After control or Shu knockdown (n = 3 each), OSCs were lysed in binding buffer [50 mM Tris-HCl (pH 8.0), 150 mM KOAc, 5 mM Mg(OAc)2, 5 mM DTT, 0.1% NP-40, 2 µg/mL leupeptin, 2 µg/mL pepstatin A, 0.5% aprotinin, 10% glycerol] and centrifuged to remove insoluble debris. The lysates were then incubated with anti-Shu antibody bound to Dynabeads Protein G (Thermo Fisher Scientific) at 4°C for 2 h. The beads were washed five times with binding buffer and twice with 50 mM ammonium bicarbonate. Proteins on the beads were digested by adding 400 ng of trypsin/Lys-C mix (Promega) at 37 °C overnight. The digests were reduced, alkylated, acidified, and desalted using GL-Tip SDB (GL Sciences). The eluates were evaporated in a SpeedVac concentrator and dissolved in 0.1% trifluoroacetic acid and 3% acetonitrile (ACN). LC-MS/MS analysis of the resultant peptides was performed on a nanoElute 2 coupled with a timsTOF HT mass spectrometer (Bruker). The peptides were separated on a 75-μm inner diameter × 150 mm C18 reversed-phase column (Nikkyo Technos) using a segmented linear gradient: 5%–20% ACN from 0 to 40 min, 20%–35% ACN from 40 to 60 min, followed by a rapid increase to 95% ACN from 60 to 61 min, and held at 95% ACN from 61 to 75 min. Data acquisition was conducted in dia-PASEF mode with an MS1 m/z range of 100–1700 and an ion mobility (1/K0) range of 0.6–1.6. The ramp time was set to 100 ms with a 100% duty cycle. The MS2 polygon was manually defined based on the region where peptides were predominantly detected, specifically using the following four vertices: Point 1 (m/z = 300, 1/K0 = 0.64), Point 2 (m/z = 300, 1/K0 = 0.8), Point 3 (m/z = 770, 1/K0 = 0.97), and Point 4 (m/z = 770, 1/K0 = 1.15). Within the defined polygon, an 8 Da isolation window was employed with an overlap of 0.5 Da for each window, resulting in 62 windows per cycle and a cycle time of 3.36 s. Protein identification was carried out in DIA-NN (version 1.9)81 against a Drosophila melanogaster (UP000000803) in silico spectral library. Parameters for constructing the library included: trypsin as the digestion enzyme, one allowed missed cleavage, peptide lengths ranging from 7 to 45 amino acids, precursor charges of 2 to 4, and a fragment ion m/z range of 200–1800. Features such as library-free search, deep learning-based retention time and ion mobility predictions, N-terminal methionine cleavage, and cysteine carbamidomethylation were enabled. In the DIA-NN search settings, MS1 and MS2 mass accuracies were set to auto, neural network classifiers were configured for single-pass mode, and the quantification strategy was set to QuantUMS (high precision).