Design
Based on the size and structure of our synthetic DNA (gBlocks), we designed our entire cloning strategy around Gibson Assembly (for A2, A3, B2, B1) and MEGAWHOP (for A1). We anticipated a high-efficiency, one-step build for all five enzymes into our P31 backbone.

Build
We performed the assembly reactions and transformed the resulting constructs into NEB Stable E. coli for cloning and propagation.

Test
We verified successful assembly using colony PCR and Next Generation sequencing. A2 and B2 were successfully cloned and verified on the first attempt. A1 and B1 sequencing results showed a point mutation. A3 failed to yield colonies.

Design
Based on the failure, we designed new primers to overlap the point mutation of A1 and B1. We also added a DPNI digest step to remove template DNA. We repeated the Gibson Assembly for A3 with more caution.

Build
We successfully corrected two previously unsuccessful constructs (A1 and B1) by using the designed primers and then used PCR to amplify the remaining gBlocks. A3 successfully yielded colonies on the LB plate. We purified the plasmids and sent A1, A3, and B1 for sequencing. After sequencing, the plasmids with the correct sequence were transformed onto T7 Express cells.

Test
Final verification using gel electrophoresis and Next Generation sequencing confirmed the correct sequence for A1 and A3. B1 still exhibited a point mutation outside of the coding region, which would not affect the protein expression. This achievement meant all five engineered plasmids were ready for characterization assays.

Learn
This success validated our repair process. We achieved our goal of building all five constructs by using data from a partial failure to inform a successful change in our engineering strategy. We then proceeded to the protein expression optimization cycle to ensure we could produce the required amount of functional enzyme.

Design
We adopted 2 different autoinduction media protocols from Simplified Autoinduction Media by The Bumbling Biochemist and Studier’s Autoinduction media (Studier, 2005).

Build
We inoculated T7 Express cells transformed with A1, A2, A3, B1, and B2 directly into 50 mL tubes containing each type of media and incubated at 30°C for 24 hours.

Test
None of the enzymes turned pink in either medium.

Learn
We realized that 50 mL of culture volume was too large for visible color development, as the expressed protein was too dilute to turn the entire medium pink.

Design
We designed an improved experiment using only 3 mL of each media to increase the chance of growing colonies. Colonies were grown a day before inoculation with autoinduction media. We also tested them under different time and temperature according to Table 1 to find the optimal condition.

Build
3 mL liquid culture tubes were first grown in LB medium containing ampicillin at 30°C for 24 hours. A 1:500 dilution was then prepared by transferring 6 μL of the LB – ampicillin culture into 3 mL of autoinduction media, which was then incubated at different conditions for an additional 48 – 72 hours.

Test
None of the cultures turned pink in either medium for all temperature/time conditions.

Learn
Dilution of 1:500 did not work. The high dilution likely reduced the initial cell density, limiting bacterial growth even in 3 mL of media. Additionally, choosing pink colonies would likely improve the chances of obtaining cells that actively produce the target protein.

Design
This trial followed the same setup as Trial 1, but instead of using diluted cultures, visibly pink colonies were directly inoculated into autoinduction media and incubated under varying temperature and time conditions in Table 1.

Build
Visibly pink colonies from LB plates were picked and inoculated directly in the media.

Test
A1, A2, A3, and B2 successfully developed the "Very pink" color and yielded high cell density with Simplified Autoinduction media. However, the B1 culture failed to turn pink neither on the LB plate nor after the initial incubation and subsequent days in our media, indicating a specific failure in its protein expression or stability.

Learn
We learned that the protocol was insufficient for B1. The problem was likely due to B1's specific sequence, making it prone to toxicity, degradation, or inclusion body formation. Our redesign aimed to implement a troubleshooting protocol specifically for B1, testing alternative, less-strenuous expression methods, such as Studier's medium (Studier, 2005) and IPTG induction, and different temperature conditions to trigger expression differently.

Design
Based on the failure of B1 transgenic expression in autoinduction media, we designed to test two alternative induction methods. First, we tried Studier’s standard autoinduction media, as it worked for other enzymes and contains additional minerals/nutrients. Secondly, we chose IPTG induction (manual induction after reaching an optical density of 0.6 at 600 nm), as this was the method for protein expression utilized in previous literature that expressed this enzyme.

Build
We started new liquid cultures of B1 and tested the new conditions. For the IPTG cultures, we monitored growth and added 0.1 uM IPTG at OD600 = 0.6.

Test
Both the Studier’s culture and the IPTG cultures were monitored over several days, but neither trial produced a visible pink color. This confirmed that the protein was not expressed via an induction method.

Learn
Ultimately, the lack of visible expression in all methods prevented us from purifying and characterizing B1, leading us to focus our final functional assays solely on the four successfully expressed enzymes (A1, A2, A3, B2).

Design
We designed a qualitative colorimetric assay protocol which tests each enzyme against its specific p-nitrophenol (pNP) substrate. Successful cleavage releases a yellow pNP group which can be used to visually confirm enzyme activity.

Build
We used the purified enzymes obtained from the optimized A1, A2, A3, B2, and CA expression runs. We prepared individual reactions using their specific substrates (e.g., A1 against its pNP substrate). B1 was excluded due to the lack of successful protein expression/purification.

Test
The A1 and B2 reactions immediately developed a yellow color after incubation, clearly demonstrating their ability to cleave their respective substrates. This visually confirmed the successful expression of functional protein. However, A2 and A3 showed no visible color change.

Design
We designed a quantitative assay to measure how absorbance changes when combined with substrates at different concentrations. We created a pNP standard curve to establish the relationship between absorbance and pNP concentration and solve for Michaelis-Menten kinetics values (Figure 17).

Build
We used the purified enzymes (A1, A2, A3, B2, CA) combined with varying concentrations of their matching substrate to spectrophotometrically measure absorbance at A405 over a 3 hour period.

Test
A1 and B2 had quantifiable data with varying absorbances for each concentration, while A2, A3, and CA showed similar absorbances to the blank and negative controls.

Learn
Both A1 and B2 had varying levels of absorbance in relation to the concentration of substrate, values that we used to determine the Km and Vmax in accordance with Michaelis-Menten kinetics. We were not able to perform this analysis with A2, A3, and CA since they did not have any detectable activity.

Design
We designed our one-pot enzymatic conversion based on the previous study on Akkermansia enzymes. From our meeting with ASU Professor of Immunohematology, Jeffery Wolz, we designed a tube-test crossmatch hemagglutination assay to determine blood compatibility.

Build
Enzymes (A1–B2) were prepared from protein purification of several autoinduction cultures. Then, A and O type porcine red blood cells were isolated from whole blood and washed with a conversion buffer. Reactions with corresponding enzymes were run on a tube shaker at 22 C for 30 minutes, 1 hour, and 2 hours.

Test
Following blood conversion, samples of each reaction were taken and diluted to 3-5% suspension to be tested against human AB serum and/or O plasma. Agglutinates were qualitatively measured by viewing agglutination in 2 mL centrifugation tubes. Additionally, converted samples were tested for blood type on commercial EldonCards and compared to controls (unprocessed blood).

Learn
While this assay cycle first allowed us to view agglutination, which was clear clumping of the cells in the plasma/serum solution, it was difficult to identify clear differences between experimental groups. For example, viewing of agglutinated cells requires gentle shaking of the tube, which can be difficult if overshaking occurs. Possibly, some macroscopic results can be skewed by improper technique, so we decided to use microscopic images to identify agglutination instead.
Secondly, testing a 30-50% suspension of converted blood cells on the blood typing card resulted in reaction with the control field, which invalidates the result of the card. We hypothesized that the saline solution used to dilute the red blood cells may have resulted in the undesired reaction.

Design
Based on information from the previous cycle, we set aside an additional reaction for the enzyme treatments that would be tested on the EldonCard. 4 Degrees was chosen for all reactions to assess whether they would be functional while stored in colder temperatures. The cold-adapted enzyme targeting the B antigen (also the alpha-gal antigen on porcine blood) was also crucial to test at 4 C, since it was hypothesized that activity is likely better at lower temperatures.

Build
Using enzymes and washed red blood cells from the previous iteration, reactions for A and O blood with their corresponding enzymes were run on a tube rotator at 4 C for 30 minutes, 1 hour, and 2 hours. Following conversion, blood was diluted to 3-5%. Reactions used in the EldonCard were diluted to 30-50% using the previously separated plasma of the converted to reflect original blood components.

Test
The 3-5% blood suspensions were crossmatched against either O plasma or AB human serum. The reactions for the EldonCard from the last iteration (at 22 C) were repeated and tested for the presence of A and a-gal antigens.

Learn
Because we used microscope imaging in this iteration, agglutination was easier to view compared to viewing in test tubes. Here, we learned that A1 was capable of cleaving antigens on A type porcine red blood cells, but activity at the concentration we used (1 uM) was not enough to completely cleave all antigens and eliminate agglutination.
For the cold-adapted enzyme, we learned that it was likely that the enzyme works better at 4 C due to reduced agglutination in the microscope images. We learned that design for future experiments would have to include crossmatching with samples containing antibodies against the extended antigen structures. This way, we can isolate specific extended structures and identify whether or not they are being cleaved.

Design
To ensure our enzyme conversion system can function at blood storage temperatures (1–6 °C), we planned a workflow to identify enzymes within the GH families of interest (A2–GH35A, A1–GH36A, A3–GH95A, B1–GH110A, B2–GH20A) that are likely cold-adapted.

Build
Using the CAZy database, we obtained datasets of enzyme information for each required GH family. We compiled a list of genera that, based on literature, are of organisms that are native to colder temperatures, and therefore, likely to have cold-adapted enzymes. We filtered the list of enzymes to only include those from our genus list. Using the NCBI protein IDs, we verified that the shortlisted enzymes exist, have the required function, and are likely to work within our desired temperature range.
Test
We used this workflow to manually shortlist enzymes in the GH36 and GH110 families. The shortlisting resulted in a significant decrease in the number of entries.

Learn
Doing this process manually requires a substantial amount of time and manpower, especially considering some of our GH families had over 25000 enzyme entries in the database. To increase efficiency, we decided to automate this process using code.

Design
We decided to automate and build a preliminary code. We compared protein IDs from CAZy with the NCBI Protein database to output organisms that contain the name of the enzyme we were looking at.

Build
After gathering the enzyme list with their names, we used a CSV file with the filtered proteins from the psychrophilic genera from CAZy and used that to compare with a DataFrame with filtered Protein IDs with organisms from the NCBI Protein database using Entrez.esummary.

Test
We got 19142 counts for GH family GH36 before the NCBI, and only 15 counts for GH family GH36 after passing it through the NCBI Database. Moreover, we got 1244 counts for GH family GH110 before the NCBI, and only 61 counts for GH family GH110 after passing it through the NCBI Database.

Learn
The problem we ran into was that we had no way of determining whether the organisms were cold-adapted or not. While the list was made to have that concept in mind, we realized we needed to confirm that the enzymes themselves were psychrophilic.
We solved the problem immediately by researching the source of the species and shortlisted any that were likely to function below 10 degrees Celsius.

Design
The code could handle small databases; we decided to improve the code with a focus on the characterization of only A and B antigens, and compare structural similarities between those. We decided to do so by testing the existing code on complete GH families (GH20, GH35, GH95), to improve the clarity of the output.

Build
We modified the code so that it could display results as a dictionary instead of a list, with each entry including the enzyme name, accession ID, and organism name. The updated code then ran on the full GH family datasets to check performance and accuracy.

Test
The code successfully reduced entries and accurately filtered and categorized most enzymes, as verified by manual inspection. Processing time for roughly 18000 entries was 5 to 6 hours, suggesting it could handle large datasets with moderate efficiency.

Learn
The filtration system was functional but limited by runtime and minor data mismatches. An important update that could eventually be done would be a method for the code to significantly reduce time and address any of these small inconsistencies in formatting found with the NCBI titles.

Design
Filtered Project 1 hits for psychrophilic α-galactosidases predicted to target the B-antigen chain for cold-active conversion during pre-transfusion (4–10 °C). Selected two candidates: Polaribacter sejongensis AUC20683.1 and Colwellia sp. ARD46000.1.

Build
Cloned the Colwellia candidate (ARD46000.1) and prepared porcine Type O whole blood for assays; B2 (ExtB-cleaving) enzyme available for combination testing.

Test

To test the effectiveness of the Colwellia cold-adapted (CA) enzyme on blood, we treated O-type porcine blood for 2 hours at two different temperatures. While this ɑ-galactosidase enzyme was intended to cleave the human B antigen, it is also capable of cleaving the ɑ-galactose antigen on porcine blood, which is detectable by human antibodies. Figure 1A highlights that at 22 C, the CA and B2 enzymes are not capable of cleaving the ɑ-galactose antigen on porcine blood, resulting in detection by human antibodies and severe clumping. However, Figure 1B and 1C indicate that treatment with CA alone and CA + B2 is capable of ɑ-galactose cleavage, as shown by a reduced amount of agglutinated cells and an increased prevalence of single red blood cells.

Learn
The Colwellia cold-adapted enzyme is functional at low temperature, supporting the cold-active conversion concept. Room-temperature treatment is insufficient, underscoring the need to operate near storage temps.

Design
After identifying the potential psychrophilic enzymes in Project 1, we aimed to determine whether there were any structural and functional similarities between the cold-adapted and the mesophilic enzymes.

Build
We started with a plan of action to complete the CLUSTAL sequence alignment; first, identify a list of 5 enzymes per group (thermophilic, mesophilic, and psychrophilic), then obtain the protein sequences from the NCBI Protein Database, run the CLUSTALW program, and compare the regions of similarity in the groups. Using this, we performed multiple sequence alignments between psychrophilic enzymes and known mesophilic counterparts to identify the conserved residues and motifs associated with catalytic activity.

Test
The CLUSTAL alignments revealed that key catalytic residues and active site motifs were conserved across both enzyme types, which suggested that psychrophilic enzymes retained functional similarities to their mesophilic counterparts. This could be said to indicate a strong cold activity potential in the selected psychrophilic enzymes.

Learn
The sequence alignment confirmed that psychrophilic enzymes could, under low temperature conditions, perform similar functions due to the structural similarities. These results allowed for further comparative analysis within the same GH families to explore the regions of adaptation that they share.

Design
We aimed to identify the regions of similarity among the enzymes in the GH families used across different bacterial species. This analysis would help determine whether a specific structural region is responsible for the psychrophilic adaptations.

Build
Using literature review and database analysis, we compared the sequences of psychrophilic and mesophilic enzymes within the GH36 and GH110 families. We examined the conserved domains and active site architecture, with a focus on the flexibility and stability at lower temperatures.

Test
Comparative analysis showed that there were regions of similarity between the enzymes in the same family. Psychrophilic enzymes displayed certain amino acid substitutions consistent with an enhanced degree of flexibility, which is a known feature of cold-adapted enzymes.

Learn
Through the intra-familial comparison, the observed psychrophilic and mesophilic enzymes preserve key structural motifs, differing mainly in peripheral or stabilizing regions. These variations are likely to contribute to temperature-specific activity, like that looked for in cold-active enzyme optimization.

Design
After a literature search on strategies to leukodeplete immune cells from whole blood, we learned the only available strategy was to filter out RBCs. We chose a different strategy of extracting white blood cells (WBCs) using their natural affinity for some materials. We chose SCOBY hydrogels as a potential filter material because of their partial chitin content. Literature searches showed that immune cells have a high affinity for chitin, however SCOBY hydrogels were not inflammatory, and already being tested as wound dressings. Based on this we hypothesized they would have the ability to both filter plasma and trap white blood cells in blood samples. A successful design would result in a low-cost material that could improve on the function of WBC filters.

Build
The SCOBY hydrogel was first decellularized by incubation with 1M NaOH overnight, followed by 3 washes in deionized water over another 24h. The decellularized SCOBY was processed into thinner fragments to improve flow-through and WBC binding. Decellularizing the SCOBY removed all color indicating chemicals other than the bacteriocellulose/chitin matrix had been removed. Potentially making it easier to push blood through the hydrogel while still providing sites for binding.

Test
We examined the effects of dehydration/rehydration and decellularization, and tracked how well it worked in reducing plasma vs. binding white blood cells. Weighing the SCOBY material pre and post hydration, we found the hydrogel was 99% water by mass. We then exposed the SCOBY to pork blood samples with WBC labeled using trypan blue stain. Pork blood had fewer WBCs than estimated for healthy blood, but all trypan labeled cells were removed from our blood samples and visibly labeled the SCOBY surface, allowing us to successfully validate the filtering process.

Learn
Pork blood had low white blood cell levels, but the SCOBY did bind white blood cells. Its potential impact on plasma components remains undetermined. This was useful because it showed that natural biomaterials can interact with blood cells, and it gave us first-hand experience with decellularization methods that we could apply to other materials later. We also realized that if WBCs only bound to the SCOBY surface there would be a limit to leukodepletion in a higher volume of blood. This had to be solved in future iterations.

Design
To optimize the leukoreducing capability of SCOBY, we looked for processing methods to maximize its surface area within the bag. We reasoned that a freeze thaw cycle of the hydrogel would create pores and surface disruptions to increase its surface area for WBC binding.

Build
We again decellularized the SCOBY like in iteration 1, and then tested 3 preservation conditions- air-dried, lyophilized, and frozen, thawed, then lyophilized.

Test
To look at the SCOBY surface structure we used scanning electron microscopy (SEM) on lyophilized SCOBY. We placed 1mm x 1mm x 0.1mm slices of SCOBY onto carbon tape and applied a sputtered gold coat to preserved structures and increase SEM contrast. Samples were prepped onto a gold stage, and analyzed with SEM. We identified that the frozen, thawed, and lyophilized sample altered SCOBY’s surface topography, increasing the surface area for leukocytes to bind to.

Learn
We learned that the optimal processing technique for SCOBY within our filtration bag is freezing, thawing, and lyophilizing. Further testing is required to quantify leukoreduction compared to traditional methods.

Design
To integrate our purified enzymes with our blood type conversion hardware we sought a way to preserve enzymes for long periods of time without significant loss in enzyme activity. We began by testing the viability of freeze drying enzymes, using a sucrose stabilizing solution.

Build
10ul aliquots of purified enzymes were mixed with 500 ul of 30% sucrose, and lyophilized using the lab’s freeze-dryer. Lyophilized enzyme was kept at -80 for 2 weeks.

Test
We rehydrated enzymes to perform a colorimetric substrate assay. We then compared the enzymatic activity using data collected from enzymes before and after lyophilizing to determine losses in enzyme activity. Before finalizing our results, we used Michaelis-Menten kinetics to evaluate enzyme performance. The values used for this method, Vmax and Km, allowed us to compare and contrast.

Learn
We got hands-on experience using the lyophilizer and preparing freeze-dried enzymes. This process was new to the team, so part of the “build” stage was learning how to prepare samples and operate the machine properly. This iteration helped us build confidence with enzyme preservation techniques and set the foundation for longer-term and future storage solutions.
While resuspending the lyophilized enzymes, we discovered that our desired concentration was not feasible for some of our samples; the high amounts of sucrose in the tube would not be properly dissolved in the low volumes required to reach a concentration of 0.1 ug/ml of enzyme. A key lesson of the process was to aliquot enzymes not based on volume but on higher molar concentrations, making resuspension easier.
Overall, we learned lyophilization’s effect on enzyme activity was almost minimal but affected each enzyme differently. A1 retained some activity but showed signs of inconsistency, resulting in inaccurate Vmax and Km measurements while B2 maintained its activity with an increased Vmax and decreased Km values, suggesting lower substrate affinity but a higher potential reaction rate. The remaining enzymes showed no activity after lyophilization, suggesting that the process was not effective in preservation for all proteins. From these tests, analyzing the differences helped us better understand how to interpret Vmax and Km values before and after via enzyme lyophilization.

Design
The first idea was to build the sensor system around an Arduino, since it’s affordable and flexible for prototyping. We focused on simple circuit features (basic wiring, sensor input, microcontroller output) and making sure it could connect with the biosensor concept.

Build
We created CAD prototypes for housings, testing different shapes and sizes to fit the Arduino and wiring. The goal was to make it lightweight, accessible, and sustainable so anyone could use or rebuild it easily. We developed an initial prototype of the potentiostat circuit using wires on a breadboard.

Test
We performed early runs to check if the Arduino could handle the circuit setup and if the sensor inputs responded the way we expected.

Learn
We concluded that turning our breadboard design into a printed circuit board would both simplify its design, increase its reproducibility, and allow for easier testing.

Design
We designed the first version of our custom PCB potentiostat as an Arduino shield in KiCad to run cyclic voltammetry and linear sweep voltammetry by applying a voltage to the working electrode, allowing the counter electrode to adjust based on the rate of reaction occurring, and the reference electrode to maintain a stable voltage.

Build
We designed, ordered, and soldered the board together and set up the circuit for initial runs. Furthermore we soldered a screen printed electrode with a working, reference, and counter electrode onto our circuit. We also included a potentiometer to adjust the resistance between the reference electrode and operational amplifier so we could determine the optimal resistance to reduce noise.

Test
We tested the circuit using a test solution of CuSO4 and HCl to determine whether it could detect the rate of a redox reaction. We took 50 measurements at a high resistance (~10kΩ) and 50 measurements at a low resistance (~1kΩ).

Learn
We concluded that a higher resistance (~10kΩ) greatly reduced noise, but found limitations in the design because it had to be connected to a laptop to collect data and it used an Arduino which is bulky in size and has suboptimal memory. In future iterations, we aimed to optimize our printed circuit board by switching to an ESP32 and to allow data to be collected wirelessly.

Design
An ESP32 version was used to add Wi-Fi and improve the setup. This meant re-thinking both the mechanical layout of the board and the electrical design to fit in new features. We also designed housing to hold the electrode and reagent bottles. We swapped out the Arduino for an ESP32, and from the Nanostat design, we altered the ESP32-PICO-D4 to an ESP32-PICO-V3, which has a newer chip core (ECO V3) with more features than the D4’s ECO V1 chip, and has 16KB of RTC SRAM and security enhancements, which gives it better flash encryption and RSA support. This enhances our circuit’s ability to efficiently collect data and promotes the security of the data collected that is sent over a network to an external device. We also included a port for a 3.7V battery to power the circuit so a laptop would not be required.

Build
We optimized the printed circuit board based on the design of the Nanostat. Furthermore, we designed housing to hold the new printed circuit board, the electrode, and the holder in SolidWorks.

Test
We tested the new board with the same test solution from iteration 2 (CuSO4 and HCl), and then also with lectin-NOS, arginine, and treated blood. Using the NanoStat platform, we collected data wirelessly, sent over a network, and found the sensor data to be comparable to the previous iteration, but with a more compact and wireless design.

Learn
This design greatly reduced the size from the PCB from iteration 2 and increased its overall portability and ease of use through wireless data collection. In future designs, optimization of the electrode would greatly improve the user experience by minimizing interaction with blood. One way to achieve this is by utilizing capillary action by directly printing the working, counter, and reference electrode onto a membrane with metal ink and allowing reagents to flow through the membrane and into contact with the electrodes rather than pipetting directly onto them.