Metabolic modeling elucidates the distinctive landscape of Pancreatic Ductal Adenocarcinoma (PDAC) cells

Andrea Goertzen Author

04/05/2021 Added

14 Plays

Description

In this work, two metabolic models of a healthy and cancerous pancreatic cell were reconstructed to identify potential therapeutic targets for treatment.

Searchable Transcript

Search:

[00:00:00.000]Hello my name is Andrea Goertzen, and I
[00:00:04.007]spent the past school year working in the
[00:00:06.397]Systems and Synthetic Biology Laboratory
[00:00:09.007]in the department of Chemical and
[00:00:11.227]Biomolecular Engineering here at UNL.
[00:00:13.507]So today I'm going to talk about my work
[00:00:15.517]with mapping the metabolic landscape of
[00:00:17.931]Pancreatic Ductal Adenocarcinoma cells
[00:00:20.861]Pancreatic Ductal Adenocarcinoma or PDAC
[00:00:24.665]is the third deadliest cancer in the
[00:00:26.579]United States, with a survival rate of
[00:00:28.704]only 8.2% over five years. This can be
[00:00:32.637]attributed to the lack of early diagnostic
[00:00:34.758]markers as well as resistance to
[00:00:36.687]chemotherapy and other treatments.
[00:00:38.825]My project focuses on metabolic
[00:00:41.511]modeling as a tool for identifying
[00:00:43.656]therapeutic targets for PDAC treatment.
[00:00:46.760]Metabolic models contain all of the genes,
[00:00:49.710]reactions, and metabolites known to be in
[00:00:52.067]a type of tissue. Specifically, I looked
[00:00:54.954]at pathways that seem unregulated in PDAC
[00:00:57.723]cells to sustain their rapid proliferation
[00:01:00.748]By understanding the key metabolic
[00:01:02.927]adaptations of PDAC cells, we can then
[00:01:06.128]identify these features as targets for
[00:01:08.430]therapeutic treatment. For this work, I
[00:01:12.356]reconstructed two models of a human
[00:01:14.128]pancreas - a PDAC model and a healthy cell
[00:01:17.632]model, and then compared their metabolic
[00:01:20.111]profiles. Next I'm going to talk a little
[00:01:22.259]bit more about how these models are
[00:01:24.271]reconstructed. The central DOGMA of
[00:01:27.868]biology tells us that information is
[00:01:29.902]carried from DNA, to RNA, to proteins,
[00:01:33.189]which then catalyze a reaction. Because
[00:01:35.657]of this, if we have a genome with all of
[00:01:38.098]the genetic information of a cell, we can
[00:01:40.174]know the reactions that take place in this
[00:01:42.212]cell. And once we have a set of reactions
[00:01:46.256]from that genetic information, it's
[00:01:48.193]important to check for imbalances in
[00:01:50.294]formulas and reactions. This is because
[00:01:52.256]the model needs to demonstrate a
[00:01:55.042]consistent, accurate flow of metabolites
[00:01:57.389]and imbalances would interrupt this.
[00:01:59.887]These reactions and formulas are checked
[00:02:02.517]against several biochemical databases.
[00:02:05.058]The next step of building a model is to
[00:02:08.771]construct a hypothetical reaction to
[00:02:10.721]account for the accumulation of macro-
[00:02:12.919]molecules, such as amino acids, lipids,
[00:02:15.645]and so on. This reaction is purely
[00:02:18.386]artificial, it does not occur in real life
[00:02:21.304]but we call it the biomass reaction. It is
[00:02:24.306]used to simulate cellular growth. Once we
[00:02:28.091]have a network of reactions, there will
[00:02:30.361]inherently be gaps. The next step in the
[00:02:32.845]process is to use biochemical databases
[00:02:35.724]to find the missing functionalities of
[00:02:37.991]the network that need to be added to fill
[00:02:40.457]these gaps and complete the model.
[00:02:42.872]Last, we assemble these reactions in the
[00:02:46.501]form of a stoichiometric matrix, where
[00:02:49.262]each row is a metabolite and each column
[00:02:51.952]is representative of a reaction. The
[00:02:54.551]values in the matrix are the
[00:02:56.240]stoichiometric coefficients of the
[00:02:58.221]metabolites for each reaction. For example
[00:03:00.989]if we look at this second column, which
[00:03:02.889]represents a single reaction, we see that
[00:03:06.492]one of a certain metabolite is consumed
[00:03:09.317]and two of one metabolite and three of
[00:03:12.435]another metabolite are produced. Once
[00:03:17.311]we have this mathematical structure, we
[00:03:19.441]then apply an important assumption called
[00:03:21.550]pseudo steady-state. For each metabolite,
[00:03:24.381]the rate of change of its concentration is
[00:03:26.524]equal to the sum of its stoichiometric
[00:03:28.831]coefficient times the flux of every
[00:03:31.149]reaction that it is involved in. At pseudo
[00:03:34.081]steady-state, the change in concentration
[00:03:36.214]of all the molecules over time is zero. So
[00:03:40.226]the mass balance equations define the
[00:03:43.981]biological system as a mathematical
[00:03:46.236]solution space, indicated by this higher
[00:03:48.764]order cone shape. The constraints for this
[00:03:51.452]cone are given by the steady-state
[00:03:55.013]assumption as well as upper lower bounds
[00:03:57.979]for each reaction imposed on the model.
[00:04:00.712]On this feasible space, we can use
[00:04:03.111]different optimization tools to analyze
[00:04:05.166]the metabolism. One of these is called
[00:04:07.871]Flux Balance Analysis, or FBA. FBA sets
[00:04:12.658]biomass as an objective function, and then
[00:04:15.785]finds the flux through every other
[00:04:17.450]reaction in the model subject to given
[00:04:19.917]constraints to optimize the flux through
[00:04:22.533]the biomass reaction. Stoichiometric
[00:04:27.055]models were constructed for both PDAC and
[00:04:30.110]healthy pancreas cells in this work using
[00:04:32.090]a tool called iMAT. iMAT uses gene
[00:04:35.146]expression data to assign scores to
[00:04:37.340]reactions, and the algorithm uses these
[00:04:40.369]scores to predict which reactions will
[00:04:42.767]be present in the model. So for this work
[00:04:45.813]we started with a genome scale metabolic
[00:04:48.112]model for human metabolism named Human1.
[00:04:50.574]And in this model we see that genes are
[00:04:53.301]represented by diamond shapes, metabolites
[00:04:55.346]by blue circles, and reactions by the
[00:04:57.450]rectangles. We then overlaid gene
[00:05:00.715]expression data to identify which
[00:05:03.302]reactions are more likely to be present in
[00:05:05.693]the model. For example, G1 is highly
[00:05:09.009]expressed, indicated by its green color,
[00:05:11.785]gene G6 is yellow, and that indicates
[00:05:16.630]that it's moderately expressed, and G3 is
[00:05:18.848]lowly expressed, indicated by its red
[00:05:21.059]color. By assigning these high, moderate,
[00:05:24.480]and low scores of the genes to their
[00:05:26.693]corresponding reactions, iMAT produces a
[00:05:29.778]model that includes all the reactions
[00:05:32.051]likely to be active. For example, G4 was a
[00:05:36.680]highly expressed gene, so its
[00:05:38.796]corresponding reaction R4 is likely to be
[00:05:42.414]present in the pancreas model. G7, on the
[00:05:45.859]other hand, was lowly expressed, so its
[00:05:48.399]corresponding reaction, R7, does not
[00:05:52.358]appear in our tissue specific model.
[00:05:54.988]The results of these model analyses show
[00:05:59.453]that certain pathways are more or less
[00:06:02.771]active in PDAC cells relative to healthy
[00:06:06.144]pancreas cells. This table shows a
[00:06:09.581]snapshot of the percentage of reactions in
[00:06:11.965]a PDAC cell whose flux space shrunk or
[00:06:15.027]expanded compared to a healthy cell. The
[00:06:17.752]blue bars pointing to the right indicate
[00:06:20.132]that these pathways are upregulated in
[00:06:22.417]PDAC cells relative to healthy cells. Red
[00:06:26.215]bars pointing to the left indicate that
[00:06:28.886]the pathways are downregulated in PDAC
[00:06:32.380]cells compared to healthy cells. To get a
[00:06:35.869]closer look at this, let's examine one of
[00:06:38.146]the pathways. The fatty acid elongation
[00:06:40.522]pathway was upregulated in PDAC cells,
[00:06:43.210]meaning it's more active. The increase in
[00:06:46.363]flux through this pathway indicates that
[00:06:48.800]PDAC cells likely utilizes longer fatty
[00:06:52.114]acids for synthesizing biomass. To
[00:06:57.708]reiterate, this research identified key
[00:07:00.360]metabolic adaptions of PDAC cells through
[00:07:03.773]metabolic modeling.
[00:07:05.714]To further validate these models,
[00:07:08.311]experimental data will confirm the
[00:07:10.473]identified therapeutic targets. Next, with
[00:07:15.765]additional transcriptomic, metabolic,
[00:07:18.366]and proteomic data, the aim is to develop
[00:07:20.710]the first genome-scale kinetic model of
[00:07:23.472]PDAC cells. Kinetic models paint a more
[00:07:26.987]accurate picture of cell metabolism
[00:07:29.161]because they include regulatory
[00:07:30.703]constraints based on metabolite
[00:07:33.170]concentration or enzyme activity. I would
[00:07:38.116]like to thank Mohammad Mazharul Islam
[00:07:40.273]and Dr. Saha for being mentors to me
[00:07:42.690]throughout this project. I would also like
[00:07:45.718]to say thank you to the University of
[00:07:47.605]Nebraska UCARE program for allowing me to
[00:07:49.920]participate in this research. I am happy
[00:07:52.580]to answer any questions that you have for
[00:07:54.531]me at this time.

The screen size you are trying to search captions on is too small!

You can always jump over to MediaHub and check it out there.

Comments

0 Comments

Metabolic modeling elucidates the distinctive landscape of Pancreatic Ductal Adenocarcinoma (PDAC) cells

Description

Searchable Transcript

Comments icon comment

Related Channels

Comments