Practical 5

Building a model from databases

Our aim is to construct a model of the Entner-Doudoroff pathway of glycolysis in the natural ethanol-fermenting organism Zymomonas mobilis. For examples of how such a model can be exploited, you can find two articles here and here.

Within ScrumPy, there is a module PyoCyc that can extract the required reaction equations from a downloaded copy of the MetaCyc database, starting from a list of the enzyme EC numbers. However, the BioCyc set of databases, including MetaCyc, have moved to a subscription model that makes it more difficult for us to share this with you. Hence you will need to consult the web version of the database and write the reaction equations your self.

Your task is to get the model to work so there is a viable route from glucose to ethanol.

Step 1

1. Download, by right-clicking on its link, the ScrumPy file Zymomonas.spy and save it in the directory you are going to use for this exercise.

2. Open the model file with ScrumPy. Two editor windows will open. The one for Zymomonas.spy contains some generic components of the model here that we are giving you as a starting point:

   1. An Include directive to the file !ZmED.spy which opens as the second, initially empty window into which you are going to write the pathwy reactions.

   2. A statement that WATER and PROTONS are regarded as external, freely-available species. Balancing every last PROTON is hard, so we're not going to struggle with t_hat yet.
   3. A set of transporter definitions that state that glucose (GLC), ethanol (ETOH) and carbon dioxide (CARBON-DIOXIDE) can exchange between the metabolism and the environment.
   4. A generic ATPase reaction that states that any ATP generated will be hydrolysed by other reactions in the cell.

3. Note that this means you will need to use the same metabolite names for H₂O, H⁺, etc.

Step 2

1. This list of EC numbers covers the Entner-Doudoroff pathway: EC-2.7.1.2, EC-1.1.1.363 ( or EC-1.1.1.388), EC-3.1.1.31, EC-4.2.1.12, 'EC-4.1.2.14','EC-1.1.1.44', 'EC-1.2.1.12', 'EC-2.7.2.3', 'EC-5.4.2.11', 'EC-4.2.1.11', 'EC-2.7.1.40', 'EC-4.1.1.1', 'EC-1.1.1.1']

2. Go to MetaCyc and enter the number for the first enzyme in the list: EC-2.7.1.2.

3. In the result, there is a section Reactions. If you mouse over the reaction specification, the link that shows up in your browser is the MetaCyc-assigned name fo the reaction (since an enzyme can catalyse more than one reaction, and a reaction can be catalysed by more than one enzyme.

4. Enter the rection name in the ZmED.spy file followed by a ':'; then copy the relevant reaction specification, followed by ' ~'.

5. Consider whether the MetaCyc metabolite names in the reaction match the names used for the same metabolite in previous or successive reactions. To see how this can arise as a problem, search MetaCyc for the compound D-glucose, then look at the Subclasses and Instances subsection of the result. It's partly your choice what name you use as standard, but any one metabolite must have a single, unique name in your model, and some metabolites have already been named for you in Zymomonas.spy.

6. Searching for the next enzyme, EC-1.1.1.363, returns a reaction specification containing the 'metabolite' NAD(P)+, meaning there are tow reactions possible; one with NAD+ and another with NADP+, and the corresponding products. The principal substrate in Z. mobilis is actually NAD+ (which is actually the specific reaction of EC-1.1.1.388).

7. Continue through the list adding the reactions to the !Zm.ED.spy window.
8. .

Step 3

1. Un-comment the Include directive in the Zymomonas.spy file to load your ZmED.spy and recompile the model using the drop-down ScrumPy menu option. Find out if there is a potential pathway through it by calculating the null space of the stoichiometry matrix:

   ns = m.sm.NullSpace()
   ns

   • Two windows should now be open, one for Zymomonas.spy and one for ZmED.spy. However, the null space is empty, in other words the system defined by this model cannot support a stead-state flux in any of its reactions.

2. The common possibilities for the lack of a viable route where there should be one are either missing reactions, or, when all the reactions are present, inconsistencies in metabolite definitions break the connectivity. Looking for dangling (orphan) metabolites can point to both of these issues:

   m.sm.OrphanMets()

3. If the network connectivity is being broken by inconsistent metabolite naming, edit ZmED.spy, save it, and recompile the model. Then check the null space and orphan metabolites again.
4. Once you get one or more null space vectors, you can compute the elementary modes of the model and the stoichiometry of their overall reaction to check whether you have the exected conversion of glucose to ethanol.
5. The model can now be reduced by removing reactions that will always be dead. Calculate the enzyme subsets of the model; all the dead reactions will be in the "dead" subset. Rather than deleting the reactions immediately, remove them from the model by placing a comment marker, "#", at the start of each line belonging to an unwanted reaction. Recompile the model and check that you have not broken the network.
6. How many reactions are left in ZmED.spy when you have inactivated all the dead ones?

Step 4

Time permitting, go to the next part.