Project Overview

One protein, trying to do the job of three machines

CxnA is an engineered enzyme — two natural ones fused into one — that already does two jobs most industrial processes need two separate machines for. We tried to teach it a third job, and to make it better at the two it already has. Here's the plain-language version of what that means and what we found.

Cellulose degradation, step by step

Cellulose. A chain built to resist breaking down.

Long chains of glucose, linked end to end and packed tightly together with dense hydrogen bonding — this is what a plant cell wall is mostly made of. Three different enzymes take turns pulling it apart. Meet the first.

Machine 1 — CenA (Endoglucanase)

It cuts in the middle.

CenA

CenA, an endoglucanase, cleaves random bonds inside the chain, creating new free ends for the next machine to grab.

Machine 2 — Cex (Exoglucanase)

It cuts from the ends.

Cex

Cex attacks a chain end and releases two linked glucose units — cellobiose — at a time.

Machine 3 — β-Glucosidase

It splits the last bond.

BGL

β-Glucosidase breaks the one remaining bond in cellobiose, releasing two separate glucose molecules.

The payoff

Glucose. Ready for fermentation.

The smallest unit — a simple sugar that yeast and bacteria can ferment straight into fuel.

The catch

Scattered enzymes. A random search.

CenA

Cex

BGL

CenA

Cex

BGL

Many copies of each enzyme drift through the cell at once. An intermediate has to randomly collide with the right one before it can be processed — some bump into it quickly, some wander for a while first.

The idea behind this project

One machine. Every step. One place.

CenA Cex BGL

Fuse the enzymes together and the intermediates never have to travel — no drifting, no searching, no waiting in line. That's the bet this project is testing.

In plain language

Think of cellulose as a rope that needs to become a handful of beads

Cellulose — the material in plant cell walls, and one of the most abundant organic substances on Earth — is a long, tough chain of linked sugar units. Turning it into glucose (the simple sugar that can actually be fermented into biofuel) means cutting that rope down, in stages, until it's just loose beads.

Normally, turning cellulose into glucose takes three separate enzymes working in sequence. CxnA already combines the first two jobs in a single protein. This project tried to add the third — and to make the first two faster — using directed evolution.

Most industrial setups use three separate enzymes for this, floating freely in solution. That's inefficient — separate proteins don't always find their target at the right moment, and manufacturing three proteins costs more than manufacturing one.

CxnA isn't something nature made on its own — it's an engineered fusion protein, built by joining the catalytic domains of two separate cellulose-degrading enzymes naturally found in the soil bacterium Cellulomonas fimi: an exoglucanase (Cex) and an endoglucanase (CenA), linked through a carbohydrate-binding module. It was first constructed and characterised by Duedu & French at the University of Edinburgh in 2016 [1].

This project asked two questions at once. Could directed evolution teach CxnA a third trick — splitting cellobiose into usable glucose, the one job it's missing? And could the same process make the two jobs it already does faster or more robust? Both questions were tested with the same tool: scrambling CxnA's DNA at random, then screening the resulting mutants for anything better.

Technical framing: domain architecture and mechanistic rationale

CxnA has only two domains: Cex (glycoside hydrolase family GH10, exoglucanase) and CenA (family GH6, endoglucanase), joined through a carbohydrate-binding module. There is no separate third domain in the protein — the goal for the missing β-glucosidase activity is to evolve it directly within Cex's existing active site. Cex (GH10) and the β-glucosidases of family GH1 both belong to clan GH-A, sharing a (β/α)₈ fold and an equivalent pair of catalytic glutamates — which is what makes coaxing glucosidase chemistry out of Cex's own active site structurally plausible.

Structural and docking analysis (see Results — Dry Lab) found that cellobiose can occupy Cex's active site in contact with its existing catalytic residues, GLU127 and GLU233 — the same two residues Cex already uses on its natural substrate. Separately, CenA's own active site (Asp595, Asp631, Asp666, Asp771) was analysed for mutations that might improve its existing activity (see Results — CenA).

There is no direct published precedent for evolving a GH10-family exoglucanase toward glucosidase-type activity. This project tests that gap directly, with the mutagenesis carried out by error-prone PCR across the gene — random, not computationally targeted — so the dry-lab predictions above informed where we expected activity to emerge, rather than dictating which mutations were actually tried.

Two questions, one experiment

Project A & Project B

Both questions were asked of the same mutant library, screened through the same three assays — they differ in which result would count as a "win," not in the wet-lab pipeline itself.

Project A No confirmed hit

Trifunctional evolution: adding the missing third activity

Goal: Evolve β-glucosidase-like activity into Cex's own active site, so CxnA can finish the job of turning cellobiose into glucose by itself.

How it was tested

Error-prone PCR (epPCR) across the CxnA gene to generate random mutants
X-gluc colour screen — mutants that develop a β-glucosidase-like activity turn visibly blue
MUC fluorescence assay for a precise, quantitative activity number per mutant
Guided (not dictated) by AlphaFold-based docking of cellobiose into Cex's active site

Honest framing

No mutant matched the positive control. CO12 read above wild type at 1.35× on average, but sequencing returned the parent sequence for it and for every other codon-optimised variant tested — so that figure sits inside the spread shown by genetically identical isolates, and is not evidence of an evolved activity. Reported as a real, interpretable negative result rather than a hidden failure.

Dry lab inputs

Cex structure validated against the 1FHD crystal structure · cellobiose docked next to GLU127/GLU233 · conservation analysis indicating both residues are functionally essential · computational mutation screen (FoldX + re-docking)

View dry-lab results for Project A →

Project B No confirmed hit

Activity enhancement: improving what CxnA already does

Goal: Use the same mutant library to look for variants with better exo- or endoglucanase activity than wild-type CxnA.

How it was tested

Congo red plate screen — a clearing halo means endoglucanase activity survived
MUC fluorescence assay, quantifying exoglucanase activity precisely against same-day wild-type controls
Computational candidate screen on CenA (the endoglucanase domain) independently of the random wet-lab mutants

Rationale

Project B doubles as a sanity check on the whole pipeline: CxnA's existing activities are real and measurable, so if the screens can't detect genuine known activity reliably, that would call the whole method into question. The assay cleared that bar in one direction: it resolved complete loss of function unambiguously, with T14 and T15 sitting at the dead-enzyme control. Resolving a modest gain proved harder — colonies with no detectable mutations spanned 0.57× to 1.39× of wild type, which is the real detection floor and the reason no variant could be called an improvement.

Dry lab inputs

CenA structure and conservation analysis · cellotetraose docked against catalytic Asp595/Asp631/Asp666/Asp771 · combined stability + binding mutation screen, best candidate K671S

View dry-lab results for Project B →

Broader context

Why does this matter?

Biofuel production

Enzyme cost is a major bottleneck in turning agricultural waste and energy crops into fuel — one estimate puts enzyme production at around 30% of total process cost. A single protein doing the job of three could cut that cost directly.

Biomass recycling

The same chemistry underlies waste valorisation, paper and textile bioprocessing, and soil carbon cycling. A cheaper, simpler enzyme system lowers the barrier to using any of them at scale.

Testing AI-assisted protein engineering

This project also shows where AlphaFold-based prediction sits in a directed-evolution workflow — and where it stops. The computational work set priorities for where to look and gave a reference point for interpreting what turned up; because the mutagenesis was random, none of the predicted substitutions appeared in the library, so the predictions themselves were never tested experimentally. That limit is part of the result.

Affiliation

iDEC — International Directed Evolution Competition

iDEC is the International Directed Evolution Competition — an annual student research competition focused on applying directed evolution to solve real biological problems. Teams design and execute original directed evolution experiments, with results judged on scientific rigour, innovation, and communication.

This project is affiliated with and submitted to iDEC 2026. The complete research record — including all methods, raw data, lab notebook entries, and honest discussion of limitations — is published on this site as the official project record.

Visit the iDEC website

References cited on this page

Duedu, K.O. & French, C.E. (2016). Characterization of a Cellulomonas fimi exoglucanase/xylanase–endoglucanase gene fusion which improves microbial degradation of cellulosic biomass. Enzyme and Microbial Technology, 93–94, 113–121.

Full bibliography with all project references: Results — References.