AlphaFold2 and STRING-Based Mapping of REJ-Mediated Interactions in PC1 Uncovers Intracellular Roles in ADPKD

Autosomal dominant polycystic kidney disease (ADPKD) is a prevalent genetic disorder characterised by the progressive formation of renal cysts, ultimately leading to kidney failure. Mutations in the PKD1 gene encoding polycystin-1 (PC1) are the primary cause of ADPKD, affecting millions globally [1]. The protein products of these genes—polycystin-1, polycystin-2, and fibrocystin—are localised to the primary cilium or involved in trafficking proteins to it, underscoring their importance in ciliary signalling and development [2].

While PC1 has been predominantly studied in renal function, emerging evidence suggests that its role extends beyond kidney, impacting other organs such as the liver, pancreas, and nervous system [3]. The extracellular region of PC1 contains several domains involved in cell-cell and cell-matrix interactions, critical for signalling and structural integrity. Among these, the receptor of egg jelly (REJ) domain stands out due to its putative role in intracellular signaling. Mutations within this domain have been linked to defects in protein processing and trafficking, yet its structural and functional attributes remain poorly defined. Thus, elucidating the REJ domain’s role provides a crucial step in bridging PC1’s diverse organ-specific functions with ADPKD pathology [4,5].

Despite its developmental conservation and role in GPS cleavage, the intracellular relevance of the REJ domain remains under debate. The REJ domain is involved in cleavage at the G-protein-coupled proteolytic site (GPS), a prerequisite for the functional maturation of PC1 [6,7,8]. However, little is known about the intracellular binding proteins of the REJ domain, despite its biological significance.

To address this gap, we mapped the intracellular interactome of the REJ domain using pull-down assays coupled with mass spectrometry, identifying four key binding partners: YWHAZ, DNAH11, PPIA, and PCYOX1. These novel interactions broaden the functional landscape of PC1, implicating the REJ domain in diverse cellular processes including signal transduction, intracellular trafficking, oxidative stress regulation, and protein homeostasis.

YWHAZ (14-3-3 ζ/δ) is a phospho-serine/threonine-binding protein that forms a signalling scaffold regulating various cellular pathways, including apoptosis, motility, and cytoskeletal dynamics. These proteins are found throughout the cell and influence their targets' localization, stability, and molecular interactions [9]. Complementing this role in cellular regulation, DNAH11 (Dynein Heavy Chain 11) acts as a motor protein in the outer dynein arm of motile cilia, where it is essential for generating ciliary motion and plays a crucial role in mucociliary clearance within the respiratory tract [10,11]. Enhancing this network of cellular maintenance, PPIA (Peptidyl-Prolyl Cis-Trans Isomerase A), also known as cyclophilin A, assists in protein folding and the stress response. Beyond its isomerase function, it actively participates in intracellular signaling, inflammation, and apoptosis [12,13], and is notably expressed in vascular smooth muscle cells, where it contributes to reactive oxygen species (ROS)-mediated inflammatory responses [14]. Expanding this functional diversity, PCYOX1 (Prenylcysteine Oxidase 1) is a flavin-dependent enzyme that cleaves prenyl-L-cysteines to produce cysteine and reactive aldehydes. Structural data highlight hydrophobic tunnels that facilitate isoprenoid binding and membrane association [15], while functional studies have linked PCYOX1 to the regulation of platelet aggregation, thrombosis, and lipid peroxidation [16].

These interactions suggest that the REJ domain of PC1 may have extensive roles beyond kidney-related pathways. DNAH11's involvement supports a role in ciliary function, while PPIA and PCYOX1 indicate protein homeostasis and oxidative stress regulation. These observations reinforce the hypothesis that the REJ domain contributes to a broader intracellular signalling network relevant to systemic disease.

This study investigates novel intracellular mechanisms of the REJ domain by combining experimental proteomics with advanced structural bioinformatics. Modern bioinformatics tools are critical for elucidating protein interaction networks and linking them to disease mechanisms. Pathway enrichment analyses reveal the biological significance of protein interactors, while molecular docking predicts structural dynamics and binding affinities [17]. In this study, we employed in silico methods such as the STRING database for protein–protein interaction (PPI) network analysis and biological process enrichment [18,19], as well as AlphaFold2 for 3d structural prediction of the REJ–protein complex [20]. Structural interfaces were visualised and analysed using PyMOL, providing detailed insights into molecular contacts and functional domains [21]. Together, these computational approaches complement our experimental results, offering a more comprehensive understanding of the REJ domain’s cellular roles.

In the context of ADPKD, integrating experimental and computational approaches provides a comprehensive framework for uncovering the intracellular functions of the REJ domain. This combined strategy not only advances understanding of PC1 dysfunction but also highlights potential therapeutic targets to mitigate its downstream effects.

MBP-REJ Pull-Down and Mass Spectrometric Analysis

A recombinant MBP-tagged REJ domain of polycystin-1 (PC1) was generated using the PET21-MBP(TEV)-REJ construct and expressed in E. coli. The fusion protein was purified via amylose affinity chromatography and its purity verified by SDS-PAGE and Coomassie staining. For pull-down assays, HEK293 cell lysates were prepared in RIPA buffer supplemented with protease inhibitors and incubated with immobilized MBP–REJ fusion protein under optimized binding conditions designed to minimize nonspecific interactions. Following multiple high-salt and detergent washes, bound protein complexes were eluted under native conditions. Eluted fractions were first resolved by SDS-PAGE, and discrete protein bands were visualized, excised, and subjected to in-gel trypsin digestion. Peptide fragments were analyzed by MALDI-TOF mass spectrometry, and protein identities were confirmed through database searching using MASCOT and cross-validated against established proteomic datasets. This workflow ensured high specificity in identifying REJ-associated proteins while reducing background contaminants [4,5,6].

Protein Selection and Interaction Analysis

Five proteins of interest—YWHAZ, DNAH11, PPIA, PCYOX1, and polycystin-1 (PKD1)—were selected based on previous pull-down assay results. Each was queried individually using the STRING database (version 12.0) to retrieve its direct protein–protein interactors. The species was set to Homo sapiens, and only high-confidence physical or functional interactions supported by text mining were included.

Network Construction

The combined set of the five initial proteins and their identified interactors was then used as input in STRING to generate a comprehensive PPI network. STRING was set to display only interactions with "text mining" evidence (to focus on literature-curated associations), with a minimum required interaction score of 0.4 (medium confidence to maximize inclusivity). Two distinct networks were generated. The first was the Functional and Physical Association Network, which displays edges representing functional and physical protein associations (default setting). The second was the Physical Complex Network, which displays edges indicating that the connected proteins are part of a direct physical complex (for validation of direct binding partners).

Enrichment Analysis

STRING was employed to conduct enrichment analysis on the resulting network [19,22]. The analysis encompassed the following categories:

• Biological Process from Gene Ontology (GO)
• Molecular Function from Gene Ontology (GO)
• Disease-Gene Associations from the DISEASES database
• Protein Domains and Features from InterPro

Protein Complex Structure Prediction Using AlphaFold2

The three-dimensional structure of a protein complex comprising 14-3-3 zeta (YWHAZ), cyclophilin A (PPIA), the REJ domain of polycystin-1 (PKD1), and prenylcysteine oxidase (PCYOX) was predicted using the AlphaFold2_multimer_v3 model [20,23].Amino acid sequences for YWHAZ (UniProt ID: P63104), PPIA (UniProt ID: P62937, the REJ domain of PKD1 (UniProt ID: P98161, residues 2146-2833, and PCYOX (UniProt ID: Q9UHG3) were retrieved from the UniProt database [24]. The sequences were concatenated into a single FASTA file, with chain boundaries explicitly defined to represent the four distinct subunits of the complex.

AlphaFold2 employs a deep learning architecture that integrates attention-based neural networks with evolutionary information to predict protein structures accurately. For multi-chain complexes, the multimer model captures inter-chain interactions by modelling pairwise residue distances and interface contacts, informed by multiple sequence alignments (MSAS) derived from genetic databases. Structure prediction was performed using the AlphaFold Server [25], a cloud-based platform for running AlphaFold2 models. The server was configured to generate MSAS and predict the complex structure, producing atomic coordinates, per-residue confidence scores (plddt, predicted Local Distance Difference Test), and predicted aligned error (PAE) matrices to assess structural and inter-chain reliability.

Multiple models were generated to account for prediction variability, and the model with the highest average plddt score was selected for analysis. Prediction quality was evaluated by analysing plddt scores (0–100 scale, with >90 indicating high confidence) and PAE matrices to confirm the accuracy of inter-chain interfaces. The stereochemical quality of the predicted structure was assessed using MolProbity [26]. to verify geometric correctness.

Visualisation and Analysis of Inter-Chain Binding Sites

The predicted protein complex structure was visualised using Pymol Version 3.1, Schrödinger, Inc.[21]. Individual chains (YWHAZ, PPIA, REJ domain, and PCYOX) were colored distinctly to highlight the complex’s quaternary architecture. Inter-chain contacts were mapped using Pymol’s find pairs command to identify binding sites between chains. Residue-residue contacts were calculated within a distance cutoff of 4.0 Å between heavy atoms of different chains to define interaction interfaces. Polar interactions, including hydrogen bonds, were identified using the find_hbonds function with default criteria (distance<3.5 Å, angle > 120°). Hydrophobic contacts were inferred for non-polar residues within 4.0 Å proximity.

The identified contacts were tabulated to characterise the binding interfaces between YWHAZ, PPIA, the REJ domain, and PCYOX, focusing on residues contributing to inter-chain stability. Interaction interfaces were visually inspected in Pymol to ensure consistency with PAE matrix predictions and available biochemical data [specify if applicable, e.g., known interaction motifs or literature evidence]. Structural figures of the complex and its binding sites were generated using Pymol for inclusion in the study.

Structure prediction computations were performed via the AlphaFold Server’s cloud-based infrastructure, while visualisation and contact analyses were conducted on a local computer with an Intel Core i7 and 16 GB RAM. The predicted structure and inter-chain contact analysis provided insights into the molecular interactions stabilising the YWHAZ–PPIA–REJ–PCYOX complex.

Interaction of MBP-REJ Fusion Protein Intracellular Proteins

The pull-down assay products were resolved on 15% SDS-PAGE, and protein–protein interactions were subsequently characterized by mass spectrometry [4]. Identified bands were matched to protein entries using the UniProt database, which enabled accurate determination of molecular identities. Four candidate interactors were identified in association with the MBP–REJ fusion protein: YWHAZ, DNAH11, PPIA, and PCYOX1. Among these, three proteins—YWHAZ, PPIA, and PCYOX1—were consistently detected as direct interactors of the REJ domain, highlighting their potential functional relevance to PC1. Table 1 includes the predicted molecular weight, isoelectric point (pI), protein name, and mass for each identified protein. It also shows the potential protein candidates suggested by Uniprot and EXPASY.

Table 1: Proteins Interacting with the MBP-REJ Fusion Protein Identified by MALDI-TOF MS via Pull-Down Assay.

STRING-Based Interaction Network Analysis of REJ Domain-Associated Proteins

The STRING database generated comprehensive protein-protein interaction (PPI) networks based on five initial proteins—YWHAZ, DNAH11, PPIA, PCYOX1, and PKD1—and their identified interactors. Two distinct network visualisations were created using a medium confidence interaction score (0.4) with text mining evidence (Figure 1).

Figure 1: Based on STRING Analysis, The Functional and Physical Association Network of YWHAZ, DNAH11, PPIA, PCYOX1, PKD1, and their Interactors. This Network Includes Functional and Physical Protein Associations, Highlighting Literature-Curated Interactions with a Medium Confidence Score (0.4)

Functional and Physical Association Network

The first network visualisation (Figure 1) displays a complex interaction landscape comprising functional and physical protein associations. The network reveals several distinct protein clusters organised around key hub proteins. Major interaction hubs include YWHAZ, which strongly connects with YWHAE, YWHAB, FOXO3, and 14-3-3 family proteins, suggesting important regulatory functions. Another significant hub centres around CLU (Clusterin), interconnecting with multiple apolipoprotein family members (APOA1, APOA2, APOA4, APOC3, APOL1, APOM). The PKD1 hub demonstrates connections with TRPC family proteins (TRPC1, TRPC4), PKD2, and PKHD1, highlighting its role in mechanosensation and calcium signalling pathways. PPIA (Peptidyl-prolyl cis-trans isomerase A) forms interactions with proteins involved in cellular stress responses, including H2AX and PPID.

Physical Complex Network

The second network visualisation (Figure 2) highlights direct physical protein complexes, revealing a reorganised topology with more defined functional modules. This network demonstrates stronger clustering patterns, particularly evident in the NME8-centered complex that includes DNAAF1, RSPH4A, DNAI1, DNAI2, DNAH11, CCDC114, and CCDC39/40— proteins predominantly associated with ciliary function and dynein arm assembly.