Massive Phosphoproteomics Sweep Maps the Hidden Regulatory Switches on a Key Metabolic Enzyme

A large-scale computational survey of thousands of published human phosphoproteomics experiments has identified three phosphorylation sites on pantothenate kinase 2, the mitochondrial enzyme whose failure drives a rare childhood neurodegenerative disease, as the most consistently observed regulatory marks on the protein. The analysis, which assembled 2,359 mass spectrometry datasets from human cell lines, found that the serine residues at positions 168, 169 and 189 of PANK2 are detected far more frequently than any other phosphorylation site on the enzyme, and it linked each of these sites to extensive networks of proteins whose phosphorylation rises or falls in step with them.

PANK2 has long occupied a special place in metabolism and neurology. Identified in 2001 as a causative gene for neurodegeneration with brain iron accumulation, the gene encodes a 570-amino-acid mitochondrial enzyme that catalyzes the rate-limiting phosphorylation of pantothenic acid, better known as vitamin B5, to 4′-phosphopantothenate in the coenzyme A biosynthetic pathway. Coenzyme A is an essential cofactor for more than 100 metabolic reactions, supporting fatty acid synthesis, the citric acid cycle and energy production. PANK2 activity is itself tightly controlled, subject to feedback inhibition by coenzyme A derivatives such as acetyl-CoA and modulated by metabolic cues including insulin signaling.

When PANK2 fails, the consequences are severe. More than 100 pathogenic variants of the gene have been reported, including the common missense mutations G411R and T418M, which typically abolish enzymatic activity. The result is pantothenate kinase-associated neurodegeneration, or PKAN, an autosomal recessive subtype of neurodegeneration with brain iron accumulation characterized by dystonia, parkinsonism, cognitive decline and the pathognomonic eye-of-the-tiger sign on MRI, produced by iron accumulation in the globus pallidus. Loss of PANK2 activity leads to coenzyme A deficiency, mitochondrial dysfunction, oxidative stress, impaired autophagy and ferroptosis, with elevated citrate levels and lipid peroxidation further exacerbating neuronal damage. No targeted therapies for PKAN currently exist; available treatments provide only symptomatic relief.

Against that backdrop, the new study took a systematic, data-mining approach to characterize the global phosphoproteome of PANK2. The researchers screened approximately 3,825 published human phosphoproteomics studies from the literature, retaining only those that reported at least one Class-1 PANK2 phosphosite, defined as a site with a localization probability of at least 75 percent and an A-score above 13, generated through digestion with the proteases LysC or trypsin. That filtering yielded 2,061 profile datasets, which catalog all detected phosphopeptides, and 298 differential datasets, which quantitatively compare phosphosite abundance between test and control conditions. Datasets produced with other proteases, such as chymotrypsin and Glu-C, were excluded to maintain consistency in peptide mapping.

Ranking every Class-1 PANK2 phosphosite by how often it appeared across these datasets, the team found that S168, S169 and S189 stood out as predominant sites, detected in 91, 95 and 75 datasets respectively. Notably, all three residues sit in the intermediate/regulatory domain of the protein, which spans roughly amino acids 47 to 243, outside the C-terminal catalytic core where most pathogenic PKAN missense mutations cluster. Sequence analysis with a deep-learning tool that predicts ordered and disordered regions showed that these sites map largely to intrinsically disordered segments of the protein, particularly around residues 50 to 100 and 200 to 250. The authors note that this pattern is consistent with PANK2 being classified as a kinase whose predominant phosphorylated sites lie outside its kinase domain, and that their high detection frequency across diverse experimental conditions suggests the sites are constitutively accessible and dynamically regulated under a wide range of cellular states.

To explore what these phosphorylation events might be connected to, the researchers searched for phosphosites in other proteins that were co-regulated with each predominant PANK2 site across the differential datasets. Using a one-sided Fisher’s exact test on contingency tables built for every candidate pair, followed by a series of bias-correction filters requiring support from at least three distinct published articles and three distinct experimental conditions, they assembled a high-confidence set of co-phosphoregulated proteins. The scale was striking: PANK2 S168 was positively co-phosphoregulated with 504 phosphosites and negatively with 12; S169 showed positive and negative co-phosphoregulation with 667 and 55 phosphosites respectively; and S189 displayed the most extensive network, with 806 positive and 14 negative associations.

Within these networks, the team identified 35 phosphosites in kinases predicted to act as upstream regulators of PANK2, most derived from a kinase-substrate inference framework based on positional amino acid preferences of the human serine/threonine kinome. The candidates span a remarkable functional breadth: ribosomal protein S6 kinases RPS6KB1 and RPS6KA1, the autophagy initiator ULK1, multiple MAPK-pathway components including MAP3K3, RAF1 and MAP2K2, the energy-sensing AMPK catalytic subunit PRKAA1, cytoskeletal regulators ROCK2 and CDC42BPA, and the calcium-sensing kinase CAMKK2, among others. Only one of these assignments, AKT1 at serine 126, is backed by direct biochemical validation in the literature. That single validated link is nonetheless significant, because prior work has shown that AKT phosphorylates PANK2 at Ser169 and Ser189 within conserved RXRXXS/T motifs, and that PI3K signaling promotes de novo coenzyme A synthesis from vitamin B5. The same study found that AKT phosphorylation of the related protein PANK4 relieves its inhibitory 4′-phosphopantetheine phosphatase activity, suggesting coordinated control of both the forward and reverse arms of the coenzyme A pathway.

The co-phosphoregulated kinases fell into recognizable functional groups. MAP kinase pathway signaling formed the largest category with 14 proteins, spanning multiple tiers of the cascade from MAP3K-family kinases through RAF-family members to downstream effectors. Atypical kinases and other enzymes formed the second-largest group with 13 proteins, followed by cell cycle, mitosis and centrosome regulation with 11, and PI3K/AKT/mTOR signaling and autophagy with 8, including AKT1, AKT2, ULK1 and PRKAA1. Cytoskeletal dynamics, cell polarity and motility accounted for another 8 proteins. Several of these kinases carry phosphosites with documented functional consequences in the literature, such as MAP2K4_S80, which inhibits apoptosis, and RAF1_S259, which is linked to inhibited cell differentiation and cell cycle regulation, raising the possibility of reciprocal or feed-forward regulation with PANK2.

The analysis also surfaced phosphatases and binary protein interactors within the networks. PANK2 S189 showed the most extensive phosphatase connections, co-phosphoregulating with LPIN1, INPP1, DUSP16, CDC25B and PTPN1, while S169 connected with DUSP3, PTPRA, CDC25B and MTMR4, and S168 showed fewer links, primarily with SSH3 and PPIP5K1. The overlap of CDC25B and PTPN1 across multiple sites hints at potential crosstalk between dephosphorylation events. Five binary interactors emerged from computational predictions: KSR1, NFKB1, RNF34, DCAF15 and ATL1, with RNF34 and ATL1 each co-phosphoregulating with more than one PANK2 site. When the positively co-phosphoregulated protein set was mapped onto the KEGG PI3K-AKT signaling pathway, 26 proteins were annotated pathway members, spanning receptor inputs, core components such as AKT, mTORC2 and Cdc37, energy-stress sensors including AMPK and TSC2, and downstream branches governing protein synthesis, glycogen metabolism, cell-cycle progression and survival. The authors are careful to stress that this mapping is descriptive rather than a statistical enrichment result, since the size and pervasiveness of the PI3K-AKT pathway means some overlap with any phosphoprotein list is expected.

The study’s authors are equally candid about its limitations. The integrative analysis of nearly 2,200 datasets introduces heterogeneity in cell types, stimuli and platforms that statistical filtering cannot fully eliminate without reprocessing raw data, and the inferred kinase-substrate relationships draw from databases that partially overlap with the input data, creating potential circularity. Most importantly, co-phosphoregulation reflects statistical association rather than direct causality; kinases such as AKT1, AKT2 and ULK1 are implicated as upstream regulators based on sequence motif matching and co-occurrence, and direct biochemical validation through in vitro kinase assays or site-directed phosphosite mutagenesis will be required to confirm which relationships reflect genuine PANK2 regulation. The framework is explicitly hypothesis-generating, prioritizing candidates for future experimental work in disease-relevant systems such as PANK2-knockout neurons or iPSC-derived PKAN lines. Within those boundaries, the findings position PANK2 within a broader phosphoregulatory network linking growth-factor and energy-sensing signals to coenzyme A biosynthesis, and suggest that PANK2 phosphorylation may participate in cellular processes beyond metabolism, including proliferation, stress signaling and intracellular trafficking, a possibility particularly relevant to PKAN, where neuronal vulnerability may reflect not only impaired coenzyme A production but also defective integration of metabolic and signaling pathways.

Subject of Research: Phosphoproteomic mapping of PANK2 phosphorylation sites and co-phosphoregulatory signaling networks

Article Title: Global phosphoproteomics uncovers PANK2 phosphorylation sites and co-phosphoregulation networks

Article References: Rajeev, A. C., Mahin, A., Gopalakrishnan, A. P., Shivamurthy, P. B., & Raju, R. (2026). Global phosphoproteomics uncovers PANK2 phosphorylation sites and co-phosphoregulation networks. Results in Chemistry, 31, Article 104004. https://doi.org/10.1016/j.rechem.2026.104004

Image Credits: AI Generated

DOI: 10.1016/j.rechem.2026.104004

Keywords: PANK2, PKAN, phosphoproteomics, coenzyme A, pantothenate kinase, PI3K-AKT signaling, phosphorylation, neurodegeneration with brain iron accumulation, AKT, co-phosphoregulation, mitochondria, vitamin B5