Hardware

Summary of Hardware

As described throughout our wiki, we felt that it was essential to test our constructs not only in silico and in vitro, but also in environments that were similar to those in which the devices would ultimately be deployed. Therefore, in order to study how our constructs behave in realistic conditions, we designed and built models that mimic real-world environments. Our primary hardware device is a human colon simulator, mimicking physiological conditions of the colon with a three-stage continuous culture model. We also built soil microcosms that incorporated “rain” and vegetation to provide an additional realistic environment in which we test our phage satellites. 

Many commonly-used colon models require expensive materials and specialized equipment, presenting barriers to replication across labs. Our model’s design is inspired by these relevant, validated gastrointestinal simulators, but makes significant improvements toward cost-accessibility and ease of replication. Built from inexpensive materials, the design, assembly, and operation of our colon is fully documented to allow future reproduction and use. The full hardware device comprises three models: two three-state colon models and one single-stage colon model. The three-stage colon model is designed with three consecutive compartments which move a defined media, formulated to mimic the nutritional contents and pancreatic juices present in the colon, through the system via peristaltic pumps on a fixed schedule to mimic feeding and digestion through the colon. The model is housed in an aquarium tank, which uses aquarium heaters and wavemakers to uniformly heat the system to body temperature and mix the contents of each compartment. 

Our soil microcosms are similarly designed to balance realism and replicability, with a low-cost, fully documented design. Despite the less technically complex design, the microcosms incorporate simulated rainfall, a large-scale spatial design, plant interactions, and nondestructive sampling. While these features individually have been incorporated into soil microcosm experiments, the combination of them with their experimental utilization studying bacteria, phage, and phage satellites, presents an unexplored microcosm framework. 

The two synthetic models provided our project with platforms to evaluate our constructs in representative natural ecosystems. The colon is a cost-effective, easily-reproducible simulator of the complex human colon environment, which advances the accessibility and replicability of colon models across labs. The microcosms are similarly replicable, offering an additional model environment with a combination of rainfall simulation, spatial design, and more. Together, these practical models enable future investigators to implement realistic tests in simulated colon and soil environments into their research, bringing advancements in synthetic biology closer to deployment in the real world. 

Criteria for Effective Hardware

Addressing a Need in Synthetic Biology

Our devices address a profound need in synthetic biology, namely for devices that mimic real-world environments, allowing for testing of synthetic biology constructs in realistic environments, whether that be in simulated in vivo or environmental conditions. Equally importantly, these devices need to be accessible, inexpensive, effective, and easy to build and use by anyone working in the field. Moreover, they need to be sufficiently inviting so as to encourage wet lab practitioners to test their constructs in realistic environments following in vitro tests. 

Conducting User Testing and Learning from User Feedback

The design and construction of our models were significantly informed by user testing and feedback from our wet lab team. As our models are meant to be used by synthetic biology practitioners, we conducted used testing with our wet lab team over the course of our project. Throughout the experiments involving the hardware models, our wet lab team, the primary users of our models, offered input and suggestions to improve our models and ensure ease of use. One such suggestion was labeling tubing, pumps, and effluent containers to delineate the components of each of the three models within our device, making the setup of the model clearer, and the model’s operation and troubleshooting much easier for users. Another aspect of our colon model which was derived from wet lab input was the on and off mechanism. When our wet lab team struggled to start and end runs, and pause them when needed, they encountered struggles with writing and uploading code for the Arduino. To address this, we first created annotated code for them to understand how to write and upload code to the microcontroller, and then opted to use the power supply as a simple on and off switch. Once an Arduino program has been uploaded, it runs until told to stop or removed from a power source, and begins as soon as it is connected to a power source; this feature of the Arduino enabled us to turn the colon model on by giving power to the Arduino, and turning it off by removing power. This results in a single connection toggling the model on and off, streamlining its operation. This input, along with other suggestions, helped us develop our model to be user-friendly for future operators.

Demonstrating Utility and Functionality

Our hardware models were built for the purpose of testing our project’s engineered constructs in model real-world environments. This portion of our project justifies the ‘R’ in SaPh:IRES, as an integral aspect of our project is the evaluation of our phage satellite systems in these realistic models. We conducted multiple full-length runs of our human colon model to evaluate the performance of several of our constructs in the colon environment. Our microcosms were similarly utilized to study our constructs, along with phage, in a soil environment. Experiments with both of these models produced intriguing results that greatly informed the conclusions and future directions of our project. For more information, see our experiments and results pages.

Documenting Hardware Sufficiently to Enable Reproduction

The team has documentation for both the colon and soil models that includes: a list of materials; detailed assembly and use instructions; Arduino code; experimental design; and a 3D model of the colon simulator. We also provide a writeup of considerations to improve the models’ utility and functionality (automated pH control, alternatives for inoculum, etc) as well as how other users can modify the hardware for their own specific uses.

Introduction: The Problem and Our Solution

Synthetic biology has the potential to deliver solutions to the most pressing and critical issues facing the world, from climate change and pollution to hunger to emerging diseases. However, advances in synthetic biology are hindered by challenges that synthetic constructs encounter outside the lab, impeding investigators’ ability to develop and deploy functionally effective solutions into the real world, where they ultimately need to work. 

The vast majority of synthetic biology experiments occur in highly controlled laboratory environments, where engineered biological constructs are privileged with all of the nutrients and optimal conditions they need to grow and thrive. While these experiments are fundamental to early-stage development of synthetic constructs, enabling researchers to fine-tune genetic components with precision to make rapid progress toward functional systems, they paint an incomplete picture of how these constructs will perform in a naturally-occuring environment. Natural ecosystems are dynamic and unpredictable, with fluctuating resources and environmental stresses, as well as interactions between countless organisms. Efficacy in test tubes or petri dishes does not guarantee synthetic circuitry is prepared for the instability and harsh conditions of real-world environments. This presents a critical disparity between success in the laboratory and success in the natural world, inhibiting synthetic biology from transcending the confines of the lab and tackling global challenges in diverse, heterogeneous environments. 

William & Mary 2024 believes it is essential for investigators to test their engineered products in simulated realistic environments to create an intermediate step between the controlled lab environment and uncontrollable real-world environment. Deployment in simulated natural environments reveals how constructs interact with the surrounding ecosystem, how they respond to stress, and how they adapt - or fail to adapt - to changing conditions. This allows researchers to develop more resilient, versatile, and sustainable solutions that are more likely to function effectively in the real world. The laboratory is an indispensable starting point for synthetic biology, but it is only the beginning of a long and iterative process, the endpoint of which requires robust constructs that can function in dynamic, uncontrollable, and often hostile environments in order to truly have an impact. 

SaPh:IRES makes progress toward developing real-world engineering solutions by testing our applications in two simulated environments: human colon and soil microcosms. The model colon is designed to simulate the fluid dynamics, retention time, temperature, and contents of the colon, and the soil microcosms mimic the fluid dynamics, physical soil structure, and plant interactions present in soil environments. These models provide a degree of realism which allows us to evaluate the efficacy of our satellite systems in conditions that mimic the conditions of two important environments. 

The Colon Model

Introduction and Current State of the Field

The human gastrointestinal (GI) tract is among the most complex ecosystems in the human body, and one of the most important for human health (Liang et al., 2018). The study of the gut microbiome has generated world-wide interest in the past two decades, leading to the development of a number of in vitro gastrointestinal simulators which aim to mimic the key characteristics of the human gut. However, despite the range of in vitro models that have been developed, due to the complexity of the gut, including diverse physiological, biochemical, and biological conditions, these models aren’t simultaneously able to effectively simulate all of the elements of the gut ecosystem (Roupar et al., 2021). Commonly-used in vitro GI models can be batch culture systems, single-stage or multi-stage continuous culture systems, or microbiome-host interaction models. These models share methods of replicating the gut environment: many are flushed with an inert gas, usually nitrogen, to create anaerobic conditions, many are inoculated with a fecal sample or a representative bacterial community to mimic the microbiota, and many of these models feature temperature and pH control. Each relevant GI model has its strengths and limitations, and is employed for different reasons. 

Batch culture models are among the simplest GI simulation models. They involve growing a microbial community or digesting a substrate in a single bioreactor vessel, often adjusted to a desired pH and kept under anaerobic conditions to mimic the gut environment. Typically, runs of these models are short-term, lasting 24 to 48 hours, and study the impact of substrates on the physiology and biodiversity of intestinal microorganisms. Batch culture models are relatively simple both in terms of their physical assembly and experimental design, and are inexpensive and reproducible, however their simplicity comes at a cost. Batch culture models are disadvantaged by substrate depletion and accumulation of end products of metabolism, as well as lacking simulation of more complex GI features like peristalsis, making them unviable for longer term simulations (Roupar et al., 2021). 

Single-stage continuous culture models simulate the conditions of a single section of the GI tract by maintaining a constant influx of media and efflux of waste products, allowing for the establishment of a microbial community and increasing relevance to the human gut. The continuous addition and removal of media gives this type of model more flexibility for experimental design, offering much longer runs of experiments (from days to months) and the ability to more effectively control parameters such as retention time and the amount of media added and removed. However, as this model only simulates a single stage of the GI tract, its applicability in more holistic studies of the gut is limited (Roupar et al., 2021). An example of this type of model is the Proximal - Environmental Control System for Intestinal Microbiota (P-ECSIM), which is a single-state continuous culture model designed to mimic the conditions of the proximal colon (Feria-Gervasio et al., 2011). Like some other GI models, the P-ECSIM is fed a defined media formulated to mimic the traditional western diet and inoculated with a fresh fecal sample from a healthy donor (Feria-Gervasio et al., 2011). Single-stage continuous culture models are generally used for investigating the stability and resiliency of microbial communities or the impact of specific dietary components or drugs on microbial activity and composition (Roupar et al., 2021). 

Multi-stage continuous culture models simulate the conditions of multiple sections of the GI tract, allowing for an even more relevant simulation of the gut. These models allow for the study of microbial composition and activity across different gut regions, offering a higher degree of complexity and realism to gut simulations, while retaining the strengths of single-stage models. Nevertheless, replicating a larger portion of the digestive tract, in some cases the entire thing, increases the difficulty of parameter setting, along with the assembly and operation of the model. While it is difficult to find exact figures of the initial and operational costs of these models, the high-tech specialized equipment and need for continuous creation of culture media and strict maintenance means that these models can exceed $100,000 in terms of physical components, equipment, and operational costs. This is a massive disadvantage compared to more cost-accessible models (Lorenzo et al., 2020). Many of these simulators are based on a three-part colon model published in 1988 (Gibson et al., 1988). This model comprises a smaller acidic, nutrient-rich environment in the ascending/proximal colon, which increases in volume, and gets more neutral and nutrient-sparse through the transverse/middle colon into the descending/distal colon, which flows defined media through the system via gravity (Gibson et al., 1988). Its representativeness to the human colon has been validated by comparing chemical short-chain fatty acids from the in vitro model to human intestinal samples (McFarlane et al., 1998). Due to its wide applicability and relevance, it has become a standard for all multi-stage continuous culture models of the colon (Roupar et al., 2021). Like the single-stage continuous culture models, the multi-stage models also allow for control over the retention time, pH, temperature, and other parameters, as well as expansion beyond the colon, into simulations of the full digestive tract. One of the most well-known GI models, the Simulator of the Human Intestinal Microbial Ecosystem (SHIME) is one such model that encompasses the entire GI tract, incorporating a stomach and small intestine compartment along with the three-stage colon model (Van de Wiele et al., 2015). The five double-jacketed glass vessels are connected with peristaltic pumps, which pump the defined media, consisting of nutritional feed and pancreatic juice, through the system via peristalsis, more closely resembling the movement of fluid through the GI tract. The pumps flowing in and out of the stomach and small intestine operate semi continuously, creating a fill-and-draw system for the first two vessels (Van de Wiele et al., 2015). The stomach vessel is fed three times a day with the defined nutritional feed, formulated to mimic the western diet, and pumped into the small intestine along with a synthetic pancreatic juice mix consisting of bile salts, pancreatin, and sodium bicarbonate. The colon vessel pumps operate continuously, maintaining a constant volume and controlled pH. The SHIME offers increased flexibility in its physical design, as compartments can be added or removed depending on the experimental design, and many parameters are left up to the operators control; the retention time of each of the vessels, choice of inoculum, pH profiles, feed composition, and more can be altered to fit the profiles of different populations and species (Van de Wiele et al., 2015). The incorporation of mucin-covered microcosms to the vessels adds another dimension to the SHIME, replicating the mucosal layer of the gut, and the mucosal microbiome that colonizes it (Van den Abbeele et al., 2011). 

Finally, microbiome-host interaction models utilize microfluidic technology to model the dynamic and spatially structured environment of the GI tract. These models typically co-culture human host cells and microbial cells on a microfluidic chip designed with a microchannel for precise control of fluid flow, nutrient supply, and peristalsis, along with a barrier for compounds of interest to permeate (Roupar et al., 2021.) Its technical complexity lets it provide a highly relevant simulation of many key physiological, biochemical, and physical characteristics of the gut, along with interactions between microbial and host cells, but also can inhibit its reproducibility and create challenges with the fabrication of these models (Bhatia, 2014). Due to its small size, it is harder to conduct larger-scale experiments with the model, and struggles to replicate the highly diverse microbiome present in the gut (Valiei et al., 2023). Nonetheless, microbiome-host models are an incredibly powerful tool to study disease, drug absorption and permeability, host-microbe interactions, and much more. 

These models have their own advantages and disadvantages pertaining to their applicability to the environment they try to model, but each one also comes with different benefits and limitations to adoption and implementation for a laboratory. Batch culture and single-stage continuous culture models are much easier and cost-effective to create and operate than the complex multi-stage continuous culture and microbiome-host models, making them a more ideal choice for a smaller-scale laboratory, such as an iGEM team, to adopt and utilize (Roupar et al., 2021). While companies have developed products and platforms to streamline the manufacturing and use of complex in vitro models, such as the SHIME itself developed by ProDigest, these commercialized models can be priced far out of reach for laboratories and researchers, meaning that options are limited for realistic simulators of the human gut (“Shime®”). Creating a microbiome-host interaction model requires expertise in microfabrication and system integration, along with high-tech machinery and scientific equipment (Valiei et al., 2023). Though creating a multi-stage continuous culture model still requires a not-insignificant amount of time and money, its design lends itself to being built much more easily for smaller labs. Both single and multi-stage models are typically reliant on expensive, specialized equipment which are large in footprint and limit accessibility. The Mini Colon Model, or MiCoMo, was developed to overcome these major hurdles to utilizing colon models, providing a low-cost, accessible colon simulator which is easy to assemble, small in size, and offers high throughput (Jin et al., 2022). While this model is an excellent alternative to more complex and costly models, it consists of multiple single-stage batch culture colon models, and therefore is subject to the same limitations as other batch culture or single-stage models (Jin et al., 2022, Marther et al., 2023). This is why William & Mary 2024 decided to develop our own multi-stage continuous culture model of the human colon to test our engineered satellite phage systems. We’ve designed and built a GI simulator which is fully documented, reproducible, and cost-effective to increase the accessibility of complex in vitro colon models for future iGEM teams and synthetic biology. Our model features twin three-stage colon models created using relatively low-cost materials, and its design and assembly are fully documented.

Design

Summary

Figure 1: Human Colon Model

We initially based our model design on the SHIME, a multi-stage continuous culture model of the entire digestive tract, from the stomach to a three-part colon, as the model has previously been used to study phage (Laforêt et al., 2023). The SHIME’s flexibility is one of its strongest attributes, permitting the addition or removal of vessels depending on experimental design (Van de Wiele et al., 2015). This flexibility came in handy for our hardware team when we altered our design by adding another vessel to our model and converting our digestive tract model into two three-stage colon models and a single-stage model. Our three-part colon model comprises an ascending, transverse, and descending colon, connected by peristaltic pumps and 2.5 mm ID x 5.5 mm OD silicone tubing, which pump defined media, a mix of nutritional feed and pancreatic juice, through the system at a constant rate of 25 mL/hr. The volumes and retention times of the colon vessels are based on standard SHIME runs. The defined media, consisting of a nutritional feed formulated to mimic the traditional western diet and an alkaline pancreatic juice to include digestive enzymes, is made as previously described (Van den Abeele, 2010). 

Volume (mL) Retention Time (hrs)
Ascending Colon 500 20
Transverse Colon 800 32
Descending Colon 600 24

Table 1: Volumes and Retention Times of Colon Sections.

However, our model greatly differs from the SHIME, and other multi-stage continuous culture models. Rather than the traditionally-used double-jacketed glass vessels, our model repurposes enteral delivery bags for the colon chambers. While the glass vessels provide temperature control for the chambers in traditional models, we suspend our bags in an aquarium tank filled with water, which is heated via aquarium heaters to body temperature, as well as external terrarium heaters. Aquarium wavemakers are employed in our model to move the water back and forth in the tank, sloshing the bags around with it to create a mixing mechanism. 

The aquarium tank’s top is made from a sheet of 25” x 12.8” PVC plastic, machined to rest on the edges of the tank. The top houses all of the electrical components of the model, including the pumps and an Arduino Mega 2560 Rev3 microcontroller with three stacked Adafruit Motor Shield v2s controlling the pumps. The pumps are Gikfun 12V peristaltic dosing pumps which are attached to 3D printed pump stands. The Arduino and motor shields are housed in a 3D printed case with a detachable top and spaces to allow for wire connections. 

Summary

Figure 2: Technical Drawing of the Aquarium Tank Top

Our model’s design can support up to 7 chambers: two three-stage colon models and one single-stage model, however the flexibility of the model it was originally inspired by translates to our model. By reconfiguring the tubing connections and slightly altering the code, our model can be a five-stage model like the SHIME, two three-stage colon models and a single-stage colon model like we used for this project, or other configurations depending on the needs of the experimental design. 

Many of our design decisions were made due to cost considerations: we wanted to build a low-cost, easily-reproducible multi-stage continuous culture model which future iGEM teams could build and implement to test their engineered constructs in a realistic colon environment. The raw materials required for the physical build of our model costs under $1000. 

Item Name Supplier Quantity Needed Unit Price (USD) Item Name
20 Gallon LifeGard Aquarium Tank Amazon 1 125 20 Gallon LifeGard Aquarium Tank
hygger Aquarium Heater Amazon 1 55 hygger Aquarium Heater
VIVOSUN Terrarium Heater Amazon 2 30 VIVOSUN Terrarium Heater
hygge Aquarium Wavemaker Amazon 2 60 hygge Aquarium Wavemaker
1/4" x 1/8" Weatherstrips Amazon 1 8 1/4" x 1/8" Weatherstrips
24" 48" PVC Sheet Local hardware store 1 40 24" 48" PVC Sheet
1200 mL Enteral Delivery Gravity Bag Set Amazon 7 15 1200 mL Enteral Delivery Gravity Bag Set
Adafruit Motorshield V2 Amazon 3 20 Adafruit Motorshield V2
12V power supply Amazon 3 16 12V power supply
Arduino Mega 2560 Rev3 Amazon 1 53 Arduino Mega 2560 Rev3
5V power supply Amazon 1 16.5 5V power supply
Gikfun 12V Peristaltic Pump Amazon 10 12 Gikfun 12V Peristaltic Pump
3/32" Diameter Bulkhead through barbs Amazon 3 7 3/32" Diameter Bulkhead through barbs
5.5mm OD x 2.5mm ID Tube connectors Amazon 3 9.6 5.5mm OD x 2.5mm ID Tube connectors
Electrical Tape Local hardware store 1 10 Electrical Tape
Gorilla High Strength Clear Glue Local hardware store 1 10 Gorilla High Strength Clear Glue
26-gauge Electrical Wire Local hardware store 2ft .60/ft 26-gauge Electrical Wire
6-gauge Copper wire Local hardware store 2ft 1.05/ft 6-gauge Copper wire
5 ft Roll 5.5mm OD x 2.5mm ID Silicone Tubing Amazon 1 8 5 ft Roll 5.5mm OD x 2.5mm ID Silicone Tubing

Table 2: Human Colon Model Materials List

Setup and Assembly

Summary

Figure 3: 3D Model of Human Colon model

Once all of the raw materials have been obtained, a few things need to be done. First, the PVC sheet needs to be machined to fit the aquarium tank, and holes need to be drilled into the top if not done while machining, to make the top match the technical drawing in Figure 2. Then, the pump bases and Arduino case and lid should be 3D printed. To prepare the bags, drill a single hole through the top of each bag and insert a bulkhead through-barb through it before tightening. For the electrical components, each motor shield needs to have its own I2C address. Refer to the Adafruit Motor Shield v2 documentation for setting up the motor shield (“Adafruit”). The standard address for the Adafruit Motor Shield v2 is 0x60; the binary address is determined by the jumper board toward the bottom of the motor shield. Adding a drop of solder to bridge one of these jumpers changes the binary address: the rightmost jumper is bit #0, soldering it like the picture below changes the address to 0x61. Second to the right is bit #1, which changes the binary address to 0x62. Once each motor shield has a unique address (0x60, 0x61, and 0x62), the Arduino can distinguish between them. A crucial piece of preparation is calibrating each individual pump. Due to the relatively low quality of the pumps used, manufacturing errors can and likely will cause each pump to perform slightly differently. Ensure each pump is calibrated to obtain the right flow rate before installing them onto the tank top by measuring the volume of fluid dispensed at different speeds, until settling on a speed that allows the pump to reliably output a desired volume. 

As virtually the entire model is attached to the top, the top is assembled first before the rest of the model. Once all of the components are prepared (i.e. top is machined, components are printed, etc), the first step is gluing the pump bases and Arduino case onto the top. As these dry, the bulkhead through-barbs can be inserted through the holes in the top and tightened. After this, the pumps are secured to the top, preferably by tapping holes into the pump bases and screwing the pumps in, and the Arduino Mega is placed into its case. Stack motorshields onto the Arduino one by one, and with each one make connections between a pump and a DC motor terminal(M1-M4), making note of which pump is connected to which terminal, as this is important to identify each pump in the code. Once all motor shields are stacked and all wiring connections are made to pumps, drill small pairs (1/16”) holes through the top at specified points, and insert copper wire through the top out to the bottom side of the tank top, and back up to the top side. Check there is enough space to fit a bag head through the copper loop on the bottom side, then push down and glue copper wire to the tank top. Fit the tops of the bags through each copper wire loop and close them so that the bags hang on the wire, then begin connecting the tubing between the bags so that the model resembles Figure 2. Cut tubing and connect tubing with connectors appropriately so that not too much stress is placed on the connection between the bag and the tank top, and tubing flows smoothly between the pumps and through-barbs on the tank top. Apply weatherstripping to the lip of the bottom side of the tank top, allowing the top to rest on the weatherstripping when placed on the tank. Once all of this is done, run water through the system with prepared code to test its functionality. 

Function and Operation

The C++ program uploaded to the Arduino determines the identity of each pump, the speed at which each pump will run, and the time for which each pump will run. The code is structured so that once it is uploaded, the ascending colons of both models take up 200mL of defined media for 210 seconds while the colon pumps run for 10 seconds, then every 6 minutes the colon pumps run for 10 seconds to create an average flow rate of 25 mL/hr, resulting in 200 mL of effluent being dispensed per model, before the loop starts over. This requires a fresh batch of media to be connected to the FeedAC1 and FeedAC2 pumps before the code is uploaded or the Arduino is plugged in, and a fresh batch of media to replace the empty batch every 16 hours if the media is made in 800mL batches. The pumps were calibrated with this feeding structure in mind. The Arduino program used for our colon model is shown below, and both the code and relevant 3D models are uploaded to our GitLab repository.

Summary

Figure 3: Arduino code for model human colon

The duration of a colon model run is flexible, depending on the experimental design. A standard SHIME run can last up to 8 weeks, with a 2 week period for stabilization, another 2 weeks to get baseline measurements under normal conditions, a 2-4 week treatment period, then a 2 week washout period to evaluate how long effects last, SHIME runs can be as 1 week if needed, though(Van de Wiele , 2015). We chose to operate our model with 4 week runs, comprising a 1 week stabilization period and a 3 week treatment period. Typical GI simulators inoculate their models with a fecal microbiome isolated from a one individual, or defined microbial community meant to represent the gut microbiome. As our engineered phage satellite systems could potentially confer antibiotic resistance to infected BSL-2 bacteria, we decided against inoculation with a fecal sample in order to make our model accessible to BSL-1 labs. We opted instead to perform our first run with a single defined bacteria strain, E. coli HL713, and then add a commercially available probiotic for our second run, to achieve a microbial community in our model. We decided to use Seed Health’s DS-01 probiotic, due to its diverse list of 24 bacterial strains, and because Seed claims to have tested it in the SHIME to validate its performance(“Seed”). A more in-depth look at our experimental design and relevant protocols can be found here. Read about the results of these experiments here.

Ongoing & Future Improvements

Our colon model is able to replicate many conditions of the human colon at a much lower price point than commercially available GI simulators, including the temperature, mixing, contents, and fluid mechanics of the colon. While the choice of enteral delivery bags instead of double-jacketed glass vessels greatly lowers the cost, we hypothesize it might come with other benefits, though further research is required to validate these potential benefits. The choice of enteral delivery bags instead of glass vessels may enhance the physiological relevance of our model, as the aquarium tank more closely resembles the peritoneal environment of the colon, surrounded by body fluid, and the water uniformly distributes the heat around the bags, similar to how body fluids warm the organs, potentially leading to more stable temperatures. The medical-grade silicone which makes up the bags provides a softer, more compliant material which may simulate the organ tissue of the colon better than rigid glass vessels, allowing the bags to expand and contract as they move within the tank, which may mimic the mechanical forces of the GI tract better than the magnetic stirring of traditional models. Finally, the compartmentalization of the bags within the same shared spaces also replicates the localization of the colon within a confined space. By far, though, the biggest strength of this model is its low cost and high reproducibility. With all of the materials, code, and assembly instructions documented, future researchers can utilize our model to create their own human colon simulator, allowing them to test engineered constructs in a realistic gut environment.

However, along with its strengths there are notable limitations to our model. The lack of precise scientific instruments means the temperature may not be kept exactly stable at 37 degrees Celsius, the low-quality pumps and other materials may degrade over time, eventually reducing accuracy and performance, and our model lacks some of the functionalities that high-tech multi-stage models possess. Our model does not simulate the anaerobic environment of the colon, nor does it simulate the different pH profiles of each colon section. Connecting the headspaces of the bags proved to be much more difficult than connecting the headspaces of glass vessels, and nitrogen gas purges would end up inflating the bags almost to a point of popping as the air would not flow through the headspaces well enough, and pH control with the bags also proved to be more difficult than it would be with glass vessels. The lack of a fecal microbial community also means it does not simulate the gut microbiome as accurately as it could. Additionally, the small size of the tubing caused mucin agar to clog the tubes frequently, leading to us ceasing to put mucin agar in the model. Finally, the copper wire that tethers the bags to the tank’s top corrodes over time, causing them to break, which we needed to repair for all of the bags. 

With its limitations in mind, we offer future improvements for implementing our colon model to its full potential. Using larger size tubing than we did would improve the flow of media, allow for the incorporation of mucin agar to flow easier through the system, and potentially make maintaining anaerobic conditions throughout our model easier, as gas would have an easier time flowing through the headspaces. Incorporating anaerobic conditions would involve connecting a nitrogen gas tank to the ascending colons of the models, then connecting the headspaces of the bags for each model, and fixing the tube out of the descending colon under the water level to create a bubbler, allowing the measurement of the flow rate and preventing any air from returning to the headspace of the bags. If anaerobic conditions could be effectively maintained in the model, a larger variety of inocula could be chosen, including fecal microbiomes and defined microbial communities containing obligate anaerobes, increasing the applicability of our model’s microbial culture to that of the human colon. Investing in higher-quality temperature control systems would increase the control over the temperature, as well as higher-quality peristaltic pumps to achieve more reliable flow rates and longer lifespans. Additionally, using a different material other than copper wire to suspend the bags from the top, or even copper wire that has a plastic coating around it to prevent corrosion, would greatly benefit the model. Most GI models use an automated pH controller, which allows for an even higher degree of realism, further distinguishing model compartments. Implementing this into our model was attempted in the initial design, but didn’t make into the final design. While high-quality pH probes and pH controllers would increase the cost greatly, regulating the pH of the different colon sections replicates an important aspect of the colon environment. 

Soil Microcosms

Introduction

The human colon model is the primary part of the hardware portion of our project, but it was essential to create an additional real-world environment in which to further test our constructs. Though less complex in terms of hardware development and testing, they nevertheless incorporate some novel hardware features, including a large-scale spatial design with cheap materials, nondestructive sampling, and the inclusion of regular simulated rainfall. However, the true novelty of these microcosms lies in the combination of these features with our experimental design and wet lab work. Soil microcosms are an indispensable model for synthetic biology, allowing researchers to easily simulate complex natural ecosystems on as large or small a scale as they desired. Particularly useful for agricultural, environmental, and bioremediation applications, soil microcosms are widely employed to test engineered organisms in conditions that closely simulate nature, mimicking the microbial diversity, nutrient cycling, and physical soil structure. These models provide an important, reproducible tool for synthetic biologists to study how an organism or substrate affects its surrounding ecosystem, and vice versa. 

Design

We designed our soil microcosms on a large scale to allow for horizontal spatial measurements of both the persistence of our constructs as well as their spread through the soil. The soil profile consisted of a mix of two-parts topsoil to three-parts garden soil, on top of a screen mesh layer and a layer of pea gravel at the bottom of each microcosm. To mimic plant interactions, either bush bean seeds or romaine lettuce seeds were grown in the microcosms, and watered regularly via a rainfall mechanism. These plants are fast-growing and provide a root system particularly important for E. coli growth. This rain mechanism was constructed out of sheets of ¼” thick plexiglass drilled with 5/64” diameter holes every 1.25 inches to slowly drip down water, mimicking rainfall. A stand made from spare robotics components to hold the rainfall apparatus above the microcosms, though other materials can be substituted for this purpose. The pea gravel on the bottom of each microcosm acted as a drainage layer, and 1/16” holes were drilled into the bottom of each plastic container. The containers were placed in aluminum drip pans, allowing water to drain through the microcosms into the pans, which acted as effluent collectors. 

Summary

Figure 5: Soil microcosm rainfall mechanism

We constructed our microcosms with low-cost, accessible materials so that they could be easily reproduced. The total cost of the 16 soil microcosms, excluding the rainmaker, was $269.91 or $16.87 per microcosm; see the materials list below. 

Item Name Supplier Quantity Cost (USD)
1 cu ft Timberline Top soil Lowes 3 2.48
2 cu ft MiracleGro Garden soil Lowes 3 8.47
Mainstays 5-count Aluminum pans Lowes 4 6.48
0.5 cu ft Kolor Scape Pea gravel Lowes 2 7.18
Iris Plastic containers Amazon 16 5.24
Wozelek Grow lights Amazon 2 26.99
Romaine lettuce seeds Amazon 1 5.79
Bush bean seeds Amazon 1 3.49
Mago screen roll Amazon 1 9.69
1/4" Plexiglass sheet Amazon 1 39.99

Table 3: Soil Microcosm Materials List

Setup and Assembly

To set up our microcosms, first, 1/16” diameter holes need to be drilled in the bottom of each plastic container. These holes should be evenly spaced out to allow effluent water to easily drain out of the bottom of the containers. The physical assembly of the microcosms simply involves placing the 16 plastic containers in their own drip pans, filling the bottom of each container with roughly a half-inch of pea gravel, placing two 15” by 9” cutouts of screen mesh on top of the gravel layer, and then filling each container with a soil mixture consisting of two-parts topsoil to three-parts garden soil up to roughly three-quarters of the containers’ volume. Finally, place the microcosms side by side and attach grow lights to certain microcosms to evenly distribute the grow lighting across four microcosms per grow light. 

To create the rainfall apparatus, a plexiglass sheet was cut into a rectangular shape greater in area than the tops of the plastic containers. Then the rest of the sheet was cut into four more thin pieces and glued using a waterproof glue sealant to the rectangular piece, creating a container able to hold water. Then, 5/64” diameter holes were drilled every 1.25 inches on the rectangular piece, creating a container that slowly leaks water droplets out of the bottom. A stand to hold the rainmaker above the microcosms can be constructed from a variety of materials or held above microcosms when watering; for our project it was made from spare robotics components. 

Function and Operation

Half of our microcosms were inoculated with E. coli, while the other half were inoculated with M. aichiense. Of the 8 E. coli microcosms, 4 were inoculated with transducing units of the non-targeting CRISPR system, and 4 with transducing units of an RFP system. Of the 8 M. aichiense microcosms, 4 were inoculated with a novel M. aichiense phage, and 4 with the mycobacteria phage satellites, also known as “phagelets”. Samples were taken from the soil microcosms regularly, and plated for phage and bacteria. Our microcosms included features like rainfall and nondestructive sampling, which are not commonly used in microcosm studies, but are not truly novel. It is their use in conjunction with our experimental design which constitutes novelty, as the use of spatial, nondestructive microcosms with simulated rainfall and plant interactions to study bacteria, phage, and phage satellites has never been done before. Relevant protocols and more information about our experimental design can be found on our experiments page, and their outcomes can be found on our results page.

Ongoing & Future Improvements

Our soil microcosms mimic the fluid dynamics, plant-soil interactions, and physical structure of the soil, as well as allow for measurements of the persistence and spread of our inoculants. Our rainmaker apparatus mimics rainfall onto soil, and the drainage of water through soil. However, our microcosms do not control or measure many variables, the introduction of which would improve our models and experimental design. Varying parameters like sterility, pH, oxygen levels, and soil moisture would add new dimensions to our experiments, allowing us to track optimal conditions for the performance of our engineered constructs. Additionally, to further enhance realism, the physical structure of the soil could be modified to replicate local soil structure, rather than using a mix of garden soil and topsoil. 

Conclusion and Contribution to Knowledge

Testing in realistic environments is critical to ensuring engineered constructs are able to perform outside of controlled laboratory conditions. This step allows researchers to assess viability, ecological impact, and long-term performance, providing a tool for the exploration of the real-world potential of novel biotechnologies. 

Our project SaPh:IRES aims to bring satellite phage into the fold of contemporary synthetic biology by demonstrating their expansive potential in three key applications and providing investigators with a toolkit to harness the power of satellites. We test two of these applications in simulated real-world environments of the colon and soil, helping to underscore the versatility of satellite SynBio, as we aim to bridge the gap between laboratory success and practical implementation. 

In our project, soil microcosms were used to simulate rainfall and the complex soil ecosystem, and evaluate how our engineered satellites perform in a realistic soil environment. Our colon model, adapted from commonly-used models, allowed us to explore performance in realistic gut-like conditions for potential therapeutic applications. Both of these models enabled us to test our engineered systems in relevant conditions, yielding insights into their function, resilience, and adaptability in real-world scenarios. See our results section for more information. 

Our models are fully documented and cost-effective, allowing future iGEM teams and labs to test engineered constructs in dynamic, realistic environments, and carry out the same experiments, or improve upon our designs to create even more realistic models. By providing these tools, we enable others to validate the behavior and robustness of their synthetic biology applications in soil and colon-like settings, driving forward satellite phage and synthetic biology research toward real-world solutions.

References