Team:TAS Taipei/Applied Design




We envision delivering our treatment either as purified ALDH2*1 enzymes or ALDH2*1-expressing probiotics. We tested both of these delivery forms under realistic oral conditions by accounting for two factors: the viscosity of saliva and temperature. Our test results show that both purified enzymes and genetically engineered living probiotics (we used the probiotic E. coli Nissle 1917) are able to metabolize acetaldehyde significantly faster than their mutant ALDH2*2 counterparts under simulated conditions in artificial saliva at body temperature. Finally, we manufactured and tested our own probiotics candy, and found that the ALDH2 probiotic successfully metabolizes acetaldehyde!


Through our research, we found that we needed to target the upper digestive and respiratory tract, as ALDH2 deficiency mainly increases the risks of developing esophageal and head and neck cancers (Chao et al., 2000; Yang et al., 2007; Yokoyama, 2001; Matsuo et al., 2001; Cui et al., 2009; Lee et al., 2007; Wu et al., 2013; Huang et al., 2017; Hiraki et al., 2007). When we gathered public opinion through our survey, 83% of the public indicated that they preferred either probiotics or oral medication as the method of treatment. With this in mind, we decided to manufacture a candy (similar to a throat lozenge that remains in the mouth) to deliver recombinant ALDH2*1. The candy will contain either purified ALDH2*1 or probiotic strains that constitutively express ALDH2*1.


To test our products under realistic conditions of the human oral cavity, we considered two factors: temperature and viscosity of saliva. We used the incubator to simulate mouth temperatures, which are around 37°C, and we used artificial saliva to mimic conditions in the mouth. However, we visited multiple pharmacies and learned that vendors no longer distribute artificial saliva in Taiwan. We asked Dr. Hsiu-Ying Huang, who told us that there are other dental products which have the same ingredients as artificial saliva, such as cellulose, glycerin, and xylitol (Amal et al., 2017). After looking through different available products, we decided to make our own artificial saliva using Biotene Dry Mouth OralBalance Gel, which is a concentrated gel used to treat xerostomia, more commonly known as dry mouth (visit Biotene for product details).

To mimic saliva in the mouth, we needed to dilute the concentrated Biotene gel. We determined the amount of dilution by researching the viscosity of saliva: the mean viscosity of saliva at room temperature is 2.1 centipoise (cP), and the mean viscosity of water at room temperature is 0.95 cP (Kusy et al., 1995; Tang, 2016). Thus, we needed to dilute the gel so that the resulting solution would be about 2 times as viscous as water. Conducting a ball drop test to find determine viscosity (Figure 3-1; we adapted the ball drop procedure and viscosity calculations from Tang, 2016. See Protocols in our lab notebook for details), we found that diluting 6.5 grams of Biotene gel into 50 mL of water would very closely mimic the viscosity of saliva (Tang, 2016). The viscosity of this saliva model is 2.09 cP. All subsequent experiments were performed using this artificial saliva at 37°C.
Figure 3-1. Ball drop test to mimic the viscosity of saliva. A bead falls slower in the more viscous artificial saliva in comparison to water. (A) The bead falls at an average rate of 0.278 m/s in artificial saliva. (B) The bead falls at an average rate of 0.617 m/s in water. The procedure and viscosity equations were adapted from Tang, 2016. See Protocol in our notebook for more details. (Experiment and Video: Tim H, Justin W, Phillip W, Ben K)



We purified ALDH2*1 and tested its enzymatic activity by monitoring changes in NADH concentration, using the same method described in our Experiment page.This functional test compared the activity of purified ALDH2*1 and ALDH2*2 in artificial saliva at 37°C (Figure 3-2). Our results show that ALDH2*1 increased NADH levels significantly (by about 3.5 times) compared to ALDH2*2, indicating that our purified ALDH2*1 is able to metabolize acetaldehyde efficiently under simulated human oral conditions.
Figure 3-2. Purified ALDH2*1 has a higher activity level compared to purified ALDH2*2. Enzymatic activity of purified ALDH2*1 and ALDH2*2 was tested in artificial saliva at 37°C to simulate real-life conditions in a mouth. A negative control containing only elution buffer (from the protein purification process) was also included (gray). Under these conditions, ALDH2*1 was more efficient than previous tests using water at 25°C. ALDH2*1 also metabolized significantly more acetaldehyde compared to both ALDH2*2 and the negative control. The error bars represent standard error. (Experiment & Figure: Justin W)



Our survey results showed that 53% of the people preferred delivery of treatment using probiotics. Probiotics are bacteria that are beneficial to the health of the host (Haukioja, 2010). The ability for different probiotics to compete for cell adhesion in the oral cavity makes them an excellent fit for our ALDH2*1 delivery system (Haukioja, 2010; Haukioja et al., 2006; Samot et al., 2011; Stamatova et al., 2009). Originally, our visit to the Yakult factory inspired us to use Lactobacillus casei (L. casei), a common probiotic strain, as our chassis. However, after multiple attempts, we were unable to transform our ALDH2*1 expression plasmid into L. casei. After consulting Dr. Ying Chieh Tsai, a renowned probiotic expert in Asia, we learned that transforming L. casei was uncommon in Taiwan and difficult to do. He recommended that we try two other probiotics strains, Lactococcus lactis (L. lactis) and Escherichia coli Nissle 1917 (EcN).

Lactococcus lactis Transformation

L. lactis is a gram-positive lactic acid bacteria commonly used in dairy fermentation. It is labeled as “generally recognized as safe” (GRAS) by the FDA and is one of the most well-researched probiotic species (Song et al., 2017).
Figure 3-3. L. lactis imaged using a scanning electron microscope. L. lactis samples were prepared for SEM by fixation with glutaraldehyde. (Imaging & Figure: Catherine C)
We utilized the NICE® Expression System, provided to us by Professor Yeh from NCHU, for Lactococcus lactis transformation of our ALDH2*1-expression device (BBa_K2539100). We first cloned BBa_K2539100 into the NICE® pNZ8008 vector, then transformed the plasmid into the NICE® L. lactis Strain NZ9000 by electroporation. The L. lactis bacteria was made electrocompetent by treatment with sucrose and glycerol, then electroporated at 2500 V using a BTX ECM399 Electroporator, which was lent to us by NYMU iGEM team. The cells were immediately recovered in media containing sugars and salts (0.5 M sucrose, 20 mM MgCl2, 2 mM CaCl2) for two hours, and plated on M17 plates containing chloramphenicol. Sadly, we did not see any colonies on the transformation plates.
Figure 3-4. Electroporation of L. Lactis. A) L. lactis bacteria grown on an M17 plate. B) PCR check results after cloning the ALDH2*1-expression construct (BBa_K2539100) into the transport vector (pNZ8008) using primers complementary to ALDH2*1. The expected PCR product size is around 1.6 kb. C) Electroporation setup. (Cloning, Primer Design, Electroporation: Catherine C & Justin W)

Escherichia coli Nissle 1917 Transformation

Like L. lactis, EcN is another probiotic strain which is also “generally recognized as safe” (GRAS) by the FDA (Reister et al., 2014). EcN is one of the most frequently used gram-negative oral probiotic strains in research, and has been studied for over a century (Wassenaar, 2016). A recent New York Times article, “Scientists are Retooling Bacteria to Cure Disease,” documents how researchers have inserted genes into the DNA of EcN; they have successfully conducted human trials and demonstrated that people can safely tolerate the bacteria (Zimmer, 2018).
Figure 3-5. EcN imaged using a scanning electron microscope. E. coli Nissle 1917 samples were prepared for SEM by fixation with glutaraldehyde. (Imaging & Figure: Catherine C)
We obtained a sample of EcN (Mutaflor) from Pharma-Zentrale GmbH. We treated the cells with calcium chloride, then followed our standard E. coli protocol (link to lab notebook) to transform GFP-expression constructs into the competent cells (Figure 3-6). GFP was used so we could easily determine whether transformation was successful, and E. coli K-12 was included as a positive transformation control. Both E. coli K-12 and Nissle 1917 were treated with calcium chloride and chemically transformed with GFP (Figure 3-6). Although the efficiency was much lower, we observed GFP-expressing EcN colonies, indicating that transformation was successful!
Figure 3-6. GFP was successfully transformed into EcN. (Bottom) EcN colonies were glowing after transformation, showing that transformation of GFP was successful. (Top) Transformation of E. coli K-12 with GFP was performed as a control. (Experiment and Figure: Tim H, Catherine C, Leona T)
Knowing that transformation into EcN is possible, we tried to transform both ALDH2*1 and ALDH2*2-expression constructs (BBa_K2539100 and BBa_K2539200), and well as a GFP positive control (Figure 3-7). GFP was successfully transformed into EcN (glowing green colonies). ALDH2*1 and ALDH2*2-expression constructs were also successfully transformed; we observed colonies on the plates and confirmed the plasmids by PCR with the primers VF2 and VR (Figure 3-7).
Figure 3-7. Transformation of GFP, ALDH2*1, and ALDH2*2-expression constructs into EcN. (A) We transformed EcN with both ALDH2*1 and ALDH2*2-expression constructs (BBa_K2539100 and BBa_K2539200), and well as GFP as a positive control. Transformation was successful, as we observed glowing green colonies for the GFP sample, as well as colonies on the ALDH2*1 and ALDH2*2 plates (red arrowhead points to the one bacterial colony of ALDH2*2). Only bacteria successfully transformed with our chloramphenicol-resistant plasmids can grow. (B) PCR check confirmed the transformation of EcN with ALDH2*1 and ALDH2*2-expression constructs (BBa_K2539100 and BBa_K2539200). PCR was run using VF2 and VR primers on colonies picked from the transformation plates. We observed bands at all of the expected sizes: BBa_K2539100 and BBa_K2539200 is 2.1 kb, and the GFP is 1.1 kb. (Cloning and Figure: Tim H, Catherine C, Leona T)

Testing Engineered ALDH2 E. coli Nissle 1917

We tested the ALDH2-expressing EcN in simulated oral conditions: with artificial saliva at body temperature. Using the Megazyme kit (as described in Experiments), we tested the ability of ALDH2*1 and ALDH2*2-expressing EcN to metabolize acetaldehyde by monitoring changes in NADH concentration. Bacterial cultures were grown and lysed for the test to prevent intact cells from interfering with spectrophotometer readings at 340 nm.
Figure 3-8. EcN expressing ALDH2*1 converts acetaldehyde faster than ALDH2*2. To simulate oral conditions, this test was run in artificial saliva at 37°C. The graph shows relative activity of lysates containing either ALDH2*1, ALDH2*2, or inactive ALDH2*1 (boiled to denature proteins; negative control). Over a 40-minute period, EcN carrying our ALDH2*1 construct (BBa_K2539100) produced more NADH than the mutant form (BBa_K2539200), while the negative control (boiled BBa_K2539100) did not change significantly. Error bars represent standard error. This experiment shows that our probiotics effectively metabolize acetaldehyde under realistic conditions. (Experiment & Figure: Justin W)
Our results show that ALDH2*1-expressing EcN metabolizes acetaldehyde in simulated oral conditions. Over a 40 minute period, the absorbance at 340 nm increased faster for ALDH2*1 compared to ALDH2*2, but did not significantly change for the negative control (inactive ALDH2*1) (Figure 3-8).


When we surveyed the public about possible treatment methods, 83% of the people indicated that they preferred treatment using either probiotics or oral medication. Since ALDH2 deficiency mainly increases the risks of developing esophageal and head and neck cancers, it made sense to use a candy (similar to a throat lozenge that remains in the mouth) to deliver recombinant ALDH2*1. The candy will deliver ALDH2*1 either as a purified enzyme or expressed by the probiotic strain EcN. Here, we document the production and testing of a probiotic candy.

Determining Temperature Threshold of EcN

The candy-making process consists of three stages: dissolving, cooling, and molding (Husband, 2014). We want to add probiotics to the candy during the cooling stage to make sure they remain alive. To determine the exact temperature at which our engineered probiotics should be added, we tested bacterial growth at different temperatures.

10 mL of our probiotic culture was pelleted and resuspended in 10 mL distilled water. The test was carried out by heating this probiotic mixture from 25°C to 80°C, and plating a set volume of bacteria at 5°C intervals (Figure 3-9). In addition to testing ALDH2-expressing EcN, we also used GFP-expressing EcN as a positive control to show that proteins are still functional. Only bacteria exposed to temperatures at 60°C and below grew and carried functional proteins (Figure 3-9). From these results, we conclude that ALDH2*1-expressing EcN should be added to the candy below 60°C.
Figure 3-9. EcN grows at temperatures at or below 60°C. (A) Experimental setup. EcN carrying different constructs were heated in a beaker of water and plated at 5°C increments. B) ALDH2*1 expressing probiotics grow at or below 60°C. (red arrowhead points to the one bacterial colony at 60°C). (C) GFP expressing probiotics grow at temperatures below 60°C. (D) GFP expressing probiotics glow under UV, indicating that proteins are functional. (Test & Figure: Leona T, Catherine C)

Production of Candy

To make our very own engineered EcN candy, we followed and modified a recipe from Food Network on making gummy bears. (Food Network, 2017) First, we heated 30mL of sugar and 30 mL of gelatin in a 120 mL of water. After all the ingredients fully dissolved, we pipetted 3 mL of the sugar mixture into ice cube trays and let it cool. The ALDH2-expressing EcN were grown, pelleted, and resuspended in 50 µL of water. When the sugar mixtures cooled down to 55ºC, we added in our probiotics and waited for the candy to solidify over a period of three days (Figure 3-10). When we made candy with GFP-expressing EcN, the candy glowed! We successfully incorporated living probiotics, carrying functional proteins, into the candy.
Figure 3-10. Making our EcN probiotic candy. (A) We dissolved sugar in a solution containing water, gelatin, and corn syrup. (B) Probiotic candy mixture cooling in an ice cube tray. 3 mL of candy mixture was pipetted into each well. (C) The GFP-expressing probiotic candy glows under blue light, whereas the candy without GFP-expressing EcN did not. (Test & Figure: Leona T, Catherine C)

Product Testing

Following the same procedure, we made some probiotic candies carrying ALDH2*1-expressing EcN. We wanted to test whether this probiotic candy would be effective under realistic conditions by answering two questions:
1) How long does it take for the candy to dissolve in saliva?

2) Can the probiotic candy successfully metabolize acetaldehyde?
To answer the first question, we measured the amount of time needed for the probiotic candy to dissolve in artificial saliva at 37°C. Ideally, we do not want the candy to dissolve too slowly, as that would mean that the ALDH2*1-carrying probiotics are not released into the mouth fast enough. On the other hand, we also do not want the candy to dissolve too quickly, because that might lead to most of the probiotics being swallowed right away.

We wanted our candy to dissolve at roughly the same rate real cough drops do, so we compared our candy with two commercial products, Ricola and Nin Jiom herbal candies. The candies were placed in a beaker of artificial saliva on a heat plate set at 37°C, with a stir bar to mimic saliva movement (Figure 3-12). Our probiotic candy dissolved after 60 minutes, whereas it took 54 minutes and 63 minutes for Nin Jiom and Ricola, respectively (Figure 3-11). In the lab setup, Nin Jiom dissolved after 54 minutes, but when we ate the Nin Jiom candies (different package, outside of the lab), they dissolved after an average of 13 minutes (± 2 minutes standard error, n= 6 team members). Extrapolating the data from above, we speculate that if people were to eat our probiotic candy, it should completely dissolve after about 15 minutes.
Figure 3-11. Ricola and Nin Jiom are commercial cough drops (hard candies).
Figure 3-12. Dissolution of our probiotic candy in artificial saliva at 37°C. Our probiotic candy (2.5 x 2 x 0.5 cm in dimension) was placed in a beaker of artificial saliva on a heat plate set at 37 °C. A stir bar was used to simulate saliva movement in the mouth. Under these simulated conditions, the hard candy completely dissolved after 60 minutes.
To answer the second question, we isolated ALDH2*1-expressing EcN from the dissolved probiotic candy and tested its ability to metabolize acetaldehyde. We ran a functional test (the procedure is described above and on our Experiments page) by measuring NADH concentrations (Figure 3-12). In this experiment we compared three different candies, containing either ALDH2*1-EcN probiotics, ALDH2*2-EcN probiotics, or no probiotics. The candies were first dissolved like they would be in the mouth, then the samples were centrifuged to isolate the probiotics. After decanting the supernatant, we lysed the cells and tested the cell lysates under simulated oral conditions (in artificial saliva at 37°C). Our results show that EcN expressing ALDH2*1 in the probiotic candy is functional and effective at metabolizing acetaldehyde (Figure 3-12).
Figure 3-12. Our candy carrying ALDH2*1-EcN converts acetaldehyde faster than its ALDH2*2 counterpart. To simulate realistic conditions, the candy was first dissolved (like it would in the mouth) before the EcN cell extracts were tested. A candy that did not contain any bacteria served as a negative control. The results show that our probiotic candy is effective under realistic conditions and can successfully metabolize acetaldehyde. Error bars represent standard error. (Experiment & Figure: Justin W)

Information for Manufacturers

Next, we used mathematical modeling to model how fast acetaldehyde is being made and metabolized in the mouth after alcohol consumption. With this information, we determined how much additional ALDH2*1 a deficient individual would need to process acetaldehyde at wild type rates. We determined that 1.839*10-9 mol of ALDH2*1 enzymes or an equivalent amount inside ALDH2*1-EcN probiotics (2.67 mg) would be packaged into one candy, which can be used for 15 minutes. While eating the candy, a consumer would be able to metabolize acetaldehyde at wild type rates.



Saito M. (2007). Role of FOSHU (food for specified health uses) for healthier life.
Yakugaku Zasshi. 127(3):407-16.

Yakult. (n.d.). Our Science. Retrieved from