Abstract: We investigate the relationship of protein-protein interactions with mRNA expression levels, by integrating a variety of data sources for yeast. We focus on known protein complexes that have clearly defined interactions between their subunits. We find that subunits of the same protein complex show significant coexpression, both in terms of similarities of absolute mRNA levels and expression profiles, e.g., we can often see subunits of a complex having correlated patterns of expression over a time course. We classify the yeast protein complexes as either permanent or transient, with permanent ones being maintained through most cellular conditions. We find that, generally, permanent complexes, such as the ribosome and proteasome, have a particularly strong relationship with expression, while transient ones do not. However, we note that several transient complexes, such as the RNA polymerase II holoenzyme and the replication complex, can be subdivided into smaller permanent ones, which do have a strong relationship to gene expression. We also investigated the interactions in aggregated, genome-wide data sets, such as the comprehensive yeast two-hybrid experiments, and found them to have only a weak relationship with gene expression, similar to that of transient complexes.
Introduction
This 2002 paper from the Gerstein lab at Yale is a foundational study in systems biology. It explores a deceptively simple question: If two proteins work together in a cell, are their corresponding genes expressed at the same time and in similar amounts? By integrating disparate datasets for yeast (S. cerevisiae), the authors reveal that the answer depends entirely on whether the protein interaction is a "permanent" marriage or a "transient" hookup.
What Problem Does This Paper Address?
In the early 2000s, the "genomics revolution" was producing massive amounts of data: mRNA expression levels (transcriptomics) and protein-protein interaction (PPI) maps (proteomics). A common assumption was the "guilt by association" principle—if two genes are co-expressed, their proteins likely interact.
However, researchers were finding that these two types of data didn't always align. Jansen et al. sought to bridge this gap. They wanted to determine the extent to which mRNA levels could actually predict physical protein interactions and why certain interactions seemed "invisible" to expression analysis.
Methods and Metrics
The authors employed two primary statistical metrics to compare mRNA expression and protein interactions:
- Normalized Difference ($D_{ij}$): To compare absolute expression levels (how many copies of mRNA exist), they calculated the difference between two genes as a fraction of their sum. A value near 0 indicates nearly identical abundance.
- Pearson Correlation Coefficient: To compare relative expression profiles (how levels change over time or across conditions, such as the cell cycle), they used standard correlation.
The "novel" step was their classification of protein complexes into two categories:
- Permanent: Stable complexes maintained through most cellular conditions (e.g., the ribosome, proteasome).
- Transient: Interactions that form and break apart depending on specific signals or cell cycle stages (e.g., signaling molecules, the replication complex).
Key Insights and Ideas
- The Energetic Logic of Co-expression: Permanent complexes like the ribosome show extremely high co-expression. The authors argue this is an evolutionary necessity: building half a ribosome is energetically costly and functionally useless, so the cell regulates the "parts list" with high precision.
- Transient Complexes are Modular: While large transient complexes (like the RNA polymerase II holoenzyme) show weak overall co-expression, they can often be subdivided into smaller, permanent "modules." These modules are highly co-expressed, suggesting that the cell builds stable sub-units that only come together when needed.
- The "Noise" in High-Throughput Data: The authors found that aggregated genome-wide datasets (like Y2H) show very weak correlation with expression. This is likely because these methods capture many transient interactions or include significant experimental noise, making them look more like random protein pairs than stable complexes.
Key Results
The analysis revealed a stark contrast between permanent complexes and other types of interactions.
- Permanent vs. Random: Subunits of permanent complexes had a median normalized difference of ~0.2, significantly lower than the ~0.5 observed for random protein pairs (see Figure 1).
- Statistical Significance: The synchronized expression of permanent complexes is highly unlikely to occur by chance, with distributions significantly shifted from random pairs.
- Intersection Gap: Only about 6% of the interactions found in Y2H experiments overlapped with known stable complexes, explaining why Y2H data correlates so poorly with expression profiles.
Limitations and Future Work
The authors are careful to note that mRNA levels are not a perfect proxy for protein levels. Post-translational modifications, protein degradation rates, and translation efficiency mean that even perfectly synchronized mRNA might result in different protein abundances.
Furthermore, the binary classification of "permanent vs. transient" is a simplification. In reality, there is a spectrum of stability. The authors suggest that future work should focus on more sophisticated kinetic models that account for the rate constants of translation and degradation to better predict protein behavior from transcriptomic data.
Summary updated