Hydrogen-bonding changes cause differences in imipenem breakdown activity in OXA-48 variants

This study utilizes multiscale simulations to reveal that hydrogen-bonding changes within the active site, driven by mutations in the $\beta$5-$\beta$6 loop, alter the dynamics and efficiency of imipenem deacylation and binding, thereby explaining the distinct hydrolytic activities observed among OXA-48 variants.

The Gist

The Big Picture: A Biological Lockpick and Its Broken Keys

Imagine antibiotics (like imipenem) are special keys designed to jam the locks of bacteria, stopping them from building their cell walls and killing them.

OXA-48 is a bacterial enzyme that acts like a master lockpick. Its job is to find these antibiotic keys and break them so they can't do their job. This is how bacteria become resistant to medicine.

Scientists have found several "versions" of this lockpick (called variants: OXA-163, OXA-405, and OXA-517). Some of these new versions are terrible at breaking the keys, while others are just as good as the original. The big mystery was: Why? They all look almost identical, with just tiny changes in a specific loop of the protein.

This paper uses super-computer simulations to figure out exactly why these tiny changes make such a huge difference.

---

The Main Characters

1. The Enzyme (OXA-48): The lockpick.
2. The Antibiotic (Imipenem): The key that needs breaking.
3. The "Deacylating Water" (DW): A tiny water molecule inside the enzyme that acts like a hammer. It smashes the antibiotic to break it.
4. The β5-β6 Loop: A flexible "arm" or "fence" near the active site where the breaking happens. In the different variants, this arm has been shortened or reshaped.

---

The Discovery: It's All About the "Handshake"

The researchers discovered that the efficiency of breaking the antibiotic depends entirely on how the Water Hammer (DW) holds hands with the Antibiotic Key.

1. The Perfect Grip (OXA-48 and OXA-517)

In the original enzyme (OXA-48) and the new OXA-517 variant, the "fence" (the β5-β6 loop) holds the antibiotic in a perfect position.

• The Analogy: Imagine the antibiotic is a fragile vase. The water hammer needs to hit it from a specific angle to shatter it. In these enzymes, the fence holds the vase steady, and the water hammer pushes against the vase (donating a hydrogen bond).
• The Result: The hammer hits perfectly. The antibiotic breaks easily. The enzyme works fast.

2. The Slippery Grip (OXA-163 and OXA-405)

In these two variants, the "fence" (the loop) has been cut short (deletions).

• The Analogy: Because the fence is missing a few planks, the vase (antibiotic) wobbles. The water hammer can't push against it anymore. Instead, the hammer accidentally pulls on the vase (accepting a hydrogen bond).
• The Result: It's like trying to break a vase by pulling on it instead of hitting it. It takes a massive amount of energy to do this, and the hammer often just misses. The enzyme becomes very slow and inefficient.

The "Thr213" Factor:
There is one specific amino acid in the fence called Thr213.

• In the good enzymes, Thr213 acts like a spotter in a gym. It helps organize the water molecules so the hammer knows exactly where to hit.
• In the bad enzymes, the fence is missing pieces, so the spotter is gone or confused. The water molecules get disorganized, and the hammer can't find its rhythm.

---

The Twist: Why is OXA-517 "Slow" Overall?

Here is a confusing part: OXA-517 has the same "hammer technique" as the original OXA-48 (it breaks the antibiotic fast once it grabs it). However, in real life, OXA-517 is still a worse antibiotic-killer than OXA-48. Why?

• The Problem: It's bad at grabbing the antibiotic in the first place.
• The Analogy: Imagine a baseball player (the enzyme) who is a world-class hitter (breaks the antibiotic fast). But, this player is terrible at catching the ball. The ball keeps rolling past them.
• The Cause: In OXA-517, the "fence" is slightly different, causing the antibiotic to sit in the wrong spot. It's like the player is standing two feet to the left. They can hit the ball great if they catch it, but they rarely catch it because it's not in the right place.

The Takeaway

This paper teaches us that in biology, tiny details matter immensely.

• Small changes = Big consequences: Deleting just a few amino acids in a loop (the fence) changes how water molecules arrange themselves (the handshake).
• Water is the hero: The way water molecules interact with the enzyme and the drug is the secret switch that turns the enzyme "on" or "off."
• Two ways to fail: An enzyme can fail because it can't break the drug (OXA-163/405) OR because it can't catch the drug (OXA-517).

Why does this matter?
Understanding these tiny "handshakes" helps scientists design new drugs. If we can build a new antibiotic that forces the water hammer to pull instead of push, or that jams the fence so the enzyme can't catch the drug, we can stop these super-bacteria in their tracks.

Technical Summary

1. Problem Statement

Antimicrobial resistance (AMR) is a critical global health threat, largely driven by $\beta$-lactamases, enzymes that hydrolyze $\beta$-lactam antibiotics. The OXA-48 family of Class D $\beta$-lactamases is particularly concerning due to its ability to hydrolyze carbapenems (often "last-resort" antibiotics). While the parent OXA-48 enzyme efficiently hydrolyzes imipenem, naturally occurring variants with mutations in the $\beta$5-$\beta$6 loop exhibit divergent kinetic profiles:

• OXA-163 and OXA-405: Possess deletions/substitutions in the $\beta$5-$\beta$6 loop and show significantly reduced carbapenem hydrolysis ($k_{cat}$) compared to OXA-48, despite enhanced activity against cephalosporins.
• OXA-517: Also has mutations in the $\beta$5-$\beta$6 loop but retains a $k_{cat}$ similar to OXA-48, though it exhibits a much higher $K_M$ (lower binding affinity).

The molecular basis for these differences—specifically how subtle loop mutations alter the active site dynamics to affect the rate-limiting deacylation step and substrate binding—remained unclear.

2. Methodology

The authors employed a multiscale computational approach combining Molecular Dynamics (MD) and Quantum Mechanics/Molecular Mechanics (QM/MM) simulations to investigate four enzymes: OXA-48 (wild-type), OXA-163, OXA-405, and OXA-517.

• System Setup:
  • Acylenzyme Complexes: Structures were derived from PDB (6P97, 7KHZ, 5FDH, 6HB8) or modeled by docking imipenem into apo structures. The catalytic base Lys73 was modeled as carboxylated (KCX).
  • Michaelis Complexes: Generated by breaking the Ser70-imipenem bond to simulate non-covalent binding.
  • Force Fields: Amber ff14SB for proteins, GAFF2 for imipenem, and TIP3P for water.
• Molecular Dynamics (MD):
  • Performed 120 ns unrestrained simulations for acylenzyme complexes (5 replicas each) and 10.5 ns for Michaelis complexes (10 replicas each).
  • Analyzed conformational clustering, flexibility (RMSF), water channel width (Val120-Leu158), and hydrogen bond (H-bond) networks.
• QM/MM Reaction Simulations:
  • Method: 2D Umbrella Sampling (US) using DFTB2 (semi-empirical) for the QM region (43 atoms: imipenem core, Ser70, KCX, deacylating water) and ff14SB for MM.
  • Reaction Coordinates: Proton transfer (KCX $\leftrightarrow$ Deacylating Water) and nucleophilic attack (Water $\rightarrow$ Imipenem C7).
  • Validation: Free energy barriers were corrected to the M06-2X/def2-TZVP level (DFT) to ensure accuracy.
• Binding Energy Analysis:
  • MM/GBSA calculations were used to estimate relative binding affinities ($K_M$ correlates) for the Michaelis complexes.

3. Key Contributions and Results

A. The Rate-Limiting Step and H-Bonding Patterns

The study confirmed that the formation of the tetrahedral intermediate (TI) during deacylation is the rate-limiting step. The efficiency of this step depends critically on two factors:

1. Hydration of KCX: A "less hydrated" state (KCX accepting only one H-bond from water) lowers the energy barrier compared to a fully hydrated state.
2. Deacylating Water (DW) Orientation: The most efficient deacylation occurs when the DW acts as an H-bond donor to the imipenem 6$\alpha$-hydroxyethyl group.

Key Finding: There is a strong correlation ($R^2 > 0.99$) between calculated energy barriers and experimental $k_{cat}$ values, but only when the correct DW orientation is assumed for each variant:

• OXA-48 & OXA-517: The DW predominantly acts as an H-bond donor. This results in lower energy barriers (~17.1–17.2 kcal/mol after correction), consistent with high $k_{cat}$.
• OXA-163 & OXA-405: The DW predominantly acts as an H-bond acceptor. This results in significantly higher energy barriers (~19.3–18.8 kcal/mol), explaining their low $k_{cat}$.

B. Structural Origin: The $\beta$5-$\beta$6 Loop and Thr213

The study identified that mutations in the $\beta$5-$\beta$6 loop alter the H-bond network surrounding the active site, specifically involving residue Thr213:

• OXA-48: Thr213 accepts an H-bond from a water molecule, which in turn accepts an H-bond from the imipenem hydroxyl. This network stabilizes the DW as an H-bond donor.
• OXA-517: Despite a two-residue deletion, Thr213 remains and can still form a water-mediated network that stabilizes the reactive DW orientation (though slightly less frequently than OXA-48).
• OXA-163 & OXA-405:
  • OXA-163 has a Thr213 that donates an H-bond to the backbone of residue 212, disrupting the network that supports the reactive DW orientation.
  • OXA-405 lacks Thr213 entirely.
  • Energetic Penalty: QM/MM potential energy calculations showed that forcing the DW into the "reactive donor" orientation in OXA-163 and OXA-405 incurs a significant energy penalty (5.3 and 7.5 kcal/mol, respectively) compared to OXA-48 and OXA-517 (2.6 and 2.2 kcal/mol).

C. Binding Affinity ($K_M$) Differences

While OXA-517 has a high $k_{cat}$, its overall efficiency is low due to a high $K_M$.

• MD Clustering: Revealed that in OXA-517, the imipenem binding pose shifts slightly toward the $\beta$5-$\beta$6 loop compared to OXA-48.
• MM/GBSA: Calculated a significantly lower binding affinity for OXA-517 ($-29.9$ kcal/mol) compared to OXA-48 ($-34.9$ kcal/mol). This is attributed to weakened interactions with residues Ile102, Trp105, Ser118, and Thr209 due to the shifted pose.
• OXA-163/405: Showed binding affinities similar to OXA-48, confirming that their reduced activity is purely due to the slower deacylation step, not binding issues.

4. Significance

• Mechanistic Insight: The paper provides the first atomistic explanation for why specific $\beta$5-$\beta$6 loop mutations in OXA-48 variants lead to a "switch" in carbapenemase activity. It demonstrates that activity loss is not just due to steric hindrance but is driven by subtle changes in the hydrogen-bonding network that dictate the orientation of the catalytic water molecule.
• Dynamic vs. Static: It highlights that crystal structures alone may not capture the dynamic H-bonding preferences required for catalysis; MD simulations are essential to observe the sampling of reactive vs. non-reactive states.
• Clinical Relevance: Understanding that OXA-517 retains high turnover but poor binding, while OXA-163/405 lose turnover, helps in predicting resistance profiles.
• Drug Design: The findings suggest that new $\beta$-lactamase inhibitors or antibiotics could be designed to specifically disrupt the critical water-mediated H-bond network (involving Thr213 and the deacylating water) to prevent efficient deacylation, thereby restoring antibiotic efficacy.

In summary, the study elucidates how minor loop mutations propagate through the active site to alter the hydrogen-bonding network, thereby controlling the orientation of the deacylating water and ultimately determining the enzyme's ability to hydrolyze critical carbapenem antibiotics.