Inflammatory Bowel Disease is an autoimmune disease that results in inflammation of the colon and small intestine. IBD is a huge problem in developed countries, vastly growing with no effective therapy in sight. This disease is caused by an imbalance of immune T-cells, Th17 and T regulatory, which control inflammation and immunosuppression responses respectively. We propose a novel mechanism using synthetic biology that aims to regulate this balance in vivo. This involves a two-part device: one that regulates inflammation and the other immunosuppression. We cloned a device that recognizes superoxide producers, a byproduct of inflammation and Th17 proliferation, to produce retinoic acid, which blocks the further differentiation of Th17 cells. This device has been characterized, is well behaved and capable of recognizing superoxides, like paraquat, at optimum levels of 40 to 80 uM without growth inhibition. To control immunosuppression, we cloned and mutated the trp operon so that it recognizes 5-Methyl tryptophan, a target substrate of an enzyme involved in Treg immunosuppression, and designed a potential IL-6 excretion device to regulate Treg proliferation. The behavior of this novel tryptophan sensor has been studied by our group through the development of a mathematical model. Our results suggest that our construct has potential applications as a diagnostic or therapeutic tool for post-operative Crohn's patients.
Background and Clinical Significance
Inflammatory Bowel Disease is a group of inflammatory diseases that affect the colon and small intestine. The disease comes in two varieties, either as Ulcerative Colitis (UC) or Crohn's Disease (CD). An autoimmune condition, IBD is becoming particularly prevalent in the industrialized nations of the world. Already, in the United States and Europe, the disease group affects over 4 million individuals. With a faster growth rate and higher incidence than other well known autoimmune diseases like multiple sclerosis and asthma, IBD will affect an even larger percentage of the world's population, particularly in industrializing nations like India and China, over the next thirty years. While both diseases, CD and UC, affect similar demographics, they tend to have very different symptoms that requires differential treatment methods. In Ulcerative Colitis the inflammation is usually continuous and can stretch from the colon to, in many cases, the rectum. The inflammation results in a continuous stretch of ulcers in these regions but very rarely in fistulas and abnormal passageways, a characteristic of Crohn's disease. Because of this shallow and continuous inflammation, Ulcerative Colitis can often be treated without remission by removing the affected areas of the colon or rectum.
In contrast to Ulcerative Colitis, Crohn's Disease often includes the terminal ileum, the most distal part of the small intestine, and the colon. Unlike other forms of IBD, CD also results in patchy areas of inflammation with the localization of the disease "jumping" from distal portions of the small intestine to the colon. The localized bands of inflammation are often accentuated by the presence of fistula, granulomata and deep, transmural inflammation. As a result of the nature and localization of the inflammation, Crohn's Disease is usually hard to treat through surgical resection procedures that remove affected areas of the colon. In post-operative Crohn's patients, the site of resection often results in proximal inflammation that can travel distally, resulting in a reemergence of patchy inflammation in the colon.
As a result post-operative Crohn's patients are particularly appealing demographic for whom to design treatment options since current resection methods designed for UC seem to be ineffective.
Numerous clinical studies with IBD patients have suggested that the disease is an outgrowth of cellular imbalance, namely an imbalance in two different T cell populations - Th17 and Treg. Thought to have an inverse relationship, increased naive CD4 differentiation to Th17 and associated inflammation is hypothesized to decrease naive CD4 differentiation to Treg and associated immunosuppression. In the case of IBD, experimental work suggest that Th17 strains begin to cause inflammatory damage to the microenvironment of the colon by (1) proliferating unexpectedly and (2) becoming self-recognizing. Thus, a device designed to ameliorate the effects of recurring Crohn's disease would have to curb the proliferation of Th17 cell types. Simultaneously, this device would have to increase the immunosuppressive response in the colon to decrease the damage caused by the self-recognizing Th17 cells.
Given the project's context in inflammatory bowel disease (IBD), Stanford iGEM strove to develop a therapy that addresses the weaknesses of current treatments for IBD. Taking a bioengineering approach that could avoid the drawbacks of medications and circumvent the risks inherent to surgical procedures, the team tackled IBD at its source by designing a microbial device that is capable of monitoring and regulating imbalances in ratio of T helper 17 (Th17) cells to regulatory T cells (Treg) cells. With this project as a model, the team hoped to prototype a probiotic capable of autonomously determining loss of homeostasis and exerting measures to restore balance in helper T-cell ratios.
Markers involved in Device Regulation
Because the envisioned microbial construct is a biological input/output device, it must have the appropriate sensors to detect Th17 and Treg cell balance and the appropriate generators to output factors to restore balance. First, the team identified two markers of Th17 and Treg activity: nitric oxide (NO) and 5-methyltryptophan (5MT) concentration levels. These two markers serve as the inputs to the device. In other words, the device senses concentration levels of these two markers to determine the balance between Th17 and Treg cell populations. If the inputs indicate an imbalance, the device will produce factors to restore balance. The team next identified two factors, retinoic acid (RA) and Interleukin–6 (IL-6), that can control production of Th17 and Treg cells in vivo. Thus the device is engineered to produce heterologous RA and IL-6 in response to signals from the device sensor indicating imbalanced Th17 and Treg populations.
The team chose Escherichia coli to use as the chassis for the sensors and generators. The advantages of using E. coli are that the species is well characterized, tractable for bioengineering, and perhaps most importantly, native to the human gastrointestinal microbiota (depending on strain). This last point ultimately enables our device to be assimilated into an IBD patient’s digestive tract, in close proximity to possible sites of disease. Nevertheless, an E. coli chassis is not perfect: it is difficult to construct an E. coli device that can simultaneously sense two different inputs and dynamically output two different molecules.
Inflammation and Immunosuppression Device specifics
Thus the team divided the two sensors and generators between two E. coli devices: Device 1 detects excessive Th17 inflammatory activity (through NO) and secretes an anti-inflammatory factor (RA) in response, while Device 2 detects excessive Treg-mediated immunosuppression (through 5MT) and produces an inflammatory factor (IL-6) in response. The primary role of Device 1 is to treat IBD by reducing excessive Th17 inflammatory activity. Device 1 can accomplish this by secreting RA, which inhibits Th17 cell production from naïve CD4+ T cells while at the same time boosting Treg cell production from the same population of naïve CD4+ cells. Device 2 has the opposite role; it produces and exports IL-6, which inhibits Treg cell production from naïve CD4+ cells. Device 2 serves two roles: (1) to autonomously regulate Treg populations and immunosuppressive response and (2) to prevent Device 1 from overshooting and creating an immunosuppressive state and an increased risk of colon cancer.
Alternative Sensor and Generator Arrangements
Although the team considered allocating the sensors to one device and the generators to a second device as an alternate way to divide the input and output tasks, such a scheme would require, in addition to the I/O system outlined above, a secondary I/O system that would allow the sensor device to communicate to the generator device. Moreover, such as secondary I/O system would have to be extremely robust, since the gastrointestinal extracellular environment is subject to great amounts of molecular noise. Finally, the devices in this scheme would have to untangle possible crosstalk between the two sensors and two generators. Ultimately, the team decided this division of labor was inferior to the one detailed above.
System Level Overview
Below is a system level overview of our IBD therapeutic that shows how device 1 and device 2 interact with one another and with inflammation and immunosuppressive responses. The functions of Device 1 are highlighted in green while the effects of device 2 are highlighted in purple:
Below is a link to an animation of our device function:
Device 1: Anti-Inflammatory Device
The team chose nitric oxide (NO) for the in vivo inflammation signal for two primary reasons. First, NO levels closely correspond to inflammation. In Crohn’s Disease and IBD patients NO levels are elevated up to 1000-fold. Moreover, Th17 cells induce nitric oxide production by secreting Interleukin-17 (IL-17), which stimulates inducible nitric oxide synthase (iNOS) in nearby chondrocytes. Thus, nitric oxide can serve as a clear signal of IBD conditions and surfeit Th17 activity.
Second, E. coli naturally senses NO through an endogenous superoxide stress system, making this stress system a prime candidate for bioengineering. Through kinetic and thermodynamic studies this system is predicted to have nanomolar sensitivity to NO concentrations. At the heart of this highly tuned superoxide sensor are the SoxR gene and the SoxS promoter. The SoxR gene encodes the SoxR transcription factor, which is activated by NO. In the presence of NO, SoxR induces transcription from the SoxS promoter. Given its exquisite sensitivity and robust transcriptional response, the team selected SoxR/SoxS to be the Device 1 sensor.
With the sensor identified, the team next designed the Device 1 generator. Recall that the generator must produce a factor capable of suppressing IBD inflammation. The literature provided several candidate immunosuppressives, but the team chose retinoic acid for several reasons: Recently, retinoic acid has moved into the center of inflammation research for its ability to shift differentiation of naïve CD4+ cells away from the Th17 lineage in favor of the immunosuppressive Treg lineage. In this way retinoic acid may be key to reducing damage from inflammation in autoimmune diseases such as IBD. Furthermore, retinoic acid is a small hydrophobic molecule, meaning that it can diffuse freely through cell membranes. This reduces the possibility that the output retinoic acid will become trapped inside Device 1 as inclusion bodies. Finally the biochemical production of retinoic acid begins with acetyl Coenzyme A (acetyl CoA), the ubiquitous metabolic unit of energy common to almost all life.
The generator of Device therefore consists of the retinoic acid synthesis machinery linked to the SoxR/SoxS sensor. Following sensor activation, synthesis of retinoic acid proceeds through two stages and the constituent genetic parts are located on two plasmids. The first stage involves beta carotene synthesis from acetyl CoA derivatives and is mediated by the crt-gene cluster. The crt cluster (BBa_K125005) is controlled by a constitutive promoter; this construct is located in its own plasmid. The second stage of retinoic acid synthesis involves cleavage of beta carotene and subsequent oxidation to retinoic acid. A second plasmid incorporates the two genes required for this conversion into the plasmid with the SoxR/SoxS sensor.
Overall, in an inflammatory IBD environment, elevated NO levels signal through the SoxR/SoxS sensor to induce production of retinoic acid from beta carotene, which diffuses out of Device 1 and modulates CD4+ cell differentiation toward Tregs, which are the primary immunosuppressive cells of the body.
Device 2: Anti-Immunosuppressive Device
Device 2 is the anti-immunosuppressive device: it functions by recognizing an immunosuppressive signal (5-methyltryptophan) and subsequently producing an inflammatory factor (Interleukin-6) that can activate the immune system. Device 2 is intended to regulate Device 1 so that it does not inadvertently bring the body into an immunosuppressive state vulnerable to opportunistic infection and cancer.
Device 2’s sensor must be able to detect Treg activity in a concentration dependent manner. Although Treg cells have distinct secretion profiles, the team could not identify any Treg markers to which E. coli responds directly. Instead, the team discovered that Tregs recruit antigen-presenting cells (APC) to express the enzyme indoleamine-2,3-dioxygenase (IDO), which oxidizes and degrades tryptophan. Since the Device 2 sensor’s signal is inversely related to target cell activity – reduced tryptophan levels indicate increased Treg activity and vice versa – the Device 2 sensor must act as an inverter.
Fortunately, there exists such an inverter mechanism in E. coli. The tryptophan regulon native to E. coli consists of an inducible repressor (TrpR) and the tryptophan operon (Trp operon). The presence of the Treg signal, tryptophan, induces the TrpR to dimerize. Up to three of these repressor dimers can bind to the Trp operon and prevent transcription. This system is an inverter because low tryptophan input results in high output, while high input results in low output.
Nevertheless, the caveat to using tryptophan as the Treg marker is that tryptophan is a ubitquitous amino acid common to many metabolic pathways. Moreover its concentration in the digestive tract is highly variable depending on diet. Hence a tryptophan input signal would suffer from destructive amounts of noise. As such, the team identified a synthetic tryptophan analog, 5-methyltryptophan (5MT), that is also degraded by IDO. It is neither consumed by any known human or E. coli metabolic pathway nor produced naturally. 5MT is also very weakly absorbed in the digestive tract. Hence 5MT serves as a stable signal of Treg activity. The team developed mutant TrpR repressors that exhibit higher affinity for 5MT than tryptophan and mutant Trp operons that are bound almost exclusively by the mutant TrpR repressors in order to eliminate possible crosstalk between Device 2’s sensor and the endogenous tryptophan system.
Linked to the 5MT sensor is the Device 2 generator, which produces and secretes recombinant human Interleukin-6 (hIL-6). As with Device 1, the genetic parts to the generator were split into two plasmids. One plasmid contains the hemolysin (hly) gene cluster and the tolC gene. The hly and tolC parts encode the hemolysin secretion system protein that allow transport outside the E. coli. The other plasmid consists of the 5MT sensor linked to hIL-6 tagged with the hemolysin signal sequence. This signal sequence tags hIL-6 to be transported out of the E. coli by the hemolysin secretion system.
Thus, in an immunosuppressive context, excess Tregs due to dysregulated Device 1 activity results in reduced 5MT. This signals Device 2 to produce and secrete hIL-6, which polarizes CD4+ cells away from the Treg fate.
Results and Analysis
This summer, the team built three of the four device subparts, namely the SoxR/SoxS sensor, the retinoic acid generator and the 5-methyl tryptophan mutant trp sensor. All the sensors were built with GFP downstream of the sensor itself. As a result, only the sensing capability, that is the ability of sensor to recognize inflammation or immunosuppressive signal molecules, was tested. Of these three device subparts, the SoxR/SoxS sensor was characterized on both a high and low copy plasmid and found to be functional. In the section below are the graphs relating OD to time and GFP to time for the sensor on both the high and low copy vector strains.
Note on Assays
In order to characterize our device, we needed to use an inducer which imposed superoxide stress within the cell. In vivo, this stress would take the form of nitric oxide, a downstream product related to increasing levels of Th-17 cells. However, in vitro, we needed an inducer which was more suitable for us to test. Thus, we used paraquat, which causes oxidative stress in a similar manner as nitric oxide.
Low Copy SoxR/SoxS Sensor
Experimentation done on the low copy vector indicated that our SoxR/SoxS sensor was not only capable of inducible GFP (or another downstream protein) production but also that the downstream production levels can be modulated through varying concentrations of inducer paraquat. In this case, our group demonstrated that increasing the concentration of superoxide inducer paraquat increased the production of GFP as expected. Additionally, it was noted that at higher concentrations of the inducer, the rate of change of fluorescence increased as well.
However, it is interesting to note that the induction chemical paraquat, which can be toxic to E. coli cells at large concentrations, begins to have a negative effect on the cells at higher concentrations. Specifically, it seems that at 80uM concentration of paraquat, there is a marked decrease in growth rate with time. This is shown by the OD levels, which seem to begin stabilizing around an OD600 of 1.1 three hours after induction.
While growth inhibition seems to occur at 60 and 80 uM concentrations of paraquat, the overall stability of the low copy system and its ability to sense a large range of inducer concentrations over a long period of time suggests that it is ideal sensor system for our anti-inflammatory device. In particular, by virtue of the low copy system's ability to induce GFP production over a longer time and larger range range of concentrations, we would like to use it in the diagnosis and treatment of stable post-operative Crohn's patients who need long-term, stable retinoic acid therapy.
High Copy SoxR/SoxS Sensor
The SoxR/SoxS sensor was similarly tested on a high copy vector. Both of these assays lead to our characterization of this system with respect to its production of downstream GFP through induction by superoxide stress (in this case paraquat). As shown by the graphs, our results from the high copy varied sufficiently from the low copy vector. In the case of the high copy, one should first note that there is a greater growth inhibition occurring at lower inducer concentrations. As compared to the low copy in which growth inhibition seems to occur at 60 and 80 uM, the high copy showed growth inhibition occurring around 30uM. The second factor to take note of is the greater standard deviation occurring in the OD levels. Both of these factors are most likely due to the nature of the high copy vector itself. Since the cells are more metabolically active and have a higher rate division, they are under greater stress. When coupled to the stress of the inducer, this leads to larger growth inhibition and discrepancies in the OD with varying inducer concentrations.
The differences between the high and low copy results allow us to tailor our device to meet individual patient needs. The high copy vector will have the greatest application in treating high risk, post-operative patients of Crohn's Disease. In this stage of the disease, the patient's will require a burst of high dosage retinoic acid localized to a specific section of the small intestine. This would result in a short-term, more intensive treatment. However, this system is not sustainable; eventually the treatment would switch to the low copy vector as more of a monitor device.
1. Would any of your project ideas raise safety issues in terms of:
a. researcher safety, b. public safety, or c. environmental safety?
Answer: There are no risks regarding our general system design other than the use of probiotics in humans. A few concerns regarding probiotics have been noted from previous iGEM competitors such as NYMU Taipei (2008) and CalTech (2008).
2. Is there a local biosafety group, committee, or review board at your institution?
Answer: There is a local board at Stanford University called Health and Safety at the Stanford University School of Medicine.
3. What does your local biosafety group think about your project?
Answer: We spoke with David Silberman who is the Director of Health and Safety at the Stanford Univeristy School of Medicine about using the HlyA signal sequence alone for the export of non-hazardous protein, IL-6. Dr. Silberman considered it safe to use the signal sequence as long as the pathogenic HlyA protein was not used. The Health Safety board at Stanford University found all other parts nonhazardous.
4. Do any of the new BioBrick parts that you made this year raise any safety issues?
Answer: For our second device, our export system was taken from pathogenic bacteria which uses HlyAs, BBa_K223054, to export hazardous material. We use the same proteins, the Hly gene cluster B, C, D, R and As involved in the export, however we do not include the hazardous gene, HlyA, to export IL-6.
Mucida, D., Park, Y., Kim, G., Turovskaya, O., Scott, I., Kornenberg, M., and Cheroutre, H.. 2007. Reciprocal Th17 and regulatory T cell differentiation mediated by retinoic acid. Science.317(5835): 256-60.
Xiao, S., Jin, H., Korn, T., Liu, S.M., Oukka, M., Lim, B., and Kuchroo, V.K. 2008. Retinoic acid increases Foxp3+ regulatory T cells and inhibits development of Th17 cells by enhancing TGF-beta-driven Smad3 signaling and inhibiting IL-6 and IL-23 receptor expression. J Immunol. 181(4):2277-84.
Bai, A., Lu, N., Guo, Y., Liu, Z, Chen, J., and Peng, Z. 2009. All-trans retinoic acid down-regulates inflammatory responses by shifting the Treg/Th17 profile in human ulcerative and murine colitis. J Leukoc. Biol. 86 (4): 959-69.
Nitric Oxide and IBD
Kolios, G., Valatas, V., and Ward, S.G. 2004. Nitric oxide in inflammatory bowel disease: a universal messenger in an unsolved puzzle. Immunology. 113(4): 427-37.
Herulf, M., Ljung, T., Hellstrom, P.M., Weitzberg, E., and Lundberg, J.O. 1998. Increased luminal nitric oxide in inflammatory bowel disease as shown with a novel minimally invasive method. Scand J Gastroenterol. 33(2): 164-9.
Th17 and Tregs in IBD and the GI Mucosa
Maul, J., Loddenkemper, C., Mundt, P., Berg, E., Giese, T., Stallmach, A., Zeitz, M., and Duchmann, R. 2005. Peripheral and intestinal regulatory CD4+ CD25(high) T cells in inflammatory bowel disease. Gastroenterology. 128(7): 1868-78.
Yu, Q.T., Saruta, M., Avanesyan, A., Fleshner, P.R., Banham, A.H., and Papadakis, K.A. 2007. Expression and functional characterization of FOXP3+ CD4+ regulatory T cells in ulcerative colitis. 13(2): 191-9.
Ivanov II, Frutos Rde, L., Manel, N., Yoshinaga, K., Rifkin, D.B., Sartor, R.B., Finlay, B.B., and Littman, D.R. 2008. Specific microbiota direct the differentiation of IL-17-producing T-helper cells in the mucosa of the small intestine. Cell Host Microbe. 4(4): 337-49.