Fibroblast Gene Profiling Reveals Inflammatory and Remodeling Signals in PIK3CA Overgrowth Spectrum

Researchers in Japan have used microarray-based expression profiling to chart the molecular differences between skin cells carrying and lacking a somatic PIK3CA mutation in a patient with PIK3CA-related overgrowth spectrum, or PROS. By comparing fibroblasts grown from affected skin with fibroblasts from unaffected skin of the same individual, the team identified a short list of candidate messenger RNAs and microRNAs that were markedly altered only in the mutation-bearing tissue. The findings, published as an exploratory study in Frontiers in Genetics, point toward inflammatory signaling, extracellular matrix remodeling, and developmental pathways as molecular features accompanying the segmental tissue overgrowth that defines this rare group of disorders.

PROS is an umbrella term proposed around 2013 for a set of conditions previously considered separate entities, including megalencephaly-capillary malformation syndrome and CLOVES syndrome, which combines congenital lipomatous overgrowth, vascular malformations, epidermal nevi, and skeletal anomalies. What unites them is tissue overgrowth affecting multiple organs: skin, adipose tissue, blood vessels, bones, and the nervous system. The underlying cause is not an inherited mutation present in every cell, but a postzygotic, mosaic mutation in PIK3CA, the gene encoding the p110α catalytic subunit of phosphoinositide 3-kinase alpha. Because the mutation arises during embryonic development, only a subset of cells carries it, producing patchy, segmental overgrowth rather than a uniform disorder.

The recurrent activating variants found in PROS cluster at specific hotspots. Mutations such as E542K and E545K in the helical domain disrupt the inhibitory interaction between the p110α catalytic subunit and the p85α regulatory subunit, while the kinase-domain mutation H1047R induces conformational changes that reduce p85α-mediated autoinhibition and enhance membrane-associated kinase activity. The result in both cases is increased production of phosphatidylinositol-3,4,5-trisphosphate, or PIP3, and constitutive activation of the downstream AKT–mTOR signaling cascade, a pathway central to cell growth, metabolism, and survival. The relationship between individual PIK3CA variants and specific clinical phenotypes, however, is not absolute, and the signaling networks downstream of AKT–mTOR remain only partially charted.

PIK3CA activation is among the most common genetic alterations in human cancers, yet PROS is fundamentally distinct from neoplastic disease. In PROS, the mosaic mutation promotes tissue overgrowth, hypertrophy, and developmental dysregulation rather than invasive growth and metastatic progression. The same activating variants can appear in both contexts, which suggests that the biological consequences of PI3K activation depend heavily on developmental timing, mosaic distribution, and cellular and tissue context. Historically, treatment consisted mainly of symptomatic interventions such as surgery and laser therapy. More recently, PI3K inhibitors such as alpelisib, which targets PI3Kα directly, have emerged as molecularly targeted options reported to reduce lesion volume and improve pain and functional outcomes. Their efficacy for neurological manifestations, including neurodevelopmental delay, cognitive impairment, and epilepsy, remains incompletely defined, and it is unclear whether targeted inhibition can prevent or reverse established developmental abnormalities.

The patient at the center of the new study was born at term after a normal pregnancy and was followed for macrocephaly with massive hyperostosis of the skull and psychomotor delay. Tonic seizures affecting the right side of the body appeared on postnatal day 3, and brain MRI revealed left hemimegalencephaly. Because the seizures were medically intractable, a hemispherotomy was performed at 1 year and 7 months of age, after which antiepileptic drugs were not required until seizure recurrence at 21 years. Around age 12, protrusion of the left frontal region became apparent and progressed slowly; around age 20, rapid enlargement of the left frontal region and facial overgrowth caused superior compression of the left globe. Cranial reconstruction involving the frontal bone and zygoma was performed at 21 years of age, and histopathological examination of the lesional tissue was consistent with epidermal nevus, a cutaneous manifestation known to arise from somatic mosaic PIK3CA mutations in a subset of cases. Slow regrowth continued, with pain in the protruding frontal region from approximately 29 years of age, controlled with analgesics.

Genetic diagnosis rested on whole exome sequencing of peripheral blood lymphocytes, parental samples, and both affected and unaffected skin. A PIK3CA mutation, NM_006218.3:c.3140A>G, producing the p.(His1047Arg) amino acid change, was detected exclusively in the affected skin samples. High-resolution melting analysis, which detects heteroduplex formation from sequence variants, showed an abnormal melting curve only in affected skin, while the patient’s unaffected skin and lymphocytes and his father’s lymphocytes showed identical normal profiles. This confirmed a somatic mosaic PIK3CA mutation confined to the lesion, establishing the PROS diagnosis. The experimental design then exploited this mosaicism: the team established fibroblast cultures from affected skin, designated pv-Fibro, and from unaffected skin of the same patient, designated non-pv-Fibro, and confirmed that the PIK3CA mosaic mutation was present only in pv-Fibro cells. As a reference, fibroblasts from three healthy individuals were pooled and analyzed as a single control sample, nc-Fibro. Using Affymetrix whole-transcriptome and microRNA microarrays, the researchers profiled 24,034 human mRNAs, of which 17,599 passed filtering, and 4,603 human microRNAs. Because the affected and unaffected cells came from the same genetic and environmental background, the intra-individual comparison minimized confounding from sex, treatment history, and inter-individual variability that complicates cross-patient studies.

The microarray data revealed a clear transcriptional signature. Comparing pv-Fibro with nc-Fibro, 38 genes were upregulated at least fivefold and 36 were downregulated to 0.2-fold or less. Functional enrichment analysis of the upregulated genes highlighted extracellular matrix organization, regulation of cellular responses to growth factor stimuli, axon guidance, cell–cell adhesion, and myometrial relaxation and contraction pathways. STRING network analysis supported this picture: of 407 genes annotated in the Gene Ontology term collagen-containing extracellular matrix, 7 of the 38 input genes mapped to that term with statistically significant enrichment, at p = 0.0172. The downregulated genes were enriched in pathways including ABC transporters in lipid homeostasis, regulation of blood circulation, NABA extracellular matrix glycoproteins, negative regulation of cell migration, tube morphogenesis, and regulation of osteoblast differentiation, with 3 of 36 genes mapping to an ABC transporter-related category at p = 0.0306. The authors emphasized that this transporter enrichment is a pathway-level observation whose biological significance requires further investigation.

To isolate lesion-associated changes, the team applied a two-step filter: genes had to change at least fivefold in pv-Fibro relative to nc-Fibro while showing less than a twofold change in non-pv-Fibro relative to nc-Fibro, or the mirror-image criterion for downregulation. This yielded ten candidate lesion-associated upregulated mRNAs: WFDC1, L1CAM, MX1, KCNMB1, OLFM2, MYH11, CORIN, DOK5, EPHA5, and MT1F. These fall into overlapping functional groups. WFDC1, L1CAM, and EPHA5 relate to extracellular matrix organization, cell adhesion, and contact-dependent signaling; L1CAM mediates homophilic and heterophilic adhesion through integrins and other surface receptors, while EPHA5 is a receptor tyrosine kinase governing tissue boundary formation. L1CAM, DOK5, and EPHA5 have also been implicated in nervous system development and neuronal differentiation. MX1 and MT1F connect to interferon signaling, oxidative stress responses, and cellular defense, with WFDC1’s anti-inflammatory properties potentially counterbalancing these responses. OLFM2 has been linked to adipocyte differentiation and metabolism, and CORIN regulates blood pressure and fluid balance by activating atrial natriuretic peptide. KCNMB1, MYH11, and CORIN together point to vascular smooth muscle contractility and vascular tone, suggesting vascular remodeling and altered local hemodynamic regulation within the lesion. Three mRNAs passed the downregulation filter: SFRP1, HOXB3, and NRN1. SFRP1 is an inhibitor of Wnt signaling, and its reduced expression has been associated with altered Wnt signaling and tumorigenesis in other disease contexts, raising the possibility that Wnt-related signaling is altered in the affected fibroblasts. HOXB3 is a homeobox transcription factor involved in embryonic patterning of neural crest- and mesoderm-derived tissues, so its reduction may reflect altered developmental programs associated with PIK3CA-driven overgrowth. NRN1 is a neurotrophic molecule involved in neuronal development and synaptic plasticity, though its role in fibroblasts remains unclear.

The microRNA arm of the study produced the most striking single finding. Among the 4,603 microRNAs examined, six were upregulated and four were downregulated at least fivefold in pv-Fibro, and after the same filtering, two lesion-associated candidates emerged: hsa-miR-146a-5p, upregulated approximately sixfold, and hsa-miR-146b-5p, downregulated to 0.17-fold of control levels. The miR-146 family is known to negatively regulate NF-κB signaling by targeting TRAF6 and IRAK1, functioning as feedback regulators of innate immune and inflammatory responses, so the opposing behavior of the two family members may reflect altered inflammatory signaling in PROS fibroblasts. The authors cautioned, however, that NF-κB activity was not directly measured and no functional experiments were performed, so no causal relationship is established. The two microRNAs share highly similar seed sequences yet are encoded at distinct genomic loci and subject to different transcriptional regulation, warranting isoform-specific interpretation. Computational target prediction using miRDB identified 30 predicted targets of miR-146a-5p with scores above 90, four of which, WWC2, NUMB, CARD10, and ZNF540, are expressed in fibroblast tissues, but none overlapped with the mRNAs identified in the study. TargetScanHuman 8.0 likewise predicted none of the lesion-associated mRNAs as targets, and DOK5 was predicted as a miR-146b-5p target only with a relatively low score of 52, indicating no direct correspondence could be established in this dataset.

The study’s limitations are substantial and acknowledged by the authors. It included samples from a single patient, precluding assessment of inter-individual variability, and the pooled three-donor control prevented conventional statistical comparisons; differential expression was assessed solely by fold-change thresholds without formal statistical testing. The work used dermal fibroblasts, which cannot recapitulate the cellular heterogeneity and tissue architecture of PROS lesions in vivo, and no RT-qPCR validation or functional assays confirmed the candidate molecules. The authors also note that prior single-cell long-read RNA sequencing of MCAP lesions by Wedemeyer and colleagues found PIK3CA-mutant cells enriched in a PAX3-positive fibroblast-like population with an immature mesenchymal program, findings not directly comparable to this tissue-level analysis because of differences in tissues and resolution. The researchers accordingly frame their results as exploratory and hypothesis-generating: inflammatory, tissue remodeling, and developmental signaling pathways emerge as candidates for future investigation in PROS, potentially alongside emerging strategies targeting dysregulated microRNAs such as the miR-146 family, but none can yet be considered validated markers or therapeutic targets. Validation in additional PROS cases, functional microRNA–target interaction assays, and comparative analyses with appropriate disease controls will be necessary before the molecular landscape sketched here can be confirmed as a driver of tissue overgrowth. The underlying expression datasets have been deposited in the GEO repository under accession numbers GSE349641 and GSE349642.

Subject of Research: Exploratory mRNA and microRNA profiling identifies transcriptomic alterations in the PIK3CA-related overgrowth spectrum

Article Title: Exploratory mRNA and microRNA profiling identifies transcriptomic alterations in the PIK3CA-related overgrowth spectrum

Article References: Shiohama, T., Uegami, M., Kubota, Y., Saitsu, H., Kato, M., Matsushima, J., Sawada, D., Yoshii, S., Mitsukawa, N., & Hamada, H. (2026). Exploratory mRNA and microRNA profiling identifies transcriptomic alterations in the PIK3CA-related overgrowth spectrum. Frontiers in Genetics, 17, Article 1935186. https://doi.org/10.3389/fgene.2026.1935186

Image Credits: AI Generated

DOI: 10.3389/fgene.2026.1935186

Keywords: Exploratory, mRNA, microRNA, profiling, identifies, transcriptomic, alterations, PIK3CA-related, overgrowth, spectrum, scientific research