Join IJHSB mailing list

Subscribe to the Monthly IJHSB news… mailing list to receive details about upcoming issues as they are announced.

Subscribe

Homepage
Submission
For authors

      New Article

      In silico analysis of conservation and structure of FAM163A and 14-3-3β protein interaction

      Authors: Nikoloz Tkebuchava & Lizi Chachua

      LEPL Vladimir Komarov Tbilisi Physics-Mathematics Public School N199, Tbilisi, Georgia

      Publication date: 27.7.2026

      Abstract

      Background/Objective:  FAM163A, or neuroblastoma-derived secretory protein (NDSP), is an uncharacterized protein overexpressed in neuroblastoma and squamous cell lung carcinoma. Currently, data regarding its structure, functional domains, and oncogenic interactions are lacking. This study aims to perform an in silico characterization of FAM163A to determine its 3D structure, mammalian conservation, cellular location, and potential binding interface with the 14-3-3β protein.

      Methods: Evolutionary conservation was determined using multiple sequence align­ment and phylogenetic construction. Structural analysis and intrinsically disordered re­gions (IDRs) were investigated using AlphaFold3, DeepTMHMM, NetSurfP-3.0, SignalP-6.0, and MetaPredictV3. Subcellular localization was predicted via DeepLoc-2.1. Protein-protein interactions and phosphorylation sites were analyzed using HPEPDOCK docking, 14-3-3Pred motif analysis, and NetPhos-3.1.

      Results: Analysis revealed that FAM163A has highly conserved terminal domains flanking a divergent central region and possesses an extended transmembrane he­lix. Additionally, localization predictions place the protein within the cell membrane. A significant feature is a large C-terminal region comprising intrinsically disordered segments and flexible coils (residues 84–115). Notably, the C-terminus presents a candidate 14-3-3β binding motif, characterized by a specific serine residue predicted as a phosphorylation target for PKA or RSK kinases.

      Conclusions: These findings suggest FAM163A may function as a membrane an­chor for 14-3-3β, potentially recruiting it to facilitate Raf-1 activation and sustain the MAPK/ERK pathway. Moreover, the predicted RSK-mediated phosphorylation at the C-terminus points toward the existence of a positive feedback loop that may contribute to increased cancer cell proliferation.

      1. Introduction

      The regulation of cell proliferation is governed by complex signaling networks, and their deregulation is a hallmark of oncogenesis [1]. FAM163A, initially characterized as neuroblastoma-derived secretory protein (NDSP) on human chromosome 1q25.2, was originally identified as a secreted factor overexpressed in neuroblastoma, correlating with advanced disease stages and poor prognosis [2–4]. Subsequent research expanded its pathological scope to lung squamous cell carcinoma (LUSC), where FAM163A acts as a positive regulator of the ERK signaling pathway by upregulating 14-3-3β to promote proliferation [5]. Furthermore, proteomic studies have suggested a ubiquitous role for the protein, identifying it as a “missing” mitochondrial protein [6] implicated in metabolic regulation and obesity [7].

      Despite accumulating evidence of FAM163A’s clinical relevance, a critical disparity exists between its known biological effects and its undefined molecular characteristics. While the protein influences the MAPK/ERK pathway [5] and may reside in the mitochondria [6,7], the structural basis for these activities is unknown. Specifically, ambiguity regarding whether FAM163A functions as a secreted factor [2], a transmembrane anchor, or a protein associated with the mitochondrial proteome [6] prevents the precise mapping of its 14-3-3β-binding motif and phosphorylation sites. This structural gap hinders the mechanistic understanding of how FAM163A sustains cancer cell survival.

      Although recent literature establishes that FAM163A acts as a positive regulator of the ERK pathway via 14-3-3β [5], the molecular mechanism remains unknown. By modeling the FAM163A–14-3-3β interaction, this in silico study aims to generate a structural hypothesis for how FAM163A might influence ERK pathway activation [8]. Furthermore, we use predictive algorithms to evaluate secretory and potential mitochondrial targeting signals. Based on these predictions, we hypothesize that FAM163A may act as a membrane-associated scaffold facilitating Raf-1 activation via 14-3-3β, which may, in turn, promote further phosphorylation of FAM163A, thereby creating a positive feedback loop that sustains MAPK/ERK signaling. Practically, identifying the C-terminal binding interface could expose novel vulnerabilities for targeted inhibitors.

      The primary objective of this study is to conduct a comprehensive in silico characterization of FAM163A to determine its three-dimensional structure, evolutionary conservation, and subcellular sorting signals. Utilizing the canonical human sequence (UniProt ID: Q96GL9), the scope encompasses evolutionary conservation, transmembrane topology, structural modeling, and functional motif annotation. Methodologically, the study employs a multilayered framework integrating phylogenetic investigation with structural prediction to characterize the transmembrane domain and intrinsically disordered regions. Molecular docking simulations were performed to investigate phosphorylation-dependent interactions with 14-3-3β, with specificity evaluated against negative-control models generated through in silico mutagenesis.

      The study is limited by its reliance on computational prediction algorithms without concurrent in vitro validation; consequently, the models remain theoretical. Additionally, the analysis is restricted to static structural properties and does not account for dynamic conformational changes or additional regulatory factors.

      2. Methods

      2.1. Protein Sequence Retrieval and Phylogenetic Analysis

      Protein sequences for FAM163A were retrieved from the UniProt database [10], filtering for complete sequences across 62 mammalian species. Multiple sequence alignment was performed using MUSCLE [9]. Gaps in the consensus sequence were trimmed if more than 50% of the species had a gap at that position. Conservation scores were calculated across all species sequences.

      Phylogenetic reconstruction was conducted in the R programming environment (v4.5.2) [11] using the phangorn [12], msa [13], ape [14], bio3d [15], gridExtra [16], ggplot2 [17], and ggtree [18] packages for plot generation. Phylogenetic reconstruction was performed using a maximum-likelihood (ML) approach. An initial BioNJ tree [19] was used as the starting topology. The final ML tree was then optimized using the JTT+G(4) amino acid substitution model [20], selected based on the lowest BIC score among JTT, LG, WAG, BLOSUM62, and Dayhoff, and nearest-neighbor interchange (NNI) heuristic searches. Branch support was assessed using bootstrap analysis with 1,000 replicates.

      2.2. Structural Prediction and Motif Analysis

      The full-length three-dimensional structure of human FAM163A (UniProt ID: Q96GL9) was predicted using AlphaFold 3 [21]. Structural confidence and secondary-structure features were further assessed using NetSurfP-3.0 [22]. Intrinsically disordered regions (IDRs) were identified using Metapredict V3 [23].

      To resolve discrepancies between prediction tools, a consensus rule was established: regions predicted as highly disordered by Metapredict V3 but scoring below the strict disorder threshold in NetSurfP-3.0, while still being assigned a coil secondary structure, were classified as “flexible coils.” Putative 14-3-3β-binding motifs were predicted using 14-3-3-Pred [24], and the phosphorylation potential of residues within these motifs was assessed using NetPhos-3.1 [25].

      2.3. Subcellular Localization

      Membrane topology was analyzed using DeepTMHMM [26] and SignalP-6.0 [27] to identify transmembrane domains, signal peptides, and the specific number of membrane-spanning residues. Subcellular localization and membrane specificity were determined using DeepLoc-2.1 [28]. Additionally, to corroborate the transmembrane topology, we cross-referenced the DeepTMHMM predictions with Phobius [29].

      2.4. Molecular Docking and Interaction Analysis

      Prior to docking analysis, candidate 14-3-3-binding motifs within the FAM163A sequence were identified and ranked using the 14-3-3-Pred web server [24]. To evaluate the spatial viability of membrane-proximal motifs, the three-dimensional orientation of FAM163A within a lipid bilayer was calculated using the PPM 3.0 web server [30]. To ensure accurate membrane embedding and prevent algorithmic bias from the extensive intrinsically disordered C-terminal tail, a truncated AlphaFold 3 model comprising residues 1–60 was used for the PPM calculation.

      Comparative spatial analysis of the Thr-40 motif was then performed in PyMOL by superimposing a reference 14-3-3β-binding complex onto the membrane-embedded model to assess steric clashes. Following this spatial exclusion analysis, the distal Ser-164 site was prioritized. To evaluate the interaction interface of this prioritized motif, blind protein–peptide docking simulations were performed using the HPEPDOCK server [31].

      The 14-3-3β receptor structure was obtained from the Protein Data Bank (PDB ID: 2BQ0) [32,33]. Within the motif, the serine residue was phosphorylated before the docking simulation. The peptide FTNPRAISTDV, corresponding exactly to FAM163A residues 157–167, was selected to encompass positions −7 to +3 relative to the central Ser-164 residue. The +3 extension represents the complete biological C-terminus of FAM163A, which terminates at residue 167, ensuring that the complete tail was simulated. The −7 upstream extension was selected to include the putative arginine consensus motif, Arg-160 at position −4, which is structurally required for 14-3-3 electrostatic recognition [34,35].

      To simulate the active biological state, the central serine residue was explicitly parameterized as phosphorylated serine (SEP) using HPEPDOCK’s native support for non-standard amino acids. To assess the specificity of this interaction, in silico negative controls were performed. Docking simulations were conducted using both an unphosphorylated wild-type peptide containing Ser-164 and an alanine substitution control (S→A). All three peptide variants—pSer, Ser, and Ala—were docked against the prepared 14-3-3β template using identical protocols.

      Post-docking analysis was conducted using PyMOL [36] to visualize the binding interface and identify interactions involving phosphoserine. The same post-docking analysis framework was applied to the alanine mutant to evaluate the loss of specific interactions.

      The amino acid sequences of human 14-3-3ε (UniProt ID: P62258) and 14-3-3β (UniProt ID: P31946) were retrieved from the UniProt database. Pairwise sequence alignment was performed using ClustalW [37] with default parameters. Conserved residues involved in phosphate coordination were identified by aligning the highly conserved amphipathic groove regions corresponding to helices α3 and α5. Specifically, Lys50, Arg57, Arg130, and Tyr131 in the ε isoform [38] were mapped to Lys51, Arg58, Arg129, and Tyr130 in the β isoform, respectively.

      Because the side-chain coordinates for Lys51 were missing from the deposited 2BQ0 crystal structure, this residue did not influence the rigid-body docking algorithm. Therefore, to conduct post-docking visual inspection and evaluate hydrogen bonds between the active site and the C-terminal motif, the side chain of Lys51 was reconstructed using the mutagenesis tool in PyMOL.

      The rotamer conformation was adjusted to evaluate whether the docked phosphate group was positioned within a biologically viable distance to interact with the lysine ε-amino group (NH₂). This manual reconstruction was used solely to evaluate the theoretical geometry of the basic pocket and confirm that the docked conformation was spatially capable of forming the hydrogen bonds required for phosphate coordination.

      3. Results

      3.1. Phylogenetic Analysis

      3.1.1. Sequence Conservation Analysis

      The initial multiple sequence alignment of mammalian FAM163A sequences and the resulting conservation profile revealed three segments with markedly different levels of conservation. The N-terminal region comprising amino acids 1–40 and the distal C-terminal region exhibited high sequence identity. In sharp contrast, the central region displayed extensive sequence divergence and insertion/deletion events, resulting in lower conservation scores (Figure 1).

      Figure 1. Conservation landscape of the trimmed FAM163A alignment. The plot displays the conservation score (identity) for each amino acid position.

      3.1.2. Phylogenetic Reconstruction

      The maximum-likelihood phylogenetic tree clustered species into established mammalian orders, including Primates (Homo, Macaca, and Sapajus), Rodentia (Mus, Cricetulus, and Castor), and Cetartiodactyla (Tursiops, Balaenoptera, and Bos) (Figure 2).

      Figure 2. Phylogenetic tree of mammalian FAM163A. Species are grouped according to the maximum-likelihood analysis. The x-axis represents branch length based on evolutionary change.

      Bootstrap analysis indicated high statistical support (>70%; green nodes) for terminal branches representing recent divergence events within orders. In contrast, deep internal nodes representing divergence among mammalian orders displayed bootstrap support values below 50% (red nodes), indicating that the nodes separating different clades were unresolved in the consensus tree (Supplementary Figure S1).

      Figure 3. Structural prediction, error assessment, and consensus disorder profile of FAM163A. A) AlphaFold 3 model colored according to pLDDT confidence (blue, >90; orange, <50). B) Predicted Aligned Error (PAE) plot. The heatmap displays the expected positional error for every residue pair. Dark-green regions along the diagonal indicate low error and rigid local structure, whereas diffuse light-green regions indicate high error. C) NetSurfP-3.0 disorder profile. The probability of intrinsic disorder (y-axis) is plotted against residue position (x-axis). The orange line represents the standard disorder threshold of 0.5. D) Metapredict V3 disorder profile. The red curve represents the disorder score. The shaded region indicates residues classified as disordered (score >0.5).

      3.2. Structural Investigation

      To characterize the three-dimensional structure of FAM163A, AlphaFold 3 was used to generate a high-confidence structural model, which was cross-referenced with sequence-based predictions from NetSurfP-3.0 and Metapredict V3. These three tools revealed two principal characteristics of FAM163A:

      N-terminal α-helix. AlphaFold 3 predicted a rigid α-helical conformation for the N-terminal region. High model confidence (pLDDT >90) was observed for residues 3–38 (Figure 3A,B). Secondary-structure analysis using NetSurfP-3.0 identified a similar α-helical region, with high probability scores assigned to residues 6–30. The predictions produced by the two methods overlapped across residues 6–30.

      Intrinsically disordered region. AlphaFold 3 predicted a coil structure for residues 40–167, predominantly with very low confidence scores (pLDDT <50), indicative of high spatial flexibility or an intrinsically disordered region (Figure 3A–D). To quantify this prediction formally, consensus disorder analysis was performed using Metapredict V3 and NetSurfP-3.0. Metapredict V3 predicted the entire C-terminal domain following the N-terminal α-helix to be intrinsically disordered (Figure 3D). NetSurfP-3.0 largely agreed but assigned residues 84–115 scores below the intrinsic-disorder threshold while still identifying them as a coil structure (Figure 3C). Applying our consensus rule, this segment comprising residues 84–115 was classified as a flexible coil rather than a strict intrinsically disordered region. Overall, the consensus strongly supports a predominantly disordered C-terminus.

      3.3. Subcellular Localization

      3.3.1. Subcellular Localization of FAM163A

      The subcellular localization of the protein was predicted using DeepLoc-2.1, suggesting that FAM163A is a transmembrane protein localized to the plasma membrane (Figure 4A).

      3.3.2. Transmembrane Topology

      Membrane topology was investigated using DeepTMHMM and Phobius. DeepTMHMM identified a single transmembrane core spanning residues 6–26 (Figure 4B), which physically overlaps with the continuous N-terminal α-helix comprising residues 3–38 identified in our secondary-structure predictions. The sequences flanking this transmembrane core were classified as a short N-terminal domain comprising residues 1–5 and a long C-terminal domain comprising residues 27–167. Within the C-terminal sequence, the region immediately following the membrane, comprising residues 27–38, corresponds to the remaining extramembranous segment of the α-helix before its transition into the intrinsically disordered region.

      Phobius produced similar results, predicting a transmembrane helix between residues 6 and 28 (Figure 4C). Notably, both prediction tools identified a single transmembrane helix with the C-terminus on the cytosolic side. These results suggest that FAM163A is a single-pass transmembrane protein with an extracellular N-terminus and a cytosolic C-terminus (Figure 4B,C).

      3.3.3. Distinguishing Between a Signal Peptide and a Transmembrane Helix

      To distinguish between a cleaved signal peptide and a stable transmembrane anchor, the N-terminal sequence was analyzed using SignalP-6.0. The algorithm yielded a probability score of <0.001 for the presence of a standard secretory signal peptide.

      Figure 4. Predicted subcellular localization and topology. A) DeepLoc-2.1 probability distribution. The bar chart displays localization likelihoods, and the red horizontal segments indicate the specific probability threshold for each compartment. B) DeepTMHMM topology profile. The plot shows probabilities for transmembrane (red), inside (blue), and outside (pink) regions. C) Phobius topology profile. The plot shows probabilities for transmembrane (purple), non-cytoplasmic (cyan), and cytoplasmic (teal) regions.

      3.4. Binding Motif Analysis and Molecular Docking

      3.4.1. Motif Identification and Kinase Specificity

      Sequence scanning for 14-3-3 interaction motifs using 14-3-3-Pred identified two consensus binding sites supported by two of the three methods: a proximal motif centered on Thr-40, with a consensus score of 0.573, and a distal motif centered on Ser-164, with a consensus score of 0.545.

      To evaluate the biological viability of the slightly higher-scoring Thr-40 motif, comparative spatial analysis was conducted in PyMOL. By structurally aligning the 14-3-3β-binding complex with the Thr-40 region of the membrane-embedded topology model, we observed severe spatial constraints. Accommodating the bulky 14-3-3β dimer at Thr-40 resulted in a direct steric clash with the plasma membrane (Figure 5). Because this topological hindrance would likely prevent stable dimer recruitment in vivo, Thr-40 was excluded. Therefore, based on the distal location of the Ser-164 motif within the disordered C-terminal tail and the absence of membrane-proximity constraints, this site was selected for detailed interaction analysis (Supplementary Figure S2).

      To computationally evaluate the likelihood of Ser-164 phosphorylation, kinase-specific prediction was performed using NetPhos-3.1. Ser-164 was predicted to be a phosphorylation substrate for p90 ribosomal S6 kinase (RSK; score: 0.526) and protein kinase A (PKA; score: 0.624).

      Figure 5. Structural comparison of the Thr-40 motif with the membrane-embedded FAM163A model. The 14-3-3β-binding groove (pink) was aligned with FAM163A residues 38–42 surrounding Thr-40 (yellow). Blue and red dots represent the cytoplasmic and extracellular membrane boundaries, respectively. The alignment indicates steric incompatibility between membrane-proximal Thr-40 binding and the 14-3-3β dimer.

      3.4.2. Molecular Docking Simulations

      Blind protein–peptide docking simulations were performed using the HPEPDOCK server to characterize the interaction interface between the FAM163A C-terminal motif and the 14-3-3β receptor.

      The docking algorithm generated ten top-scoring models for each peptide variant based on shape complementarity. For the phosphorylated wild-type peptide containing pSer-164, HPEPDOCK SHAPE scores ranged from −172.790 to −157.205 (Table 1). In comparison, the unphosphorylated Ser-164 control and the Ala-164 mutant produced scores ranging from −182.143 to −164.569 and from −182.063 to −163.940, respectively (Tables 2 and 3).

      Top-ranked models were analyzed to identify specific polar contacts with the conserved 14-3-3β-binding pocket residues Lys51, Arg58, Arg129, and Tyr130. Interactions were defined using an interatomic distance of <4.0 Å.

      Phosphorylated wild-type. Structural inspection revealed distinct binding modes among the top solutions. Although Models 1 and 2 produced the highest shape-complementarity scores, they failed to orient the phosphate group toward the basic pocket. Model 3 (score: −170.467) was identified as the only conformation exhibiting simultaneous phosphate polar bonds (PPBs) with all four specificity determinants: Lys51, Arg58, Arg129, and Tyr130 (Figure 6).

      Figure 6. Structural interface of the pSer-164 docking complex (Model 3). The phosphorylated serine residue is positioned within the basic binding pocket of 14-3-3β. The FAM163A C-terminal peptide is shown in yellow. Gray dashed lines indicate the network of phosphate polar bonds (PPBs) connecting the peptide phosphate group to the receptor residues Lys51, Arg58, Arg129, and Tyr130, which are shown in purple.

      Unphosphorylated control. Post-docking analysis of unphosphorylated Ser-164 revealed an absence of specific side-chain interactions. Despite producing the most negative raw SHAPE scores, none of the ten models formed the necessary polar bonds between the side chain of unphosphorylated serine and Lys51, Arg58, Arg129, or Tyr130. Three non-specific backbone polar bonds (BPBs), involving Lys51, Arg58, and Tyr130, were observed only in Model 1 (Table 2).

      Alanine mutant. Post-docking analysis of the Ala-164 mutant revealed an absence of specific side-chain interactions. None of the ten models formed polar bonds with Lys51, Arg58, or Arg129. A single non-specific backbone polar bond with Tyr130 was observed in Model 7 (Table 3).

      Table 1. Blind protein–peptide docking scores and polar bonds between pSer-164 and phosphopeptide-binding pocket residues

      ModelHPEPDOCK SHAPE scoreLys51Arg58Arg129Tyr130
      Model 1−172.790N/AN/AN/AN/A
      Model 2−172.089N/AN/AN/AN/A
      Model 3−170.467PPBPPBPPBPPB
      Model 4−162.976BPBN/AN/AN/A
      Model 5−161.048N/AN/AN/AN/A
      Model 6−160.048N/AN/AN/AN/A
      Model 7−158.818N/AN/AN/AN/A
      Model 8−158.795N/AN/AN/AN/A
      Model 9−157.966N/AN/AN/AN/A
      Model 10−157.205N/AN/AN/AN/A

      PPB, phosphate polar bond; BPB, backbone polar bond; N/A, no bond.

      Table 2. Blind protein–peptide docking scores and polar bonds between unphosphorylated Ser-164 and phosphopeptide-binding pocket residues

      ModelHPEPDOCK SHAPE scoreLys51Arg58Arg129Tyr130
      Model 1−182.143BPBBPBN/ABPB
      Model 2−182.032N/AN/AN/AN/A
      Model 3−174.421N/AN/AN/AN/A
      Model 4−173.407N/AN/AN/AN/A
      Model 5−172.657N/AN/AN/AN/A
      Model 6−170.653N/AN/AN/AN/A
      Model 7−167.148N/AN/AN/AN/A
      Model 8−165.762N/AN/AN/AN/A
      Model 9−165.325N/AN/AN/AN/A
      Model 10−164.569N/AN/AN/AN/A

      PPB, phosphate polar bond; BPB, backbone polar bond; N/A, no bond.

      Table 3. Blind protein–peptide docking scores and polar bonds between the Ala-164 mutant and phosphopeptide-binding pocket residues

      ModelHPEPDOCK SHAPE scoreLys51Arg58Arg129Tyr130
      Model 1−182.063N/AN/AN/AN/A
      Model 2−178.042N/AN/AN/AN/A
      Model 3−175.756N/AN/AN/AN/A
      Model 4−172.501N/AN/AN/AN/A
      Model 5−171.738N/AN/AN/AN/A
      Model 6−171.118N/AN/AN/AN/A
      Model 7−168.897N/AN/AN/ABPB
      Model 8−167.067N/AN/AN/AN/A
      Model 9−165.301N/AN/AN/AN/A
      Model 10−163.940N/AN/AN/AN/A

      PPB, phosphate polar bond; BPB, backbone polar bond; N/A, no bond.

      4. Discussion

      Our in silico characterization provides the first comprehensive structural model of FAM163A and challenges its historical classification as a secretory protein in neuroblastoma cells [2]. The absence of a cleavage-competent signal peptide, with a SignalP-6.0 probability of <0.001, and the prediction of a stable transmembrane helix comprising residues 6–26 suggest that FAM163A may function as a single-pass transmembrane protein with an extracellular N-terminus and a cytosolic C-terminus, rather than as a predominantly soluble secretory factor. Although prior proteomic studies have suggested mitochondrial association [7], DeepLoc-2.1 favors plasma membrane localization. However, because these conclusions are based on computational predictions, in vitro validation is required to resolve the subcellular targeting of FAM163A definitively.

      Phylogenetic reconstruction revealed a marked disparity in statistical confidence. Although most terminal nodes representing recent divergence showed high bootstrap support (>70%), deep internal nodes separating major mammalian clades were poorly resolved (<50%). This discrepancy can be attributed to the conservation profile of FAM163A.

      The protein comprises highly conserved N- and C-terminal regions separated by a faster-evolving central intrinsically disordered region. Our analysis suggests that this central domain lacks the conserved informative sites required to resolve ancient evolutionary divergences. Although the terminal regions provide a sufficient signal to group closely related species, the ancestral phylogenetic nodes remain unresolved.

      Furthermore, the marked contrast between the highly conserved terminal regions and the divergent central region suggests that the central IDR may function primarily as a flexible linker connecting the transmembrane helix to the distal C-terminal binding motif. From an evolutionary perspective, such linker regions are often subject to considerably lower sequence constraints than catalytic or binding domains, provided that they retain sufficient length and flexibility to connect functional motifs. Consequently, the specific amino acid sequence of this region may undergo relatively neutral drift, explaining the low bootstrap support for deep phylogenetic nodes. The phylogenetic signal in this region may have been lost through mutational noise over time. As long as the region remains disordered and sufficiently long to project the C-terminal signaling tail into the cytoplasm, the evolutionary pressure to conserve individual residues may remain low.

      The identification of a distal C-terminal binding motif centered on Ser-164 provides a structural hypothesis for the protein’s reported oncogenic activity [5] and supports the proposed classification of FAM163A as a transmembrane anchor protein. Although our computational screen identified two potential sites, structural comparison excluded the proximal Thr-40 motif because of topological constraints. Thr-40 is positioned immediately adjacent to the membrane-anchored α-helix. Because 14-3-3 proteins function as large dimers, our spatial alignment indicated that binding at this membrane-proximal location would produce severe steric clashes with the inner leaflet of the plasma membrane.

      In contrast, the distal location of Ser-164 within the flexible C-terminal tail may improve accessibility for the 14-3-3β dimer. Additionally, the extensive intrinsic disorder predicted in the C-terminal tail, comprising residues 40–167, may facilitate the recruitment of 14-3-3β by increasing the effective capture radius of the motif. This structural flexibility is consistent with a potential role as a signaling hub, because it may allow the C-terminus to sample a larger intracellular volume and thereby increase the probability of encountering 14-3-3β in a crowded cellular environment.

      Our docking simulations indicate that the selected phosphorylated complex is geometrically consistent with electrostatic recognition. Phosphorylated Ser-164 engages the conserved basic tetrad comprising Lys51, Arg58, Arg129, and Tyr130. The failure of both unphosphorylated Ser-164 and the alanine mutant to form these critical side-chain contacts supports a predicted phosphorylation-dependent interaction that may be regulated by RSK or PKA.

      A critical observation from the docking simulations was the discrepancy between the best-scoring models and the model displaying biologically relevant geometry. Although HPEPDOCK ranked Models 1 and 2 highest based on shape complementarity, these conformations displayed a non-canonical outward-facing geometry in which phosphoserine remained solvent-exposed. Only Model 3, despite its slightly less favorable shape score of −170.467 compared with −172.790 for Model 1, reproduced the buried inward-facing geometry required for canonical 14-3-3 interaction.

      This observation highlights a known limitation of blind protein–peptide docking, in which the protein is treated as a rigid body [31]. Such algorithms prioritize the maximization of buried surface area and favorable protein–peptide contacts, producing more negative scores. Consequently, they may overlook specific electrostatic networks and generate false-positive models with more favorable docking scores. Such models can achieve improved geometric packing by positioning the peptide backbone along the groove surface while avoiding the steric penalty associated with inserting the bulky phosphate group.

      Furthermore, the absence of the complete Lys51 side-chain coordinates in the initial receptor structure likely prevented the scoring function from recognizing the electrostatic attraction required for phosphate coordination. This geometric and electrostatic bias may explain why unphosphorylated Ser-164 and the alanine mutant produced slightly more negative raw docking scores than the phosphorylated sequence. Although biological specificity may depend on electrostatic recognition, this contribution is often underestimated by rigid-body geometric scoring functions.

      This discrepancy may also be influenced by the desolvation penalty. In an aqueous environment, phosphorylated serine is stabilized by a dense network of water molecules that form strong hydrogen bonds with the oxygen atoms of the phosphate group, producing a hydration shell surrounding PO₃⁻. For the peptide to bind correctly within the 14-3-3 groove, this water shell must be displaced, which requires an initial energetic investment. This energetic contribution is not typically represented fully in rigid-body docking algorithms. Consequently, the scoring function may favor the alanine mutant and unphosphorylated serine because they lack this bulky, hydrated group. Structural filtering, as performed here, is therefore required to distinguish potentially biologically relevant geometries from algorithmic artifacts.

      Building on the reported upregulation of 14-3-3β by FAM163A [5], our computational findings lead us to propose a “membrane recruiter” hypothesis. In the MAPK/ERK pathway, Raf-1 activation requires translocation to the plasma membrane [8]. Our modeling suggests that by anchoring 14-3-3β to the membrane through its C-terminal tail, FAM163A may increase the local concentration of 14-3-3β at the cell surface and potentially facilitate the recruitment of active Raf-1.

      Because RSK is downstream of Raf-1 activation [8], we further hypothesize that FAM163A overexpression is structurally compatible with a positive feedback loop in which RSK phosphorylates additional FAM163A molecules, which may, in turn, further support Raf-1 activation. Our findings provide a plausible mechanism through which such a positive feedback loop could sustain the hyperproliferation observed in neuroblastoma and LUSC.

      Furthermore, the DeepTMHMM and Phobius topology predictions place the entire intrinsically disordered C-terminus of FAM163A in the cytosol. This cytosolic orientation is a necessary condition for the proposed mechanistic model because it would allow the C-terminal tail and its serine motifs to remain spatially accessible for phosphorylation by kinases such as RSK and for docking with intracellular signaling hubs such as 14-3-3β.

      This study relies exclusively on computational predictions. Although the algorithms used, including AlphaFold 3 and HPEPDOCK, are state-of-the-art, they cannot fully simulate the dynamic conformational entropy of the intrinsically disordered tail. Specifically, the rigid-body docking approach required manual rotamer optimization of Lys51 to represent the induced-fit characteristics of the binding pocket. The current model also does not include the electrostatic landscape of the membrane. Because the inner leaflet of the plasma membrane is rich in negatively charged phospholipids, such as phosphatidylserine, and the 14-3-3β dimer also presents a significant negative surface charge [39], the interaction between these surfaces remains speculative. Whether this environment generates electrostatic repulsion that influences binding kinetics remains to be determined.

      Consequently, the current rigid-body docking models represent static structural snapshots. Future molecular dynamics simulations incorporating a lipid bilayer, alongside in vitro co-immunoprecipitation or binding assays, will be required to characterize these potential membrane-level effects fully. Future therapeutic strategies could target this interface; small molecules mimicking the C-terminal tail could potentially disrupt the FAM163A–14-3-3β interaction, offering a possible intervention for FAM163A-driven cancers.

      5.Conclusion

      Based on in silico predictions, this study proposes reevaluating FAM163A as a single-pass transmembrane protein with a cytosolic C-terminal tail rather than as a secreted factor. The results identify a conserved, predominantly intrinsically disordered C-terminal tail that is predicted to function as a signaling hub.

      Specifically, we characterized a putative 14-3-3β-binding motif centered on Ser-164, where phosphorylation by PKA or RSK is predicted to create a phosphorylation-dependent 14-3-3β-binding interface on FAM163A that may facilitate the recruitment of 14-3-3 dimers to the plasma membrane. This interaction is proposed to serve as a molecular scaffold for localizing Raf-1 to the plasma membrane, thereby potentially amplifying MAPK/ERK signaling and establishing a positive feedback loop through RSK activation. This mechanism may contribute to malignant proliferation in neuroblastoma and lung squamous cell carcinoma.

      Experimental validation of this computationally derived phosphorylation-dependent interface could provide a concrete and accessible target for the development of inhibitors intended to disrupt FAM163A-driven oncogenesis.

      Acknowledgments

      The authors thank the developers of the AlphaFold3, HPEPDOCK, and DeepTMHMM web servers for making their computational tools freely available to the scientific com­munity. This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

      Author Contributions

      N.T. (Nikoloz Tkebuchava) performed the phylogenetic analysis, investigated func­tional motifs, conducted the molecular docking simulations, and led the drafting of the manuscript. He was responsible for conceptualization of the study, data analysis, all data interpretation and the final visualization of structural models. L.C. (Lizi Chachua) provided support in the initial structural modeling and assisted with manuscript review. Both authors approved the final version of the paper.

      Funding Statement: This research received no specific grant from any funding agency.

      Conflicts of Interest: The authors declare no competing interests.

      Ethics Statement: Not applicable (In silico study).

      References

      1. Hanahan, D., Weinberg, R. A. (2011). Hallmarks of cancer: The next generation. Cell, 144, 646–674. https://doi.org/10.1016/j.cell.2011.02.013
      2. Vasudevan, S. A., Shang, X., Chang, S., Ge, N., Diaz-Miron, J. L., Russell, H. V., Hicks,
        M. J., Ludwig, A. D., Wesson, C. L., Burlingame, S. M., Kim, E. S., Khan, J., Yang, J., Nuchtern, J. G. (2009). Neuroblastoma-derived secretory protein is a novel secreted factor overexpressed in neuroblastoma. Molecular Cancer Therapeutics, 8, 2478–2489. https:
        //doi.org/10.1158/1535-7163.mct-08-1132
      3. Vasudevan S. A., Russell H. V., Okcu M. F., Burlingame S. M., Liu Z. J., Yang J., Nuchtern J.
        G. (2007) Neuroblastoma-derived secretory protein messenger RNA levels correlate with high-risk neuroblastoma. Journal of Pediatric Surgery, 42, 148–152. https://doi.org/ 10.1016/j.jpedsurg.2006.09.064
      4. Chen Y. L., Li X. L., Li G., Tao Y. F., Zhuo R., Cao H. B., Jiao W. Y., Li Z. H., Zhu Z. H., Fang
        F., Xie Y., Liao X. M., Wu D., Wang H. R., Yu J. J., Jia S. Q., Yang Y., Feng C. X., Yang P. C., Fei X. D., Wang J. W., Xu Y. Y., Qian G. H., Zhang Z. M., Pan J. (2022). BRD4 inhibitor GNE987 exerts anti-cancer effects by targeting super-enhancers in neuroblastoma. Cell & Bioscience, 12. https://doi.org/10.1186/s13578-022-00769-8
      5. Liu, N., Zhou, H., Zhang, X., Cai, L., Li, J., Zhao, J., Liu, Y., Wang, L., Fan, C., Li, A., Miao, Y. (2019). FAM163A, a positive regulator of ERK signaling pathway, interacts with 14-3-3β and promotes cell proliferation in squamous cell lung carcinoma. OncoTargets and Therapy, 12, 6393–6406. https://doi.org/10.2147/OTT.S214731
      6. Ronci, M., Pieroni, L., Greco, V., Scotti, L., Marini, F., Carregari, V. C., Cunsolo, V., Foti, S., Aceto, A., Urbani, A. (2018). Sequential fractionation strategy identifies three missing proteins in the mitochondrial proteome of commonly used cell lines. Journal of Proteome Research, 17, 4307–4314. https://doi.org/10.1021/acs.jproteome.8b00422
      7. Erdogan, C. S., Yavuz, Y., Ozgun, H. B., Bilgin, V. A., Agus, S., Kalkan, U. F., Yilmaz, B. (2025). Fam163a knockdown and mitochondrial stress in the arcuate nucleus of hypothala-mus reduce AgRP neuron activity and differentially regulate mitochondrial dynamics in mice. Acta Physiologica, 241, e70020. https://doi.org/10.1111/apha.70020
      8. McCubrey, J. A., Steelman, L. S., Chappell, W. H., Abrams, S. L., Wong, E. W. T., Chang,
        F., Lehmann, B., Terrian, D. M., Milella, M., Tafuri, A., Stivala, F., Libra, M., Basecke, J., Evangelisti, C., Martelli, A. M., Franklin, R. A. (2007). Roles of the Raf/MEK/ERK pathway in cell growth, malignant transformation and drug resistance. Biochimica et Biophysica Acta (BBA) – Molecular Cell Research, 1773, 1263–1284. https://doi.org/10.1016/ j.bbamcr.2006.10.001
      9. Edgar, R. C. (2004). MUSCLE: Multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Research, 32, 1792–1797. https://doi.org/10.1093/nar/ gkh340
      10. The UniProt Consortium. (2024). UniProt: The universal protein knowledgebase in 2025.
        Nucleic Acids Research, 53, 609–617. https://doi.org/10.1093/nar/gkae1010
      11. R Core Team. (2025). R: A language and environment for statistical computing [Computer software]. R Foundation for Statistical Computing. https://www.R-project.org/
      12. Schliep, K. P. (2011). phangorn: Phylogenetic analysis in R. Bioinformatics, 27, 592–593.
        https://doi.org/10.1093/bioinformatics/btq706
      13. Bodenhofer, U., Bonatesta, E., Horejs-Kainrath, C., Hochreiter, S. (2015). msa: An R package for multiple sequence alignment. Bioinformatics, 31, 3997–3999. https://doi. org/10.1093/bioinformatics/btv494
      14. Paradis, E., Schliep, K. (2019). ape 5.0: An environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics, 35, 526–528. https://doi.org/10.1093/ bioinformatics/bty633
      15. Grant, B. J., Rodrigues, A. P. C., ElSawy, K. M., McCammon, J. A., Caves, L. S. D. (2006). Bio3d: An R package for the comparative analysis of protein structures. Bioinformatics, 22, 2695–2696. https://doi.org/10.1093/bioinformatics/btl461
      16. Auguie, B. (2017). gridExtra: Miscellaneous functions for “grid” graphics (R package version 2.3) [Computer software]. https://CRAN.R-project.org/package=gridExtra
      17. Wickham, H. (2016). ggplot2: Elegant graphics for data analysis. Springer-Verlag New York.
        https://ggplot2.tidyverse.org
      18. Yu, G., Smith, D. K., Zhu, H., Guan, Y., Lam, T. T. (2016). ggtree: An R package for visualization and annotation of phylogenetic trees with their covariates and other associated data. Methods in Ecology and Evolution, 8, 28–36. https://doi.org/10.1111/2041-210X.12628
      19. Gascuel, O. (1997). BIONJ: An improved version of the NJ algorithm based on a simple model of sequence data. Molecular Biology and Evolution, 14, 685–695. https://doi. org/10.1093/oxfordjournals.molbev.a025808
      20. Jones DT, Taylor WR, Thornton JM. (1992) The rapid generation of mutation data matri­ces from protein sequences. Bioinformatics 8, 275–282. https://doi.org/10.1093/ bioinformatics/8.3.275
      21. Abramson, J., Adler, J., Dunger, J., Evans, R., Green, T., Pritzel, A., Ronneberger, O., Willmore, L., Ballard, A. J., Bambrick, J., Bodenstein, S. W., Evans, D. A., Hung, C., O’Neill, M., Reiman, D., Tunyasuvunakool, K., Wu, Z., Žemgulyte˙, A., Arvaniti, E., … Jumper, J. M. (2024). Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature, 630, 493–500. https://doi.org/10.1038/s41586-024-07487-w
      22. Høie, M. H., Kiehl, E. N., Petersen, B., Nielsen, M., Winther, O., Nielsen, H., Hallgren, J., Marcatili, P. (2022). NetSurfP-3.0: Accurate and fast prediction of protein structural features by protein language models and deep learning. Nucleic Acids Research, 50, 510–515. https://doi.org/10.1093/nar/gkac439
      23. Lotthammer, J. M., Hernández-García, J., Griffith, D., Weijers, D., Holehouse, A. S., Eme-necker, R. J. (2024). Metapredict enables accurate disorder prediction across the Tree of Life. bioRxiv. https://doi.org/10.1101/2024.11.05.622168
      24. Madeira, F., Tinti, M., Murugesan, G., Berrett, E., Stafford, M., Toth, R., Cole, C., MacK-intosh, C., Barton, G. J. (2015). 14-3-3-Pred: Improved methods to predict 14-3-3-binding phosphopeptides. Bioinformatics, 31, 2276–2283. https://doi.org/10.1093/ bioinformatics/btv133
      25. Blom, N., Sicheritz-Pontén, T., Gupta, R., Gammeltoft, S., Brunak, S. (2004). Prediction
        of post-translational glycosylation and phosphorylation of proteins from the amino acid sequence. Proteomics, 4, 1633–1649. https://doi.org/10.1002/pmic.200300771
      26. Hallgren, J., Tsirigos, K. D., Pedersen, M. D., Armenteros, J. J. A., Marcatili, P., Nielsen, H., Krogh, A., Winther, O. (2022). DeepTMHMM predicts alpha and beta transmembrane proteins using deep neural networks. bioRxiv. https://doi.org/10.1101/2022.04. 08.487609
      27. Teufel, F., Armenteros, J. J. A., Johansen, A. R., Gíslason, M. H., Pihl, S. I., Tsirigos, K. D., Winther, O., Brunak, S., v. Heijne, G., Nielsen, H. (2022). SignalP 6.0 predicts all five types of signal peptides using protein language models. Nature Biotechnology, 40, 1023–1025. https://doi.org/10.1038/s41587-021-01156-3
      28. Ødum, M. T., Teufel, F., Thumuluri, V., Armenteros, J. J. A., Johansen, A. R., Winther, O., Nielsen, H. (2024). DeepLoc 2.1: Multi-label membrane protein type prediction using protein language models. Nucleic Acids Research, 52, 215–220. https://doi.org/10.1093/ nar/gkae237
      29. Käll, L., Krogh, A., Sonnhammer, E. L. L. (2007). Advantages of combined transmembrane topology and signal peptide prediction—the Phobius web server. Nucleic Acids Research, 35, 429–432. https://doi.org/10.1093/nar/gkm256
      30. Lomize A. L., Todd S. C., Pogozheva I. D. (2022). Spatial arrangement of proteins in planar and curved membranes by PPM 3.0. Protein Science, 31, 209–220. https://doi.org/ 10.1002/pro.4219
      31. Zhou, P., Jin, B., Li, H., Huang, S. (2018). HPEPDOCK: A web server for blind pep-tide–protein docking based on a hierarchical algorithm. Nucleic Acids Research, 46, 443–
        450. https://doi.org/10.1093/nar/gky357
      32. Yang, X., Lee, W. H., Sobott, F., Papagrigoriou, E., Robinson, C. V., Grossmann, J. G., Sundström, M., Doyle, D. A., Elkins, J. M. (2006). Structural basis for protein–protein interactions in the 14-3-3 protein family. Proceedings of the National Academy of Sciences, 103, 17237–17242. https://doi.org/10.1073/pnas.0605779103
      33. Yang, X., Elkins, J. M., Fedorov, O., Longman, E. J., Sobott, L., Ball, L. J., Sundstrom, M., Arrowsmith, C., Edwards, A., Doyle, D. A. (2005). PDB ID: 2BQ0. Protein Data Bank. https://doi.org/10.2210/pdb2bq0/pdb
      34. Yaffe B. M., Rittinger K., Volinia S., Caron R. P., Aitken A., Leffers H., Gamblin J. S., Smerdon J. S., Cantley C. L. (1997). The Structural Basis for 14-3-3:Phosphopeptide Binding Specificity. Cell, 91, 961-971. https://doi.org/10.1016/S0092-8674(00)80487-0
      35. Johnson, C., Crowther, S., Stafford, M. J., Campbell, D. G., Toth, R., MacKintosh, C. (2010). Bioinformatic and experimental survey of 14-3-3-binding sites. Biochemical Journal, 427, 68–78. https://doi.org/10.1042/BJ20091834
      36. Schrödinger, LLC. (2015). The PyMOL Molecular Graphics System (Version 3.0) [Computer software]. https://www.pymol.org/
      37. Thompson, J. D., Higgins, D. G., Gibson, T. J. (1994). CLUSTAL W: Improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Research, 22, 4673–4680. https:
        //doi.org/10.1093/nar/22.22.4673
      38. Kamayirese, S., Hansen, L. A., Lovas, S. (2025). Ligand recognition by 14-3-3 proteins requires negative charges but not necessarily phosphorylation. FEBS Letters, 599, 838–847. https://doi.org/10.1002/1873-3468.15077
      39. Aljabal, G., Yap, B. K. (2020). 14-3-3σ and its modulators in cancer. Pharmaceuticals, 13.
        https://doi.org/10.3390/ph13120441

      Supplementary Material

      Code and FASTA files can be found in this github repository: https://github.com/NikoMolecule/Supplementary-Material-For-Research

      Supplementary Figure S1. Bootstrap values of the Phylogenetic Tree Node colors represent bootstrap support percentages (Green = 100%, Red = 0%).

      Supplementary Figure S2. 14-3-3Pred predictions of potential serine/threonine motifs. Blue Amino Acids were supported by two methods out of three, and Orange Amino Acids were supported by one method out of three. The red-boxed serine and neighboring sequence were used for detailed analysis