back-ground
Engineering

Engineering

sam

Overview

We aimed to develop a method to discriminate dengue virus serotypes that outperforms existing measurement methods by actively implementing an engineering cycle (Design, Build, Test and Learn). Our engineering cycle consists of the following parts.

  1. Reproducing our previous project by iGEM Tokyo Tech 2018
  2. Validating new genetic circuits for enhancing the performance of diagnosis
  3. Preparing mosquito cells for screening pseudovirus
  4. Developing fluorometers toward high-throughput analysis
  5. Establishing prediction models of the serotype’s ratio and software toward integration with wet lab experiments

Figure 1. DBTL cyclesFigure 1. DBTL cycles

Cycle1: Reproducing our previous project by iGEM Tokyo Tech 2018

Figure 2. Cycle1: Reproducing our previous project by iGEM Tokyo Tech 2018Figure 2. Cycle1: Reproducing our previous project by iGEM Tokyo Tech 2018

Design
We conducted experiments to reproduce the results by iGEM Tokyo Tech 2018 team. We reproduced the production scheme of single-round infectious particles (SRIP) and used SRIPs containing different fluorescent proteins to identify four serotypes of dengue virus (Fig. 2). First, three plasmids encoding the SRIP component were transfected into HEK293T cells and SRIPs were collected. When SRIP interacted with infection-detecting cells, we can observe four types of fluorescence.

Build & Test
We investigated the infectivity of SRIP with Vero cells (specific cell line for testing viral infectivity) to test the performance of fluorescence-based serotype detection. We confirmed the appearance of fluorescence for serotypes 1 and 4. However, we had to consider the alternative design of genetic circuits to improve the performance.

Learn
After discussions with Dr. Ryosuke Suzuki (National Institute of Infectious Diseases), we came up with a new hypothesis that the sequence introduced into the SRIP was too long, which significantly reduced the efficiency of gene fragment internalization into the SRIP (Fig. 3).

Cycle2 : Validating new genetic circuits for enhancing the performance of diagnosis

Figure 3.Cycle2 : Validating new genetic circuits for enhancing the performance of diagnosisFigure 3.Cycle2 : Validating new genetic circuits for enhancing the performance of diagnosis

Design
Using the Split-Cre system, we aimed to solve the limitation caused by sequence length by shortening the sequences contained in the single-round infectious particles (SRIP). Furthermore, we developed new pseudovirus-detectable cells. These pseudovirus-detectable cells show two different results, depending on the presence or absence of antibodies to specific serotypes in the blood (Fig. 4).

  1. Under uninfected conditions (without antibodies) When SRIP is infected, the red fluorescent protein mCherry is expressed in addition to the green fluorescent protein (GFP).

  2. Under infected conditions (with antibodies) If there are neutralizing antibodies in the serum, SRIP is neutralized and cannot infect Vero cells, and only GFP is expressed.

Build & Test
We transfected genetic parts for green fluorescence into Vero cells to verify the function of pseudovirus-detectable cells. As a result, we confirmed the expression of green fluorescent protein in Vero cells.

Learn
We found that our gene sequence can be transfected into Vero cells. We also found that green fluorescent protein can be produced from the gene sequence in Vero cells. We also found in the literature that Dengue's SRIP may be more likely to make the protein in C6/36 cells (mosquito cells).

Cycle3: Preparing mosquito cells for screening pseudovirus

Figure 4.Cycle3: Preparing mosquito cells for screening pseudovirusFigure 4.Cycle3: Preparing mosquito cells for screening pseudovirus

Design
We optimized the gene sequence to prepare pseudovirus-detectable cells for mosquitoes.

Build & Test We transfected genetic parts for green fluorescence into C6/36 cells (mosquito cells) to verify the function of pseudovirus-detectable cells. As a result, we confirmed the expression of green fluorescent protein in C6/36 cells.

Learn We found that C6/36 also produces green fluorescent proteins with the gene sequence we designed.

Cycle4: Developing fluorometers toward high-throughput analysis

Figure 5.Cycle4: Developing fluorometers toward high-throughput analysisFigure 5.Cycle4: Developing fluorometers toward high-throughput analysis

Design Also, after discussions with researchers familiar with the endemic area, we realized that the system should be usable without expensive fluorescence microscopes or other equipment. To address the needs, we decided to create a fluorometer.

Build We designed a device that can determine the presence of fluorescent substances at low concentrations. We used a camera to read the fluorescence as an image and analyzed the image to determine the presence or absence of fluorescence.

Test
We performed a test in which we measured fluorescent substances at different concentrations. We could determine the presence of fluorescence at some concentration, but not at lower concentrations. However, we could not determine the presence of fluorescence at lower concentrations.

Learn
We thought that the reason we could not determine the presence of fluorescence at low concentrations was that the excitation light was noisy when measuring fluorescence at low concentrations. In other words, we found it necessary to suppress the noise between the excitation light and the external light in order to make a clear difference even at low concentrations.

Redesign After consulting with a fluorescence measurement device manufacturer, we aimed to reduce noise from the excitation light and outside light.

Rebuild We suppressed noise from excitation light and external light by redesigning the device to have black drawing paper attached to the inside of the device.

Retest
We tested the measurement of fluorescent substances of different concentrations. We could get a clearer image than with the previous device.

Learn We understood the importance of designing devices without any external factors.

Cycle5: Establishing prediction models of the serotype’s ratio and software toward integration with wet lab experiments

Figure 6.Cycle5: Establishing prediction models of the serotype’s ratio and software toward integration with wet lab experimentsFigure 6.Cycle5: Establishing prediction models of the serotype’s ratio and software toward integration with wet lab experiments
Design
In anticipation of the availability of more data through the construction of diagnostic tools, we have attempted to construct a model that can predict the epidemic for each serotype based on data on the number of infected patients per serotype.

Build & Test We used different methods (Lasso, Random front, Simple neural network and long short-term memory) to create a prediction model based on historical dengue virus serotype 1 to 4 case count data. We confirmed high prediction accuracy for serotypes 3 and 4 data and found that different models have strengths and weaknesses.

Learn We thought we should make these models accessible for everyone, so we decided to create software (described later). In addition, we realized that the accuracy of the model could be further improved by collecting data on a larger number of infected individuals. Therefore, we believe that the integration of this predictive model with the diagnostic tool developed in the wet lab will be essential in the future.

Design We designed a website that can run the fashion forecasting model to make it usable for people unfamiliar with programming.

Build & Test We created an easy-to-read and easy-to-use website for a trend forecasting system for the number of infection cases. This application has the following advantages (2) It is possible to view past data. (3) It can run forecasting models. (4) The results are displayed in graphs.

Learn To make this system easier to use, we believe it would be more effective in predicting dengue epidemics if it could notify infected areas of an outbreak.