Unique Transcriptional Programs Identify Subtypes of AKI

Abstract

Two metrics, a rise in serum creatinine concentration and a decrease in urine output, are considered tantamount to the injury of the kidney tubule and the epithelial cells thereof (AKI). Yet neither criterion emphasizes the etiology or the pathogenetic heterogeneity of acute decreases in kidney excretory function. In fact, whether decreased excretory function due to contraction of the extracellular fluid volume (vAKI) or due to intrinsic kidney injury (iAKI) actually share pathogenesis and should be aggregated in the same diagnostic group remains an open question. To examine this possibility, we created mouse models of iAKI and vAKI that induced a similar increase in serum creatinine concentration. Using laser microdissection to isolate specific domains of the kidney, followed by RNA sequencing, we found that thousands of genes responded specifically to iAKI or to vAKI, but very few responded to both stimuli. In fact, the activated gene sets comprised different, functionally unrelated signal transduction pathways and were expressed in different regions of the kidney. Moreover, we identified distinctive gene expression patterns in human urine as potential biomarkers of either iAKI or vAKI, but not both. Hence, iAKI and vAKI are biologically unrelated, suggesting that molecular analysis should clarify our current definitions of acute changes in kidney excretory function.

Keywords: acute kidney injury; biomarkers; renal ischemia; transcriptional profiling; volume depletion.

Figure 1.

Limited overlap of gene expression between iAKI and vAKI. Differential gene expression was most prominent in the OSOM in the iAKI model and in the ISOM in the vAKI model. The heatmaps portray only the significant DEGs (q-value<0.01). Gene expression was z-score–transformed on a per-gene basis and then hierarchically clustered.

Figure 2.

Different patterning of iAKI and vAKI pathways. (A) Functional analyses using gene set enrichment analysis (GSEA) against KEGG, Reactome, Biocarta, and PID Pathway databases. Each row (thin lines) demonstrates a pathway found in one or more of the queried databases. Significant pathway enrichment is represented in a binary manner (enrichment=red; de-enrichment=blue; unchanged=white; false discovery rate <25%). (B) Functional analysis using GSEA was supplemented with signaling pathway impact analysis leveraging the topologic information available from canonical signaling pathways. Each horizontal division (boxes) contains an aggregate of gene sets ascribable to a known signaling pathway analyzed by one or more of the queried databases (i.e., KEGG, Reactome, Biocarta, PID). Each shaded row within the division represents an individual gene set analyzed by a single database; most signaling pathways were analyzed by multiple gene sets and databases. Pathway activation (red) or inhibition (green) was determined by signaling pathway impact analysis; the depth of shading of red or green reflects the degree of GSEA enrichment or de-enrichment.

Figure 3.

Detection of novel biomarkers in mouse. Immunofluorescence (CK20) and in situ hybridization (Tacstd2, Lcn2) demonstrate specificity for iAKI. Note that CK20 (red) was expressed in proximal tubules (megalin=green) and in intercalated cells (IC) (inset: CK20=red; AQP2=white). Note that Tacstd2 was expressed in distal nephron segments. Lcn2 RNA was expressed by thick ascending limbs of Henle (TALH) as well as ICs of the collecting ducts in iAKI. Timp2 and Igfbp7 RNA are shown for comparison. Note the expression of these genes in scattered glomerular (G) and tubular cells.

Figure 4.

Detection of novel biomarkers in human. (A) Secreted proteins (CHI3L1, TROP2, TPA, CK20) were elevated in most human iAKI urine but not in most vAKI urine. Conversely, secreted protein PAPPA2 was elevated in many vAKI urines, and some normal urines, but it is degraded in iAKI patients. Blue bars denote canonical molecular weights of the analyte. (B) Secreted proteins (NGAL, VDBP, TIMP2, IGFBP7) are shown for comparison and demonstrate different degrees of specificity of iAKI patients, as listed. Blue bars denote canonical molecular weights of the analyte. (C) Proteolysis of vAKI urinary PAPPA2 by iAKI urine was rescued by protease inhibitors, pepstatin A, or combinations of inhibitors. PAPPA2 was also degraded by cathepsin D.