Multi-Omics: TRIM28 Links Kidney Injury to Urine Signals
- Aug 23
- 4 min read
Acute kidney injury has no approved disease-modifying drug, and the tissue where the injury program is visible, a kidney biopsy, is almost never available in routine care. A group writing in Inflammation attacked that gap with a multi-omics design running from human single-cell atlases down to a protein measurable in voided urine. Single-cell, bulk, and spatial transcriptomics were stacked against kidney proteomics from a mouse ischemia-reperfusion model and urine proteomics from 20 AKI patients. Only candidates surviving every layer advanced. TRIM28 came out on top.
Key Takeaways
Multi-omics integration across single-cell, spatial, and proteomic layers nominated TRIM28 as a regulator of tubular injury that is also detectable in patient urine.
TRIM28 sits upstream of IL-17 and ACT1 signaling in hypoxia-stressed human tubular cells, making it a mechanistic node and a candidate drug target at once.
The clinical arm was 20 patients against 20 controls, so TRIM28 is a candidate urinary marker, not a validated one.
Stacking Multi-Omics Layers from Kidney Tissue to Urine
The single-cell backbone came from 6 control and 8 AKI kidneys, 112,912 cells after quality control, resolving into 12 clusters and eleven annotated renal populations. Bulk RNA-seq on a separate cohort added 6,045 differentially expressed genes, and LC-MS/MS ran on sham versus ischemia-reperfusion mouse cortex plus urine from 20 patients and 20 healthy controls. Intersecting transcript and protein lists, then ranking survivors by random forest, is the step in this multi-omics integration that does the real filtering. Proximal tubule cells shrank as a fraction of the AKI atlas, matching earlier mouse injury profiling from Kirita and colleagues (2020).
Key Findings
Five genes survived every layer: TRIM28, HNRNPH1, ARHGEF10L, C1RL, and UCHL3 held up across transcript and protein evidence, with TRIM28 carrying the highest random forest feature importance.
TRIM28 tracks tightly with PKD2: correlation reached r = 0.965 (P = 6.3 x 10-9), while the link to CEBPA ran the other way at r = -0.787.
The protein reaches urine: TRIM28 was detected directly in the human urine proteome and differed in abundance between AKI patients and controls, giving the tissue finding a non-invasive readout.
Bidirectional control of IL-17 signaling: in HK-2 cells under hypoxia/reoxygenation, knockdown lowered IL-17 immunoreactivity, ACT1, and phospho-p65, while overexpression pushed all three up.
Figure 1. Single-cell atlas of the injured kidney. Panel A shows the UMAP embedding of 12 transcriptional clusters recovered after quality control and batch correction. Panel B assigns those clusters to renal populations, including proximal and distal tubule, loop of Henle, collecting duct, endothelium, mesenchyme, leukocytes, podocytes, and progenitor-like cells. Panel C is a bubble plot of canonical markers, dot size for percentage of expressing cells and color for mean expression. Panel D compares cell type proportions between control and AKI kidneys, and Panel E summarizes cluster-specific GO and KEGG enrichment. Adapted from Wang et al. (2026), Inflammation.
Narrowing Five Candidates to One
Overlap alone rarely settles which gene to chase. Feature importance put TRIM28 clearly ahead of the other four, and deconvolution showed M0 macrophages and neutrophils expanding while memory B cells, resting NK cells, and CD8+ T cells contracted. Spatial maps placed TRIM28 unevenly across the tissue, with hotspots in proximal tubule and progenitor-like cells rather than a diffuse rise.
Mechanism and a Druggable Handle
Perturbation in HK-2 cells supplied the causal step correlations cannot. Knocking TRIM28 down softened the IL-17 and ACT1 response to hypoxia/reoxygenation and cut phospho-p65 without moving total NF-kB; overexpression did the reverse. Structure-based screening produced HY-N10592 at a docking score of -14.557, with CETSA supporting engagement inside cells. That compound work is early, and the score says more about the screen than about anything a medicinal chemistry team could act on today.
Where Cross-Layer Integration Gets Expensive
The hard part of a design like this isn't the biology, it's keeping transcript and protein measurements comparable across two species, four datasets, and separate instrument runs. Where the layers come off one injection, as they do in the single-injection multi-omics analysis we run on Omni-MS at Dalton, batch correction across plates stops competing with sample volume for the same aliquot. That alignment is what makes a cross-omics overlap in multi-omics biomarker discovery worth acting on instead of worth repeating.
Frequently Asked Questions
What is multi-omics analysis used for?
Multi-omics analysis measures two or more molecular layers, such as transcripts, proteins, metabolites, or lipids, in one biological system and reads them together. Teams reach for it when a single layer leaves the mechanism ambiguous. In biomarker programs the goal is usually to trace a tissue-level mechanism out to something measurable in blood or urine.
Can a multi-omics study find biomarkers you can measure in urine?
Yes, and this acute kidney injury study is a concrete example. TRIM28 was nominated from kidney tissue transcriptomics and proteomics, then detected directly in the urine proteome of AKI patients. The 20-versus-20 cohort is too small to call it validated, but the tissue-to-urine bridge held.
How many samples does a multi-omics biomarker study need?
Discovery phases commonly run 20 to 50 subjects per group, enough to rank candidates but not to pin an effect size. Validation usually needs a few hundred, ideally from an independent site. Sample volume matters as much as subject count when several assays draw from one aliquot.
Conclusion
What's believable here is the nomination: TRIM28 earned its rank across independent layers and two species, and it appears where a clinician can actually sample it. What isn't proven is that measuring it beats existing AKI markers, since no diagnostic performance is reported for the urinary signal. The sensible next move for renal biomarker programs is a targeted assay in a powered validation cohort.
Related Reading
See how we run these analyses in one lab: Dalton's multi-omics CRO services.
Citation
Wang, K., Wang, H., Zhang, Y., Zhang, Z., Xu, C., Zhao, H., Ma, J., Man, J., & Yang, L. (2026). Multi-omics integration of single-cell and spatial transcriptomics with tissue and urine proteomics identifies TRIM28 as a translationally relevant candidate regulator in acute kidney injury. Inflammation, 49(1), 176. https://doi.org/10.1007/s10753-026-02521-7
Note
This blog post summarizes findings from the above-cited research. Figures are adapted from the original publication. For full details, please refer to the source article.
By Seungjun Yeo, CEO at Dalton Bioanalytics. Specializing in multi-omics mass spectrometry for drug discovery and biomarker research.