A novel adipose loss-of-function mutant in Drosophila

ABSTRACT

To identify genes required for brain growth, we took an RNAi knockdown reverse genetic approach in Drosophila. One potential candidate isolated from this effort is the anti-lipogenic gene adipose (adp). Adp has an established role in the negative regulation of lipogenesis in the fat body of the fly and adipose tissue in mammals. While fat is key to proper development in general, adp has not been investigated during brain development. Here, we found that RNAi knockdown of adp in neuronal stem cells and neurons results in reduced brain lobe volume and sought to replicate this with a mutant fly. We generated a novel adp mutant that acts as a loss-of-function mutant based on buoyancy assay results. We found that despite a change in fat content in the body overall and a decrease in the number of larger (>5 µm) brain lipid droplets, there was no change in the brain lobe volume of mutant larvae. Overall, our work describes a novel adp mutant that can functionally replace the long-standing adp⁶⁰ mutant and shows that the adp gene has no obvious involvement in brain growth.

KEYWORDS:

• Drosophila
• adipose
• neurodevelopment
• fat body
• lipogenesis
• CRISPR

Introduction

Many genes associated with human disease were discovered and studied through model organisms, including genes required for brain growth [1–4]. Human variants in critical neurodevelopmental genes can cause microcephaly, which is a rare neurodevelopmental disorder characterized by a reduced occipital frontal circumference (OFC) of two standard deviations (SD) or more below the mean for a child’s age and sex [5–7]. Half of all known causes of microcephaly are from genetic mutations, suggesting that additional cases of genetic microcephaly might be found by assessing genes for neurodevelopment defects in model organisms [3,4,6–8]. We hypothesized that by using a reverse genetics approach, we would isolate novel genes required for neurodevelopment and linked to human disease. We used commercially available Drosophila RNAi lines to knock down candidate genes in developing brains and screened third-instar larval Drosophila for brain size differences. Through this screen, we identified a potentially novel candidate gene in neurodevelopment, adipose (adp).

Adipose was first characterized in the 1950s when a wild strain of fly, adp⁶⁰, was isolated in Africa [9]. These adp⁶⁰ flies had a 23-base pair deletion in the middle of the adp gene, leading to increased lipogenesis [10–12]. Adult adp⁶⁰ flies were resistant to starvation compared to Oregon R controls, and both the larvae and adults were shown to have higher triglyceride levels [10–12]. Further work validated that adp functions as an inhibitor of lipogenesis, and the adp⁶⁰ flies carried a loss of function allele, demonstrating an increase in fat storage. While the lipogenesis phenotypes of adp⁶⁰ have been well characterized, it is currently unknown whether adp plays a role in neurodevelopment.

The importance of fat during brain development has been well demonstrated across species. In vertebrates, fat is necessary to produce the myelin sheath covering axons. In the developing fly brain, neural stem cells, called neuroblasts, reside in stem cell niches protected by lipid droplets [13,14]. No work has shown whether adp is necessary to produce brain lipid droplets in larval Drosophila, but we hypothesized that adp is necessary for neurodevelopment and is likely to function through lipid droplet production in the brain. In this paper, we generated a novel adp mutant that behaves as a loss of function mutant that replicates high-fat content as previously shown. We also demonstrate that adp is not necessary for neurodevelopment in the fly nor for lipid droplet production in the brain.

Results

In an effort to identify conserved pathways required for brain growth and novel players in neurodevelopment, we screened a collection of Drosophila genes for function in the brain using in vivo RNAi. We crossed UAS-RNAi flies to either inscuteable-GAL4 (insc-GAL4) [15] or neuronal Synaptobrevin-GAL4 (nSyb-GAL4) to knock down genes of interest in neuronal stem cells or post-mitotic neurons, respectively. Developing brains from late third-instar larvae were collected for volumetric analysis to assess brain lobe growth [8,16]. Knockdown of adipose (adp) in both neuronal stem cells and post-mitotic neurons resulted in significantly reduced brain lobe volume compared to the control knockdown of GFP (Figure 1a–f).

Figure 1. RNAi knockdown of adp but not the presumed adp⁶⁰ mutant results in significantly reduced brain lobe volume. Stereroscope images of third-instar larval brains just prior to pupation with (A, D) GFP knockdown (VALIUM22-EGFP.ShRNA.1) and (B, E) adp knockdown (TRiP.HMC006600) in neural stem cells (A-C, insc-GAL4) or post-mitotic neurons (D-F, nSyb-GAL4). (C) Brain lobe volume of neural stem cell knockdown of GFP (control) or adp in third-instar larvae; each dot represents one brain lobe (n = 8–10). Knockdown of adp in neural stem cells results in significantly reduced brain lobe volume compared to control (independent t-test, t = 4.932, df = 16, p = 0.0002). (F) Brain lobe volume of post-mitotic neuronal knockdown of GFP (control) or adp in third-instar larvae, each dot represents one brain lobe (n = 9–10). Knockdown of adp in post-mitotic neurons results in significantly reduced brain lobe volume compared to control (independent t-test, t = 3.694, df = 17, p = 0.0018). Stereoscope images of third-instar larval brains just prior to pupation of (G) Oregon-R and (H) the presumed adp⁶⁰ fly stocks. (I) Brain lobe volume of Oregon-R (control) or adp⁶⁰ third-instar larvae, each dot represents one brain lobe (n = 8–9). The presumed adp⁶⁰ mutant does not differ in brain lobe volume compared to an Oregon-R control (independent t-test, t = 0.9658, df = 15, p = 0.3495). Sequencing of the presumed adp⁶⁰ stock showed no mutation in the adp gene.

Generally, total larval number in each vial is not controlled in our standard RNAi crosses. However, competition for resources can influence animal and tissue growth in some mutants and conditions [17]. To test whether adp RNAi brain volume was affected by growth conditions, we determined how larval crowding modified our results. Since neuronal adp knockdown had the most robust results, we looked at brain size in a medium animal density (50–65 embryos) and high animal density (150 embryos) condition and found that control (GFP RNAi) brain volume was not different between crowding conditions (Supplemental Figure S1a), indicating that brain volume in general is not affected by crowding. However, adp knockdown animals had reduced brain volume that was significantly smaller than control in the high animal density condition (Supplemental Figure S1c), but not the medium animal density condition (Supplemental Figure S1b). Therefore, Adp regulation of brain growth is affected by crowding and competition for resources. Together, these results indicate that adp is involved in neurodevelopment.

To support the idea that Adp is important during neurodevelopment, we determined which cell types express adp in the larval brain. We generated 36 single molecule fluorescent in situ hybridization (smFISH) DNA primary probes spanning the adp mRNA region and a single fluorescently labelled secondary probe to use in larval brain RNA in situ. adp RNA appears to be expressed throughout the brain at a moderate level with no cell specificity (Supplemental Figure S2a,b). To ensure the documented signal was due to the presence of RNA and was not background, we treated larval brains with RNase prior to smFISH and found the signal was almost completely abolished (Supplemental Figure S2c). These results indicate the smFISH signal in (Supplemental Figure S2a,b) is RNA and likely the adp transcript. Interestingly, our adp smFISH probes also produce bright puncta in the nucleus of some cells which are not eliminated after treatment with RNase. These puncta probably correspond to the adp DNA locus and serve as an ideal control for hybridization.

To verify that reduced brain lobe volume resulted from a loss of adp and not off-target RNAi effects, we assessed a known loss-of-function mutation called adp⁶⁰ [9,10]. Adult male adp⁶⁰ flies are resistant to starvation, and both adult males and third-instar larvae have increased triglyceride content compared to wild-type animals, indicating that Adp functions to inhibit lipogenesis [10–12]. However, third-instar larval adp⁶⁰ brains were not significantly different in size compared to Oregon-R control brains (Figure 1(g–i)). This negative result could mean that adp is either unnecessary for brain development or the mutation in adp⁶⁰ may no longer be present.

We aimed to verify the mutation by sequencing the presumed adp⁶⁰ fly with the same primers described in the initial characterization [11]. The published mutant contains a 23-base pair deletion that removes nucleotides 1153 to 1176 in exon 2, resulting in an early stop codon in the predicted protein. Sanger sequencing established that the adp genomic sequence in the available adp⁶⁰ mutants matches the reference genome with no deletion present in both the main and backup adp⁶⁰ Bloomington stocks, indicating that the presumed adp⁶⁰ stock is no longer a mutant allele of adp.

Due to the lack of an available adp mutant, we sought to generate our own loss-of-function mutant using the TRiP-CRISPR toolbox [18–28]. We crossed flies expressing Cas9 in the germline (nanos-Cas9) together with flies expressing single guide RNA for adp targeting base pairs 221–243 in the first exon. F1 males are expected to have germline mutations in adp, and F2 founder flies were isolated to generate 10 independent mutant adp lines. Initially, three lines were genetically characterized using PCR and Sanger sequencing of the adp locus. Surprisingly, all three contained small INDELs in the guide RNA target region, resulting in early stop codons in the predicted protein. adp INDEL lines were crossed to w¹¹¹⁸ for three generations to remove potential off-target mutations. The remaining 7 lines have not been sequenced. We decided to focus on a single mutant, which we named adp¹. The adp¹ mutant contains a frameshift mutation (c.237_238insT) predicted to result in an early stop codon in the first exon (p.Asp134Ter) (Figure 2b). The adp¹ mutant is homozygous fertile and viable. Due to how early the predicted truncation mutation appears, we predict adp¹ to be a loss-of-function mutation.

Figure 2. adp¹ has higher fat content than controls. (a) Genomic region of adp and location of duplication line used for rescue. (b) Sequencing of fly adp¹ showed a frameshift mutation (c.237_238insT) resulting in an early stop codon p.Asp134Ter marked with the asterisk. (c) Buoyancy assay comparing control (w¹¹¹⁸), adp¹, and adp¹ plus the genomic duplication noted in (A), each dot represents a replicate experiment of around 20–30 larvae per genotype (total n sizes from all replicates: control n = 100, mutant n = 139, rescue n = 125). Loss of function adp¹ larvae float at a lower density than control larvae, while a genomic duplication of fly adp is able to rescue (repeated measures one-way ANOVA, F = 117.9, dF = 14, p = 0.0001, with Tukey’s multiple comparisons control vs. adp¹ p = 0.0002, control vs adp¹ plus duplication p = 0.0033, and adp¹ vs adp¹ plus duplication p = 0.0035).

Previous research shows that the loss of adp results in increased fat stores, so we verified this observation with our new mutation before assessing for neurodevelopmental phenotypes [9–12]. Using a buoyancy assay, we evaluated changes in fat storage at the third-instar larval stage [29]. adp¹ mutant and w¹¹¹⁸ control third-instar larvae were floated in a sucrose solution where density was increased until all larvae were floating. The density at which all larvae floated was recorded and analysed. As expected, adp¹ mutants float at a lower density than w¹¹¹⁸ controls, indicating an increase in fat content compared to controls (Figure 2c). To show that the increase in fat storage was due to mutations in adp and not background variability or off-target effects of CRISPR mutagenesis, we introduced a 80 kb genomic duplication line containing the adp locus and again tested buoyancy. We were able to significantly rescue the fat phenotype, indicating that the loss of adp causes increased fat stores (Figure 2c).

Having confirmed that adp¹ displays similar loss of function phenotypes based on previous research, we wanted to know if adp loss results in brain growth perturbations to validate our RNAi data. The larval brain contains lipid droplets that help maintain the stem cell niche, so we first assessed whether adp¹ has changes in lipid droplet number that would indicate changes in lipogenesis in the brain [13,14]. We performed Nile Red staining in third-instar larval brains and quantified the total number of lipid droplets in adp¹ and w¹¹¹⁸ brains [13,30]. While adp¹ mutants display no significant difference in the total number of lipid droplets per micrometre cubed, they do have significantly less droplets over 5 µm in diameter (Figure 3a–d). This suggests that there is a change in lipogenesis in the brain, and adp may be involved in controlling lipid droplet size

Figure 3. adp¹ shows no difference in number of brain lipid droplets nor brain lobe volume. Nile red staining in the third-instar brain lobe showed no difference between (A) controls (w¹¹¹⁸) and (B) adp¹ in the total number of lipid droplets, quantified in (C), where each dot represents a single brain lobe (independent t-test, t = 1.154, dF = 8, p = 0.2819). (D) adp¹ mutants have significantly less lipid droplets greater than 5 µm in diameter, where each dot represents a single brain lobe (independent t-test, t = 2.456, dF = 8, p = 0.0396). Immunohistochemistry staining of third-instar larval (E) w¹¹¹⁸ control and (F) adp¹ brains showing Deadpan (Dpn, stem cells, magenta) and DAPI (DNA, green). (G) Brain lobe volume of w¹¹¹⁸ (control) or adp¹ third-instar larvae, each dot represents one brain lobe (n = 9–11). The adp¹ mutant does not differ in brain lobe volume compared to a w¹¹¹⁸ control (independent t-test, t = 0.2020, df = 18, p = 0.8422).

Finally, we wanted to determine if adp loss-of-function affected brain growth to replicate our RNAi data in a loss-of-function model (see Reagent Table for specific antibodies used). We quantified the brain lobe volume of adp¹ and w¹¹¹⁸ third-instar larvae but found no significant difference, suggesting that adp is not necessary for brain growth (Figure 3e–g). Interestingly, adp¹ animals are generally healthy and survive well at a high animal density, suggesting that the crowding effect documented in adp knockdown (Supplemental Figure S1) does not replicate in adp mutant animals. Our results also suggest that the initial RNAi results we found may have been due to off-target effects or a result of cell-specific knockdown.

Discussion

In this study, we generated a novel adp mutant to investigate its role in neurodevelopment. We showed that the previous standard adp mutant available from the Bloomington Stock centre, adp⁶⁰, did not carry the described mutation. Our new mutant, adp¹, acts as a loss-of-function exhibiting the expected phenotype of increased fat stores but failed to show any changes in neurodevelopment.

Previous discovery and description of the adp⁶⁰ fly demonstrated adp’s role in the negative regulation of lipogenesis [9–12]. Adult adp⁶⁰ flies had increased survival in starvation scenarios, and both larvae and adults exhibited increased triglyceride storage in the fat body. This research validated the use of adp⁶⁰ as a negative control for lipogenesis research [29]. When we looked at brain lobe volume of adp⁶⁰ compared to Oregon R controls, we found no difference. We do note that the Oregon R controls are significantly smaller than other controls used in this study. However, we found no difference in adp⁶⁰ brain lobe volume compared to all other controls in this study. We hypothesize that brain lobe volume difference in Oregon R flies could be due to background variation. We do not believe background variation or off-target mutations affect adp¹ [1 because it was backcrossed to w¹¹¹⁸ for three generations. Since we did not see a reduction in brain lobe volume in the presumed adp⁶⁰ line, we wanted to validate the presence of the characterized adp⁶⁰ mutation. Our sequencing of the adp⁶⁰ stock obtained from the Bloomington Stock Center using the primers described in the original publication showed sequence that was identical to the reference genome [11]. The 23 base pair deletion was no longer present in this stock, making it wild-type. In order to perform our own tests of the role of adp in neurodevelopment, we generated a new loss-of-function mutant.

Our adp¹ mutant larvae float at a lower density than controls, equating to a higher fat-to-muscle ratio, consistent with an adp loss-of-function phenotype (Figure 2C). Despite the previous literature on adp⁶⁰ not specifically quantifying the buoyancy of adp mutant larvae, most buoyancy protocols suggest using adp⁶⁰ larvae as a control for higher fat content [29]. These results validated that our newly generated mutant acts as a loss-of-function and demonstrated that our mutant could functionally replace adp⁶⁰ as a control in lipogenesis research.

While adp has not previously been linked with neurodevelopment, expression data show that transcript and protein are present in the third-instar larval central nervous system [31–34]. This expression profile and our RNAi knockdown results (Figure 1a–f) indicated that adp might have a role in brain development. Since Adp is involved in lipogenesis, we assessed whether this change in brain size was partially due to crowded growing conditions. When looking at GFP knockdown alone, we saw no difference in brain size between medium and high animal densities, indicating that brain size in general is not affected by crowding (Supplemental Figure S1a). However, we did find that adp knockdown brain volume phenotypes are only present in high animal density conditions (Supplemental Figure S1c). Interestingly, even though adp mutants thrive in higher density conditions, no brain size difference can be documented. There may be some interplay with adp dosage, cell-specific knockdown, and larval crowding that affects the growth of the brain.

Lipid droplets provide protection for neuroblasts in hypoxic environments, allowing them to remain proliferative, and protect the neuroblasts from assault by reactive oxygen species [13,14]. Disruption of lipid droplet production in glia leaves neuroblasts vulnerable to hypoxic conditions [13]. Adp inhibits triglyceride storage but has not previously been linked to lipid droplet production in the larval brain. Here, we show that adp loss-of-function does not affect the total number of lipid droplets in the third-instar larval brain but decreases the number of droplets greater than 5 µm in diameter (Figure 3a–d), indicating that adp may regulate lipid droplet size.

Despite our RNAi data indicating that adp may function in the developing brain, we failed to see a difference in brain size in our mutant (Figure 3e–g). Adp¹ acts as a loss-of-function mutant based on its mutation, fat phenotypes, and ability to rescue with duplication, so the lack of brain size phenotype confirms adp is not necessary for proper brain development. RNAi can have off-target effects, but genetic compensation can also occur. It has been shown in numerous organisms that phenotypic differences can exist between knockout and knockdown animals [35,36]. Compensatory genetic networks can arise in germline mutations allowing adaptation as the animal develops, which often negates deleterious effects [35,36]. Alternatively, post-transcriptional or post-translational effects might also prevent detrimental phenotypes [35,36]. Adp could also have different roles in different cell types in the brain. When Adp function is lost in the whole animal, some cell-type specific effects could balance out, leading to no obvious phenotypes in the mutant. Despite this, we are confident that adp is not a vital component of brain volume regulation.

Methods

Fly lines

The following fly lines were used: adp RNAi (P{TRiP.HMC06600}attP40), EGFP RNAi (P{VALIUM22-EGFP.shRNA.1}attP40), inscuteable-GAL4 (P{w[+mW.hs]=GawB}insc[Mz1407]) [15], neuronal Synaptobrevin-GAL4 (P{y[+t7.7] w[+mC]=nSyb-GAL4.P}attP2), adp⁶⁰ [9,10], Oregon-R [9,10], nanos-Cas9 (P{y[+t7.7] v[+t1.8]=nos-Cas9.R}attP2) [18], adp snRNA:U6:96Ac (P{y[+t7.7] v[+t1.8]=TKO.GS04840}attP40) [18–28], adp¹(this study), w¹¹¹⁸ [37,38], w[1118]; Dp(2;3)GV-CH321-48F17, PBac{y[+mDint2] w[+mC]=GV-CH321-48F17}VK00031. All flies were maintained at 25°C and grown on Archon glucose formula medium in plastic vials. Crosses were performed at the temperature indicated (18°C, 25°C, or 29°C). Brain volume measurements were conducted in late wandering 3rd-instar larvae identified by gut clearance and extruding spiracles [8,16].

Due to the Bloomington adp⁶⁰ stock no longer containing the described mutation, the authors asked multiple labs that previously worked with the allele for a copy, but none were able to provide one.

RNAi knockdown of adp

Male adp RNAi and EGFP RNAi flies were crossed with either insc-GAL4 or nSyb-GAL4 females for knockdown in neural stem cells or post-mitotic neurons, respectively. Crosses were set at 29°C at the same time for each experiment. Third-instar larvae were selected for brain lobe volume analysis.

Immunohistochemistry for brain volume

Late third-instar larval brains were dissected and immediately fixed in 4% paraformaldehyde in Phosphate Buffered Saline + 0.3% TritonX (PBST) for 20 min [8,16]. Brains were washed with PBST three times for 5 min and blocked twice with PBST + 1% w/v Bovine Serum Albumin (PBSTB) for 30 min before blocking with PBSTB + 5% Normal Donkey Serum (NDS) for 30 min. Brains were incubated with 1:1000 rat anti-Deadpan (neuroblast marker, Abcam, ab195173) in PBSTB overnight at 4°C. Primary antibody was removed from brains before washing three times with PBSTB for 20 minutes. Next, the brains are incubated with secondary antibodies 1:500 Donkey anti-Rat fluorophore 647 (Jackson ImmunoResearch Laboratories Inc., 712-605-153) and 1:1000 DAPI for one hour at room temperature. Finally, brains were washed with PBST four times for 10 min before being mounted for confocal microscopy.

Power analyses for brain lobe volume

We used the program G*Power to assess the power of our n sizes used in this study. We decided to use the data from the nsyb-GAL4 knockdown to assess the power. We selected t tests for ‘Test Family’ and ‘Means: Difference between two independent means (two groups)’. We performed a ‘Post hoc: Compute achieved power – given α, sample size, and effect size’ for the ‘Type of power analysis’ since we are using previously collected data. We selected Two tails, and left the ‘α error probability’ as .05. Our n sizes were 9 and 10 for GFP RNAi and adp RNAi, respectively. We then calculated the Cohen’s d with their calculator. We typed in the averages (GFP RNAi = 4.791, adp RNAi = 3.803) and standard deviations (GFP RNAi = 0.5923, adp RNAi = 0.5730) and the software calculated d = 1.7006163. We then calculated the power which was 93.64%. We therefore conclude that our n sizes of around 10 brains per condition is sufficiently powered to detect differences in brain lobe volume.

Larval crowding assay

Virgin female nSyb-GAL4 flies were crossed with male adp RNAi or EGFP RNAi on grape plates set in embryo collection chambers at 29°C. Embryos were collected off grape plates 18–24 h later, prior to hatching [17]. Embryos were placed in blue food vials in two conditions: medium animal density (50–65 embryos) or high animal density (150+ embryos) [17]. The high animal density suggests crowding and competition for resources. Brains from late wandering third-instar larvae were isolated and measured for brain lobe volume as previously described.

Generation of CRISPR mutants

The adp¹ mutant was generated from the TRiP-CRISPR stocks and TRiP-CRISPR Knockout (TRiP-KO) protocol [18–28]. Ten nanos-Cas9 females were crossed with 6 adp sgRNA males at 25°C to generate germline mutations in adp [18–21,24,27]. Fifteen F1 male flies (y,v,sc,sev; adp sgRNA/+; nanos-Cas9/+) were then crossed sswith 15 y,v,sc,sev; lethal/CyO females to isolate mutant animals. Both nanos-Cas9 and adp sgRNA constructs are tagged with y⁺,v⁺. F2 individuals were selected for y⁻ and v^– to ensure the removal of the Cas9 and sgRNA sequences and balance adp mutations. Once F3 larvae appeared, the founder F2 individual was removed from the tube for sequencing. Ten mutant stocks were established using this method. All work in the rest of this paper was performed with the adp¹ mutant. The adp¹ mutant was backcrossed to w¹¹¹⁸ for three generations to remove any extraneous mutations that may have occurred during mutagenesis. Backcrossing also allows for w¹¹¹⁸ to be used as a control. Both the adp¹ mutant and the adp duplication line were crossed into the same balancer line before being double balanced for rescue.

Sequencing and primers

Founder F2 adults were squished with fresh squishing buffer (10 mM Tris pH 8.2, 25 mM NaCl, 1 mM EDTA, and 200 µg/mL Proteinase K). The lysate was incubated for 30 minutes at 37°C degrees and then for 10 minutes at 85°C. 2 µL of the lysate was used for sequencing. The following primers were used for PCR: 5’-AACAAGTGTCATAATCCTATCCACAGCA-3’ and 5’-TGCATGCAGCCAATATAGATCAAGATG-3’. PCR products were purified and sequenced using the same primers. Sequencing of adp¹ showed a single insertion c.237_238insT resulting in an early stop codon p.Asp134Ter.

Buoyancy assay

To indirectly test the fat content of adp¹, we performed a buoyancy assay [29]. Approximately 20–40 late third instar larvae were collected from either adp¹ or w¹¹¹⁸ vials and placed in 50 mL conical tubes with a starting solution of 11.5 mL PBS and 9 mL 20% w/v Sucrose in PBS. Samples were swirled and inverted 2–3 times and settled for 2 min. The number of larvae floating was counted. 1 mL 20% sucrose was added to the conical, and samples were swirled, inverted, and settled. The number of floating larvae was recorded after each 1 mL addition. Additional sucrose was added, and floating larvae were counted until all larvae floated in each genotype. This experiment was repeated with 5 additional cohorts. The average sucrose concentration for all larvae to float in each genotype was compared using a paired t-test in Prism.

Nile red staining

Late third-instar larval brains from adp¹ and w¹¹¹⁸ were dissected in PBS and then fixed with 4% paraformaldehyde in PBS for 30 min [30]. The brains were washed with PBST three times for 20 min each before incubating overnight at 4°C in 1 ug/mL Nile Red in PBST. Finally, brains were washed twice with PBST for 30 min each before being mounted for imaging.

smFISH

Probes

We utilized the free ProbeDealer code for MATLAB to generate smFISH probes for adp [39]. The code, instructions, and Drosophila database are at https://campuspress.yale.edu/wanglab/ProbeDealer/. The adp input sequence was the whole mRNA transcript FASTA file from FlyBase. We selected to make 36 ‘sequential RNA fish’ probes outputted as ‘primary probe sequences’. The code automatically puts a 20-nucleotide secondary probe binding sequence on the 5’ and 3’ ends of each probe, making the final product 70 nucleotides along. We chose to remove the 20-nucleotide sequence from the 3’ end. The probes were ordered from Integrated DNA Technologies (IDT) with the “ssDNA oligo plate, 25 nmole scale, standard desalting” option. A single secondary probe was ordered also from IDT with the “ssDNA oligo, 250 nmole scale, HPLC purification” option using the 20-nucleotide secondary sequence and attaching Alexa Fluor 594 tag to the 3’ end.

Oligos were resuspended in RNase free water to 100 µM, and 50 µL each probe was combined into a single solution with total probe concentration as 100 µM.

All solutions were treated with DEPC or filtered to remove potential RNases and kept RNase free throughout the protocol. Vessels were treated with RNase ZAP and rinsed with RNase free solutions. Third-instar larval brains were dissected in PBS, fixed in 4% paraformaldehyde in PBS + 0.3% TritonX for 20 min. Samples were rinsed with PBS + 0.3% Tween 20 three times, washed for 15 min at room temp with PBS + 0.3% Tween 20 three times, then incubated in wash buffer (10% deionized formamide in 2X SSC) for 5 min at 37°C. Primary and secondary probes were diluted 1:250 in hybridization buffer (10% deionized formamide and 10% dextran sulphate in 2X SSC), incubated with tissue overnight at 37°C with gentle shaking on a Thermoshaker in the dark. Samples were rinsed three times in wash buffer, washed three times for 15 min in wash buffer at room temperature, and rinsed with PBS + 0.03 Tween 20 +DAPI. Brains were mounted in Slowfade Gold.

Confocal microscopy

All the imaging was performed on a Zeiss LSM 710 confocal microscope with the 40X water immersion lens. A single brain lobe was centred in frame [20,21]. Zoom was set to 0.7, and scanning was done at speed 9. Z-stacks were set to encompass the entire z-range of the lobe, and the stack size was set at 2 µm.

Nile Red

The Alexa-fluor 488 channel was used for Nile Red imaging [40]. The frame size was 1024 × 1024 with a line averaging of 2.

smFISH

Images were optimized for the secondary probe (Alexa Fluor 594). A LD C-Apochromat 40×/1.1 W Korr M27 lens was used. All images were taken on the same day with the same laser power (5), gain (959), pinhole (1 airy unit), and scanning parameters (frame: 1024 × 1024, line average: 16, scan speed: 6). Post imaging processing changed the signal max from 255 to 115 in Imaris on all images.

Volume

Z stacks were set using Deadpan signal (647 channel) for volumetric analysis. The frame size was 512 × 512.

Volume analysis

Analysis was performed using the IMARIS software with the surface function. To compute volume, we drew surfaces around every 5th z-stack, including the two farthest ends of the brain. The stacks were then compiled, and volume was generated automatically. The average volume was compared using a t-test in Prism.

Nile red analysis

Analysis was performed using the IMARIS software with the surface and spots functions. First, all images were set to the same brightness and contrast settings (Minimum of 35.46, maximum of 255, and gamma or 2.12). Volume was computed under the surface tab as described above, then a mask of the surface was generated to be used as a region of interest in spots. Under the spots tab, we used the automatic spots counter with an estimated XY diameter of 1.96523 units, and the quality filter was set to 20%, allowing for the most lipid droplets to be counted without generating false positives. We then looked through the brain to remove false positives and add in missed droplets, therefore manually ensuring all lipid droplets were accounted for without false spots being counted. To count the large lipid droplets, we manually counted all droplets that were greater than 5 μm in diameter. For each brain, the number of lipid droplets was divided by the volume to compute the number of droplets/µm³. The average number of lipid droplets/µm [3] was then compared between groups with a t-test in Prism.

Statistical analysis

All statistical analyses were performed using GraphPad Prism software. Independent t-tests for brain volume of neural stem cell knockdown, post-mitotic neuronal knockdown, Oregon-R vs adp⁶⁰, lipid droplet analysis, and w¹¹¹⁸ vs adp¹ were done by selecting ‘t-tests’ under column analyses and ‘Unpaired’ under Experimental Design and assuming Gaussian distribution and equal SD. The paired t-test for WDTC1 rescue of buoyancy was performed by selecting ‘t-tests’ under column analysis and ‘Paired’ under Experimental Design and assuming Gaussian distribution and equal SD. Finally, the repeated measures one-way ANOVA for the buoyancy in Figure 2c was performed by selecting ‘One-Way ANOVA’ under Column Analyses, ‘each row represents matched, or repeated measures, data’ under Experimental Design, assume Gaussian distribution, and not assuming sphericity therefore using Geisser-Greenhouse correction. Under the ‘Multiple Comparisons’ tab, ‘Compare the mean of each column to the mean of every other column’ to allow for identification of rescue phenotypes. The p-values reported in figure legends were the p value under the Repeated Measures ANOVA Summary and the adjusted p values from the multiple comparisons.

Acknowledgments

Stocks obtained from the Bloomington Drosophila Stock Center (NIH P40OD018537) were used in this study. Human ORFs used in this study were provided by the University of Utah HSC Core Research Facilities as part of a partnership with the Huntsman Cancer Institute and individual contributing investigators. Thank you to members of the Link lab for thorough reading of the manuscript. We acknowledge the information provided by FlyBase (NHGRI awards U41HG000739 and U24HG010859) using release FB2023_04³⁸.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Strains and plasmids are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19336934.2024.2352938

Additional information

Funding

This material is based upon work supported by the University of Utah, NIH NIAID grants R56AI170857 and R01AI170857, and National Science Foundation Graduate Research Fellowship Program under Grant No. 2139322. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.