CLONING

1. Plasmid construction

We designed and assembled plasmids containing a T7 promoter and terminator, ribosome binding site (RBS), 10X His Tag, red fluorescent reporter gene mScarlet-I, and one of the five target exoglycosidase sequences from Akkermansia muciniphila (Jensen et al., 2024). The enzyme sequences were designated as follows: A1 - AmGH36A, A2 - AmGH35A, A3 - AmGH95B, B1 - AmGH110A, and B2 - AmGH20A. Figure 1 shows the constructs of 5 plasmids. All the sequences were synthesized through IDT.

2. Cloning

Our constructs were built using MEGAWHOP (A1) and Gibson Assembly methods (A2, A3, B1, B2). The constructs were first transformed into NEB Stable Competent E. coli cells, followed by colony PCR and gel electrophoresis to confirm the presence of the correct insert size. Plasmids from positive colonies were then purified by miniprep and sent for sequencing to verify sequence accuracy. We transformed each construct into T7 Express cells for protein production. We verified successful cloning results via gel electrophoresis, sequencing, and visualization of a pink color in the colonies. Glycerol stocks were prepared from the positive T7 colonies for long-term storage and future use.

2.1. Gel electrophoresis Results

NEB 1kb Plus DNA Ladder was used in all gel electrophoresis experiments.

2.2. Sequencing Results

2.3. T7 Expression Results

Our plasmid constructs contain mScarlet-I reporter, enabling direct visualization of successful protein expression as colonies will develop a pink color (Fig. 12 A-C, D). However, B1 colonies did not display the expected pink color (Fig. 12D). We hypothesize that the B1 likely encodes a product that is toxic upon expression, eliminating pink expressing colonies and leaving only white, non-expressing cells.

3. Protein Purification

3.1. Autoinduction media experiment

To optimize protein production, we compared two different autoinduction media formulations under multiple growth conditions. The first was Studier’s standard autoinduction media (Studier, 2005), and the second was a simplified formulation without the use of salts and metals, adapted from The Bumbling Biochemist (“Autoinduction for Recombinant Protein Overexpression,” 2024). The experiment was performed for all five enzyme plasmids and protein expression was monitored by the intensity of the pink color. Table 1 summarizes different testing conditions and representative results of this experiment. The data shown were obtained using cultures starting from B2 colonies, and similar trends were observed across the other enzyme constructs.

Cultures starting with a 1:500 dilution failed to produce any visible protein expression. Direct colony inoculation successfully led to protein expression in both media types. Overall, the simplified autoinduction media produced brighter pink cultures than Studier’s medium, with the exception of Conditions 3a and 3b, where Studier’s formulation yielded slightly stronger color. The most effective condition for both media was Condition 2, which consistently produced the brightest pink cultures observed after two days (Figure 13). Based on these results, we selected Condition 2b for use in subsequent protein purification experiments.

3.2. Protein purification

Using two protocols from NEBExpress and Irvine Valley College , we adopted our own protein purification protocol. Throughout the purification process, flow-through samples from each step were collected and analyzed by SDS-PAGE. Results of A1, A2, A3, B2 show a slightly larger band size than expected bands (Figure 14). This may be due to the incomplete denaturation process of protein although the samples were heated at 95℃ for 5 minutes. Cold-adapted enzyme from Colwellia (denoted as CA) displayed a smaller band size, which may explain its lack of activity in subsequent assays. Enzyme concentrations were assessed using a NanoDrop spectrophotometer, with the dye-specific mode confirming the amount of the mScarlet-I.

COLORIMETRIC SUBSTRATE ASSAY

Colorimetric assay was performed to evaluate whether our enzymes were functional. Each enzyme was tested against its specific substrate, which, upon cleavage, released a p-nitrophenol (pNP) group, turning the solution yellow. The concentration of p-nitrophenol can be quantified at absorbance 405 nm (Figure 15). Because we were unable to successfully purify B1 enzyme, we elected to use GH110 enzyme from Colwellia sp. to substitute for the following assays, which cleave the same sugar bond.

1. Enzymatic Function Verification

Small-scale colorimetric tests were performed to verify the enzymatic function, with results shown in Figure 16. Each well contained 20 μL of 10 mM substrate stock solution, 4 μL of enzyme, 4 μL of 10X GlycoBuffer 1 (500 mM sodium acetate, 50 mM CaCl2, pH 5.5), 4 μL of purified BSA (100 μg/mL). Negative control used 4 μL of heat-killed enzyme. The remaining volume was adjusted with water when needed to reach a total of 40 μL. The A1 and B2 reactions developed a yellow solution after incubation, showing their ability to cleave their respective substrates. However, A2, A3, and B1 using cold-adapted enzyme from Colwellia sp. show a visible color change.

2. Enzymatic Activity Quantification

2.1. Standard Curve

Figure 17 shows a pNP standard curve measured at A405 was generated using concentrations from 80 µM to 10 mM and fit to a linear regression of y = 0.0202x + 0.1236 (R2 = 0.9807).

2.2. Michaelis – Menten Kinetics Model

To quantify the enzyme activity, reactions were performed for our enzymes (A1, A2, A3, CA, and B2) across a range of substrate concentrations (0.15625 mM, 0.3125 mM, 0.625 mM, 1.25 mM, 2.5 mM, 5 mM) using 0.1 μM of enzyme. Absorbance at 405 nm was recorded every minute over a 3-hour period to track product formation. Each reaction was run in triplicate, and averaged results were used to calculate Michaelis – Menten constants (Vmax and Km) for enzyme efficiency. A1 and B2 enzymes showed a consistent activity across all substrate concentrations, further validating previous assay results. In contrast, A2, A3, and CA showed no detectable activity during the 3-hour period. The overall experiment was repeated in duplicate. Figure 18 shows a representative Michaelis – Menten plot for A1 at 24°C, with Vmax = 0.257 mM/min and Km = 1.2 mM, while A3 demonstrated a slower binding rate with Vmax = 0.133 mM/min and Km = 0.38 mM.

PORCINE HEMAGGLUTINATION ASSAY

1. Conversion and Crossmatch Testing of A Type Porcine Blood

To verify the blood types of the A and O type porcine blood, a blood typing test was performed using a commercial EldonCard with anti-A and anti-B antibody reagents. A type was crossmatched with plasma from O blood to verify incompatibility between blood types. Then, both types were crossmatched with human AB serum to verify incompatibility with human blood.

First, porcine blood type A was handled in different ways to serve as control groups. Type A packed red blood cells (RBCs) washed with our conversion buffer and crossmatched with human AB serum showed clear agglutination (Figure 19A). Similarly, unwashed type A RBCs and washed RBCs crossmatched with type O plasma also displayed agglutination (Figures 19B and red arrow in 19C), though some single RBCs were visible in Figure 19C (green arrow). As shown in Figure 19D, unwashed type A whole blood exhibited strong agglutination in the anti-A field and weak agglutination in the anti-B field. This confirmed that the porcine blood sample used was type A and also contained ⍺-Gal, which is structurally similar to the B antigen .

Porcine type A blood was treated with various enzyme combinations at 22°C. Samples treated with A1 enzyme alone and crossmatched with type O plasma (Figure 20A) displayed noticeably fewer clumps and more single cells, suggesting reduced agglutination compared to untreated controls. Treatment with A1, A2, and A3 combined (Figure 20B) did not show a clear difference in agglutination compared to treatment with A1 alone. This is consistent with colorimetric assays indicating limited or no activity for A2 and A3. When all five enzymes were combined and crossmatched with human serum (Figure 20C), mild agglutination persisted, indicating that the enzyme that targets a-gal/the B antigen was not effective. EldonCard testing (Figure 20D) of blood treated with the full enzyme mix showed weaker agglutination in the anti-A field while the anti-B field remained similarly weak compared to control groups. A1 was confirmed to exhibit the most effective antigen-cleaving activity, while additional testing with multiple replicates or lectin-based assays for H-antigen detection would be required to further validate antigen conversion efficiency of A2 and A3.

2. Conversion and Crossmatch Testing of O Type Porcine Blood

To serve as a positive control for enzymatic conversion, porcine type O blood samples were tested at 22°C to confirm the absence of A and B antigens, as well as the presence of ⍺-Gal. Washed O-type RBCs crossmatched with human serum (Figure 21A) showed minimal agglutination, while unwashed O-type RBCs (Figure 21B) displayed similar results. When tested on an EldonCard (Figure 21C), the O-type blood showed weak agglutination in the anti-B field, which may be due to nonspecific interactions.

Porcine type O blood was treated with the cold-adapted enzyme (CA) and B2 at different temperatures. Samples treated with CA + B2 at 22°C (Figure 22A) showed visible clumping, indicating little to no enzymatic cleavage. Treatment with C alone at 4°C (Figure 22B) resulted in more single cells, suggesting that the cold-adapted enzyme may perform more effectively at lower temperatures. However, when CA + B2 were combined at 4 °C (Figure 22C), no significant qualitative difference was observed compared to treatment with CA alone. EldonCard testing (Figure 22D) of the same sample showed very minimal reduction in agglutination in the anti-B field, which was also seen in untreated A-type RBCs in Figure 19D. It remains uncertain whether the porcine blood used expressed the extended B antigen targeted by B2. While pigs have the same gene B3GALNT1 that encodes the glycosyltransferase responsible for this antigen, cleavage by B2 is difficult to determine qualitatively by reduction in agglutination. Additionally, because the serum sample was pooled from multiple donors, it is unclear whether it contained antibodies against the extended B structure. Given that the cold-adapted enzyme CA showed no verified activity in prior assays, the observed reduction in agglutination could reflect either minimal B2 activity or improved CA performance at 4°C. Further testing using B2-only controls, verified extended-B antigen samples, and multiple biological replicates will be necessary to confirm enzyme specificity and function.

DRY LAB

Due to the strict temperature requirements our kit would need to be stored at, considering cold-adapted enzymes was a crucial step in this project. There is, however, a lack of literature regarding cold-adapted enzymes for blood group modifications. Therefore, we developed our own code with the goals of identifying potential cold-adapted enzymes and streamlining the broader process of searching and organizing enzymes from large databases.

1. Computational Identification of Cold-Adapted Enzymes from CAZy and NCBI

To identify cold-adapted enzymes, the CAZy database was cross-compared to the NCBI database. Manual shortlisting to find enzymes within their respective families was successful for enzymes GH110 and GH35. For enzymes GH20, GH36, and GH95, an automated filtration process was done in Colab. The general process is outlined in Figure 23.

The enzymes were further validated for use by shortlisting any that could function under 10 °C. Code was confirmed to be working on smaller databases, as two psychrophile enzymes were found that fit our parameters for alpha-galactosidase: Polaribacter sejongensis and Colwellia sp. PAMC 2182. Constructs shown in Figure 24. Both enzymes were ordered through IDT, with a codon-optimized version of Colwellia sp. PAMC 2182. Gibson Assembly was used to insert the sequence into the backbone. The constructs were first transformed into NEB Stable Competent E. coli cells, which was followed by colony PCR and gel electrophoresis to confirm their presence, which were then miniprepped and nanodropped. Only Colwellia sp. PAMC 2182 was successfully cloned into the backbone as shown in Figure 25 and the sequence was confirmed as shown in Figure 26.

2. Temperature-Based Sequence Alignment and Structure Analysis

We then aimed to find similar regions between the three temperature groups of enzymes, thermophilic, mesophilic, and psychrophilic, that allow for cold temperature functionality. This was to recognize mutation sites in our original 5 enzymes in order to make them cold-adapted.

Using CLUSTAL sequence alignment, we identified 5 enzymes within each temperature range, and found there were no significant regions of similarity for the thermophilic and psychrophilic enzymes. We then ran protein sequences within the Alteromonadales order, the order of Colwellia sp. PAMC 2182 found in project 1. By doing this, we were able to compile enzymes from 5 families, the Colwelliaceae (other than Colwellia sp. PAMC 21821 ), Alteromonadalaceae, Pseudoalteromonadaceae, Psychromonadaceae, and Shewanellaceae, for a total of 25 enzymes in the GH110 family. Sequence alignments between the enzymes, including Colwellia sp. PAMC 2182, was run in CLUSTAL. Between Colwellia sp. PAMC 21821 and Akkermansia muciniphila, distinct regions of similarity were found as shown in Figure 27.

HARDWARE: Development of UNIglobin Enzymatic Blood Conversion Kit

We developed a hardware conversion kit that would detect residual antigens left over after a round of enzyme cleavage to ensure safety before a transfusion. This kit includes enzymes required for the blood conversion, a miniaturized biosensor to measure antigen cleavage, and a filter that removes leukocytes and plasma while keeping red blood cells and platelets.

1. Lyophilized Enzymes

Small aliquots of each enzyme (10 µL) were mixed with 50% sucrose and lyophilized using a lyophilizer, then stored at – 20°C. Their activity was quantified and compared with recently purified enzymes using the Michaelis – Menten model to evaluate whether lyophilization is a viable method for maintaining enzymatic function over long storage periods. Among the tested enzymes, A1 and B2 retained measurable activity; however, the A1 dataset showed high variability and could not be accurately fitted to least squares regression to generate Michaelis – Menten graph. Therefore, Vmax and Km of A1 were not determined. In contrast, B2 produced a clear reaction with Vmax = 0.39 mM/min and Km = 1.42 mM, indicating that the lyophilized enzyme required a higher substrate concentration to reach half of its maximum velocity compared to recently purified B2.

2. Miniaturized Electronic Antigen Biosensor (MEAB)

To complement our enzymatic conversion system, we developed a miniaturized electronic biosensor capable of detecting nitric oxide (NO) released when our lectin is bound to O type blood. The biosensor integrates a custom-built circuit that measures peak current from NO redox reactions, which correlates with the amount of antigen bound. We successfully assembled the biosensor and validated its functionality at the circuit level. Figure 29 shows the biosensor phototype.

To test functionality of the sensor, we conducted preliminary testing using voltammetry with a solution of CuSO4, HCl, and water. As shown in Figure 30, an increasing voltage was applied while the resulting current was measured, implying that the sensor could detect redox reactions. This signal will be used to determine the rate at which the nitric oxide synthase forward reaction occurs, which can then be correlated to the amount of antigen bound.

Similarly to the process of cloning enzymes, we successfully cloned the Aleuria Aurantia Lectin (AAL) sequence together with Bacillus subtilis Nitric Oxide Synthase (bsNOS) sequence in our plasmid backbone. Figure 31 shows the designed sequence map similar to the 5 enzymes, Figure 32 presents the sequencing confirmation for the construct, Figure 33 gives a visualization of the protein expression with T7 express cells. This is then purified to obtain AAL+bsNOS protein using in our MEAB.

Cyclic voltammetry testing was performed to test the NO detection of the MEAB using varying AAL- bsNOS complex concentrations. Figure 34 shows current per unit of voltage vs. time for three conditions - porcine O blood, porcine A blood, and enzymatically treated porcine blood. In a paired t-test, type O (positive control) and treated blood were highly significant compared to type A (negative control) (p<0.001).

3. Filtration Bag

Access to centrifuges for blood processing is limited in low-resourced communities, and costly leukoreduction filters are not practical. Our two-part filter provides a low-cost solution: SCOBY cellulose binds leukocytes, while a PTFE membrane allows plasma to pass through. After filtration, red blood cells and platelets remain in the bag for collection, reducing the probabilities of immune reactions and disease transmission due to residual antibodies and leukocytes.

4. SCOBY

Immunofluorescence staining with DAPI and anti-CD45 antibody staining on a SCOBY sample to assess leukocyte binding (Figure 36). However, the number observed was lower than we expected. Further optimization is needed to maximize the effective surface area of SCOBY or to enhance leukocyte binding.

SCOBY was treated using different processing techniques: air-drying, lyophilizing, and freezing/lyophilizing to increase binding effectiveness. Scanning electron microscopy images in Figure 37 show freezing and lyophilizing SCOBY greatly alters surface topography, increasing the surface area that leukocytes can bind to. We also determined that the pore size of SCOBY ranges from 0.2 µm - 1.5 µm, which suggests that SCOBY could be used to filter small proteins and antibodies within plasma.

5. PTFE

We obtained our PTFE membrane from JVLAB but accidentally purchased the hydrophobic version instead of the hydrophilic one required for plasma separation. Due to time constraints, we were unable to replace it, though the design remains compatible with hydrophilic PTFE for future testing.

FUTURE WORK

Moving forward, we plan to build on our progress by refining several aspects of the project. In the wet lab, future experiments are needed to troubleshoot the expression of enzyme B1. Alongside this, we aim to optimize the conditions for purifying larger amounts of enzymes and investigate the lack of activity observed in enzymes A2 and A3. Further work into testing cold-adaptive and lyophilized enzymes to evaluate their activity and stability under different conditions will need to be performed to improve future versions of the blood conversion kit. On the hardware side, the next steps include obtaining the appropriate hydrophilic PTFE membrane and developing improved prototypes of the conversion kit, which will then undergo systematic testing and optimization.

REFERENCES

[1] Autoinduction for recombinant protein overexpression. (2024, June 7). The Bumbling Biochemist. https://thebumblingbiochemist.com/365-days-of-science/autoinduction/

[2] Jensen, M., Stenfelt, L., Ricci Hagman, J., Pichler, M. J., Weikum, J., Nielsen, T. S., Hult, A., Morth, J. P., Olsson, M. L., & Abou Hachem, M. (2024). Akkermansia muciniphila exoglycosidases target extended blood group antigens to generate ABO-universal blood. Nature Microbiology, 9(5), 1176–1188. https://doi.org/10.1038/s41564-024-01663-4

[3] Studier, F. W. (2005). Protein production by auto-induction in high-density shaking cultures. Protein Expression and Purification, 41(1), 207–234. https://doi.org/10.1016/j.pep.2005.01.016