Inter-organ Wingless/Ror/Akt signaling regulates nutrient-dependent hyperarborization of somatosensory neurons

  1. Yasutetsu Kanaoka
  2. Koun Onodera
  3. Kaori Watanabe
  4. Yusaku Hayashi
  5. Tadao Usui
  6. Tadashi Uemura  Is a corresponding author
  7. Yukako Hattori  Is a corresponding author
  1. Graduate School of Biostudies, Kyoto University, Japan
  2. Research Center for Dynamic Living Systems, Kyoto University, Japan
  3. AMED-CREST, Japan
  4. JST FOREST, Japan

Abstract

Nutrition in early life has profound effects on an organism, altering processes such as organogenesis. However, little is known about how specific nutrients affect neuronal development. Dendrites of class IV dendritic arborization neurons in Drosophila larvae become more complex when the larvae are reared on a low-yeast diet compared to a high-yeast diet. Our systematic search for key nutrients revealed that the neurons increase their dendritic terminal densities in response to a combined deficiency in vitamins, metal ions, and cholesterol. The deficiency of these nutrients upregulates Wingless in a closely located tissue, body wall muscle. Muscle-derived Wingless activates Akt in the neurons through the receptor tyrosine kinase Ror, which promotes the dendrite branching. In larval muscles, the expression of wingless is regulated not only in this key nutrient-dependent manner, but also by the JAK/STAT signaling pathway. Additionally, the low-yeast diet blunts neuronal light responsiveness and light avoidance behavior, which may help larvae optimize their survival strategies under low-nutritional conditions. Together, our studies illustrate how the availability of specific nutrients affects neuronal development through inter-organ signaling.

Editor's evaluation

Nutrition profoundly affects neural development. The Uemura lab previously reported that C4da neurons elaborate complex dendrites when larvae grow on low-yeast diets, a phenomenon called neural sparing. In the current study, they define the molecular mechanism underlying the nutrition-mediated phenomenon and identify that the inter-organ Wingless/Ror/Akt pathway between the neuron and its adjacent muscles is necessary and sufficient to mediate dendrite over-branching in the low-yeast condition.

https://doi.org/10.7554/eLife.79461.sa0

Introduction

The physiological state of an organism influences organogenesis throughout the body. Among many external factors affecting the physiological state, nutrition in early life has a profound impact (Bhutta et al., 2017). This is particularly the case with neural development, which is highly metabolically demanding. A large amount of energy is consumed to control neural stem cell division, form complex dendrites and long axons in myriad neuronal cell types, and ultimately construct functional neural circuits (Prado and Dewey, 2014; Georgieff et al., 2015). Compared to metabolic regulation of neural stem cell proliferation (Homem et al., 2015), little is known about how nutritional status is conveyed to developing neurons and how those neurons regulate growth in response to such a signal (Shimada-Niwa and Niwa, 2014; Shimono et al., 2014; Liu et al., 2017).

Dietary nutrients are absorbed by the digestive tract and circulated throughout the body, and they are sensed by organs including the nervous system (Chantranupong et al., 2015). Those organs communicate the nutritional status to each other by secreting signaling molecules, either low-molecular-weight metabolites or macromolecules such as soluble proteins and lipoprotein particles, to elicit tissue-specific responses; and it is this inter-organ communication network that coordinates organogenesis with body growth (Droujinine and Perrimon, 2016; Texada et al., 2020). In the nervous system, neurons sense circulating nutrients directly or by way of signaling molecules derived from other tissues, so there exist diverse modes of nutrient sensing (Morton et al., 2014; Jayakumar and Hasan, 2018).

The above-mentioned regulatory mechanisms of nutrient-dependent neuronal development can be explored at the molecular level with appropriate model neurons; and one particularly amenable model is the Drosophila class IV dendritic arborization (C4da) neuron located in the larval periphery (Grueber et al., 2002). Dendritic arbors of C4da neurons extensively cover the body wall, and they are elaborated two-dimensionally between the epidermis and the body wall muscles. C4da neurons in larvae respond to noxious thermal, mechanical, and light stimuli and provoke robust avoidance behaviors (Tracey et al., 2003; Hwang et al., 2007; Xiang et al., 2010; Zhong et al., 2010; Guntur et al., 2015; Chin and Tracey, 2017). In the context of adaptation of growing animals to nutritional environments, it has been shown more recently that the C4da neurons sense amino acid deprivation by an amino acid transporter, Slimfast, at a late larval stage, which contributes to overcoming the nutritional stress, thereby allowing pupariation (Jayakumar et al., 2016; Jayakumar et al., 2018). In addition, we and another group have shown that dendrites of C4da neurons become more complex when larvae are reared on low-yeast diets compared to high-yeast diets (Figure 1A and B, and Figure 1—figure supplement 1A; Watanabe et al., 2017; Poe et al., 2020). We designate this counterintuitive phenotype as hyperarborization. Although the entire larval development takes longer on the low-yeast diet compared to the high-yeast diet (Figure 1—figure supplement 1A), it is unlikely that the hyperarborization is a simple consequence of the longer larval stage (Watanabe et al., 2017; see also Results). Therefore, it has been assumed that the low-yeast diet is deficient in select nutrients, which causes the phenotype. However, the identities of such key nutrients responsible for the hyperarborization phenotype have heretofore not been determined.

Figure 1 with 2 supplements see all
A mixture of vitamins, metal ions, and cholesterol ameliorates C4da neuron hyperarborization.

(A and B) Representative output images of DeTerm. DeTerm automatically detects dendritic terminals of C4da neuron ddaC in larvae reared on a high-yeast diet (HYD; A) or a low-yeast diet (LYD; B). Red points indicate detected branch terminals. (C–E) The numbers of branch terminals detected by DeTerm (C), coverage areas of dendrites (D), and densities of branch terminals (the number of branch terminals/coverage area; E) of individual neurons, on HYD or LYD (Student’s t-test, n=6). Boxes show the 25th–75th percentiles. The central lines indicate the medians. Whiskers extend to the most extreme data points, which are no more than 1.5 times the interquartile range. Boxes and points for HYD data and those for LYD data are colored red and blue, respectively, in this and subsequent figures. (F) Two-dimensional (2D) plot of the dendritic area and the number of branch terminals of each neuron. The ellipses represent the 95% confidence intervals, which are clearly separated for HYD and LYD. This plot shows a positive correlation between the area of the dendritic field and the number of branch terminals. (G–I) Images of ddaC neurons in larvae reared on LYD (G), LYD +vitamin + metal ion +cholesterol + other ingredients (LYD +VMCO; H), or LYD +VMCO + 1 x amino acids (LYD +VMCO + 1 x AAs; I). (J and K) Quantitative analysis of ddaC on LYD +VMCO or LYD +VMCO + 1 x AAs. (J) 2D plot. Note that the ellipse of LYD +VMCO and that of LYD +VMCO + 1 x AAs are located between those of HYD and LYD. (K) Densities of branch terminals (One-way ANOVA and Tukey’s HSD test, n=8–10). (L–N) Images of ddaC neurons in larvae reared on LYD +metal ion+cholesterol + other ingredients (LYD +MCO; L), LYD +vitamin + cholesterol +other ingredients (LYD +VCO; M), or LYD +vitamin + metal ion +other ingredients (LYD +VMO; N). (O–R) 2D plots of ddaC on LYD +MCO (O), LYD +VCO (P), or LYD +VMO (Q), and densities of branch terminals (R, Steel test, n=10). The elipses of these diets largely overlap with that of LYD and clearly or almost separare from that of HYD (O–Q). (S) Images of ddaC neurons in larvae reared on LYD +vitamin + metal ion +cholesterol (LYD +VMC). (T and U) 2D plots of ddaC neurons in larvae reared on LYD +VMC (T) and densities of branch terminals (U, Steel test, n=8). The ellipse of LYD +VMC is located between those of HYD and LYD (T). Boxplots in (K, R and U) are depicted as in (C). *p<0.05, **p<0.01, and ***p<0.001. Scale bars, 100 μm.

A wealth of genetic analyses on standard foods has revealed numerous regulators of dendrite morphogenesis working either in cell-autonomous or non-cell autonomous manners (Jan and Jan, 2010; Dong et al., 2015; Valnegri et al., 2015). Some of the cell-autonomous mechanisms include those related to intake and synthesis of metabolites: amino acid transporter SLC36/Pathetic (Path) (Lin et al., 2015) and a critical regulator of fatty acid synthesis, sterol regulatory element binding protein (SREBP) (Meltzer et al., 2017; Ziegler et al., 2017). Concerning the non-cell autonomous mechanisms, direct interactions between C4da neurons and one of the adjacent tissues, the epidermis, have been well characterized with the help of anatomical approaches under both light and electron microscopes (Yang and Chien, 2019). Some portions of dendritic branches are attached to the extracellular matrix, and the attachment is mediated by signaling between an epidermally derived semaphorin ligand Sema-2b and its receptor Plexin B (PlexB) on the dendrite (Meltzer et al., 2016), as well as between a TGF-β ligand Maverick (Mav)-Ret receptor combination (Hoyer et al., 2018). Other portions of dendritic arbors are wrapped by epidermal cells, so overall the dendrite arbor is embedded in the epidermis locally (Han et al., 2012; Kim et al., 2012; Tenenbaum et al., 2017; Jiang et al., 2019). In contrast to the above dendrite-epidermis interaction, there is much less evidence for signaling between muscles and dendrites, despite their proximity to dendrites and their large volume in the body (Yasunaga et al., 2010). Furthermore, when considering the relationship between nutritional status and C4da neurons, little is known about how exactly the dietary information is remotely transmitted from the gut to the neurons. To address these unsolved questions, it is critical to efficiently quantify the effects of various nutritional and genetic conditions on this nutrition-dependent hyperarborization. For this purpose, we developed DeTerm, a software program for automatic detection of dendritic branch terminals (Figure 1A and B; Kanaoka et al., 2019).

Here, we show that C4da neurons increase their dendritic terminal density on a low-yeast diet (LYD) compared to a high-yeast diet (HYD) due to a concurrent deficiency in vitamins, metal ions, and cholesterol. We then identified an extrinsic factor and an intracellular signaling axis that jointly enable C4da neurons to respond to the LYD nutritional status. On LYD, Akt and its upstream receptor tyrosine kinase Ror in the neuron are required for the hyperarborization. In a paracrine fashion, Wingless (Wg) produced by body wall muscles activates Akt by way of Ror and contributes to the hyperarborization. In muscles of larvae on the HYD, Stat92E, a transcription factor in the JAK/STAT pathway, was more highly expressed and negatively regulated wg expression, whereas the LYD resulted in lower expression of Stat92E, which partly contributed to higher expression of wg. Together, our studies illustrate how nutritional environments impact neuronal development through the Wg-Ror-Akt pathway between the neuron and closely located muscles. As for the neuronal function, we found that LYD blunted light responsiveness of class IV neurons and larval light avoidance behavior, which may help larvae optimize their survival strategies under low-nutritional conditions.

Results

A mixture of vitamins, metal ions, and cholesterol ameliorates the hyperarborization

Our analysis using the software program called DeTerm established that both the number of branch terminals per neuron and the density of terminals (terminal number/arbor size) were higher on LYD than on HYD (Figure 1C-E). In addition to these box plots, we drew two-dimensional plots with the dendritic area on the X-axis and the number of branch terminals on the Y-axis, and the numerical features of dendrites of C4da neurons on HYD and those on LYD were clearly separated (Figure 1F). Therefore, in the subsequent analyses, we mainly focused on the density of branch terminals (Figure 1E) and the separation in 2D plots (Figure 1F) to evaluate the hyperarborization phenotypes.

Yeast is one of the main ingredients in Drosophila laboratory foods, and it has been primarily considered as a source of amino acids. We suspected the possibility that LYD is deficient in amino acids and that is the cause of the phenotype. Therefore, we first examined whether supplementation of LYD with amino acids would ameliorate the hyperarborization. However, the addition of an essential amino acid solution, an amino acid mix, or peptone resulted in only slight or no restoration of the phenotype (Figure 1—figure supplement 1B–L, and see details in the legend). To more comprehensively search for nutrients responsible for the hyperarborization, we used fractions of a fully chemically defined or holidic medium for Drosophila (Piper et al., 2014; Piper et al., 2017) and examined which fraction or which combinations of the fractions were able to ameliorate the phenotype (Figure 1—figure supplement 1A). Addition of four fractions other than amino acids, which comprise vitamins (V), metal ions (M), cholesterol (C), and other ingredients (nucleic acids and lipid-related metabolites: O), to LYD significantly rescued the hyperarborization (Figure 1G, H, J and K; see also the legend of Figure 1J). We named this diet LYD + VMCO. Further supplementation of amino acids to LYD + VMCO did not improve the degree of the rescue (Figure 1K). Importantly, the phenotype was not restored without any one of three fractions, namely, vitamins, metal ions and cholesterol (Figure 1L–R). On the other hand, the fraction designated other ingredients was dispensable for amelioration of the phenotype (Figure 1S-U). These results suggest that the concurrent deficiency in vitamins, metal ions, and cholesterol contributes to the hyperarborization phenotype.

Larval developmental timing on LYD + VMC was comparable to that on LYD Figure 1—figure supplement 1A; nonetheless the hyperarborization phenotype was blunted on LYD + VMC. We then examined whether extension of larval growth cause the hyperarborization by testing other dietary or genetic interventions. We previously compared dendrite morphologies between larvae reared on a low-sugar diet and those on a high-sugar diet that delays larval development (Musselman et al., 2011), and we reported that the hyperarborization does not occur on the high-sugar diet (Watanabe et al., 2017). We expanded this approach and analyzed the effect of the sugar overload on dendrite branching in a quantitative manner. When we observed dendrites in larvae reared on HYD supplemented with sucrose at the same timing as those on LYD (8–9 days AEL), they did not become more complex compared to those on HYD alone (Figure 1—figure supplement 2A–E). Moreover, we observed the C4da neurons in larvae with dlip8 overexpressed in wing imaginal discs, which is sufficient to extend the larval stage (Colombani et al., 2012). This genetic intervention did not affect dendrite complexity (Figure 1—figure supplement 2F–J). Altogether, these results suggest that the prolonged larval period was not the primary cause of the hyperarborization phenotype.

Akt and receptor tyrosine kinase Ror are required in C4da neurons to hyperarborize their dendrites

To investigate the molecular mechanism underlying the hyperarborization phenotype on LYD, we focused on intracellular signaling factors that have been reported to sense nutritional status in other cellular contexts. Thus, we examined whether C4da neuron-specific knockdown (KD) of any of these factors would affect this diet-selective phenotype (Figure 2A–J and Figure 2—figure supplement 1A–P). To identify candidate genes whose KD attenuated the hyperarborization phenotype, we investigated how much the HYD ellipse and the LYD ellipse approached each other or overlapped in the 2D plot. We also compared the terminal density using two-way analysis of variance (ANOVA) throughout this study, unless described otherwise (see p-values in the aforementioned figures).

Figure 2 with 6 supplements see all
Akt and receptor tyrosine kinase Ror are required in C4da neurons to hyperarborize their dendrites.

(A–F) Images of control ddaC neurons (A and D), or Akt knockdown (KD) ddaC neurons using UAS-Akt RNAiv2902 (B and E) or UAS-Akt RNAiBL33615 (C and F), on HYD (A–C) or LYD (D–F). (G–J) Quantitative analysis of effects of Akt KD using UAS-Akt RNAiv2902 (G and H) or UAS-Akt RNAiBL33615 (I and J). (G and I) 2D plots. (H and J) Densities of branch terminals. As indicated by the p-values, there was a significant interaction between diet and genotype on terminal density in both Akt KD experiments (two-way ANOVA, n=6). That is, compared to the difference between HYD and LYD in control C4da neurons, that difference in Akt KD neurons using UAS-Akt RNAiv2902 or UAS-Akt RNAiBL33615 was significantly smaller, suggesting that the hyperarborization was blunted by the Akt KD. (K–P) Images of control (K and M) or Ror knocked down ddaC (L and N) on HYD (K and L) or LYD (M and N). 2D plot (O) and densities of branch terminals (p, two-way ANOVA, n=8). (Q–V) Images of ddaC in wild-type (WT; Q and S) or Ror4 mutant larvae (R and T) on HYD (Q and R) or LYD (S and T). The ddaC neurons were visualized by expressing ppk-CD4:tdGFP. 2D plot (U) and densities of branch terminals (V, two-way ANOVA, n=8). Boxplots in (H, J, P, and V) are depicted as in Figure 1C. ***p<0.001. Scale bars, 100 μm.

Among KD phenotypes of the candidate genes, we were interested in an Akt kinase (Akt) KD in one of the RNAi lines, which impacted hyperarborization but left overall dendritic architecture relatively intact (v2902; Figure 2B, E, G and H). This Akt KD in the v2902 line resulted in neither apparent downsizing of the arbor area (Figure 2G) nor overt decreases in branch length on HYD (Figure 2—figure supplement 2A and B), contrasting with diet-independent severe phenotypes observed in another Akt RNAi line (BL33615; Figure 2C, F, I and J and Figure 2—figure supplement 2C and D). Our subsequent analyses showed that v2902 was less effective in knocking down Akt than BL33615 (Figure 2—figure supplement 3A–H). We interpreted these results as follows: the severe reduction of Akt function in the BL33615 line impairs growth of dendritic branches, as shown previously (Parrish et al., 2009), whereas the mild reduction in the v2902 line mostly secures the basal activity of Akt necessary for growth, but it affects hyperarborization on LYD in a relatively selective manner (Figure 2—figure supplement 2G–I). We further knocked down genes that constitute the signaling pathways of Akt (Figure 2—figure supplement 1Q-AJ), and found that inhibition of TOR signaling components, such as Target of rapamycin (Tor) or Ribosomal protein S6 kinase (S6k) also ameliorated the phenotype (Figure 2—figure supplement 1Q–AJ). A similar effect caused by inhibition of Tor was described by Poe et al., 2020.

Various secreted factors are known to function as inter-organ communication factors in response to nutritional conditions (Droujinine and Perrimon, 2016). We therefore hypothesized that, in larvae on LYD, C4da neurons receive signaling molecules from other tissues, leading to the hyperarborization via the Akt/Tor signaling pathway. As candidate receptors upstream of Akt, we focused on receptor tyrosine kinases (RTKs; Sopko and Perrimon, 2013), and conducted C4da neuron-specific KD screenings of 20 RTK genes (Supplementary file 2 and Figure 2—figure supplements 46; see ‘RTK screening’ in Materials and methods). One of the positive hits in our primary screening was RTK-like orphan receptor (Ror; Figure 2—figure supplement 4C and D) and we confirmed that the Ror KD significantly suppressed the hyperarborization in the secondary screening (Figure 2K–P). Moreover, we observed C4da neurons in Ror4 null mutant larvae and showed that they recapitulated the result of the KD (Figure 2Q–V). These results suggest that Ror and Akt are required in C4da neurons to hyperarborize their dendrites on LYD. Other positive hits included the known upstream regulators of Akt, Insulin-like receptor (InR) or Anaplastic lymphoma kinase (Alk). However, we could not definitively conclude whether InR and Alk contribute to the hyperarborization phenotype, due to inconsistent KD results among multiple experiments (Supplementary file 2 and Figure 2—figure supplement 5).

To further characterize the phenotypes of the Ror KD and the v2902 Akt KD line, we measured the total length of branches per neuron (dendrite length) and dendrite length/area on each diet (Figure 2—figure supplement 2A, B, E, F). For both KD lines, values for length were higher on LYD compared to HYD in control C4da neurons. The v2902 line reduced dendrite length on both diets, although the decrease on HYD was marginal (Figure 2—figure supplement 2A and B), whereas Ror KD only decreased dendrite length on LYD (Figure 2—figure supplement 2E and F). This difference may reflect a restricted role for Ror in the response to a deficiency in the key nutrients, as opposed to a more general requirement of Akt for branch growth irrespective of the diets (Figure 2—figure supplement 2G, H, J).

Wg in muscles is more highly expressed on LYD and promotes dendritic branching of C4da neurons

Ror binds to Wnt ligands and triggers intracellular signaling cascades (Ripp et al., 2018; van Amerongen and Nusse, 2009). We therefore knocked down wingless (wg), Wnt2, Wnt4, or Wnt5 in either of the two tissues adjacent to C4da neurons: epidermal cells and muscles. We observed that wg KD in muscles using either Mhc-GAL4 or mef2-GAL4 suppressed the hyperarborization phenotype (Figure 3A–F and Figure 3—figure supplement 1A–F). In contrast, epidermal KD of wg had no effect on the phenotype (Figure 3—figure supplement 1G–L). The requirement of Wg for the hyperarborization was further confirmed by the finding that the hyperarborization effect was dampened in C4da neurons in the whole-body wg mutant (hypomorphic wg1/amorphic wg l-8; Figure 3G–L).

Figure 3 with 1 supplement see all
Wg in muscles is expressed more highly on LYD and promotes dendritic branching of C4da neurons.

(A–F) Images of ddaC neurons in control larvae (A and C) or larvae with wg KD in muscles (B and D), on HYD or LYD. 2D plot (E) and densities of branch terminals (F, two-way ANOVA, n=6). (G–L) Images of ddaC neurons in WT (G and I) or wg1/l-8 larvae (H and J) on HYD or LYD. 2D plot (K) and densities of branch terminals (L, two-way ANOVA, n=8). (M–P) Muscles in larvae reared on HYD (M), LYD (N), or LYD +VMC (O) were stained for Wg. The signal intensities are represented by the indicated color code. (P) Quantification of the mean Wg immunofluorescence intensity in muscle 9, one of the closest muscles to the ddaC neuron (Steel Dwass test, n=18–23). (Q–S) Images of wg-Gal4 muscles driving the expression of RedStinger, which is DsRed tagged with a nuclear localization signal. Muscles, on HYD (Q) or LYD (R) were stained with phalloidin (green), an antibody to DsRed (red), and DAPI (blue). The signal intensities of DsRed are represented by the indicated color code, and white dashed circles indicate outlines of nuclei (Q’and R’). (S) Quantification of DsRed intensity in nuclei of muscle 9 (Wilcoxon-Mann-Whitney test, n=13–15). (T–Y) Images of ddaC neurons in control larvae (T and V) or larvae with wg overexpression at 29 °C in muscles (U and W) on HYD or LYD. 2D plot (X) and densities of branch terminals (Y, Wilcoxon-Mann-Whitney test, n=8–10). Experiments were couducted at both 29°C and 25°C, but only results at 29 °C are shown. At 29 °C, the effect of wg overexpression was expected to be higher, and in fact, branch terminal density increased on both HYD and on LYD, but only on HYD at 25 °C. The increase in the branch terminal density elicited by wg overexpression on both diets was less dramatic than the difference in the branch terminal density due to the diets in each respective genotype. The ddaC neurons were visualized by expressing ppk-CD4:tdGFP. Boxplots in (F, L, P, S, and Y) are depicted as in Figure 1C. *p<0.05, **p<0.01, and ***p<0.001. Scale bars, 100 μm (A-D, G-J, M-O, and T-W), 25 μm (Q-R’).

We then examined whether Wg is differentially expressed in muscles between larvae reared on HYD and those on LYD. Immunostaining using an anti-Wg antibody showed stronger signals in LYD-fed larvae (Figure 3M, N and P). These stronger signals indeed represented increased amounts of endogenous Wg because knocking down wg decreased the intensity (Figure 3—figure supplement 1M–Q). We also asked whether wg expression is up-regulated on LYD at the transcriptional level. We expressed RedStinger, DsRed tagged with a nuclear localization signal, under the knocked-in wg-GAL4 driver that reflects the endogenous expression pattern of wg (Bosch et al., 2020). Nuclear RedStinger signals in muscles were stronger in larvae on LYD (Figure 3Q-S), indicating that LYD up-regulated wg transcription compared to HYD. We further tested whether muscle-derived Wg promotes dendritic branching of C4da neurons. For this purpose, we overexpressed wg in muscles and found that those larvae increased the number of dendritic terminals per neuron on both HYD and LYD (Figure 3T–Y), strengthening the role of the muscle–C4da neuron communication in hyperarborization. Importantly, addition of vitamins, metal ions, and cholesterol to LYD significantly suppressed the up-regulation of Wg on LYD (Figure 3M-P). Together with the effect of these compounds on dendritic branching (Figure 1S-U), we hypothesized that wg expression in muscles is enhanced by a concurrent deficiency in vitamins, metal ions, and cholesterol in LYD, and that muscle-derived Wg promotes dendritic branching of C4da neurons.

Wg-Ror-mediated activation of Akt in C4da neurons evokes the hyperarborization

Wnt signaling is engaged in diverse contexts of neuronal development and regeneration (Green et al., 2014; He et al., 2018; Endo and Minami, 2018; Nye et al., 2020; Weiner et al., 2020). In a previous study in Drosophila, responses to dendrite injuries were investigated using class I da (C1da) and C4da neurons. This showed that Ror, a seven-pass transmembrane receptor Frizzled (Fz), and downstream components including Disheveled (Dsh) and Axin (Axn) are required for dendrite regeneration (Nye et al., 2020). We therefore examined whether these genes and other components of Wnt signaling affect the hyperarborization phenotype (Figure 4—figure supplement 1 and Figure 4—figure supplement 2). Knocking down fz2 significantly ameliorated the hyperarborization (Figure 4—figure supplement 1A, C, D, F, H, J). Not only fz2 KD neurons, but also fz2 null mutant neurons showed less prominent hyperarborization compared to the control neurons (Figure 4—figure supplement 1K–P). These results are consistent with the proposed function of Ror as a Wnt co-receptor with Fz2 (Ripp et al., 2018). In addition to fz2, KD of fz, KD of downstream components (dsh and Axn), or expression of a dominant-negative form of Bsk also significantly blunted the hyperarborization (Figure 4—figure supplement 1B, E, G, I and Figure 4—figure supplement 2 [B, I, O and U], [G, N, T and Z], and [C, J, P and V]). However, we question whether all of these ameliorated phenotypes share the same underlying mechanism with those of Ror or fz2 KD (see the legend of Figure 4—figure supplements 1 and 2, and DISCUSSION). Altogether, our results suggest that among the known components of Wnt signaling in Drosophila, at least Fz2 cooperates with Ror in transducing the external signal to evoke the hyperarborization.

Ror is also reported to activate the PI3K/Akt/mTor signaling pathway in lung adenocarcinoma or multiple myeloma (Liu et al., 2015; Frenquelli et al., 2020). We therefore hypothesized that Wg-Ror signaling activates Akt signaling in C4da neurons on LYD, leading to the hyperarborization. To test this hypothesis and to clarify the relationship between Wg-Ror and Akt at the molecular level, we examined how genetic manipulations of Wg-Ror signaling affect Akt activity levels in C4da neurons (Figure 4A–J). The specificity of the p-Akt antibody in C4da neurons was validated in two ways: (1) p-Akt signals were significantly reduced by Akt KD Figure 2—figure supplement 3A-H; (2) expression of myr-Akt, a constitutively activated membrane-anchored form of Akt (Stocker et al., 2002), dramatically increased the p-Akt signal strength (Figure 2—figure supplement 3I and J). Using this antibody, we first examined how the p-Akt level in C4da neurons differs between larvae reared on HYD and LYD. Immunostaining showed that the p-Akt level in C4da neurons was higher on LYD than on HYD (Figure 4A, A’, C, C’ , and E). In contrast, Ror KD neurons from larvae on LYD showed reduced p-Akt levels compared to those on HYD (Figure 4B, B’, D, D’ , and E). Furthermore, wg overexpression in muscles increased the p-Akt level in C4da neurons on HYD (compare Figure 4F’ with 4 G’), which became comparable to the level on LYD (compare Figure 4G’ with Figure 4I’; see quantification in Figure 4J). These results suggest that Akt signaling in C4da neurons is activated by muscle-derived Wg, and this activation is mediated by Ror in the neurons.

Figure 4 with 3 supplements see all
Wg-Ror-mediated activation of Akt in C4da neurons evokes the hyperarborization.

(A–E) Control ddaC (A, A’, C and C’) or Ror KD ddaC neurons (B, B’, D and D’) were stained for p-Akt (A’-D’) and co-imaged with a C4da neuron marker ppk-CD4:tdTom (A–D). The signal intensities of p-Akt correspond to the indicated color code, at right, and white dashed circles indicate the cell bodies of ddaC neurons. (E) Quantification of p-Akt intensity in cell bodies of control or Ror knocked down ddaC neurons (Wilcoxon-Mann-Whitney test, n=22–28). (F–J) ddaC in control larvae (F, F’, H and H’) or larvae with wg overexpression in muscles (G, G’, I and I’) were stained for a pan-sensory neuron marker (22C10; F–I) and for p-Akt (F’-I’). (F–I) Magenta arrows indicate the cell bodies of ddaC neurons. (F’-I’) The intensities of p-Akt signals correspond to the indicated color code, at right, and white dashed circles indicate the cell bodies of ddaC neurons. (J) Quantification of p-Akt intensity in control larvae or larvae with wg overexpression in muscles (Wilcoxon-Mann-Whitney test, n=12–18). (K–U) Images of UAS-control RNAi and UAS-control expressing ddaC neurons (K and O), UAS-Ror RNAi and UAS-control expressing ddaC neurons (L and P), UAS-control RNAi and UAS-myrAkt, a constitutively active form of Akt, expressing ddaC neurons (M and Q), or UAS-Ror RNAi and UAS-myrAkt expressing ddaC neurons (N and R), on HYD or LYD. We used UAS-grnd RNAi, which had no significant impact on the hyperarborization phenotype (Figure 2—figure supplement 1A and E, 1I, and 1 M), as the UAS-control RNAi and UAS-CD4:tdTom as the UAS-control. (S–W) Quantitative analysis of combinatorial effects of Ror KD and myrAkt expression. (S–U) 2D plots. (V and W) Densities of branch terminals on HYD (V) or LYD (W) (Steel Dwass test, n=8–9). Boxplots in (E, J, V and W) are depicted as in Figure 1C. *p<0.05 and **p<0.01. Scale bars, 25 μm (A-D’ and F-I’), 100 μm (K–R).

We further examined whether the activation of Akt itself evokes hyperarborization even in the absence of the upstream Ror-mediated signaling (Figure 4K–W). Expression of myr-Akt in C4da neurons increased the terminal density even on HYD, regardless of whether Ror was knocked down or not (compare Figure 4K with 4 M and 4 N; see also 4T, 4 U and 4 V). This result suggests that Akt activation in the neurons plays a pivotal role for the hyperarborization. Our result is consistent with a previous finding that overexpression of the wild-type form of Akt causes a significant increase in dendrite coverage of the epidermis (Parrish et al., 2009).

Somatosensation of Drosophila larvae depends on C4da and 3 other classes of da neurons. Among them, class I da (C1da) and class III da (C3da) neurons function in proprioception and gentle-touch sensation, respectively (Hughes and Thomas, 2007; Hwang et al., 2007; Im and Galko, 2012; Yan et al., 2013; Tsubouchi et al., 2012). We also examined whether the hyperarborization phenotype and p-Akt upregulation are observed in these classes. We previously reported that the hyperarborization phenotype is not seen in ddaD and ddaE C1da neurons (Watanabe et al., 2017). The p-Akt level showed no significant difference between HYD and LYD in ddaE neurons (Figure 4—figure supplement 3A–C). On the other hand, a C3da neuron, ddaF, showed an increase in both the dendritic terminal number and the p-Akt level on LYD (Figure 4—figure supplement 3D–I), similar to C4da neurons. Increased branch terminals of C3da neurons on a low-yeast diet was also reported previously (Poe et al., 2020). Our results raise the possibility that, along with C4da neurons, C3da neurons share the Akt-driven branching mechanism in response to the low-nutrient condition.

Stat92E partially contributes to downregulation of Wg expression and suppresses hyperarborization on HYD

Given that Wg expression in muscles is higher on LYD (Figure 3M–P), and the differential expression impacts the dendritic branching of C4da neurons (Figure 3T–Y), we then asked how Wg expression in muscles is regulated in the nutrient-dependent manner. To search for upstream regulators of the Wg expression, we performed RNA-seq analysis on mature whole larvae that were reared on either diet. We identified 3854 differentially expressed genes between the diets (Figure 5A, Figure 5—figure supplement 1, and Supplementary file 3). Among these, we focused on a transcriptional factor in the JAK/STAT pathway, Stat92E, which is more highly expressed on HYD than LYD (Figure 5B). Also informing our decision, it was reported that Stat92E is a negative regulator of wg expression in the eye imaginal disc (Ekas et al., 2006). We used a Stat92E reporter strain (Bach et al., 2007) and found that Stat92E reporter expression in muscle was higher on HYD (Figure 5C–E). We therefore hypothesized that, in muscles of larvae on HYD, higher expression of Stat92E downregulates Wg expression, thereby suppressing the hyperarborization phenotype. To test this hypothesis, we knocked down Stat92E in muscles, and this led to increased Wg levels compared to the control muscles on HYD (Figure 5F–J; compare 5 F with 5 G). Furthermore, knocking down Stat92E or hopscotch (hop) encoding JAK in muscles promoted hyperarborization of C4da neurons in larvae reared on HYD (Figure 5K–T and Figure 5—figure supplement 2A, B, F, G, K, O). In contrast, overexpression of hop in muscles ameliorated the hyperarborization on LYD (Figure 5U–Z). These results indicate that JAK/STAT signaling contributes to the downregulation of wg expression in muscles and the suppression of dendritic hyperarborization on HYD (Figure 7).

Figure 5 with 2 supplements see all
Downregulation of Wg expression by Stat92E on HYD suppresses the hyperarborization phenotype.

(A and B) Plots of whole-body RNA-seq of wandering 3rd instar larvae reared on HYD or LYD. (A) The fold change (LYD/HYD) in read counts was plotted against average counts per million mapped reads (CPM) for HYD and LYD. Dots that are statistically supported (FDR ≤0.05) are colored (red for HYD >LYD and blue for HYD <LYD). (B) Plot of transcripts per million (TPM) of Stat92E. Adjusted p-value with Benjamini & Hochberg correction (FDR) is indicated. (C–E) Muscles of 10 x Stat-GFP larvae on HYD (C) or LYD (D). The signal intensities of GFP correspond to the indicated color code. (E) Quantification of 10 x Stat-GFP intensity in muscle 9 (Student’s t-test, n=20–27). (F–J) Muscles of control larvae (F and H) or larvae with Stat92E KD in muscles (G and I) on HYD or LYD were stained for Wg. The signal intensities correspond to the indicated color code. (J) Quantification of the mean Wg immunofluorescence in muscle 9 (Wilcoxon-Mann-Whitney test, n=17–22). (K–T) Images of ddaC neurons in control larvae (K and N), larvae with Stat92E KD in muscles (L and O), or larvae with hop KD in muscles (M and P), on HYD or LYD. 2D plots (Q and R) and densities of branch terminals (S and T, Wilcoxon-Mann-Whitney test, n=8–9). (U–Z) Images of ddaC neurons in control larvae (U and W) or larvae with hop overexpression in muscles (V and X), on HYD or LYD. 2D plot (Y) and densities of branch terminals (Z, two-way ANOVA, n=8). (AA-AF) Images of ddaC neurons in control larvae (AA and AC) or larvae with upd2 KD in the fat body and hemocytes (AB and AD), raised on HYD or LYD. 2D plot (AE) and densities of branch terminals (AF, Wilcoxon-Mann-Whitney test, n=8). The ddaC neurons were visualized by expressing ppk-CD4:tdGFP. Control data in (R) and (T) are shared with (Y) and (Z). Boxplots in (B, E, J, S, T, Z and AF) are depicted as in Figure 1C. *p<0.05, **p<0.01, and ***p<0.001. Scale bars, 100 μm.

It was previously reported that Upd2 secreted from the fat body activates JAK/STAT signaling through transmembrane receptor Domeless (Dome) in GABAergic neurons in the adult brain, which project onto insulin producing cells (IPCs), thereby regulating systemic growth in a nutritional-status-dependent manner (Rajan and Perrimon, 2012). It has also been shown that the secretion of Upd2 or Upd3 from hemocytes promotes the expression of a Stat92E reporter in larval muscle (Yang et al., 2015). These studies prompted us to address whether any Upds from the fat body or hemocytes, and Dome in muscles, contribute to the hyperarborization phenotype. Knocking down upd2, but not upd or upd3, in the fat body and hemocytes resulted in an increased terminal density on HYD (Figure 5AA-AF and Figure 5—figure supplement 2S-AN). This effect of upd2 KD in the fat body and hemocytes is similar to that of Stat92E or hop KD (Figure 5K–T) and that of wg overexpression in muscles (Figure 3T–Y). Enhanced branching on HYD was also seen in a dome KD in one out of three RNAi lines (Figure 5—figure supplement 2C–E, H-J, L-N and P-R ). Although it is necessary to verify KD of dome in the future, these results are suggestive of the role of fat body (and hemocytes)–muscle inter-organ communication through a Upd2-Stat92E pathway in suppressing the hyperarborization phenotype on HYD. To address whether the key nutrients (vitamins, metal ions, and cholesterol) increase Stat92E expression in muscles, we examined the reporter expression on LYD supplemented with or without VMC. However, the addition of VMC to LYD did not increase the signal intensity of the Stat92E reporter (data not shown). This result contrasts with the decreased level of Wg in response to the key nutrients (Figure 3M–P). The Stat92E KD caused only a marginal Wg increase on HYD compared to the difference in the amount of Wg between HYD and LYD (Figure 5F, H and J). Considering these results, it is likely that an additional unknown molecular mechanism other than the JAK/STAT pathway contributes to the high VMC-mediated downregulation of Wg in muscles (Figure 7).

LYD blunts light responsiveness of C4da neurons and larval light avoidance behavior

C4da neurons sense noxious thermal, mechanical, and light stimuli (Chin and Tracey, 2017). We therefore examined how our dietary conditions affect the electrophysiological activity of C4da neurons and larval behavior (Figure 6). First, we compared firing activities of C4da neurons in larvae that were reared on either HYD or LYD. As a noxious stimulus, we illuminated entire arbors of recorded neurons with blue light (Xiang et al., 2010; Terada et al., 2016). We used extracellular recording to monitor both spontaneous and evoked activities (Figure 6A–C). The frequency of spontaneous firing was higher in C4da neurons from larvae reared on LYD than on HYD (Figure 6D). Regarding the response to the light stimulus, all relevant parameters, i.e., the firing frequency, the change amount, and the change rate, were lower on LYD than on HYD (Figure 6E–G, see definition of the parameters in the legend), indicating that C4da neurons on LYD are less sensitive to the stimulus. Next, we examined whether the blunted light responsiveness of the neurons affects larval avoidance behavior. It was reported that Drosophila larvae prefer dark places to avoid noxious light, and this light avoidance behavior requires the activity of C4da neurons (Yamanaka et al., 2013; Imambocus et al., 2022). We speculated that the blunted light responsiveness of larvae on LYD may cause declines in their light avoidance behavior, and this may allow larvae to continue their search for high-nutrient food. To address this possibility, we conducted light/dark choice assays in which larvae reared on HYD or LYD were allowed to choose between dark and bright areas. We found that both foraging 3rd instar and wandering 3rd instar larvae on LYD showed lower preference for dark places than larvae on HYD (Figure 6H and I). Furthermore, when Ror was knocked down in C4da neurons, differences in light avoidance behavior between the diets tended to be smaller than the control larvae (Figure 6H and I). Our results suggest that the hyperarborization of C4da neurons is associated with blunted light avoidance behavior.

LYD blunts light responsiveness of C4da neurons and larval light avoidance behavior.

(A) A schematic diagram outlining the electrophysiological analysis. Firing activities of C4da neurons v’ada were recorded by measuring the extracellular membrane potential. After spontaneous firings were recorded for about 20 s, activities during blue light irradiation were monitored for 5 s. (B and C) Firing activities of C4da neurons on HYD or LYD. (B) Raster plots of firing. Blue shading indicates the 5 s blue light irradiation. Each row in the plots represents the data for a single cell. (C) Peristimulus time histograms calculated at 250 ms bins, on HYD (red line) or LYD (blue line). (D–G) Quantitative analysis of the firing activities. (D) Spontaneous firing frequency. (E) Firing frequency during blue light irradiation. (F) Change amount of the firing response to the blue light stimulus calculated by subtracting [spontaneous firing frequency] from [firing frequency during blue light irradiation] (G) Change rate of the firing response to the blue light stimulus calculated by dividing [firing frequency during blue light irradiation] by [spontaneous firing frequency]. (Wilcoxon-Mann-Whitney test, n=15). (H and I) Schematic diagram of light/dark choice assays and dark preference index of foraging 3rd instar larvae on agar plates (H), and wandering 3rd instar larvae in plastic tubes (I). Control larvae and larvae with Ror KD in C4da neurons were tested. (Wilcoxon-Mann-Whitney test, n=12–16) Boxplots in (D–I) are depicted as in Figure 1C. *p<0.05, **p<0.01, and ***p<0.001.

Discussion

Collectively, our studies illustrate how selective nutrients in the food impact neuronal development through inter-organ signaling (Figure 7). Yeast has long been considered as a rich source of amino acids for Drosophila; however, our results suggest that C4da neurons increase their dendritic terminal density on the LYD due to a combined deficiency in vitamins, metal ions, and cholesterol. This result is unexpected, because previous studies on nutrition-dependent cell growth has focused primarily on the TOR signaling pathway, which is activated by amino acids (González et al., 2020; Liu and Sabatini, 2020). The addition of the above nutrient trio to LYD did not restore the branch terminal density to the same level as HYD. This may indicate that the balance of concentrations among these nutrients was not optimized or that unknown nutrients may need to be added along with these nutrients. In addition, we showed that up-regulation of Wg expression in muscle on LYD was suppressed by supplementation of these components to the diet. This regulation may be achieved at least at a transcriptional level possibly by an interplay between a hypothetical nutrient-responsive module in the cis-element of wg and transcription factors and/or epigenetic machineries that require vitamins, vitamin-derived metabolites, metal ions and cholesterol (Harris et al., 2016; Kambe et al., 2016). Further investigations are necessary to understand the detailed molecular mechanisms underlying the combined effects of these components on wg expression. It has been reported that increasing or decreasing the amount of yeast in foods causes various responses in Drosophila (Bass et al., 2007; Okamoto and Nishimura, 2015), and the approach used in this study may help to clarify which nutrients are the key factors that cause those responses.

Model of the low-nutrient dependent dendritic hyperarborization.

Compared to C4da neurons of HYD-fed larvae (top), those of LYD-fed larvae (bottom) increase in dendritic branching (hyperarborization) due to a combined deficiency of vitamins, metal ions, and cholesterol in the food (‘Low VMC’). In the LYD-fed larvae, wg expression in muscles is higher than in muscles in the HYD-fed larvae; secreted Wg is bound by receptors Ror and Fz2 in C4da neurons, which in turn hyperactivate Akt signaling, thereby promoting dendrite branching. In the C4da neurons, other receptors (e.g. InR or Alk) upstream of Akt and intracellular components of Wnt signaling (‘Other factors’ such as Dsh and Bsk) might also contribute to the hyperarborization phenotype. The cellular response of C4da neurons is associated with a whole-animal level response (blunted light avoidance behavior). In the HYD-fed larvae, wg expression is suppressed partly by Upd2-Hop-Stat92E signaling and partly by an abundant VMC-mediated unknown molecular mechanism (red T bar and ‘?’). See Results and Discussion for details.

Inter-organ Wg/Ror/Akt signaling-mediated the hyperarborization phenotype

Previous studies demonstrated the coordinate growth control of dendrites of C4da neurons and the epidermis (Parrish et al., 2009; Jiang et al., 2014), from which a separate model underlying dendritic hyperarborization has evolved (Poe et al., 2020). In that model, stress sensor FoxO is expressed less in neurons than in neighboring epidermal cells, which results in lower levels of autophagy and less suppression of Tor signaling in the neuron, thereby ensuring dendritic growth even under a low-yeast condition (Poe et al., 2020). In contrast to the above model, our model highlights the signaling between neurons and another adjacent tissue: muscles secrete Wg, while C4da neurons express the receptor complex Ror-Fz2 on their cell surface. Therefore, it is likely that both the extrinsic Wg-dependent mechanism and the FoxO-dependent intrinsic sparing mechanism work together to generate the hyperarborization phenotype.

Ror is mainly expressed in the nervous system. No significant abnormalities in neuronal morphology including that of the C4da neurons were observed in Ror mutants under standard dietary conditions (Ripp et al., 2018; Nye et al., 2020). Consistent with these reports, there was no significant difference in morphological features of dendritic arbors of C4da neurons between Ror mutant or KD larvae and control larvae under the nutrient-rich HYD condition (Figure 2L and R). Because Ror is required for the hyperarborization under the hypotrophic condition (LYD in this study; Figure 2N and T) and for dendrite regeneration through the regulation of microtubule nucleation (Nye et al., 2020), it could be an adaptive agent that copes with environmental stress or damage. The pathway for microtubule nucleation includes Dsh and Axn, whose KD appeared to ameliorate the hyperarborization of C4da neurons. It remains to be seen how Dsh and Axn contribute to the hyperarborization on LYD, and whether the Wg-Ror-Akt pathway and the Ror-mediated microtubule-nucleation mechanism intersect. A number of other RTKs, such as insulin/IGF receptors and EGFR, activate Akt in other cellular contexts such as growth and proliferation of various stem cells and mammalian cancer cells (Shim et al., 2013; Butti et al., 2018). It is likely that InR also functions upstream of Akt in C4da neurons (Parrish et al., 2009; Shimono et al., 2014; Poe et al., 2020). Future studies will explore whether InR and other RTKs indeed function upstream of Akt in the context of the Wg/Ror/Akt signaling, and if so, how these various inputs are integrated by Akt to realize the nutritional status-dependent dendrite branching of C4da neurons (Figure 7).

The muscle is not only an energy-consuming organ, but it also plays an important role in regulating metabolic signaling through inter-organ communication with other tissues such as the brain and the fat body (Bretscher and O’Connor, 2020). In the adult stage, for example, muscle-derived Wg regulates lipid storage in the fat body (Lee et al., 2014). Our study revealed that muscle-derived Wg, which is up-regulated in response to low levels of vitamins, metal ions, and cholesterol, regulates dendrite branching of C4da neurons in the larval stage. Therefore, the muscle functions as a mediator of the nutritional status to other peripheral tissues in both growing and adult stages, and Wnt signaling may play a pivotal role in fulfilling this metabolic response function throughout the life cycle.

In our search for regulatory mechanisms of wg expression, we found that Stat92E reporter expression was higher in muscles on HYD than on LYD. This finding is reminiscent of Stat92E reporter expression in a population of GABAergic neurons in the adult brain, which project onto the Insulin producing cells (IPCs) and inhibit the release of Dilps (Rajan and Perrimon, 2012). This reporter expression in the GABAergic neurons also varies in a nutritional-status-dependent manner: the expression is higher on a standard laboratory food containing yeast compared to a sucrose-only condition. Our results suggest that Upd2-Stat92E signaling partially contributes to downregulation of Wg in muscles and suppression of the hyperarborization phenotype. When stat92E was knocked down in muscles, Wg increased on HYD; however, that increase was marginal compared to the difference in the level of Wg expression between HYD and LYD (Figure 5F, G, H and J). Moreover, the addition of a combination of vitamins, metal ions, and cholesterol to LYD did not increase the signal intensity of a Stat92E reporter (data not shown). Therefore, it is likely that in the HYD-fed larvae, an unknown VMC-mediated molecular mechanism also contributes to downregulation of wg in muscles (red T bar with ‘?’ in Figure 7). Further investigation is required to elucidate how Wg expression in muscles is controlled at the molecular level in such a key nutrient(s)-dependent manner and how the Upd2-Stat92E pathway contributes to the entire mechanism of inter-organ communication.

Physiological roles of C4da neurons and the hyperarborization phenotype

What are the implications of the inter-organ signaling mechanism controlling dendritic branches in the context of nutritional adaptation? It has been reported that a wide range of animals tend to take more risks when they are hungry (Symmonds et al., 2010; Filosa et al., 2016; Padilla et al., 2016; Bräcker et al., 2013). Our electrophysiological analysis indicates that C4da neurons on LYD decrease light-evoked response (Figure 6A–G). Consistently, larvae reared on LYD displayed decreases in their dark preference, compared to those reared on HYD, and they explored bright places, which is potentially risky for their survival (Control of Figure 6H and I). This difference between the diets tended to become smaller once the inter-organ signaling mechanism was suppressed in C4da neurons (ppk >Ror RNAi in Figure 6H and I). These results imply that the hyperarborization of C4da neurons on LYD might contribute to blunting light avoidance behavior, although we cannot exclude the possibility that the Ror KD might affect neuronal functions through other mechanisms than the dendritic morphological change. Our study raises the possibility that nutrient-dependent development of somatosensory neurons plays a role in optimizing a trade-off between searching for high-nutrient foods and escaping from noxious environmental threats. Although a recent study described the circuitry required for the larval light avoidance behavior, it remains unclear whether the possible modifications of neural circuits downstream of C4da neurons take part in this behavioral transition (Imambocus et al., 2022). The identification of the downstream circuits would allow further study of the relationship between nutrient-dependent neural differentiations and evoking risk-taking behavior.

In contrast to our results, a previous study reported that larvae with hyperarborized C4da neurons react more quickly to noxious heat (Poe et al., 2020). While light-induced Ca2+ activity in C4da neurons decreases, thermal nociceptive behavior increases during 2nd and 3rd instar larval periods (Jaszczak et al., 2022), which indicates that these nociceptive responses are regulated in the opposite direction or in distinct fashions. Therefore, seemingly contradictory results between the previous study and ours may be due to different regulatory mechanisms of the sensory modalities.

The relationship between nutritional status and neural development has often been studied epidemiologically (Prado and Dewey, 2014; Bhutta et al., 2017). Our study, which presents a mechanism by which quantitative changes in specific nutrients act on neuronal morphology and operate through inter-organ signaling, provides a stepping stone for future explorations of molecular mechanisms linking nutrition and development of other neuronal cell types and in other animal species.

Materials and methods

Drosophila strains and fly culture

Request a detailed protocol

Fly strains used in this study are listed in Key Resources Table. Our stocks are usually reared on a laboratory standard diet (Watanabe et al., 2017). Adult males and virgin females that had developed on the standard diet were collected and crossed on the standard diet for 3–5 days. Then, the adults were transferred into vials containing HYD or LYD, which were identical to the semidefined medium (SDM)-based diet (8% Y) and the SDM-based diet (0.8% Y), respectively (see Supplementary file 1 and its legend in Watanabe et al., 2017). After an egg-laying interval, the adult flies were cleared in every experiment and wandering 3rd instar larvae that came out of individual diets were used. Larvae were reared under noncrowded conditions at 25 °C in all the experiments except the wg overexpression experiments at 29 °C. Our recombinant DNA experiments follow Kyoto University Regulations for Safety Management in Recombinant DNA Experiments under protocol # 210059.

Experimental diets

Request a detailed protocol

We cooked the high yeast diet (HYD) or low yeast diet (LYD) based on semidefined media (SDM) as described previously (Watanabe et al., 2017). The original SDM recipe is described at the Bloomington Drosophila Stock Center https://bdsc.indiana.edu/information/recipes/germanfood.html. HYD and LYD were composed of brewer’s yeast (MPBio 2903312), glucose (Wako 049–31165), sucrose (Wako 196–00015), peptone (Fluka 82303), and agar (Matsuki Kanten). The complete compositions of these diets can be found in Supplementary file 1. After the ingredients were mixed, water was added to a final volume of 200 ml, followed by autoclaving. Once the foods had cooled, 1.2 ml propionic acid (Nacalai Tesque, 29018–55) and 2 ml 10% butyl p-hydroxybenzoate (Nacalai Tesque, 06327–02) in 70% ethanol were added. The foods were then dispensed into vials and left overnight before use.

For the supplementation with essential amino acid solution, we used 50 x MEM EAA solution (Wako 132–15641). Each fraction of holidic medium (Piper et al., 2014; Piper et al., 2017) other than amino acids (vitamins, cholesterol, metal ion and other ingredients) was added to LYD at 10 times the concentration in holidic medium. Amino acids mixture from holidic medium was added at a 1 x or 3 x concentration. The complete compositions of diets used in nutrient supplementation experiments can be found in Supplementary file 1.

Imaging and quantification for assessing dendritic morphology

Request a detailed protocol

Images of ddaC (C4da) or ddaF (C3da) neurons in A3–A5 segments were acquired in live whole-mount larvae as described (Hattori et al., 2013; Parrish et al., 2009; Matsubara et al., 2011). Protocols for single-cell labeling (MARCM) were as previously described (Shimono et al., 2014). For quantification of the number of dendritic branching terminals, we drew an outline of the dendritic field as a region of interest (ROI) by connecting the outermost dendritic terminals with the Adobe Photoshop path tool. Then, dendritic branching terminals inside the ROI of ddaC were automatically counted using DeTerm (Kanaoka et al., 2019). Concerning counting short spikes of ddaF, results obtained by DeTerm were corrected manually. In addition, the area size of the ROI was measured as the dendritic coverage size. Dendrite length of ddaC neurons was measured using Fiji/ImageJ as previously described (Poe et al., 2017). Briefly, images of dendrites were processed sequentially by Gaussian Blur, Auto Local Threshold, Particles4, and Skeletonize (2D/3D), and the length of the 1-pixel-width skeleton was measured inside the ROI. Some representative control images and control data are shared by multiple figures. See figure legends.

Preparation of larvae with developmental delay

Request a detailed protocol

In the experiments in which larval growth was delayed by excess sucrose, we collected eggs as previously described (Watanabe et al., 2019) and placed them on either HYD, LYD, or HYD +sucrose in vials. The complete composition of HYD + sucrose can be found in Supplementary file 1. In a dilp8 overexpression experiment (Colombani et al., 2012), the adult flies were allowed to oviposit on HYD or LYD for 24 hr. The timing of the neuronal observations under each condition is indicated in Figure 1—figure supplement 2.

RTK screening

Request a detailed protocol

We conducted two rounds of screening. In the primary screening, we intended to enhance KD efficacy and used Gr28b.c-GAL4 and ppk-GAL4 together. We acquired images of 3–8 knocked-down neurons for each gene on each diet, and then visually judged whether hyperarborization was blunted or not. We selected nine genes (Ror, InR, Alk, htl, Egfr, Pvr, Ddr, dnt, and drl) for the secondary screening, in which we used only ppk-GAL4 because Gr28b.c-GAL4 is expressed in a small subset of neurons in the central nervous system in addition to C4da neurons in the peripheral nervous system (Xiang et al., 2010). See Supplementary file 2 for names of the 20 RTK genes and stock numbers of RNAi lines used.

Immunostaining

Request a detailed protocol

Dissected wandering 3rd instar larvae were fixed in a 1:10 dilution of Formaldehyde Solution (Nacalai Tesque, 16222–65) in PBS plus 0.05% Triton X-100 for 30 min, then washed three times in PBS plus 0.1% Triton X-100 (PBST). After blocking in PBST plus 2% bovine serum albumin for 30 min, primary antibodies listed in Key Resources Table were added, then incubated overnight at 4 °C. After three successive washes, secondary antibodies were added, then incubated for 1 hr at room temperature. Finally, samples were mounted using FluorSave Reagent (Calbiochem). Most of the images were acquired with a Nikon C1 laser scanning confocal microscope coupled to a Nikon Eclipse E-800 microscope. The images in Figure 2—figure supplement 5 were acquired with a ZEISS LSM 800 microscope.

Quantification of signal intensity

Request a detailed protocol

To quantify signal levels in muscle, we made Z-stack images and chose muscle 9, one of the closest muscles to ddaC, for measuring the signal intensity. For quantification of Wg or 10 x Stat-GFP signals, we measured the signal intensity inside a 19 μm or a 27 μm square ROI, respectively. Three ROIs were drawn for each muscle, and the average value was calculated. For quantification of RedStinger driven by wg-GAL4, the signal intensity inside nuclei that were identified by DAPI signals was measured. Then, the values from 2 to 7 nuclei in each muscle were averaged. For quantification of p-Akt levels in cell bodies of C4da ddaC, C1da ddaE or C3da ddaF neurons, we selected the single section containing the strongest signal in the neurons and measured signal intensities inside a 1.7 μm square ROI located on the abdominal side of nuclei of the neurons. However, for the quantification in ddaF neurons, the ROI was placed on the ventral side of nuclei only when we could not identify the border between ddaF and ddaC.

Electrophysiology

Request a detailed protocol

Extracellular single-unit recordings in wandering 3rd instar larvae were performed as previously described (Terada et al., 2016; Onodera et al., 2017). We recorded the activity of v’ada of C4da neurons, which showed hyperarborization on LYD. For blue light irradiation, the 460–495 nm light at 72mW/mm2 power was projected onto larvae for 5 s. The light spot was 1.5 mm in diameter. Peristimulus time histograms were calculated at 250 ms bins. The mean spontaneous firing frequencies were quantified in the 20 s window preceding the light stimuli. The mean firing frequencies during the light stimulation were quantified in the 5 s entire window. The firing changes were calculated by subtracting the mean spontaneous firing rate from the light-evoked one (Change amount) or using a ratio of the mean spontaneous firing rate divided by the mean light-evoked one (Change rate).

Light/dark choice assay

Request a detailed protocol

Light/dark choice assays were performed as previously described with modifications (Yamanaka et al., 2013). For the assay of foraging larvae, we used foraging 3rd instar larvae one day before they start wandering. We prepared 2% agar plate with a lid half of which was covered with black tape and 20 larvae were placed along the junction between light and dark sides. After the plates were illuminated for 15 min with white LED light (OHM, ODS-LKL6-W) at 700 lux, the number of larvae in both dark and light areas were counted. In some trials, one or two larvae dug into the agar. Such larvae were excluded from calculation of dark preference index. For the assay of wandering larvae, two opposed plastic tubes were joined by transparent scotch tape and one of the vials was covered with black tape. After 16 wandering larvae reared on HYD or LYD were put near the junction of the tubes, they were illuminated by the 700 lux light for 15 min, then the number of larvae in both dark and light areas were counted. Dark preference index was calculated as follows:

((Number of larvae in dark)(Number of larvae in light))/(Total number of larvae)

RNA-seq

Request a detailed protocol

Protocols for sample preparation and data analysis of RNA-seq were essentially as described in Watanabe et al., 2019. To prepare each replicate, RNA was extracted from five whole bodies of male wandering third-instar larvae. The following procedures are different from Watanabe et al., 2019: (1) the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (NEB, E7760) was used for library preparation. (2) RNA-sequencing was performed on an Illumina NextSeq 500 system using single end reads. (3) All raw sequencing data were trimmed using TrimGalore (ver. 0.6.0, Cutadapt ver. 1.18; DOI:10.5281/zenodo.5127899, DOI:10.14806/ej.17.1.200) with -clip_R1 13 option. (4) Gene-based read counts were obtained using htseq-count (ver. 0.11.3; Anders et al., 2015) with -s reverse -a 10 options. (5) Differential expression analysis was performed on the count data using a generalized linear model (GLM) in the edgeR Bioconductor package (ver. 3.30.3; McCarthy et al., 2012; Robinson et al., 2010). All the RNA-sequencing data have been deposited and are available in the DDBJ Sequence Read Archive. The accession numbers for the data are DRR311224-DRR311229 (BioProject accession number: PRJDB12048).

Statistical analysis

Request a detailed protocol

R (R Core Team) was used for stastical analysis. Values of P<0.05 were considered statistically different. Student’s t-test or the Wilcoxon-Mann-Whitney test was used for two-group comparisons, and Dunnett’s test, Steel test, or Steel-Dwass tests were used for multiple comparisons. We used two-way analysis of variance (ANOVA) to analyze interactive effects between genotype and diet. On the other hand, we used the two-group comparison tests (Student’s t-test or the Wilcoxon-Mann-Whitney test) when we simply focused on whether a genetic manipulation itself affected dendrite branching on the same diet. Statistical tests used, the exact sample size (n), and p values are shown in Supplementary file 4. R was also used to draw 95% confidence ellipses. See also figure legends for details.

Appendix 1

Appendix 1—key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
genetic reagent (Drosophila melanogaster)ppk-GAL4 UAS-mCD8:GFPGrueber et al., 2007 (https://doi.org/10.1242/dev.02666)N/A
genetic reagent (D. melanogaster)Gr28b.c-GAL4 UAS-mCD8:GFPXiang et al., 2010 (https://doi.org/10.1038/nature09576)N/A
genetic reagent (D. melanogaster)ppk-CD4-tdGFP1bHan et al., 2011 (https://doi.org/10.1073/pnas.1106386108)N/A
genetic reagent (D. melanogaster)ppk-CD4-tdGFP8Han et al., 2011 (https://doi.org/10.1073/pnas.1106386108)N/A
genetic reagent (D. melanogaster)ppk-CD4-tdTom4aHan et al., 2011 (https://doi.org/10.1073/pnas.1106386108)N/A
genetic reagent (D. melanogaster)Ror4Ripp et al., 2018 (https://doi.org/10.1242/bio.033001)N/A
genetic reagent (D. melanogaster)ppk-GAL4Bloomington Drosophila Stock CenterStock #: 32079
genetic reagent (D. melanogaster)Mhc-GAL4Schuster and Davis, 1996 (https://doi.org/10.1016/s0896-6273(00)80,197x)N/A
genetic reagent (D. melanogaster)Cg-GAL4Bloomington Drosophila Stock CenterStock #: 7011
genetic reagent (D. melanogaster)apts78j wg1KYOTO Stock CenterStock #: 107069
genetic reagent (D. melanogaster)wgl-8 cn1 bw1 speck1KYOTO Stock CenterStock #: 107019
genetic reagent (D. melanogaster)wg-GAL4Bloomington Drosophila Stock CenterStock #: 83627
genetic reagent (D. melanogaster)UAS-RedStingerBloomington Drosophila Stock CenterStock #: 8547
genetic reagent (D. melanogaster)UAS-Akt RNAi (BL33615)Bloomington Drosophila Stock CenterStock #: 33615
genetic reagent (D. melanogaster)UAS-Akt RNAi (v2902)Vienna Drosophila Resource CenterStock #: 2902
genetic reagent (D. melanogaster)UAS-Ror RNAiNational Institute of GeneticsStock #: 4926 R-1
genetic reagent (D. melanogaster)UAS-wg RNAiVienna Drosophila Resource CenterStock #: 6692
genetic reagent (D. melanogaster)UAS-Stat92E RNAi (BL33637)Bloomington Drosophila Stock CenterStock #: 33637
genetic reagent (D. melanogaster)UAS-hop RNAiBloomington Drosophila Stock CenterStock #: 32966
genetic reagent (D. melanogaster)UAS-upd2 RNAi (5988 R-1)National Institute of GeneticsStock #: 5988 R-1
genetic reagent (D. melanogaster)UAS-hopBloomington Drosophila Stock CenterStock #: 79033
genetic reagent (D. melanogaster)UAS-myrAktBloomington Drosophila Stock CenterStock #: 50758
genetic reagent (D. melanogaster)UAS-wg.H.T:HA1KYOTO Stock CenterStock #: 108488
genetic reagent (D. melanogaster)10XSTAT92E-GFPBloomington Drosophila Stock CenterStock #: 26197
genetic reagent (D. melanogaster)UAS-InR RNAi (BL31594)Bloomington Drosophila Stock CenterStock #: 31594
genetic reagent (D. melanogaster)UAS-InR RNAi (BL51518)Bloomington Drosophila Stock CenterStock #: 51518
genetic reagent (D. melanogaster)UAS-grnd RNAiVienna Drosophila Resource CenterStock #: 43454
genetic reagent (D. melanogaster)UAS-wgn RNAiVienna Drosophila Resource CenterStock #: 9152
genetic reagent (D. melanogaster)UAS-mth RNAiVienna Drosophila Resource CenterStock #: 102303
genetic reagent (D. melanogaster)UAS-babo RNAiBloomington Drosophila Stock CenterStock #: 25933
genetic reagent (D. melanogaster)UAS-Tor.TEDBloomington Drosophila Stock CenterStock #: 7013
genetic reagent (D. melanogaster)UAS-S6k.KQBloomington Drosophila Stock CenterStock #: 6911
genetic reagent (D. melanogaster)UAS-Tif-IA RNAiVienna Drosophila Resource CenterStock #: 20336
genetic reagent (D. melanogaster)UAS-Thor RNAiVienna Drosophila Resource CenterStock #: 35439
genetic reagent (D. melanogaster)UAS-foxo RNAiBloomington Drosophila Stock CenterStock #: 32427
genetic reagent (D. melanogaster)UAS-Alk RNAi (v11446)Vienna Drosophila Resource CenterStock #: 11446
genetic reagent (D. melanogaster)UAS-Alk RNAi (BL107083)Vienna Drosophila Resource CenterStock #: 107083
genetic reagent (D. melanogaster)mef2-GAL4Bloomington Drosophila Stock CenterStock #: 27390
genetic reagent (D. melanogaster)R38F11-GAL4Bloomington Drosophila Stock CenterStock #: 50014
genetic reagent (D. melanogaster)UAS-fz RNAiVienna Drosophila Resource CenterStock #: 43075
genetic reagent (D. melanogaster)UAS-fz2 RNAiNational Institute of GeneticsStock #: 9739 R-1
genetic reagent (D. melanogaster)“GAL45-40 UAS-Venus:pm SOP-FLP#42; tubPGal80 FRT2A”KYOTO Stock CenterStock #: 109950
genetic reagent (D. melanogaster)“w*; FRT2A”KYOTO Stock CenterStock #: 106623
genetic reagent (D. melanogaster)“y w hs-flp; fz2C2 FRT2A”Chen and Struhl, 1999 (https://doi.org/10.1242/dev.126.23.5441)N/A
genetic reagent (D. melanogaster)UAS-dsh RNAiVienna Drosophila Resource CenterStock #: 101525
genetic reagent (D. melanogaster)UAS-bsk.DNBloomington Drosophila Stock CenterStock #: 6409
genetic reagent (D. melanogaster)UAS-DAAM RNAiVienna Drosophila Resource CenterStock #: 24885
genetic reagent (D. melanogaster)UAS-arm RNAiVienna Drosophila Resource CenterStock #: 7767
genetic reagent (D. melanogaster)UAS-norpA RNAiBloomington Drosophila Stock CenterStock #: 31113
genetic reagent (D. melanogaster)UAS-Axn RNAiBloomington Drosophila Stock CenterStock #: 31703
genetic reagent (D. melanogaster)UAS-Stat92E RNAi (BL31318)Bloomington Drosophila Stock CenterStock #: 31318
genetic reagent (D. melanogaster)UAS-dome RNAi (v106071)Vienna Drosophila Resource CenterStock #: 106071
genetic reagent (D. melanogaster)UAS-dome RNAi (BL32860)Bloomington Drosophila Stock CenterStock #: 32860
genetic reagent (D. melanogaster)UAS-dome RNAi (BL34618)Bloomington Drosophila Stock CenterStock #: 34618
genetic reagent (D. melanogaster)UAS-upd RNAi (BL33680)Bloomington Drosophila Stock CenterStock #: 33680
genetic reagent (D. melanogaster)UAS-upd RNAi (v3282)Vienna Drosophila Resource CenterStock #: 3282
genetic reagent (D. melanogaster)UAS-upd2 RNAi (BL33949)Bloomington Drosophila Stock CenterStock #: 33949
genetic reagent (D. melanogaster)UAS-upd3 RNAi (BL32859)Bloomington Drosophila Stock CenterStock #: 32859
genetic reagent (D. melanogaster)UAS-upd3 RNAi (BL28575)Bloomington Drosophila Stock CenterStock #: 28575
genetic reagent (D. melanogaster)rn-GAL4St Pierre et al., 2002 (https://doi.org/10.1242/dev.129.5.1273)N/A
genetic reagent (D. melanogaster)UAS-dilp8Colombani et al., 2012 (https://doi.org/10.1126/science.1216689)N/A
genetic reagent (D. melanogaster)UAS-htl RNAi (v6692)Vienna Drosophila Resource CenterStock #: 6692
genetic reagent (D. melanogaster)UAS-htl RNAi (BL35024)Bloomington Drosophila Stock CenterStock #: 35024
genetic reagent (D. melanogaster)UAS-Egfr RNAiVienna Drosophila Resource CenterStock #: 43267
genetic reagent (D. melanogaster)UAS-Pvr RNAiVienna Drosophila Resource CenterStock #: 13502
genetic reagent (D. melanogaster)UAS-Ddr RNAiVienna Drosophila Resource CenterStock #: 29720
genetic reagent (D. melanogaster)UAS-dnt RNAiNational Institute of GeneticsStock #: 17,559 R-3
genetic reagent (D. melanogaster)UAS-drl RNAiBloomington Drosophila Stock CenterStock #: 29602
genetic reagent (D. melanogaster)UAS-Eph RNAiBloomington Drosophila Stock CenterStock #: 28511
genetic reagent (D. melanogaster)UAS-otk RNAiBloomington Drosophila Stock CenterStock #: 25790
genetic reagent (D. melanogaster)UAS-sev RNAiBloomington Drosophila Stock CenterStock #: 31274
genetic reagent (D. melanogaster)UAS-btl RNAiVienna Drosophila Resource CenterStock #: 110277
genetic reagent (D. melanogaster)UAS-Cad96Ca RNAiVienna Drosophila Resource CenterStock #: 1089
genetic reagent (D. melanogaster)UAS-CG10702 RNAiVienna Drosophila Resource CenterStock #: 27052
genetic reagent (D. melanogaster)UAS-Drl-2 RNAiVienna Drosophila Resource CenterStock #: 40484
genetic reagent (D. melanogaster)UAS-Nrk RNAiVienna Drosophila Resource CenterStock #: 9653
genetic reagent (D. melanogaster)UAS-Ret RNAiVienna Drosophila Resource CenterStock #: 107648
genetic reagent (D. melanogaster)UAS-tor RNAiVienna Drosophila Resource CenterStock #: 36280
genetic reagent (D. melanogaster)UAS-Tie RNAiVienna Drosophila Resource CenterStock #: 26879
genetic reagent (D. melanogaster)Gal419-12Xiang et al., 2010 (https://doi.org/10.1038/nature09576)N/A
antibodyanti-Wg (Mouse monoclonal)Developmental Studies Hybridoma BankCat# 4D4, RRID: AB_528512IF (1:15)
antibodyanti-Futch (Mouse monoclonal)Developmental Studies Hybridoma BankCat# 22C10, RRID: AB_528403IF (1:20)
antibodyanti-phospho-Akt (Rabbit polyclonal)Cell SignalingCat# 9271 SIF (1:100)
antibodyanti-DsRed (Rabbit polyclonal)ClontechCat# 632496IF (1:250)
antibodyanti-Mouse IgG Alexa Fluor 488 (Goat polyclonal)InvitrogenCat# A11029IF (1:1000)
antibodyanti-Rabbit IgG Alexa Fluor 488 (Goat polyclonal)InvitrogenCat# A11034IF (1:1000)
antibodyanti-Mouse IgG Alexa Fluor 546 (Goat polyclonal)InvitrogenCat# A11030IF (1:1000)
chemical compound, drugBrewer’s YeastMPBioCat# 2903312
chemical compound, drugYeast extractSigma-Aldrich FlukaCat# 70161
chemical compound, drugPeptone from casein, enzymatic digestSigma-Aldrich FlukaCat# 82303
chemical compound, drugGlucoseWakoCat# 049–31165
chemical compound, drugSucroseWakoCat# 196–00015
chemical compound, drugMgSO4WakoCat# 132–00435
chemical compound, drugCaCl2WakoCat# 031–00435
chemical compound, drugagarMatsuki KantenN/A
chemical compound, drugAgar Purified, powderNacalai TesqueCat# 01162–15
chemical compound, drugpropionic acidNacalai TesqueCat# 29018–55
chemical compound, drugbutyl p-hydroxybenzoateNacalai TesqueCat# 06327–02
chemical compound, drug50 x MEM Essential Amino Acids SolutionWakoCat# 132–15641
chemical compound, drugBSANacalai TesqueCat# 01863–77
chemical compound, drugAlexa Fluor 488 PhalloidinInvitrogenCat# A12379
chemical compound, drugFluorSave ReagentCalbiochemCat# 345789
chemical compound, drugFormalinNacalai TesqueCat# 16222–65
chemical compound, drugDAPINacalai TesqueCat# 19178–91
software, algorithmDeTermKanaoka et al., 2019 (
https://doi.org/10.1111/gtc.12700)
N/A
software, algorithmRR Core TeamRRID:SCR_001905
software, algorithmFijiNIHRRID:SCR_002285
software, algorithmPhotoshopAdobeRRID:SCR_014199
otherLED desk lightOHMCat# ODS-LKL6-WThis LED light was used in Light/dark choice assay

Data availability

All the RNA-sequencing data have been deposited and are available in the DDBJ Sequence Read Archive. The accession numbers for the data are DRR311224-DRR311229 (BioProject accession number: PRJDB12048).

The following data sets were generated
    1. Hattori Y
    2. Kanaoka Y
    3. Uemura T
    (2021) DDBJ Sequence Read Archive
    ID DRR311224-DRR311229. Transcriptome analysis of male Drosophila larvae reared on two different diets.

References

Decision letter

  1. K VijayRaghavan
    Senior and Reviewing Editor; National Centre for Biological Sciences, Tata Institute of Fundamental Research, India
  2. Fengwei Yu
    Reviewer; Temasek Life Sciences Laboratory, Singapore
  3. Cheng-Ting Chien
    Reviewer; Institute of Molecular Biology, Academia Sinica, Taiwan

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Inter-organ Wingless/Ror/Akt signaling regulates nutrient-dependent hype-arborization of somatosensory neurons" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and K VijayRaghavan as the Senior Editor. The following individuals involved in the review of your submission have agreed to reveal their identities: Fengwei Yu (Reviewer #1); Cheng-Ting Chien (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

One of the reviewers has consolidated and summarized our consultations on the individual reviews. These are given below. Please attend to each concern while responding. Please all see the individual reviews below.

1) Are Wg levels increased in muscles when Upd2, Dome, or stat93E KD is knocked down. Since wg transcription could be regulated in a complex way, the Dome/JAK/STAT pathway is likely insufficient to be the main or only regulator. This is also echoed by a result described in the Discussion that Stat92E was not elevated by VMC supplement in LYD. The authors should limit their description on the significance of these results such as the description in Abstract (and Figure 7) and conclusion on lines 276-279 in Results. Also, they could include more discussion on the possibility of other pathways.

2) Could authors compare the KD efficiency in two Akt RNAi lines (v2902 and BL33615) (or are they known in previous literature)? They also need to test the specificity of anti-Akt antibodies using these akt RNAi lines. Are anti-Akt signals gone when Akt is knocked down?

3) Was the dendrite hyper-arborization phenotype also observed in other types of da neurons? Did you see similar pAkt upregulation in other da neurons (class I-III) under the condition of LYD diets? If the authors have the results, it should be very nice to demonstrate the general and significant effects of their findings and mechanism.

4) The Akt mutant phenotype is quite different from the Ror mutant phenotype based on the images shown in Figure 2. It seems to affect more parameters of dendrite structure than Ror. The authors should elaborate on the phenotypic difference between Akt and Ror mutants, perhaps using different quantification approaches. It is therefore appears unlikely that Akt, which has a stronger and different phenotype from Ror, only mediates Ror signaling. Indeed, as the authors mentioned in the Discussion, Akt has previously been shown to regulate ddaC dendrite branching through signals from epithelial cells, and it has also been previously demonstrated, that, unlike Ror, Akt affects dendrite growth on standard diets (Parrish, 2009). The only data that suggests that Ror functions upstream of Akt are the pAkt staining shown in Figure 4. The authors need to confirm the specificity of the Akt antibody. In addition, the authors should address this concern by modifying the model in the Discussion (such as to include the possibility of excess growth by Ror knockdown neurons in LYD seems likely to be related to the function of Ror in mediating microtubule nucleation, the possibility of integrating InR and AlK in the pathway, which they have shown in this study in the supplemental data, as well as in a previous study (Poe, 2020)).

5) It looks like the major change for wg RNAi in muscle is a change in arbor area in LYD- indeed this often seems to drive or at least contribute to the phenotype (except in Akt loss). Is this because animals are overall larger? Since the arbors cover the body- is this increase in size related to overall size? If so, is the dendrite difference secondary to a much broader change in animal size? Are the same conclusions reached if a total number of terminals is shown?

6) The issue of lengthened developmental time on LYD is problematic and not sufficiently addressed. In a previous study on this topic (Poe), the continued growth of ddaC neurons was interpreted as resilience to malnutrition and shown to be controlled by lower levels of a stress transcription factor, foxo, than in other cells. In this interpretation, the ddaC neurons just continue to grow at their normal pace while the overall development of the animal is slowed. How is this view reconciled with the current model? Which conditions change developmental timing in this study? Do any of the genetic manipulations rescue developmental speed- and if so, is dendrite architecture changing secondarily to this?

7) In figure 6 – it would be helpful, if feasible speedily, to include Ror k/d in the electrophysiological assays. The behavioral assays look somewhat inconclusive in terms of whether Ror rescues the LYD phenotype.

8) On page 9 the authors mentioned screening 20 RTKs. It would be helpful to include this data. Certainly, two others they include – InR and Alk – look like they have similar phenotypes to Ror and it would be helpful to know the specificity.

9) There are two ways of statistics when manipulating gene activities (RNAi or OE) in HYD and LYD. Direct comparisons of two samples were used mostly in the main Figures (except Figure 5Z), while "difference in two differences" were shown mainly for several Supplemental figures (2, 3, 4, 5). It is not clear why the authors used two different ways. Are there specific reasons for doing so? Could not the direct comparisons of two samples (such as for main figures) could cope with descriptions in the text and easier to comprehend.

Reviewer #1 (Recommendations for the authors):

Nutrition profoundly affects neural development. The Uemura lab previously reported that C4da neurons elaborate complex dendrites when larvae grow on low-yeast diets, a phenomenon called neural sparing. In this current study, they elegantly show that vitamins, metal ions, and cholesterol, but not amino acids, are critical for these hyperaborization defects. They then define the molecular mechanism underlying the nutrition-mediated phenomenon. They identify that the Wingless/Ror/Akt pathway between the neuron and its adjacent muscles is necessary and sufficient to mediate dendrite over branching in the low-yeast condition. Moreover, they also identify a systematic Upd2-Stat92E pathway which is activated in muscles by the fat body. As a result, the low-yeast condition can unsensitized C4da neuron functions which may help larva to survive better under limited nutrition conditions. Overall, this is a comprehensive and important study that provides mechanistic insights into the question of how neurons over-branch in low-yeast diets. The experiments were well designed, and the results were properly interpreted. Most of the figures are well presented. Overall, this is an interesting and rigorously conducted study and the current version is ready to be published.

Reviewer #2 (Recommendations for the authors):

1. Akt knockdowns in two lines (v2902 and BL33615) display qualitative suppression of dendrite hyperarborization in Figure 2. In v2902 KD, while the dendritic arbor looks more normal in both HYD and LYD, the BL33615 KD showed quite a dramatic reduction even in HYD, and also in LYD, rendering both sets of data groups clustering on a distinct area (Figure 2I), instead of in between two controls (HYD and LYD) such as in Figure 2G. While these two lines have been used before, I am wondering if the BL33615 has a stronger KD effect and if the basal Akt is necessary for a fundamental process in normal dendrite development (which is still preserved in the v2902 line), irrespective of nutrient condition. Could authors compare the KD efficiency in these two lines (or are they known in previous literature)?

2. The wg signal derived from muscles is required for dendrite hyperarborization (Figure 3) is a very nice set of data, although the effect by overexpression of wg in muscle is limited in HYD (Figure 3Y). It could be due to further control such as from epidermal cells shown in Poe et al. (2020) eLife. The experiment was performed at 29C to increase Wg expression, and the dendrite morphology was not normal. I am not asking for more experiments, but the authors could describe/explain what is the results at 25C instead of using 29C for readers not familiar with the approach? Have the authors tried using 2 copies of UAS-wg to increase wg expression at 25C?

3. There are two ways of statistics used when manipulating gene activities (RNAi or OE) in HYD and LYD. Direct comparisons of two samples were used mostly in the main Figures (except Figure 5Z), while "difference in two differences" were shown mainly for several Supplemental figures (2, 3, 4, 5). It is not clear why the authors used two different ways. Are there specific reasons for doing so? I thought the direct comparisons of two samples (such as for main figures) could cope with descriptions in the text and be easier to comprehend.

4. The RNAi KD effects of Dome/JAK/STAT pathway components on inducing dendrite hyperarborization in HYD are quite variable! Three Dome RNAi KDs had no effect and one stat92E KD had an effect while the other had not! What are the effects on Wg expressions in muscles with these genetic manipulations? Since wg transcription could be regulated in a complex way, the Dome/JAK/STAT pathway is likely insufficient to be the main or only regulator. This is also echoed by a result described in the Discussion that Stat92E was not elevated by VMC supplement in LYD. The authors should limit their description on the significance of these results such as the description in Abstract (and Figure 7) and conclusion on lines 276-279 in Results. Also, they could include more discussion on the possibility of other pathways.

5. (line 217-218) "However, these results need careful interpretation (see the Figure 4—figure supplement 2 legend)" From the legend "However, knocking down dsh or blocking JNK signaling tended to increase the branch density over the control genotype on HYD, which may consequently reduce the differences in densities of dendritic terminals between the diets." I do not quite understand the reasoning here why some increases in HYD in the KD may reduce the differences in LYD? The explanation in the Figure legend is not clear!

Reviewer #3 (Recommendations for the authors):

The Akt phenotype is quite different from the Ror one based on the images shown in Figure 2. It looks like it causes a baseline phenotype in HYD, and overall seems to affect more parameters of dendrite structure than Ror. It is therefore unlikely that Akt, which has a stronger and different phenotype from Ror, is the downstream mediator of Ror signaling. Indeed, as the authors mention in the discussion, Akt has previously been shown to regulate ddaC dendrite branching through signals from epithelial cells, and it has also been previously demonstrated, that, unlike Ror, Akt affects dendrite growth on standard diets (Parrish, 2009).

In this study in the supplemental data, as well as in a previous study (Poe, 2020) (that interprets the overgrowth of ddaC neurons on LYD in a somewhat different way) InR has been shown to be important. It is not clear why it is dismissed in this study when it is much better established as a regulator of Akt than Ror.

The fact that Ror and Akt are interpreted as having similar effects on dendrite growth makes me very concerned about the quantitation method used. It is somewhat opaque and also at times dismissed (for example the growth phenotypes of InR and Alk are described as difficult to interpret and so sidelined, but it is unclear why).

The only data that suggests that Ror functions upstream of Akt are the phosphor-Akt staining shown in Figure 4. The antibody used is different from one previously used in da neurons in Parrish 2009 and looks like it is from a commercial source. With the recognition that much of the reproducibility crisis in science is due to poorly validated antibodies, it is essential to include key controls to validate this antibody- including loss of signal when Akt is knocked down. The differences in signal do not look particularly robust, for example when compared to those shown in Parrish 2009, and as the entire link between Ror and Akt rests on this data, it is imperative to be very sure that it is correct, and ideally do some additional experiments to determine whether Ror actually acts through Akt.

I would also suggest considering an alternate model, which seems much more likely based on data in the supplement and previous data on Ror function in ddaC neurons. To promote branching during dendrite regeneration Ror functions through canonical Wnt signaling proteins including dsh and Axin (Nye, 2020), which these authors also show have branching phenotypes in their assay. This Ror function is part of a Wnt signaling pathway that controls microtubule nucleation and also includes arrow, fz, and fz2 (Nye, 2020 and Weiner, 2020). The failure to exhibit excess growth by Ror knockdown neurons in LYD seems likely to be related to the function of Ror in mediating microtubule nucleation in these cells.

It looks like the major change for wg RNAi in muscle is a change in arbor area in LYD – indeed this often seems to drive or at least contribute to the phenotype (except in Akt loss). Is this because animals are overall larger? Since the arbors cover the body – is this increase in size related to overall size? If so, is the dendrite difference secondary to a much broader change in animal size? Are the same conclusions reached if the total number of terminals is shown?

I do not understand why some of the results are dismissed, but others are not. For example: why do the dsh k/d and bskDN results in more careful interpretation than, say the fz2 or fz knockdown? The reason given is that there is a slight increase in branching in HYD, but it also looks like that might be the case in Ror mutants, while the opposite is seen in Ror k/d.

There issue of lengthened developmental time on LYD is problematic and not sufficiently addressed. In a previous study on this topic (Poe), the continued growth of ddaC neurons was interpreted as resilience to malnutrition and shown to be controlled by lower levels of a stress transcription factor, foxo, than other cells. In this interpretation, the ddaC neurons just continue to grow at their normal pace while the overall development of the animal is slowed. How is this view reconciled with the current model? Which conditions change developmental timing in this study? Do any of the genetic manipulations rescue developmental speed- and if so, is dendrite architecture changing secondarily to this?

Correct controls are needed for RNAi experiments. Control RNAis should be paired with other transgenes rather than no RNAis to control for Gal4 dilution by expression of multiple transgenes. For example, in Figure 4 myr-Akt is compared to Ror k/d + myr-Akt. Ror k/d + myr-Akt should be compared to control RNAi + myr-Akt.

In Figure 4Supp1 it states in the legend that both fz and fz2 knockdown were different from control, but only fz2 is mentioned in the main text. In this figure, images are shown for an fz2 mutant, but no quantitation is shown.

Figure 5 – data seems a little preliminary. It might be better to figure out the pathway in which Ror acts rather than add another piece.

In figure 6 – it would be helpful to include Ror k/d in the electrophysiological assays. The behavioral assays look somewhat inconclusive in terms of whether Ror rescues the LYD phenotype.

https://doi.org/10.7554/eLife.79461.sa1

Author response

Essential revisions:

One of the reviewers has consolidated and summarized our consultations on the individual reviews. These are given below. Please attend to each concern while responding. Please all see the individual reviews below.

1) Are Wg levels increased in muscles when Upd2, Dome, or stat93E KD is knocked down. Since wg transcription could be regulated in a complex way, the Dome/JAK/STAT pathway is likely insufficient to be the main or only regulator. This is also echoed by a result described in the Discussion that Stat92E was not elevated by VMC supplement in LYD. The authors should limit their description on the significance of these results such as the description in Abstract (and Figure 7) and conclusion on lines 276-279 in Results. Also, they could include more discussion on the possibility of other pathways.

We agree with the reviewers that it is likely that signaling pathways other than the Dome/JAK/STAT could also contribute to downregulation of Wg on the high-yeast diet (HYD). This is because we conducted two experiments to address whether the Dome/JAK/STAT is a primary contributor to downregulating Wg on HYD, but the results were not supportive of such a possibility: First, when stat92E was knocked down in muscles, Wg increased on HYD; however, that increase was marginal (Figure 5F and 5G) compared to the change in the amount of Wg between larvae on HYD and on LYD (Figure 5F and 5H; see the quantitative data in 5J). Second, stat92E was not elevated by VMC supplementation on LYD (data not shown) as described in the Discussion in the original manuscript. Accordingly, we have followed the reviewers’ advice, carefully limiting our interpretation of the results, and we discuss the possible contribution of the hypothetical signaling pathways in the Abstract (lines 23-27), Introduction (lines 105-108), Results (lines 338-347) and Discussion (lines 441-448), as well as our model (Figure 7).

2) Could authors compare the KD efficiency in two Akt RNAi lines (v2902 and BL33615) (or are they known in previous literature)? They also need to test the specificity of anti-Akt antibodies using these akt RNAi lines. Are anti-Akt signals gone when Akt is knocked down?

The reviewers are requesting the KD efficiency in two Akt RNAi lines (v2902 and BL33615) and the specificity of the anti-phospho-Akt (Serine 473) antibody employed in our study (Cell Signaling). The KD efficiency in those RNAi lines and the specificity of the antibody have been documented in other developmental contexts in previously published literature. For example, when Akt is knocked down in the v2902 line, p-Akt signals are absent in the wing imaginal disc (see Figure 3E in Santabarbara-Ruiz et al. PLoS Genetics, 2019).

We have performed additional experiments and evaluated the KD efficiency in the lines and the specificity of the antibody in the class IV da neuron ddaC as follows:

1. p-Akt signals were significantly reduced in either the v2902 or BL33615 line (Figure 2—figure supplement 3A-3H).

2. The KD efficiency differed between the two lines, which is consistent with the phenotypic differences. The BL33615 line, which gave more severe dendrite phenotypes (Figure 2A-2J), showed weaker p-Akt signals than v2902 (Figure 2—figure supplement 3A-3H of the revised version).

3. Expression of myr-Akt, a constitutively activated membrane-anchored form of Akt (Stocker et al., Science, 2002) in class IV da neurons, increased the signal level of p-Akt (Figure 2—figure supplement 3I and 3J).

We added the above results in our revised manuscript (lines 175-176 and 268-272 in Results).

3) Was the dendrite hyper-arborization phenotype also observed in other types of da neurons? Did you see similar pAkt upregulation in other da neurons (class I-III) under the condition of LYD diets? If the authors have the results, it should be very nice to demonstrate the general and significant effects of their findings and mechanism.

The reviewer is asking whether the hyperarborization phenotype and p-Akt upregulation are also observed in other classes of da neurons. We addressed this question by performing additional quantitative data analyses in class I and III da neurons:

1. We previously reported that the hyperarborization phenotype is not seen in class I da neurons, ddaD and ddaE (Figure 2 in Watanabe et al., Genes to Cells, 2017). We quantified p-Akt levels in class I da neuron ddaE and found that the p-Akt level showed no significant difference between HYD and LYD (Figure 4—figure supplement 3A-3C in the revised manuscript).

2. On the other hand, class III da neuron ddaF showed increases both in the dendritic terminal number and in the p-Akt level on LYD compared to HYD (Figure 4—figure supplement 3D-3I), much like class IV da neurons.

Our results are consistent with a previous study (Poe et al., eLife, 2020), which examined dendrite morphologies (but not p-Akt signals), raising the possibility that the Akt-driven branching mechanism in response to the low-nutrient condition in class IV da neurons is shared by other classes of da neurons, including class III. We described and discussed these results in our revised manuscript (lines 288-300 in Results).

4) The Akt mutant phenotype is quite different from the Ror mutant phenotype based on the images shown in Figure 2. It seems to affect more parameters of dendrite structure than Ror. The authors should elaborate on the phenotypic difference between Akt and Ror mutants, perhaps using different quantification approaches. It is therefore appears unlikely that Akt, which has a stronger and different phenotype from Ror, only mediates Ror signaling. Indeed, as the authors mentioned in the Discussion, Akt has previously been shown to regulate ddaC dendrite branching through signals from epithelial cells, and it has also been previously demonstrated, that, unlike Ror, Akt affects dendrite growth on standard diets (Parrish, 2009). The only data that suggests that Ror functions upstream of Akt are the pAkt staining shown in Figure 4. The authors need to confirm the specificity of the Akt antibody. In addition, the authors should address this concern by modifying the model in the Discussion (such as to include the possibility of excess growth by Ror knockdown neurons in LYD seems likely to be related to the function of Ror in mediating microtubule nucleation, the possibility of integrating InR and AlK in the pathway, which they have shown in this study in the supplemental data, as well as in a previous study (Poe, 2020)).

The reviewers pointed out that the Akt-knocked down phenotype is quite different from the Ror-knocked down or mutant phenotype (Figure 2) and gave helpful suggestions. Before explaining our additional analysis on this matter, we addressed the specificity of the p-Akt antibody in class IV da neurons and the differential KD efficiency in two Akt RNAi lines in our response to Essential Revisions 2, and we have consolidated our model where Ror functions upstream of Akt (Figure 4).

To elaborate the phenotypic difference between Akt KD and Ror KD, we followed the reviewers’ advice and introduced a different quantification approach. In addition to dendrite “branching,” we evaluated “elongation” in our revised manuscript (Figure 2—figure supplement 2). We measured the total length of branches per neuron (dendrite length) and also divided the total length by the arbor size (dendrite length/area). Box plots of dendrite length/area (Figure 2—figure supplement 2B, 2D, and 2F) and that of terminal number/area (Figure 1E) showed obvious “hyperelongation” as well as “hyperarborization” of the control class IV da neuron on LYD compared to HYD. In addition to the box plots, we drew two-dimensional plots with the dendritic area on the X-axis and the dendrite length on the Y-axis (Figure 2—figure supplement 2A, 2C and 2E), which showed that the numerical features of dendrite length of the control neurons on HYD and those on LYD were clearly separated (red solid ellipse and blue solid ellipse, respectively).

Compared to the control neurons, strong Akt KD in the BL33615 line severely impaired both elongation (Figure 2—figure supplement 2C and 2D) and branching (Figure 2C, 2F, 2I and 2J), irrespective of the diets. On the other hand, Akt KD in the v2902 line mildly affected the hyperelongation phenotype (Figure 2—figure supplement 2A and 2B); and it did ameliorate hyperarborization on LYD (Figure 2B, 2E, 2G and 2H). These results indicate that the basal activity of Akt is required for both elongation and branching on HYD and LYD (Figure 2—figure supplement 2G-2I), which is consistent with a previous report that demonstrated the requirement of Akt for the regulation of dendritic morphology of class IV da neurons on standard laboratory food (Parrish et al., Neuron, 2009).

Our critical findings are: (1) Ror KD or a Ror mutation blunted both elongation and branching only on LYD (Figure 2—figure supplement 2E and 2F and Figure 2K-2V); and (2) the diet-dependent phenotype of the mild Akt KD by the v2902 line was partly similar to the Ror KD or mutant phenotype. Together with the evidence for the function of Ror upstream of Akt (Figure 4), our result reinforces the proposed role of the Ror/Akt signaling pathway in response to LYD. We have drawn diagrams of how the signaling pathway is working or defective and how it affects dendrite branching as well as elongation under individual genetic and dietary conditions (Figure 2—figure supplement 2G-2J). Our model underscores the notion that Akt mediates signaling from multiple upstream receptors, including Ror. We also revised the text throughout the manuscript (lines 169-180, 202-210, and 422-425) and Figure 7.

Regarding questions related to the function of Ror in mediating microtubule nucleation and the possibility of integrating InR and AlK in the pathway, please see our detailed responses to major points 2, 5, and 7 of reviewer #3.

5) It looks like the major change for wg RNAi in muscle is a change in arbor area in LYD- indeed this often seems to drive or at least contribute to the phenotype (except in Akt loss). Is this because animals are overall larger? Since the arbors cover the body- is this increase in size related to overall size? If so, is the dendrite difference secondary to a much broader change in animal size? Are the same conclusions reached if a total number of terminals is shown?

The reviewers are concerned about the possibility that the suppression of hyperarborization by wg knockdown in muscle may be a secondary effect of an increase in arbor size or larval body size (compare the blue solid ellipse with the blue dotted one in Figure 3E). We did not measure larval body size and cannot answer whether the wg knocked down larvae were bigger than control larvae on LYD. However, we believe that the suppression of hyperarborization by wg knockdown was not a secondary effect of the increased body size, for the following reasons:

1. As mentioned in the legend for Figure 1, 2D-plots of the control larvae, such as Figure 1F, show (1) a positive correlation between the area of the dendritic field and the number of branch terminals, and (2) a clear separation of 95% confidence ellipses of the numerical values between HYD and LYD (compare the blue ellipse with the red one). Therefore, instead of simply comparing the number of branch terminals between the diets, we focused on changes in the density of terminals (Figure 1E) and how far the pair of red and blue ellipses are separated in the 2D plot (Figure 1F) to evaluate the hyperarborization phenotypes. When wg was knocked down in muscles on LYD, its ellipse (the blue dotted one in Figure 3E) shifted closer to or overlapped with ellipses of the larvae on HYD (the red solid and dotted ones in Figure 3E). This result suggests that wg RNAi in muscles blunted the hyperarborization phenotype despite the increase in the dendritic area.

2. We also analyzed the effects of wg knockdown using another muscle GAL4 driver (Figure 3—figure supplement 1A-1F) and wg hypomorphic mutations (Figure 3G-3L). The distribution of the arbor size on LYD was not much affected, neither by the knockdown nor the mutation (compare blue solid circles with open circles in Figure 3—figure supplement 1E and Figure 3K); and again, the ellipse of the knockdown or the mutant larvae shifted closer to or overlapped with those of larvae on HYD (red ellipses in the 2D plots, respectively). Together with the box plots of the terminal number/area (Figure 3—figure supplements 1F and Figure 3L), our results strongly suggest that reduced function of wg attenuates the hyperarborization phenotype without causally changing the body size.

6) The issue of lengthened developmental time on LYD is problematic and not sufficiently addressed. In a previous study on this topic (Poe), the continued growth of ddaC neurons was interpreted as resilience to malnutrition and shown to be controlled by lower levels of a stress transcription factor, foxo, than in other cells. In this interpretation, the ddaC neurons just continue to grow at their normal pace while the overall development of the animal is slowed. How is this view reconciled with the current model? Which conditions change developmental timing in this study? Do any of the genetic manipulations rescue developmental speed- and if so, is dendrite architecture changing secondarily to this?

The reviewers are concerned about the effect of the lengthened developmental time on LYD on dendrite architecture. The critical question is whether the hyperaborization phenotype is a secondary consequence of the longer larval stage on LYD than that on HYD. We gathered three lines of evidence against this possibility, which consist of the data already shown in the original submission and new data from two additional experiments. We explain each in the details below and also in the revised manuscript.

1. As stated in the original manuscript, we collected wandering 3rd instar larvae and imaged class IV da neurons as scheduled throughout our dietary or genetic interventions (6-7 days after egg laying (AEL) on HYD and 9-10 days AEL on LYD or LYD plus supplements; Figure 1—figure supplement 1A). On LYD plus any combinations of nutrients, larval developmental timings were essentially the same as for LYD; nonetheless, the hyperarborization phenotype was blunted on LYD+VMC(O) (Figure 1G-1U). Moreover, throughout the testing of control genotypes and all genetic interventions, the timings on LYD were similarly longer than those on HYD; notwithstanding, Ror KD (Figure 2K-2P), a Ror mutation (Figure 2Q-2V), wg KD in muscles (Figure 3A-3F) or wg mutations (Figure 3G-3L) blunted the hyperarborization phenotype compared to that of the WT dendritic arbors. These results indicate that our dietary or genetic interventions ameliorated hyperarborization without changing the developmental timing on LYD.

2. As briefly stated in the Introduction in the original version of the manuscript, we previously compared dendrite morphologies between a low-sugar diet and a high-sugar diet (Musselman et al., Dis Model Mech, 2011), the latter of which delays larval development, and we reported that the hyperarborization is not observed between those diets (Watanabe et al., Genes to Cells, 2017). We have now expanded this approach and analyzed the effect of the sugar overload on the arborization in a quantitative manner (Figure 1—figure supplement 2A-2E). The larval stage was longer on HYD supplemented with excess sucrose (HYD + sucrose) than on HYD; however, the dendritic arbors of class IV da neurons did not become more complex (Figure 1—figure supplement 2C-2E). Thus, we showed that an increase in the amount of sucrose in HYD, which extended larval development, was not associated with hyperarborization.

3. Finally, we addressed whether any of the genetic manipulations that cause larval developmental delay are associated with an increase in dendrite complexity on a standard diet or not. Many genetic manipulations are reported to cause developmental delays; however, most of those also result in large increases in the body size (e.g., McBrayer et al., Developmental Cell, 2007), which may complicate the matters in solving our key question. We therefore chose dlip8 expression in wing imaginal discs, which is sufficient to extend the larval stage with a minimum effect on body growth (Colombani et al., Science 2012). This genetic intervention caused a mild delay in larval development, but it was not associated with hyperarborization (Figure 1—figure supplement 2F-2J), providing an additional piece of evidence allaying the concern about the effect of the lengthened developmental time on dendrite complexity.

The reviewers may wish to propose an experiment in which a genetic manipulation rescues larval developmental delay on LYD, and then examine whether the hyperarborization remains or not. This would be an interesting--but complex—experiment, which would require comparisons between genotypes, diets, and timings. We believe that our data above sufficiently dismiss the simple possibility of the hyperaborization phenotype as a secondary consequence of the longer larval stage on LYD; thus, we propose our model wherein class IV da neurons are programmed to arborize in excess due to the indispensable role of Wg/Ror/Akt signaling in response to combined VMC deficiency. As discussed in the original manuscript and in the revised version, the Poe et al. paper highlights the relationship between the neuron and the adjacent epidermis, where the key molecular mechanism is the differential expression of the transcription factor FoxO. In contrast, our study unearths the communication between the neuron and another adjacent tissue, the body wall muscle, which upregulates Wg in response to low-nutrient conditions. Both mechanisms can coexist and contribute to the continued growth of ddaC neurons while the overall development of the animal is slowed.

7) In figure 6 – it would be helpful, if feasible speedily, to include Ror k/d in the electrophysiological assays. The behavioral assays look somewhat inconclusive in terms of whether Ror rescues the LYD phenotype.

We appreciate the reviewer’s advice. Unfortunately, it has been difficult to rapidly conduct electrophysiological assays using larvae of multiple genotypes on different diets, due to limited human resources and access to apparatuses that must be shared by many other projects in the lab. As discussed in the original submission and also in the revised version, the identification of the downstream circuits would allow further studies, including electrophysiological analysis along the circuit, but this is beyond the scope of the present study.

8) On page 9 the authors mentioned screening 20 RTKs. It would be helpful to include this data. Certainly, two others they include – InR and Alk – look like they have similar phenotypes to Ror and it would be helpful to know the specificity.

Thank you for the suggestion. Supplementary file 2 lists the names of the 20 RTK genes, stock numbers of RNAi lines, GAL4 drivers, and effects of individual RNAi lines. We conducted two rounds of screening. In the primary screening, we intended to enhance the knockdown efficacy and used Gr28b.c-GAL4 and ppk-GAL4 together. We acquired images of 3-8 knocked-down neurons for each gene on each diet, and then visually judged whether hyperarborization was blunted or not. We selected nine genes (Ror, InR, Alk, htl, Egfr, Pvr, Ddr, dnt, and drl) for the secondary screening, in which we used ppk-GAL4 only, because Gr28b.c-GAL4 is expressed in a small subset of neurons in the central nervous system in addition to class IV da neurons in the peripheral nervous system (Xiang et al., Nature, 2010). Screening the 20 RTK genes is described in a new section in Materials and Methods in the revised manuscript (lines 542-550).

Representative images for each gene knockdown are shown in Figure 2—figure supplement 4 (data for the primary screening of 18 genes), Figure 2—figure supplement 5 (data for the primary and secondary screenings of InR and Alk and data for the secondary screening of htl), and Figure 2—figure supplement 6 (data for the secondary screening of Egfr, Pvr, Ddr, dnt, and drl). Regarding the KD data for InR or Alk, see our explanations and interpretations in the response to point 2 of reviewer #3.

9) There are two ways of statistics when manipulating gene activities (RNAi or OE) in HYD and LYD. Direct comparisons of two samples were used mostly in the main Figures (except Figure 5Z), while "difference in two differences" were shown mainly for several Supplemental figures (2, 3, 4, 5). It is not clear why the authors used two different ways. Are there specific reasons for doing so? Could not the direct comparisons of two samples (such as for main figures) could cope with descriptions in the text and easier to comprehend.

The hyperarborization phenotype is defined based on the difference in the terminal density between the two diets (double-headed arrows in Author response image 1). In most analyses in this study, we examined which genetic manipulation blunted the hyperarborization. As rendered in Author response image 1, the gene KD in the lefthand plot did not blunt the hyperarborization phenotype, whereas the gene KD in the righthand plot did. To make a statistical decision, we employed 2-way ANOVA, which was also used in the previous study (Poe et al., eLife, 2020), to test the interaction between the two variables: genotype and diet. When we first introduce 2-way ANOVA in the revised manuscript (Figure 2H), we briefly explain why we needed to employ that particular statistical test (lines 162-168 in Results). The use of 2-way ANOVA is now also described in the Statistical Analysis section in Materials And Methods and in the figure legends.

On the other hand, when comparing the effects of genotypes under the same dietary conditions (Figures 3Y, 4V, 4W, 5S, 5T,5AF, and Figure 5—figure supplement2), we simply focused on whether a genetic manipulation itself affected dendrite branching. For example, in the Mhc > wg experiment (Figure. 3Y), we were interested in whether muscle-derived Wg promoted dendrite branching, not whether its effect varied according to the diets. We added this description to the Statistical Analysis section in Materials And Methods.

Author response image 1

Reviewer #2 (Recommendations for the authors):

1. Akt knockdowns in two lines (v2902 and BL33615) display qualitative suppression of dendrite hyperarborization in Figure 2. In v2902 KD, while the dendritic arbor looks more normal in both HYD and LYD, the BL33615 KD showed quite a dramatic reduction even in HYD, and also in LYD, rendering both sets of data groups clustering on a distinct area (Figure 2I), instead of in between two controls (HYD and LYD) such as in Figure 2G. While these two lines have been used before, I am wondering if the BL33615 has a stronger KD effect and if the basal Akt is necessary for a fundamental process in normal dendrite development (which is still preserved in the v2902 line), irrespective of nutrient condition. Could authors compare the KD efficiency in these two lines (or are they known in previous literature)?

We thank for the reviewer’s important suggestion and addressed this point as described in Essential revision 2:

The reviewers are requesting the KD efficiency in two Akt RNAi lines (v2902 and BL33615) and the specificity of the anti-phospho-Akt (Serine 473) antibody employed in our study (Cell Signaling). The KD efficiency in those RNAi lines and the specificity of the antibody have been documented in other developmental contexts in previously published literature. For example, when Akt is knocked down in the v2902 line, p-Akt signals are absent in the wing imaginal disc (see Figure 3E in Santabarbara-Ruiz et al. PLoS Genetics, 2019).

We have performed additional experiments and evaluated the KD efficiency in the lines and the specificity of the antibody in the class IV da neuron ddaC as follows:

1. p-Akt signals were significantly reduced in either the v2902 or BL33615 line (Figure 2—figure supplement 3A-3H).

2. The KD efficiency differed between the two lines, which is consistent with the phenotypic differences. The BL33615 line, which gave more severe dendrite phenotypes (Figure 2A-2J), showed weaker p-Akt signals than v2902 (Figure 2—figure supplement 3A-3H of the revised version).

3. Expression of myr-Akt, a constitutively activated membrane-anchored form of Akt (Stocker et al., Science, 2002) in class IV da neurons, increased the signal level of p-Akt (Figure 2—figure supplement 3I and 3J).

We added the above results in our revised manuscript (lines 175-176 and 268-272 in Results).

2. The wg signal derived from muscles is required for dendrite hyperarborization (Figure 3) is a very nice set of data, although the effect by overexpression of wg in muscle is limited in HYD (Figure 3Y). It could be due to further control such as from epidermal cells shown in Poe et al. (2020) eLife. The experiment was performed at 29C to increase Wg expression, and the dendrite morphology was not normal. I am not asking for more experiments, but the authors could describe/explain what is the results at 25C instead of using 29C for readers not familiar with the approach? Have the authors tried using 2 copies of UAS-wg to increase wg expression at 25C?

The reviewer is asking about the effect of wg overexpression in muscles on dendrite arborization and our experimental design (Figures 3T-3Y). In addition to the overexpression at 29°C (Figures 3T-3Y), we did perform the same experiment at 25°C and found that the dendrite density increased on HYD (Author response image 2A and 2B). We added this description in the revised Figure3 legend. We have not used 2 copies of UAS-wg to increase its expression.

As the reviewer pointed out, the increase in the dendrite density by wg overexpression on either diet was less dramatic compared to the difference between the diets in each genotype at 29°C (Figures 3T-3Y), which is now stated as such in the revised version (lines 1027-1029). However, we do not discuss how this occurs, because it could be due to a number of possibilities. For example, the wg overexpression might not override the regulation by the epidermis, as the reviewer implied. While it would be quite interesting to pursue, it is also beyond the scope of this study to unravel the mechanisms underlying the limited dendrite growth due to wg overexpression.

Author response image 2

3. There are two ways of statistics used when manipulating gene activities (RNAi or OE) in HYD and LYD. Direct comparisons of two samples were used mostly in the main Figures (except Figure 5Z), while "difference in two differences" were shown mainly for several Supplemental figures (2, 3, 4, 5). It is not clear why the authors used two different ways. Are there specific reasons for doing so? I thought the direct comparisons of two samples (such as for main figures) could cope with descriptions in the text and be easier to comprehend.

We addressed this point as described in Essential revision 9:

The hyperarborization phenotype is defined based on the difference in the terminal density between the two diets (double-headed arrows in Author response image 1). In most analyses in this study, we examined which genetic manipulation blunted the hyperarborization. As rendered in Author response image 1, the gene KD in the lefthand plot did not blunt the hyperarborization phenotype, whereas the gene KD in the righthand plot did. To make a statistical decision, we employed 2-way ANOVA, which was also used in the previous study (Poe et al., eLife, 2020), to test the interaction between the two variables: genotype and diet. When we first introduce 2-way ANOVA in the revised manuscript (Figure 2H), we briefly explain why we needed to employ that particular statistical test (lines 162-168 in Results). The use of 2-way ANOVA is now also described in the Statistical Analysis section in Materials And Methods and in the figure legends.

On the other hand, when comparing the effects of genotypes under the same dietary conditions (Figures 3Y, 4V, 4W, 5S, 5T,5AF, and Figure 5—figure supplement2), we simply focused on whether a genetic manipulation itself affected dendrite branching. For example, in the Mhc > wg experiment (Figure 3Y), we were interested in whether muscle-derived Wg promoted dendrite branching, not whether its effect varied according to the diets. We added this description to the Statistical Analysis section in Materials And Methods.

4. The RNAi KD effects of Dome/JAK/STAT pathway components on inducing dendrite hyperarborization in HYD are quite variable! Three Dome RNAi KDs had no effect and one stat92E KD had an effect while the other had not! What are the effects on Wg expressions in muscles with these genetic manipulations? Since wg transcription could be regulated in a complex way, the Dome/JAK/STAT pathway is likely insufficient to be the main or only regulator. This is also echoed by a result described in the Discussion that Stat92E was not elevated by VMC supplement in LYD. The authors should limit their description on the significance of these results such as the description in Abstract (and Figure 7) and conclusion on lines 276-279 in Results. Also, they could include more discussion on the possibility of other pathways.

We described our response to this point in detail in Essential revision 1 and clarified the limited contribution of the JAK/STAT pathway in the revised version as the reviewer suggested.

The reviewer stated “The RNAi KD effects of Dome/JAK/STAT pathway components on inducing dendrite hyperarborization in HYD are quite variable! Three Dome RNAi KDs had no effect and one stat92E KD had an effect while the other had not.” This statement may stem from an unfortunate misunderstanding of our data. Stat92E knockdown in muscles in both of the two RNAi lines similarly resulted in enhanced dendritic branching on HYD (Figure 5L, 5O, 5Q, and 5S, and Figure 5—figure supplement 2B, 2G, 2K and 2O). These results are consistent with our hypothesis that Stat92E negatively regulates wg in muscles, and consequently, branching of class IV da neurons on HYD. This enhanced branching was also seen with the knockdown of dome in one of the three RNAi lines (Figure 5—figure supplement 2C, 2H, 2L, and 2P), knockdown of hop (Figure 5M, P, R, and T), and knockdown of upd2 in both of the two RNA lines (Figures 5AB, 5AD, 5AE, and 5AF, and Figure 5—figure supplement 2V, 2AB, 2AG and 2AL). Therefore, it seems reasonable to speculate that the Upd2- Stat92E pathway functions in suppressing the hyperarborization on HYD. However, we have no data to explain why only one out of the three dome RNAi lines promoted dendritic branching on HYD; thus, it will be necessary to verify the involvement of Dome in the future (lines 334-336 of the revised version).

5. (line 217-218) "However, these results need careful interpretation (see the Figure 4—figure supplement 2 legend)" From the legend "However, knocking down dsh or blocking JNK signaling tended to increase the branch density over the control genotype on HYD, which may consequently reduce the differences in densities of dendritic terminals between the diets." I do not quite understand the reasoning here why some increases in HYD in the KD may reduce the differences in LYD? The explanation in the Figure legend is not clear!

We addressed this concern in the revised manuscript, and offer our explanation, referring to Figure 4—figure supplement 2, as follows:

In our studies, we focused on the difference in dendrite terminal density between the two diets (HYD and LYD) and pursued dietary and genetic interventions that blunted this hyperarborization (see Author response image 1 in Essential revision 9). Reduction or loss of Ror function attenuated dendrite branching on LYD, while there were marginal effects on HYD (Figures 2K-2V in the revised manuscript). On the other hand, dsh knockdown or expression of a dominant-negative form of Bsk not only reduced dendrite branching on LYD but also increased dendritic branching on HYD (Figure 4-supplement 2B-2C, 2I-2J, 2O-2P and 2U-2V in the revised manuscript). Given this phenotypic difference on HYD between Ror and dsh or bsk, we speculate that Dsh and Bsk may regulate dendrite branching through a different mechanism than the low-nutrient dependent Wg-Ror pathway. We therefore did not focus on these factors further in this study.

Nonetheless, it is possible that these factors contribute to the hyperarborization phenotype either independently of the Wg-Ror-Akt pathway or by acting on Akt in a different manner from Wg-Ror signaling. Considering this possibility, we added a description to the Discussion and modified Figure 7.

Reviewer #3 (Recommendations for the authors):

The Akt phenotype is quite different from the Ror one based on the images shown in Figure 2. It looks like it causes a baseline phenotype in HYD, and overall seems to affect more parameters of dendrite structure than Ror. It is therefore unlikely that Akt, which has a stronger and different phenotype from Ror, is the downstream mediator of Ror signaling. Indeed, as the authors mention in the discussion, Akt has previously been shown to regulate ddaC dendrite branching through signals from epithelial cells, and it has also been previously demonstrated, that, unlike Ror, Akt affects dendrite growth on standard diets (Parrish, 2009).

We thank the reviewer’s insightful suggestion. We addressed this point as described in Essential revision 4:

The reviewers pointed out that the Akt-knocked down phenotype is quite different from the Ror-knocked down or mutant phenotype (Figure 2) and gave helpful suggestions. Before explaining our additional analysis on this matter, we addressed the specificity of the p-Akt antibody in class IV da neurons and the differential KD efficiency in two Akt RNAi lines in our response to Essential Revisions 2, and we have consolidated our model where Ror functions upstream of Akt (Figure 4).

To elaborate the phenotypic difference between Akt KD and Ror KD, we followed the reviewers’ advice and introduced a different quantification approach. In addition to dendrite “branching,” we evaluated “elongation” in our revised manuscript (Figure 2—figure supplement 2). We measured the total length of branches per neuron (dendrite length) and also divided the total length by the arbor size (dendrite length/area). Box plots of dendrite length/area (Figure 2—figure supplement 2B, 2D, and 2F) and that of terminal number/area (Figure 1E) showed obvious “hyperelongation” as well as “hyperarborization” of the control class IV da neuron on LYD compared to HYD. In addition to the box plots, we drew two-dimensional plots with the dendritic area on the X-axis and the dendrite length on the Y-axis (Figure 2—figure supplement 2A, 2C and 2E), which showed that the numerical features of dendrite length of the control neurons on HYD and those on LYD were clearly separated (red solid ellipse and blue solid ellipse, respectively).

Compared to the control neurons, strong Akt KD in the BL33615 line severely impaired both elongation (Figure 2—figure supplement 2C and 2D) and branching (Figure 2C, 2F, 2I and 2J), irrespective of the diets. On the other hand, Akt KD in the v2902 line mildly affected the hyperelongation phenotype (Figure 2—figure supplement 2A and 2B); and it did ameliorate hyperarborization on LYD (Figure 2B, 2E, 2G and 2H). These results indicate that the basal activity of Akt is required for both elongation and branching on HYD and LYD (Figure 2—figure supplement 2G-2I), which is consistent with a previous report that demonstrated the requirement of Akt for the regulation of dendritic morphology of class IV da neurons on standard laboratory food (Parrish et al., Neuron, 2009).

Our critical findings are: (1) Ror KD or a Ror mutation blunted both elongation and branching only on LYD (Figure 2—figure supplement 2E and 2F and Figure 2K-2V); and (2) the diet-dependent phenotype of the mild Akt KD by the v2902 line was partly similar to the Ror KD or mutant phenotype. Together with the evidence for the function of Ror upstream of Akt (Figure 4), our result reinforces the proposed role of the Ror/Akt signaling pathway in response to LYD. We have drawn diagrams of how the signaling pathway is working or defective and how it affects dendrite branching as well as elongation under individual genetic and dietary conditions (Figure 2—figure supplement 2G-2J). Our model underscores the notion that Akt mediates signaling from multiple upstream receptors, including Ror. We also revised the text throughout the manuscript (lines 169-180, 202-210, and 422-425) and Figure 7.

In this study in the supplemental data, as well as in a previous study (Poe, 2020) (that interprets the overgrowth of ddaC neurons on LYD in a somewhat different way) InR has been shown to be important. It is not clear why it is dismissed in this study when it is much better established as a regulator of Akt than Ror.

For the following reasons, we focused on Ror, rather than InR or Alk, in this study. We have added a new Supplementary file 2 and Figure 2—figure supplement 5.

1) As for InR, while knockdown of InR by one RNAi line, which was used in Poe et al., attenuated the hyperarborization phenotype (Figure 2—figure supplement 5G-5L), another RNAi line (not used in Poe et al.) had no significant effect on the phenotype (Figure 2—figure supplement 5A-5F). Therefore, we could not convincingly conclude that InR contributes to the low-nutrient dependent hyperarborization.

2) Alk knockdown in the primary screening (using ppk-GAL4 + Gr28b.c-GAL4) resulted in blunted hyperarborization. However, knockdown using 2 RNAi lines in the secondary screening (using ppk-GAL4) had no significant effect (Supplementary file 2 and Figure 2—figure supplement 5M-5X).

3) On the other hand, Ror knockdown consistently suppressed the hyperarborization phenotype in both the primary and the secondary screening (Figure 2—figure supplement 4C and 4D and Figures 2K-2P). Moreover, mutant analyses recapitulated the knockdown results (Figures 2Q-2V). We therefore decided to investigate the Ror-mediated mechanism.

Future studies will explore whether other RTKs including InR and Alk indeed function upstream of Akt in the context of the Wg/Ror/Akt signaling, and if so, how these various inputs are integrated by Akt. Accordingly, we modified the Result (lines 198-201), Discussion (lines 422-425) and our model (Figure 7).

The fact that Ror and Akt are interpreted as having similar effects on dendrite growth makes me very concerned about the quantitation method used. It is somewhat opaque and also at times dismissed (for example the growth phenotypes of InR and Alk are described as difficult to interpret and so sidelined, but it is unclear why).

To elaborate the phenotypic differences and similarities between the Ror KD and the Akt KD, we followed the reviewers’ advice (introduction of different quantification approaches in Essential revision 4) and evaluated dendrite “elongation” in addition to “branching”. As described in detail in our reply to Essential revision 4, this quantitative analysis revealed the phenotypic similarity between the Ror KD and the Akt mild KD.

The only data that suggests that Ror functions upstream of Akt are the phosphor-Akt staining shown in Figure 4. The antibody used is different from one previously used in da neurons in Parrish 2009 and looks like it is from a commercial source. With the recognition that much of the reproducibility crisis in science is due to poorly validated antibodies, it is essential to include key controls to validate this antibody- including loss of signal when Akt is knocked down. The differences in signal do not look particularly robust, for example when compared to those shown in Parrish 2009, and as the entire link between Ror and Akt rests on this data, it is imperative to be very sure that it is correct, and ideally do some additional experiments to determine whether Ror actually acts through Akt.

The antibody used in Parrish et al. (2009) was no longer commercially available, and we had no choice but to use a different product. We opted to use the anti-phospho-Akt antibody (Cell Signaling) because it was used in a previous study that showed clear reduction of the signal in the wing imaginal disc when Akt was knocked down by ptc-GAL4 (see Figure 3E in Santabárbara-Ruiz et al., PLOS Genetics, 2019). As described in detail in Essential revisions 2, we validated the specificity of the Cell Signaling antibody in class IV da neuron, ddaC. We believe that our immunostaining data for Ror knocked down neurons (Figures 4A-4E) and wg overexpressing larvae (Figures 4F-4J) convincingly indicate that Wg-Ror signaling regulates Akt activity in class IV da neurons.

I would also suggest considering an alternate model, which seems much more likely based on data in the supplement and previous data on Ror function in ddaC neurons. To promote branching during dendrite regeneration Ror functions through canonical Wnt signaling proteins including dsh and Axin (Nye, 2020), which these authors also show have branching phenotypes in their assay. This Ror function is part of a Wnt signaling pathway that controls microtubule nucleation and also includes arrow, fz, and fz2 (Nye, 2020 and Weiner, 2020). The failure to exhibit excess growth by Ror knockdown neurons in LYD seems likely to be related to the function of Ror in mediating microtubule nucleation in these cells.

The reviewer suggests that the Ror-mediated microtubule nucleation mechanism (Nye et al. 2020) may contribute to the hyperarborization phenotype. This is because Knockdown of dsh or Axin appeared to blunt hyperarborization (Figure 4—figure supplement 2B, 2I, 2O, 2U for dsh and 2G, 2N, 2T, 2Z for Axin). However, we found a phenotypic difference between Ror knockdown and the knockdown of dsh or Axin, and consider that Dsh and Axin might regulate dendrite branching through a mechanism different from the low-nutrient dependent Wg-Ror pathway (see our reply to comment 7 below). We added this point to the Discussion (lines 415-418 in DISCUSSION).

It looks like the major change for wg RNAi in muscle is a change in arbor area in LYD – indeed this often seems to drive or at least contribute to the phenotype (except in Akt loss). Is this because animals are overall larger? Since the arbors cover the body – is this increase in size related to overall size? If so, is the dendrite difference secondary to a much broader change in animal size? Are the same conclusions reached if the total number of terminals is shown?

We addressed this point as described in Essential revision 5:

The reviewers are concerned about the possibility that the suppression of hyperarborization by wg knockdown in muscle may be a secondary effect of an increase in arbor size or larval body size (compare the blue solid ellipse with the blue dotted one in Figure 3E). We did not measure larval body size and cannot answer whether the wg knocked down larvae were bigger than control larvae on LYD. However, we believe that the suppression of hyperarborization by wg knockdown was not a secondary effect of the increased body size, for the following reasons:

1. As mentioned in the legend for Figure 1, 2D-plots of the control larvae, such as Figure 1F, show (1) a positive correlation between the area of the dendritic field and the number of branch terminals, and (2) a clear separation of 95% confidence ellipses of the numerical values between HYD and LYD (compare the blue ellipse with the red one). Therefore, instead of simply comparing the number of branch terminals between the diets, we focused on changes in the density of terminals (Figure 1E) and how far the pair of red and blue ellipses are separated in the 2D plot (Figure 1F) to evaluate the hyperarborization phenotypes. When wg was knocked down in muscles on LYD, its ellipse (the blue dotted one in Figure 3E) shifted closer to or overlapped with ellipses of the larvae on HYD (the red solid and dotted ones in Figure 3E). This result suggests that wg RNAi in muscles blunted the hyperarborization phenotype despite the increase in the dendritic area.

2. We also analyzed the effects of wg knockdown using another muscle GAL4 driver (Figure 3—figure supplement 1A-1F) and wg hypomorphic mutations (Figure 3G-3L). The distribution of the arbor size on LYD was not much affected, neither by the knockdown nor the mutation (compare blue solid circles with open circles in Figure 3—figure supplement 1E and Figure 3K); and again, the ellipse of the knockdown or the mutant larvae shifted closer to or overlapped with those of larvae on HYD (red ellipses in the above 2D plots, respectively). Together with the box plots of the terminal number/area (Figure 3—figure supplements 1F and Figure 3L), our results strongly suggest that reduced function of wg attenuates the hyperarborization phenotype without causally changing the body size.

I do not understand why some of the results are dismissed, but others are not. For example: why do the dsh k/d and bskDN results in more careful interpretation than, say the fz2 or fz knockdown? The reason given is that there is a slight increase in branching in HYD, but it also looks like that might be the case in Ror mutants, while the opposite is seen in Ror k/d.

To address this concern, we added a brief description in the Results (lines 254-260), and we provide an explanation below in the figure legend (lines 1304-1311):

In our studies, we focused on the difference in dendrite terminal density between the two diets (HYD and LYD) and pursued dietary and genetic interventions that blunted this hyperarborization (see Author response image 1 in Essential revision 9). Reduction or loss of Ror function attenuated dendrite branching on LYD, while there were marginal effects on HYD (Figure 2K-2V in the revised manuscript). On the other hand, dsh knockdown or expression of a dominant-negative form of Bsk not only reduced dendrite branching on LYD but also increased dendritic branching on HYD (Figure 4-supplement 2B-2C, 2I-2J, 2O-2P and 2U-2V in the revised manuscript). This also appears to be the case with the Axin knockdown, although the effect was not statistically significant (Figure 4—figure supplement 2G, 2N, 2T and 2Z). Given this phenotypic difference on HYD between Ror and dsh, bsk, or Axin, we speculate that Dsh, Bsk and Axin might regulate the dendrite branching through a mechanism different from the low-nutrient dependent Wg-Ror pathway. We therefore did not focus on these factors further in this study.

There issue of lengthened developmental time on LYD is problematic and not sufficiently addressed. In a previous study on this topic (Poe), the continued growth of ddaC neurons was interpreted as resilience to malnutrition and shown to be controlled by lower levels of a stress transcription factor, foxo, than other cells. In this interpretation, the ddaC neurons just continue to grow at their normal pace while the overall development of the animal is slowed. How is this view reconciled with the current model? Which conditions change developmental timing in this study? Do any of the genetic manipulations rescue developmental speed- and if so, is dendrite architecture changing secondarily to this?

We understood that this reviewer’s concern is very important, performed two additional experiments, and described our response in Essential revision 6:

The reviewers are concerned about the effect of the lengthened developmental time on LYD on dendrite architecture. The critical question is whether the hyperaborization phenotype is a secondary consequence of the longer larval stage on LYD than that on HYD. We gathered three lines of evidence against this possibility, which consist of the data already shown in the original submission and new data from two additional experiments. We explain each in the details below and also in the revised manuscript.

1. As stated in the original manuscript, we collected wandering 3rd instar larvae and imaged class IV da neurons as scheduled throughout our dietary or genetic interventions (6-7 days after egg laying (AEL) on HYD and 9-10 days AEL on LYD or LYD plus supplements; Figure 1—figure supplement 1A). On LYD plus any combinations of nutrients, larval developmental timings were essentially the same as for LYD; nonetheless, the hyperarborization phenotype was blunted on LYD+VMC(O) (Figure 1G-1U). Moreover, throughout the testing of control genotypes and all genetic interventions, the timings on LYD were similarly longer than those on HYD; notwithstanding, Ror KD (Figure 2K-2P), a Ror mutation (Figure 2Q-2V), wg KD in muscles (Figure 3A-3F) or wg mutations (Figure 3G-3L) blunted the hyperarborization phenotype compared to that of the WT dendritic arbors. These results indicate that our dietary or genetic interventions ameliorated hyperarborization without changing the developmental timing on LYD.

2. As briefly stated in the Introduction in the original version of the manuscript, we previously compared dendrite morphologies between a low-sugar diet and a high-sugar diet (Musselman et al., Dis Model Mech, 2011), the latter of which delays larval development, and we reported that the hyperarborization is not observed between those diets (Watanabe et al., Genes to Cells, 2017). We have now expanded this approach and analyzed the effect of the sugar overload on the arborization in a quantitative manner (Figure 1—figure supplement 2A-2E). The larval stage was longer on HYD supplemented with excess sucrose (HYD + sucrose) than on HYD; however, the dendrites of class IV da neurons did not become more complex (Figure 1—figure supplement 2C-2E). Thus, we showed that an increase in the amount of sucrose in HYD, which extended larval development, was not associated with hyperarborization.

3. Finally, we addressed whether any of the genetic manipulations that cause larval developmental delay are associated with an increase in dendrite complexity on a standard diet or not. Many genetic manipulations are reported to cause developmental delays; however, most of those also result in large increases in the body size (e.g., McBrayer et al., Developmental Cell, 2007), which may complicate the matters in solving our key question. We therefore chose dlip8 expression in wing imaginal discs, which is sufficient to extend the larval stage with a minimum effect on body growth (Colombani et al., Science 2012). This genetic intervention caused a mild delay in larval development, but it was not associated with hyperarborization (Figure 1—figure supplement 2F-2J), providing an additional piece of evidence allaying the concern about the effect of the lengthened developmental time on dendrite complexity.

The reviewers may wish to propose an experiment in which a genetic manipulation rescues larval developmental delay on LYD, and then examine whether the hyperarborization remains or not. This would be an interesting--but complex—experiment, which would require comparisons between genotypes, diets, and timings. We believe that our data above sufficiently dismiss the simple possibility of the hyperaborization phenotype as a secondary consequence of the longer larval stage on LYD; thus, we propose our model wherein class IV da neurons are programmed to arborize in excess due to the indispensable role of Wg/Ror/Akt signaling in response to combined VMC deficiency. As discussed in the original manuscript and in the revised version, the Poe et al. paper highlights the relationship between the neuron and the adjacent epidermis, where the key molecular mechanism is the differential expression of the transcription factor FoxO. In contrast, our study unearths the communication between the neuron and another adjacent tissue, the body wall muscle, which upregulates Wg in response to low-nutrient conditions. Both mechanisms can coexist and contribute to the continued growth of ddaC neurons while the overall development of the animal is slowed.

Correct controls are needed for RNAi experiments. Control RNAis should be paired with other transgenes rather than no RNAis to control for Gal4 dilution by expression of multiple transgenes. For example, in Figure 4 myr-Akt is compared to Ror k/d + myr-Akt. Ror k/d + myr-Akt should be compared to control RNAi + myr-Akt.

We have followed the reviewer’s advice, examined the effect of myr-Akt (the constitutively activated membrane-anchored form of Akt) using correct controls, and totally replaced the previous data (Figure 4K-4W of the original version) with the new data (Figure 4K-4W of the revised version). Fortunately, the results in the original version were recapitulated: Expression of myr-Akt in class IV da neurons increased the terminal density, regardless of whether Ror was knocked down or not. This result strengthened the idea that Akt activation in the neurons plays a pivotal role for the hyperarborization. Our choices of the controls are described in the legend, and the exact genotypes are listed in Supplementary file 4.

In Figure 4Supp1 it states in the legend that both fz and fz2 knockdown were different from control, but only fz2 is mentioned in the main text. In this figure, images are shown for an fz2 mutant, but no quantitation is shown.

We added our interpretation of the fz KD experiment to the legend of Figure 4—figure supplement 1 as follows:

Both 2D-plots and boxplots of dendrite density show that the degree of attenuation of dendrite branching on LYD is much higher in fz2 KD neurons than fz KD neurons (Figure 4—figure supplement 1A-1J). Therefore, we mentioned only the result of fz2 knockdown in the main text. However, we do not rule out the possibility that Fz may be involved in the hyperarborization phenotype.

We added the quantification data for the fz2 mutant in Figure 4—figure supplement 1O and 1P and showed that hyperarborization was strongly suppressed in the fz2 mutant neurons.

Figure 5 – data seems a little preliminary. It might be better to figure out the pathway in which Ror acts rather than add another piece.

We believe that the addition of the Dome/JAK/STAT pathway is informative because of its connection with wg expression in muscles. As described in Essential revision 1, we carefully clarified the limited contribution of the Dome/JAK/STAT pathway in the revised version.

In figure 6 – it would be helpful to include Ror k/d in the electrophysiological assays. The behavioral assays look somewhat inconclusive in terms of whether Ror rescues the LYD phenotype.

We replied to this suggestion in Essential revision 7:

We appreciate the reviewer’s advice. Unfortunately, it has been difficult to rapidly conduct electrophysiological assays using larvae of multiple genotypes on different diets, due to limited human resources and access to apparatuses that must be shared by many other projects in the lab. As discussed in the original submission and also in the revised version, the identification of the downstream circuits would allow further studies, including electrophysiological analysis along the circuit, but this is beyond the scope of the present study.

Reference

Colombani J, Andersen DS, Léopold P. 2012. Secreted Peptide Dilp8 Coordinates Drosophila Tissue Growth with Developmental Timing. Science (1979) 336:582–585. doi:10.1126/science.1216689

McBrayer Z, Ono H, Shimell MJ, Parvy JP, Beckstead RB, Warren JT, Thummel CS, Dauphin-Villemant C, Gilbert LI, O’Connor MB. 2007. Prothoracicotropic Hormone Regulates Developmental Timing and Body Size in Drosophila. Dev Cell 13:857–871. doi:10.1016/j.devcel.2007.11.003

Nye DMR, Albertson RM, Weiner AT, Ian Hertzler J, Shorey M, Goberdhan DCI, Wilson C, Janes KA, Rolls MM. 2020. The receptor tyrosine kinase Ror is required for dendrite regeneration in Drosophila neurons. PLoS Biol 18. doi:10.1371/journal.pbio.3000657

Palanker Musselman L, Fink JL, Narzinski K, Ramachandran PV, Sukumar Hathiramani S, Cagan RL, Baranski TJ. 2011. A high-sugar diet produces obesity and insulin resistance in wild-type Drosophila. Dis Model Mech 4:842–849. doi:10.1242/dmm.007948

Parrish JZ, Xu P, Kim CC, Jan LY, Jan YN. 2009. The microRNA bantam Functions in Epithelial Cells to Regulate Scaling Growth of Dendrite Arbors in Drosophila Sensory Neurons. Neuron 63:788–802. doi:10.1016/j.neuron.2009.08.006

Poe AR, Xu Y, Zhang C, Lei J, Li K, Labib D, Han C. 2020. Low FoxO expression in Drosophila somatosensory neurons protects dendrite growth under nutrient restriction. eLife 9:1–47. doi:10.7554/eLife.53351

Santabárbara-Ruiz P, Esteban-Collado J, Pérez L, Viola G, Abril JF, Milán M, Corominas M, Serras F. 2019. Ask1 and Akt act synergistically to promote ROS-dependent regeneration in Drosophila. PLoS Genet 15:e1007926. doi:10.1371/journal.pgen.1007926

Stocker H, Andjelkovic M, Oldham S, Laffargue M, Wymann MP, Hemmings BA, Hafen E. 2002. Living with Lethal PIP3 Levels: Viability of Flies Lacking PTEN Restored by a PH Domain Mutation in Akt/PKB. Science (1979) 295:2088–2091. doi:10.1126/science.1068094

Watanabe K, Furumizo Y, Usui T, Hattori Y, Uemura T. 2017. Nutrient-dependent increased dendritic arborization of somatosensory neurons. Genes to Cells 22:105–114. doi:10.1111/gtc.12451

Xiang Y, Yuan Q, Vogt N, Looger LL, Jan LY, Jan YN. 2010. Light-avoidance-mediating photoreceptors tile the Drosophila larval body wall. Nature 468:921–926. doi:10.1038/nature09576

https://doi.org/10.7554/eLife.79461.sa2

Article and author information

Author details

  1. Yasutetsu Kanaoka

    Graduate School of Biostudies, Kyoto University, Kyoto, Japan
    Contribution
    Resources, Data curation, Formal analysis, Funding acquisition, Investigation, Visualization, Methodology, Writing – original draft
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1835-3248
  2. Koun Onodera

    Graduate School of Biostudies, Kyoto University, Kyoto, Japan
    Contribution
    Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4203-9865
  3. Kaori Watanabe

    Graduate School of Biostudies, Kyoto University, Kyoto, Japan
    Contribution
    Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2887-3690
  4. Yusaku Hayashi

    Graduate School of Biostudies, Kyoto University, Kyoto, Japan
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  5. Tadao Usui

    Graduate School of Biostudies, Kyoto University, Kyoto, Japan
    Contribution
    Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0507-1495
  6. Tadashi Uemura

    1. Graduate School of Biostudies, Kyoto University, Kyoto, Japan
    2. Research Center for Dynamic Living Systems, Kyoto University, Kyoto, Japan
    3. AMED-CREST, Tokyo, Japan
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Writing – original draft, Project administration, Writing – review and editing
    For correspondence
    tauemura@lif.kyoto-u.ac.jp
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7204-3606
  7. Yukako Hattori

    1. Graduate School of Biostudies, Kyoto University, Kyoto, Japan
    2. JST FOREST, Tokyo, Japan
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing
    For correspondence
    yhattori@lif.kyoto-u.ac.jp
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5977-8501

Funding

Japan Agency for Medical Research and Development (JP18gm1110001)

  • Tadashi Uemura

Japan Society for the Promotion of Science (15H02400)

  • Tadashi Uemura

Japan Society for the Promotion of Science (21H00251)

  • Yukako Hattori

Japan Society for the Promotion of Science (21K06186)

  • Yukako Hattori

Japan Society for the Promotion of Science (20J15084)

  • Yasutetsu Kanaoka

Japan Science and Technology Agency (JPMJFR2051)

  • Yukako Hattori

Naito Foundation

  • Yukako Hattori

Japan Foundation for Applied Enzymology

  • Yukako Hattori

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The reagents and genomic datasets were provided by the Drosophila Genetic Resource Center at Kyoto Institute of Technology, National Institute of Genetics, the Bloomington Stock Center, Vienna Drosophila Resource Center, FlyBase, and the Developmental Studies Hybridoma Bank maintained by the University of Iowa. We thank T Kondo and Y Sando for performing RNA-sequencing; J A Hejna for polishing the manuscript; T Kambe, R Niwa, T Jovanic, N Yamanaka, S Goulas, Y Shimada-Niwa, N Okamoto and other members of the Uemura laboratory for discussions and their technical assistance; M M Rolls, A Wodarz, T Igaki, T Ito, M Nakamura, M Yamazaki, M Sato, P Leopold, Y Sanaki and T Nishimura for kindly providing reagents; and Y Xiang for sharing unpublished results.

Ethics

Our recombinant DNA experiments follow Kyoto University Regulations for Safety Management in Recombinant DNA Experiments under protocol # 210059.

Senior and Reviewing Editor

  1. K VijayRaghavan, National Centre for Biological Sciences, Tata Institute of Fundamental Research, India

Reviewers

  1. Fengwei Yu, Temasek Life Sciences Laboratory, Singapore
  2. Cheng-Ting Chien, Institute of Molecular Biology, Academia Sinica, Taiwan

Publication history

  1. Received: April 13, 2022
  2. Preprint posted: May 1, 2022 (view preprint)
  3. Accepted: December 11, 2022
  4. Version of Record published: January 17, 2023 (version 1)

Copyright

© 2023, Kanaoka et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 413
    Page views
  • 46
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Yasutetsu Kanaoka
  2. Koun Onodera
  3. Kaori Watanabe
  4. Yusaku Hayashi
  5. Tadao Usui
  6. Tadashi Uemura
  7. Yukako Hattori
(2023)
Inter-organ Wingless/Ror/Akt signaling regulates nutrient-dependent hyperarborization of somatosensory neurons
eLife 12:e79461.
https://doi.org/10.7554/eLife.79461

Further reading

    1. Developmental Biology
    Marianne E Emmert, Parul Aggarwal ... Roger Cornwall
    Research Article Updated

    Neonatal brachial plexus injury (NBPI) causes disabling and incurable muscle contractures that result from impaired longitudinal growth of denervated muscles. This deficit in muscle growth is driven by increased proteasome-mediated protein degradation, suggesting a dysregulation of muscle proteostasis. The myostatin (MSTN) pathway, a prominent muscle-specific regulator of proteostasis, is a putative signaling mechanism by which neonatal denervation could impair longitudinal muscle growth, and thus a potential target to prevent NBPI-induced contractures. Through a mouse model of NBPI, our present study revealed that pharmacologic inhibition of MSTN signaling induces hypertrophy, restores longitudinal growth, and prevents contractures in denervated muscles of female but not male mice, despite inducing hypertrophy of normally innervated muscles in both sexes. Additionally, the MSTN-dependent impairment of longitudinal muscle growth after NBPI in female mice is associated with perturbation of 20S proteasome activity, but not through alterations in canonical MSTN signaling pathways. These findings reveal a sex dimorphism in the regulation of neonatal longitudinal muscle growth and contractures, thereby providing insights into contracture pathophysiology, identifying a potential muscle-specific therapeutic target for contracture prevention, and underscoring the importance of sex as a biological variable in the pathophysiology of neuromuscular disorders.

    1. Developmental Biology
    2. Genetics and Genomics
    Ankit Sabharwal, Mark D Wishman ... Stephen C Ekker
    Research Advance Updated

    The clinical and largely unpredictable heterogeneity of phenotypes in patients with mitochondrial disorders demonstrates the ongoing challenges in the understanding of this semi-autonomous organelle in biology and disease. Previously, we used the gene-breaking transposon to create 1200 transgenic zebrafish strains tagging protein-coding genes (Ichino et al., 2020), including the lrpprc locus. Here, we present and characterize a new genetic revertible animal model that recapitulates components of Leigh Syndrome French Canadian Type (LSFC), a mitochondrial disorder that includes diagnostic liver dysfunction. LSFC is caused by allelic variations in the LRPPRC gene, involved in mitochondrial mRNA polyadenylation and translation. lrpprc zebrafish homozygous mutants displayed biochemical and mitochondrial phenotypes similar to clinical manifestations observed in patients, including dysfunction in lipid homeostasis. We were able to rescue these phenotypes in the disease model using a liver-specific genetic model therapy, functionally demonstrating a previously under-recognized critical role for the liver in the pathophysiology of this disease.