AMPylation Regulates 5′-3′ Exonuclease PLD3 Processing

Abstract

The 5′-3′ exonuclease phospholipase D3 (PLD3) is a single-pass transmembrane protein undergoing sequential post-translational modifications by N-glycosylation, AMPylation, and proteolytic cleavage. The substrates of PLD3 5′-3′ exonuclease activity are single-stranded DNAs and RNAs, which act as ligands for Toll-like receptors and trigger a downstream proinflammatory response. Although PLD3 has primarily been studied in immune cells, recent findings indicate its enrichment in neurons, where it plays a role in regulating axonal fitness in Alzheimer’s disease. However, the regulatory mechanisms governing the proteolytic processing of PLD3 into its catalytically active soluble form and its functional roles in both immune and neuronal cells remain unclear. Here, we describe the functional implications of PLD3 AMPylation, its direct interaction with the protein adenylyltransferase (FICD), and changes in PLD3 processing in Parkinson’s disease patient–derived neurons. We identified PLD3 AMPylation sites within the protein's soluble region and showed that mutation of these sites hampers PLD3 activation and its catalytic activity. Overexpression of FICD AMP transferase accelerates PLD3 degradation and induces cellular stress response. Together, our findings demonstrate a critical role of AMPylation in PLD3 processing and regulation of its catalytic activity and provide new insights into the protein’s transport and localization to lysosomes. The observation that PLD3 regulation in Parkinson’s disease–derived neurons is altered compared with healthy neurons further highlights its role in neurodegenerative diseases.

Graphical Abstract

Highlights

• •Mapping AMPylation sites of PLD3 by data-independent acquisition mass spectrometry.
• •FICD’s AMPylation activity reduces PLD3 protein levels.
• •AMPylation of PLD3 regulates TLR9 response in Cal-1 cells.
• •PLD3 processing is cell-type specific and altered in Parkinson’s patients.

In Brief

We have identified AMPylation sites in the soluble region of the 5′-3′ exonuclease phospholipase D3 (PLD3). Point mutations at these sites impair PLD3’s proteolytic processing, dimerization, and catalytic activity. Our study reveals critical functional roles of AMPylation in regulating PLD3 processing, its interaction with the protein AMP-transferase FICD, and provide further evidence highlighting its involvement in neurodegenerative diseases.

In humans, two AMP-transferases, the endoplasmic reticulum (ER)–resident protein adenylyltransferase FICD and the mitochondrial Selenoprotein O (gene: SELENOO), have been described to date (1, 2, 3). FICD contains a highly conserved Fic domain, which was first identified in bacteria, and is characterized by the active site sequence HPFIDGNGRTS in human cells (4, 5). The bifunctional FICD catalyzes both protein AMPylation, which involves the transfer of an AMP to a target protein, and deAMPylation, the reverse process (Fig. 1A). Switching between the two functions is governed by an autoinhibitory α-helix. A point mutation of the critical glutamate E234 to glycine within the autoinhibitory helix disables the enzyme’s ability to escape the highly active AMP-transferase state, which is characterized by its lack of deAMPylation activity (4, 6, 7, 8, 9, 10, 11). Furthermore, the catalytic activity is regulated through FICD homodimerization and the ratio of Mg²⁺ to Ca²⁺ ions (6, 12). At the cellular level, FICD is a low abundant protein residing in the ER (13). Although FICD was initially identified as the Huntingtin-interacting protein, it has been linked to neurodegeneration in Caenorhabditis elegans, shown to accelerate neuronal differentiation in human cerebral organoids, and most recently associated with diabetes in human insulin–producing beta cells (10, 14, 15, 16, 17, 18, 19).

Fig. 1.

Proteomics of protein AMPylation.A, schematic representation of the proposed regulation of FICD through dimerization and the autoinhibitory α-helix. B, mass spectrometric (MS) properties and fragmentation of the AMP group. C, structure of the pro-N6pA probe. D, chemical proteomics workflow used for profiling of protein AMPylation by in-gel analysis and MS-based proteomics. E, schematic drawing of PLD3 domains and its proteolytic processing. AlphaFold v2.0-generated prediction of PLD3 was used for visualization (75, 76). F, Western blot analysis of different cell types using anti-PLD3 antibody to show differences in expression levels and proteoforms. hiPSCs, human-induced pluripotent stem cells, 10 days; iNGNs, neurons after 10 days differentiation from hiPSCs overexpressing neurogenin-1 and -2, physiological neurons were differentiated for 5 (young) or 10 (mature) weeks from hiPSCs. Equal amounts (10 μg) of total protein were loaded onto SDS-PAGE. FICD, protein adenylyltransferase; PLD3, phospholipase D3; pro-N6pA, N⁶-propargyl adenosine phosphoramidate.

The major AMPylation/deAMPylation substrate of FICD is the chaperone and unfolded protein response (UPR) regulator, heat shock protein family A member 5 (HSPA5), also known as BiP or GRP-78 (13, 20, 21). The chaperone activity of HSPA5 is inhibited by AMPylation following UPR stress release, before HSPA5 levels return to basal levels (7, 22). Another FICD AMPylation substrate has been validated in vitro, the peptidase cathepsin B, which is inhibited by AMPylation (23). However, the function and regulation of FICD-catalyzed AMPylation/deAMPylation activity and its downstream signaling role in different cell types remain poorly understood.

Direct protein AMPylation status determination by, for example, mass spectrometry (MS)–based proteomics remains challenging because of the low abundance of AMP moiety–containing peptides, poor ionization because of the presence of the negatively charged phosphate group, and the fact that AMP group readily fragments during peptide fragmentation hampering the identification in mass spectra (Fig. 1B) (23, 24, 25). Several complementary strategies have been used to characterize protein AMPylation, including radioactive-labeled AMPylation substrates, anti-AMPylation antibodies, and fluorescence- or affinity-tagged ATP analogs to enrich and visualize AMPylated proteins (26, 27, 28, 29, 30). The main drawback of ATP-based analogs is that they can be only applied in cell lysates, resulting in loss of subcellular compartmentalization. Therefore, we recently developed the N⁶-propargyl adenosine phosphoramidate prodrug (pro-N6pA, Fig. 1C), which renders the nucleotide probe cell permeable, while avoiding the need for initial phosphorylation. This allows monitoring AMPylation in living cells (Figs. 1D and S1) (23, 31). Of note, the inherent competition with endogenous ATP restricts the absolute quantification of labeled proteins. The terminal alkyne on N⁶-propargyl is used for downstream analyses, enabling the pull-down of probe-labeled proteins by affinity purification and subsequent comparison to control or other treatment conditions.

The pro-N6pA probe has been applied to profile protein AMPylation using quantitative MS-based chemical proteomics in human cancer cell lines and during the differentiation of human-induced pluripotent stem cells (hiPSCs) into neurons. This approach led to the identification of more than 100 AMPylated proteins from various subcellular localizations, including the ER, cytoplasm, and lysosomes (23, 31). Significant changes in AMPylation were particularly observed on lysosomal proteins, ACP2, ABHD6, and PLD3, during neuronal differentiation (31). Interestingly, neutralization of the lysosomal compartment in neuroblastoma cells by treatment with monensin or bafilomycin A1 resulted in increased AMPylation of PLD3 and the identification of an additional group of AMPylated proteins, including NOTCH2 and APP (32).

PLD3 is a type II transmembrane protein with a characteristic phospholipase fold, exhibiting exonuclease activity toward single-stranded DNA and RNA (Fig. 1E) (33, 34, 35, 36, 37). It is synthesized as a full-length (75 kDa) protein in the ER, where it undergoes N-glycosylation (37). PLD3 is then trafficked through the endolysosomal pathway to lysosomes, where it primarily localizes and undergoes proteolytic cleavage (37). This cleavage generates a soluble, truncated form of PLD3 (50 kDa). Furthermore, PLD3 was shown to undergo additional modification by AMPylation (31). However, the modification sites remained unidentified, representing a major bottleneck for functional studies. The soluble form of PLD3 has been shown to negatively regulate the immune receptor Toll-like receptor 9 (TLR9) by degrading its ligands and to positively regulate TLR7 by generating ligands for its distinct binding pockets (38).

PLD3 expression levels and proteoform ratios were shown to change dramatically during neuronal differentiation: In iPSCs, full-length PLD3 is present at relatively low levels, whereas the amount of catalytically active, soluble PLD3 rapidly increases during neuronal differentiation. Interestingly, during neuronal maturation, an increasing amount of soluble PLD3 becomes AMPylated, which inhibits its exonuclease catalytic activity (31).

PLD3 is well known for its involvement in lysosomal function. Mitochondrial DNA (mtDNA) has been identified as a key physiological substrate of PLD3 during mitophagy. Studies in PLD3^−/− cells have reported impaired mitophagy and the accumulation of mtDNA in lysosomes, highlighting a crucial role for PLD3 in maintaining lysosomal homeostasis (34). Given the central role of lysosomes and associated pathways in neurodegenerative diseases, PLD3 has garnered growing interest in this context. Indeed, PLD3 polymorphism has been identified as genetic risk factors for late-onset Alzheimer’s disease (AD), with risk variants promoting amyloid-β accumulation (33). In addition, altered PLD3 levels have been shown to impact axonal function in the context of AD, likely through the modulation of endolysosomal biogenesis (39). Furthermore, PLD3 regulates mtDNA degradation in a PINK1-dependent manner during mitophagy. Disruption in this pathway has also been linked to hyperactivation of the STING signaling (34). Interestingly, both PINK1, a crucial protein in mitophagy and a known risk factor for Parkinson’s disease (PD), and STING signaling, which is implicated in neuroinflammation and neurodegeneration, are closely associated with PD pathogenesis (40, 41). Due to emerging evidence on the relevance of PLD3 in neurodegeneration, along with its regulated expression during neuronal development and neuron-specific processing, PLD3 has the potential as a pharmacological target for neurodegenerative diseases (33, 39, 42, 43, 44, 45). However, the mechanism and cell type–dependent differences regulating PLD3 AMPylation, a newly identified post-translational modification (PTM) of PLD3 that may directly impact its exonuclease activity, remain unclear (31). Clarifying these regulatory mechanisms is essential for evaluating the therapeutic potential of PLD3 and guiding the development of novel targeted interventions in neurodegenerative diseases.

Here, we established an MS-based proteomics approach to identify the AMPylation sites on PLD3 and to characterize the functional consequences of its modification. We found that PLD3 and FICD directly interact with each other, suggesting that FICD might be a canonical AMP-transferase modifying PLD3. Complementary proteomics analyses showed that FICD regulates PLD3 homeostasis and processing in living cells. Moreover, PLD3 processing is delayed in PD patient–derived neurons, highlighting its functional role in neurodegeneration.

Experimental Procedures

Culturing of HEK293T, Hap1 and Cal-1 cells

Human embryonic kidney 293T (HEK293T) (received from Prof Dr Thomas Carell Lab) and Hap1 cells (received from Prof Dr Lucas Jae Lab) were grown in Dulbecco's modified Eagle's medium—high glucose or Iscove's modified Dulbecco's medium, respectively. Both media were supplemented with 10% fetal bovine serum and 2 mM l-alanyl-l-glutamine. Cells were maintained at 37 °C in a 5% CO₂ atmosphere. Cal-1 cells were cultured in RPMI-1640 medium supplemented with 10% heat-inactivated fetal calf serum, 100 U/ml penicillin–streptomycin, 1 mM sodium pyruvate, 2× GlutaMAX, 10 mM Hepes, and 1× minimal essential medium nonessential amino acid. Cells were maintained in a humidified incubator at 37 °C with 5% CO₂.

Cultivation of Neural Precursor Cells and Midbrain Neurons

hiPSC-derived inducible neurogenins (inducible overexpression of neurogenin-1 and -2) and physiological neurons were generated as previously described (31). The generation of neural precurfigsor cells (NPCs) and the differentiation to midbrain dopaminergic neurons was performed as previously described (46). Briefly, NPCs were cultivated on 12-well plates coated with Geltrex (Gibco) in N2B27 medium, consisting of 50% Dulbecco's modified Eagle's medium/F12 Glutamax, 50% neurobasal medium, 1:200 N2, 1:100 B27, and 1:100 penicillin–streptomycin. The N2B27 medium was further supplemented with 3 μM CHIR 99021, 0.5 μM purmorphamine, 10 μM SB-431542, and 150 μM ascorbic acid. After washing once with PBS, NPCs were split in 1:3 ratios and detached with Accutase (Sigma–Aldrich). For the differentiation into midbrain dopaminergic neurons, the medium was changed to N2B27, supplemented with 100 ng/ml FGF8, 1 μM purmorphamine, and 200 μM ascorbic acid for 7 days with media changes every other day. On day 8, medium was changed to maturation medium, consisting of N2B27 medium with 10 ng/ml brain-derived neurotrophic factor, 10 ng/ml glial cell line–derived neurotrophic factor, 1 ng/ml transforming growth factor beta 3, 500 μM cAMP, 0.5 μM PurMA, and 200 μM ascorbic acid. For the final replating on day 9, the maturation medium was supplemented with 10 μM Rho-associated protein kinase inhibitor Y-27632(2HCl). On day 11, PurMA was removed from the medium, and cells were cultivated until the 16th day of maturation, with media changes every other day.

Probe Treatment in Living Cells

Cells were seeded at a density of 2.5 × 10⁶ cells per 10 cm dish in 10 ml of culture medium. Following plating, cells were either treated with probe (100 μM pro-N6pA) or an equivalent volume of dimethyl sulfoxide (DMSO) as a control. After the addition of the probe and, if required, 1 µM Monensin, cells were incubated for 16 h before being harvested. Harvesting was performed by washing the cells twice with 2 ml of ice-cold Dulbecco's PBS, scraping them in 1 ml of Dulbecco's PBS, and centrifuging at 250g for 4 min at 4 °C. Following centrifugation, the supernatant was removed, and the cell pellets were either frozen at −80 °C or immediately subjected to cell lysis.

Cell Lysis

Cells were lysed using 100 to 200 μl of lysis buffer (1% NP-40, 0.2% SDS, 20 mM Hepes, pH 7.5) via rod sonication (10 s, 20% intensity). Cells intended for subsequent immunoprecipitation (IP) were lysed in 1 ml of lysis buffer (50 mM Tris–HCl, 150 mM NaCl, 2.5% Tween-20, pH 7.4) by shaking for 30 min at 4 °C. The lysates were then clarified by centrifugation (13,000g, 4 °C, 10 min), and the supernatant was transferred into a new 1.5 ml tube. Lysates were either stored at −80 °C or processed immediately.

Protein Concentration Measurement

Protein concentration measurement was performed with a Pierce Bicinchoninic Acid (BCA) Protein Assay Kit (Thermo Scientific).

Transient Overexpression in Human Cell Culture

Cells were seeded with either 2.5 × 10⁶ cells in 10 ml of culture medium in a 10 cm dish or 5 × 10⁶ cells in 15 ml for a 15 cm dish, respectively. Seeded cells were then incubated for 24 h at 37 °C with 5% CO₂. The following day, a transfection mixture was prepared by dissolving either 10 μg or 15 μg of plasmid DNA in 1 ml or 1.5 ml of serum-reduced medium (Opti-MEM). After mixing, 30 μl or 45 μl of polyethyleneimine (PEI) was added. The mixture was mixed again and incubated for 20 min at room temperature. Meanwhile, the cell medium was replaced with fresh, prewarmed medium before adding the transfection mixture. The cells were then incubated for another 24 h at 37 °C with 5% CO₂. On the subsequent day, cells were either harvested as described or treated with probe/DMSO. Control samples that are not transfected are referred to as “untreated” (UT) samples.

Depletion of FICD and SELENOO in Hap1 Cells

Hap1 cell lines depleted for FICD or SELENOO were generated by utilizing the CRISPR–Cas9 system. Hap1 cells were seeded to 4 × 10⁵ cells per 24 well and transiently transfected 24 h later using Turbofectin (OriGene) and OptiMEM. For FICD, the pX330 vector (Addgene; catalog no.: 42230) containing a single-guide RNA (sgRNA) (GGCGGTGACTGAACCGAAAT) targeting FICD exon 1 was cotransfected with a vector containing a blasticidin resistance cassette flanked by TIA sites derived from zebrafish, as described previously (47). The next day, cells were selected with 25 μg/ml blasticidin (Invivogen). Cells were selected again 4 days post transfection to ensure integration of the blasticidin resistance cassette. Clonal progeny was generated, and knock-in of the blasticidin resistance cassette into the FICD gene was confirmed by PCR amplification of genomic DNA and parts of the resistance cassette using the following primers: 5′-GCTGCAAACAGCTAATGCACATTG-3′, 5′-GCGGGCCATTTACCGTAAGTTATG-3′, 5′-TCTTGGCTCCTTGCAGATCC-3′. Knock-in was confirmed by Sanger sequencing. SELENOO-depleted cells were constructed by cotransfection of two pX330 vectors with the sgRNAs (CCGTAGGACCTTGCGACCGT, GCCAGGCTGCCCTATACACT) targeting the second and last exons of SELENOO, respectively. A vector conferring transient resistance toward puromycin was cotransfected. About 24 h post transfection, cells were selected with 1 μg/ml puromycin. Clonal progeny was generated and genotyped by PCR using the primers 5′-GTTACACGCGTTCCCTCCTT-3′ and 5′-GGGGGAACAATGACCACAGAG-3′. Genotypes were confirmed by Sanger sequencing.

Lentiviral Transduction

Lentiviruses were generated by transfecting HEK293T cells with pFUGW_Blasticidin, containing the different transgenes, and the helper plasmids, pMDLg/pRRE, pRSV-rev, and pCMV-VSV-G, using PEI Max. After 72 h, the lentiviral supernatants were harvested, and PLD3-deficient Cal-1 cells were transduced for 48 h and afterward selected with blasticidin (10 μg/ml). The polyclonal cell population was then used for further experiments.

Cal-1 Stimulation

Cal-1 cells (100,000 cells/well in a 96-well plate) were primed with hIFN-γ (10 ng/ml) for 6 h and afterward stimulated with 1 μg/ml R848, 5 μM CpG^S, or 5 μM CpG^O DNA. Supernatants of Cal-1 cells were harvested after 16 h of incubation at 37 °C and afterward used to measure hIFN-β by ELISA.

Enzyme-Linked Immunosorbent Assay

hIFN-βTW with PLD3 ELISA was conducted according to the supplier’s protocol (Novus Biologicals).

Immunoblotting of PLD3 WT and PLD3 Mutants Expressed in Cal-1 Cells

To confirm the expression of PLD3 WT and PLD3 mutants in PLD3-deficient Cal-1 cells, 1 × 10⁶ cells were lysed in DISC buffer (150 mM NaCl, 50 mM Tris [pH 7.5], 10% glycerol, and 1% Triton X-100) supplemented with cOmplete protease inhibitor cocktail for 10 min on ice and afterward centrifuged for 10 min at 16,000g. The supernatants were collected, 6× Laemmli buffer was added (60 mM Tris [pH 6.8], 9.3% DTT [w/v], 12% SDS [w/v], 47% glycerol [v/v], and 0.06% bromophenol blue [w/v]), and the samples were denatured for 5 min at 95 °C. Subsequently, the samples were separated by Tris–lycine SDS-PAGE and transferred onto 0.45 µm nitrocellulose membrane. Membranes were blocked in 5% milk for 1 h at room temperature and afterward incubated with indicated primary and corresponding secondary antibodies. Chemiluminescent signals were recorded with a CCD camera.

Whole Proteome Analysis Using SP3 Workflow

Experimental Design and Statistical Rationale

To quantify protein levels, a full proteome analysis was conducted using DMSO-treated control lysates. Initially, 20 μg of protein from each lysate (in at least triplicates) was diluted to a total volume of 50 μl with lysis buffer. A 1:1 mixture of hydrophilic and hydrophobic carboxylate beads (Cytiva) were then equilibrated. For each sample, 20 μl of premixed carboxylate beads were placed into a 96-well plate and washed three times with 100 μl of H₂O (MS grade). Following the final wash, the lysates were directly added to the magnetic beads. Upon addition of 60 μl of absolute ethanol (EtOH), the beads were incubated for binding at room temperature for 5 min at 850 rpm. Subsequently, the 96-well plate was transferred to a Hamilton Microlab Prep pipetting robot to minimize pipetting errors. The plate was then placed onto a magnet to remove supernatant-containing unbound components. The magnetic beads were washed thrice with 80% EtOH in H₂O (MS grade) and once with acetonitrile (ACN). For each wash, the beads were incubated for 1 min at room temperature while shaking at 850 rpm in the washing solution, followed by placement of the plate on the magnet to remove the wash solution. Finally, the beads were dissolved in 60 μl of 100 mM ammonium bicarbonate buffer, and 1 μl of sequencing-grade trypsin (0.5 mg/ml; Promega) was added for overnight digestion at 37 °C while shaking at 650 rpm. The following day, the 96-well plate was returned to the magnet, and the supernatant containing the peptide mixture was transferred to a clean 1.5 ml tube. The beads were manually washed twice with 50 μl and 30 μl of 1% formic acid (FA), respectively. Each time, the plate was incubated for 5 min at 40 °C while shaking at 850 rpm. The supernatants were combined in the initial 1.5 ml tube, which was then reattached to the magnet. Finally, the supernatant was transferred to an MS vial and subjected to LC–MS/MS analysis. Unless mentioned otherwise, whole proteome samples were prepared in two independent experiments, each performed in technical triplicates derived from one biological replicate.

Enrichment of Modified Proteins Using Small-Scale SP2E Workflow

Experimental Design and Statistical Rationale

The procedure was carried out using the Hamilton Microlab Prep pipetting robot in a 96-well plate format (with at least duplicates for each condition) (32, 48). Initially, 100 μg of in cellulo probe–labeled (or DMSO control) lysates were diluted to 40 μl with lysis buffer. CuAAC-mediated click reaction mixture including 0.4 μl of biotin-N₃ (10 mM in DMSO) and 0.25 μl of TBTA (16.7 mM in DMSO) were then added. In addition, either 0.4 μl of Tris(2-carboxyethyl)phosphine hydrochloride (TCEP) (condition A: 100 mM in MS-grade water [MS-H₂O]) or 1.2 μl TCEP (condition B: 100 mM MS-H₂O) were added. Condition A was applied to experiments shown in Figures 2, A, D, E, 4, E, F, S7 and S15. Condition B was applied to experiments shown in Figures 2C, 4H, S3, S18, C and D. The reaction was initiated by adding 0.8 μl of CuSO₄ (50 mM in MS-H₂O) and incubating at room temperature, 600 rpm for 1.5 h. Following incubation, 40 μl of urea (8 M in MS-H₂O) was added to the click reaction to reach a final volume of 80 μl. A 1:1 mixture of hydrophobic and hydrophilic carboxylate-coated magnetic beads (Cytiva) was washed three times with 100 μl of MS-H₂O. The reaction mixture was then transferred onto 20 μl of the prewashed beads, and all further steps were carried out by the pipetting robot. The workflow starts with the addition of 100 μl absolute EtOH, mixing, and incubation at room temperature, 850 rpm for 5 min. Subsequently, the beads were washed three times with 150 μl of 80% EtOH in MS-H₂O and once with 150 μl MS-grade ACN. Proteins were eluted three times with 60 μl of 0.2% SDS in PBS at 40 °C, 850 rpm for 5 min from the carboxylate-coated beads, and transferred onto 50 μl pre-equilibrated streptavidin-coated magnetic beads (New England Biolabs) by a 3× wash with 100 μl of 0.2% SDS in PBS. For streptavidin–biotin complex formation, beads were incubated at room temperature, 800 rpm for 1 h with eluted proteins. Subsequently, beads were washed three times with 150 μl of 0.1% NP-40 in PBS, two times with 150 μl of 6 M urea (in MS-H₂O), and two times with 150 μl of MS-H₂O by incubating at room temperature, 800 rpm for 1 min between each wash step. After the last wash step, 50 μl of tetraethylammonium bromide buffer (50 mM in MS-H₂O) was added, and digestion was carried out overnight with 1.5 μl sequencing-grade trypsin (0.5 mg/ml, Promega) at 37 °C, 600 rpm. The next day, supernatants were transferred into new 1.5 ml tubes, and beads were washed manually once with 20 μl of tetraethylammonium bromide buffer (50 mM in MS-H₂O) and once with 20 μl of 0.5% MS-grade FA (in MS-H₂O) by incubating each time at 40 °C, 600 rpm for 5 min. To the combined fractions, 0.9 μl MS-grade FA was added, placed onto the magnet for 15 min, and transferred into new 1.5 ml tubes. These samples were placed onto the magnet for 2 min, the supernatant was transferred into MS vials to ensure that no beads were in the samples and subjected to LC–MS/MS measurement. If not mentioned otherwise, enrichment samples were prepared in two independent experiments, each performed in technical triplicates derived from one biological replicate.

Fig. 2.

Identification and validation of PLD3 AMPylation sites.A, HEK293T cells transiently overexpressing full-length PLD3 were treated with pro-N6pA and labeled with TAMRA-azide via CuAAC. The figure shows a fluorescence scan and Western blot of untreated (UT) control and PLD3-overexpressing HEK293T cells. The increased fluorescence signal around 55 kDa indicates AMPylation of the soluble PLD3 form. One representative blot of three independent experiments is shown. B, representative MS² spectra of newly identified AMPylation sites Y323 (HR, DIA, 10 ppm), S380 (standard, DIA, 10 ppm), and S365 (HR, DIA, 10 ppm). In total, sites were identified in eight spectra for S365 (four containing natural AMPs and four with N⁶-propargyl AMP) and two independent replicates; eight spectra for Y323 (eight containing natural AMPs) in three independent replicates; two spectra for S380 (two natural AMPs) and three independent replicates (Fig. S4). C, fluorescence scan of PLD3 single point mutants, Y323A, S365A, and S380A, as well as the triple mutant. Strong or complete depletion of pro-N6pA probe labeling for all three single and the triple mutant corroborated validates these three as the main AMPylation. The Western blot of PLD3 shows that Y323A and triple mutant fails to produce the soluble PLD3 form. One representative blot of two independent experiments is shown. D, AMPylation profiling indicates the decrease in AMPylation of PLD3 variants (Fig. S7). E, fluorescence-based analysis of the PLD3 Y323F point mutant shows that in this mutant, AMPylation on the other sites is retained. One representative gel of two independent experiments is shown. F, PLD3 Y323F retains formation of the soluble form but without catalytic activity (Fig. 2E). One representative blot of two independent experiments is shown. G, box plot visualizing the decrease in PLD3 exonuclease catalytic activity, n = 3. To eliminate background exonuclease activity, fluorescence intensity was corrected by subtracting the signal from untreated cells. DIA, data-independent acquisition; HEK293T, human embryonic kidney 293T cell line; HR, high resolution; PLD3, phospholipase D3; pro-N6pA, N⁶-propargyl adenosine phosphoramidate.

Fig. 4.

FICD-mediated AMPylation accelerates PLD3 degradation in HEK293T cells.A, proteomics analysis of PLD3 level changes in coexpression experiment of PLD3 with FICD variant. Of note, FICD was found and quantified in all replicates of FICD-overexpressing cells but only in one replicate from untreated cells. Representative result from two independent runs. Samples were prepared in triplicates (n = 3) from one protein lysate. B, Western blot using anti-PLD3 of whole cell lysate confirms decreasing amount of full-length and soluble PLD3 when coexpressed with FICD E234G. One representative blot of three independent experiments is shown. C, volcano plots visualizing changes in protein expression on whole proteome level in HEK293T cells overexpressing FICD variants. Representative result from two independent runs. Samples were prepared in triplicates (n = 3) from one protein lysate. D, profile plots of selected upregulated and downregulated proteins upon FICD variant overexpression. E, pro-N6pA-based enrichment of AMPylated proteins followed by Western blot using anti-PLD3. One representative blot of two independent experiments is shown. F, pro-N6pA-based enrichment of AMPylated proteins followed LC–MS/MS analysis. The bar plots show the fold enrichment of AMPylated proteins. Representative result from two independent runs. Samples were prepared in triplicates (n = 3) from one protein lysate. G, box plot visualizing the drop of PLD3 exonuclease catalytic activity when coexpressed with FICD variants, n = 3. To eliminate background exonuclease activity, fluorescence intensity was corrected by subtracting the signal from untreated cells. H, SDS-PAGE-based analysis of pro-N6pA reported AMPylation upon coexpression of PLD3 WT with FICD variants. One representative blot of two independent experiments is shown. FICD, protein adenylyltransferase; HEK293T, human embryonic kidney 293T cell line; PLD3, phospholipase D3; pro-N6pA, N⁶-propargyl adenosine phosphoramidate.

Enrichment of Modified Proteins Using Large-Scale SP2E Workflow

Experimental Design and Statistical Rationale

About 400 μg of proteins from in cellulo control or probe-treated lysates were diluted to 200 μl with lysis buffer (each condition prepared in at least duplicates). CuAAC-mediated click reagents including 2 μl of biotin-PEG-N₃ (10 mM in DMSO), 6 μl of TCEP (100 mM in MS-H₂O), and 0.25 μl of TBTA (83.5 mM in DMSO) were then added, vortexed, and briefly spun. The reaction was initiated by adding 4 μl of CuSO₄ (50 mM in MS-H₂O) and incubating at room temperature, 600 rpm for 1.5 h. After completion of the incubation, 200 μl of urea (8 M in MS-H₂O) was added to the click reaction to reach a final volume of 400 μl. A 1:1-mixture of hydrophobic and hydrophilic carboxylate-coated magnetic beads (Cytiva) was washed three times with 500 μl of MS-H₂O, and the reaction mixture was then transferred onto the 50 μl per sample of prewashed beads, followed by the addition of 600 μl absolute EtOH, mixing, and incubation at room temperature, 950 rpm for 5 min. The beads were washed three times with 500 μl of 80% EtOH in MS-H₂O. Proteins were eluted twice with 500 μl of 0.2% SDS in PBS at room temperature, 950 rpm for 5 min from the carboxylate-coated beads, and transferred onto 50 μl pre-equilibrated streptavidin-coated magnetic beads (New England Biolabs) by a 3× wash with 1 ml of 0.2% SDS in PBS. For streptavidin–biotin complex formation, beads were incubated at room temperature, 950 rpm for 1 h with eluted proteins. Subsequently, beads were washed three times with 500 μl of 0.1% NP-40 in PBS, two times with 500 μl of 6 M urea (in MS-H₂O), and two times with 500 μl of MS-H₂O by mixing and briefly spinning between each wash step. After the last wash step, 25 μl of 5× Laemmli buffer (10% [w/v] SDS, 50% [v/v] glycerol, 25% [v/v] β-mercaptoethanol, 0.5% [w/v] bromophenol blue, 315 mM Tris–HCl, 100 mM DTT, pH 6.8) was added and incubated at 95 °C for 5 min. Supernatants were transferred into new Eppendorf tubes and either stored at −80 °C or directly loaded onto an SDS-Gel for subsequent Western blot (WB) analysis. If not mentioned otherwise, enrichment samples were prepared in two independent experiments, each performed using one biological replicate.

Pulse-Chase Experiment Using pro-N6pA

To analyze the dynamics of AMPylation in living cells, time-dependent delabeling of the pro-N6pA probe was performed. HEK293T cells were seeded and treated with the pro-N6pA probe ("pulse") for 16 h, as described above. DMSO-treated cells served as controls. At defined time points (t = 0, 6, 24, 48, and 72 h) after the initial incubation, cells were analyzed ("chase"). To achieve this, samples were washed twice with PBS, and cells at each time point were harvested and stored at −80 °C. Fresh, prewarmed culture medium was added to the remaining cells, and incubation was continued. Control dishes were harvested at t = 0 h. After all samples were collected, cells were lysed, and protein concentrations were determined using a BCA assay. Enrichment analysis using small-scale SP2E workflow followed by LC–MS/MS analysis was then performed. The experiment was conducted once with biological triplicates.

PLD3 Exonuclease Activity Assay

Experimental Design and Statistical Rationale

For quantitative measurement of PLD3 5′ exonuclease activity, lysates were prepared in Tris-lysis buffer (Tris-buffered saline with 1% [v/v] Triton X-100 and 1 tablet of protease inhibitor). After collecting the whole cell lysates as described previously, lysates were diluted in triplicates to a final volume of 100 μl in MES reaction buffer (50 mM MES, 200 mM NaCl, pH 5.5). The final concentration was adjusted to 50 ng/μl for UT cells, 1 ng/μl for WT PLD3, and 5 ng/ml for PLD3 point mutants in a lumox multiwell 96 plate (Sarstedt). Each sample was measured in technical triplicates derived from one biological sample. The reaction was initiated by adding 100 pmol of quenched FAM-ssDNA substrate (6-FAM-ACCATGACGTTCCTG-BMN-Q535 from Biomers.net, with ∗ indicating a phosphorothioate bond). Fluorescence emission at 528 nm (excitation at 485 nm) was measured using a microwell plate reader (Tecan) over a 4-h period, with measurements taken every 5 min at 37 °C. For evaluation, a substrate control without lysate and a lysate control without substrate were measured alongside the samples. The corrected fluorescence intensity, Fc(t), was calculated by subtracting the mean fluorescence intensity of UT cells, thereby removing any background exonuclease activity as well as the values from the two controls at each time point from the mean fluorescence intensity.

$$F_c\left(t\right)=F_m\left(t\right)-F_{UT}\left(t\right)-\left(\frac{{dF}_s\left(t\right)}{dt}+\frac{{dF}_L\left(t\right)}{dt}\right)xt$$

With: F_C(t) = calculated sample intensity

F_m(t) = measured sample intensity

F_UT(t) = measured sample intensity from UT cells

dFI_S(t) = measured intensity of the substrate control

dF_L(t) = measured intensity of the lysate control

t = time

IP Using Anti-FLAG Affinity Gel

For the IP of overexpressed FLAG-tagged protein, anti-FLAG M2 Affinity Gel (Thermo Fisher) was used. The gel was equilibrated according to the manufacturer's protocol (Part II: Resin preparation). The IP procedure was carried out as described in the section “FLAG Fusion Protein Immunoprecipitation” of the manual. All centrifugation steps were carried out at 4 °C for 1 min and 6000g. Briefly, 40 μl of gel suspension per reaction (equivalent to 20 μl of packed gel volume) were used. As controls, either a reagent blank with no protein or protein lacking the FLAG-tag (negative control) was employed. Triplicates were prepared for each condition: duplicates for LC–MS/MS analysis and one replicate for WB analysis. For each reaction, 1 mg of in cellulo probe–treated (or DMSO control) lysates were added to the equilibrated gel matrix, and the volume was adjusted to 1 ml with lysis buffer (50 mM Tris–HCl, 150 mM NaCl, Tween-20, pH 7.4). The gel was incubated for 2 h at 4 °C with shaking to facilitate protein binding to the matrix. After incubation, the supernatant was removed, and the gel was washed twice with 0.5 ml of wash buffer (50 mM Tris–HCl, 150 mM NaCl) and twice with MS-H₂O. Next, an in-gel CuAAC click reaction was performed. To the gel, 1 μl of TAMRA-N₃ (10 mM in DMSO), 1 μl TCEP (100 mM in H₂O), and 0.125 μl TBTA (83.5 mM in DMSO) were added, and the mixture was adjusted to a final volume of 100 μl with 1× PBS. The reaction was initiated by adding 2 μl CuSO₄ (50 mM in H₂O) and incubated for 1.5 h at room temperature with mixing at 650 rpm. Samples intended for WB analysis were centrifuged at room temperature for 1 min at 6000g, the supernatant was discarded, and the proteins were eluted from the gel by adding 25 μl of 1× Laemmli buffer and heating for 5 min at 95 °C and 850 rpm. The supernatants were transferred to fresh tubes and loaded onto the SDS-Gel. Samples destined for LC–MS/MS analysis were centrifuged, the supernatant was discarded, and 80 μl ammonium bicarbonate buffer (125 mM in MS-H₂O) and 10 μl TCEP (100 mM in MS-H₂O) were added. The sample was heated to 56 °C for 30 min for reduction. Afterward, 10 μl chloroacetamide (400 mM in MS-H₂O) was added, and the mixture was incubated for 30 min at room temperature for alkylation. For digestion, 1.5 μl of sequencing-grade trypsin (Promega) was added, and the samples were incubated overnight at 37 °C. The next day, samples were vortexed and spun down for 30 s at room temperature at 6000g. Trypsin activity was quenched by acidification using 2 μl of 100% MS-grade FA. Samples were desalted using 50 mg SepPak C18 columns. The columns were equilibrated by adding 1 ml of 100% ACN, followed by 1 ml 80% ACN/0.5% FA in MS-H₂O, and finally three times with 1 ml 0.5% FA in MS-H₂O. The sample was then loaded and washed three times with 1 ml 0.5% FA in MS-H₂O (gravity flow). The column was removed, the aperture was dried by applying a vacuum, and finally, the proteins were eluted into fresh 1.5 ml tubes by adding 250 μl 80% ACN/0.5% FA in MS-H₂O (gravity flow) followed by 250 μl 80% ACN/0.5% FA in MS-H₂O (vacuum). The peptide mixture was dried in the SpeedVac for 2.5 h at 35 °C, and the dried peptide mixture was dissolved in 30 μl 1% FA before transfer into MS vials. Samples were analyzed using short and long HPLC gradients combined with high-resolution MS and MS/MS measurement and the data-independent acquisition (DIA) method. If not mentioned otherwise, IP samples were prepared in two independent experiments, each performed in biological triplicates. For each condition (pro-N⁶pA probe or DMSO control) on each occasion, duplicates were utilized for LC–MS/MS measurements, whereas one replicate was allocated for SDS-Gel analysis. In addition, the duplicates intended for LC–MS/MS analysis were measured three times.

Endoglycosidase H-Digest

For cleaving N-linked glycosylation of glycoproteins, recombinant Endoglycosidase H (NEB) was employed. Glycoprotein (20 μg), 1 μl of 10X Glycoprotein Denaturing Buffer, and H₂O were combined to create a total reaction volume of 10 μl. Glycoproteins were denatured by heating the reaction at 100 °C for 10 min. Subsequently, 2 μl of 10X GlycoBuffer 3 and 1 μl of EndoH were added, and the mixture was filled up to 20 μl with H₂O. The reaction mixture was then incubated for 1 h at 37 °C, and successful deglycosylation was assessed by SDS-PAGE and subsequent WB analysis.

PNGaseF Digest

For removing almost all N-linked oligosaccharides from glycoproteins, PNGaseF (NEB) was employed. Glycoprotein (20 μg), 1 μl of Glycoprotein Denaturing Buffer (10X), and H₂O were combined to make a 10 μl total reaction volume. Glycoproteins were denatured by heating the reaction at 100 °C for 10 min. Denatured samples were spun down, and 2 μl GlycoBuffer 2 (10×), 2 μl 10% NP-40, and 6 μl H₂O were added to make a total reaction volume of 20 μl. Then, 1 μl PNGase F was added, the reaction mixture was gently mixed and incubated for 1 h at 37 °C, and successful deglycosylation was assessed by SDS-PAGE and subsequent WB analysis.

Chemical Transformation of Bacterial Cells

For chemical transformation, chemically competent E. coli DH5-alpha cells stored at −80 °C were thawed for 10 min on ice. Then, 50 ng of plasmid DNA or 5 μl of the Kinase, Ligase, and DpnI (KLD) reaction mixture (see Site-directed mutagenesis section) was added directly to 50 μl of cells, and the tube was gently flicked three times. The cells were incubated for 30 min on ice and then exposed to 42 °C for 30 s, a process known as heat shock. The cells were immediately placed on ice for 5 min for regeneration before the addition of 950 μl of SOC medium (NEB) and incubation for 1 h at 37 °C at 180 rpm. In the meantime, LB-Agar (1.3% Agar) plates supplemented with 100 μg/ml ampicillin were prewarmed at 37 °C. After incubation, 100 μl of the cell suspension was spread onto prewarmed LB-Agar plates and evenly distributed using glass beads by shaking the plate under sterile conditions. The LB-Agar plates were then incubated overnight at 37 °C and subsequently stored at 4 °C or directly used for plasmid amplification.

Plasmid DNA Isolation Using Mini/Midi Preparation Kit

For plasmid DNA isolation, either the Mini (small scale) or Midi (large scale) preparation kit from NEB was utilized. For Mini or Midi preparation, 3 ml or 100 ml, respectively, of sterile LB medium supplemented with 100 μg/ml ampicillin was prepared. A single clone from the overnight LB agar plate was selected using a pipette tip and subsequently added to the medium. The medium was then incubated for 12 to 16 h at 37 °C at 180 rpm. Plasmid DNA was isolated following the manufacturer's protocol for the Plasmid Miniprep/Midiprep Kit (NEB). The DNA was eluted in deionized H₂O. DNA concentration was measured using a Nanodrop, and the DNA sequence was confirmed by Sanger sequencing conducted by Genewiz.

Q5 Site–Directed Mutagenesis

For introduction of point mutations to plasmids, the Q5 site–directed mutagenesis kit (NEB) was used. To validate the AMPylation sites in PLD3, point mutants of the appropriate sites were prepared. For this purpose, a primer pair containing the mutated amino acid for each of the three potential AMPylation sites (Y323, S365, and S380) was designed. The triple mutant containing all three mutations was prepared in an iterative manner. Here, purified DNA template of the first point mutant previously confirmed by sequencing was used for the introduction of the second- and third-point mutation by repeating the described steps with different primer pairs. The forward and reverse oligonucleotides as well as the applied annealing temperature (T_a) for cloning the different PLD3 construct are listed in Table S1. The soluble form of PLD3 was cloned starting from amino acid position 61 and adding an N-terminal IgK-signal peptide using the full-length WT PLD3 with a C-terminal FLAG(3×)-tag as a template. For cloning the catalytically inactive point mutant FICD H363A, the following primer pair was used together with the template containing WT FICD. For introduction of C-terminal FLAG(3×)-tag to FICD WT or mutants (H363A and E234G), a primer pair containing the FLAG(3×)-tag sequence (5′-DYKDHDGDYKDHDIDYKDDDDK-3′) was prepared. For all constructs, the sequence identity was validated by Sanger sequencing (Genewiz) using the appropriate sequencing oligonucleotides listed in Table S2.

Polymerase Chain Reaction

PCR was performed for cloning purposes using plasmid DNA as a template. Here, the procedure was performed according to the manufacturers’ protocol (Q5 site–directed mutagenesis). The thermocycling conditions applied are listed in Table S3. PCR product formation was validated by 1% agarose gel electrophoresis.

KLD Reaction

After PCR, KLD reaction was performed for circularization of the linear product based on the manufacturers’ protocol (Q5 site–directed mutagenesis). Here, the PCR product is first phosphorylated (kinase), then ligated (ligase), and residual template DNA is decomposed by DpnI enzyme (only methylated DNA is a substrate). The reaction mixture was prepared, mixed by pipetting up and down 5 to 10 times, and incubated for 5 min at room temperature. The KLD mixture was used for subsequent bacterial transformation.

1%-Agarose Gel

For validation of PCR product formation, 1% agarose gel electrophoresis was conducted. The agarose was dissolved in 40 ml 1× 40 mM Tris, 20 mM acetic acid, 1 mM EDTA buffer, and heated up in the microwave. After cooling down, 2 μl Gel stain was added and the gel was polymerized. As a standard, 1 kB DNA ladder (Carl Roth) was used. To the PCR product, 6× stain (NEB) was added (c_f = 1×). The electrophoresis was conducted for 1 h at 100 V in 0.5× Tris, acetic acid, and EDTA buffer. PCR products were detected under UV light.

WB Analysis

For WB analysis, 20 μg of cell lysate was used. Proteins were denatured using 5× Laemmli buffer (10% [w/v] SDS, 50% [v/v] glycerol, 25% [v/v] β-mercaptoethanol, 0.5% [w/v] bromophenol blue, 315 mM Tris–HCl, 100 mM DTT, pH 6.8) and diluted to a final concentration of 1×. Then, samples were boiled for 5 min at 95 °C and subsequently loaded onto a 10% SDS-Gel, thereby allowing protein separation by size. The separated proteins were transferred onto a polyvinylidene fluoride membrane in a semidry manner. For this purpose, the SDS-Gel and filter paper were equilibrated in transfer buffer (48 mM Tris, 39 mM glycine, 0.0375% [m/v] SDS, and 20% [v/v] methanol) for 5 min at room temperature. The polyvinylidene fluoride membrane was first activated for 1 min in MeOH and then equilibrated in transfer buffer for 5 min at room temperature. To allow transfer, the filter paper was placed on the bottom with the membrane on top followed by the SDS-Gel and another filter paper on top. Then, the protein transfer was carried out for 30 min at 25 V using a semidry blotter (Bio-Rad). Afterward, nonspecific binding sites were blocked by incubating the membrane for 1 h at room temperature in blocking solution (0.5 g milk powder in 10 ml PBST [PBS/0.5% (v/v) and Tween]). Subsequently, 10 ml of primary antibody diluted in blocking solution with specificity for the protein of interest was added, and the mixture was incubated for 1 h up to overnight at 4 °C. The membrane was washed three times for 10 min with PBST before 1 ml of the secondary antibody diluted in 10 ml of blocking solution was added. As loading control, fluorophore-coupled anti-GAPDH antibody was added during the secondary antibody incubation. After 1 h of incubation at room temperature in the dark, the membrane was washed three times for 10 min with PBST while shaking at room temperature. Then, a 1:1 mixture composed of 400 μl of ECL substrate and 400 μl peroxide solution was prepared and added to the membrane for chemiluminescence signal detection. Finally, images of the WB were taken by developing using the Amersham Imager 680 (GE Healthcare) machine.

Lysosomal IP

For isolation of the lysosomal fraction using antihemagglutinin (HA) beads, lysosomal membrane protein TRPML1 with HA-tag was overexpressed in HEK293T cells. For each condition, two 150 cm dishes were seeded with each 5 × 10⁶ cells. Cells were incubated for 24 h at 37 °C and 5% CO₂. The next day, cells were transfected using 15 μg plasmid DNA containing TRPML1 gene with 3x HA tag and again incubated for 24 h at 37 °C and 5% CO₂. As a control, UT (not transfected) HEK293T cells were used. The next day, cells were washed twice with ice-cold PBS and collected by scraping in 7 ml KPBS I buffer (10 mM KH₂PO₄, 136 mM KCl, pH 7.25, freshly added protease inhibitor tablet) per plate into a fresh 15 ml falcon. The falcon was spun down for 5 min at 200g and 4 °C. The supernatant was discarded, and the pellet was resuspended in 1 ml KPBS I buffer. Then, 80 μl of magnetic anti-HA beads per condition were equilibrated thrice with KPBS I buffer by pipetting up and down. Beads were resuspended in 80 μl KPBS I buffer. Cell suspension was transferred to a tissue grinder (VWR) to ensure cell lysis by preserving organellar membrane. For this, cells were homogenized by 30× strokes, thereby avoiding air bubble formation. Homogenate was transferred to fresh 1.5 ml tubes. An aliquot of 20 μl was used as “whole cell lysate (L)” fraction and diluted with 20 μl lysis buffer and 5 μl 5× Laemmli for subsequent SDS-PAGE analysis. The homogenate was centrifuged for 2 min at 4 °C and 2000g, and the supernatant containing lysosomal fraction was transferred directly to equilibrated beads and resuspended by pipetting twice up and down. To allow binding, beads were incubated for 10 min at 4 °C under continuous motion. The tubes were placed on a magnet, and the supernatant containing the “flow through” fraction was removed after taking an aliquot of 20 μl and diluting with 20 μl lysis buffer and 5 μl 5× Laemmli. Next, beads were washed twice with 500 μl KPBS III buffer (100 mM KH₂PO₄, 25 mM KCl, 150 mM NaCl [pH 7.2], freshly added protease inhibitor tablet) and once with KPBS II buffer (100 mM KH₂PO₄, 25 mM KCl, pH 7.2, freshly added protease inhibitor tablet). After each washing step, the mixture was transferred to a fresh 1.5 ml tube to avoid contamination of unbound fraction. Finally, the last washing solution was aspirated, and the proteins were eluted from the beads with 120 μl elution buffer (KPBS I buffer, 0.5% NP-40, freshly added protease inhibitor tablet). Elution occurred while incubating beads at 4 °C for 30 min while in continuous motion. Tubes were placed on the magnet, and the supernatant was transferred to fresh 1.5 ml tubes and spun down for 10 min at 4 °C and 13,000g. The supernatant was transferred to fresh 1.5 ml tubes, and a 40 μl aliquot of “IP” fraction was diluted with 10 μl 5× Laemmli for subsequent SDS-PAGE analysis. Eluates of lysosomal immunoprecipitation (LysoIP) were stored after snap-freezing in liquid N₂ at −80 °C. Samples for SDS-PAGE analysis were heated up to 95 °C for 5 min, and 15 μl of each fraction was loaded onto a 10% SDS-Gel followed by WB analysis. For whole proteome analysis via LC–MS/MS, protein concentration of eluates was measured via BCA assay and continued with SP3 workflow (as described). The experiment was conducted at two different time points, preparing biological duplicates for TRPML1-HA–overexpressing cells and one replicate for UT control cells on each occasion. LysoIP samples were prepared in duplicates from each lysate for LC–MS/MS measurements. One replicate was allocated for SDS-Gel analysis.

In-Gel Fluorescence Analysis

For in-gel fluorescence analysis, 100 μg of in cellulo probe–labeled or DMSO-control lysates were diluted to 100 μl with lysis buffer (1% NP-40, 0.2% SDS, 20 mM Hepes, pH 7.5). Click reagents, such as 1 μl of TAMRA-N₃ (10 mM in DMSO), 3 μl of TCEP (100 mM in MS-H₂O), and 0.125 μl of TBTA (83.5 mM in DMSO), were added, vortexed, and spun down. The reaction was initiated by the addition of 2 μl of CuSO₄ (50 mM in MS-H₂O), and the mixture was incubated at room temperature while shaking at 600 rpm for 1.5 h. Proteins were precipitated with 400 μl precooled acetone (4:1 ratio) at −20 °C for at least 1 h and were pelleted for 10 min at 4 °C and 13,000g. Protein pellet was reconstituted in 80 μl 0.2% SDS in PBS. The total protein lysate (20 μg) per sample was boiled for 5 min at 95 °C with 5× Laemmli buffer, loaded and run with 150 V for 1 h in 1× running buffer (25 mM Tris, 0.192 M glycine, and 0.1% [m/v] SDS) on a 10% SDS-PAGE gel to electrophoretically separate the proteins. In-gel fluorescence was scanned with an Amersham Imager 680 (GE Healthcare). For loading control, proteins were stained with a Coomassie staining solution (0.25% Coomassie Blue R-250, 10% acetic acid, and 50% MeOH) and properly destained with destaining solution (20% MeOH, 10% acetic acid).

Purification of Recombinant PLD3 WT, PLD3(S365A), and PLD3(S380A) Mutants

Purification of recombinant PLD3(S365A) and PLD3(S380A) was conducted as previously described (PMID: 38697119). In brief, PLD3 mutants lacking the N-terminal and the transmembrane domains were fused to the Igk leader sequence, followed by a 6xHistag and a HRV3C cleavage site, and cloned into the piggyBac vector system (49). Constructs were electroporated into RCH-ACV cells using a Gene Pulser device (Bio-Rad), and stable pools expressing the mutants were selected with blasticidin (10 mg/ml) and puromycin (2.5 mg/ml). Protein expression was induced with doxycycline (1 mg/ml), after cells reached a density of 4 × 10⁶ cells/ml. Supernatants containing the mutants were collected after 5 days, filtered, and rotated for 2 h at 4 °C in the presence of nickel–nitrilotriacetic acid (agarose beads. Subsequently, nickel–nitrilotriacetic acid beads were washed three times with wash buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole, pH = 8.0), and afterward, PLD3 mutants were eluted from the beads (50 mM NaH₂PO₄, 300 mM NaCl, 300 mM imidazole, pH = 8.0). Buffer was exchanged using an Amicon (cutoff 10 kDa) into a buffer containing 30 mM Hepes, 100 mM NaCl, pH = 7.2.

PLD3 Dimerization Assay

Mass photometric analysis was used to analyze in vitro dimerization of recombinant WT PLD3, PLD3(S365A), and PLD3(S380A) mutants, using a Refeyn TwoMP mass photometer. Prior to each measurement, proteins were diluted to a final concentration of 50 nM in sterile-filtered buffer composed of 50 mM NaOAc and 100 mM NaCl, pH 4.5. Each measurement was recorded as a 60s movie, and the resulting data were processed using the AcquireMP software. Each measurement was conducted twice.

In Vitro DNase Assays

CpG^O-DNA (100 ng) was incubated with indicated amounts of enzymes in assay buffer (50 mM NaAc, 100 mM NaCl, pH = 4.5) for 20 min at 37 °C. Subsequently, 2× RNA loading dye was added, and the mixture was heated at 95 °C for five additional minutes. The resulting fragments were separated and detected on a urea gel. Each DNase assay was conducted twice.

Urea Gels

DNA degradation products of PLD3 and mutant PLD3’s were visualized on urea gels. Gels were prepared according to the manufacturer’s instructions using SequaGel Concentrate, SequaGel Diluent, and SequaGel buffer. The gels were run at 250 V for 60 min in 1× TBE buffer (containing 100 mM Tris, 100 mM boric acid, and 2 mM EDTA). Afterward, the gels were stained with SYBR Gold Nucleic Acid Gel Stain for 5 min, followed by imaging.

Cell Imaging

Semiconfluent HEK293T cell cultures were grown on coverslips (ibidi, 8-well μ-slide). Transient overexpression of FLAG-tagged PLD3 WT or point mutants (Y323A, Y323F, S365A, and S380A) and the triple mutant (Y323A–S365A–S380A) was performed using PEI, as described previously. After 24 h of incubation, cells were washed with PBS+ (containing MgCl₂ and CaCl₂). Cells were then fixed at room temperature for 20 min using a fixation solution (4% formaldehyde/4% sucrose in PBS+), followed by a single PBS+ wash. Residual formaldehyde was quenched with quenching solution (50 mM NH₄Cl in PBS+) for 10 min at room temperature. After three additional PBS+ washes, cells were permeabilized with 0.1% Triton X-100 in PBS+ for 15 min at room temperature. To block nonspecific binding, cells were incubated in a blocking solution (2% bovine serum albumin [BSA]/2% fetal bovine serum in PBS+) for 30 min at room temperature. Meanwhile, primary antibodies were diluted in 1% BSA/0.05% Triton X-100 and incubated with the cells overnight at 4 °C. The next day, cells were washed three times with PBS+ before incubation with the corresponding secondary antibody (diluted in 1% BSA/0.05% Triton X-100) for 1 h at room temperature in the dark. After two final PBS+ washes, cells were mounted with a drop of ibidi mounting medium. Imaging was performed using a Leica Laser Scanning Confocal Microscope with a 100× lens (numerical aperture: 1.40). Image acquisition was conducted using Leica Las X software (Leica Microsystems GmbH).

LC–MS/MS Measurement

MS measurements were conducted using an Orbitrap Eclipse Tribrid Mass Spectrometer (Thermo Fisher Scientific) connected to an UltiMate 3000 Nano-HPLC (Thermo Fisher Scientific) via a Nanospray Flex (Thermo Fisher Scientific) and field asymmetric ion mobility spectrometry (FAIMS) interface (Thermo Fisher Scientific). Peptides were initially loaded onto an Acclaim PepMap 100 μ-precolumn cartridge (5 μm, 100 Å; 300 μM ID × 5 mm; Thermo Fisher Scientific), followed by separation at 40 °C on a PicoTip emitter (noncoated, 15 cm, 75 μm ID, 8 μm tip; New Objective) packed in-house with Reprosil-Pur 120 C18-AQ material (1.9 μm, 150 Å; Dr A. Maisch GmbH). LC buffers comprised MS-H₂O (A) and ACN (B), both supplemented with 0.1% FA. The short gradient ran from 4 to 35.2% B over 60 min (0–5 min: 4%, 5–6 min: 7%, 7–36 min: 24.8%, 37–41 min: 35.2%, 42–46 min: 80%, and 47–60 min: 4%) at a flow rate of 300 nl/min. The long gradient spanned 105 min, with a B gradient from 4 to 35.2% (0–5 min: 4%, 5–6 min: 7%, 6–75 min: 24.8%, 75–90 min: 35.2%, 90.1–95 min: 80%, and 95–95.1–105 min: 4%) at a flow rate of 300 nl/min.

Data-Independent Acquisition

FAIMS was conducted with a single CV set at −45 V. Each DIA cycle consisted of 1 MS¹ scan followed by 30 MS² scans. The mass spectrometer operated in DIA mode with the following parameters: Polarity: positive; MS¹ Orbitrap resolution: 60 k for standard or 120 k for high resolution; MS¹ automatic gain control (AGC) target: standard; MS¹ maximum injection time: 50 ms; MS¹ scan range: m/z 200 to 1800 or if mentioned m/z 120 to 1700; RF lens: 30%; precursor mass range: m/z 500 to 740; isolation window: m/z 4; window overlap: m/z 2; MS² Orbitrap resolution: 30 k or 60 k for high resolution; MS² AGC target: 200%; MS² maximum injection time: auto; higher-energy collisional dissociation (HCD) collision energy: 35%; RF lens: 30%; and MS² scan range: auto.

Data-Dependent Acquisition

FAIMS was performed with two alternating CV settings, −50 V and −70 V. The mass spectrometer operated in dd-MS² mode with the following parameters: Polarity: positive; MS¹ Orbitrap resolution: 240 k; MS¹ AGC target: standard; MS¹ maximum injection time: 50 ms; MS¹ scan range: m/z 375 to 1500; RF lens: 30%; MS² Orbitrap resolution: 15 k; MS² AGC target: standard; MS² maximum injection time: 35 ms; HCD collision energy: 30%; RF lens: 30%; MS² cycle time: 1.7 s; intensity threshold: 1.0e4 counts; included charge states: 2 to 6; and dynamic exclusion: 60 s.

Data-Independent Acquisition-Neural Network

First, ∗.raw files were converted to ∗.mzML format using “MSConvert” from the “ProteoWizard” software package (http://www.proteowizard.org/download.html) with the following settings: “peakPicking” filter with “vendor msLevel = 1”; “Demultiplex” filter with parameters “Overlap Only” and “mass error” set to 10 ppm. Subsequently, they were analyzed with DIA-NN 1.8.1, and peptides were searched against the UniProt database for Homo sapiens (taxon identifier: 9606) with included contaminants and decoys (193632 entries in total). The DIA-neural network (NN) settings were as follows: FASTA digest for library-free search/library generation: enabled; Deep learning–based spectra, retention times, and ion mobility prediction: enabled; missed cleavages: 2; maximum number of variable modifications: 3; modifications: N-term M excision, carbamidomethylation, oxidation (M), and N-term acetylation; Precursor charge range: 2 to 6; precursor range: m/z 500 to 740; fragment ion range: m/z 200 to 1800; precursor and protein false discovery rate (FDR) level: 1%; match between runs: enabled; library generation: smart profiling; quantification strategy: Robust LC; mass accuracy: 0; and scan windows: 0.

FragPipe

MS ∗.raw files were converted to ∗.mzML format with “MSConvert” with the following settings: “peakPicking” filter with “vendor msLevel = 1,” “titleMaker” filter was enabled. Peptides were searched in FragPipe 1.8 against UniProt database for H. sapiens (taxon identifier: 9606) with included contaminants and decoys (193632 entries in total). For protein digestion, trypsin (cuts at KR) was chosen with two missed cleavages. Due to the huge variety of settings, general workflows provided by FragPipe were used and subsequently modified accordingly. All results were viewed using the integrated FP-PDV viewer and directly from the psm.tsv results table. Results in the psm.tsv file were filtered for being “unique” and for q value “<0.1%.” In the following, most important features were listed for both, AMPylation and N-glycosylation site identification. Only hits fulfilling these parameters were taken into consideration.

AMPylation Site Identification Using MSFragger-DIA Search Engine

For AMPylation site identification, general workflows for DIA data, DIA_SpecLib_Quant and DIA_SpecLib-Umpire, offered by FragPipe were used and adjusted accordingly (Table S4).

N-Glycosylation Site Identification Using MSFragger-Glyco Search Engine

Glycosylation site identification was performed using template glycol workflow provided by MSFragger-Glyco. The following two basic workflows were applied and successfully resulted in detection of N-glycosylation sites: glyco-N-HCD and glyco-N-open-HCD (open search), both for collision-induced dissociation/HCD fragmentation of N-glycopeptides. Glycoproteomics results were viewed using the integrated FP-PDV viewer and directly from the psm.tsv results table.

Volcano Plot Preparation From Quantified Data (DIA-NN)

For statistical analysis, the “report.pg_matrix.tsv” table was used in Perseus 1.6.14.0.21 (Max Planck Institute of Biochemistry). Then, quantified values were log₂-transformed and columns were assigned with either control or probe treated. Subsequently, the groups were filtered based on the experimental background (at least two valid values out of three columns in at least one group or at least two valid values out of three columns in each group). Further, missing values were replaced from a normal distribution. -log₁₀(p values) were obtained by a two-sided one-sample Student’s t test over replicates with the initial significance level α = 0.05 adjustment by the multiple testing correction method of Benjamini and Hochberg (FDR = 0.05) using the volcano plot function. Finally, the volcano plot values from Perseus software (1.6.10.43) were transferred to OriginPro 2022 (9.9.5.171; OriginLab Corporation, Northampton) and visualized properly.

Principal Component Analysis

For principal component analysis (PCA), quantified values were log₂-transformed, and columns were assigned to the cell lines tested (Hap1 WT, Hap1 FICD KO, or Hap1 SELENOO KO) using Perseus software (1.6.10.43). Subsequently, the groups were filtered, and missing values were replaced based on normal distribution. PCA was conducted using the integrated tool in Origin 2022b (9.9.5.171).

Statistical analysis

All statistical analyses were conducted using Origin 2022b (9.9.5.171). For multiple group comparisons against each other, a one-way ANOVA with Tukey's range test was performed, also correcting for multiple comparisons to a significance level of α ≤ 0.05. For comparisons between only two groups, a two-sided t test was employed with a significance level of α ≤ 0.05.

Results

PLD3 Processing is Cell Type Dependent

We previously showed that in neurons, the soluble form is the major PLD3 proteoform (31). Here, we additionally compared the expression levels and proteoforms of PLD3 in several cell types (Fig. 1F). We observed that in cancer cell lines and plasmacytoid dendritic cells (Cal-1), the full-length, membrane-bound PLD3 is the main form, whereas in neurons undergoing neuronal differentiation as well as in fully differentiated neurons, the pool of PLD3 is mainly composed of proteolytically cleaved soluble PLD3 (Fig. 1F). Quantification of the total PLD3 levels revealed that on average 17-fold more PLD3 is present in neurons in comparison to other tested cell types (Fig. S2). Interestingly, the highest level of full-length PLD3 in non-neuronal cells was determined in Hap1 cells, a near haploid cell line derived from chronic myeloid leukemia (50).

AMPylation Delabeling is Substrate Dependent

Both AMPylation and deAMPylation can be catalyzed by FICD, which inherently complicates deconvolution of processes that may result in opposing downstream effects. In order to determine dynamics of protein deAMPylation, we performed pro-N6pA pulse-chase labeling in HEK293T cells followed by LC–MS/MS analysis. HEK293T cells were first incubated with pro-N6pA for 16 h. Afterward, the cell culture media were exchanged for probe-free media and cells were grown further for up to 72 h (Fig. S3 and Table S5). The cells collected directly after media exchange showed known AMPylated proteins as expected. However, PLD3 was completely deAMPylated after 6 h, whereas other AMPylated proteins, such as PPME1 and SCPEP1, were still significantly enriched after 24 and 48 h. Taken together, pro-N6pA pulse-chase delabeling experiment points toward protein-specific differences in deAMPylation kinetics, suggesting that AMPylation may be dynamically regulated in a protein target-dependent manner.

Multiple Sites Within the Soluble Region of PLD3 are AMPylated

Identifying AMPylation sites on endogenous proteins in living cells remains a significant challenge. Thus, it was unclear whether the absence of reported AMPylation sites on PLD3 reflects a true lack of modification, which was unlikely because of the functional relevance of this PTM, or whether it was due to technical challenges associated with detecting such sites, such as low abundancy of soluble AMPylated PLD3 in the cell lines used, resulting in a limited pool of detectable modified peptides. In addition, the poor ionization efficiency of phosphate group–containing modified peptides and the lability of N-glycosidic bond during fragmentation contribute to the challenge (24, 25, 27). To increase the amount of putatively AMPylated PLD3 peptides for MS analysis, we linked full-length PLD3 to a C-terminal 3xFLAG-tag (PLD3-FLAG) enabling efficient IP of the protein from whole cell lysates. To first test whether overexpression affected PLD3 processing or its pro-N6pA labeling pattern, PLD3-FLAG was transiently overexpressed in HEK293T cells, followed by treatment with either pro-N6pA or DMSO as a control. The following WB analysis confirmed the high expression levels of PLD3-FLAG protein. In-gel fluorescence analysis utilizing the CuAAC with TAMRA-azide to specifically label AMPylated proteins in whole cell lysates revealed a significant increase in the fluorescent band around 55 kDa, corresponding to luminal soluble PLD3 (Fig. 2A). Lysates from nontransfected cells, referred to as UT, were compared with those from PLD3-overexpressing cells. A strong fluorescence signal was observed in probe-treated samples, in contrast to the DMSO-treated control (Fig. 2A). Together, these results indicated proper processing of PLD3 and successful incorporation of pro-N6pA into PLD3. The combined WB and in-gel fluorescence analysis showed a notable increase in PLD3 AMPylation, confirming the feasibility of our approach.

Next, the overexpressed PLD3 was immunoprecipitated using anti-FLAG agarose beads followed by on-beads digest with trypsin for MS analysis. Digested samples were analyzed by different LC–MS/MS methods on Orbitrap Tribrid Eclipse using short (35 min) and long (90 min) nanoLC gradients coupled to MS analysis running in DIA or data-dependent acquisition mode combining standard high-resolution Orbitrap for MS¹ and faster low-resolution linear ion trap for MS². While the data-dependent acquisition approach did not reveal AMPylation sites, analysis of DIA data with MSFragger-DIA (51, 52) spectral search with natural AMPylation (+329.0525) or N⁶-propargyl AMP (+367.0682) on Ser, Thr, and Tyr as variable modification revealed multiple AMPylation sites on Y323, S365, and S380 of PLD3 with high confidence (Figs. 2B and S4). Natural AMP and the N⁶-propargyl AMP group could be detected at all sites showing even distribution of our AMP homolog in the target protein. While the N⁶-propargyl AMP was not found in DMSO-treated controls, with the exception of a single poor-quality spectrum for S365, which can be assigned as false positive because of known challenging FDR estimation for modified peptides and complexity of MS² spectra in DIA mode (53). To further confirm the identified AMPylation sites, a set of the newly prepared PLD3 IP samples were remeasured using DIA method with high-resolution of MS². With this approach, we confirmed that all sites contain AMP group. Interestingly, residue S380 was recently shown to be involved in homodimeric PLD3 interactions and to be located within the dimer interface, suggesting that the AMPylation status may modulate PLD3 catalytic activity (35). Furthermore, the Y323 site is in spatial proximity to a transmembrane domain, suggesting a potential role in regulation of PLD3 proteolytic cleavage (35).

To validate the identified AMPylation sites, PLD3 point mutants were generated in which the respective Ser or Tyr were replaced by alanine residue. The PLD3 point mutants were subsequently overexpressed in HEK293T cells and incubated with pro-N6pA to determine the AMPylation status by in-gel fluorescence. For PLD3 Y323A, the fluorescent signal based on pro-N6pA was completely depleted (Fig. 2C). For both, the S365A and S380A mutant, a strong decrease in fluorescent signal was observed pointing toward lack of AMPylation on these sites (Fig. 2C). The complementary WB analysis of the PLD3 forms revealed the absence of the luminal PLD3 for Y323A and the triple mutant containing all three point mutations, and somewhat lower levels for mutants S365A and S380A, thus changing the ratio between full length and soluble PLD3 (Fig. S5). The total PLD3 levels of all point mutants were comparable with overexpressed WT PLD3 as shown by WB and confirmed by proteomics (Figs. 2C and S5 and Table S6). Interestingly, while there was no major change in the whole proteome of cells expressing the single PLD3 point mutants, the expression of the triple PLD3 mutant exhibited strong dysregulation of whole proteome, with over 500 proteins being significantly upregulated and 160 downregulated (Fig. S5). The Gene Ontology term search points toward changes in RNA metabolic processes, RNA processing, and ribonucleoprotein complex biogenesis (Fig. S6). The overall decrease of AMPylation on PLD3 was confirmed by pull-down of AMPylated proteins using the pro-N6pA probe, coupled via CuAAC with biotin, followed by affinity enrichment and LC–MS/MS analysis (SP2E, Figs. 2D and S7, Tables S7 and S8). Of note, HEK293T cells also contain endogenous PLD3.

The unexpected complete depletion of PLD3 AMPylation on all identified sites in the single-point mutant Y323A led us to construct an additional point mutant exchanging Y323 to structurally more similar phenylalanine. In this mutant, AMPylation at residues S365 and S380 could be partially restored (Fig. 2E). The proteolytic processing of the PLD3 Y323F mutant was restored as well, as observed by WB analysis (Fig. 2F). Taken together, the described MS-based experiments provided direct confirmation of PLD3 AMPylation and identified three modified amino acid side chains, Y323, S365, and S380, within the soluble domain of PLD3.

AMPylation is Necessary for PLD3 Catalytic Activity in HEK293T Cells

Next, we asked how the point mutations of the AMPylation sites influence the PLD3 5′-3′ exonuclease activity. A previously described PLD3 activity assay was utilized, which relies on the cleavage of a short ssDNA substrate releasing fluorescent signal upon scission (36). The activity assay was conducted in cell lysates overexpressing either WT PLD3 or its point mutants. The catalytic activity of PLD3 Y323A and triple mutant was completely inhibited, whereas the other PLD3 variants, including S365A, S380A, and Y323F, displayed significantly decreased activities compared with the WT enzyme (Fig. 2G and Table S9). Moreover, the observed decrease in the catalytic activity of PLD3 point mutants could not be only explained by lower amount of the soluble form itself because the drop in PLD3 soluble form was not proportional to change in its catalytic activity (Figs. 2G and S5). For example, while there was only 32% less soluble PLD3 S365A in the cell lysate in comparison to PLD3 WT, the exonuclease catalytic activity dropped by 92%. That further corroborates an AMPylation-dependent process leading to correct PLD3 activation. We speculate that the PLD3 Y323A mutation may cause improper folding and proteolytic processing of the full-length protein, resulting in a lack of the active soluble form. Because of S380 localization in the dimerization interface, the loss of the activity might have resulted from dimerization failure or folding changes. Therefore, PLD3 (WT, S365A, and S380A) were purified from RCH-ACV cells and analyzed for their ability to form stable dimers by mass photometry (Fig. S8). We observed that both PLD3 point mutants failed to form stable dimer, which is in line with their reduced catalytic activity (Fig. S9).

Next, we asked whether overexpression of PLD3 WT and point mutants may lead to subcellular mislocalization in HEK293T cells. However, cell imaging of endogenous PLD3 and overexpressed PLD3-FLAG variants did not show distinguishable changes in subcellular distribution as determined by colocalization with ER, endosomal, and lysosomal markers (Fig. S10). Both endogenous PLD3 and overexpressed PLD3 variants localized mainly to endosomes and ER in HEK293T cells.

In human cancer cell lines, immune Cal-1 cells, and iPSCs, the major PLD3 form is the full-length protein, whereas during neuronal differentiation and in differentiated neurons, the soluble form becomes more abundant. Finally, in mature neurons, the main form is AMPylated PLD3, leading to inhibition of its catalytic activity (Figs. 1F, S11, B and D) (31). This suggested a crucial role for PLD3 AMPylation and processing during neuronal development. Comparing PLD3 exonuclease activity across different cell lines indicated that the amount of soluble non-AMPylated PLD3 is important for its catalytic activity (Fig. S11A). Furthermore, based on the activity data of PLD3 Y323A, Y323F, S365A, and S380A mutants, we propose that PLD3 AMPylation might be a necessary transient modification leading to release of the soluble active PLD3, which in neurons might be halted at the stage of AMPylated soluble form to block its exonuclease activity (Fig. S11, B–D). The soluble active PLD3 might be then swiftly released by deAMPylation from pool of AMPylated soluble form. Together, these data showed that AMPylation was necessary to establish the catalytic activity of PLD3.

The AMP-Transferase FICD Directly Interacts with PLD3

To elucidate whether the ER-resident AMP-transferase FICD interacts with PLD3, we utilized a PLD3-FLAG construct to coimmunoprecipitate (co-IP) interacting proteins. To account for unspecific background binding proteins, we compared the co-IP from HEK293T cells overexpressing PLD3 with or without the FLAG tag. Since endogenous FICD levels are low and difficult to detect by both WB and MS-based proteomics, we coexpressed a WT FICD-encoding vector to increase the FICD levels. First, WB with an anti-PLD3 antibody confirmed efficient pull-down of PLD3-FLAG from the lysate (Fig. 3A), whereas no enrichment of PLD3 was observed in the control (protein without the FLAG-tag) (Figs. 3A and S12). Next, WB with an anti-FICD antibody revealed the corresponding FICD band, which was absent in the co-IP from PLD3 without the FLAG tag (Fig. 3B). To further validate the FICD–PLD3 interaction, we now overexpressed WT FICD-FLAG and WT PLD3 (without FLAG tag). As expected, the co-IP of FICD-FLAG showed a clear enrichment of PLD3 (Fig. 3B), thereby confirming the previous result. Interestingly, only an intermediate form of PLD3 was observed, which did not correspond to the ∼72 kDa full-length form or the soluble PLD3 form around 55 kDa. Next, the interaction of FICD and PLD3 was corroborated by MS-based proteomics (Fig. 3F and Table S10) to show also an overlap of 70 proteins between the two IPs including heat-shock proteins involved in UPR HSPA5 and HSPA8 (Fig. 3G). The other commonly shared enriched proteins, LRRC59, RAB18, RABL3, TMED1, and TMED2, suggested potential crosstalk between pathways associated with cell differentiation, intracellular protein transport, and vesicular protein trafficking. Together, these data point toward direct interaction between FICD, PLD3, and HSPA5, suggesting a crosstalk between UPR and TLR9 signaling.

Fig. 3.

FICD interacts with the intermediate form of PLD3 in HEK293T cells.A, Western blot analysis before and after immunoprecipitation (IP) of PLD3-FLAG using anti-FLAG beads demonstrates co-IP of FICD. One representative blot of two independent experiments is shown. B, Western blot analysis before and after IP of FICD-FLAG using anti-FLAG beads for crossvalidation confirms PLD3–FICD interaction and further elucidates an intermediate form of PLD3 interacting with FICD. C, truncated luminal PLD3^-AA61 equipped with N-terminal signal peptide. D, Western blot analysis of PLD3-FLAG before and after incubation with PNGase F glycosidase specifically cleaving N-linked glycoproteins showing almost quantitative N-glycosylation of full-length PLD3. E, Western blot analysis of PLD3^-AA61 before and after incubation with Endo H glycosidase showing almost quantitative N-glycosylation of full-length PLD3. F, volcano plots visualizing the results of co-IP MS-based proteomics of PLD3-FLAG and FICD-FLAG constructs. Red circles—significantly enriched proteins, blue dots—proteins found significantly enriched in both PLD3 and FICD co-IPs, and gray circles—not significantly enriched proteins. Co-IPs were prepared in triplicates (n = 3) from one protein lysate. G, Venn diagram showing the overlap between significantly enriched proteins in PLD3 and FICD co-IPs. FICD, protein adenylyltransferase; HEK293T, human embryonic kidney 293T cell line; PLD3, phospholipase D3.

To better characterize PLD3 form that was found co-IPed with FICD (Fig. 3B), we next examined the glycosylation status of overexpressed PLD3. Of note, the soluble PLD3 is known to be stabilized by N-glycosylation; in the mouse protein on residues N97, N132, N234, N282, and N385 (35, 37). To confirm that the N-glycosylation takes place in our expression system, the MS data from PLD3 IP were searched by MSFragger for N-glycosylation sites to corroborate previously identified N236 and N387 glycosylation sites on human PLD3 (54, 55). Next, the lysates containing overexpressed full-length PLD3 or truncated soluble PLD3 (Fig. 3C) were incubated with glycosidases PNGase F and Endo H (Figs. 3, D, E and S13). The comparison of the bands observed before and after the PNGase F and Endo H incubation showed almost quantitative N-glycosylation of both full-length and soluble PLD3.

FICD AMPylation Activity Decreases PLD3 Levels in HEK293T Cells

FICD catalyzes both AMPylation and deAMPylation. Switching between the two functionalities is mediated by an autoinhibitory α-helix, which is contingent on the critical glutamate residue E234 (4). To investigate whether the processing of PLD3 into different forms is influenced by FICD activity, we coexpressed WT PLD3-FLAG and three variants of FICD—WT FICD, the highly active FICD E234G, which lacks deAMPylation activity and the catalytically inactive mutant FICD H363A (1). First, the LC–MS/MS-based whole proteome analysis revealed that FICD E234G leads to significantly decreased levels of PLD3 compared with WT FICD and inactive FICD H363A (Fig. 4A and Table S11). Complementary WB analysis confirmed the overall lower protein levels of PLD3 because of FICD E234G expression, which rendered the soluble PLD3 levels difficult to detect (Fig. 4B). This suggested that FICD might regulate PLD3 levels directly through AMPylation and indirectly by switching a yet unknown regulatory loop. To account for effects caused by coexpression that might not be excluded by catalytically inactive FICD H363A control, coexpression of PLD3 and mScarlet fluorescent protein with ER targeting N-terminal KDEL-sequence was performed (Fig. S14). Similar to all FICD variants and PLD3 co-expression, mScarlet-KDEL lead to lower efficiency of PLD3 expression rendering the catalytically inactive FICD H363A mutant as the suitable control for the co-expression experiments. Furthermore, expression of FICD E234G led to significant changes on whole proteome level (Fig. 4C and Table S11). Interestingly, FICD E234G–induced decrease in PLD3 levels was countered by an increase of HSPA5 and other UPR-associated proteins, including HSPA6, HSB1, DNAJB1, and DNAJB4, whereas no change was observed for the expression of housekeeping genes such as H3-2 (Fig. 4D and Table S11). There is an overlap of five proteins, which are upregulated when FICD E234G is overexpressed in comparison to all other conditions. This set of upregulated proteins include BAG3, HSPB1, HSPA6, DNAJB1, and DNAJB4, suggesting increasing cellular stress and induction of heat-shock proteins. Only coagulation factor V was significantly downregulated in FICD E234G cells comparing with other FICD mutants. The most dysregulated proteins were observed when comparing only AMPylating FICD E234G with catalytically inactive FICD H363A variant. From a total of 181 dysregulated proteins, 152 proteins are upregulated, including TPM3, CHP1, GLA, COPRS, NRAS, HRAS, B4GALT7, and COPRS (Fig. 4C). The Gene Ontology term cellular component analysis of these significantly upregulated proteins suggests that AMPylation is involved in protein secretion (Fig. S16).

Next, the analysis of PLD3 AMPylation status using the pro-N6pA probe upon coexpressions of FICD variants and subsequent WB with anti-PLD3 antibody showed a strong overall decrease in the amount of AMPylated soluble PLD3 (Fig. 4E). Surprisingly, in the case of WT PLD3 together with FICD E234G coexpression, we observed strictly the intermediate form of PLD3 pointing toward malfunctioning downstream processing of PLD3, perhaps caused by missing deAMPylation activity (Fig. 4E). Furthermore, the MS-based analysis of enriched AMPylated proteins showed significant FICD self-AMPylation confirming the previously described in vitro observations (Figs. 4F, S15 and Table S12) (56). Analysis of PLD3 AMPylation showed a significant increase in fold enrichment between FICD WT and FICD E234G (Fig. 4F). The normalization to the total and soluble form of PLD3 suggested that FICD might indeed catalyze AMP transfer to PLD3 (Fig. 4, B and F). Of note, the MS-based analysis determines only the relative AMPylation ratio as it compares control to pro-N6Pa-enriched samples. Third, the PLD3 exonuclease activity has dropped accordingly correlating with the decreased levels of soluble PLD3 (Fig. 4G). The negligible difference between FICD WT and the inactive E234G mutant may be attributed to the complex activity of FICD and the presence of endogenous FICD. The activity drops to similar levels as for PLD3 point mutants lacking the AMPylation sites (Fig. 2G). Finally, the in-gel analysis of pro-N6pA labeling confirmed the low amounts of soluble PLD3 when coexpressed with FICD variants and corresponding disappearance of fluorescent signal of soluble PLD3, which was visible only for WT PLD3 (Fig. 4H). Together, FICD AMPylation activity leads to decrease of both full-length and soluble PLD3 forms. The FICD E234G lacking deAMPylation activity obstructs the formation of soluble PLD3 and yielded only the intermediate PLD3 form.

Full-Length PLD3 is the Major Constituent of Lysosomes

FICD is localized to the ER, whereas the PLD3-soluble catalytically active domain was previously described to be present in lysosomes, where it is transported through the endosomal pathway (37). However, the aforedescribed interaction between PLD3 and FICD suggests that PLD3 mainly interacts with proteins localized to the ER and Golgi (Fig. 3F). This raised the question, in which subcellular compartment AMPylation and deAMPylation are taking place and what the major form of PLD3 in lysosomes is. To this end, we used LysoIP with the transient overexpression of lysosomal TRPML1 Ca²⁺ channel bearing a triple human influenza HA tag, which is exposed to cytosol (Fig. 5A) (57). The efficient enrichment of the lysosomal fraction was determined by immunoblotting using LAMP1 and by LC–MS/MS proteomics to determine additional lysosomal markers, LAMP2 and CTSD, both confirming the successful isolation of lysosomal fraction (Figs. 5, B, C, S17 and Table S13). Surprisingly, we observed solely the full-length PLD3 form present in the enriched lysosomal fraction, whereas both the soluble and full-length PLD3 forms were clearly found in original lysates before the LysoIP (Fig. 5C). Together, LysoIP suggests that PLD3 in lysosomes is present in its full-length and not its soluble form as thought previously in HEK293T cells.

Fig. 5.

LysoIP reveals full-length PLD3 as the major PLD3 form in lysosomes.A, scheme of TRPML1-3xHA construct used for LysoIP, an enrichment experiment comparing protein background binding from untreated (UT) cells with lysates from TRPML1-3xHA–containing cells. TRPML1 consists of six transmembrane helices and an extracytosolic/luminal domain, the hallmark of the TRPML family (77). The 3xHA tag is fused to the C terminus. B, profile plots of PLD3 and lysosomal markers obtained from LC–MS/MS analysis of lysate and after lysosomal IP. UT control or TRPML1-3xHA tag containing HEK293T cells. Of note, the protein of interest, such as PLD3, was found and quantified in all replicates from TRPML1-HA–containing cells, whereas only in two or less replicates from UT cells. One replicate represents an individual LysoIP sample. C, Western blot with anti-PLD3 of whole lysate (L), flow through (FT), and lysosomal IP fraction for TRPML1-overexpressing or UT control HEK293T cells. One representative blot from two independent LysoIP experiments is shown. HEK293T, human embryonic kidney 293T cell line; IP, immunoprecipitation; LysoIP, lysosomal immunoprecipitation; PLD3, phospholipase D3.

Hap1 FICD-Depleted Cells Corroborate Cell Type Specificity of PLD3 AMPylation

First, we analyzed the overall AMPylation profile in Hap1 utilizing the pro-N6pA probe for enrichment followed by LC–MS/MS analysis (Fig. S18 and Table S14). However, there was no enrichment observed for neither HSPA5 nor PLD3, whereas other previously identified AMPylated proteins, such as PPME1 and Cathepsin B, were found significantly enriched (Fig. S18). Because of the bidirectional FICD catalytic activity, we further investigated whether depletion of FICD would lead to PLD3 AMPylation, which may suggested the presence of a yet unknown AMP-transferase. Therefore, we targeted FICD with sgRNAs in near haploid Hap1 cells to create a FICD knockout cell line. The sgFICD line was complemented by an sgSELENOO line to analyze both thus far two known AMP-transferases. We determined the changes on whole proteome level, which were induced by depletion of FICD or SELENOO (Fig. S18 and Table S15). PCA revealed distinct and significant changes in protein expression (Fig. S18). More than 300 proteins were downregulated in both cell lines. The most downregulated proteins, including SKP2 and FBXL6, are involved in G2/M transition of mitotic cell cycle and regulating protein localization to the membrane, such as TCAF1, GPC3, and GPC4. Interestingly, heat-shock protein HSPB1 was one of the most downregulated proteins in the sgFICD cells and also significantly downregulated in sgSELENOO cells (Fig. S18). In contrast, HSPB1 was found upregulated in cells overexpressing FICD E234G; see later. Furthermore, the 85 proteins upregulated in both sgFICD and sgSELENOO cells include ALDOC, PCK2, PNPLA2, SLC22A18, KCTD12, and PARK7 pointing toward the role in carbohydrate metabolism and oxidative stress regulation (Fig. S18).

Surprisingly, we observed decrease of PLD3 exonuclease activity in both sgSELENOO and sgFICD cells, when compared with parent Hap-1 cells, which suggests dysregulation of cellular response to CpG^O-DNA stimulation (Fig. S11). Finally, inhibition of the lysosomal activity by monensin in both WT and knockout cells resulted in further depletion of AMPylation on cysteine cathepsins (Fig. S18). Of note, monensin was previously used to increase AMPylation on PLD3 and other proteins (32). Taken together, modulation of AMP-transferase activity in Hap-1 cells corroborates cell type specificity of protein AMPylation.

Depletion of PLD3 AMPylation Leads to TLR9 Activation in Cal-1 Cells

To assess whether PLD3 AMPylation mutants lack the ability to process single-stranded DNA in immune cells, we made use of the pDC-like immortalized cell line, Cal-1. It was previously described that PLD3 degrades DNase 2-generated ligands for TLR9 in mice (42). Consequently, it was shown that CpG^O-DNA did not trigger a TLR9 response in WT Cal-1 cells but in the absence of PLD3, indicating that PLD3 also negatively regulates the TLR9 response in Cal-1 cells (38). We therefore stably expressed WT PLD3 and PLD3 AMPylation mutants in PLD3-deficient Cal-1 cells (Fig. 6A) and stimulated those cells with CpG^O-DNA as well as TLR agonist R848 as a control (Fig. 6, B and C). As expected, WT Cal-1 cells did not respond to CpG^O-DNA stimulation, whereas CpG^O stimulation in PLD3-deficient cells induced robust IFN-β release (Fig. 6B). Overexpression of the WT PLD3 protein in PLD3-deficient cells reversed this effect. PLD3 S365A and PLD3 S380A overexpression also suppressed the CpG^O response, similar to the WT PLD3 protein, indicating that PLD3 S365A and PLD3 S380A are still functional in degrading DNA in Cal-1 cells. Interestingly, overexpression of PLD3 Y232A and the triple mutant PLD3 (S365A, S380A, and Y323A) and subsequent stimulation with CpG^O resulted in partial IFN-β release for PLD3 Y232A and robust IFN release for PLD3 (S365A, S380A, and Y323A), mimicking the PLD3-deficient cell line. TLR agonist R848 control stimulation resulted in similar IFN-β release in all conditions tested (Fig. 6C). Stable expression of mCherry in PLD3-deficient Cal-1 cells was further used as a control and showed similar behavior to the PLD3-deficient cell line. The nonhydrolyzable phosphorothioate-modified CpG oligonucleotides were used as a control to ensure the cleavage selectivity (Fig. 6D). Furthermore, PLD3 S365A, S380A, and Y323A mutant resulted in somewhat smaller changes on whole proteome compared with overexpression in HEK293T cells (Figs. 6E, S19, S20, and Table S16). Together, in Cal-1 pDC-like immune cells, dysregulation of PLD3 AMPylation leads to aberrant TLR9 responses in PLD3 Y323A and triple mutant–containing cells.

Fig. 6.

PLD3-dependent TLR9 activity in Cal-1 cells and PLD3 processing in PD patient–derived neurons.A, Western blot of WT and PLD3^−/− cells reconstituted with PLD3-FLAG, indicated PLD3-FLAG mutants, or mCherry (FLAG). One representative blot of three independent experiments is shown. B, WT Cal-1, PLD3^−/−, or PLD3^−/− cells reconstituted with WT PLD3, indicated PLD3 mutants, or mCherry were unstimulated or stimulated with CpG^O, and IFN-β release was measured by ELISA. Data are depicted as mean ± SEM of n = 3 independent experiments. C, WT Cal-1, PLD3^−/−, or PLD3^−/− cells reconstituted with WT PLD3, indicated PLD3 mutants, or mCherry were unstimulated or stimulated with R848, and IFN-β release was measured by ELISA. Data are depicted as mean ± SEM of n = 3 independent experiments. D, WT Cal-1, PLD3^−/−, or PLD3^−/− cells reconstituted with WT PLD3, indicated PLD3 mutants, or mCherry were unstimulated or stimulated with CpG^S, and IFN-β release was measured by ELISA. Data are depicted as mean ± SEM of n = 3 independent experiments. E, whole proteome analysis of Cal-1 cells expressing PLD3 point mutants (Fig. S19). F, Western blotting of PLD3 in NPCs from a healthy donor compared with midbrain neurons from a PD patient with SNCA duplication. In PD-derived neurons, higher levels of PLD3 intermediate forms were observed. G, proposed post-translational processing of PLD3 resulting in three different forms. In green rectangle are PLD3 forms regulated by FICD AMPylation and deAMPylation activity, which is governed by its inhibition of α-helix. FICD, protein adenylyltransferase; IFN-β, interferon beta; NPC, neural precursor cell; PD, Parkinson’s disease; PLD3, phospholipase D3; TLR9, Toll-like receptor 9.

PLD3 Forms are Altered in PD Patient–Derived Neurons

The function of PLD3 has been previously linked to neurodegenerative conditions, such as AD (58). We next explored the relevance of PLD3 processing in neurodegenerative conditions. Given the many similarities between different neurodegenerative diseases, we set to examine whether PLD3 processing also plays a role in PD patient–derived neurons. We characterized PLD3 processing profile in hiPSC-derived NPCs and midbrain neurons from a PD patient carrying a heterozygous duplication of the SNCA gene and a healthy donor. The SNCA gene encodes α-synuclein, whose aggregation plays a crucial pathological role in PD pathogenesis. SNCA gene multiplications, including duplication and triplication, have been shown to induce parkinsonism by increasing α-synuclein levels and promoting its aggregation, leading to neuronal loss. Thus, hiPSC-derived neurons from SNCA multiplication patients provide a unique model to investigate α-synuclein-mediated protein dyshomeostasis and other pathways contributing to neurodegeneration (59, 60, 61). Our previous study demonstrated that midbrain neurons with an SNCA duplication exhibited neurite degeneration (46, 61), as well as selective loss of midbrain dopaminergic neurons (62), the key neuronal population lost in PD. To explore PLD3 forms in PD patient–derived neurons, we employed a small molecule–based differentiation protocol to generate midbrain neurons through NPCs (46, 63). This approach allowed us to model both neurodevelopment and disease processes from the same individual at a cellular level.

Confirming our observation in hiPSC and neurons (Fig. 1F), the main forms in proliferating NPCs were the full-length form. The predominant form in differentiated neurons is soluble PLD3 (Fig. 6F). Interestingly, in PD patient–derived midbrain neurons, we observed the presence of intermediate and full-length PLD3, when compared with control neurons, suggesting a delayed or impaired processing of PLD3. These findings provide a new insight into PLD3 processing and suggest dysregulation of PLD3 in PD.

Discussion

Protein PTMs can swiftly regulate protein catalytic activity, protein–protein interactions, and protein subcellular localization. Therefore, PTMs provide the cell with an efficient and fast mechanism to adjust metabolic pathways by reacting to changes in the intracellular and extracellular environments. Indeed, protein PTMs have been shown to play an important role in many biological processes, whereas their dysregulation can be often associated with a disease state (64, 65). The number and stoichiometry of many known protein PTMs differ dramatically in cell types, leading to the concept of proteoforms (65, 66). A proteoform is understood as an individual protein isoform and splice variant with defined PTM status. The different proteoforms resulting from expression of a gene may have different subcellular localization, catalytic activity, and interactions with other proteins. Thus far, multiple protein PTMs have been closely associated with neurodegenerative diseases for example by influencing aggregation properties of Tau and α-synuclein (67, 68). The first step to deconvolute such complex networks and dynamic changes of proteoforms resulting in the onset of neurodegeneration or reversing the metabolism toward the homeostasis is understanding properties of the individual protein “players” or proteoforms, including plethora of ever emerging protein PTMs. Protein AMPylation is a reversible modification on the side-chain hydroxyl group of Tyr, Ser, or Thr residues (69). In human cells, the most prominent AMPylated protein is HSPA5, the main activator of UPR. HSPA5 is AMPylated by FICD to inhibit its chaperone activity once the ER stress is released and elevated HSPA5 expression is decreasing to its basal level (13, 20, 21). In addition, the pool of AMPylated HSPA5 in the cells can provide quick access to active HSPA5 by FICD-catalyzed deAMPylation (22). The obstacle to address functionally FICD-related AMPylation/deAMPylation is its low abundancy in most cell lines, aforementioned dual catalytic activity, and yet unclear mechanism(s) allowing FICD to switch between AMPylation and deAMPylation. Interestingly, a Drosophila FICD mutant lacking the deAMPylation activity was not viable, and induction of excessive AMPylation by the hyperactive FICD mutant led to scrambled eye development (10). Moreover, overexpression of FICD in developing cerebral cortex organoids revealed accelerated neuronal differentiation from NPCs, which resulted in the partial failure of newly produced neurons to migrate into the cortical layer (23). Another intriguing question concerns FICD cosubstrate selectivity, as it seems highly likely that different nucleotide analogs serve as a substrate in living cells (30, 70). Furthermore, several dozen additional proteins were identified to be AMPylated, many of them lysosomal proteins known to be associated with neurodegeneration (18, 71).

The function of PLD3 has been extensively studied in immune cells where it regulates the activation of TLR7 and TLR9 by cleaving ssRNA and ssDNA substrates (35, 38, 42, 72). Based on a Genome-Wide Association Study, PLD3 was linked to AD, but the putative mechanism by which PLD3 contributes to the pathology remains controversial. Studies with neuroblastoma cells identified mtDNA as the major physiological source of PLD3 DNA substrate (33, 43, 44). Recently, an AD mouse model showed that lack of PLD3 activity improves axonal fitness and hence neuronal networking (39). While neurons are not immunologically active, the signaling pathways may be employed for other function as suggested for excitatory CA1 neurons to employ TLR9 inflammatory signaling to establish features characteristic for memory assemblies (73). As shown by us here and previously, there is a striking difference of PLD3 levels and proteoforms between immune and neuronal cells (31). The major PLD3 form in neurons is the proteolytically cleaved soluble PLD3 with highest AMPylation stoichiometry in mature neurons. Pull-down of pro-N6pA-modified PLD3 shows that only the proteolytically processed soluble PLD3 form is AMPylated. There are several cognate questions remaining to answer about PLD3 function and processing in general and specifically in neurons, which might be important for understanding the mechanism relevant for the role in neurodegeneration.

In this work, we focused on the characterization of PLD3 AMPylation and its functional implications. We suggest that AMPylation is necessary for PLD3 proteolytic processing and 5′-3′ exonuclease activity. Our study identifies three AMPylation sites (Y323, S365, and S380) in PLD3 and establishes AMPylation as a critical processing step required for production of catalytically active soluble PLD3 (Fig. 6G). The dimerization study showed that mutation of the AMPylation sites lead equally to dimerization failure and reduction of catalytic activity. Although the PLD3 S365A and S380A mutants exhibited reduced catalytic activity in vitro compared with WT PLD3, the residual activity was still sufficient to suppress TLR9 responses to CpG-DNA in cells. The follow-up studies in engineered cell lines need to be carried out to fully characterize the impact of each AMPylation site.

The direct interaction between PLD3 and FICD, together with coexpression experiments, suggests that FICD regulates AMPylation of PLD3 in living cells. In combination, accumulation of soluble PLD3 in neurons and the observed functional role in neurodegenerative diseases renders AMPylation a plausible mechanism, which tightly regulates downstream TLR9 signaling. The identification of whether this regulatory cascade may offer desired pharmacological opportunities to improve neuronal fitness in neurodegeneration remains to be explored.

The critical aspect of PLD3 AMPylation–deAMPylation is its subcellular localization. FICD is an ER-localized protein, which we showed by IP, interacts predominantly with the nonglycosylated form of PLD3. Cellular imaging of endogenous PLD3, overexpressed FLAG-PLD3, and PLD3 point mutants suggests mainly ER and endosomal localization in HEK293T cells. Therefore, AMPylation and deAMPylation events might take place either directly in the ER or at ER–endosome junctions. We hypothesize that the non-AMPylated PLD3 may stay in endosomes or be further transported into lysosomes where it is proteolytically cleaved into catalytically active soluble form. The processes leading to accelerated degradation of PLD3 upon FICD overexpression need to be yet investigated.

An interesting finding of this study is the significant increase in soluble PLD3 in hiPSC-derived neurons during differentiation or in differentiated states, compared with proliferating cell lines (Fig. 1F) or autologous proliferating neural precursor cells (Fig. 6F). This distinct PLD3 processing pattern suggests a potential role in neuronal development. Notably, our neurons were derived from different individuals and generated using distinct differentiation protocols for cortical (Fig. 1F) (74) and midbrain (Fig. 6F) lineages (63), highlighting the robustness of this observation, independent of differentiation method, hiPSC donor, or neuronal lineage. Importantly, our results indicate impaired PLD3 processing in PD patient–derived neurons, characterized by a more abundant mixture of full-length and intermediate PLD3 forms compared with healthy control neurons (Fig. 1F). This links for the first time dysregulated PLD3 processing to PD in patient-derived neurons. Neurodegenerative diseases share common pathological features, including protein aggregation, despite disease-specific protein deposits. Disruptions in protein homeostasis, resulting from factors, such as ER stress and lysosomal dysfunction, are widely implicated in neurodegeneration and are also observed in hiPSC-derived neurons from patients carrying SNCA gene multiplication dysfunction (59, 60). Thus, impaired PLD3 processing may result from disruptions in protein quality control systems triggered by an increased SNCA gene dosage. Moreover, a detrimental feedback loop between PLD3 and protein quality control is also possible. Since PLD3 functions as a lysosomal exonuclease, our findings support the notion that proper PLD3 processing is essential for its enzymatic function and neuronal homeostasis. Delayed or aberrant PLD3 processing could exacerbate lysosomal disturbances, potentially affecting the clearance of aggregated proteins associated with neurodegenerative diseases. Further studies are needed to elucidate the mechanistic links between PLD3, protein quality control systems, and neurodegeneration. It also remains to be determined whether PLD3 dysregulation is specific to PD or represents a common mechanism across neurodegenerative disorders. Understanding these pathways will be crucial in assessing the potential of PLD3 as a therapeutic target for neuroprotection and regeneration.

Data Availability

All MS data are available via ProteomeXchange with identifier PXD057707.

Supplemental data

This article contains supplemental data (78, 79, 80, 81, 82, 83, 84).

Conflict of interest

The authors declare no competing interests.