Our project is focused on the degradation of pectins, specifically homogalacturonan, to increase the speed at which composting takes place. The end product we hope to create is a cell lysate which we can apply to a compost pile containing enzymes that follow the degradation pathway of pectin. Our modeling has served both to understand and improve our system and its proficiency at breaking down homogalacturonan, as well as combating problems which face the application of our system, namely its thermostability.
Investigating Potential Organisms
When our team selected the enzymes we intended to use in our system we were faced with the question of what organism we wanted to source our enzymes from. We had a number of different options to choose from for each enzyme so we compiled data to make a better informed decision. This data included the protein’s molecular weight, length, extinction coefficient, isoelectric point, and in some cases optimal temperature. In cases where this data was not found the cell was simply left blank. The data for Pnl, PelB, and PelC follows:
This data was obtained by members of our team through UniProt and was used, in combination with a paper by Keggi and Doran-Peterson (2020), to decide which organism we would source our proteins from for future engineering. Ultimately we decided on the organism Paenibacillus amylolyticus, for multiple reasons. Firstly, enzymes from P. amylolyticus can degrade pectins without the use of an extracellular methylesterase (1), which is beneficial for us as it skips an entire portion of the degradation pathway. Secondly, the enzymes we chose are able to degrade both methylated and unmethylated homogalacturonan when obtained from P. amylolyticus (1). Finally, PelC and PelB work well together, with PelC giving PelB access to the internal demethylated regions of homogalacturonan (1). These factors come together to make P. amylolyticus an appealing organism to source our enzymes from.
Creating Homology Models
Unfortunately, the proteins from P. amylolyticus do not have 3D crystal structures available. In order to gain a better understanding of how our proteins function, we decided to generate a homology model using related, known structures of pectin degradation enzymes. The structures will also allow us to complete molecular dynamics simulations as well. Pnl, PelB and PelC were modeled using SWISS-MODEL and iTASSER. For Pnl, the sequences identified showed only 23% identity and also very low similarity (Table 4).
Figure 1: Homology models for Pnl from SWISS-MODEL (left), iTASSER (middle), with the structures aligned (right).
The structure of Pnl shows the polypeptide strand forming into the canonical 𝛃-helix fold, common among pectate lyase enzyme families. Due to the low sequence similarity the N-terminus was not able to be modeled through these methods.
The sequence used to model PelB had 93% identity and 96% similarity (Table 5), meaning that this structure could be used much more reliably for function predictions.
Figure 2: PelB modeled using SWISS-MODEL (red), iTASSER (orange), and sequences aligned (right).
Both SWISS-MODEL and iTASSER modeling of PelB resulted in an overall 𝛃-helix structure, typical of PL1 enzymes. Several alpha-helices decorate the exterior of the inner 𝛃-sheets. Similar to Pnl, PelB has additional amino acids at the N-terminus that were unable to be modelled using this software. However, predictions using QUARK, has shown that this region in both proteins fold into an extended alpha-helix.
Figure 3: Modelling the N-terminus of PelB using QUARK results in alpha-helical structures.
Unfortunately, PelC also had low sequence identity and similarity, 34 and 47%, respectively (Table 6) with the amino acid sequence that was used to predict the structure. As with Pnl, the low sequence similarity between these enzymes amino acid sequences means that we have to be careful when making any interpretations from the structures.
Table 6: Protein used to model P. amylolyticus PelC enzyme
The protein architecture again shows the 𝛃-helix fold common among polysaccharide lyase enzymes (figure 4). Similarly to Pnl and PelB, PelC also has an N-terminus that whose structure could not be predicted through these methods, was shown to have an alpha-helix using QUARK.
Figure 4: Structural predictions of PelC using SWISS-MODEL (green) and iTASSER (blue). The structures were then aligned on the right.
Improving the Thermostability of Pectin Degradation Enzymes
In an interview we conducted with Bill Macmillan, a waste and recycling engineer for the city of Lethbridge, we were given the advice that a system which could control the temperatures of the compost pile would provide more utility to its customers. Unfortunately controlling the conditions of an entire composting pile is out of reach for our project, but nevertheless the advice spurred us to improve the thermostability of our system. We came to realize that thermostability is extremely important for our project as composting piles can reach temperatures above forty-five degrees Celsius (2). These high temperatures may appear to benefit us - the optimal temperatures of all the proteins are between 55 and 70 degrees Celsius - but it is important to ensure that our system can remain stable over long periods of time. Bearing this in mind we began work on improving the thermostability of our system.
We completed multiple sequence alignments of our target proteins and analyzed the data we had to create a general consensus sequence. The consensus sequence for Pnl can see seen below:
We then analyzed this sequence with Popmusic to determine the sites at which point mutations would be most effective at improving thermostability. Team members then implemented these changes to the Pnl strand. Two different sets of point mutations were implemented, one with thermostability increases greater than fifteen , and one with increases limited to between five and eight.
Changes greater than fifteen to the Pnl consensus sequence:
With these new sequences we have hopefully created a more thermostable version of our project that is better suited to the high temperatures compost piles can reach.
We used the wildtype enzyme as a starting point to model the Pnl mutant with an increased melting temperature (Figure 5).
Figure 5: Structural prediction of Pnl thermostable variant.
- Keggi and Doran-Peterson. (2020) The homogalacturonan deconstruction system of Paenibacillus amylolyticus 27C64 requires no extracellular pectin methyltransferase and has significant industrial potential. Applied Environmental Microbiology. 86: e02275-19