Kidney Co-Expression Map Uncovers Circadian and Metabolic Shifts Before Cysts Form

Autosomal dominant polycystic kidney disease begins its molecular work long before the fluid-filled cysts that define it become visible. In a study published in the Journal of Molecular Medicine, researchers applied a cross-species co-expression analysis to kidneys from inducible Pkd1-knockout mice and found that coordinated transcriptional programs, including circadian regulators, sterol biosynthesis and suppressed mitochondrial pathways, were already measurable during the morphologically normal precystic phase. Because standard differential expression detected only scattered single-gene changes at these early stages, the modular approach allowed the team to resolve weak, distributed signals into coherent biological programs.

The team, led by Cesare A. Marino, Steven J. Kunnen, Bob van de Water and Dorien J. M. Peters, used Weighted Gene Co-expression Network Analysis, or WGCNA, which clusters genes into modules based on shared transcriptional control. Rather than building a new mouse-specific network, they projected log2 fold-change vectors from the Pkd1-mutant kidneys onto 347 curated rat kidney co-expression modules from the TXG-MAPr framework, a reference built from large compendia of compound-treated rat kidneys with matched histopathology. Mouse gene identifiers were converted to rat orthologs using the Rat Genome Database mapping, retaining 78 to 79 percent of expressed genes.

The mouse model, iKsp-Pkd1del, carries a kidney-specific deletion of Pkd1 exons 2 through 11 induced with tamoxifen at postnatal day 38 or 90. Kidneys were harvested at precystic stages of 2, 3 and 6 weeks post-induction, at a moderate stage of 12.4 weeks, and at an advanced stage of 16.4 weeks, alongside non-induced controls. Only males were used to eliminate sex-related variation. Disease stage was confirmed by plasma blood-urea nitrogen, kidney weight to body weight ratios and histology, and the precystic samples corresponded to those in a previously published 2018 dataset.

Across the precystic stages, when cysts and tubular dilations were virtually absent, conventional differential expression detected few significantly dysregulated genes despite more than 10,000 expressed transcripts per dataset, using thresholds of adjusted p value below 0.05 and absolute log2 fold change of at least 1. The authors note that early transcriptional alterations following Pkd1 loss are subtle at the single-gene level and insufficient to define coherent programs by differential expression alone. Moderate and advanced stages, by contrast, showed hundreds of differentially expressed genes, including the stress-response genes Atf3, Egr2 and Socs3 and the tubular-injury markers Havcr1 and Timp1.

Module prioritization retained 42 transcriptionally active modules, 21 of them already deregulated during the precystic phase and 30 at moderate-to-advanced disease, with 9 shared between the two. Together these modules captured 8,212 of 11,888 mapped orthologues. Three temporal patterns emerged across the 42 modules: progressive activation with disease severity, an early peak with moderate decline, and persistent early repression. The early-activated module 217 includes Pkd2, the gene encoding polycystin-2, the principal interactor of polycystin-1, and its high eigengene activity at the earliest time points suggests early perturbation of the polycystin complex, although the module is small and functionally heterogeneous and showed no significant pathway enrichment.

The functional identities of the early modules were striking. Module 33 was enriched for circadian rhythm terms and was strongly upregulated from week 2, with transcription-factor enrichment pointing to autoregulatory clock control by BMAL1, PER1 and CLOCK, which were uniquely associated with this module. Module 124, enriched for cholesterol biosynthesis under SREBP2 control, pointed to early metabolic rewiring that the authors suggest may support membrane integrity, ciliary signaling and vesicular trafficking in early cystogenesis before declining as epithelial polarity is lost. In the opposite direction, modules 28m and the late-stage module 96m were enriched for mitochondrial oxidative phosphorylation and showed early, sustained repression consistent with energy-metabolism suppression after Pkd1 loss.

Inflammatory and cytokine signaling also emerged early and progressed with disease stage. Modules 29 and 64m were strongly enriched for TNF-alpha/NF-kappaB pathways, with module 29 additionally capturing hypoxia and MAPK signaling consistent with a general stress-response role. Transcription-factor enrichment identified NF-kappaB1 and RELA targets in seven and five modules respectively, with p53 and JUN each associated with four modules, and STAT3 among the regulators of the inflammatory branch. Two further immune modules, 56 and 7m, were more strongly induced at moderate and advanced stages, while late proliferative and plasticity responses were captured by modules 5m, enriched for E2F targets, and 10m, enriched for epithelial-mesenchymal transition genes.

Spatial validation using a previously published 10x Genomics Visium dataset from kidneys with focal cyst formation after scattered Pkd1-deletion confirmed that these programs occupy distinct renal microdomains. The circadian module 33 showed a predominantly cortical distribution in control tissue, with local enrichment in pericystic areas of small to moderate cysts and reduced expression around the largest cysts, consistent with its declining eigengene score. The mitochondrial modules 28m and 103 showed broadly reduced expression around cysts of all sizes, supporting spatially pervasive mitochondrial repression, while the inflammatory modules 29 and 64m showed localized enrichment around cystic epithelia and adjacent stroma, confirming focal cytokine activation in the cyst microenvironment.

The authors are careful to frame the work as a discovery and framework analysis rather than a causal demonstration. Eigengene scores capture module-level trends but may obscure heterogeneous dynamics within individual genes. The data are bulk RNA-seq integrating diverse kidney cell types, projected onto rat rather than mouse modules, which could attenuate cell- and species-specific effects, although incomplete ortholog mapping causes false negatives rather than spurious signal because unmapped genes are absent rather than misassigned. Transcription-factor enrichment relied on the curated TRRUST-v2 database and did not directly infer TF activity or binding, so the reported associations represent statistical overrepresentation and should be read as putative regulatory links. The eigengene cutoff itself is a distribution-based prioritization filter rather than a significance test, with statistical inference resting on the limma-estimated fold changes used as input.

Placed against prior evidence, the findings suggest a layered trajectory of cystogenic remodeling. Genetic disruption of Bmal1/Arntl accelerates cystogenesis and fibrosis in Pkd1-deficient mice, and Pkd1 loss is known to dampen clock gene oscillations in cystic kidneys and cell-autonomously in renal epithelial cells, supporting a bidirectional link between polycystin-1 loss and circadian dysregulation. Proteomic and metabolomic studies have likewise pointed to early glycolytic reprogramming and impaired fatty-acid oxidation in PKD, consistent with the view that mitochondrial dysfunction is an initiating rather than terminal event. In this study, circadian disruption, sterol-pathway activation, mitochondrial suppression and low-magnitude inflammatory engagement emerged by week 2 as an intrinsic epithelial phase, followed by strengthening inflammatory and injury responses between weeks 3 and 6, and culminating in chronic immune-fibrotic remodeling at moderate and late stages that underlies cyst expansion. Because the modules are externally defined rather than rebuilt from the input data, the authors attribute this resolution to the projection framework itself, and they position the modular map as a scaffold for stage-specific biomarker discovery and therapeutic targeting in ADPKD.

Subject of Research: Early transcriptional module changes in precystic autosomal dominant polycystic kidney disease

Article Title: Weighted co-expression reveals early circadian and metabolic dysregulation in precystic kidneys

Article References: Marino, C. A., Kunnen, S. J., van de Water, B., & Peters, D. J. M. (2026). Weighted co-expression reveals early circadian and metabolic dysregulation in precystic kidneys. Journal of Molecular Medicine, 104(1), Article 126. https://doi.org/10.1007/s00109-026-02731-4

Image Credits: AI Generated

DOI: 10.1007/s00109-026-02731-4

Keywords: ADPKD, Pkd1, WGCNA, TXG-MAPr, circadian dysregulation, mitochondrial repression, SREBP2, NF-kappaB, spatial transcriptomics, precystic phase, kidney, transcriptomics