Gene expression profiling of bovine raw milk as a new tool to monitor the inflammatory status of the udder : a pilot study

This pilot study demonstrates that gene expression profiling of raw milk using RT-qPCR can effectively differentiate milk samples based on specific immune cell types (neutrophils, macrophages, and T lymphocytes), offering a promising alternative to traditional somatic cell counting for monitoring the inflammatory status of the bovine udder.

The Gist

The Big Picture: Listening to the Udder's "Whispers"

Imagine a dairy cow's udder as a busy city. When the city is healthy, the streets are quiet, and the "police force" (immune cells) is small and relaxed. But when a bacterial invader (mastitis) attacks, the city goes into lockdown. The police swarm in, traffic jams occur, and the whole place gets chaotic.

For decades, farmers have tried to measure this chaos by simply counting the number of police cars (cells) in the milk. This is called the Somatic Cell Count (SCC).

• Low count? The city is probably fine.
• High count? The city is under attack.
• The Problem: Sometimes the count is in the "middle." Is the city just having a busy day, or is it in the middle of a riot? The old method can't tell the difference. It's like trying to guess if a party is a celebration or a brawl just by counting how many people are in the room.

The New Idea: Reading the "Signs" Instead of Counting Heads

This study proposes a smarter way to check the udder's health. Instead of just counting the cells, the researchers decided to listen to what the cells are saying.

Every type of immune cell (neutrophils, macrophages, and T-lymphocytes) has a unique "voice" or gene expression profile.

• Neutrophils are the "first responders" (like paramedics rushing to a fire).
• Macrophages are the "cleanup crew" (like sanitation workers).
• T-lymphocytes are the "specialized detectives" (like FBI agents).

The researchers asked: If we take a sample of raw milk and read the genetic "diary" of the cells inside, can we tell exactly what kind of emergency is happening, even if the total number of cells isn't huge?

How They Did It: The "DNA Detective" Experiment

1. The Training Phase: First, they went to the lab and sorted out pure groups of these three cell types (like separating red, blue, and green marbles). They read the genes of each group to learn their specific "voices." They found 36 specific genes that act like unique fingerprints for each cell type.
2. The Test Phase: They took milk samples from 38 different cow quarters. Some were healthy, some were slightly irritated, and some were very inflamed.
3. The Analysis: They didn't just count the cells. They extracted the RNA (the genetic messages) from the raw milk and checked which of those 36 "fingerprint" genes were active.

The Results: Four Distinct "Moods"

By looking at the gene patterns, the computer grouped the milk samples into four distinct clusters, revealing the true story behind the numbers:

• Cluster 1 (The Peaceful City): Low cell count, mostly "cleanup crew" (macrophages) and "detectives." This looks like a healthy, calm udder.
• Cluster 2 (The Alarm is Ringing): Low cell count, but the "paramedics" (neutrophils) are starting to wake up. The city is quiet, but the police are mobilizing. This might be an early infection that the old cell-count method would miss.
• Cluster 3 (The Cleanup Phase): High cell count, but the "paramedics" are leaving and the "cleanup crew" is taking over. This suggests the infection is being resolved or is a chronic, low-level issue.
• Cluster 4 (The Full Riot): High cell count, and the "paramedics" are everywhere. This is a full-blown active infection.

Why This Matters

The researchers compared their "gene listening" method to the traditional "cell counting" method (flow cytometry). They found that the gene method matched the cell counts perfectly but gave much more detail.

• The Analogy: Imagine two rooms with 50 people in each.
  • Old Method: "There are 50 people. We don't know what's happening."
  • New Method: "There are 50 people, but 40 of them are wearing firefighter gear and holding hoses. This is a fire!" or "There are 50 people, but they are all wearing party hats. This is a celebration!"

The Takeaway

This study is a pilot (a small test run), but it shows great promise. It proves that we can use the genetic "voice" of milk to diagnose udder health much more accurately than just counting cells.

• Current Limit: It's currently a lab technique that requires expensive equipment, so you can't use it on a farm with a handheld device yet.
• Future Hope: In the future, this could lead to a test that tells farmers exactly what is wrong with a cow's udder before the cow even shows symptoms. This could help farmers use antibiotics only when truly necessary, saving money and keeping cows healthier.

In short: Instead of just counting the crowd, this new tool listens to the crowd's conversation to figure out if they are celebrating, fighting, or cleaning up a mess.

Technical Summary

1. Problem Statement

Mastitis (inflammation of the mammary gland) is a major economic and welfare issue in dairy farming. Current monitoring relies heavily on Somatic Cell Count (SCC) as an indirect indicator of infection. However, SCC has significant limitations:

• Ambiguity in the "Gray Zone": SCC values between 100,000 and 200,000 cells/mL do not definitively distinguish between healthy and infected quarters.
• Lack of Specificity: High SCC indicates inflammation but does not identify the specific cell types involved (neutrophils, macrophages, lymphocytes) or the stage of inflammation (acute vs. resolving).
• Limitations of Differential Counts: While Differential Somatic Cell Count (DSCC) and flow cytometry offer better resolution, they often require complex sample processing (centrifugation) which can lead to cell loss (particularly macrophages in the cream layer) and rely on user-defined thresholds that vary by herd and lactation stage.

The authors hypothesized that gene expression profiling of raw milk could serve as a functional surrogate to characterize the inflammatory status of udder quarters more precisely, independent of cell sorting or complex thresholds.

2. Methodology

The study employed a multi-step approach combining cell sorting, high-throughput RT-qPCR, and flow cytometry on raw milk samples.

• Sample Collection:

  • Cohort: 38 quarter milk samples from 28 Holstein and 10 Normande cows.
  • Selection Criteria: Samples were selected to cover a wide SCC range (17,000 to 3,603,000 cells/mL) and various lactation stages (35–391 days in milk). No clinical mastitis cases were included.
  • Processing: Samples were collected aseptically as foremilk.

• Gene Panel Selection (Pilot Phase):

  • Cell Sorting: Macrophages (CD45+ G1- CD14+), Neutrophils (CD45+ G1+), and T-lymphocytes (CD45+ CD3+) were purified from three milk samples using flow cytometry.
  • Screening: RNA was extracted from these purified populations to screen a panel of 95 candidate genes.
  • Selection: Based on hierarchical clustering and PCA, 36 genes were selected as specific markers for the three cell types:
    • Neutrophils: 15 genes (Cluster "PMN2").
    • Macrophages: 15 genes (Cluster "Macro").
    • T-lymphocytes: 11 genes (Cluster "Lc").

• Gene Expression Profiling:

  • RNA Extraction: Directly from raw milk samples (200 µL) using Trizol, avoiding centrifugation steps that might lose cells.
  • Quantification: High-throughput RT-qPCR was performed using the Fluidigm BioMark HD System (96.96 GE Dynamic Array).
  • Analysis: Expression data were normalized against reference genes (ACTB, GAPDH) and a theoretical control. Hierarchical clustering and Principal Component Analysis (PCA) were used to group samples.

• Validation:

  • Flow Cytometry: The same milk samples were analyzed via flow cytometry to determine the actual proportions of neutrophils, macrophages, and T-lymphocytes.
  • Bacteriology: Standard plating was performed to identify bacterial presence (Gram-positive vs. Gram-negative/contaminants).
  • Statistical Comparison: The dendrograms derived from gene expression were compared to those from flow cytometry using Baker's index to assess correlation.

3. Key Contributions

• Novel Diagnostic Approach: The study proposes a shift from counting cells (SCC/DSCC) to assessing the functional state of milk cells via gene expression.
• Direct Raw Milk Analysis: The protocol extracts RNA directly from raw milk without preliminary centrifugation, potentially capturing cell populations (like macrophages) that are often lost in the cream layer during standard processing.
• Unsupervised Clustering: The method identifies distinct inflammatory states without relying on pre-defined SCC thresholds or user-defined cut-offs.

4. Key Results

• Gene Panel Validation: The initial screening successfully identified 36 genes with distinct expression patterns specific to neutrophils, macrophages, or T-lymphocytes.
• Sample Clustering: Hierarchical clustering of the 36 genes in raw milk samples revealed four distinct clusters:
  1. Cluster 1: Low SCC, high T-lymphocyte gene expression, high macrophage gene expression. Interpreted as healthy/low inflammation.
  2. Cluster 2: Low SCC, high T-lymphocyte gene expression, but elevated neutrophil gene expression compared to Cluster 1. Interpreted as early inflammation/neutrophil recruitment.
  3. Cluster 3: High SCC, high macrophage gene expression, lower neutrophil expression. Interpreted as resolving inflammation or chronic infection.
  4. Cluster 4: High SCC, high neutrophil gene expression. Interpreted as acute inflammation.
• Correlation with Flow Cytometry:
  • The clustering based on gene expression strongly correlated with the cell proportions determined by flow cytometry (Baker's index p < 0.01).
  • Samples in Clusters 3 and 4 showed significantly higher neutrophil proportions, while Clusters 1 and 2 showed higher lymphocyte/macrophage proportions.
• Bacteriological Correlation: Cluster 4 (acute inflammation) showed a tendency to contain Gram-positive bacteria, consistent with high SCC and active infection.
• Differentiation of Low SCC: Crucially, the method differentiated between Cluster 1 and Cluster 2 samples, both of which had low SCC (<100,000 cells/mL) but different underlying inflammatory statuses (healthy vs. early recruitment).

5. Significance and Limitations

Significance:

• Improved Diagnostic Precision: This method can detect inflammatory changes in the "gray zone" of SCC where traditional methods fail, potentially allowing for earlier intervention.
• Functional Insight: It provides information on the type of immune response (e.g., neutrophil-driven vs. macrophage-driven) rather than just the quantity of cells.
• Future Applications: While currently a laboratory tool, this approach paves the way for medium-to-high throughput phenotyping strategies for selective dry cow therapy and routine herd health monitoring.

Limitations:

• Pilot Scale: The study involved a small number of samples (n=38), limiting statistical power for definitive pathogen-specific associations.
• Accessibility: The method requires specialized laboratory equipment (Fluidigm system, RT-qPCR) and is not yet feasible for on-farm diagnostics.
• Epithelial Cell Influence: The study could not fully quantify epithelial cells via flow cytometry. Since RNA in raw milk may also originate from epithelial cells or fat globules, the profiles reflect a mix of immune and epithelial functionality.
• Variable Factors: Further studies are needed to account for the effects of parity, lactation stage, and milking time (foremilk vs. hindmilk) on gene expression profiles.

Conclusion:
The study successfully demonstrates that gene expression profiling of raw milk is a viable, high-resolution tool for monitoring udder health. It can distinguish between different inflammatory states (healthy, early recruitment, resolving, acute) that are indistinguishable by SCC alone, offering a promising avenue for future diagnostic developments in dairy science.