Drosophila tweety Facilitates Autophagy to Regulate Mitochondrial Homeostasis and Bioenergetics in Glia

Abstract

Mitochondria support the energetic demands of the cells. Autophagic turnover of mitochondria serves as a critical pathway for mitochondrial homeostasis. It is unclear how bioenergetics and autophagy are functionally connected. Here, we identify an endolysosomal membrane protein that facilitates autophagy to regulate ATP production in glia. We determined that Drosophila tweety (tty) is highly expressed in glia and localized to endolysosomes. Diminished fusion between autophagosomes and endolysosomes in tty-deficient glia was rescued by expressing the human Tweety Homolog 1 (TTYH1). Loss of tty in glia attenuated mitochondrial turnover, elevated mitochondrial oxidative stress, and impaired locomotor functions. The cellular and organismal defects were partially reversed by antioxidant treatment. We performed live-cell imaging of genetically encoded metabolite sensors to determine the impact of tty and autophagy deficiencies on glial bioenergetics. We found that tty-deficient glia exhibited reduced mitochondrial pyruvate consumption accompanied by a shift towards glycolysis for ATP production. Likewise, genetic inhibition of autophagy in glia resulted in a similar glycolytic shift in bioenergetics. Furthermore, the survival of mutant flies became more sensitive to starvation, underlining the significance of tty in the crosstalk between autophagy and bioenergetics. Together, our findings uncover the role for tty in mitochondrial homeostasis via facilitating autophagy, which determines bioenergetic balance in glia.

Graphical Abstract

Introduction

The Tweety homologs (TTYHs) are evolutionarily conserved membrane proteins in multicellular eukaryotes. The founding member was identified in the flightless genomic locus of Drosophila melanogaster and was named tweety (tty) [1]. Humans possess three paralogs, namely TTYH1, TTYH2, and TTYH3. TTYH2 and TTYH3 are broadly expressed across different cell types, whereas TTYH1 is predominantly expressed in brain and testes [2–7]. The two Drosophila orthologs, tty and CG3638, share higher homology with TTYH1 than with TTYH2/3 [8]. According to modENCODE RNA-sequencing database, tty expression is more restricted to brain compared to the ubiquitously expressed CG3638 [3].

Differential expressions of mammalian TTYHs have been associated with pathological conditions. Elevated expression levels of TTYHs were reported in multiple cancer types [9–13]. In rat brain, epileptic events upregulated Ttyh1 expression in astrocytes [2]. Indeed, single-cell transcriptomic analyses of mouse and human brains revealed that TTYH1 is abundantly expressed in astrocytes but not in neurons [14, 15]. Notch signaling in mouse neural stem cells, the precursor of neurons and astrocytes, was found to require Ttyh1 expression [16, 17]. Genomic deletion of Ttyh1 in mouse was initially found to be embryonic lethal [18], though was reported to be viable in a later study [17]. Notably, reduced TTYH1 expression levels were found in brain samples from Alzheimer’s disease and Parkinson’s disease patient [19, 20], implicating its potential role in neurodegeneration pathogenesis.

Although TTYH1 and TTYH2 are designated as understudied proteins by the National Institute of Health [21], the subcellular localization and molecular function of TTYHs remain inconclusive. TTYHs were detected at the plasma membrane when ectopically expressed in mammalian cell lines [22–24]. Endogenous Ttyh1 protein was localized to intracellular vesicular structures in glia and membrane protrusion in glioma [2, 10]. In addition to the undefined primary site of function, the molecular function of TTYH proteins also remains ambiguous. An early study found that both Drosophila and mammalian TTYHs form chloride-conducting ion channels when overexpressed in mammalian cells [23, 25]. Subsequent investigations proposed that TTYHs form volume-regulated anion channels in brain cells [24, 26, 27]. An arginine residue was identified as critical to channel activity [26], although we found that it is only conserved among TTYHs of tetrapods (data not shown). Recently, the chloride channel proposal was contested by two reports on the high-resolution structures of TTYHs [28, 29]. TTYHs were found to lack anion conductance and ion channel structural features [28, 29]. The arginine residue proposed to confer channel activity is located far away from the transmembrane region [28, 29]. Interestingly, both studies revealed a highly conserved hydrophobic pocket located at the membrane-solution interface [28, 29]. It is unknown, however, whether such structural feature is associated with membrane trafficking function.

We reasoned that a genetically tractable model will facilitate the illumination of TTYHs. Given that both TTYH1 and Drosophila tty show similar brain-enriched expression patterns, we utilized the Drosophila model and its versatile genetic tools to interrogate the biological function of tty in brain cells. Our study identifies endolysosomes as the primary site of function for tty and TTYH1 in glia. Using a series of genetically encoded fluorescent sensors expressed in primary glia, we define tty’s roles in facilitating autophagy, mitochondrial homeostasis, and mitochondrial metabolism. Interestingly, our data indicate a functional coupling between tty deficiency and energy production. We measured cellular ATP levels and found that glia exhibited a shift towards glycolytic energy production upon tty or autophagy deficiency. Our findings thus identify the endolysosomal function of tty in mediating the crosstalk between autophagy and bioenergetics in glia.

Methods

Fly strains and husbandry.

Flies were grown at 25°C with standard fly food (3 L of food contained 25.6 g agar, 80.6 g brewer’s yeast, 190.6 g of cornmeal, 40.3 g of sugar, 16.6 g of propionic acid, and 13.3 g of tegosept). For some climbing and locomotor activity assays, chemical feeding was performed using Formula 4–24 instant food (Carolina Biological Supply) infused with DMSO, paraquat, rotenone, or AD4. The following fly strains were obtained from Bloomington Drosophila Stock Centre (stock number): tty^MI12269 (57910), tty^T2A-Gal4 (93947), Repo-GAL4 (7415), Act5C-GAL4 (4414), UAS-nls-mCherry (38424), UAS-hPABP[30] (9419), UAS-GFP-LAMP1 (42714), UAS-EYFP-mito (7194), UAS-mCherry-Atg8a (37750), UAS-GFP-mCherry-Atg8a (37749), UAS-MitoTimer (57323), UAS-mito-roGFP2-Grx1 (67664), tub-mito-roGFP2-Grx1 (67669), UAS-mito-PyronicSF (94536), UAS-Atg5^RNAi (34899), UAS-Atg8a^RNAi (58309), and UAS-Atg1 (60734). UAS-tty^RNAi (KK101659) was obtained from Vienna Drosophila Resource Centre. The UAS-PercevalHR strain has been described previously [31, 32].

Generation of transgenic fly strains.

Two genomic-rescue strains, GR1 and GR2, that contain genomic fragments encompassing the tty locus were generated for the study. GR1 carries a 21-kb BAC clone (CH322–154M13) docked to VK00027 on the 3^rd chromosome. GR2 (Dp(1;3)DC379) was generated by the Gene Disruption Project [33], and carries a duplicated X-chromosome fragment docked to VK00033 on the 3^rd chromosome. UAS-HA-tty-FLAG was generated as follows: Forward primer containing the hemagglutinin tag coding sequence and reverse primer containing the FLAG tag coding sequence were used to PCR-amplify the tty coding sequence of the longest transcript variant using the BAC clone (CH322–154M13) as template. The amplicon was then subcloned to the notI site of pUAS-attB vector. The resulting construct, pUAS-HA-tty-FLAG-attB, was docked to VK00027 on the 3^rd chromosome to create the transgenic strain. UAS-TTYH1-RFP was generated as follows: cDNA of C-terminally RFP-tagged human TTYH1 (NM_001005367.2) was synthesized by Cyagen. The cDNA was subcloned to the xhoI site of pUAS-attB vector. The resulting construct, pUAS-TTYH1-RFP-attB, was docked to VK00003a on the 2^nd chromosome to create the transgenic strain.

Chemicals and antibodies.

AD4 (HY-110256) and UK-5099 (HY-15475) were obtained from Med Chem Express. RNAsin (N2615) was obtained from Promega. 2-deoxy-D-glucose (25972), Concanavalin A (C2010), D-Mannitol (M4125), insulin (I6634), L-cysteine (C7352), Protease inhibitor (P8340), RNAzol RT (R4533), Schneider’s medium (S0146), and SYBR green JumpStart Taq Ready-Mix (S4438), were obtained from Sigma-Aldrich. Dynabeads Protein G (10003D), Fetal Bovine Serum (10082147), and High-capacity cDNA reverse transcription kit (4368814) were obtained from Thermo-Fisher. Papain (LS003124) was obtained from Worthington.

Mouse monoclonal anti-FLAG antibody (8146) and rabbit monoclonal anti-HA antibody (3724) were obtained from Cell Signalling. Mouse monoclonal anti-GFP antibody (N86/38), mouse monoclonal anti-Rab7 antibody (Rab7), mouse monoclonal anti-Repo (8D12), and mouse monoclonal anti-α-tubulin (12G10) were obtained from Developmental Studies Hybridoma Bank. Mouse monoclonal anti-insect Cathepsin L (MAB22591) was obtained from R&D Systems. Rabbit polyclonal anti-RFP antibody (600-401-379-RTU) was obtained from Rockland. Donkey anti-Mouse IgG Secondary Antibody Alexa Fluor 488 (A21202), donkey anti-Mouse IgG Secondary Antibody Alexa Fluor 555 (A31570), donkey anti-Mouse IgG Secondary Antibody Alexa Fluor 647 (A31571), donkey anti-Rabbit IgG Secondary Antibody Alexa Fluor 488 (A21206), donkey anti-Rabbit IgG Secondary Antibody Alexa Fluor 555 (A31572), and mouse monoclonal anti-GFP antibody (MA5–15256) were obtained from Thermo-Fisher. Mouse monoclonal anti-HA antibody (H3663) was obtained from Sigma-Aldrich.

Drosophila activity assay.

Assays were performed in an incubator set at 25°C and 12-hour light-dark cycle. Locomotor activity was recorded using a 32-sample Drosophila Activity Monitor (DAM2; TriKinetics). A single 1-week-old fly was enclosed in a hollow 65 mm glass rod plugged with agar (5% sucrose and 2% agar) at one end and cotton at the other end. Vials were bisected by a transverse infra-red (IR) beam, and the number of times each fly crossed the IR beam per minute was reported. Flies were allowed to acclimatize in the monitor for at least half a day before data collection for a 24-hour period.

Drosophila climbing assay.

A cohort of 10 to 20 1-week-old flies were enclosed in a transparent 50-mL measuring cylinder (08–550D, Fisher Scientific). The cylinder was marked with 11 vertical levels of approximately 12-mm height each. Movement of flies was recorded by a cell phone camera as video file. The cylinder was tapped 3 times on a sponge pad to send all the flies to the bottom. After standing the cylinder still for about 20 seconds, the cylinder was tapped again. The cycle repeated for 1 additional time to constitute 3 trials for each climbing assay. To analyse the recorded video, playback was frozen at the 5-second timepoint after cylinder tapping. Distribution of flies within the 11 vertical levels was counted. Number of flies reaching each level was averaged from the 3 trials to give the mean distribution. Median climbed height was calculated from the mean distribution for each individual assay. Each cohort of flies was assayed once for a median climbed height. At least 4 independent cohorts were analysed for each experimental condition (genotype and treatment).

Survival analysis.

One-to-three-days-old adult flies were collected in vials (≤20 flies per vial) containing 3% agar dissolved in water. At least 4 vials of flies were set up for each experimental condition. Fly vials were kept in an incubator set at 25°C and 12-hour light-dark cycle. Dead flies in each vial were counted every day until all flies were dead.

Glial mRNA tagging.

For each biological replicate, the heads of 200 flies expressing UAS-hPABP driven by repo-GAL4 were fixed in 1 mL of PBS containing 1% formaldehyde and 0.5% Triton X-100 for 30 minutes at 4°C [30]. Fixation was blocked by adding 140 μL of 2M glycine and washing with PBS. Samples were then homogenised in 800 μL of DEPC-treated extraction buffer containing 150 mM NaCl, 50 mM HEPES buffer at pH 7.6, 1 mM EGTA, 15 mM EDTA, 10% glycerol, 10U/mL RNAsin (Promega) and 1x Protease inhibitor (Sigma-Aldrich). Lysates were centrifuged at 13,000 g for 10 min at 4 °C, where the collected supernatants were incubated with 100 μL Dynabeads Protein G (Invitrogen) coated with mouse monoclonal anti-FLAG antibody (9A3; Cell Signalling) overnight at 4 °C with a rotatory shaker. The beads were washed 4 times with the extraction buffer, before incubated with 100 μL of elution buffer (50 mM Tris–HCl at pH 7.0, 10 mM EDTA, 1.3% SDS and 10U/mL RNAsin) at 65 °C for 30 minutes to release the mRNA. RNAzol (Sigma-Aldrich) was applied to purify the sample according to manufacturer’s instructions.

Real-time quantitative PCR.

For whole-fly RNA extraction, 10 adult flies (1-week-old) were homogenized in 600 μL of RNAzol (Sigma-Aldrich). Reverse transcription of total RNA to cDNA was performed using the High-Capacity cDNA Reverse Transcription kit (Life-Tech), followed by real-time qPCR using SYBR Green JumpStart Taq ReadyMix (Sigma-Aldrich). The tweety mRNA levels relative to rp49 mRNA was calculated by the Delta Delta C_T (ΔΔC_T) method using CFX Maestro (Bio-Rad) software. The primers used are as follow:

rp49 (housekeeping)

Forward: 5’-AGCATACAGGCCCAAGATCG-3’

Reverse: 5’-TGTTGTCGATACCCTTGGGC-3’

tweety

Forward: 5’-GTCAGCAGGCCCACC-3’

Reverse: 5’-TTGGACTCGTTGAGCGG-3’

Primary cell culture of Drosophila glia.

The procedures for dissociation and culturing of primary glia from Drosophila has been previously described with slight modifications [31, 32]. In brief, wandering third-instar larvae were first sterilized by immersing in 70% ethanol and rinsing with sterile H₂O. Dissection was performed in filtered Schneider’s medium (Sigma-Aldrich) containing 10% FBS and 1% antibiotic/antimycotic solution (Sigma-Aldrich). Dissected brains were briefly washed with Schneider’s medium before transferring to filtered HL-3 solution (70 mM NaCl, 5 mM KCl, 1 mM CaCl₂, 20 mM MgCl₂, 10 mM NaHCO₃, 115 mM sucrose, 5 mM trehalose, and 5 mM HEPES) with the addition of 0.423 mM L-cysteine (Sigma-Aldrich) and 5 U/mL papain (Worthington). After 20 minutes of enzymatic digestion, the brains were washed twice with Schneider’s medium before dissociating the cells in Schneider’s medium with 50 μg/mL of insulin (Sigma-Aldrich). The cell suspension was then pipetted into 35 mm glass-bottom dishes (Cellvis) precoated with concanavalin A (Sigma-Aldrich). Primary Drosophila neurons and glial cells were co-cultured in a cell culture incubator at 25°C for 4 days prior to experiment. During the incubation period, the culture medium was replaced in a daily basis. For experiments, glia cells were identified based on cell morphology and glial-specific biosensors.

Live-cell imaging of primary glia expressing genetically encoded fluorescent reporters.

Time-lapse images of live primary cultured cells were acquired by a ZEISS Axio Observer microscope equipped with a Colibri 5 LED light source and a 90 HE LED multi-bandpass filter set and controlled by the ZEISS ZEN Blue software. The LED emits excitation wavelength/bandwidth of 385/30 nm, 469/38 nm, 555/30 nm, and 631/33 nm. Emission signals were recorded by a ZEISS Axiocam 705 camera.

GFP-mCherry-Atg8a.

Culture media was replaced with HL-3 solution. A 63x objective (Zeiss) was used for the measurements. Briefly, primary glia expressing GFP-mCherry-Atg8a driven by repo-GAL4 were sequentially excited at 469 nm and 555 nm to generate fluorescence emission [34]. The recorded signals were then analysed by CellProfiler (Broad Institute), where the mCherry signals were selected to identify punctate structures representing intracellular autophagosomes. GFP and mCherry signal intensities within each identified autophagosome were measured. The mean average mCherry/GFP ratio of all identified autophagosomes represents the rate of fusion between autophagosomes and acidic endolysosomes and was normalized to the mCherry/GFP ratios of wildtype glia.

MitoTimer.

Culture media was replaced with HL-3. A 63x objective (Zeiss) was used for the measurements. Briefly, primary glia expressing MitoTimer driven by repo-GAL4 were sequentially excited at 469nm and 555nm to generate fluorescence emissions Em_469ex and Em_555ex, respectively [35]. The recorded signals were then analysed by CellProfiler (Broad Institute), where the Em_469ex signals were selected to identify punctate structures representing mitochondria. Em_469ex (green) and Em_555ex (red) signal intensities within each identified mitochondrial structure were measured. The average Em_555ex/Em_469ex ratio of all identified mitochondrial structures was normalized to the Em_555ex/Em_469ex ratios of wildtype glia.

mCherry-Atg8a and EYFP-mito colocalization.

Culture media was replaced with HL-3. A 100x oil objective (Zeiss) was used for the measurements. Briefly, primary glia expressing mCherry-Atg8a and EYFP-mito driven by repo-GAL4 were sequentially excited at 469nm and 555nm to generate fluorescence emission. The recorded signals were then analysed by CellProfiler (Broad Institute), where the mCherry signals were selected to identify punctate structures representing individual autophagosomes. EYFP and mCherry signal intensities within each identified autophagosome were measured. A single autophagosome was considered mitochondria-containing when the mCherry/EYFP ratio was < 1 in amplitude. Number of mitochondria-containing autophagosomes and total number of identified autophagosomes for each cell were counted to calculate the percentage of mitochondria-containing autophagosomes in individual cells.

mito-roGFP2-Grx1.

Culture media was replaced with HL-3. A 63x objective (Zeiss) was used for the measurements. For glial measurements, repo-GAL4 was used to drive mito-roGFP2-Grx1 expression. For neuronal measurements, mito-roGFP2-Grx1 was driven by an upstream tubulin promoter (tub-mito-roGFP2-Grx1), and neurons were identified by their morphology and relatively stronger fluorescence emission. Mito-roGFP2-Grx1 was sequentially excited at 385 nm and 469 nm to generate fluorescence emissions Em_385ex and Em_469ex, respectively [36], where the signals were recorded consecutively for 10 minutes. Hydrogen peroxide (50 μM) was applied at the end of the experiment as a positive control to verify that the recorded signal represents the mitochondrial redox homeostasis. Data from the timepoint 1-minute before H₂O₂ addition was selected for comparison between genotypes. The recorded time-lapse images were analysed with Zen 3.3 (Zeiss), where all mitochondria were selected as the region of interest (ROI). Amplitudes of the Em_385ex/Em_469ex emission ratio reports the mitochondrial GSH/GSSG ratio [36], and hence, redox homeostasis between genotypes.

mito-PyronicSF.

Culture media was replaced with HL-3. A 100x oil objective (Zeiss) was used for the measurements. Briefly, mito-PyronicSF was sequentially excited at 385 nm and 469 nm to generate fluorescence emissions [37], where the signal was recorded for 10 minutes. Baseline was established 1 min before the addition of UK-5099 (10μM), while endpoint was established at the end of the experiment. The recorded signals were then analysed by CellProfiler (Broad Institute), where the signals from 469 nm excitation (Em_469ex) were selected to create object masks for identifying punctate structures representing intracellular mitochondria. The amplitude of the mean Em_469ex/Em_385ex ratio for all identified mitochondria indicates relative mitochondrial pyruvate levels upon normalization to values from wildtype glia. Representative traces were achieved by plotting the average Em_469ex/Em_385ex ratio of mitochondria in a single cell of each genotype at indicated time points.

PercevalHR.

Culture media was replaced with mannitol HL-3 (70 mM NaCl, 5 mM KCl, 1 mM CaCl₂, 20 mM MgCl₂, 10 mM NaHCO₃, 120 mM mannitol, and 5 mM HEPES). For AD4 treatment groups, cells were cultured in Schneider’s medium containing 40 ng/mL AD4 for 3 days before the experiment. A 63x objective (Zeiss) was used for the measurements. Briefly, PercevalHR was sequentially excited at 385 nm and 469 nm to generate fluorescence emissions Em_385ex and Em_469ex, respectively [38], where the signals were recorded for 15 minutes. Steady-states for all genotypes were established at 1-minute before the addition of 2-DG (10 mM), whereas endpoints were established at 10-minute after 2-DG addition. The recorded time-lapse images were analysed with Zen 3.3 (Zeiss), where whole cell bodies were selected as the region of interest (ROI). At least 22 cells from ≥3 independent culture dishes were analysed for each experimental condition. The Em_469ex/Em_385ex emission ratio represents the Perceval (ATP/ADP) ratio in each cell. The relative PercevalHR (ATP/ADP) ratio was calculated by normalizing the mean ratios of each genotype to their corresponding steady-state values. The percentage change in PercevalHR ratio was obtained by averaging the percent reduction of individual cells over the course of the experiment.

Immunofluorescence.

Cells or dissected tissues were fixed with 4% paraformaldehyde in PBS at pH 7.4 for 10 to 30 minutes at room temperature. Samples were then washed three times with ice-cold PBS. Permeabilization was done by incubating with 0.1% Trition X-100 in PBS, before washing with PBS for three times. Samples were incubated overnight at 4°C in primary antibodies dilution containing 1% BSA and 0.1% Trition X-100 in PBS. After primary antibody incubation, samples were washed three times with PBS and incubated with the fluorophore-conjugated secondary antibodies in 1% BSA in PBS for 1 hr at room temperature in the dark. Samples were immersed in Antifade Mounting Medium with DAPI (Vector) after three washes with PBS in the dark. Sample images were taken using confocal microscopy (Zeiss LSM 980 with Airyscan2 or Zeiss LSM 510 confocal microscopes; 63x objective with variable digital zoom and resolution) with appropriate laser power and filter selection. Representative images were processed and generated using Zen 3.3 and ImageJ.

Data presentation and statistical analyses.

Comparisons between two groups were performed using Student’s t-test, Mann-Whitney test, or log-rank test; Multiple comparisons were performed by analysis of variance (ANOVA). Excel (Microsoft) and Prism 8 (GraphPad) were used for statistical analyses. Statistical significance was defined as P < 0.05. P-values are shown as asterisks in the figure: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Boxplots show median, interquartile range (box), and data range (whiskers). Bar charts show mean average (bar) and standard error mean (whiskers). Data points on boxplots and bar charts are values from individual cells, individual flies, independent fly cohorts, or biological replicates. Numbers of independent experiments and analysed samples are listed in Supporting Information Table 1.

Results

Despite being the founding member, so far there have been no functional study on tty in Drosophila. According to the modENCODE high-throughput RNA-Seq expression data [3], tty expression is enriched in Drosophila central nervous system. Two independent alleles have been generated: tty^{CR60103-TG4.0} (tty^TG4) and tty^MI12269 (tty¹). tty^TG4 allele contains an artificial GAL4-expressing exon inserted into the third intron that follows the first coding exon (Figure 1A) [39]. Hence, tty^TG4 is expected to express GAL4 under the tty gene regulatory elements while arresting transcription of the remainder of the gene. tty¹ allele contains a transposon inserted into the third coding exon (Figure 1A) [40], and is thus expected to disrupt mRNA transcription. We found that tty mRNA expression in fly heads was abolished in both tty¹ and trans-heterozygous tty^1/TG4 flies as measured by quantitative reverse transcription polymerase chain reaction (qRT-PCR) (Figure 1B). Ubiquitous expression of a short-hairpin RNA (shRNA) against tty (tty^RNAi) significantly reduced tty mRNA levels in fly heads (Figure 1B). To examine the endogenous expression pattern of tty, we used tty^TG4 to drive the expression of nls-mCherry. We observed mCherry signals overlapping with glial marker Repo (Figure 1C) [41, 42], indicating that tty is endogenously expressed by brain glial cells. To validate that tty expression is lost in tty¹ mutant glial cells, we expressed poly(A)-binding protein (PABP) using a glial driver to isolate mRNA from glial cells before quantifying tty expression (Figure 1D) [43]. Consistent with the whole-head expression quantifications (Figure 1B), tty mRNA expression in glial cells was abolished in tty¹ mutant flies (Figure 1E).

Figure 1. Tweety is expressed in glia.

(A) Top: Genomic structure of the tweety (tty) gene in Drosophila melanogaster. Boxes represent exons. Protein-coding regions are colored in grey. Bottom: The transgenic insertion locations of tty^TG4 and tty¹ alleles are indicated by colored triangles.

(B) Relative amount of whole-body tty mRNA measured by real-time quantitative PCR. Both insertion strains (tty^TG4 and tty¹) and ubiquitous knockdown of tty (Act5C>tty^RNAi) exhibited significantly reduced tty mRNA levels. **** p<0.0001; t-test compared to w¹¹¹⁸.

(C) tty is expressed in larval brain glia. The tty^TG4 gene-trap expresses GAL4 to activate nls::mCherry expression in the same cells. Immunostaining of larval brains revealed colocalization between glial marker Repo and mCherry signals in the surface glia of brain lobe and ventral nerve cord (VNC).

(D) Schematic of mRNA purification from glial cells. Glia-specific repo-GAL4 drives the expression of FLAG-tagged human polyA-binding protein (hPABP).

(E) Loss of tty mRNA in mutant glial cells. Glial total mRNA purified from adult tty¹ mutant fly heads contained significantly diminished levels of tty mRNA. ** p<0.01; t-test compared to WT.

(F and G) Global or glial loss of tty impairs climbing ability of adult flies. Median climbed heights in 5 seconds of the indicated genotypes are normalized to the mean of the respective controls (tty¹/+ in F and RNAi controls in G). GR1 and GR2 are independent genomic fragments that contain full-length tty gene. repo-GAL4 and Eaat1-GAL4 drive expression in all glia and glutamate transporter-expressing glial cells, respectively. ** p<0.01; t-test.

(H) Global or glial loss of tty reduces locomotor activity of adult flies. CantonS and w¹¹¹⁸ were used as wild-type controls. *** p<0.001; t-test.

Given the glia-specific expression of tty, we examined whether tty deficiency impacts nervous system functions. We examined the motor behavior as its alteration is a common phenotype of neuronal dysfunction [44, 45]. Although tty gene is located near the flightless locus and was named after a flightless cartoon bird [46], we observed that tty¹ mutant flies were capable of flight. To examine and analyze locomotor function, we performed climbing and locomotor activity assays on adult flies. Compared to heterozygous control flies, tty¹ flies climbed lower in height (Figure 1F). Two independent genomic rescue constructs that encompass the tty locus restored climbing ability in tty¹ flies (Figure 1F). Supporting the notion that tty is functionally expressed in glial cells, knockdown of tty in all glia or glutamate transporter (Eaat1)-expressing glia also resulted in climbing deficit (Figure 1G). Likewise, flies with whole-body loss of tty (tty¹ and tty^1/TG4) or pan-glial knockdown of tty (repo>tty^RNAi) were less active compared to controls flies (Figure 1H). The locomotor phenotypes upon global loss of tty or glial-specific knockdown was relatively mild compared to other Drosophila mutations that primarily impair neurons [32, 44, 45], which is not surprising given that neurons are the actual mediators of locomotion with glia providing predominantly providing metabolic and trophic support to neurons [47]. These results demonstrate that tty is functionally expressed in glial cells and its loss impairs locomotor function.

Next, we sought to determine the subcellular site of function for tty. We created transgenic strains to express hemagglutinin (HA)-tagged tty and red fluorescent protein (RFP)-tagged human TTYH1, ortholog of tty [1], under GAL4-UAS control. We found that in larval ventral nerve cord, the coexpressed HA-tty and TTYH1-RFP were both localized to Rab7-positive vesicles, indicative of endolysosomal compartment (Figure S1A). To better resolve subcellular localization, we performed 2-D culture of primary brain cells isolated from fly larvae co-expressing HA-tty and fluorescent reporters in glia (Figure 2A). Both tty and human TTYH1 colocalized with endogenous Rab7 in primary glial cells (Figure 2B–C and Figure S1B). The colocalization with GFP-tagged LAMP1 further validates that tty is primarily localized to the endolysosomal compartment (Figure 2D). Endolysosomes are degradative organelle responsible for processing cargoes captured by autophagy [48]. We found endogenous Atg1 and co-expressed Atg8a, markers of autophagosomes [49, 50], colocalized with tty (Figure 2E–F), suggesting that tty-positive endolysosomes are terminal site for autophagy.

Figure 2. Tweety is localized to lysosome and is required for autophagy.

(A) Schematic of primary brain cell culture. Fluorescent reporters used in this study were expressed in glial cells using repo-GAL4. Brain cells from dissected third-instar larval brains were dissociated with enzymes before culturing on glass-bottomed dishes. After 3 days of culture, fluorescence signals were acquired by confocal microscopy or wide-field epi-fluorescence microscopy.

(B) Colocalization of tty (HA-tagged) and human TTYH1 (RFP-tagged) with endogenous endolysosomal marker Rab7. Confocal images show a single primary cultured glial cell. Dashed boxes indicate zoom-in regions shown to the right of the images.

(C − F) Colocalization of tty with endolysosome (Rab7), lysosome (LAMP1), and autophagosome (Atg1 and Atg8a) markers. Confocal images show single primary cultured glial cells. Dashed boxes indicate zoom-in regions shown to the right of the images. HA-tagged tty was expressed in glia (repo-GAL4) alone (C), with GFP-tagged LAMP1 (D), with mCherry-tagged Atg8a (F), or with endogenously expressed GFP-tagged Atg1 (E). Scale bars represent 5 μm.

(G and H) Reduced expression and maturation of cathepsin-L (CTSL) upon loss of tty. Representative Western blot images and quantifications for CTSL expression in adult fly heads. *** p<0.001, ** p<0.01, * p<0.05; t-test.

(I and J) tty is required for maintaining autophagic activity in glia. (I) Representative fluorescence image of single primary glial cells of wild-type (WT) and tty mutant (tty¹) larvae expressing tandem GFP-mCherry-tagged Atg8a in glia (repo-GAL4). High fusion rate between autophagosomes and endolysosomes reduces GFP fluorescent signals by pH quenching in the endolysosomal compartment, resulting in increased mCherry-to-GFP fluorescence ratio. (J) Quantification of Atg8a fluorescence ratio (mCherry/GFP) in glial cells of the indicated genotypes. Rescue with a genomic copy (GR) or ectopic expression of human TTYH1 in glia (repo>TTYH1) significantly increased fusion rate in tty¹ mutant glia. **** p<0.0001, *** p<0.001, ** p<0.01; t-test.

Lysosomal enzymes confer the degradative function of endolysosomal compartment. Certain lysosomal enzymes, such as cathepsins, require maturation via proteolytic cleavage in endolysosomes [51, 52]. Given that tty might be required for endolysosomal function, we examined whether tty deficiency impacts cathepsin-L (CTSL) maturation. In wild-type fly heads, we found that CTSL expression was increased by starvation (Figure 2H). Loss of tty reduced CTSL expression and rendered its expression unresponsive to starvation (Figure 2H). Levels of mature CTSL were not affected in fed flies, but were significantly reduced in starved tty¹ mutant flies (Figure 2H). Since defective cathepsin expression and maturation are hallmarks of lysosomal dysfunction [49, 51, 53], these findings suggest that tty deficiency impairs endolysosomal function which is critical to degradative processes such as autophagy.

We speculated that loss of tty in endolysosomes could impact autophagy, a homeostatic process that requires lysosomes to complete degradation. Fusion between autophagosomes and endolysosomes delivers autophagic cargoes to endolysosomes, where degradation of autophagosomes happens. We observed that the autophagosomal marker, mCherry-Atg8a, expressed in glial cells exhibited higher levels in adult fly heads of tty¹ mutants compared to those in wild-type (Figure S1C–D), suggesting diminished degradation of autophagosomes. The fusion rate between autophagosomes and endolysosomes can be measured using fluorescent reporter GFP-mCherry-Atg8a [54]. Quenching of GFP fluorescence in the acidic endolysosomes results in higher mCherry-to-GFP signal ratio, which represents higher rate of delivery of autophagosomal cargo to endolysosomes. We expressed and analyzed the fluorescence signals of GFP-mCherry-Atg8a in glial cells isolated from wild-type and tty¹ larval brains (Figure 2I). We found that tty deficiency led to reduced fusion between autophagosomes and endolysosomes, which was rescued by introducing a genomic rescue construct to the flies (Figure 2J). These results demonstrate that tty is required for maintaining autophagic activity in glia. Notably, ectopic expression of human TTYH1 in tty¹ mutant glia also increased the fusion rate to higher levels than in wild-type glia (Figure 2J), suggesting that human TTYH1 is sufficient to promote autophagic activity and the functional conservation between fly tty and human TTYH1.

Autophagy is a critical determinant of organelle homeostasis [49, 55]. Old or damaged mitochondria are captured and removed by autophagy. Defective autophagy and endolysosomal dysfunction impair mitochondrial turnover [56–59]. The impeded delivery of autophagic cargo to endolysosomes in tty-deficient glia might therefore result in slow autophagic degradation of mitochondria. We first examined the mitochondria captured by autophagosomes. Mitochondrial marker EYFP-mito and autophagosome marker mCherry-Atg8a were expressed in wild-type and tty¹ glia (Figure 3A). Upon analyzing the colocalized fluorescent signals, we found that a higher percentage of autophagosomes contained undegraded mitochondria in tty-deficient glia (Figure 3B). This suggests that while mitochondria are readily captured by autophagosomes, their degradation by lysosomes is delayed upon tty deficiency. To test if loss of tty impairs mitochondrial turnover, we utilized MitoTimer to report mitochondrial turnover in glial cells [60–62]. MitoTimer fluorescence irreversibly shifts from green to red as a function of time (Figure 3C) [60–62]. We found that MitoTimer exhibited higher red-to-green fluorescence ratio in tty-deficient glia, indicative of impaired mitochondrial turnover (Figure 3D–E). Consistent with a role of autophagy in facilitating mitochondrial turnover, genetic inhibition of autophagy by expressing shRNA against Atg5 (Atg5^RNAi) also impaired mitochondrial turnover in glia (Figure 3E). Together, these findings demonstrate that tty in endolysosomes regulates mitochondrial homeostasis via autophagic degradation of mitochondria.

Figure 3. tweety is required for mitochondrial turnover.

(A) Colocalization between mitochondria (EYFP-mito; green) and autophagosomes (mCherry-Atg8a; red) in wild-type (WT) and tty¹ mutant (tty¹) glia. Both fluorescent markers were expressed by glial driver repo-GAL4. Shown are representative images of primary glial cells.

(B) Mitochondria accumulate in autophagosome upon loss of tty. Quantification for the colocalization between mitochondria and autophagosomes as in (A). Boxplots (box: median ± interquartile range; whisker: all data range) displays percentage of colocalization in 25 wild-type and 36 tty¹ glial cells. * p<0.05; t-test.

(C) Principle of using MitoTimer to monitor mitochondrial turnover. The green fluorescence of MitoTimer irreversibly shifts to red fluorescence as mitochondria age.

(D) Representative fluorescence images and ratio-images of single live wild-type (WT) and tty¹ mutant (tty¹) glial cells expressing MitoTimer.

(E) Reduced mitochondrial turnover upon loss of tty or genetic inhibition of autophagy in glia. Shown are boxplots (box: median ± interquartile range; whisker: all data range) of MitoTimer fluorescence ratio of the indicated genotypes. **** p<0.0001, ** p<0.01; t-test.

Mitochondria are the main intracellular producers of reactive oxygen species (ROS) such as superoxide anion and hydrogen peroxide. The homeostasis of mitochondrial ROS is regulated by scavenging system to mitigate elevated ROS levels and prevent oxidative damage. Elevated ROS production can be caused by compromised mitochondrial quality control [63]. We reasoned that the impaired autophagic turnover of mitochondria in tty-deficient glia might induce mitochondrial oxidative stress. To quantitatively assess mitochondrial oxidative stress, we expressed mitochondria-targeted redox sensor mito-roGFP2-Grx1 in glial cells (Figure 4A) [36, 64]. The roGFP2-Grx1 sensor is coupled with the glutathione redox balance in mitochondrial matrix, which constitutes the major ROS scavenging system [63]. Glutathione oxidation resulted from elevated mitochondrial oxidative stress is rapidly equilibrated by the roGFP2-Grx1 at the expense of roGFP2 oxidation [36]. Fluorescence excitation of oxidized roGFP2 shifts from 488 nm to 405 nm [36, 64]. Ratiometric analyses of the roGFP2 emission signals from the two excitation wavelengths can thus be used to quantify mitochondrial oxidative stress [36, 64]. Conforming with our hypothesis, we observed that tty-deficient glia exhibited elevated mitochondrial oxidative stress, which was rescued by introducing a genomic rescue construct (Figure 4B). Glial overexpression of tty or human TTYH1 significantly reduced mitochondrial oxidative stress, regardless of tty deficiency (tty¹) or not (wild-type) (Figure 4C). These data indicate that tty is necessary and sufficient to mitigate mitochondrial oxidative stress in glia.

Figure 4. Elevated mitochondrial oxidative stress in tweety mutant glia.

(A) Representative fluorescence images and ratio-images of single live wild-type (WT) and tty¹ mutant (tty¹) glial cells expressing the ratiometric mitochondrial oxidative stress sensor Mito-Grx1-roGFP2.

(B) Elevated mitochondrial oxidative stress upon loss of tty in glia. Mito-Grx1-roGFP2 fluorescence ratios of primary glial cells of the indicated genotypes are normalized to the mean of the wild-type (WT) ratios. Higher ratio values indicate elevated oxidative stress. *** p<0.0001; Mann-Whitney test.

(C) Glial (repo-GAL4) overexpression of tty (+ tty) or human TTYH1 (+ TTYH1) significantly reduced mitochondrial oxidative stress in both wild-type and tty¹ mutant glia. Data points for “WT” are the same as in Figure 4B. *** p<0.0001; one-way ANOVA.

(D) Loss of tweety sensitizes flies to oxidative stress inducers. Climbing ability of tty¹ mutant flies with (tty¹+GR1) or without (tty¹) genomic rescue was assessed. Boxplots display median climbed height normalized to the mean of the control (tty¹+GR1 with DMSO). Flies were fed with food containing vehicle (0.1% DMSO), rotenone (50 μM), or paraquat (1 μM) for 3 days prior to climbing assay. n.s.: not significant, * p<0.05; t-test compared to “DMSO”.

(E) Antioxidant partially restores mitochondrial turnover in tty¹ mutant glia. Boxplots show MitoTimer fluorescence ratio of primary glial cells of the indicated genotypes. Primary cells were treated with or without 40 ng/mL AD4 in culture medium for 3 days prior to imaging experiments. ** p<0.01, * p<0.05; Mann-Whitney test.

(F and G) Exogenous antioxidant alleviates locomotor phenotypes of tweety mutant flies. Climbing ability (F) and locomotor activity (G) of wild-type and tty1 flies were assessed and quantified after feeding with food containing 40 μg/mL AD4 (N-acetyl cysteine amide) for 1 week after eclosion. *** p<0.001, ** p<0.01, * p<0.05; t-test.

We speculated that impaired ROS scavenging underlies the elevated mitochondrial oxidative stress upon loss of tty. If so, tty-deficient flies would be more sensitive to oxidative insults. Exposure to high quantity of rotenone and paraquat elicits oxidative damage, neurotoxicity, and locomotor dysfunction in flies [65–68]. We tested the impact of these oxidative stress inducers on the climbing ability of tty-deficient flies. We fed flies with food infused with rotenone (50 μM), paraquat (1 μM), or vehicle for 3 days before testing their climbing ability. Neither rotenone nor paraquat affected climbing performance in tty¹ flies carrying a genomic rescue construct (Figure 4D). In contrast, both rotenone and paraquat exacerbated the climbing phenotype in tty¹ mutant flies (Figure 4D). Thus, loss of tty diminishes the capacity to mitigate elevated oxidative stress induced by rotenone and paraquat at the respective dosage.

Does oxidative stress account for the altered locomotor function in tty-deficient flies? We tested whether promoting ROS scavenging could rescue the locomotor phenotypes in tty¹ mutant flies. N-acetylcysteine amide (AD4), an antioxidant that acts as glutathione precursor, alleviates mitochondrial oxidative stress in cell and animal models [69, 70]. Feeding flies with food containing N-acetylcysteine or AD4 has been shown to reduce ROS, increase locomotor activity, and extend lifespan [69–72]. We found that treatment of cultured tty¹ glia with AD4 slightly reduced MitoTimer fluorescence ratios (Figure 4E). However, the MitoTimer ratios were not restored to the same levels as in wild-type glia (Figure 4E), indicating that the defective mitochondrial turnover persisted. We fed adult tty¹ mutant flies with AD4 for one week, and found that both climbing ability and locomotor activity were significantly improved compared to vehicle control (Figure 4F–G). Again, neither of the locomotor parameters was rescued to comparable levels exhibited by wild-type flies (Figure 4F–G). Hence, enhancing ROS scavenging is not sufficient to restore locomotor function in tty¹ mutant flies. The results also implicate that mitochondrial oxidative stress is only a partial outcome of diminished autophagic activity upon loss of tty.

Mitochondria process metabolites to support ATP production and biosynthesis. The altered mitochondrial homeostasis upon loss of tty could potentially impact the metabolic functions of mitochondria. Pyruvate, the end product of glycolysis, is a major mitochondrial substrate for bioenergetics or anaplerosis [73]. Thus, pyruvate consumption by mitochondria serves as an indicator of mitochondrial metabolic activity. Mitochondrial pyruvate carrier (MPC) regulates mitochondrial uptake of pyruvate for metabolic consumption (Figure 5A) [74, 75]. MPC activity is sensitive to specific MPC1 blocker UK-5099 which prevents further import of pyruvate into mitochondrial matrix for consumption [74, 76]. We utilized the genetically encoded fluorescent sensor, mito-PyronicSF [77], to measure mitochondrial pyruvate levels in wild-type and tty-deficient glia (Figure 5B). Coupled with acute inhibition of pyruvate import by UK-5099, mito-PyronicSF has been used to track the depletion of pyruvate as a readout of mitochondrial pyruvate consumption [77, 78]. The steady-state mitochondrial pyruvate levels were comparable between wild-type and tty¹ glia (Figure 5B–C), suggesting that loss of tty does not impact mitochondrial uptake of pyruvate. As expected, UK-5099 application resulted in rapid depletion of mitochondrial pyruvate (Figure 5B–C). Within a 5-minute window after UK-5099 application, wild-type glia exhibited a steeper reduction of mitochondria pyruvate compared to tty¹ glia (Figure 5D–E). Thus, loss of tty reduces mitochondrial pyruvate consumption in glia. Interestingly, autophagy inhibition by knocking down Atg5 also impaired mitochondrial consumption of pyruvate (Figure 5D–E). These findings indicate altered mitochondrial metabolism of energy substrate upon perturbation of autophagy or tty-associated endolysosomal process.

Figure 5. Loss of tweety impairs mitochondrial pyruvate consumption in glia.

(A) Schematic of mitochondrial import of pyruvate via mitochondrial pyruvate carrier (MPC). UK-5099 is a selective MPC inhibitor. Mitochondrial pyruvate (Pyruvate_mito) can be measured by ratiometic fluorescence sensor mito-PyronicSF.

(B) Representative ratio-images of single live wild-type and tty¹ glial cells expressing mito-PyronicSF.

(C) Representative time-lapse traces of mito-PyronicSF fluorescence ratio of primary glial cells measured by live-cell imaging. Data points are mean ± SEM of mito-PyronicSF ratios from wild-type or tty¹ glial cells acquired from a single imaging experiment.

(D and E) Quantifications of mito-PyronicSF fluorescence before (UK-5099 −) and 5 minutes after application of 10 μM UK-5099 (UK-5099 +) to the bath solution. Boxplots display ratio values normalized to the mean values from wild-type glial cells imaged on the same day (D) and percentage change of the ratio values (E). **** p<0.0001, *** p<0.001; t-test.

The bioenergetics of a cell is largely determined by the ATP production from glycolysis and mitochondrial oxidative phosphorylation (OXPHOS). The altered mitochondrial pyruvate consumption in tty-deficient glia prompted us to examine whether ATP production is altered. ATP/ADP ratio can be measured using the genetically encoded ratiometric sensor PercevalHR [79]. We and others have used PercevalHR coupled with inhibitors targeting metabolic enzymes to assess bioenergetics in both mammalian and Drosophila cells [31, 32, 80]. Here, we measured ATP/ADP ratios in PercevalHR-expressing glia isolated from wild-type, tty¹, and tty¹+GR1 (genomic rescue in tty¹ background) larvae (Figure 6A). We found that steady-state ATP/ADP levels were not significantly affected by the loss of tty (Figure 6B), despite an impaired mitochondrial pyruvate metabolism. This suggests that tty-deficient glia are able to maintain ATP production potentially by shifting the dependence on glycolysis versus mitochondrial OXPHOS. We therefore monitored ATP/ADP ratios in response to glycolytic blocker 2-deoxyglucose (2-DG). 2-DG inhibits hexokinase, which catalyzes the rate-limiting first step in glycolysis. Upon 2-DG inhibition, production of both ATP and pyruvate by glycolysis ceases, leaving mitochondrial OXPHOS as the sole ATP producer. Application of 2-DG to isolated glia resulted in a rapid and substantial ATP/ADP decline (Figure 6C–E). While pyruvate supplementation acutely raised mitochondrial pyruvate concentration [77], it was not able to restore the ATP/ADP decline due to 2-DG (Figure S2A–B). This suggests that glycolysis, but not mitochondrial pyruvate metabolism, accounts for the decline in ATP levels in response to 2-DG. While both control and tty-deficient glia exhibited a rapid response to 2-DG, loss of tty led to a significantly faster and larger decline in ATP/ADP ratio (Figure 6C–E). Thus, glycolysis contributes more to steady-state ATP levels in the absence of tty. These data indicate that loss of tty induces glycolysis upregulation to maintain ATP production in glia.

Figure 6. Loss of tweety shifts glial bioenergetics towards glycolysis-dependent ATP production.

(A) Representative ratio-images of single live wild-type and tty¹ glial cells expressing ratiometric ATP/ADP sensor PercevalHR.

(B) Steady-state PercevalHR fluorescence ratios of primary glial cells of the indicated genotypes. n.s. not significant; one-way ANOVA.

(C) Time-lapse traces of PercevalHR fluorescence ratio of primary glial cells measured by live-cell imaging. Data points are mean ± SEM of ratios from the primary glial cells of the indicated genotypes, and are normalized to ratio values at 1 minute before 2-DG (10 mM) application. Dashed line indicates 10-minute time-point after 2-DG application.

(D and E) ATP:ADP levels are more sensitive to glycolytic blockage upon loss of tweety in glia. Quantification of PercevalHR fluorescence at 10 minutes after 2-DG application to the glial cells of the indicated genotypes. Boxplots display percentage change of ratio values (D) and ratio values (E). *** p<0.0001, ** p<0.01; Mann-Whitney test.

(F and G) Quantification of PercevalHR fluorescence before (G) and 10 minutes after (H) 2-DG application to the glial cells treated with 40 ng/mL AD4 in culture medium for 3 days prior to imaging experiments. Boxplots display steady-state ratio values (G) and percentage change of ratio values (H). n.s. not significant, * p<0.05; Mann-Whitney test.

(H and I) Genetic inhibition of autophagy phenocopies loss of tweety. Changes in PercevalHR fluorescence ratio upon 2-DG application to primary glial cells expressing short hairpin RNAs against Atg5 (Atg5^RNAi), Atg8a (Atg8a^RNAi), and mCherry (mCherry^RNAi; as control) are shown as time-lapse traces (I) and percentage change of ratio values (J). *** p<0.001, ** p<0.01; t-test.

(J and K) Autophagy induction does not restore sensitivity to glycolytic blockage in tweety mutant glia. PercevalHR fluorescence ratio at steady-state (K) and 10 minutes after 2-DG application (L) to wild-type glial cells and Atg1-overexpressing (repo>Atg1) wild-type and tty¹ glial cells. Data points for “WT” are the same as in Figure 6B. **** p<0.0001, *** p<0.001; Mann-Whitney test.

(L) Schematic showing the function of tweety in autophagy, mitochondrial turnover, and bioenergetics in glia. Old or damaged mitochondria are engulfed in autophagosomes to be delivered to endolysosomes for enzymatic degradation. Loss of tweety impedes fusion between autophagosomes and endolysosomes, causing a slow mitochondrial turnover. Old mitochondria which exhibit elevated oxidative stress accumulate. Mitochondrial metabolism such as pyruvate oxidation diminishes. Consequently, loss of tweety stipulates glial cells to ramp up glycolysis for ATP production.

The cooccurrence of diminished mitochondrial turnover and upregulated glycolysis-dependent ATP production suggests that the bioenergetic shift might stem from defective mitochondrial homeostasis. Since mitochondrial turnover in tty-deficient glia was partially rescued by antioxidant AD4 (Figure 4F), we tested whether AD4 could restore the ATP/ADP response to 2-DG. We found that AD4 treatment of cultured glial cells did not alter steady-state ATP/ADP ratio (Figure 6F). In response to 2-DG, the ATP/ADP decline in tty¹ glia was not relieved by AD4 treatment and remained significantly larger compared to wild-type glia (Figure 6G). These data demonstrate that mitochondrial oxidative stress does not underlie the bioenergetic shift upon tty deficiency. Collectively, these findings support the model that elevated oxidative stress is a parallel outcome of impaired autophagic turnover of mitochondria.

To examine whether impaired autophagy results in the same bioenergetic shift observed in tty-deficient glia, we genetically inhibited autophagy and subsequently measured the ATP/ADP dynamics. Compared to non-targeting control, knockdown of Atg5 or Atg8a, autophagy genes required for autophagosome formation [49, 81], induced a significantly larger decline in ATP/ADP ratios in response to 2-DG (Figure 6H–I). These data indicate that autophagy impairment is sufficient to induce glycolysis upregulation for ATP production in glia. The bioenergetic shift upon autophagy inhibition mimics tty deficiency (Figure 6C–E), and supports the notion that tty facilitates autophagy (Figure 2). To confirm that tty functions downstream of autophagosome formation to regulate bioenergetics, we ectopically promoted autophagy initiation by overexpressing Atg1 [56, 82]. Interestingly, steady-state ATP/ADP ratios were significantly lower in tty¹ glia upon Atg1 overexpression (Figure 6J). In response to 2-DG, the ATP/ADP decline remained significantly larger in tty¹ glia (Figure 6K), demonstrating that autophagy initiation is upstream of tty function. Atg1 is sufficient to promote mitochondrial fission and autophagic turnover [56, 57, 82]. However, this process requires endolysosomes to complete the degradation and recycling of mitochondrial materials. Our findings thus suggest that defective recycling of mitochondrial materials may underlie the steady-state ATP deficit during autophagy induction in the absence of tty. Furthermore, we tested if starvation, which induces autophagy, would have any impact on the survival of tty-deficient flies. Complete nutrient withdrawal by supplying only water-infused agar was detrimental to the survival of both wild-type and tty¹ flies, resulting in death within six days (Figure S3A). However, tty¹ flies exhibited a significantly faster decline in survival in response to starvation. Likewise, flies with glial-specific knockdown of tty, Atg5, or Atg8a also succumbed to starvation faster than control flies (Figure S3B–C). These data are in line with the observed bioenergetic shift in tty-deficient and autophagy-defective glia. Taken together, our findings demonstrate that tty-mediated autophagy regulates bioenergetic balance in glia (Figure 6M). Loss of tty impairs autophagic turnover of mitochondria and induces an upregulation of glycolysis-dependent ATP production.

Discussion

The subcellular localization and cellular function of TTYH proteins have not been well defined. In this study, we determined that Drosophila tweety (tty), the founding member of the TTYH family, is expressed in glial cells and primarily localized to endolysosomes, where it facilitates autophagy by allowing efficient fusion of autophagosomes with endolysosomes. Remarkably, human TTYH1 rescues autophagic deficit in tty-deficient glia, demonstrating functional conservation between the orthologs. Notably, our findings are consistent with the reported mouse brain-derived autolysosomal proteome, which contained Ttyh1 and Ttyh3 [83]. Using genetically encoded fluorescent reporters, we found that tty is required for mitochondrial homeostasis via autophagic turnover of mitochondria. Loss of tty in glia impairs locomotor functions and sensitizes animals to oxidative stress inducers. Whereas alleviating oxidative stress only partially restores mitochondrial homeostasis and locomotor functions, our findings reveal that tty deficiency results in bioenergetic alteration. Consistent with a deficit in mitochondrial metabolism, tty-deficient flies are sensitive to starvation and succumb to death faster than wild-type flies. In glia, loss of tty stipulates an upregulation of glycolysis to maintain steady-state ATP production. Furthermore, we found that such bioenergetic shift also occurs in autophagy-deficient glia. Thus, our findings uncover tty-mediated autophagy as a regulator of bioenergetics in glia.

Acute inhibition of glycolysis resulted in rapid reduction of ATP levels in fly glia, indicating that glia rely on glucose oxidation as the major fuel source for ATP production. This is in contrast to fly neurons whose ATP levels were not acutely altered by 2-DG [84]. Our finding conforms with observations that glial cells such as astrocytes possess high glycolytic activity and can survive without mitochondrial respiration [85–87]. The importance of glycolysis to Drosophila glia is exemplified by the neurodegeneration induced by defective glial glycolysis [87, 88]. On the other hand, bioenergetics in neuron appear to rely on OXPHOS driven by mitochondrial consumption of energy substrates such as pyruvate and amino acids that can be derived from glia [87, 89, 90]. Glucose uptake and breakdown by neurons may be used to support redox homeostasis via pentose phosphate pathway [91]. It stands to reason that, during homeostasis, glia perform glycolysis to produce both ATP for self-demand and energy substrates to be sent to neurons. However, the larger ATP decline in tty-deficient glia in response to 2-DG suggests that other fuel metabolism also contributes to glial ATP production. The concurrent mitochondrial dyshomeostasis and slow pyruvate consumption, resulted from loss of tty, implicate that mitochondrial respiration contributes to ATP production in glia. Indeed, recent studies have shown that mitochondrial fatty acid breakdown and OXPHOS also support glial functions [92–96]. While both glycolysis and OXPHOS may contribute to glial bioenergetics, energy-expensive processes, such as glutamate uptake [97, 98], might demand accelerated ATP production in glia. Whether glycolysis or OXPHOS is preferred for these on-demand circumstances remains to be determined.

By uncovering the cellular function of tty, our study reveals autophagy as a regulator of glial bioenergetics. Defective autophagy is expected to impact the capacity of mitochondrial respiration due to diminished mitochondrial turnover. It was not clear, however, whether and how glial bioenergetics adapt to deficit in autophagy. We found that both tty deficiency and autophagy inhibition led to a glycolytic shift of ATP levels in glia. These results demonstrate that glial bioenergetic balance is subject to autophagy activity. While genetic induction of autophagy did not alter steady-state ATP levels or its response to 2-DG in wild-type glia, steady-state ATP levels in tty-deficient glia were reduced upon autophagy upregulation. These observations further highlight the significance of autophagic and lysosomal functions in regulating bioenergetics [58, 59]. Interestingly, it was reported that autophagy suppresses glycolytic metabolism by degrading hexokinase 2 in liver cancer [99]. Hence, derepression of glycolysis in addition to reduced mitochondrial respiration upon autophagy inhibition might underlie the glycolytic shift in glial bioenergetics.

Previous reports have put forth two distinct molecular functions for TTYH proteins. Earlier studies which involved heterologous overexpression suggested that drosophila tty and mammalian TTYHs form chloride channels at the plasma membrane [23, 25, 26]. While rodent Ttyh1 was also found in endosomal and lysosomal compartments in glial cells [2, 10], it is unclear whether its putative chloride channel activity would impact vesicular trafficking in these compartments. Lysosomal chloride ions regulate lysosomal functions such as enzyme activity and vesicular fusion [100, 101]. Loss of tty in the endolysosomal compartment could potentially lead to reduced luminal chloride level if it mediates transmembrane chloride flux. However, recent findings have revealed high-resolution protein structures of TTYHs that refute chloride channel function [28, 29]. TTYHs lack structural features that could support anionic conductance [28, 29]. Moreover, the amino acid previously proposed to confer ion conductivity (R165 in TTYH1) is located at the distal tip of the extracellular domain, away from the transmembrane region [28, 29]. Notably, R165 is only conserved among TTYH orthologs in tetrapods but not in other species such as drosophila. These latest findings argue against the notion that tty and TTYH1 directly mediate lysosomal chloride transport, although a function as auxiliary protein to bona fide lysosomal chloride transporters cannot be ruled out. Interestingly instead, both structural studies reported a highly conserved lipid-binding domain in mammalian TTYH orthologs, implicating function in transporting membrane lipids [28, 29]. We performed sequence alignment and observed that drosophila tty also shares high homology with mammalian TTYHs within the putative lipid-binding domain (data not shown). Given that membrane lipids are known determinants of lysosomal functions [102–104], an alternative model that tty and TTYH1 mediate lysosomal membrane lipid transport warrants future validation.

Main Points.

• Tweety homologs are located in endolysosomes to facilitate autophagic turnover of mitochondria in glia.

• Loss of tweety impairs mitochondrial homeostasis and metabolism, prompting glia to upregulate glycolysis for ATP production.

• Tweety mediates the crosstalk between autophagy and bioenergetics.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.