Copper-stress-associated cell-state remodeling and auditable candidate prioritization in AML12 hepatocytes

Abstract

Background: Copper imbalance is relevant to liver injury, Wilson disease, metabolic liver disease and copper-ionophore cancer therapy. Cuproptosis is a literature-defined copper-dependent death concept, but its downstream structural events, candidate boundary and cross-cell transferability remain unresolved. A 2026 Cell study proposed copper-induced MYH9 polymerization and actin disruption as an external mechanism template. That template can guide candidate ranking, but it cannot substitute for independent validation in AML12 cells. Methods: I reanalyzed GSE300382, a mouse AML12 experiment containing control, CuCl2 plus elesclomol (Cu/ES), Cu/ES plus taurocholic acid (TCAu) and TCAu alone, with three biological replicates per group. Associated SRA reads were uniformly processed against GRCm39 to generate integer counts; DESeq2 was the primary differential-expression and factorial model, and edgeR and limma-voom were used for confirmation on the same 12 samples. The public FPKM matrix was analyzed with limma only as a historical comparison with the original public matrix, not as the primary inference framework. A prespecified universe of 77 structural, copper-handling, metabolic and stress-related genes was assessed in template-referenced, leave-template-out and template-reduced class-adjusted models; MYH9 was used only as a fixed external template and was not included in candidate ranking. HPA was used only as human expression context with zero weight. PXD037278 was removed because its raw-file-to-condition mapping could not be verified. Results: DESeq2 identified broad Cu/ES-associated changes in both copper-containing contrasts (12,414 and 12,677 genes at FDR =1). edgeR identified 12,470 and 12,676 genes, and limma-voom identified 13,432 and 13,515 genes for the same contrasts. TCAu alone versus control produced a much smaller response (353, 343 and 379 genes, respectively). Count-based results were directionally concordant for 15 of 17 prespecified key genes; Myh9 and Mt2 were discordant, Atp7a was concordant but not significant under limma-voom, and FDX1 was negative and significant under all three count methods. The factorial interaction was method-sensitive and was retained only as a secondary technical sensitivity. Copper-containing conditions showed increased proteostasis and redox-defense signatures and reduced mitochondrial, cell-cycle, adhesion, cytokinesis and actin-myosin transcript programs. Count-level composition changed: genes with count>0 increased, the count>=10 set remained nearly stable, and the ribosomal read fraction rose; FPKM-based detected-gene and ribosomal metrics are reported separately. Candidate rankings were sensitive to template and class priors. MYH9 was used only as an external template and was excluded from the candidate ranking; its separate sensitivity term moved from rank 2 to rank 48 when template-linked evidence was removed, which describes model behavior rather than a biological rank. CD2AP, RDX, PDHA1, FDX1, LIAS and PDHB retained higher model-based priority within the prespecified universe in some models, not data-supported mechanisms. ATP7A-MYH9 was a direction-agnostic transcript association, and MYH9-ANLN remained an external interaction hypothesis. Structural context: Ten AlphaFold3/ColabFold v2 input specifications were generated but not executed, and no predicted coordinates or copper-conditioned results exist. Three homologous or orthologous templates completed copper-free, short-timescale molecular dynamics; these exploratory calculations do not provide binding, copper-coordination or AML12-complex evidence. Conclusion: Cu/ES is associated with broad AML12 cell-state remodeling and supports template-dependent model-based prioritization across structural, copper-handling, metabolic and stress-response candidate layers. The main contribution is a transparent, computationally auditable candidate-priority framework in which each evidence domain can be removed and audited separately. A step-by-step evidence-chain check limits same-study support to the transcript and pathway levels and leaves transcription-factor and chromatin steps as testable hypotheses. No independent biological replication was achieved; the evidence boundary is stated once below. Evidence boundary statement: This is a hypothesis-generating reanalysis of public transcriptomic data. Copper binding, MYH9 polymerization/acetylation, DLAT lipoylation/release, F-actin/G-actin fractionation, paired AML12 chromatin measurements, genetic perturbation and rescue were not performed. The study therefore supports copper-stress-associated cell-state changes, analytical concordance and candidate-priority relationships, but not cuproptosis execution, direct protein interaction or causal mechanism. Later sections refer to this statement rather than repeating the full boundary. Terminology (MYH9-related names): Four objects are kept strictly separate. The MYH9 external-template score is a fixed prior from external literature used to compute similarity; it is not a candidate gene in this study. The MYH9-as-a-candidate sensitivity term is a separate diagnostic computed only to audit template dependence, and is reported outside the candidate ranking. The template-derived MYH9 term is the ranking object produced by the MYH9 prior in the full-template score; its position reflects how the score was built, not a biological rank. The candidate ranking excluding MYH9 is the ranking table for real candidate genes; MYH9 does not appear in that table or in its figures. The MYH9-as-a-candidate sensitivity term moves from rank 2 to rank 48 when template-linked evidence is removed; this describes model behavior and must never be read as a biological candidate rank for MYH9 in AML12.

Abstract as supplied by the indexing source. © the authors / publisher.