Identification and validation of CCL28 and CCL5 as chemokine biomarkers and pro-inflammatory mediators in acute pancreatitis: an integrated bioinformatics and experimental study

Abstract

Acute pancreatitis (AP) is a common inflammatory disease of the exocrine pancreas. Its course ranges from a mild, self-limited illness to severe disease with organ failure. Current clinical tools do not reliably identify patients at risk of progression early in the disease course, and few molecular markers reflect the inflammatory signals that amplify pancreatic injury. Chemokines coordinate immune-cell recruitment during inflammation, but the regulation and functional roles of CCL28 and CCL5 in AP remain uncertain. We analysed a public pancreatic transcriptomic dataset containing 87 AP and 32 control samples. Differentially expressed genes were intersected with a curated ImmPort chemokine list, and candidate genes were prioritised with least absolute shrinkage and selection operator regression, support-vector-machine recursive feature elimination and random forest analysis. Diagnostic performance was estimated by receiver-operating-characteristic analysis with 95% confidence intervals. Immune-cell composition was inferred from human transcriptomic profiles with CIBERSORT using the LM22 signature, and pathway activity was assessed by single-sample gene-set enrichment analysis. We then tested the candidates in 266-6 pancreatic acinar cells, an adeno-associated-virus-mediated CCL28-knockdown mouse model of L-arginine-induced AP, and a clinical cohort comprising healthy controls and patients with AP or severe AP. NF-κB activation, tumour necrosis factor-α (TNF-α), interleukin-1β (IL-1β), pancreatic histology and serum amylase and lipase were assessed as functional readouts. The transcriptomic screen yielded four chemokine candidates: CCL28, CCL4, CCL5 and XCL2. All three feature-selection methods converged on CCL28 and CCL5, which separated AP from control samples with internal AUCs of 0.866 (95% CI 0.797–0.926) and 0.811 (95% CI 0.719–0.890), respectively. In serum, CCL28 and CCL5 were higher in AP than in controls and rose further in severe AP. L-arginine stress increased CCL5 expression and activated NF-κB signalling in 266-6 cells, whereas CCL28 knockdown reduced CCL5 induction and phosphorylation of IκBα and p65. In mice, CCL28 silencing lowered pancreatic CCL28 and CCL5, reduced serum amylase and lipase, decreased TNF-α and IL-1β, and preserved acinar architecture. Transcriptomic and tissue mRNA signals did not always match the protein and circulating measurements, consistent with a boundary of this study and a need to examine post-transcriptional regulation. CCL28 and CCL5 behave as up-regulated, pro-inflammatory chemokines in AP at the validated protein and serum levels. The data place CCL28 upstream of CCL5 and NF-κB signalling and support CCL28 and CCL5 as candidate severity biomarkers. CCL28 also emerges as a plausible therapeutic target, although these conclusions require confirmation in larger independent cohorts and receptor-level mechanistic studies.