Macromolecular Modeling Blog ™

An undergraduate group of students from the University of Washington, managed to redesign an enzyme to efficiently break down a model peptide for Gluten, achieving an increase in activity of hundreds fold over the only current clinically tested treatment for Gluten intolerance – Winning iGEM2011. All this, over the summer. Oh, and they also managed to produce Diesel in E. Coli.

Congratulations to the University of Washington for winning iGEM 2011 (The International Genetically Engineered Machine competition). iGEM is an international undergraduate Synthetic Biology competition. Student teams are given a kit of biological parts at the beginning of the summer from the Registry of Standard Biological Parts. Working at their own schools over the summer, they use these parts and new parts of their own design to build biological systems and operate them in living cells.

The winning group championed two amazing projects, as they coined it: “make it or break it” – make it: Diesel production in e. coli; break it: Gluten destruction.

In this post (based largely on the very informative group’s Wiki) I want to focus on the latter project: the redesign of an enzyme to more efficiently break gluten in a low pH environment.

It is estimated that around 0.5-1% of the population (US & Europe) suffer from gluten intolerance (celiac) which is an adverse reaction to gluten proteins found in wheat, barley, and rye products. The glutens invoke an immune response in the digestive tract of genetically predisposed individuals resulting in inflammation of the gut, impeding the absorption of nutrients. There are currently no effective therapies for this lifelong disease except the total elimination of glutens from the diet.

Proline (P)- and glutamine (Q)-rich components of gluten known as ‘gliadins’ appear to be responsible for the bulk of the immune response in most patients. Their high PQ content protects gliadin oligopeptides from degradation by gastrointenstinal endoproteases, but also presents a target for drug design. Any peptidase capable of cleaving at or near the P-Q bond while remaining active at the temperature and harsh pH of the stomach would have pharmacological potential as a therapy for celiac.

One candidate enzyme, currently in clinical trials, utilizes a prolyl endopeptidase (PEP) from Sphingomonas capsulata (SC) to hydrolyze gliadins. Unfortunately, the enzyme’s optimal activity is at pH 7. It is therefore pertinent to identify new candidate enzymes that have both activity on the gliadin PQ structural motif, as well as optimal activity at gastric pH levels during digestion around pH 4.

The UW team, rather than focusing primarily on substrate specificity when choosing the candidate, identified an enzyme already capable of catalyzing peptide hydrolysis at gastric pH levels, regardless of peptide substrate specificity. Upon identification of such an enzyme they used computational tools to reengineer its substrate specificity for enhanced activity on immunogenic gliadin peptides with the common PQ structural motif.

Kumamolisin, isolated from the thermoacidophilic bacterium Alicyclobacillus sendaiensis strain NTAP-1, has been shown to be produced in a recombinant host, and exhibits significant enzymatic activity at pH 2.5 and above. Its maximal activity is at about pH 4.0. It is this robust activity under acidic, gastric conditions that makes Kumamolisin so promising for the development of a pill for gluten intolerance

Specificity studies have been conducted on the native enzyme, which is most efficient at hydrolyzing peptides at the carboxyl side after the PR dipeptide motif. The combination of Kumamolisin having the desired enzymatic activity under gastric pH levels, and the P2 subsite already having the specificity for the proline from the desired PQ motif makes it a highly attractive candidate for further engineering. In addition, several crystal structures of this enzyme have been solved, allowing for the use of computational design techniques.

The team has used FoldIt as an interface to Rosetta. Within FoldIt, they designed Kumamolisin’s crystal structure in complex with a model PQLP peptide that recurs frequently in gluten, thus mimicking gluten as a substrate.

The amino acid residues around the active site of Kumamolisin were modified, attempting to decrease the free energy of, and thus stabilize, the system. Estimations of free energy were based on the FoldIt score (representing Rosetta’s score) Using this method, they proposed over 100 novel mutants, each of which could potentially increase Kumamolisin’s proteolytic activity on gluten.

To test the designs, the group have developed a whole cell lysate  fluorescence based assay that allowed to perform a rough screen of a large number of mutants to estimate relative activity on breaking down a PQLP peptide. Mutants that showed the most increase in activity from the wild-type Kumamolisin were purified and tested against the wild-type and against SC-PEP using the same fluorescence metric designed for the whole cell lysate assay (now with a known concentration).

At pH 4 wild type Kumamolisin was shown to have 6.5 fold higher activity than SC-PEP (the enzyme currently in clinical trials for breaking down gluten) for hydrolysis of the model PQLP peptide, validating its selection as template.

The whole cell lysate assay performed on over 100 mutants designed with FoldIt, found several designs with up to 10-fold higher activity over WT Kumamolisin. The best mutant (validated also in the purified enzyme assay) showed an 11-fold increase with just one point mutation.

Successful mutations were combined to construct a second library for screening. After designing a collection of combinatorial mutants, drawing from successful mutations discovered in the first round, the assay was repeated. From the initial screen of the combinatorial mutants, enzymes displaying around 50 times better activity than native Kumamolisin on breaking down PQLP were found.

By combining two of the top groups of mutations from the first round, an over 100-fold increase in activity on breaking down PQLP from the wild-type enzyme was achieved. This variant enzyme is ultimately 784 times better at breaking down PQLP than SC-PEP, the enzyme currently in clinical trials for treating gluten intolerance.

In my eyes this story proves two things:

1. FoldIt, as much as it was paraded of late for its crowd sourcing abilities is an excellent tool in itself for interfacing with Rosetta and lowering the energy barrier to utilize it.

2. Challenging undergraduates can result in solving most of the world’s problems (over the summer no less!)