Re-analysis of Transcriptomic and Proteomic Data Using Multi-Omics Approaches Identifies Biomarkers of Diabetes-Associated Complications in an INS Mutant Pig Model

By re-analyzing transcriptomic and proteomic data from the MIDY pig model using advanced multi-omics approaches, this study identifies ADAMTS17 as a novel biomarker linking diabetes-associated immune dysfunction and delayed wound healing through ECM-immune interplay.

The Gist

Imagine you have a very sick pig. This isn't just any pig; it's a special "super-model" pig created in a lab to mimic a severe form of diabetes called MIDY (Mutant Insulin-Induced Diabetes of Youth).

In this pig, the "blueprint" for making insulin is broken. Instead of building a perfect insulin key, the factory keeps making misshapen, useless keys. These broken keys pile up, clog the factory, and eventually destroy the workers (the beta cells) who are supposed to make insulin. The result? The pig has high blood sugar and needs insulin shots, just like a human with poorly controlled diabetes.

Scientists have studied this pig before, looking at its genetic code (transcriptomics) and its protein building blocks (proteomics). But they might have missed some subtle clues.

The "Re-Do" Project
The authors of this paper decided to take a second look at the old data. Think of it like a detective re-examining a crime scene with a new, high-tech magnifying glass. They used a special digital toolkit called SurfacOmics (their own invention) and combined two different types of data to see the whole picture, not just the pieces.

What They Found: The "Construction Crew" and the "Security Guard"
In a healthy body, there's a constant balance between construction (building tissues) and security (immune system). In this diabetic pig, the balance was off.

1. The Broken Fence (COL6A3): They found that a key structural protein, COL6A3 (think of this as the concrete in a fence), was missing. The "fence" holding the tissues together was getting weak.
2. The Overactive Demolition Crew (ADAMTS17): They found a protein called ADAMTS17 was way too high. If COL6A3 is the concrete, ADAMTS17 is the demolition crew with a sledgehammer.
   • The Analogy: Because the demolition crew (ADAMTS17) is working overtime, they are breaking down the fence (COL6A3) faster than it can be rebuilt. This explains why diabetic wounds heal slowly—the "construction site" is being constantly torn down.

The "Silent Alarm" (Immune System)
Here is the most interesting part. Usually, when you have an infection or injury, your immune system sounds the alarm (sending out proteins like CRP, CD40, etc.).

But in this diabetic pig, the alarm was silent. The immune markers were low. It's as if the pig's security system was asleep at the wheel.

• The study found a strong link: The more the demolition crew (ADAMTS17) was working, the quieter the security guard (immune system) became.
• This suggests that the broken-down tissue (ECM) might be confusing the immune system, making the body less able to fight off infections or heal wounds.

The "Detective's Report" (Biomarkers)
The researchers used their SurfacOmics tool to pick the best "suspects" to watch in the future.

• Top Suspect #1 (Genetic): ADAMTS17. It's the demolition crew leader.
• Top Suspect #2 (Protein): FGG (Fibrinogen). This is a clotting protein that helps stop bleeding and start healing.

The tool confirmed that these two are the best "flags" to wave if you want to detect these specific diabetes complications early.

Why Does This Matter?
Think of diabetes complications (like slow-healing wounds or infections) as a house that is falling apart because the construction crew is confused and the security guard is asleep.

This paper tells us:

1. We found the demolition crew (ADAMTS17) that is tearing things down too fast.
2. We found the sleeping security guard (low immune markers).
3. We have a new checklist (biomarkers) to spot these problems early.

The Bottom Line
By re-analyzing old data with new, smart computer tools, the team found a hidden connection: The same process that breaks down tissue in diabetes might also be turning off the body's immune defense. This gives scientists new targets for drugs that could stop the demolition crew and wake up the security guard, helping diabetic patients heal faster and stay safer.

Technical Summary

1. Problem Statement

• Clinical Challenge: Identifying reliable biomarkers to distinguish disease states and understand the pathophysiology of diabetes-associated complications remains a significant hurdle, particularly in complex conditions like Mutant Insulin-Induced Diabetes of Youth (MIDY).
• Model Limitations: While the porcine MIDY model (caused by the INSC94Y mutation) effectively mimics severe hyperglycemia and $\beta$-cell loss due to misfolded insulin, previous studies have not fully elucidated the downstream molecular alterations, specifically regarding extracellular matrix (ECM) remodeling and immune dysfunction.
• Data Gap: Existing transcriptomic and proteomic data from the MIDY model (specifically from Backman et al., 2019) had been analyzed primarily in isolation. There was a need for an integrative, multi-omics re-analysis to uncover novel biomarkers and mechanistic links between metabolic failure and immune/ECM dysregulation.

2. Methodology

The study employed a rigorous integrative multi-omics approach to re-analyze publicly available data from the MIDY porcine model.

• Data Sources:
  • Transcriptomics: Raw FASTQ files from GEO repository (Accession: GSE122029).
  • Proteomics: LC-MS/MS data from ProteomeXchange (Accession: PXD011536).
  • Cohort: 2-year-old female transgenic MIDY pigs ($n=4$) vs. wild-type (WT) littermates ($n=5$).
• Preprocessing:
  • Initial datasets contained 34,328 genes and 2,535 proteins.
  • After strict quality filtering and preprocessing, 19,664 genes and 489 proteins were retained for downstream analysis.
• Analytical Frameworks:
  • Univariate & Multivariate Analysis: Principal Component Analysis (PCA) for group separation; Differential Expression Analysis (DESeq2) for transcriptomics.
  • Multi-Omics Integration: Utilized the mixOmics R package (v6.23.4) with:
    • sPLS (Sparse Partial Least Squares): For unsupervised feature selection.
    • DIABLO & mbPLS: For supervised multiblock integration to identify discriminative features across omics layers.
  • Network Analysis: STRINGdb (5% FDR, confidence 0.15) to map co-expression modules (metabolic vs. immune).
  • Biomarker Prioritization: Used an in-house tool, SurfacOmics, to evaluate biomarker potential based on expression differences, subcellular localization, and assayability (e.g., antibody availability).

3. Key Contributions

• Novel Biomarker Discovery: Identification of ADAMTS17 (A Disintegrin And Metalloproteinase With Thrombospondin Motifs 17) as a top-tier transcriptomic biomarker and Fibrinogen Gamma (FGG) as a top proteomic biomarker specific to the MIDY state.
• Mechanistic Insight: Proposed a novel link between ECM remodeling and immune dysfunction in diabetes. The study suggests that elevated ADAMTS17 drives ECM degradation (specifically collagen), which correlates with a suppressed immune response.
• Methodological Validation: Demonstrated the efficacy of the SurfacOmics tool in prioritizing assayable targets by integrating multi-omics signals with subcellular localization data.
• Re-analysis of Existing Data: Showed that advanced integrative computational approaches can extract deeper biological insights from existing public datasets that were missed in original single-omics analyses.

4. Key Results

• Group Separation: PCA of the proteomic data clearly separated MIDY and WT groups. The primary drivers of variance were metabolic enzymes HMGCS2 and RDH16, corroborating previous findings.
• Differential Expression:
  • Proteomics: COL6A3 (collagen type VI alpha 3 chain) was significantly reduced in MIDY, while ADAMTS17 was significantly elevated.
  • Transcriptomics: 17 genes showed a log-fold change > 1. No proteins met the strict FDR < 0.01 threshold for differential abundance, but trends in COL6A3, ADAMTS17, and CRP were noted.
• Correlation Analysis:
  • A strong negative correlation ($r = -0.48$) was found between ADAMTS17 and COL6A3, suggesting ADAMTS17 drives collagen degradation.
  • Immune Suppression: Immune markers (CXCR2, CD40, CD59) and CRP were significantly lower in the MIDY group compared to WT.
  • Inverse Relationship: Strong negative correlations were observed between ADAMTS17 and immune markers (CD40, CD59, CXCR2), implying that high ADAMTS17 levels accompany reduced immune signaling.
• Network Modules: STRING analysis revealed two distinct modules: a metabolic module and an immune module. The immune module was downregulated in MIDY.
• Biomarker Validation (SurfacOmics):
  • ADAMTS17 and FGG were ranked as top candidates.
  • Despite FGG not being a surface protein, the tool flagged it as "assayable" due to the availability of commercial polyclonal antibodies for cytoplasmic detection via immunofluorescence.
  • MBL2 (Mannose-Binding Lectin 2) was identified as a co-correlated gene, reinforcing the link between ECM remodeling and immune response.

5. Significance

• Pathophysiological Understanding: The study shifts the perspective on MIDY from a purely metabolic defect to a condition involving ECM-immune interplay. It suggests that chronic insulin deficiency leads to a specific signature of ECM degradation (via ADAMTS17) coupled with immune incompetence (reduced CRP and immune markers).
• Therapeutic Targets: ADAMTS17 is proposed as a potential therapeutic target for mitigating diabetes-associated complications, particularly delayed wound healing and immune dysfunction, which are common in diabetic patients.
• Translational Potential: By identifying biomarkers with high "marker potential scores" and confirmed assayability (via SurfacOmics), the study provides a roadmap for developing diagnostic assays for diabetic complications.
• Computational Utility: The successful application of multi-omics integration (sPLS, DIABLO) and the in-house SurfacOmics tool highlights the value of computational re-analysis in extracting novel biological insights from established animal models, potentially accelerating biomarker discovery for Type 1 Diabetes (T1D) and related complications.

Note: The authors acknowledge that while the MIDY model mimics severe insulin deficiency, it lacks the autoimmune component inherent to human T1D. Therefore, further experimental validation is required to confirm these findings in broader diabetic contexts.