Team:MIT/Engineering
Engineering Success
We will describe some of the iterations of our engineering design process below.
1. RESEARCH
ITERATION 1: We began our work as a team primarily interested in COVID-19 and the lesser known immune system impacts of the disease resulting in symptoms such as the cytokine storm. We searched the literature for clinical data on elevated cytokine levels in COVID-19 patients, while taking into consideration that our anti-inflammatory actuator protein may need to target whatever cytokines we choose to sense. After doing extensive literature research on COVID-19 cytokine storms (CCS), we found that IP-10 and MCP-3 are the cytokines whose blood plasma levels most strongly correlate with CCS. These two cytokines have downstream signaling pathways that include the nuclear transcription factors (nTFs) ELK-1 and NF-κB, respectively.
ITERATION 2: After electing to sense IP-10 and MCP-3, we researched potential circuitry for processing elevated levels of these cytokines. This involved looking through existing systems that process cytokine inputs through an AND gate in other inflammatory disease states, and systems for post-transcriptional mRNA-level regulation.
ITERATION 3: After deciding on a PERSIST-based system as our AND gate, we had to solidify our choice of anti-inflammatory actuator protein. After some research, we found our options included receptor antagonists, decoy receptors, and single chain variable fragments (scFvs).
Read more on our Design page!
2. IMAGINE
ITERATION 1: At the start of our engineering design process, we imagined our engineered system would be able to sense the CSS through the levels of the biomarkers ELK-1 and NF-κB, process these inputs through an AND gate, and respond to the CSS and correlated hyperinflammation by releasing an anti-inflammatory output protein.
ITERATION 2: We imagined potentially having our circuit produce a graded, analog response to cytokine levels, as opposed to a sigmoidal, digital one. However, we did not find success in building analog circuitry for mammalian cells, as described in DESIGN ITERATION 2.
ITERATION 3: We imagined having an anti-inflammatory actuator protein that would be tunable and able to target different cytokines so that we could compare our circuit’s effectiveness in decreasing levels of various cytokines.
Read more on our Implementation page!
3. DESIGN
ITERATION 1: We first designed our cytokine sensing circuits. Since we would be sensing levels of the biomarkers ELK-1 and NF-κB, we decided to have two plasmids, utilizing the CMV promoter with a responsive element for either ELK-1 or NF-κB.
ITERATION 2: We then considered different options for processing our cytokine inputs: cascades, bacterial analog circuitry, and Programmable Endonucleolytic Scission-Induced Stability Tuning (PERSIST). When trying to apply the analog circuit design to mammalian cells, we ran into difficulties finding a system mechanistically analogous to the bacterial AraC protein that was used in the bacterial circuit. Our final circuit used PERSIST for post-transcriptional mRNA-level regulation.
ITERATION 3: While considering our options of anti-inflammatory actuator protein (receptor antagonists, decoy receptors, scFvs), we chose scFvs because they more specific and stable than decoy receptors and are available for a wide variety of cytokines, rather than just one, as is the case with utilizing a receptor antagonist.
Read more on our Design page!
4. BUILD
ITERATION 1: We built a compartmental, mechanistic ODE-based model of our circuit using MATLAB©️’s SimBiology. Our model aimed to simulate the intracellular reaction dynamics of components of our system, utilizing both mass action and repressor Hill function equation parameters from established biological phenomena, as well as the context of our system--extracellular effects on systemic plasma. We started off translating our circuit directly into the SimBiology model.
ITERATION 2: Once we had our initial SimBiology model of our circuit, we wanted to test our output protein’s effectiveness on various cytokines. Thus, to gauge the effect of AND gate output on systemic plasma, we added an extracellular cytokine network map of 9 different cytokines.
Read more on our Model page!
5. TEST
We used our SimBiology model to classify the effectiveness of scFv on different cytokines in our cytokine network map. We observed plasma ELK-1 levels for each cytokine as it was targeted by scFv. By identifying the lowest peaks of ELK-1, we can identify which target for our scFv would diminish the effects of a cytokine storm during the most severe circumstances. These lowest peaks corresponded to IFN-γ and IL-1 as targets of scFv binding.
Another state we looked into was the steady-state levels of ELK-1. By lowering the steady state cytokine levels, our system is able to diminish the cytokine storm further. We once again identified IFN-γ and IL-1 as strong candidates for our scFv target, based on the steady state metric.
Read more on our Results page!
6. LEARN & IMPROVE
We had frequent meetings with a diverse cohort of experts in synthetic biology to get feedback on our engineered system.
We met with our faculty advisor, Professor Ron Weiss, weekly to discuss advances in our circuitry and SimBiology model. Through these meetings, we often looped back to step 1 of the engineering design cycle and tweaked our system every week. We also met with postdoctoral researchers and graduate students in the Weiss Lab, in particular Noreen Wauford (PERSIST system), Sebastian Palacios (SimBiology model), and Fabio Caliendo (multimodal regulation in synbio systems).
We also discussed implications of our project if it were to be put into clinical use, with MIT political science professor Ken Oye and bioethicist Francoise Baylis.
Read more on our Human Practices page!
We will describe some of the iterations of our engineering design process below.
1. RESEARCH
ITERATION 1: We began our work as a team primarily interested in COVID-19 and the lesser known immune system impacts of the disease resulting in symptoms such as the cytokine storm. We searched the literature for clinical data on elevated cytokine levels in COVID-19 patients, while taking into consideration that our anti-inflammatory actuator protein may need to target whatever cytokines we choose to sense. After doing extensive literature research on COVID-19 cytokine storms (CCS), we found that IP-10 and MCP-3 are the cytokines whose blood plasma levels most strongly correlate with CCS. These two cytokines have downstream signaling pathways that include the nuclear transcription factors (nTFs) ELK-1 and NF-κB, respectively.
ITERATION 2: After electing to sense IP-10 and MCP-3, we researched potential circuitry for processing elevated levels of these cytokines. This involved looking through existing systems that process cytokine inputs through an AND gate in other inflammatory disease states, and systems for post-transcriptional mRNA-level regulation.
ITERATION 3: After deciding on a PERSIST-based system as our AND gate, we had to solidify our choice of anti-inflammatory actuator protein. After some research, we found our options included receptor antagonists, decoy receptors, and single chain variable fragments (scFvs).
2. IMAGINE
ITERATION 1: At the start of our engineering design process, we imagined our engineered system would be able to sense the CSS through the levels of the biomarkers ELK-1 and NF-κB, process these inputs through an AND gate, and respond to the CSS and correlated hyperinflammation by releasing an anti-inflammatory output protein.
ITERATION 2: We imagined potentially having our circuit produce a graded, analog response to cytokine levels, as opposed to a sigmoidal, digital one. However, we did not find success in building analog circuitry for mammalian cells, as described in DESIGN ITERATION 2.
ITERATION 3: We imagined having an anti-inflammatory actuator protein that would be tunable and able to target different cytokines so that we could compare our circuit’s effectiveness in decreasing levels of various cytokines.
3. DESIGN
ITERATION 1: We first designed our cytokine sensing circuits. Since we would be sensing levels of the biomarkers ELK-1 and NF-κB, we decided to have two plasmids, utilizing the CMV promoter with a responsive element for either ELK-1 or NF-κB.
ITERATION 2: We then considered different options for processing our cytokine inputs: cascades, bacterial analog circuitry, and Programmable Endonucleolytic Scission-Induced Stability Tuning (PERSIST). When trying to apply the analog circuit design to mammalian cells, we ran into difficulties finding a system mechanistically analogous to the bacterial AraC protein that was used in the bacterial circuit. Our final circuit used PERSIST for post-transcriptional mRNA-level regulation.
ITERATION 3: While considering our options of anti-inflammatory actuator protein (receptor antagonists, decoy receptors, scFvs), we chose scFvs because they more specific and stable than decoy receptors and are available for a wide variety of cytokines, rather than just one, as is the case with utilizing a receptor antagonist.
4. BUILD
ITERATION 1: We built a compartmental, mechanistic ODE-based model of our circuit using MATLAB©️’s SimBiology. Our model aimed to simulate the intracellular reaction dynamics of components of our system, utilizing both mass action and repressor Hill function equation parameters from established biological phenomena, as well as the context of our system--extracellular effects on systemic plasma. We started off translating our circuit directly into the SimBiology model.
ITERATION 2: Once we had our initial SimBiology model of our circuit, we wanted to test our output protein’s effectiveness on various cytokines. Thus, to gauge the effect of AND gate output on systemic plasma, we added an extracellular cytokine network map of 9 different cytokines.
5. TEST
We used our SimBiology model to classify the effectiveness of scFv on different cytokines in our cytokine network map. We observed plasma ELK-1 levels for each cytokine as it was targeted by scFv. By identifying the lowest peaks of ELK-1, we can identify which target for our scFv would diminish the effects of a cytokine storm during the most severe circumstances. These lowest peaks corresponded to IFN-γ and IL-1 as targets of scFv binding.
Another state we looked into was the steady-state levels of ELK-1. By lowering the steady state cytokine levels, our system is able to diminish the cytokine storm further. We once again identified IFN-γ and IL-1 as strong candidates for our scFv target, based on the steady state metric.
6. LEARN & IMPROVE
We had frequent meetings with a diverse cohort of experts in synthetic biology to get feedback on our engineered system. We met with our faculty advisor, Professor Ron Weiss, weekly to discuss advances in our circuitry and SimBiology model. Through these meetings, we often looped back to step 1 of the engineering design cycle and tweaked our system every week. We also met with postdoctoral researchers and graduate students in the Weiss Lab, in particular Noreen Wauford (PERSIST system), Sebastian Palacios (SimBiology model), and Fabio Caliendo (multimodal regulation in synbio systems). We also discussed implications of our project if it were to be put into clinical use, with MIT political science professor Ken Oye and bioethicist Francoise Baylis.
This page was written by Erin Shin