter cell fated for differentiation. In addition to being reduced in number, orb2 NSCs are also smaller. These phenotypes contrast with brat mutant NSCs, which fail to inhibit self-renewal in the daughter cell. Instead, both cells grow and proliferate, leading to brain tumors (Betschinger et al., 2006; Lee et al., 2006). Taken together, these findings suggest that Brat and Orb2 likely have distinct functions in NSCs versus mature neurons.

One way in which Brat regulates cell growth is through coordination of ribosome biogenesis and translational control (Gui et al., 2023). One possibility raised by our study is that orb2 may be required for NSC growth during NSC reactivation. NSC reactivation requires feeding and nutrient sensing (Britton & Edgar, 1998). Because we used an age-matching paradigm dependent upon larval intake of blue-colored food throughout this study (see Methods and (Hailstock et al., 2023)), any orb2-dependent response to NSC reactivation must occur post-feeding. We speculate that Orb2 functions in either the transmission of the cell growth signals themselves or in the physical enlargement of the NSCs. These processes may involve the recently defined role for Orb2 in enabling the translation of rare-codon-enriched mRNAs (Stewart et al., 2024). Further investigation is required to elucidate precisely how Orb2 influences NSC specification and growth.

MATERIALS AND METHODS

Fly stocks

The following strains and transgenic lines were used: y¹w¹¹¹⁸ (Bloomington Drosophila Stock Center (BDSC) #1495) was used as the WT control unless otherwise noted. orb2 brains were isolated from the homozygous null allele, orb2^Δ36 (P. Schedl, Princeton University (Xu et al., 2012). orb2 null embryos were sorted against a GFP balancer (BDSC #58479). Tissue-specific depletion of orb2 was by orb2^RNAi (P{TRiP.JF02376}attP2; BDSC #27050). UAS-mCherry_NLS} (BDSC #38425) was used to visualize GAL4 expression pattern, and orb2[+GFP] (BDSC #94349; gift from K. Keleman, (Krüttner et al., 2012)) is a GFP knock-in expressing Orb2-GFP at the native orb2 locus and with endogenous orb2 regulatory elements was used to detect Orb2 localization. A UAS-orb2B-GFP transgene was used in overexpression experiments (gift K. Keleman, (Krüttner et al., 2012)). UAS transgenes were driven by wor-GAL4 (P{wor.GAL4.A}2; BDSC #56555), insc-GAL4 (P{GawB}insc^Mz1407; BDSC #8751), elav-GAL4 (P{w[+]elav.GAL4}), elav^c155-GAL4 (P{GawB}elav^C155; BDSC #458), repo-GAL4 (P{GAL4}repo; BDSC #7415), c855a-GAL4 (P{GawB}c855a; BDSC #6990), brp-GAL4 (gift from A.-S. Chiang and K.-L. Feng, National Tsing Hua University; (Chen et al., 2023)), or act-GAL4 (P{w[+mC]=Act5C-GAL4}; BDSC #4414). Temperature-sensitive GAL80 (P{GAL80^ts}10; BDSC #7108) was recombined with orb2^RNAi. orb2^ΔQ was used to assess functional requirement of the poly-Q domain (gift from P. Schedl, Princeton University; (Keleman et al., 2007). CG4162^RNAi (P{TRiP.HMJ22843}attP40; BDSC #60473) and CG13928^RNAi (P{TRiP.HM05031}attP2; BDSC #28545) were used for cofactor depletion. brat^fs1 and brat^k06028 (gifts from C.Y. Lee, University of Michigan; (Arama et al., 2000)) were used to generate brat^fs1/brat^k06028 trans-heterozygote animals. brat^k06028 is a null allele (Sonoda & Wharton, 2001). brat^fs1 is a loss-of-function allele defined by a G774D mutation in the NHL domain (Sonoda & Wharton, 2001) and was recombined onto the orb2 chromosome to generate brat^fs1, orb2^Δ36 recombinant animals. UAS-brat^FL:myc (gift from C.Y. Lee, University of Michigan; (Komori et al., 2014) was recombined with orb2^RNAi to overexpress brat. All strains were maintained on Bloomington formula cornmeal-agar media (Lab-Express, Inc.; Ann Arbor, MI) at 18, 25, or 29°C in light and temperature-controlled incubators.

Generation of orb2 mutant alleles

orb2^RBD1* and orb2^RBD1/2* mutants were generated by CRISPR using the Scarless approach (Bier et al., 2018) using the method of (Port & Bullock, 2016). Multicistronic target amplicons were generated by PCR and inserted into BbsI-digested pCFD5 (Addgene #73914) by Gibson assembly using the following 5’–3’ forward (F) and reverse (R) primers with gRNA underlined:

• 1F: GCGGCCCGGGTTCGATTCCCGGCCGATGCACAGCAGGAAGGCATATCCCTGTTTTAGAGCTAGAAATAGCAAG and
• 2R: ATTTTAACTTGCTTTTCTAGCTCTAAAACCGTAGTCAGCCAATTTCCTCTGCACCAGCCGGGAATCGAACCC

For homologous recombination, gene blocks were synthesized (Twist Bioscience) then sequentially ligated into pHD-DsRed-Scarless (Drosophila Genomics Resource Center (DGRC), #1364). DNA plasmids were injected by BestGene (Chino Hills, CA) into attP40 nos-Cas9 (BDSC#78781 or TH00788.N). The PAM sequences at both target cut sites were mutated with silent mutations to prevent recutting. Two unique CRISPR lines were recovered and sequence validated using primers:

• 3orb2RRMF: GCTGGGTGGAAATGGTGG and
• 4orb2RRMR: GCGTGTGTTGTCTCTAATGGT

Alanine substitutions were designed for four amino acids (two in each RRM), following the design of (Krüttner et al., 2012). One isolate had mutations at all expected sites despite alternate amino acid substitutions in RRM2: Y492A, F494A, R601P, and F604S. One mutant had mutations only in RRM1: Y492A and F494A. Mutant lines were backcrossed to alphatub-piggyBacK10 (BDSC #32070) to excise the DsRed marker, then maintained over the TM6C, Sb,Tb balancer chromosome.

CRISPR mutant orb2^ΔZZ was generated according to the method of (Gratz et al., 2014) as described in (Low et al., 2025). The orb2^ΔZZ mutation deletes AA634–704 spanning the ZZ domain of Orb2B, and the CRISPR mutants were screened and confirmed by DNA sequencing.

Hatch rate analysis

Embryos from y¹w¹¹¹⁸ control, orb2^Δ36, and heterozygotes were collected on yeasted grape juice agar plates and ~200 embryos were transferred at 0–2h AEL to fresh plates and aged for 72 h at 25°C. Unhatched embryos were counted from each plate and compared to controls. orb2^Δ36 virgins and males were mated to generate M-Z− embryos, while orb2 heterozygotes (M-Z+) were derived by crossing null virgin females to control males, to assess maternal versus zygotic contributions. Data presented are mean ± SD from three biological replicates.

Larval age matching

Larval samples were synchronously staged as described by (Hailstock et al., 2023). Briefly, Drosophila adults were allowed to seed grape juice agar plates supplemented with yeast paste for 48 h at 25°C. Plates were changed every 24 h, after which flies were allowed 30 minutes to clear previously fertilized eggs. Embryos were collected for 4 h on a fresh plate and incubated at 25°C until first instar larvae hatched. First instar larvae were transferred into vials containing Drosophila culturing medium supplemented with yeast paste dyed with bromophenol blue, then incubated at 25°C until crawling third instar larvae emerged. Age-matched larvae that both displayed foraging behavior and partial gut clearance of the blue yeast paste were then selected for dissection (Truman & Bate, 1988).

Temperature-shift assay

For temperature-shift experiments, we used GAL80^TS to modulate expression of orb2^RNAi. Animals were reared at 18 °C to inhibit orb2^RNAi expression or at 29 °C allow orb2^RNAi expression. act-GAL4 virgins were mated to GAL80^TS, orb2^RNAi recombinant animals in cages for 48 h at 18°C. At 24–48h AEL, orb2 expression was inhibited in first instar larvae by shifting animals to 29°C. To inhibit orb2 activity in embryos, the same parental genotypes were mated at 29°C for 48 h prior to harvesting 4–7 h embryos. Embryos were maintained at 29°C for 24–48h, and first instar larvae were shifted to 18°C to allow orb2^RNAi expression. Age-matched third instar larvae were selected 48–72 h after temperature-shift as described in (Hailstock et al., 2023). Experimental groups were compared to GAL80^TS, orb2^RNAi controls lacking GAL4 raised at 18°C.

Immunofluorescence

Age-matched crawling third instar larval brains were used for dissections and prepared for immunofluorescence as previously described (Lerit et al., 2014). Brains were dissected in Schneider’s Drosophila Medium (ThermoFisher Scientific, #21720024), fixed in 9% paraformaldehyde for 15 min, blocked in PBT buffer [Phosphate Buffered Saline (PBS) supplemented with 1% bovine serum albumin (BSA) and 0.1% Tween-20] for one hour at room temperature prior to overnight incubation in primary antibodies in PBT with nutation at 4°C. Samples were further blocked with modified PBT (2% BSA, 0.1% Tween-20, and 4% normal goat serum (NGS) in PBS) for one hour before incubation for two hours at room temperature with secondary antibody and DAPI.

Embryos were fixed in a 1:4 solution of 4% paraformaldehyde:heptane for 20 min and devitellinized in methanol (Rothwell & Sullivan, 2007). Fixed embryos were rehydrated, blocked in PBTx buffer (PBS supplemented with 0.3% Triton X-100 and 1% BSA), and incubated overnight at 4°C with primary antibodies diluted in PBTx supplemented with 2% NGS. After washing, embryos were further blocked in PBTx supplemented with 2% NGS and incubated for 2 h at room temperature with secondary antibodies and DAPI (10 ng/ml, ThermoFisher Scientific, #D1306). All samples were mounted in Aqua-Poly/Mount (Polysciences Inc., #18606–20) prior to imaging.

The following primary antibodies were used: guinea pig anti-Orb2 (1:1000, gift from K. Si, Stowers institute for Medical Research; (Mastushita-Sakai et al., 2010)), mouse anti-elav 9F8A9 (1:500, DSHB), mouse anti-Orb2 2D11 and 4GS (1:25 each, DSHB), mouse anti-repo 8D12 (1:2000, DSHB), mouse anti-Prospero MR1A (1:500, DSHB), rabbit anti-DCP-1 (1:500, Cell Signaling Technology #9578), rabbit anti-Dpn (1:500, gift from Y.N. Jan, University of California San Francisco; (Bier et al., 1992)), rabbit anti-pH3 (1:1000, Millipore #05570), and rat anti-Mira (1:500, Abcam, #ab197788). For embryos, guinea pig anti-Dpn (1:500, gift from J. Skeath, Washington University of St. Louis; (Skeath et al., 2017)) was used. Secondary antibodies were Alexa Fluor 488, 568, or 647 (1:500, Molecular Probes) and DAPI (10 ng/ml, Invitrogen).

Microscopy

Images were acquired on a Nikon Ti-E system fitted with a Yokagawa CSU-X1 (Yokogawa Corp. of America) spinning disk head, Hamamatsu Orca Flash 4.0 v2 digital complementary metal oxide-semiconductor (CMOS) camera (Hamamatsu Corp.), Perfect Focus system (Nikon), and Nikon LU-N4 solid state laser launch (15 mW; 405, 488, 561, and 647 nm) The following Nikon objectives were used: 100× 1.49-NA Apo Total Internal Reflection Fluorescence oil immersion, 40× 1.3-NA Plan Fluor oil immersion, and 20× 0.75-NA Plan Apo. Images were acquired at 25°C through Nikon Elements AR software on a 64-bit HP Z440 workstation (Hewlett-Packard).

Live Imaging

Age-matched larval brains were dissected as described in (Lerit et al., 2014) and prepared for live imaging as described in (Hannaford et al., 2019). Briefly, explanted brains were transferred to a drop of fibrinogen-supplemented media (10mg/mL fibrinogen (Sigma Aldrich #341573) added to Schneider’s Drosophila Medium (ThermoFisher Scientific #21720024)) in a glass bottom imaging dish (MatTek #P35G-1.5–14-C). The brain was oriented with the dorso-anterior central brain contacting the glass, and the clot was polymerized by the addition of thrombin at 100μg/mL (Sigma Aldrich #T7513). After polymerization, 500 μL of medium was added on top of the clot for the duration of imaging.

Image Analysis

Images were resolved using Fiji (National Institutes of Health; (Schindelin et al., 2012)), to separate or merge channels, crop regions of interest, generate maximum-intensity projections, and adjust brightness and contrast. Adobe Photoshop and Adobe Illustrator software (Adobe, Inc.) were used to assemble images into panels.

Larval brain volumetric analysis

Volume measurements were obtained in Imaris 9.9, 10.1.0 and 10.1.1 software (Oxford Instruments) using the 3D surface tool to define the optic lobe region of interest on DAPI-visualized brains and statistics function to calculate surface volume. Brain volume is represented by measurement from one lobe per brain and normalized to the control mean (Link et al., 2019). NSC and neuroepithelium volume measurements were obtained using the 3D surface tool to detect objects with machine learning function to subtract background. Cell volume was calculated using the statistics function. The experimenter was blinded to genotype prior to analysis.

Cell number and volume

For embryos, 4–7 h WT and orb2 embryos were collected and stained with anti-Dpn antibodies. Maximum projected images of Dpn+ cells were counted in Fiji using the Cell Counter (O’Brien et al., 2016) plugin. The experimenter was blinded to genotype prior to analysis.

For larvae, NSCs were identified as large, Mira+ cells located in the central brain. NSC counts are reported from a single optic lobe per sample. Mitotic index was calculated as the number of Mira+, pH3+ NSCs per total Mira+ NSCs. Mitotic events were identified by the presence of pH3 signal. Single-channel .tif raw images for mitotic event, glia, and neurons were segmented in three dimensions using Python scripts adapted from the Allen Institute for Cell Science Cell Segmenter (Chen et al., 2020) as described in (Ryder & Lerit, 2020). To score cell death, events were identified by the presence of Dcp-1 in Mira+ cells using the Cell Counter (O’Brien et al., 2016) plugin in Fiji. The experimenter was blinded to genotype prior to analysis.

To quantify neurons per brain lobe, age matched WT and orb2 third instar larvae expressing elav^C155-GAL4 and UAS-mCherry_NLS were live imaged. Images were processed with Imaris 9.9 using the spot tool to detect objects and the number of objects were counted with the statistics function. The experimenter was blinded to genotype prior to analysis.

Single cell RNA-seq analysis

Single cell RNA sequencing data was downloaded from GEO Dataset GSE135810. Count matrices were imported into R, filtered, and quality controlled as described in the Github repository companion to the original publication (Corrales et al., 2022) using SeuratV5 (Hao et al., 2024). Cells were grouped into 67 clusters and then assigned a cell type by comparing the positively differentially expressed genes to known biomarkers as defined by Corrales et al. Supplementary Spreadsheet 19. Visualization was completed in R using Seurat visualization functions.

Immunoblotting

Larval brain extracts were prepared from 10–20 crawling third instar larva dissected in Schneider’s Drosophila Medium, removed of imaginal discs, transferred to fresh media, then rinsed once in cold PBST (PBS with 0.1% Tween-20). Samples were homogenized on ice in 30 μL of fresh PBST using a cordless motor and plastic disposable homogenizer, supplemented with 5X SDS loading buffer to 2X final concentration, boiled for 10 minutes at 95 °C, then stored at −20 °C or immediately resolved on a commercial 10% polyacrylamide gel (Bio-Rad, #4568023). Protein concentration was measure using the Pierce BCA Protein assay kit (Thermo Scientific, #23227) as per manufacturer’s instructions and measured on Synergy H1 Multi-Mode plate reader. Proteins were either transferred to a 0.2 μm nitrocellulose membrane (Amersham Protran, #10600001) or PVDF membrane (Bio-Rad, #1620177) by wet transfer in a buffer containing 25 mM Tris-HCl pH 7.6, 192 mM glycine, 10% methanol, and 0.02% SDS at 4 °C or semi-dry transfer by Trans-Blot Turbo Transfer System (Bio-Rad). Membranes were blocked in 3% BSA or 5% dry milk in TBST (Tris-buffered saline, 0.05% Tween-20), washed with TBST, and incubated overnight at 4 °C with primary antibodies. After washing with TBST, membranes were incubated for 2 h in secondary antibodies diluted 1:2500 in TBST. Bands were visualized with Clarity ECL substrate (Bio-Rad, #1705061) on a ChemiDoc imaging system (Bio-Rad). Densitometry was measured in ImageJ and protein levels were normalized to the loading control as described in (Stael et al., 2022).

The following primary antibodies were used: mouse anti-Actin JLA20 (1:1000, DHSB), mouse anti-Brat 3A9 (1:1000, gift from W.Shi, Institute of Genetics and Developmental Biology, China; (Shi et al., 2013)), and mouse anti-Orb2 2D11 and 4G8 (1:25 dilution each; DSHB). Secondary antibodies: goat anti-mouse HRP (1:2500 or 5000, ThermoFisher #31430) and goat anti-rabbit HRP (1:5000, ThermoFisher # 656120). Densitometry was measured in Adobe Photoshop and protein levels are normalized to the loading control.

Statistical methods

Data were plotted and statistical analysis was performed using GraphPad Prism software. Data were first tested for any outliers using a ROUT test with a Q= 1% and normality using the D’Agostino and Pearson normality test. This was followed by Student’s two-tailed t test, ANOVA, Fisher’s exact test, or the appropriate nonparametric tests. Data were plotted with mean ± SD displayed. Data shown are representative results from at least two independent replicate experiments as indicated in the figure legends.

Data availability

All code for object detection and measurements (Fig. S4 A-C, G-L) are available on Github at https://github.com/pearlryder/rna-at-centrosomes (Ryder & Lerit, 2020; Lucas et al., 2021). Full-size, replicated immunoblots are available to view on FigShare: https://doi.org/10.6084/m9.figshare.29155505.v1.

Data Availability Statement

Uncropped immunoblots are available online: https://doi.org/10.6084/m9.figshare.29155505.v1.