Advertisement

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Metabolic and behavioral effects of neurofibromin result from differential recruitment of MAPK and mTOR signaling

  • Valentina Botero,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing

    Affiliation Neuroscience and Pharmacology, University of Iowa, Iowa City, Iowa, United States of America

    ⨯
  • Jenifer Barrios,

    Roles Formal analysis, Investigation, Writing – review & editing

    Affiliation Neuroscience and Pharmacology, University of Iowa, Iowa City, Iowa, United States of America

    ⨯
  • Anneke Knauss,

    Roles Formal analysis, Investigation, Visualization, Writing – review & editing

    Affiliation Neuroscience and Pharmacology, University of Iowa, Iowa City, Iowa, United States of America

    ⨯
  • Greta Dahlen,

    Roles Formal analysis, Investigation, Visualization, Writing – review & editing

    Affiliation Neuroscience and Pharmacology, University of Iowa, Iowa City, Iowa, United States of America

    ⨯
  • Ethan Rosendahl,

    Roles Investigation, Visualization, Writing – review & editing

    Affiliation Neuroscience and Pharmacology, University of Iowa, Iowa City, Iowa, United States of America

    ⨯
  • Kenneth J. Colodner,

    Roles Investigation, Writing – review & editing

    Affiliation Program in Neuroscience and Behavior, Mount Holyoke College, South Hadley, Massachusetts, United States of America

    ⨯
  • Seth M. Tomchik

    Roles Conceptualization, Funding acquisition, Methodology, Resources, Supervision, Writing – original draft, Writing – review & editing

    seth-tomchik@uiowa.edu

    Affiliations Neuroscience and Pharmacology, University of Iowa, Iowa City, Iowa, United States of America, Stead Family Department of Pediatrics, University of Iowa, Iowa City, Iowa, United States of America, Iowa Neuroscience Institute, University of Iowa, Iowa City, Iowa, United States of America, FOE Diabetes Research Center, University of Iowa, Iowa City, Iowa, United States of America

    ⨯

Abstract

Neurofibromatosis type 1 results from mutations in the NF1 gene and its encoded neurofibromin protein. This condition produces multiple symptoms, including tumors, behavioral alterations, and metabolic changes. Molecularly, neurofibromin mutations affect Ras activity, influencing multiple downstream signaling pathways, including MAPK (Raf/MEK/ERK) and PI3K/Akt/mTOR signaling. This pleiotropy raises the question of which pathways could be targeted to treat the disease symptoms, and whether different phenotypes driven by neurofibromin mutations exhibit similar or diverging dependence on the signaling pathways downstream of Ras. To test this, we examined metabolic and behavioral alterations in the genetically tractable Drosophila neurofibromatosis type 1 model. In vivo genetic analysis revealed that behavioral effects of neurofibromin were mediated by MEK signaling, with no necessity for Akt. In contrast, metabolic effects of neurofibromin were mediated by coordinated actions MEK/ERK and Akt/mTOR/S6K/4E-BP signaling. At the systemic level, loss of neurofibromin dysregulated metabolism via molecular effects in interneurons and muscle. These changes were accompanied by altered muscle mitochondria morphology, with no concomitant changes in neuronal ultrastructure or neuronal mitochondria. Overall, this suggests that neurofibromin mutations affect multiple signaling cascades downstream of Ras, which differentially affect metabolic and behavioral neurofibromatosis type 1 phenotypes.

Author summary

The genetic disorder neurofibromatosis type 1 drives multiple symptoms in humans, including increasing risk for behavioral disorders and altering metabolism. The genetic mutations that underlie the disorder change cellular signaling pathways in multiple ways. In this study, we asked whether the multiple symptoms in the disorder could be caused by the different changes in cellular signaling pathways by testing them one at a time in a powerful genetic model, the fruit fly, Drosophila melanogaster. Mimicking the disorder by reducing levels of the neurofibromin protein increases the activity of the Ras signaling pathway and activates several downstream signaling pathways, including MEK/ERK and Akt/mTOR. Flies with reduced Nf1 exhibit several characteristics that are reminiscent of the human symptoms, including behavioral changes and metabolic alterations. With a series of genetic experiments, we examined how the two cellular signaling pathways drive behavioral changes and metabolic alterations. The results showed that multiple signaling molecules were required for Nf1-dependent effects on metabolism, while the behavioral effects were more selectively modulated by one of the two pathways (MEK/ERK). These results suggest that different symptoms in neurofibromatosis type 1 may result from different cellular signaling mechanisms. If so, they may be independently targetable with different interventions. The study sets the stage to examine other signaling pathways and to model other symptoms in the disease in animal models.

Citation: Botero V, Barrios J, Knauss A, Dahlen G, Rosendahl E, Colodner KJ, et al. (2026) Metabolic and behavioral effects of neurofibromin result from differential recruitment of MAPK and mTOR signaling. PLoS Genet 22(3): e1012061. https://doi.org/10.1371/journal.pgen.1012061

Editor: J. Elliott Robinson, Cincinnati Children's Hospital Medical Center Burnet Campus: Cincinnati Children's Hospital Medical Center, UNITED STATES OF AMERICA

Received: July 22, 2025; Accepted: February 16, 2026; Published: March 5, 2026

Copyright: © 2026 Botero et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: Raw data were uploaded to the Iowa Research Online repository and are available under DOI: https://doi.org/10.25820/data.007850.

Funding: This research was supported by the National Institutes of Health R01 NS114403 to SMT, R01 NS124716 to SMT, R01 NS126361 to SMT, R01 NS097237 to SMT, R21 NS124198 to SMT, F31 NS124245 to VB, Department of Defense NF230039 to SMT, and an award from the Roy J. Carver Charitable Trust to SMT. VB, JB, AK, GD, ER, and SMT received salary support from the National Institutes of Health and SMT received salary support from the Department of Defense. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Neurofibromatosis type 1 (NF1) is a multisystemic genetic disorder characterized by a spectrum of clinical manifestations, including tumors such as cutaneous and plexiform neurofibromas, optic gliomas, and malignant peripheral nerve sheath tumors. Cutaneous symptoms include neurofibromas and a range of cutaneous/skeletal/ocular symptoms. Additional manifestations include neuronal and growth/metabolism-related symptoms such as short stature, metabolic changes, sleep disruption, cognitive/behavioral alterations, and fatigue [1–12]. The condition is caused by loss of function mutations in the NF1 gene, which encodes a large protein called neurofibromin (Nf1). Nf1 is a major Ras GAP, which directly binds Ras, accelerates the hydrolysis of Ras-GTP, and thereby functions as a negative regulator of Ras signaling [13,14]. Loss of Nf1 upregulates the activity of multiple downstream pathways, including mitogen-activated protein kinase (MAPK) signaling (Raf/MEK/ERK) and PI3K/Akt/mTOR. Further, cAMP/PKA levels are reduced [12,15–20].

The MAPK signaling pathway, MEK in particular, is a target for treatment of plexiform neurofibromas in NF1 [21,22], though with potentially significant adverse reactions [23–25]. This raises several questions. First, could other NF1 symptoms be targeted with a similar approach? NF1 patients exhibit a range of nontumor symptoms, including increased frequency of learning deficits, attention-deficit/hyperactivity disorder, and autism spectrum disorder [26]. These cognitive and behavioral symptoms are presumably driven by neuronal dysfunction caused by the mutations in the NF1 gene. Animal models exhibit deficiencies in behavioral tasks and neuronal function indicative of cognitive dysfunction [2,12,27–29]. These result from alterations in Ras/ERK and/or dopamine/cAMP signaling. The potential efficacy of targeting MEK (or other MAPK signaling molecules) for treating cognitive and behavioral symptoms in NF1 is unknown. A second major question is whether other signaling pathways could be targeted to treat NF1 symptoms – either alone or in combination with MEK – possibly with fewer side effects.

Metabolic alterations represent an understudied aspect of NF1 which may contribute to the development of other symptoms [11,12,30–32]. Patients experience a range of metabolic changes, including reduced cerebral glucose metabolism in the thalamus [33], lower blood glucose [34], reductions in muscle mass [35], decreased muscular strength and force [32,36], alterations in muscle growth and metabolism, and increased resting energy expenditure [32]. Similarly, mouse models of NF1 exhibit metabolic phenotypes, including altered muscle growth and muscle lipid mobilization [30,37–39]. In the fruit fly, Drosophila melanogaster, loss of Nf1 alters metabolism [12], increasing energy expenditure [11], reducing triglyceride [11] and glycogen [40] levels, increasing fat turnover [11], increasing ROS production and decreasing lifespan [41], shifting locomotor activity [42–45], and altering metabolism-sleep interactions [40,46]. Some of these effects are mediated by Nf1 in neurons [11,40,42–46], suggesting that central control of metabolism is affected. Though patient and animal model studies show metabolic dysfunction caused by the loss of Nf1 function, the underlying mechanisms are unclear. Furthermore, it is unknown whether the molecular mechanisms underlying metabolic alterations are similar to those driving cognitive and behavioral alterations.

Since there are multiple phenotypes in NF1, and the neurofibromin protein is a Ras GAP that exerts pleiotropic effects on multiple signaling pathways downstream of Ras, a major question is whether different downstream signaling pathways have different involvement in NF1 phenotypes. For instance, are the NF1 cognitive/behavioral symptoms more dependent on one pathway (e.g., MEK) while the metabolic alterations more dependent on another (e.g., mTOR)? We tested this using the genetically tractable Drosophila NF1 model, which allows precise dissection of signaling pathway function in vivo. Using this model, we compared the signaling requirements for two NF1 phenotypes in Drosophila: behavioral alterations and metabolic dysregulation.

Results

Metabolic effects of neurofibromin are mediated by coordinated actions of MEK/ERK and Akt/mTOR/S6K/4E-BP

To mechanistically dissect how mutations in neurofibromin affect organismal physiology, we first examined its effect on metabolic regulation in Drosophila. Loss of Nf1 disrupts metabolic homeostasis, increasing CO2 production (and O2 consumption) [11,12,47], which correlates with organismal metabolic rate [48]. We utilized this phenotype to probe the cellular signaling mechanisms of Nf1 effects on metabolism (Fig 1A). Prior metabolic studies focused on male flies, so we initially compared CO2 production between nf1P1 mutants and wCS10 controls in both sexes. In both males and females, there was a significant increase in CO2 production in nf1P1 mutants relative to controls (Fig 1B). Since the effect appeared in both sexes, all subsequent experiments were conducted with male flies unless otherwise noted. The Nf1 metabolic effect maps to a set of neurons labeled with the PCB-Gal4 driver [11]. When knocking Nf1 down in these neurons, CO2 production was increased (Fig 1C). The Nf1-dependent metabolic alteration is Ras dependent [11]. To examine the dependence of the metabolic phenotype on signaling cascades downstream of Ras, we implemented an in vivo genetic analysis approach. Nf1 was knocked down pan-neuronally with RNAi [11,43,44], either alone or in combination with knockdown of a second signaling molecule downstream of Ras (Fig 1D). Each RNAi was expressed transgenically under Gal4/UAS control (using the pan-neuronal nSyb-Gal4 in these experiments), with knockdown efficiency enhanced by coexpression of UAS-dicer-2 [49].

thumbnail
Download:
Fig 1. Loss of Nf1 affects metabolism via a neuronal MEK/ERK-dependent mechanism.

A. Diagram of the respirometry setup used to measure metabolic rate via CO2 production. B. Normalized CO2 production in genomic nf1P1 mutant and wCS10 control males and females. **p < 0.01, ***p < 0.001 (Šidák, n = 6 males; n = 5 females). C. Normalized CO2 production in flies with Nf1 knockdown in metabolism-regulating PCB-Gal4 + neurons compared to heterozygous Gal4/+ and UAS/ + controls. ***p < 0.001 re: both controls (Šidák, n = 5). D. Diagram showing the MAPK arm of the Nf1/Ras signaling pathway. Green arrows represent the direction of signaling change following loss of Nf1. Red X marks show the molecules targeted for in vivo genetic analyses in panels G-H. E. Nf1 knockdown efficiency in flies harboring RNAi targeting Nf1, MEK, or both, measured with quantitative PCR. n.s.: not significant (Šidák, n = 5). F. Representative western blot of total ERK and phosphorylated ERK in controls (nSyb-Gal4/+), and flies with pan-neuronal MEK knockdown, Nf1 knockdown, or Nf1 + MEK knockdown. Samples derived from the same experiment and processed in parallel, with β-tubulin as loading control. Representative blot shown from one of two experiments. G. CO2 production (arbitrary units; AU) in flies harboring RNAi targeting Nf1, MEK, or both, along with controls. ***p < 0.001; n.s.: not significant (Šidák, n = 8). H. CO2 production in flies harboring RNAi targeting Nf1, ERK, or both, along with controls. **p < 0.01, ***p < 0.001; n.s.: not significant (Šidák, n = 10-12).

https://doi.org/10.1371/journal.pgen.1012061.g001

To validate the double RNAi approach, we tested whether introducing a second RNAi altered Nf1 mRNA relative to a single Nf1 RNAi via quantitative PCR (qPCR). Nf1 mRNA levels were significantly reduced in both the single Nf1 knockdown and double Nf1 + MEK knockdowns, and there was no significant difference in Nf1 mRNA expression between the single and double knockdowns (Fig 1E). This demonstrated approach was effective and that there was no dilution of the Nf1 knockdown when expressing a second RNAi line using the same Gal4 driver. As further validation, we examined how knocking down Nf1 and/or MEK affected downstream ERK protein phosphorylation. Western blot analysis showed increased phosphorylated ERK (pERK) following Nf1 knock down in neurons without affecting total ERK (Fig 1F). This demonstrated that the double RNAi lines effectively modulated the targeted signaling pathways (MEK/ERK), providing additional validation of the genetic approach.

We tested which signaling cascade(s) downstream of Ras were necessary for Nf1-dependent metabolic modulation, first targeting the MAPK signaling cascade by knocking down Nf1 along with MEK (Fig 1D). MEK is a known modifier of neurofibromatosis type 1 phenotypes and therapeutic target for treatment of plexiform neurofibromas [21,22]. Knocking down MEK alone did not affect metabolic rate (CO2 production), but combining it with Nf1 knockdown occluded the effect of Nf1 knockdown on metabolic rate (Fig 1G). Moving one step downstream in the MAPK pathway, we tested the effect of knocking down Nf1 and MEK using the same double RNAi approach. Knocking down ERK alone did not affect metabolic rate, yet it also occluded the Nf1 effect on metabolic rate when Nf1 and ERK were knocked down simultaneously (Fig 1H). This suggested that MEK/ERK signaling is required for Nf1-dependent metabolic modulation. These data complement prior results demonstrating that upregulation of ERK activity via expression of a constitutively-active ERK mutant (rlSEM) in otherwise wild-type animals increases metabolic rate [11].

The mechanistic target of rapamycin (mTOR) is a well-studied nutrient sensor and metabolic regulator [50]. It is downstream of Nf1 and activated by Ras-PI3K-Akt (Fig 2A). Alterations in Akt and mTOR modulate growth and cancer phenotypes in NF1 [51–54]. To test the role of mTOR, and its activators/effectors in Nf1-dependent metabolic regulation, we carried out double knockdown experiments as above. There are multiple PI3K isoforms (in Drosophila, PI3K92E, PI3K59F, PI3K68D, and PI3K21B) and some PI3K isoforms can functionally compensate for one another [55]. To circumvent this potential confound, we targeted the loss-of-function manipulation one step downstream, to Akt, which exists as a single isoform in Drosophila [56]. Knocking down Akt on its own did not affect metabolic rate (Fig 2B). However, when Akt and Nf1 knockdown were combined, the Akt knockdown occluded the metabolic effect of Nf1 knockdown (Fig 2B). Thus, in addition to MEK/ERK, the Nf1 metabolic phenotype was additionally dependent on Akt. Given this positive result with Akt, we moved further down the mTOR signaling pathway, testing Raptor, an mTOR complex 1 obligate subunit. Knocking down Raptor did not affect metabolism on its own, but, like Akt, occluded the Nf1 effect on metabolism (Fig 2C). mTOR is a protein kinase that plays key roles in cell growth and metabolism, in part via regulating protein synthesis [57]. Activation of mTOR is positively coupled to protein synthesis via activation of S6K and inhibition of the translational repressor 4E-BP [50,58]. Therefore, we tested whether targeting these downstream effectors could modulate the Nf1 metabolic effect. Knocking down both S6K (Fig 2D) and 4E-BP (Fig 2E) occluded the Nf1 metabolic effect. This demonstrated that targeting two mTOR complex 1 effectors ameliorated the Nf1 metabolic phenotype.

thumbnail
Download:
Fig 2. Nf1 metabolic effects require mTOR in addition to MAPK signaling mechanisms.

A. Diagram highlighting the mTOR arm of the Nf1/Ras signaling pathway. Green arrows represent the direction of signaling change following loss of Nf1. Red X marks show the molecules targeted for in vivo genetic analyses in panels B-E. B. CO2 production (arbitrary units; AU) in flies bearing an RNAi targeting Nf1, Akt, or both, along with controls. *p < 0.05, **p < 0.01; n.s.: not significant (Dunn’s test, two-sided; n = 8-9). C. CO2 production in flies bearing an RNAi targeting Nf1, Raptor, or both, along with controls. **p < 0.01, ***p < 0.001; n.s.: not significant (Šidák, n = 9-10). D. CO2 production in flies bearing an RNAi targeting Nf1, S6K, or both, along with controls. ***p < 0.001; n.s.: not significant (Šidák, n 8-10). E. CO2 production in flies bearing an RNAi targeting Nf1, 4E-BP, or both, along with controls. *p < 0.05, **p < 0.01; n.s.: not significant (Šidák, n = 10).

https://doi.org/10.1371/journal.pgen.1012061.g002

These data suggest several major conclusions. First, Nf1 affects metabolism through coordinated actions of MAPK and mTOR signaling. Second, the Nf1 metabolic effect is susceptible to modulation of either pathway. Third, the metabolic effect can be rescued by downregulating any one of multiple nodes within the mTOR signaling pathway, including multiple molecules along the vertical axis of the pathway (Akt, Raptor, downstream effectors), as well as at least two independent parallel nodes within the pathway (S6K and 4E-BP). Thus, the metabolic phenotype is highly sensitive to any one of multiple perturbations that push signaling back toward normal levels.

Neurofibromin mediates metabolic effects via Ras signaling independent of cAMP

Loss of Nf1 decreases cAMP and PKA levels in flies (and other species, including mice, fish, and humans) (Fig 3A) [15,17,20,59–62]. Therefore, we tested whether reduced cAMP levels contribute to the metabolic phenotype. As a first pass, we knocked down the type-1 adenylyl cyclase Rutabaga [63] with RNAi expressed in PCB+ neurons. This did not significantly alter metabolic rate relative to the heterozygous Gal4/+ or UAS/ + controls (Fig 3B). There was a trend toward an increase in CO2 production in the experimental knockdown group relative to the UAS-RutRNAi/ + control. However, it was not statistically significant, there was little difference between the experimental group (median = 53.4) and the Gal4/ + control (median = 52.9), and it could not account for the large increase in the Nf1 knockdown (median = 70.2). This suggested that dysregulation of cAMP signaling by PCB-Gal4 + neurons did not mimic the Nf1 metabolic phenotype. In a complementary approach, we moved one step downstream in the signaling pathway, knocking down the PKA catalytic subunit 1 (PKA-C1). Similarly, knocking down PKA-C1 did not mimic the Nf1 metabolic effect (Fig 3C). In this case, there was a significantly higher CO2 production in the knockdown groups relative to the UAS-PKA-C1RNAi/ + control (p < 0.001), but again there was no significant difference relative to the Gal4/ + control (p = 0.70; median Gal4/+ = 50.0, Gal4 > UAS-PKA-C1RNAi = 53.54). In neither of these experiments did the CO2 production rise near the level of the Nf1 knockdown. These data suggest that reducing cAMP/PKA signaling does not produce the metabolic effect observed in Nf1 loss-of-function conditions.

thumbnail
Download:
Fig 3. Manipulating cAMP and PKA enzyme expression did not mimic Nf1 effects on metabolic rate.

A. Diagram showing Nf1 and its connection to cAMP generation and PKA activity. B. CO2 production (arbitrary units; AU) in flies expressing RNAi targeting the Rut adenylyl cyclase or Nf1 in metabolism-regulating PCB-Gal4 + neurons. ***p < 0.001; n.s.: not significant (Dunn’s test, two-sided; n = 6). C. CO2 production in flies expressing RNAi targeting the catalytic PKA-C1 subunit or Nf1 in PCB-Gal4 + neurons. ***p < 0.001; n.s.: not significant (Šidák, n = 5-6).

https://doi.org/10.1371/journal.pgen.1012061.g003

A previous study reported that expression of Nf1 with a patient-derived mutation in the Ras GAP-related domain of Nf1 failed to rescue the Nf1 metabolic effects [11], initially implicating Ras signaling as the potential mechanism. The data above show that knocking down MAPK or mTOR pathways downstream of Ras can rescue the Nf1 effects on metabolism (Figs 2 and 3), and that reducing cAMP/PKA levels does not phenocopy the Nf1 effect (Fig 3). Thus, Nf1 affects metabolism via a neuronal mechanism that depends on Ras and downstream pathways that function largely independently of cAMP/PKA.

Loss of Nf1 modulates metabolic rate via interneurons, with additional contributions from muscle

After establishing the role of downstream signaling in Nf1 effects on behavior and metabolic regulation, we sought to identify the anatomical location where Nf1 acts to produce these effects, particularly in terms of metabolism. PCB-Gal4 is the most restricted driver that produces a phenotype when used to knock Nf1 down [11]. The driver contains several distinct subsets of neurons, as well as additional tissues in adult animals (Fig 4). We visualized the neuronal subsets labeled by the driver by expressing mCD8::GFP (Fig 4A-4C). These include sensory neurons from the halteres and wings (Fig 4B), interneurons in the central brain and ventral nervous cord (VNC) (Fig 4A), and insulin-producing cells [11]. To explore the potential roles of various tissues on Nf1-dependent metabolic modulation, we knocked Nf1 down in a variety of metabolically relevant tissues with RNAi. Loss of Nf1 in insulin-producing cells does not affect metabolism [11], leaving sensory neurons and interneurons as remaining candidates. To test the role of sensory neurons, we knocked down Nf1 in neurons that innervate the campaniform sensillae (CS), a type of sensory neuron in the wings and halteres (Fig 4B). The DB331-Gal4 driver labels haltere CS afferents [64], and knocking down Nf1 with this driver did not detectably alter metabolic rate (Fig 4E). In a complementary experiment, we knocked Nf1 down in wing campaniform sensillae using R12C07-Gal4 [64]. Similar to DB331, we did not detect a difference in metabolic rate when knocking Nf1 down with R12C07 (Fig 4E). This suggests that while the PCB-Gal4 driver drives expression in campaniform afferents, these neurons are not responsible for the alterations in metabolic rate.

thumbnail
Download:
Fig 4. Neurons and other tissues labeled by the PCB-Gal4 driver that potentially contribute to Nf1 metabolic effects.

A. Diagram and GFP image of the haltere tract passing through the ventral nerve cord (VNC). Left: diagram of the VNC. ProNm: prothoracic neuromeres, AMNp: accessory metathoracic neuromeres, MesoNm: mesothoracic neuromeres, MetaNm: metathoracic neuropil, ANm: abdominal neuromere. Ant: anterior, lat: lateral. Right: immunostained VNC from a PCB-Gal4 > UAS-mCD8::GFP fly counterstained with the neuronal marker bruchpilot (brp). B. Diagram of campaniform sensillae (CS) in the wing and haltere, along with a GFP image showing that PCB-Gal4 labels CS in the haltere. C. Diagram showing the location of the corpora cardiaca (CC) and oenocytes, along with an image of the PCB-Gal4 driver showing labeling in the brain and adjacent corpora cardiaca (CC). The tissue was counterstained with brp. D. VNC with PCB-Gal4 + cells labeled with nuclear-localized GFP.nls. Cell bodies were counted and marked with dots. Each neuromere is outlined with dashed white lines the left hemisphere, and the number of labeled cells is noted on the left side. E. Quantification of CO2 production when Nf1 was knocked down using drivers expressing in CS sensory neurons (DB331-Gal4 [p = 0.19, ANOVA, n = 7-9], R12C07-Gal4 [p = 0.26, ANOVA), CC (R64D11-Gal4 [p = 0.14, ANOVA, n = 5-6]), oenocytes (OK72-Gal4 [p = 0.03, Kruskal-Wallis, n = 8-9], desat1-Gal4 [P < 0.001, Kruskal-Wallis, n = 7-8]), or muscle (Mef2 [p < 0.001, ANOVA, n = 5], c179 [p < 0.001, ANOVA, n = 5], 24B [p = 0.04, ANOVA, n = 4-5, R22H05 [p < 0.001, ANOVA, n = 7-10). *p < 0.05, ***p < 0.001 re: both controls (Šidák, n = 4-10).

https://doi.org/10.1371/journal.pgen.1012061.g004

Next, we examined PCB-Gal4 for any non-neuronal Nf1 metabolic contributions. In addition to the brain and ventral nervous system, the Gal4 driver strongly labels neurosecretory cells in the corpora cardiaca (CC) (Fig 4C). A previous study reported that knocking Nf1 down with either of two CC drivers, Akh-Gal4 and Tk-Gal4 did not affect metabolic rate [11]. Nonetheless, given the strong expression of PCB-Gal4 in the CC, we further challenged the role of this driver in Nf1-dependent metabolic regulation here. Knocking down Nf1 with an additional CC-selective Gal4 driver, R64D11-Gal4 [65], did not alter metabolism (Fig 4E), confirming that Nf1 in the CC does not cell-autonomously regulate metabolism. Beyond the CC, additional cell types regulate metabolism, including the fat body and oenocytes (Fig 4C). The fat body is a lipid storage site that functions similarly to mammalian adipose tissue [66]. Knockdown of Nf1 in the fat body does not affect metabolism, suggesting that it is not the primary site of Nf1-mediated metabolic effects [11,67]. Oenocytes are hepatocyte-like neurosecretory cells in the fly abdomen that respond to nutrient state, synthesize very long-chain fatty acids, and regulate energy homeostasis [68] (Fig 4C). Knocking Nf1 down with either of two oenocyte-expressing drivers, OK72 and desat [69,70], did not significantly affect metabolic rate (Fig 4E). There was a slight decrease in the experimental group relative to controls with these drivers, resulting in a main effect in the ANOVA, but the experimental group did not significantly differ from both Gal4/+ and UAS/ + groups using either driver. Furthermore, the trend was in the downward direction and therefore could account for the Nf1 effect on metabolism. Thus, we conclude that Nf1 does not impact metabolic rate via functions in the CC, fat body, or oenocytes.

The above data implicated neurons in the PCB-Gal4 driver. This driver labels a set of neurons in the central brain and VNC [11]; the metabolic phenotype is suppressed by the ventral nervous system-biased tsh-Gal80 repressor [11], suggesting that the VNC is a likely locus of Nf1 metabolic effects. To gain insight into the neurons that could be responsible for the phenotype, we labeled nuclei of PCB-Gal4 + cells with nuclear-localized GFP (GFP.nls) and counted the labeled cells in the VNC in one fly (Fig 4D). There were 981 labeled cell bodies across the VNC, with 208 in the prothoracic neuromeres (ProNm), 148 in the accessory metathoracic neuromeres (AMNp), 213 in the mesothoracic neuromeres (MesoNm), 257 in the metathoracic neuropil (MetaNm), and 155 in the abdominal neuromere (ANm) (Fig 4D). The VNC contains approximately 23,000 neurons [71], so the 981 labeled neurons in the PCB-Gal4 driver represent approximately 4.3% of the total neurons that have cell bodies within the VNC. Overall, this suggests that PCB-Gal4 + non-neuronal cells (oenocytes and corpora cardiaca), neurons innervating the VNC (campaniform sensillae afferents in the haltere tract), as well as other candidates (insulin-producing cells, peptidergic cells, and fat body [11]) are not responsible for the Nf1 effect on metabolism. Thus, interneurons within the VNC are the likely major modulators of Nf1-dependent neuronal metabolic effects.

Muscle tissue is an additional candidate for non-neuronal Nf1 effects on metabolism [30,35,72]. Muscle movement and maintenance are energetically costly – in insects, skeletal (flight) muscles consume the largest amount energy in proportion to the animal’s weight [73,74]. Additionally, Nf1 exerts effects on metabolic parameters in mammalian muscle [30,38,72,75]. To test whether loss of Nf1 in muscle – in addition to neurons - contributes to the organismal metabolic phenotype, we knocked Nf1 down using four Gal4 drivers that include muscle expression: Mef2, c179, 24B, and R22H05 (Fig 4E) [64,76,77]. Metabolic rate was increased when Nf1 was knocked down with Mef2, c179, and R22H05, but not 24B. Among these drivers, 24B is expressed in muscle broadly through development, but its expression drops progressively after eclosion [78]. Therefore, the lack of effect when knocking Nf1 down with 24B suggests that Nf1 could impact metabolism via post-developmental effects in adult flies, though differences in the expression patterns between the drivers (across muscle or off-target outside muscle) could also be responsible. The effect size of the metabolic effect when knocking Nf1 in muscle (Fig 4E) was smaller than when knocking down Nf1 with the PCB-Gal4 driver (Fig 1C) or pan-neuronal knockdown [11]. Thus, the metabolic effect may be dominated by neuronal expression in the driver(s); for instance, Mef2-Gal4 contains robust expression in the brain (along with skeletal and visceral muscle) [79]. Overall, these data suggest that interneurons within the VNC are the major source of the Nf1 metabolic effect, with additional contribution from Nf1 in muscle.

Nf1 deficiency altered muscle mitochondria morphology

Alterations in metabolism could result from changes in mitochondrial structure and/or function. Loss of Nf1 impacts ROS production [41] and mitochondria bioenergetics in an oncogenic context [31]. To test whether mitochondria were altered, we first labeled neuronal mitochondria pan-neuronally with a genetically-encoded, mitochondrial-targeted fluorescent protein, and analyzed mitochondria in the VNC (Fig 5A-5D). At the light microscopy level, we detected no significant differences in mitochondria number between controls and nf1P1 mutants (Fig 5B). There were also no differences in mitochondrial volume (Fig 5C) or sphericity (Fig 5D) between nf1P1 mutants and controls. Analysis of electron micrographs taken from the protocerebrum neuropil area were unremarkable, revealing no qualitative differences between nf1P1 mutants and wCS10 controls with respect to features such as overall neuropil morphology, synaptic T-bar and vesicle distribution and size, and membrane integrity (Fig 5E and 5F). Quantification of mitochondrial area revealed a decrease in mitochondria size in protocerebral neurons (Fig 5G and 5H). While the difference was statistically significant, it was modest in magnitude, and there was little shift in the distribution of mitochondria sizes (Fig 5G). Thus, these data suggest that there were no major changes in either mitochondrial regulation (e.g., via fission/fusion) or neuronal ultrastructure in nf1 mutant neurons.

thumbnail
Download:
Fig 5. Nf1 effects in muscle contribute to the metabolic alterations.

A. Diagram of ventral nerve cord with neuronal mitochondria labeled with GFP. B. Neuronal mitochondrial counts in control flies and nf1P1 mutants (unpaired t-test; n = 5-6). C. Neuronal mitochondrial volume in control flies and nf1P1 mutants (unpaired t-test; n = 5-6). D. Neuronal mitochondrial sphericity in control flies and nf1P1 mutants (unpaired t-test; n = 5-6). E. Representative transmission electron micrograph of protocerebral neuropil from a control (wCS10) fly. F. Representative transmission electron micrograph of protocerebral neuropil from an nf1P1 mutant fly. G. Histogram of neuronal mitochondria size, quantified from electron micrographs from the Drosophila protocerebrums of two flies. Measurements are 2D cross-sectional area in μm2, graphed in 0.05 μm2 width bins (n = 227 [wCS10], 207 [nf1P1]). H. Individual mitochondria area measurements, quantified from electron micrographs (same values plotted in panel G). **p < 0.01 (Mann-Whitney). I. Representative transmission electron micrograph of flight muscle from a control (wCS10) fly. Junctions between adjacent mitochondria are highlighted with black arrowheads. J. Representative transmission electron micrograph of flight muscle from an nf1P1 mutant fly. Spacing between adjacent mitochondria is highlighted with black arrowheads. K. Histogram of muscle mitochondria size, quantified from electron micrographs from the Drosophila flight muscle of two flies. Measurements are 2D cross-sectional area in μm2, graphed in 0.5 μm2 width bins (n = 185 [wCS10] and 189 nf1P1]). L. Individual mitochondria area measurements, quantified from electron micrographs (same values plotted in panel). ***p < 0.001 (Mann-Whitney).

https://doi.org/10.1371/journal.pgen.1012061.g005

As loss of Nf1 affected metabolism via effects in muscle (as well as neurons), we asked whether muscle morphology and mitochondria were affected by Nf1 deficiency. To test this, we examined electron micrographs of nf1P1 mutants and controls. Ultrastructural analysis of Drosophila flight muscle revealed qualitative and quantitative changes in mitochondria. There was visibly abnormal spacing between adjacent mitochondria in nf1P1 mutant muscle (Fig 5I and 5J). Further, the mitochondria were larger in nf1P1 mutants, exhibiting an increase in cross-sectional area relative to wCS10 controls, which was accompanied by a rightward shift in the distribution in the mutants (Fig 5K and 5L). Thus, there were detectable changes in muscle mitochondria ultrastructure. Loss of Nf1 affected neurons and muscle differently at the ultrastructural level, with muscle exhibiting notable alterations in mitochondrial structure.

Neurofibromin exerts behavioral effects via MEK, with no requirement for Akt signaling

The above data suggest Nf1 signals through MAPK and Akt/mTOR to regulate metabolism. In addition to modulating systemic physiological phenotypes such as metabolic rate, Nf1 also regulates behavioral phenotypes. Do behavioral phenotypes exhibit similar dependence on both MAPK and mTOR signaling pathways? To test this, we zeroed in on a robust Nf1-dependent behavioral phenotype: spontaneous grooming. Genomic nf1 mutations or pan-neuronal Nf1 loss of function increases spontaneous grooming in Drosophila in an Nf1 Ras GAP-related domain-dependent manner [43]. We used this phenotype here to test whether each arm of the signaling pathway downstream of Nf1 was necessary for Nf1-dependent effects on behavior. Flies were monitored in open field arenas for 5 minutes and grooming behavior was quantified (Fig 6A). The Nf1 effect on grooming is present in both males and females and indistinguishable in magnitude [45], so we used males here to avoid egg accumulation in the arenas.

thumbnail
Download:
Fig 6. Loss of Nf1 alters a behavioral phenotype, grooming, via effects on MEK, but not Akt.

A. Diagram of the open field arena containing a fly, with a camera recording above. B. Quantification of grooming in flies expressing wild-type MEK (UAS-MEKwt) or a constitutively active mutant MEK (UAS-MEKE203K). ***p < 0.001; n.s.: not significant (Šidák, n = 13-16). C. Diagram showing Nf1-Ras signaling and the downstream Raf/MEK and PI3K/Akt signaling pathways. Green arrows represent the direction of signaling change following loss of Nf1. Red X marks show the molecules targeted for in vivo genetic analyses in panels D-E. D. Quantification of grooming in flies harboring RNAi targeting Nf1, MEK, or both, along with controls. ***p < 0.001; n.s.: not significant (Šidák, n = 12-16). E. Quantification of grooming in flies harboring RNAi targeting Nf1, Akt, or both. ***p < 0.001; n.s.: not significant (Šidák, n = 9-13).

https://doi.org/10.1371/journal.pgen.1012061.g006

Focusing on grooming as a behavioral readout, we asked whether MAPK and/or Akt signaling was necessary for Nf1-dependent effects. MAPK signaling has been previously implicated in metabolic modulation via gain-of-function approaches [11], but whether upregulating this pathway affects grooming is unknown. We first tested this by expressing a constitutively active MEK mutant, MEKE203K [80]. Expressing wild-type MEK pan-neuronally did not affect grooming levels, but expressing MEKE203K significantly increased grooming frequency (Fig 6B), reminiscent of the nf1 mutant phenotype. To dissect the requirement for signaling molecules downstream of Ras, we employed the same in vivo genetic analysis approach that was used to dissect the metabolic signaling, knocking down Nf1 in combination with a signaling molecule downstream of Ras (Fig 6C and 6D). Knocking down Nf1 in neurons (with nSyb-Gal4) increased spontaneous grooming frequency (Fig 6D), as expected [43–45]. Knockdown of MEK did not, on its own, affect behavior (Fig 6D). However, when knocking down both MEK and Nf1, the MEK knockdown occluded the behavioral effect of knocking Nf1 down (Fig 6D). This suggested that MEK was necessary for the Nf1 effect on the behavioral phenotype.

Nf1 modulates multiple signaling pathways downstream of Ras, including Raf/MEK and PI3K/Akt signaling pathways (Fig 6C). Therefore, we next tested whether PI3K/Akt signaling is involved in regulating the behavioral phenotype (excessive grooming caused by loss of Nf1 in neurons). As with the metabolic phenotype, we targeted Akt (rather than PI3K) here. In contrast to MEK, knocking down Akt pan-neuronally (with nSyb-Gal4) did not occlude the Nf1 effect on behavior – there was no significant difference in grooming between the single Nf1 knockdown and double Nf1 + Akt knockdown (Fig 6E). This suggested that reducing the expression of MEK, but not Akt restored a behavioral phenotype. Thus, Nf1 affected behavior via selective effects on MAPK signaling in this signaling and phenotypic context.

Modulation of mTOR/S6K rescues Nf1 behavioral effects independent of Akt

The above data revealed that behavioral effects of Nf1 knockdown were occluded by secondary loss of MEK but not Akt. In contrast, metabolic effects were occluded by both MAPK and mTOR pathway manipulations. There is significant crosstalk between signaling pathways, including ERK and mTOR [81] (Fig 7A). Moreover, mTOR is a central metabolic regulator that is activated by multiple signaling pathways such as AMPK [50,58]. We therefore asked whether the behavioral effects of loss of Nf1 could be rescued by modulating signaling at/below the level of mTOR, where multiple signaling pathways converge (Fig 7A). To address this, we first tested whether pan-neuronal knockdown of Raptor (with nSyb-Gal4) occluded the Nf1 effect on grooming. Raptor knockdown alone did not affect behavior but double knockdown with Nf1 occluded the effect (Fig 7B). Next, we moved one step downstream in the pathway, testing whether S6K occluded the Nf1 effect on grooming. Double knockdown of Nf1 and S6K also occluded the effect of Nf1 knockdown alone (Fig 7C). Thus, knocking down either Raptor or S6K rescued behavior in an Nf1 deficient background. Since MEK, but not Akt, was necessary for behavioral rescue, this suggests that either crosstalk between the pathways (Fig 7A) could be responsible or additional inputs to mTOR (e.g., AMPK) could modulate behavioral responses independent of Akt.

thumbnail
Download:
Fig 7. Behavioral effects of Nf1 rely on signaling downstream of mTOR.

A. Diagram showing Nf1-Ras signaling and the downstream MAPK and mTOR signaling pathways. Green arrows represent the direction of signaling change following loss of Nf1. Red X marks show the molecules targeted for in vivo genetic analyses in panels B-C. B. Quantification of grooming in flies harboring RNAi targeting Nf1, Raptor, or both. **p < 0.01, ***p < 0.001; n.s.: not significant (Dunn’s test, two-sided; n = 14-16). C. Quantification of grooming in flies harboring RNAi targeting Nf1, S6K, or both. **p < 0.01, ***p < 0.001; n.s.: not significant (Šidák, n = 13-15).

https://doi.org/10.1371/journal.pgen.1012061.g007

Discussion

Neurofibromatosis type 1 is a monogenetic disorder that affects diverse and complex behavioral, cellular, and organismal phenotypes. While the disorder results from mutations in a single gene, its protein product regulates a variety of diverging downstream signaling pathways. This raises the question of whether multiple neurofibromatosis type 1 phenotypes exhibit similar or differing dependence on the signaling pathways downstream of Nf1 and Ras? To test this, we implemented in vivo genetic analysis of signaling pathways in Drosophila - a robust platform for dissecting the action of individual signaling molecules on organismal phenotypes. Two human disorder-relevant phenotypes – behavior (grooming) and metabolic modulation – exhibited differing dependence on signaling pathways immediately downstream of Nf1/Ras. Nf1-dependent behavioral alteration was rescued by manipulation of MEK but not Akt. In contrast, the Nf1 metabolic phenotype was rescued by both MEK and Akt (Fig 8).

thumbnail
Download:
Fig 8. Summary diagram showing Nf1-Ras signaling and the downstream MAPK and mTOR signaling pathways.

The signaling pathways implicated in behavioral and metabolic signaling are highlighted. Molecules that were tested with double RNAi in both assays are shown (PI3K is included as an upstream signaling molecule in the mTOR pathway but was not tested).

https://doi.org/10.1371/journal.pgen.1012061.g008

Results from the present study raise an important question – does one treatment fit all for neurofibromatosis type 1 symptoms? Answering this question will require follow up from several angles. The fundamental premise will need to be tested in additional preclinical models, including mammalian models (ultimately humans). It will also need to be extended to additional phenotypes. The two phenotypes selected for the present study represent a subset of the alterations in Nf1 animal models, which are themselves a subset of the symptoms in patients. This study focused on Ras signaling and, in turn, the two most well-studied signaling pathways downstream of Ras. Further studies will be required to understand if/how additional signaling pathways contribute to behavioral and metabolic alterations (among other phenotypes) and whether they also exhibit variable/differential involvement in neurofibromatosis type 1 phenotypes. Addressing these broad questions could open new avenues to treat neurofibromatosis type 1 symptoms in novel ways.

Multiple signaling pathways have been implicated in neurofibromatosis type 1 phenotypes, including Ras/MAPK, Ras/PI3K/mTOR, and cAMP/PKA. The commonly studied MAPK signaling pathway is implicated in multiple phenotypes, and MEK is the sole target for current FDA-approved treatments [21,22]. FDA-approved drugs target multiple molecules implicated in the present study, including MEK, ERK, and Akt. These molecules are currently used primarily for treatment of various cancers and organ transplant rejection. mTOR is targeted with the FDA-approved drugs sirolimus (rapamycin), temsirolimus, and everolimus [82]. MEK is targeted by trametinib and selumetinib (used to treat NF1 cancers), along with two others [83]. A recently approved drug, capivasertib, targets Akt [84]. All of these drugs are inhibitors, which could potentially help to normalize the elevated signaling in neurofibromatosis type 1. If certain pathways/molecules are differentially effective targets for treating certain symptoms in humans, some drugs (alone or in combinations) may be particularly effective in treating those symptoms relative to others. Thus, the specific drug or combination therapy could be tailored to the suite of symptoms being targeted, similar to combination therapies for cancers. Drosophila express only one isoform each of MEK, ERK, and AKT, in contrast to the multiple isoforms found in mammals. This simplified genetic analysis in the present study, as it allowed us to target a single molecule for genetic epistasis tests. However, it may result in drug effects in humans being more complex, both in terms of dose dependence and effects across different phenotypes.

While this study focused on two readouts of metabolism and behavior – CO2 production (metabolism) and grooming (behavior), multiple other phenotypes are likely impacted by these signaling pathways. Even within a single broadly defined domain, multiple phenotypes could be differentially altered by MAPK and/or mTOR. For instance, loss of Nf1 alters many metabolic parameters, including CO2 production, O2 consumption, ATP production, NAD+ metabolism, mitochondrial function, glycogen levels, triglyceride levels, lipid turnover, starvation survival, ROS production, sleep-metabolism interactions, and lifespan [11,12,40,41,46,85]. Similarly, Nf1 impacts a range of behavioral phenotypes, including sleep and circadian rhythms [62,86–88], associative learning [20,60,89–91], social behavior [92], and tactile hypersensivity [93], among others. Each of these phenotypes could be differentially regulated by MAPK and mTOR signaling. In addition, each of these phenotypes could exhibit additional dependence on other signaling pathways beyond MAPK and mTOR signaling, such as RalGEF/Ral, PLC, and more. The present study provides proof-of-concept that some Ras effector pathways differentially modulate organismal phenotypes. Additional study will be necessary to determine how this plays out further within and across each set of pathways and phenotypes.

Moving down the pathways revealed evidence of crosstalk. Akt was not required for Nf1-dependent modulation of grooming, yet downstream mTOR complex 1 signaling molecules, such as Raptor and S6K were. This suggests that Nf1-dependent activation of mTOR occurred via crosstalk between signaling cascades downstream of Ras (Fig 8). Crosstalk between MAPK and mTOR signaling pathways is one way in which this could occur. Multiple points of crosstalk between these pathways have been reported, including ERK-mediated phosphorylation of Raptor and cross-activation via intermediates such as RSK and TSC1/2 [81]. The existence of such crosstalk facilitates potential treatment of NF1 behavioral phenotypes in several ways. For instance, it could be possible to target a small set of upstream signaling molecules (e.g., MEK) or any one of a larger set of downstream signaling molecules representing their direct targets (e.g., ERK) or indirect targets (e.g., mTOR). Targeting deep levels of the mTOR signaling pathway (Raptor and S6K) normalized both Nf1 behavioral effects and metabolism, suggesting that moving further down the signaling pathways may result in broader effects across phenotypes. Further preclinical studies will be needed to identify viable targeting approaches.

In contrast to the behavioral phenotype, Nf1’s metabolic effects were Akt-dependent. In the metabolic context, Akt likely functions via activation of mTOR – Akt activates the Raptor-containing mTOR complex 1 via a signaling pathway involving TSC1/2 and Rheb [57]. mTOR complex 1 is a conserved central metabolic regulator [94], exerting effects on cellular growth and metabolic processes [57]. The dependence of the Nf1 metabolic phenotype on both MAPK and Akt/mTOR signaling pathways is reminiscent of signaling in various cancers. For instance, both MEK and Akt are hyperactivated in optic gliomas, and inhibition of either pathway reduces several optic glioma phenotypes [53]. Similarly, inhibition Akt and MEK synergistically suppresses NF1 malignant peripheral nerve sheath tumor (MPNST) growth [95]. The cellular mechanism remains to be fully elucidated, though it could involve metabolic alterations; Akt promotes mitochondrial fusion and elongation [96]. Whether this represents a similar mechanism to that driving alterations in organismal metabolic alterations is unclear – we detected no major changes in neuronal morphology, though there was evidence for changes in muscle mitochondria.

mTOR is hyperactivated in neurofibromatosis type 1 and regulates some neurofibromatosis type 1 phenotypes, particularly tumors [97,98]. Both Ras/ERK and cAMP/PKA signaling regulate aspects of cognition in animal models [60,99,100]. Ras signals through a variety of additional effectors, including RalGEF/RalA, TIAM1, PLC, RASSF1/5. Several of these pathways have been implicated in neurofibromatosis type 1 tumor phenotypes. RalGEF/RalA is a Ras effector pathway that is overactivated in NF1 malignant peripheral nerve sheath tumors (MPNST) [101,102]. RASSF1 expression is altered in NF1 MPNST via DNA promotor methylation-induced silencing, and this is associated with poor prognosis [103]. Ras-TIAM1-Rac signaling is implicated in proliferation of plexiform neurofibromas, as genetic disruption of Rac prevents tumor formation [104]. Combination therapies targeting multiple pathways may be an effective strategy both for cancers [105,106] and other phenotypes, such as the cognitive and behavioral symptoms.

Nf1 regulates cAMP in a Ras-dependent manner via the activation of atypical protein kinase C (PKCζ) in neurons [107]. In the present study, we did not find evidence for strong cAMP effects on the Nf1 metabolic effect, as cAMP/PKA manipulations did not phenocopy the nf1 mutant phenotype. One caveat is that – given this lack of phenocopy – we did not attempt to normalize cAMP/PKA levels in the mutant background. In addition, there was a slight trend toward an increase in the median in PKA knockdown. Thus, while it cannot account for the Nf1 effect, there could be a contribution of cAMP/PKA signaling that is not dominant in the effect observed here.

At the circuit/systemic level, Nf1 exerts a primary effect in neurons to modulate the Nf1 metabolic phenotype, with a secondary contribution from muscle tissue. The critical neurons are some subset of the 981 VNC interneurons labeled by the PCB-Gal4 driver. In addition to the interneurons the PCB-Gal4 driver labels multiple other cell types, including campaniform sensillae in the halteres and wings, corpora cardiaca, and fat body. Tissue-specific RNAi experiments in the present and a prior study [11] revealed no effect of knocking Nf1 down in these cell types, suggesting that these tissues may not contribute. Oenocytes represent an additional potential site of systemic metabolic effects, which we tested here. Caveats to these experiments include the possibility of potential shifts in Gal4 expression patterns of the drivers over development/lifespan [108], developmental effects of Nf1 on neurons [43], “off-target” expression sites [109], or nonlinear expression levels across labeled cells. To minimize the potential for these confounds, we have tested multiple lines covering each of these cells/tissues. Thus, multiple lines of evidence support a role for both neurons and muscle in mediating Nf1 effects on metabolism.

Subcellular changes in nf1 mutants included differences in muscle mitochondrial ultrastructure between nf1 mutants and wild-type controls. In the nervous system, there were no notable differences in overall neuronal morphology or synaptic structure between nf1 mutants and controls, though there was a small shift in the distribution of mitochondrial size. In muscle, there was a more pronounced shift in mitochondrial size, which was accompanied by a qualitative change in mitochondrial appearance (increased spacing). This change in mitochondrial appearance was subtle, yet readily detectable in Drosophila flight muscle since the mitochondria line up end-to-end. In comparison, mitochondria in mice with biallelic Nf1 inactivation in muscle appear normal, though there is significant muscle atrophy, metabolic dysregulation, and ultimately lethality [30]. The mitochondrial changes in Drosophila were reminiscent of swollen mitochondria reported in another study [85]. Loss of nf1 drives mitochondrial dysregulation [41], which could contribute to changes in mitochondrial regulation and morphology.

Changes in CO2 production in the nf1 mutants could arise from several cellular sources. These include pyruvate decarboxylation, the tricarboxylic acid (Kreb’s) cycle, and the pentose phosphate pathway. While the present study did not examine the cellular source of the CO2 production, CO2 production correlates well with energy expenditure and is considered to be a proxy for metabolic rate [48]. Changes in metabolic rate can result from alterations in mitochondrial regulation via fusion and fission, which are detectable at the morphological level. Comparing the nf1 mutants to control animals, we found that there were no large-scale changes in mitochondrial regulation within neurons, despite a strong neuronal dependence of the Nf1 metabolic effect. Yet there were more robust changes in mitochondrial morphology in muscles in the nf1 mutants, which also exhibit Nf1-dependent metabolic effects. There are multiple potential explanations for these observations. First, Nf1 could affect metabolism in neurons via mitochondrial effects that do not involve fusion or fission (e.g., effects on electron transport chain activity). Second, Nf1 could metabolism in neurons via mitochondria-independent effects (e.g., altering glycolysis). Third, Nf1 could act in neurons but affect metabolism via systemic effects (e.g., altering release of peptides such as insulin-like peptides, adipokinetic hormone, or others [85]).

Overall, this study revealed how loss of Nf1 modulates two organismal phenotypes (grooming behavior and metabolism) through effects on two major Ras effector pathways: MAPK and mTOR. The results suggest that multiple nodes of the pathway could be targeted to approach NF1 phenotypes, particularly the metabolic alterations. Yet some of the phenotypes exhibit differential dependence on certain signaling nodes, a feature that could be exploited to target different NF1 phenotypes independently. While MAPK and mTOR are two of the major, well-studied Ras effectors, Ras signals through a wide range of effectors. Future studies will be required to address these signaling pathways more broadly. The diversity and complexity of Ras signaling represents both a challenge and an opportunity for identifying new therapeutic targets for NF1 and other Rasopathies.

Materials and methods

Drosophila husbandry and stocks

Flies were raised on Drosophila cornmeal/agar food medium according to standard protocols and housed at 25 °C, 60% relative humidity, on a 12:12 h light:dark cycle. The nf1P1 mutation was backcrossed six generations into the control genetic background wCS10. RNAi lines were obtained from the Vienna Drosophila RNAi Center, Gal4/ + control crosses consisted of an empty attP control line (VDRC #60100). UAS-dicer-2 was included to potentiate the RNAi effect [49], and was included in all genotypes containing the UAS-RNAi. Male flies were used for all experiments, unless otherwise specified, to prevent egg accumulation in behavioral and respirometry chambers.

Behavioral analysis

Spontaneous grooming was quantified in an open field [44,110]. Individual flies, 5–10 days post-eclosion, were aspirated into an open field area, 15.4 mm in diameter and 2.85 mm in height, consisting of an opaque white PLA boundary wall covered on the top and bottom with clear polycarbonate sheets. The arena was illuminated from below with light from white light-emitting diodes that was diffused through a sheet of white acrylic. Light intensity was measured at 720 lm/m2 in the location of the fly. Videos were recorded at 7.5 frames per second, 1,616 x 1,240, lossless Motion JPEG 2000 compression using a camera mounted above the chamber (FLIR Teledyne Blackfly S) fitted with a 25 mm lens (Edmund Optics). Five-min videos were recorded 30 min following the introduction into the open-field arena. Grooming was manually scored via frame-by-frame analysis, recording the start and stop frame for each grooming bout. Grooming was calculated as the percent of total time grooming during the 5-minute video.

Metabolic analysis

CO2 production was quantified via respirometry [11,48]. Respirometers were constructed by gluing together a 1 ml pipette tip and 50 µl capillary micropipette. Soda lime was placed into each pipette tip between two foam pieces. Flies were sorted under CO2 anesthesia and allowed to recover for at least 24 h before beginning the experiment. Four flies of the same genotype were aspirated into each respirometer, and the top was sealed with non-hardening modeling clay. Pipette seal was monitored and any that leaked were excluded from analysis. Sixteen respirometers were hung on a custom-made rack in a latch-lid chamber. One control respirometer was left empty in each experiment. The bottom of the chamber was filled with a red water-based dye and the chamber lid was closed/sealed with vacuum grease. The chamber was placed in an incubator and allowed to equilibrate at 25 °C for 1 h. Images were captured every 15 min with PhenoCapture 3.3. The liquid meniscus level was measured in each respirometer after 3 h using Fiji 2.0. Where appropriate, data were normalized to the mean of the control genotypes.

Quantitative polymerase chain reaction (qPCR)

Heads were collected from adult flies (5–10 days post-eclosion; 30 heads per genotype). Each sample was added to QIAzol Lysis Reagent (Qiagen, catalog #79306) and homogenized using a mortar and pestle. RNA was isolated with the Qiagen RNeasy Lipid Tissue Mini Kit. Complimentary DNA library was constructed using LunaScript RT Master Mix (New England Biolabs, catalog #E3025L) with the Random Primer Mix (New England Biolabs, catalog #S1330S). Nf1 gene expression was quantified in fly head samples using Luna Universal qPCR Master Mix (New England Biolabs, catalog #M3003L). Primer sequences were as follows: 5’ -CTTTTGGCACGTTTCGAGGAT-3’ (Nf1_F), 5’ -GGTAGCGCGATATGTGGATCA-3’ (Nf1_R), 5’ -ATGCTAAGCTGTCGCACAAATG-3’ (Rpl32_F), and 5’ -GTTCGATCCGTAACCGATGT-3’ (Rpl32_R). Five biological replicates, with three technical replicates each, were analyzed for each genotype. Analysis of gene expression was performed using standard ΔΔCt analysis. Gene expression was normalized to Rpl32, and log2 fold change was calculated.

Western blot analysis. For analysis of pERK/ERK/β-tubulin, lysates of five-day old adult fly heads were prepared using RIPA buffer supplemented with Pierce Protease and Phosphatase inhibitors (Thermo Scientific). Protein samples were mixed with NuPAGE 4x LDS Sample Buffer (Invitrogen), resolved on 4–15% Tris-Glycine Mini-PROTEAN TGX Stain-Free Protein Gels (Bio-Rad) with Tris Glycine SDS. Gels were transferred to polyvinylidene difluoride (PVDF) membranes and 5% BSA in TBST was used for blocking. Primary antibodies used were rabbit pERK (CST, 9101, 1:1000), rabbit ERK (CST, 4695, 1:1000), and mouse anti-β-tubulin (DHSB, E7, 1:10000). The membranes were washed with TBST, followed by secondary antibody incubation for 1 hour at room temperature. Secondary antibodies used: goat anti-Mouse IgG (H + L) Alexa 800 (Invitrogen, A32730, 1:10000) and donkey anti-Rabbit IgG (H + L), Alexa 680 (Invitrogen, A10043, 1:5000). Membranes were stripped using NewBlot PVDF stripping buffer (LI-COR, 928–40032) after blotting with pERK/β-tubulin, then reblocked in 5% BSA and incubated with ERK/β-tubulin antibodies overnight. Images were captured with an Odyssey XF Imaging System (LI-COR).

Immunohistochemistry, light microcopy, and mitochondria analysis

Five to ten-day-old adult flies were dissected fixed in 1% paraformaldehyde in in Schneider’s Drosophila medium and processed as previously described [111]. Samples were stained with primary antibodies for 3 hours at room temperature and 4 °C overnight, followed by secondary antibodies for 3 hours at room temperature and 4 days at 4 °C. Incubations were performed in blocking serum (3% normal goat serum). Samples were mounted in Vectashield (Vector Laboratories) for imaging. The following antibodies were used: rabbit anti-GFP (1:1000, Invitrogen), mouse anti-brp (nc82) (1:50, DSHB), goat anti-rabbit IgG, and goat anti-mouse IgG (1:800, Alexa 488 or Alexa 633, respectively, Invitrogen). Mitochondria were fluorescently labeled by expressing UAS-4mtGCaMP3 and immunostaining the GFP subunit of the GCaMP molecule. Cellular nuclei were labeled with DAPI (1:1000, Invitrogen). A Leica SP8 confocal microscope with LAS X software was used to obtain images following standard protocols. Mitochondria were measured as previously described [112]. From each fly, a 15 μm3 z stack was collected from the right ventral nervous system, processed with ClearView deconvolution (Imaris) using 50 iterations, and mitochondrial parameters (count, volume, sphericity) were quantified in Imaris.

Transmission electron microscopy (TEM) and ultrastructure analysis

Drosophila head and flight muscle ultrastructure was imaged following standard electron microscopy procedures using a Ted Pella Bio Wave processing microwave with vacuum attachments. The tissue was covered in 2% paraformaldehyde, 2.5% glutaraldehyde, in 0.1 M sodium cacodylate buffer at pH 7.2. After dissection the heads were incubated for 3 days in fixative in a cold room rotator. All processing from fixative to ethanol were carried out on ice. The pre-fixed heads were then fixed again in the microwave vacuum processor, followed by 3x purified water rinses, post-fixed with 1% aqueous osmium tetroxide, and rinsed again 3x with purified water. Concentrations from 30-100% of ethanol were used for the initial dehydration series, followed with propylene oxide as the final dehydrant. Samples were gradually infiltrated with 3 ratios of propylene oxide and Embed 812, finally going into 3 changes of pure resin under vacuum. Samples were allowed to infiltrate in pure resin overnight on a rotator. The samples were embedded into flat silicone molds and cured in the oven at 62 °C for three days. The polymerized samples were thin-sectioned at 48–50 nm and stained with 1% uranyl acetate for ten minutes followed by lead citrate for one minute before TEM examination. Grids were viewed in a JEOL 1400 Plus transmission electron microscope at 80kV. Images were captured using an AMT XR-16 mid-mount 16 mega-pixel digital camera.

Statistical analysis

Normality of data was assessed with the D’ Agostino Pearson Test. In figures, box plots graph the median as a line, the interquartile range (IQR) as a box, and whiskers extend to the min/max values. Hypothesis testing was carried out using t tests or ANOVA followed by Šidák’s multiple comparisons tests (parametric), or Wilcoxon rank-sum test or Kruskal-Wallis omnibus test followed by Dunn multiple comparisons tests (nonparametric). Two-way comparisons were carried out with a two-way ANOVA followed by Šidák’s multiple comparisons tests. For RNAi and Gal4-mediated overexpression, the experimental group was compared to heterozygous Gal4/+ and UAS/ + controls and considered positive only if it significantly differed from both controls in the same direction. For double RNAi experiments, all pairwise comparisons were statistically analyzed; for clarity, the three most informative comparisons are shown in the figures: the double knockdown vs. each single knockdown, and each single knockdown vs. the other. Statistics and graphing were carried out with Graphpad Prism, version 10.1.1.

Acknowledgments

We thank Lita Duraine and the Baylor College of Medicine Jan Duncan Neurological Research Institute microscopy facility for assistance with transmission electron microscopy and the University of Iowa Central Microscopy Resource Facility for assistance with confocal microscopy. In addition, we are grateful to David Gutmann, Corina Anastasaki, Jonathan Payne, J. Elliot Robinson, Robert Kesterson, and the members of the CABIN task force for feedback on Nf1 aspects of the project, as well as Nathan Mohar and Lori Wallrath for helpful discussions on muscle components of the project. We thank Ronald L. Davis for UAS-4mtGCaMP3 and UAS-RFP, Michael Dickinson for DB331-Gal4, and James Walker and Andre Bernards for nf1 mutants. We thank Linda Buckner and Robert Svetly for administrative assistance.

References

  1. 1. Hyman SL, Shores A, North KN. The nature and frequency of cognitive deficits in children with neurofibromatosis type 1. Neurology. 2005;65(7):1037–44. pmid:16217056
  2. 2. Diggs-Andrews KA, Gutmann DH. Modeling cognitive dysfunction in neurofibromatosis-1. Trends Neurosci. 2013;36(4):237–47. pmid:23312374
  3. 3. Constantino JN, Zhang Y, Holzhauer K, Sant S, Long K, Vallorani A, et al. Distribution and Within-Family Specificity of Quantitative Autistic Traits in Patients with Neurofibromatosis Type I. J Pediatr. 2015;167(3):621-6.e1. pmid:26051969
  4. 4. Eijk S, Mous SE, Dieleman GC, Dierckx B, Rietman AB, de Nijs PFA, et al. Autism Spectrum Disorder in an Unselected Cohort of Children with Neurofibromatosis Type 1 (NF1). J Autism Dev Disord. 2018;48(7):2278–85. pmid:29423604
  5. 5. Garg S, Plasschaert E, Descheemaeker M-J, Huson S, Borghgraef M, Vogels A, et al. Autism spectrum disorder profile in neurofibromatosis type I. J Autism Dev Disord. 2015;45(6):1649–57. pmid:25475362
  6. 6. Hyman SL, Arthur Shores E, North KN. Learning disabilities in children with neurofibromatosis type 1: subtypes, cognitive profile, and attention-deficit-hyperactivity disorder. Dev Med Child Neurol. 2006;48(12):973–7. pmid:17109785
  7. 7. Morris SM, Acosta MT, Garg S, Green J, Huson S, Legius E, et al. Disease Burden and Symptom Structure of Autism in Neurofibromatosis Type 1: A Study of the International NF1-ASD Consortium Team (INFACT). JAMA Psychiatry. 2016;73(12):1276–84. pmid:27760236
  8. 8. Plasschaert E, Van Eylen L, Descheemaeker M-J, Noens I, Legius E, Steyaert J. Executive functioning deficits in children with neurofibromatosis type 1: The influence of intellectual and social functioning. Am J Med Genet B Neuropsychiatr Genet. 2016;171B(3):348–62. pmid:26773288
  9. 9. Walsh KS, Vélez JI, Kardel PG, Imas DM, Muenke M, Packer RJ, et al. Symptomatology of autism spectrum disorder in a population with neurofibromatosis type 1. Dev Med Child Neurol. 2013;55(2):131–8. pmid:23163951
  10. 10. Vassallo G, Mughal Z, Robinson L, Weisberg D, Roberts SA, Hupton E, et al. Perceived fatigue in children and young adults with neurofibromatosis type 1. J Paediatr Child Health. 2020;56(6):878–83. pmid:31916647
  11. 11. Botero V, Stanhope BA, Brown EB, Grenci EC, Boto T, Park SJ, et al. Neurofibromin regulates metabolic rate via neuronal mechanisms in Drosophila. Nat Commun. 2021;12(1):4285. pmid:34257279
  12. 12. Botero V, Tomchik SM. Unraveling neuronal and metabolic alterations in neurofibromatosis type 1. J Neurodev Disord. 2024;16(1):49. pmid:39217323
  13. 13. Martin GA, Viskochil D, Bollag G, McCabe PC, Crosier WJ, Haubruck H, et al. The GAP-related domain of the neurofibromatosis type 1 gene product interacts with ras p21. Cell. 1990;63(4):843–9. pmid:2121370
  14. 14. Shen MH, Harper PS, Upadhyaya M. Molecular genetics of neurofibromatosis type 1 (NF1). J Med Genet. 1996;33(1):2–17. pmid:8825042
  15. 15. Walker JA, Gouzi JY, Long JB, Huang S, Maher RC, Xia H, et al. Genetic and functional studies implicate synaptic overgrowth and ring gland cAMP/PKA signaling defects in the Drosophila melanogaster neurofibromatosis-1 growth deficiency. PLoS Genet. 2013;9(11):e1003958. pmid:24278035
  16. 16. Dasgupta B, Dugan LL, Gutmann DH. The neurofibromatosis 1 gene product neurofibromin regulates pituitary adenylate cyclase-activating polypeptide-mediated signaling in astrocytes. J Neurosci. 2003;23(26):8949–54. pmid:14523097
  17. 17. Tong J, Hannan F, Zhu Y, Bernards A, Zhong Y. Neurofibromin regulates G protein-stimulated adenylyl cyclase activity. Nat Neurosci. 2002;5(2):95–6. pmid:11788835
  18. 18. The I, Hannigan GE, Cowley GS, Reginald S, Zhong Y, Gusella JF, et al. Rescue of a Drosophila NF1 mutant phenotype by protein kinase A. Science. 1997;276(5313):791–4. pmid:9115203
  19. 19. Wolman MA, de Groh ED, McBride SM, Jongens TA, Granato M, Epstein JA. Modulation of cAMP and ras signaling pathways improves distinct behavioral deficits in a zebrafish model of neurofibromatosis type 1. Cell Rep. 2014;8(5):1265–70. pmid:25176649
  20. 20. Buchanan ME, Davis RL. A distinct set of Drosophila brain neurons required for neurofibromatosis type 1-dependent learning and memory. J Neurosci. 2010;30(30):10135–43. pmid:20668197
  21. 21. Gross AM, Wolters PL, Dombi E, Baldwin A, Whitcomb P, Fisher MJ, et al. Selumetinib in Children with Inoperable Plexiform Neurofibromas. N Engl J Med. 2020;382(15):1430–42. pmid:32187457
  22. 22. Moertel CL, Hirbe AC, Shuhaiber HH, Bielamowicz K, Sidhu A, Viskochil D, et al. ReNeu: A Pivotal, Phase IIb Trial of Mirdametinib in Adults and Children With Symptomatic Neurofibromatosis Type 1-Associated Plexiform Neurofibroma. J Clin Oncol. 2025;43(6):716–29. pmid:39514826
  23. 23. Abdel-Rahman O, ElHalawani H, Ahmed H. Risk of Selected Cardiovascular Toxicities in Patients With Cancer Treated With MEK Inhibitors: A Comparative Systematic Review and Meta-Analysis. J Glob Oncol. 2015;1(2):73–82. pmid:28804776
  24. 24. Avery RA, Trimboli-Heidler C, Kilburn LB. Separation of outer retinal layers secondary to selumetinib. J AAPOS. 2016;20(3):268–71. pmid:27108842
  25. 25. Baldo F, Magnolato A, Barbi E, Bruno I. Selumetinib side effects in children treated for plexiform neurofibromas: first case reports of peripheral edema and hair color change. BMC Pediatr. 2021;21(1):67. pmid:33549085
  26. 26. Chisholm AK, Haebich KM, Pride NA, Walsh KS, Lami F, Ure A, et al. Delineating the autistic phenotype in children with neurofibromatosis type 1. Mol Autism. 2022;13(1):3. pmid:34983638
  27. 27. Gutmann DH, Anastasaki C, Gupta A, Hou Y, Morris SM, Payne JM, et al. Cognition and behavior in neurofibromatosis type 1: report and perspective from the Cognition and Behavior in NF1 (CABIN) Task Force. Genes Dev. 2025;39(9–10):541–54. pmid:40127956
  28. 28. Robinson JE, Coughlin GM, Hori AM, Cho JR, Mackey ED, Turan Z, et al. Optical dopamine monitoring with dLight1 reveals mesolimbic phenotypes in a mouse model of neurofibromatosis type 1. eLife. 2019;8. pmid:31545171
  29. 29. Brown JA, Emnett RJ, White CR, Yuede CM, Conyers SB, O’Malley KL, et al. Reduced striatal dopamine underlies the attention system dysfunction in neurofibromatosis-1 mutant mice. Hum Mol Genet. 2010;19(22):4515–28. pmid:20826448
  30. 30. Sullivan K, El-Hoss J, Quinlan KGR, Deo N, Garton F, Seto JTC, et al. NF1 is a critical regulator of muscle development and metabolism. Hum Mol Genet. 2014;23(5):1250–9. pmid:24163128
  31. 31. Masgras I, Ciscato F, Brunati AM, Tibaldi E, Indraccolo S, Curtarello M, et al. Absence of Neurofibromin Induces an Oncogenic Metabolic Switch via Mitochondrial ERK-Mediated Phosphorylation of the Chaperone TRAP1. Cell Rep. 2017;18(3):659–72. pmid:28099845
  32. 32. de Souza MLR, Jansen AK, Rodrigues LOC, Vilela DLS, Kakehasi AM, Martins AS, et al. Increased resting metabolism in neurofibromatosis type 1. Clin Nutr ESPEN. 2019;32:44–9. pmid:31221289
  33. 33. Apostolova I, Derlin T, Salamon J, Amthauer H, Granström S, Brenner W, et al. Cerebral glucose metabolism in adults with neurofibromatosis type 1. Brain Res. 2015;1625:97–101. pmid:26335059
  34. 34. Martins AS, Jansen AK, Rodrigues LOC, Matos CM, Souza MLR, de Souza JF, et al. Lower fasting blood glucose in neurofibromatosis type 1. Endocr Connect. 2016;5(1):28–33. pmid:26631381
  35. 35. Souza M, Jansen A, Martins A, Rodrigues L, Rezende N. Body composition in adults with neurofibromatosis type 1. Rev Assoc Med Bras (1992). 2016;62(9):831–6. pmid:28001256
  36. 36. Souza JF, Passos RLF, Guedes ACM, Rezende NA, Rodrigues LOC. Muscular force is reduced in neurofibromatosis type 1. J Musculoskelet Neuronal Interact. 2009;9(1):15–7. pmid:19240363
  37. 37. Tritz R, Benson T, Harris V, Hudson FZ, Mintz J, Zhang H, et al. Nf1 heterozygous mice recapitulate the anthropometric and metabolic features of human neurofibromatosis type 1. Transl Res. 2021;228:52–63. pmid:32781282
  38. 38. Wei X, Franke J, Ost M, Wardelmann K, Börno S, Timmermann B, et al. Cell autonomous requirement of neurofibromin (Nf1) for postnatal muscle hypertrophic growth and metabolic homeostasis. J Cachexia Sarcopenia Muscle. 2020;11(6):1758–78. pmid:33078583
  39. 39. Kossler N, Stricker S, Rödelsperger C, Robinson PN, Kim J, Dietrich C, et al. Neurofibromin (Nf1) is required for skeletal muscle development. Hum Mol Genet. 2011;20(14):2697–709. pmid:21478499
  40. 40. Maurer GW, Malita A, Nagy S, Koyama T, Werge TM, Halberg KA, et al. Analysis of genes within the schizophrenia-linked 22q11.2 deletion identifies interaction of night owl/LZTR1 and NF1 in GABAergic sleep control. PLoS Genet. 2020;16(4):e1008727. pmid:32339168
  41. 41. Tong JJ, Schriner SE, McCleary D, Day BJ, Wallace DC. Life extension through neurofibromin mitochondrial regulation and antioxidant therapy for neurofibromatosis-1 in Drosophila melanogaster. Nat Genet. 2007;39(4):476–85. pmid:17369827
  42. 42. van der Voet M, Harich B, Franke B, Schenck A. ADHD-associated dopamine transporter, latrophilin and neurofibromin share a dopamine-related locomotor signature in Drosophila. Mol Psychiatry. 2016;21(4):565–73. pmid:25962619
  43. 43. King LB, Boto T, Botero V, Aviles AM, Jomsky BM, Joseph C, et al. Developmental loss of neurofibromin across distributed neuronal circuits drives excessive grooming in Drosophila. PLoS Genet. 2020;16(7):e1008920. pmid:32697780
  44. 44. King LB, Koch M, Murphy KR, Velazquez Y, Ja WW, Tomchik SM. Neurofibromin Loss of Function Drives Excessive Grooming in Drosophila. G3 (Bethesda). 2016;6(4):1083–93. pmid:26896440
  45. 45. Suarez GO, Kumar DS, Brunner H, Knauss A, Barrios J, Emel J, et al. Neurofibromin Deficiency Alters the Patterning and Prioritization of Motor Behaviors in a State-Dependent Manner. J Neurosci. 2025;45(16):e1531242025. pmid:39965929
  46. 46. Brown EB, Zhang J, Lloyd E, Lanzon E, Botero V, Tomchik S, et al. Neurofibromin 1 mediates sleep depth in Drosophila. PLoS Genet. 2023;19(12):e1011049. pmid:38091360
  47. 47. Brown EB, Zhang J, Lloyd E, Lanzon E, Botero V, Tomchik S, et al. Neurofibromin 1 mediates sleep depth in Drosophila. openRxiv. 2022.
  48. 48. Yatsenko AS, Marrone AK, Kucherenko MM, Shcherbata HR. Measurement of metabolic rate in Drosophila using respirometry. J Vis Exp. 2014;(88):e51681. pmid:24998593
  49. 49. Dietzl G, Chen D, Schnorrer F, Su K-C, Barinova Y, Fellner M, et al. A genome-wide transgenic RNAi library for conditional gene inactivation in Drosophila. Nature. 2007;448(7150):151–6. pmid:17625558
  50. 50. Saxton RA, Sabatini DM. mTOR Signaling in Growth, Metabolism, and Disease. Cell. 2017;169(2):361–71. pmid:28388417
  51. 51. Johannessen CM, Johnson BW, Williams SMG, Chan AW, Reczek EE, Lynch RC, et al. TORC1 is essential for NF1-associated malignancies. Curr Biol. 2008;18(1):56–62. pmid:18164202
  52. 52. Ullrich NJ, Prabhu SP, Reddy AT, Fisher MJ, Packer R, Goldman S, et al. A phase II study of continuous oral mTOR inhibitor everolimus for recurrent, radiographic-progressive neurofibromatosis type 1-associated pediatric low-grade glioma: a Neurofibromatosis Clinical Trials Consortium study. Neuro Oncol. 2020;22(10):1527–35. pmid:32236425
  53. 53. Kaul A, Toonen JA, Cimino PJ, Gianino SM, Gutmann DH. Akt- or MEK-mediated mTOR inhibition suppresses Nf1 optic glioma growth. Neuro Oncol. 2015;17(6):843–53. pmid:25534823
  54. 54. Scheidenhelm DK, Cresswell J, Haipek CA, Fleming TP, Mercer RW, Gutmann DH. Akt-dependent cell size regulation by the adhesion molecule on glia occurs independently of phosphatidylinositol 3-kinase and Rheb signaling. Mol Cell Biol. 2005;25(8):3151–62. pmid:15798201
  55. 55. Vanhaesebroeck B, Guillermet-Guibert J, Graupera M, Bilanges B. The emerging mechanisms of isoform-specific PI3K signalling. Nat Rev Mol Cell Biol. 2010;11(5):329–41. pmid:20379207
  56. 56. Chowański S, Walkowiak-Nowicka K, Winkiel M, Marciniak P, Urbański A, Pacholska-Bogalska J. Insulin-Like Peptides and Cross-Talk With Other Factors in the Regulation of Insect Metabolism. Front Physiol. 2021;12:701203. pmid:34267679
  57. 57. He L, Cho S, Blenis J. mTORC1, the maestro of cell metabolism and growth. Genes Dev. 2025;39(1–2):109–31. pmid:39572234
  58. 58. Panwar V, Singh A, Bhatt M, Tonk RK, Azizov S, Raza AS, et al. Multifaceted role of mTOR (mammalian target of rapamycin) signaling pathway in human health and disease. Signal Transduct Target Ther. 2023;8(1):375. pmid:37779156
  59. 59. Guo HF, The I, Hannan F, Bernards A, Zhong Y. Requirement of Drosophila NF1 for activation of adenylyl cyclase by PACAP38-like neuropeptides. Science. 1997;276(5313):795–8. pmid:9115204
  60. 60. Guo HF, Tong J, Hannan F, Luo L, Zhong Y. A neurofibromatosis-1-regulated pathway is required for learning in Drosophila. Nature. 2000;403(6772):895–8. pmid:10706287
  61. 61. Walker JA, Tchoudakova AV, McKenney PT, Brill S, Wu D, Cowley GS, et al. Reduced growth of Drosophila neurofibromatosis 1 mutants reflects a non-cell-autonomous requirement for GTPase-Activating Protein activity in larval neurons. Genes Dev. 2006;20(23):3311–23. pmid:17114577
  62. 62. Machado Almeida P, Lago Solis B, Stickley L, Feidler A, Nagoshi E. Neurofibromin 1 in mushroom body neurons mediates circadian wake drive through activating cAMP-PKA signaling. Nat Commun. 2021;12(1):5758. pmid:34599173
  63. 63. Levin LR, Han PL, Hwang PM, Feinstein PG, Davis RL, Reed RR. The Drosophila learning and memory gene rutabaga encodes a Ca2+/Calmodulin-responsive adenylyl cyclase. Cell. 1992;68(3):479–89. pmid:1739965
  64. 64. Dickerson BH, de Souza AM, Huda A, Dickinson MH. Flies Regulate Wing Motion via Active Control of a Dual-Function Gyroscope. Curr Biol. 2019;29(20):3517-3524.e3. pmid:31607538
  65. 65. Lim SY, You H, Lee J, Lee J, Lee Y, Lee K-A, et al. Identification and characterization of GAL4 drivers that mark distinct cell types and regions in the Drosophila adult gut. J Neurogenet. 2021;35(1):33–44. pmid:33326321
  66. 66. Musselman LP, Kühnlein RP. Drosophila as a model to study obesity and metabolic disease. J Exp Biol. 2018;221(Pt Suppl 1):jeb163881. pmid:29514880
  67. 67. Grönke S, Beller M, Fellert S, Ramakrishnan H, Jäckle H, Kühnlein RP. Control of fat storage by a Drosophila PAT domain protein. Curr Biol. 2003;13(7):603–6. pmid:12676093
  68. 68. Huang K, Liu Y, Perrimon N. Roles of Insect Oenocytes in Physiology and Their Relevance to Human Metabolic Diseases. Front Insect Sci. 2022;2:859847. pmid:38468774
  69. 69. Kim D-H, Shin M, Jung S-H, Kim Y-J, Jones WD. A fat-derived metabolite regulates a peptidergic feeding circuit in Drosophila. PLoS Biol. 2017;15(3):e2000532. pmid:28350856
  70. 70. Li J, Huang K, Dibra I, Liu Y, Perrimon N, Simons M. Desaturase-dependent secretory functions of hepatocyte-like cells control systemic lipid metabolism during starvation in Drosophila. Res Sq. 2024;:rs.3.rs-5566817. pmid:39711529
  71. 71. Takemura S y, Hayworth KJ, Huang GB, Januszewski M, Lu Z, Marin EC. A connectome of the male Drosophila ventral nerve cord. eLife Sciences Publications, Ltd. 2024.
  72. 72. Summers MA, Quinlan KG, Payne JM, Little DG, North KN, Schindeler A. Skeletal muscle and motor deficits in Neurofibromatosis Type 1. J Musculoskelet Neuronal Interact. 2015;15(2):161–70. pmid:26032208
  73. 73. Chatterjee N, Perrimon N. What fuels the fly: Energy metabolism in Drosophila and its application to the study of obesity and diabetes. Sci Adv. 2021;7(24):eabg4336. pmid:34108216
  74. 74. Wegener G. Flying insects: model systems in exercise physiology. Experientia. 1996;52(5):404–12. pmid:8641375
  75. 75. Gutman DH, Andersen LB, Cole JL, Swaroop M, Collins FS. An alternatively-spliced mRNA in the carboxy terminus of the neurofibromatosis type 1 (NF1) gene is expressed in muscle. Hum Mol Genet. 1993;2(7):989–92. pmid:8364582
  76. 76. Zappia MP, Frolov MV. E2F function in muscle growth is necessary and sufficient for viability in Drosophila. Nat Commun. 2016;7:10509. pmid:26823289
  77. 77. Manseau L, Baradaran A, Brower D, Budhu A, Elefant F, Phan H, et al. GAL4 enhancer traps expressed in the embryo, larval brain, imaginal discs, and ovary of Drosophila. Dev Dyn. 1997;209(3):310–22. pmid:9215645
  78. 78. Barwell T, DeVeale B, Poirier L, Zheng J, Seroude F, Seroude L. Regulating the UAS/GAL4 system in adult Drosophila with Tet-off GAL80 transgenes. PeerJ. 2017;5:e4167. pmid:29259847
  79. 79. Weaver LN, Ma T, Drummond-Barbosa D. Analysis of Gal4 Expression Patterns in Adult Drosophila Females. G3 (Bethesda). 2020;10(11):4147–58. pmid:32917721
  80. 80. Marmion RA, Simpkins AG, Barrett LA, Denberg DW, Zusman S, Schottenfeld-Roames J, et al. Stochastic phenotypes in RAS-dependent developmental diseases. Curr Biol. 2023;33(5):807-816.e4. pmid:36706752
  81. 81. Mendoza MC, Er EE, Blenis J. The Ras-ERK and PI3K-mTOR pathways: cross-talk and compensation. Trends Biochem Sci. 2011;36(6):320–8. pmid:21531565
  82. 82. Wang D, Eisen HJ. Mechanistic Target of Rapamycin (mTOR) Inhibitors. Handb Exp Pharmacol. 2022;272:53–72. pmid:35091825
  83. 83. Han J, Liu Y, Yang S, Wu X, Li H, Wang Q. MEK inhibitors for the treatment of non-small cell lung cancer. J Hematol Oncol. 2021;14(1):1. pmid:33402199
  84. 84. Mullard A. FDA approves first-in-class AKT inhibitor. Nat Rev Drug Discov. 2024;23(1):9. pmid:38049466
  85. 85. Sofela FA, Lopez Valencia M, Jongens TA, Sehgal A. Effects of Nf1 on sleep behavior are mediated through starvation caused by deficits in SARM1 dependent NAD+ metabolism. openRxiv. 2024.
  86. 86. Bai L, Lee Y, Hsu CT, Williams JA, Cavanaugh D, Zheng X, et al. A Conserved Circadian Function for the Neurofibromatosis 1 Gene. Cell Rep. 2018;22(13):3416–26. pmid:29590612
  87. 87. Williams JA, Su HS, Bernards A, Field J, Sehgal A. A circadian output in Drosophila mediated by neurofibromatosis-1 and Ras/MAPK. Science. 2001;293(5538):2251–6. pmid:11567138
  88. 88. Bai L, Sehgal A. Anaplastic Lymphoma Kinase Acts in the Drosophila Mushroom Body to Negatively Regulate Sleep. PLoS Genet. 2015;11(11):e1005611. pmid:26536237
  89. 89. Hannan F, Ho I, Tong JJ, Zhu Y, Nurnberg P, Zhong Y. Effect of neurofibromatosis type I mutations on a novel pathway for adenylyl cyclase activation requiring neurofibromin and Ras. Hum Mol Genet. 2006;15(7):1087–98. pmid:16513807
  90. 90. Gouzi JY, Moressis A, Walker JA, Apostolopoulou AA, Palmer RH, Bernards A, et al. The receptor tyrosine kinase Alk controls neurofibromin functions in Drosophila growth and learning. PLoS Genet. 2011;7(9):e1002281. pmid:21949657
  91. 91. Georganta E-M, Moressis A, Skoulakis EMC. Associative Learning Requires Neurofibromin to Modulate GABAergic Inputs to Drosophila Mushroom Bodies. J Neurosci. 2021;41(24):5274–86. pmid:33972401
  92. 92. Moscato EH, Dubowy C, Walker JA, Kayser MS. Social Behavioral Deficits with Loss of Neurofibromin Emerge from Peripheral Chemosensory Neuron Dysfunction. Cell Rep. 2020;32(1):107856. pmid:32640222
  93. 93. Dyson A, Ryan M, Garg S, Evans DG, Baines RA. Loss of NF1 in Drosophila Larvae Causes Tactile Hypersensitivity and Impaired Synaptic Transmission at the Neuromuscular Junction. J Neurosci. 2022;42(50):9450–72. pmid:36344265
  94. 94. Lee G, Chung J. Discrete functions of rictor and raptor in cell growth regulation in Drosophila. Biochem Biophys Res Commun. 2007;357(4):1154–9. pmid:17462592
  95. 95. Watson AL, Anderson LK, Greeley AD, Keng VW, Rahrmann EP, Halfond AL, et al. Co-targeting the MAPK and PI3K/AKT/mTOR pathways in two genetically engineered mouse models of schwann cell tumors reduces tumor grade and multiplicity. Oncotarget. 2014;5(6):1502–14. pmid:24681606
  96. 96. Parra V, Verdejo HE, Iglewski M, Del Campo A, Troncoso R, Jones D, et al. Insulin stimulates mitochondrial fusion and function in cardiomyocytes via the Akt-mTOR-NFκB-Opa-1 signaling pathway. Diabetes. 2014;63(1):75–88. pmid:24009260
  97. 97. Dasgupta B, Yi Y, Chen DY, Weber JD, Gutmann DH. Proteomic analysis reveals hyperactivation of the mammalian target of rapamycin pathway in neurofibromatosis 1-associated human and mouse brain tumors. Cancer Res. 2005;65(7):2755–60. pmid:15805275
  98. 98. Johannessen CM, Reczek EE, James MF, Brems H, Legius E, Cichowski K. The NF1 tumor suppressor critically regulates TSC2 and mTOR. Proc Natl Acad Sci U S A. 2005;102(24):8573–8. pmid:15937108
  99. 99. Anastasaki C, Woo AS, Messiaen LM, Gutmann DH. Elucidating the impact of neurofibromatosis-1 germline mutations on neurofibromin function and dopamine-based learning. Hum Mol Genet. 2015;24(12):3518–28. pmid:25788518
  100. 100. Cui Y, Costa RM, Murphy GG, Elgersma Y, Zhu Y, Gutmann DH, et al. Neurofibromin regulation of ERK signaling modulates GABA release and learning. Cell. 2008;135(3):549–60. pmid:18984165
  101. 101. Kikuchi A, Demo SD, Ye ZH, Chen YW, Williams LT. ralGDS family members interact with the effector loop of ras p21. Mol Cell Biol. 1994;14(11):7483–91. pmid:7935463
  102. 102. Bodempudi V, Yamoutpoor F, Pan W, Dudek AZ, Esfandyari T, Piedra M, et al. Ral overactivation in malignant peripheral nerve sheath tumors. Mol Cell Biol. 2009;29(14):3964–74. pmid:19414599
  103. 103. Danielsen SA, Lind GE, Kolberg M, Høland M, Bjerkehagen B, Sundby Hall K, et al. Methylated RASSF1A in malignant peripheral nerve sheath tumors identifies neurofibromatosis type 1 patients with inferior prognosis. Neuro Oncol. 2015;17(1):63–9. pmid:25038505
  104. 104. Mund JA, Park S, Smith AE, He Y, Jiang L, Hawley E, et al. Genetic disruption of the small GTPase RAC1 prevents plexiform neurofibroma formation in mice with neurofibromatosis type 1. J Biol Chem. 2020;295(29):9948–58. pmid:32471868
  105. 105. Kohlmeyer JL, Lingo JJ, Kaemmer CA, Scherer A, Warrier A, Voigt E, et al. CDK4/6-MEK Inhibition in MPNSTs Causes Plasma Cell Infiltration, Sensitization to PD-L1 Blockade, and Tumor Regression. Clin Cancer Res. 2023;29(17):3484–97. pmid:37410426
  106. 106. Ortega-Bertran S, Fernández-Rodríguez J, Magallón-Lorenz M, Zhang X, Creus-Bachiller E, Diazgranados AP, et al. Triple Combination of MEK, BET, and CDK Inhibitors Significantly Reduces Human Malignant Peripheral Nerve Sheath Tumors in Mouse Models. Clin Cancer Res. 2025;31(5):907–20. pmid:39786423
  107. 107. Anastasaki C, Gutmann DH. Neuronal NF1/RAS regulation of cyclic AMP requires atypical PKC activation. Hum Mol Genet. 2014;23(25):6712–21. pmid:25070947
  108. 108. Delandre C, McMullen JPD, Marshall OJ. Dynamic changes in neuronal and glial GAL4 driver expression during Drosophila aging. Genetics. 2025;229(3):iyaf014. pmid:39950543
  109. 109. Weaver LN, Ma T, Drummond-Barbosa D. Analysis of Gal4 Expression Patterns in Adult Drosophila Females. G3 (Bethesda). 2020;10(11):4147–58. pmid:32917721
  110. 110. Szebenyi AL. Cleaning behaviour in Drosophila melanogaster. Animal Behaviour. 1969;17(4):641–51.
  111. 111. Jenett A, Rubin GM, Ngo T-TB, Shepherd D, Murphy C, Dionne H, et al. A GAL4-driver line resource for Drosophila neurobiology. Cell Rep. 2012;2(4):991–1001. pmid:23063364
  112. 112. Wang X, Davis RL. Early Mitochondrial Fragmentation and Dysfunction in a Drosophila Model for Alzheimer’s Disease. Mol Neurobiol. 2021;58(1):143–55. pmid:32909149