iiinova — where your digital self keeps evolving

Roberto Chica Lab @chicalab.bsky.social · 06/10/2026

Mechanistic analyses show that multistate design preserves a balanced reaction coordinate whereas single-state optimization over-stabilizes one target state, redistributes energy barriers, and creates new kinetic bottlenecks. Designing one state ≠ designing catalysis.

Figure showing mechanistic experiments comparing the parent enzyme BH32.12, the multistate-designed variant BH32.M6, and the single-state-designed variant BH32.S8.
Panel A: Kinetic isotope effect experiments probe one specific step of the four-step reaction pathway. BH32.12 and BH32.M6 exhibit similar isotope effects, indicating similar rate-limiting behavior, whereas BH32.S8 shows no kinetic isotope effect, suggesting a shifted kinetic bottleneck.
Panel B: Solvent viscosity experiments measure the contribution of diffusion-controlled processes to catalysis. BH32.12 and BH32.M6 display similar viscosity dependencies, while BH32.S8 behaves differently, revealing an altered balance between chemical and diffusion-controlled steps.
Overall, the figure shows that multistate enzyme design yields a balanced catalytic mechanism similar to that of the parent enzyme BH32.12, whereas single-state design creates a kinetic bottleneck earlier in the mechanism.

100

Roberto Chica Lab @chicalab.bsky.social · 06/10/2026

Why does it work? Structural analyses show that multistate design preserves an active site compatible with multiple transition states, whereas single-state optimization does not.

Figure showing crystal structures and conformational ensemble analyses of the parent enzyme BH32.12 and designed variants.

Panel A: Overlay of the protein structures showing that extensive active-site mutations do not alter the overall protein fold.

Panel B: Comparison of a flexible region spanning residues 181-193. This region is partially disordered in the parent enzyme but becomes ordered in the designed variants, indicating local structural remodeling.

Panel C: Close-up views of the catalytic His23-Arg124 dyad, highlighting differences in side-chain conformations and active-site geometry among the variants.

Panel D: Analysis of the conformational ensemble of catalytic His23. Experimental conformations are compared with those predicted during design. BH32.12 and the best multistate-designed variant sample a broad range of conformations that closely match the transition-state models, whereas the single-state-designed variant samples a narrower set of conformations with reduced agreement.

Panel E: Comparison of catalytic Arg124 conformations with those predicted for the four transition states along the reaction pathway. The best multistate design variant maintains conformations that more closely match the designed catalytic states than the single-state variant.

Overall, the figure shows that the most active multistate-designed enzyme best retains the catalytic conformations predicted to support multiple transition states during catalysis.

100

Roberto Chica Lab @chicalab.bsky.social · 06/10/2026

The results were striking: ✅ >100-fold higher bisubstrate catalytic efficiency than conventional single-state designs ✅ Lead variant outperformed the extensively evolved parent enzyme ✅ Higher enantioselectivity (93% ee vs 86% ee)

Figure showing the performance and properties of designed Morita-Baylis-Hillmanase (MBHase) enzymes. Panel A is a bar graph comparing product conversion for multiple designed variants and control enzymes. Several multistate design variants achieve substantially higher conversion than earlier benchmark enzymes, with six out of ten exceeding the performance of BH32.8

Panel B is a three-dimensional protein structure of BH32.12. Residues mutated during design are highlighted as colored spheres clustered around the enzyme active site.

Panel C is a bar graph comparing catalytic efficiency and enantioselectivity of MBHase variants. The lead multistate design variant, BH32.M6, displays the highest overall performance, exceeding the parent enzyme BH32.12 in both catalytic efficiency and enantiomeric excess.

Overall, the figure demonstrates that multistate enzyme design can generate highly active and stereoselective catalysts, with the best variant outperforming previous generations of evolved and designed MBHases.

100

Roberto Chica Lab @chicalab.bsky.social · 06/10/2026

Instead of designing a static active site, we design for a dynamic reaction coordinate. To test the idea, we redesigned the evolved de novo Morita-Baylis-Hillmanase BH32.12, a four-step carbon-carbon bond-forming enzyme.

Diagram illustrating the workflow of multistate enzyme design. Multiple protein backbone conformations are combined with four transition-state models (TS1–TS4) representing different steps along a catalytic reaction. Each transition state is shown as a colored molecular structure surrounded by catalytic amino acid side chains in white, arranged in idealized geometries. The transition-state models are fitted onto different protein conformations to identify structures that best accommodate them without steric clashes. Active-site residues, highlighted in orange, are then computationally optimized to stabilize all transition states simultaneously. The resulting design produces a single enzyme sequence capable of adopting four distinct conformations corresponding to different stages of the catalytic pathway. For comparison, a conventional single-state design strategy is shown using only one protein conformation and a single transition-state model representing the rate-limiting step (TS3). The figure emphasizes that multistate design explicitly accounts for the full reaction coordinate, whereas single-state design focuses on only one catalytic state.

110

Roberto Chica Lab @chicalab.bsky.social · 17/02/2026

5/6 Molecular dynamics simulations helped dissect structural determinants of catalysis. Our catalytic‑competency score, i.e. the fraction of snapshots with both key catalytic contacts formed, tracks well with experimental activity across variants.

100

Roberto Chica Lab @chicalab.bsky.social · 17/02/2026

3/6 We solved the KempTIM1 crystal structure (blue) bound to a transition‑state analogue. The designed lid (magenta) shows excellent agreement with the model, and the ligand binds in a similar pose to the design model, enabling the intended catalytic contacts.

100

Roberto Chica Lab @chicalab.bsky.social · 17/02/2026

2/6 After reviewer feedback, we boosted activity and hit rate. Keeping the same active‑site design, we improved solubility and expression via surface redesign, yielding KempTIM1, one of the most active first‑round de novo enzymes reported to date (kcat/KM = 21,000 M⁻¹ s⁻¹).

100

Roberto Chica Lab @chicalab.bsky.social · 17/02/2026

1/6 Thrilled to announce our updated preprint on the CANVAS method for de novo enzyme design, in collaboration with @birtehoecker.bsky.social. We show how CANVAS builds custom lids onto minimal TIM barrels, transforming them into efficient catalysts. www.biorxiv.org/content/10.1...

152

Roberto Chica Lab @chicalab.bsky.social · 04/12/2025

2/ Speakers of the 6th Protein Engineering Canada Conference include: Bill DeGrado Joelle Pelletier Tim Whitehead Anastassia Vorobieva Lucy Colwell Ai Niitsu @joannas.bsky.social @nickpolizzi.bsky.social @possuhuanglab.bsky.social @paolalaurino.bsky.social @stephanhammer.bsky.social and more!

032

Roberto Chica Lab @chicalab.bsky.social · 04/12/2025

1/ Save the date! The 6th Protein Engineering Canada Conference will be held on June 22-24 in Ottawa, Canada. Abstract submission and registration are open! More information here: event.fourwaves.com/pec2026/pages

Announcement for the 6th Protein Engineering Canada Conference, to be held June 22nd-24th in Ottawa, Canada. The image shows a protein structure in front of a picture of Parliament Hill and Chateau Laurier in Ottawa.

163

Roberto Chica Lab @chicalab.bsky.social · 29/07/2025

Our crystal structure validated the designed fold, confirming that the lid was correctly folded. However, a subtle 1.8 Å lid shift disrupted a key catalytic contact, likely contributing to the modest activity. But structural analysis reveals paths to improve activity in the next round of design!

The crystal structure (blue) aligns closely with the design model (minimal TIM barrel and lid colored white and magenta, respectively).

120

Roberto Chica Lab @chicalab.bsky.social · 29/07/2025

One of our designs, KempTIM4, showed catalytic efficiency comparable to many first-round de novo Kemp eliminases generated by traditional methods.

Michaelis-Menten plot of KempTIM4 showing saturation kinetics.

110

Roberto Chica Lab @chicalab.bsky.social · 29/07/2025

Using CANVAS, we designed a structural lid onto a minimal, de novo TIM barrel to anchor catalytic residues and create an active site for the Kemp elimination reaction.

Building a custom lid onto a minimal, de novo TIM barrel using CANVAS.

110

Roberto Chica Lab @chicalab.bsky.social · 29/07/2025

TIM barrels are among nature’s most powerful enzyme scaffolds but making them from scratch with catalytic function has been a challenge. Enter CANVAS: a computational pipeline combining Triad, RFdiffusion & ProteinMPNN to customize minimal TIM barrels into functional enzymes.

110

Roberto Chica Lab @chicalab.bsky.social · 28/02/2025

Molecular dynamics simulations showed that distal mutations enhance active-site accessibility—either by loosening loops covering the active site or widening bottlenecks for substrate entry & product exit. The enzyme breathes more efficiently! 🌬️ (5/6)

100

Roberto Chica Lab @chicalab.bsky.social · 28/02/2025

Kinetic solvent viscosity effects & stopped-flow experiments showed that distal mutations don’t just tweak structure—they accelerate substrate binding & product release. (4/6)

100

Roberto Chica Lab @chicalab.bsky.social · 28/02/2025

Crystal structures showed that active-site mutations pre-organize the catalytic machinery. But distal mutations? They subtly tune conformational dynamics—enhancing productive substates & reshaping the energy landscape of the catalytic cycle. (3/6)

100

Roberto Chica Lab @chicalab.bsky.social · 28/02/2025

We engineered "Core" and "Shell" variants of three evolved Kemp eliminases to dissect the effects of active-site vs. distal mutations. Core mutations dramatically boosted catalysis. Shell mutations alone? Not much—until they worked together in evolved enzymes. 🔍 (2/6)

110

Roberto Chica Lab @chicalab.bsky.social · 27/02/2025

Crystal structures revealed that distal mutations trigger large-scale conformational shifts in an active-site loop, making the active site more open. Interestingly, these mutations aren't located on the loop—making their effects hard to predict!

110

Roberto Chica Lab @chicalab.bsky.social · 27/02/2025

We studied a computationally designed & evolved retro-aldolase, creating two variants: 🔹 RA95-Core – Active-site mutations only 🔹 RA95-Shell – Distal mutations only Comparing them to the evolved RA95.5-8F revealed surprising insights!

110

Roberto Chica Lab @chicalab.bsky.social · 25/03/2024

Protein Engineering Canada Conference. 1 week left to benefit from early-bird discount on registration! event.fourwaves.com/pec2024/pages

010

Roberto Chica Lab @chicalab.bsky.social · 01/03/2024

Inspiration for an excellent beer!

120

Roberto Chica Lab @chicalab.bsky.social · 25/01/2024

Save the date! The 5th Protein Engineering Canada Conference will take place at the University of Toronto on June 26-28, 2024. event.fourwaves.com/pec2024

010

Roberto Chica Lab @chicalab.bsky.social · 06/11/2023

Save the date! The 5th Protein Engineering Canada Conference will take place at the University of Toronto on June 26-28, 2024. For more info: event.fourwaves.com/pec2024

010

Roberto Chica Lab @chicalab.bsky.social · 03/11/2023

We show that our enzyme design approach outperforms other protocols because the crystallographic data underpinning our backbone ensembles provide the highest-fidelity insight into the true conformational ensemble of the enzyme, including accurate representation of catalytically-competent substates.

000

Roberto Chica Lab @chicalab.bsky.social · 03/11/2023

Here, we use conformational ensembles generated from dynamics-based refinement against X-ray diffraction data to design artificial enzymes with efficiencies comparable to those obtained after several rounds of directed evolution.

110