Exploration of the structural and functional diversity in the metamorphic RfaH subfamily

This study utilizes AlphaFold2 predictions and experimental validation to identify a distinct clade of constitutively active, monomorphic RfaH homologs associated with virulence operons, thereby supporting a stepwise evolutionary model where RfaH specialized from an active NusG-like ancestor into an autoinhibited fold-switching regulator.

The Gist

The Story of the "Shape-Shifting" Protein

Imagine a protein called RfaH as a tiny, biological security guard working inside a bacterial cell. Its job is to decide which genes (the cell's instruction manuals) get read and which ones stay locked away.

Most of the time, this guard is lazy and asleep. It's wearing a heavy coat (an alpha-helical hairpin) that covers its face and keeps it from doing any work. This is the "auto-inhibited" state. It only wakes up when it sees a specific signal (a specific DNA sequence called ops) on a long, dangerous instruction manual (a virulence gene). When it sees the signal, it rips off its coat, transforms into a completely different shape (a beta-barrel), and becomes an active, energetic worker that helps the cell build weapons to infect other organisms.

This ability to change its entire 3D shape to do a different job is called being a metamorphic protein. It's like a Transformer toy that is a truck when parked but becomes a robot when it needs to fight.

The Big Discovery: Not All Guards Change Shape

For a long time, scientists thought all RfaH proteins worked this way: they were all shape-shifters that needed a signal to wake up.

However, this paper suggests that nature is more creative than we thought. The researchers used a super-smart AI (called AlphaFold2) to look at thousands of RfaH proteins from different bacteria. They found a hidden group of RfaH proteins that never change shape.

Think of it this way:

• The Shape-Shifters (Metamorphic): These are the guards who sleep in a coat and only wake up when they see the "ops" signal. They are careful and selective.
• The "Always-On" Guards (Monomorphic): The AI predicted that about 14% of these proteins are born wearing their "work uniform" (the beta-barrel shape). They don't have a coat to take off. They are constantly active, ready to work immediately without needing a signal.

How They Proved It

The researchers didn't just trust the AI; they tested it in the lab.

1. The Simulation (The AI): They fed the DNA sequences of these proteins into the AI. For the "Always-On" group, the AI predicted they would stay in the active shape, even without a signal.
2. The Lab Test (The Reality Check): They put these "Always-On" proteins into bacteria that were missing their own RfaH.
   • Result: The "Always-On" proteins worked perfectly! They activated the genes even when the "ops" signal was missing. They were like a guard who never sleeps and starts working the moment they are hired.
   • Contrast: The "Shape-Shifter" proteins only worked when the signal was present. If the signal was missing, they stayed asleep.

Why Does This Matter? (The Evolutionary Mystery)

The paper solves a mystery about how these proteins evolved.

Scientists used to think RfaH evolved from a simpler protein called NusG (a standard, non-shifting worker). The theory was:

1. First, the protein was a simple, always-on worker (like NusG).
2. Then, it evolved a "coat" (the alpha-helix) to stop itself from working too much, so it could be selective.
3. Finally, it learned to take the coat off only when needed.

This paper confirms that step 1 and step 2 actually exist in nature today.

The "Always-On" RfaH proteins the team found are living fossils. They are the modern-day versions of that ancient, simple worker. They haven't evolved the "coat" yet, so they are always active.

The "Location" Clue

The researchers also found a clue about where these proteins live in the bacterial genome:

• The Shape-Shifters: They usually live far away from the genes they control. They need to travel (in trans) to find their target. Because they are dangerous if they are always on, they need a "coat" to keep them safe until they find the right signal.
• The "Always-On" Guards: They live right next door to the genes they control (in cis). Because they are so close to their target, they don't need a coat to stop them from working. They can just turn on the genes immediately. It's like a factory manager who lives inside the factory; they don't need a key to get in, they just start working.

The "Confused" Middle Ground

The study also found a third group: proteins that look like they are halfway between the two shapes. They are a mix of alpha-helices and beta-barrels.

• Analogy: Imagine a Transformer toy that is stuck halfway between a truck and a robot. It's wobbly and unstable.
• Result: These proteins didn't work very well in the lab. They were likely evolutionary "experiments" that didn't quite succeed in becoming either a stable shape-shifter or a stable "always-on" guard.

The Takeaway

This paper is like finding a new branch on the family tree of life. It shows that proteins don't have to be complicated shape-shifters to be useful. Some bacteria have kept the "simple, always-on" version of RfaH because it works perfectly for their specific needs.

It proves that evolution is a step-by-step process. We can see the "ancestors" (the always-on proteins) and the "modern descendants" (the shape-shifters) living side-by-side in nature today, helping us understand how complex biological machines evolve over millions of years.

Technical Summary

1. Problem Statement

Metamorphic proteins are unique biomolecules capable of switching between two distinct, stable native folds encoded by a single amino acid sequence. The RfaH protein, a virulence factor in Enterobacteriaceae, is a quintessential metamorphic protein. It exists in an autoinhibited state (an $\alpha$-helical hairpin, $\alpha$CTD) that binds to its N-terminal domain (NTD), and an active state (a $\beta$-barrel, $\beta$CTD) that resembles the constitutively active paralog NusG.

While the mechanism of RfaH fold-switching is well-characterized in E. coli, the evolutionary history of this family remains debated. A prevailing hypothesis suggests that RfaH evolved from NusG via gene duplication, initially existing as a monomorphic, constitutively active protein (acting in cis on nearby genes) before acquiring autoinhibition to regulate distant genes (in trans). However, no extant, functional monomorphic RfaH homologs had been experimentally identified, leaving a gap in understanding the evolutionary transition from NusG-like ancestors to metamorphic RfaH.

2. Methodology

The authors employed a multi-disciplinary approach combining deep learning, molecular dynamics, phylogenetics, and in vivo experimentation:

• Large-Scale Structure Prediction (AlphaFold2/ColabFold):

  • Retrieved ~3,942 RfaH homolog sequences from three databases: ColabFoldDB, InterPro, and a curated "Genomic Cluster" (sequences with known genomic context).
  • Predicted structures using ColabFold (an implementation of AlphaFold2).
  • Classified the C-terminal domains (CTD) of predicted structures into three categories based on secondary structure content:
    1. $\alpha$CTD: Metamorphic (autoinhibited).
    2. $\beta$CTD: Monomorphic (constitutively active, NusG-like).
    3. Mixed $\alpha/\beta$CTD: Intermediate or unstable states.
  • Validated classifications using interdomain salt-bridge distances (E48-R138 equivalent) and pLDDT confidence scores.

• Computational Simulation (TMD & AF2Rank):

  • Performed Targeted Molecular Dynamics (TMD) simulations to force the E. coli RfaH fold-switch between $\alpha$ and $\beta$ states.
  • Used AF2Rank (a framework repurposing AlphaFold2 as a scoring function) to evaluate the quality of TMD-generated decoy structures. This determined if "mixed" states observed in homolog predictions were plausible intermediates or artifacts.

• Phylogenetic and Genomic Analysis:

  • Constructed a maximum likelihood phylogenetic tree to correlate structural predictions with evolutionary clades.
  • Analyzed genomic neighborhoods to determine if RfaH homologs were located near long virulence operons (suggesting in cis regulation) or distant from them (in trans).

• Experimental Validation (In Vivo Assays):

  • Selected 9 representative homologs (3 metamorphic, 3 monomorphic, 3 mixed) from the Genomic Cluster.
  • Conducted luciferase reporter assays in an E. coli $\Delta$rfaH strain.
  • Tested activity under varying conditions: presence/absence of the ops (operon polarity suppressor) signal and the Ribosome Binding Site (RBS).
  • Compared activity against wild-type E. coli RfaH and the constitutively active E48A mutant.

3. Key Contributions

• Discovery of Extant Monomorphic RfaH: The study provides the first evidence of naturally occurring, functional RfaH homologs that are monomorphic (locked in the active $\beta$-barrel state) rather than metamorphic.
• Validation of AlphaFold2 for Fold-Switching: Demonstrates that while AlphaFold2 typically predicts the dominant state, specific sequence contexts (particularly in homologs) can lead to predictions of the alternative state, revealing evolutionary intermediates.
• Characterization of "Mixed" States: Identified a subset of homologs with mixed $\alpha/\beta$ CTD structures, hypothesizing them as evolutionary or structural intermediates that are less stable and less functional.
• Functional Decoupling: Showed that monomorphic RfaH homologs function independently of the ops signal, contrasting with metamorphic RfaH which requires ops for activation.

4. Key Results

• Structural Diversity:

  • Out of 20,000 predicted structures, **66%** were predicted as metamorphic ($\alpha$CTD), ~14% as monomorphic ($\beta$CTD), and ~9% as mixed $\alpha/\beta$.
  • The monomorphic predictions were concentrated in a distinct phylogenetic clade.
  • Salt Bridge Analysis: Metamorphic predictions showed short E48-R138 distances (stabilizing the autoinhibited state), while monomorphic predictions showed large distances (>13 Å), consistent with an open, active conformation.

• Functional Assays:

  • Metamorphic Homologs: Required the ops signal and showed activity only when the RBS was absent (mimicking E. coli RfaH behavior).
  • Monomorphic Homologs: Exhibited constitutive activity comparable to the E48A mutant. They activated transcription-translation coupling even in the absence of the ops signal and functioned regardless of RBS presence.
  • Mixed Homologs: Showed significantly reduced activity, particularly when the RBS was absent, suggesting structural instability prevents efficient ribosome recruitment.

• Genomic Context:

  • Metamorphic RfaH: Typically located far from their target operons (acting in trans).
  • Monomorphic RfaH: Predominantly located immediately adjacent to or within long virulence-associated operons (e.g., LPS biosynthesis), supporting the hypothesis of in cis regulation.

• Simulation Insights:

  • TMD simulations combined with AF2Rank confirmed that "mixed" $\alpha/\beta$ structures represent high-energy intermediates or unstable states, consistent with the low pLDDT scores and poor in vivo performance of these homologs.

5. Significance

• Evolutionary Model Confirmation: The findings strongly support a stepwise evolutionary model for RfaH:
  1. Ancestral State: A monomorphic, NusG-like protein acting in cis on nearby genes (constitutively active).
  2. Transition: Evolution of autoinhibition ($\alpha$CTD) to allow regulation of distant genes (in trans) and prevent interference with essential NusG functions.
  3. Extant Diversity: The existence of monomorphic RfaH homologs in modern bacteria suggests that some lineages have retained the ancestral state, while others evolved the metamorphic switch.
• Methodological Impact: This work highlights the utility of AlphaFold2 not just for static structure prediction, but for discovering functional diversity and evolutionary intermediates within protein families, provided the analysis accounts for sequence-specific variations in MSA inputs.
• Biological Insight: It clarifies how bacteria regulate virulence genes, showing that while E. coli uses a complex switch mechanism, other pathogens may rely on simpler, constitutive mechanisms for specific operons, offering new targets for antimicrobial strategies.