As outlined in our project design, we aim to verify both the binding activity of our binding domain to the target material and the antimicrobial efficacy of our functional domain, defensins. The overall project workflow is illustrated in Fig. 0. Our approach involves screeninng and validating the binding and functional domains separately before combining them. To screen and validate the binding domain, we linked it to a chromoprotein, enabling visual confirmation of its successful attachment to the target material. This fusion of the binding domain and chromoprotein is referred to as BDC. For the functional domain, we first expressed defensin independently then expressed SUMO-linked defensin for enhanced stability. Ultimately, our goal is to fuse both domains into a single protein, where the binding domain is connected to SUMO, which is in turn linked to the functional domain. This fusion protein is designed to impart antimicrobial properties to first-aid materials.

Construction of binding domain-chromoprotein expression strain

In Cycle I, our goal is to obtain a fusion protein composed of a binding domain linked to a chromoprotein—namely mTurquoise, sfGFP, fwYellow, gfasPurple, mRFP, and eforRed—to enable visualization of the ability of the binding domains—specifically CBM2, CBM3, CBM5, CBMxx, α1, and α2; respectively—to integrate with the target material.

We performed codon optimization of CBM2, mTurquoise, α1, CBMxx, and gfasPurple, obtaining them through DNA synthesis. The fragments CBM3, sfGFP, fwYellow, mRFP, α2, and eforRed were obtained via PCR amplification using the primers Alyssa-F1/R1, Qiao-F/R, Carl-F/R, Arendelle-F/R, Hugo-F2/R2, and Hugo-F1/R1 from vector A038, vector pKeystone012, vector J23102-11 fwYellow, vector pSEVA321-rhaR-rhaS-RiboJ-RFP, pET28a-α2β1-HNP1, and pKeystone009, respectively. We obtained the vectors A038 from iGEM Team LINKS China 2021; pKeystone012 and pKeystone009 from iGEM Team Keystone 2023; and J23102-11 fwYellow from LinkSpider Lab (Shenzhen, China). The DNA fragment encoding CBM5 was obtained through SOE PCR using the primer pairs CBM5MMP-F1/R1, CBM5MMP-F2/R2, and CBM5MMP-F3/R3. The primer sequences are shown in experiments.

All final binding domain DNA fragments and chromoprotein DNA fragments, except for α2 and eforRed, were ligated into the digested plasmid pLINKS2400 expression vector by Golden Gate assembly; α2 and eforRed were assembled into pLINKS2400 through Gibson assembly. Escherichia coli DH5a competent cells were transformed and screened in LB medium containing kanamycin. The plasmids were then confirmed by colony PCR and DNA sequencing for the presence of the correct sequence. The correct plasmids were named pLINKS2407, pLINKS2408, pLINKS2410, pLINKS2405, pLINKS2406 and pLINKS2431 for plasmids carrying CBM2-mTurquoise, CBM3-sfGFP, CBM5-fwYellow, α1-mRFP, α2-eforRed and CBMxx-gfasPurple, respectively. These plasmids were then extracted and transformed into E. coli BL21 (DE3) competent cells. pLINKS2431 is later transformed into SHuffle T7 cell for expression of CBMxx-gfasPurple in supernatant.

For optimal conditions, E. coli BL21 (DE3) and E. coli SHuffle T7 cells containing the correct plasmids pLINKS2407, pLINKS2408, pLINKS2410, pLINKS2405, pLINKS2406, and pLinks2431 were cultured overnight in fresh LB medium at 37 °C. Take 400 µL of the overnight culture and inoculate it into 400 mL of LB medium containing kanamycin. Incubate the culture at 37°C until OD-600 reaches 0.6-0.8. Then add IPTG with a final concentration of 0.3 mM and continue incubation at 30°C with shaking at 250 rpm for 3.5 hours. For α2-eforRed and CBMxx-gfasPurple fusion proteins, we observed suboptimal expression under these conditions. Thus, we modified the protocol by using a lower IPTG concentration of 0.1 mM and reducing the temperature to 20°C, followed by an extended induction period of 14 hours. After induction, the cells were collected by centrifugation at 4 °C and 8500 rpm for 3.5 minutes. The pellet was suspended in 40 ml of 20 mM Tris-HCl, pH 8.0, and sonicated on ice with an appropriate tip at 25 Hz for 30 minutes (3 seconds working, 6 seconds free) until the cells were lysed. The supernatant of the cell lysates was collected by centrifugation at 4 °C and 8,500 rpm for 10 minutes. SDS-PAGE was then performed to analyze correct protein expression, and fluorescence was detected under yellow-light to visualize protein expression.

Verification of binding domain function

To test the binding domain protein fusion with cellulose, two rows of proteins, including two of the BDCs targeted to bind to cellulose (CBM2-mTurquoise and CBM3-sfGFP) and a fluorescent protein control (mTurquoise and sfGFP), were added to a 48-well plate, all suspended in 20mM Tris-HCl. Row 1 contained gauze soaked in the protein solution, while Row 2 contained the protein only. After soaking, washing with isosopranol and drying (more details in experiments), the results indicate that CBM2 and CBM3 are capable of binding with cellulose, as shown by the brighter fluorescent after washing(Fig. 2A).

For BDCs intended to bind to collagen/chitosan and sodium alginate hydrogel mixture, we conducted procedures as outlined in the experiments. It is important to note that the α1 and α2 binding domains are intended to interact specifically with type IV collagen. However, due to budget constraints in our lab, we assessed their binding ability using type II collagen instead. This choice may explain the comparatively lower brightness observed in the α binding domains incorporated into the sodium alginate and collagen mixture after washing. We opted to use α2 to verify the functionality of our combined domain (the binding domain fused with the functional domain) because of its relatively greater brightness. CBM5 and CBMxx, on the other hand, both displayed moderate ability to bind to their target materials as indicated by their relative deepness in color compared to the control group. Thus, we continued to use both binding domains to verify our combined domain functions.

After verifying the binding domain function, we next wanted to verify if the antimicrobial domain is functioning normally. Due to the extremely low content of defensins in human blood and the high cost of chemical synthesis, we used Escherichia coli to express the selected four human recombinant defensins (HNP1, HNP4, HD5 and HBD3). However, the prokaryotic expression of defensins is not ideal, either the expressed defensins lack function, or the leakage of expressed defensins kills the cells.

In addition, we also synthesized the immature forms of two defensins (preproHNP1 and preproHD5). Xie et al reported that the preproHNP1 expressed in BL21(DE3) can be cleaved by the endogenous serine protease in E. coli to produce the mature HNP1, so E. coli can accumulate more defensins after reaching a higher cell density.

Based on this, we replaced the chassis with SHuffle T7, so that the defensins expressed in E. coli can be correctly folded and form disulfide bonds, thereby improving the stability of the defensins.

Construction of defensin expression strains

Defensin fragments were obtained by SOE-PCR as described in the protocol. Those fragments were ligated into the digested plasmid pLINKS2400 expression vector by golden gate assembly. Escherichia coli DH5a competent cells were transformed and screened in LB medium containing kanamycin. The plasmids were then confirmed by colony PCR (figure 3a) and DNA sequencing for the presence of the correct sequence. The correct plasmids were named pLINKS2401, pLINKS2402, pLINKS2403, and pLINKS2404 for plasmids carrying HNP1, HNP4, HD5, and HBD3, respectively. These plasmids were then extracted and transformed into E. coli SHuffle T7 competent cells.

For optimal conditions, E. coli SHuffle T7 cells containing correct plasmid pLINKS2401 ~ pLINKS2404 were cultured in fresh LB medium for overnight at 37 °C. Next day, the overnight cultures were subcultured in 4 ml LB medium with Kanamycin on a rotary shaker (220 rpm) at 30 °C for 3.5 h until OD600 reached 0.6~0.8. For expression of the defensins, E. coli SHuffle T7 cells containing pLINKS2400 were then cultured in 300 ml LB medium at 37 °C for 3 h until OD600 reached 0.6 to 0.8, the expression of the gene was induced by addition of 0.3 mM IPTG and conducted for 3.5h at 30 °C. The cells were collected by centrifugation at 4 °C, 12000 rpm for 3 min. The pellet was suspended in 500 ul 20mM Tris-HCl，pH8.0 and sonicated in ice with an appropriate tip at 25 Hz for 4 min (3 s working, 6 s free) until cells were lysed. Tris-Tricine-SDS-PAGE was then performed to analyze the expression of proteins in the cell, as shown in Figure 3b. The result shows that for the defensins without the pro segment, only HNP1 were expressed, but only in small amounts. Moreover, we observed that these strains continued to rapidly increase in cell number after IPTG induction, which is contrary to the expected results. If the produced defensin is toxic to E. coli, the growth of cells would be rapidly inhibited. We speculate that there are two reasons why the defensins are not produced and is not toxic. One reason is that the molecular weight of the defensin is too small (less than 5 kDa), making it easily degraded by endogenous proteases. The second reason is that the addition of 10 amino acids at the N-terminus, including a 6×His tag, affects the correct folding of the defensins. Therefore, we have decided to change our strategy and use protease cleavage to release the defensins, and after N-terminal protease cleavage, it does not contain any amino acid residues.

Meanwhile, ProHNP1 and ProHD5 were both produced, but had not been cleaved by protease naturally expressed inE. coliSHuffle T7 into mature form of defensins and thus didn’t inhibit the growth of E. coli.

Due to the unsuccessful attempts to express defensins directly in Escherichia coli, we decided to produce defensins using an enzymatic cleavage method. The SUMO tag, known for enhancing solubility and leaving no amino acid residues at the N-terminus after cleavage, has been widely utilized in the production of antimicrobial peptides (AMPs) in previous studies (Zhu et al., 2021). Building on this, we fused the SUMO tag to the C-terminus of the functionally validated binding domain CBM3 and the N-terminus of the functionally unvalidated HNP1, aiming to generate defensins through enzymatic cleavage.

Construction of CBM3-SUMO-Defensins strain

We obtained CBM3! fragment through amplification of A038 using primers Alyssa-F1 and Alyssa-R1, and SUMO! from pET28a-BsaI site1-SUMO-BsaI site2 using primers Henry-F1 and Henry-R1. Defensin fragments were obtained via amplification of their corresponding vectors pLINKS2401 ~ pLINKS2404. The CBM3!, SUMO!, and defensin fragments were ligated into the digested plasmid pLINKS2400 expression vector via golden gate assembly. The process of transformation and confirmation is the same as above. The confirmed plasmids were named pLINKS2416, pLINKS2417, pLINKS2418, and pLINKS2419, corresponding to plasmids carrying the HNP1, HNP4, HD5, and HBD3 genes（fig. 4a）, respectively. These verified plasmids were then extracted and transformed into E. coli SHuffle T7 competent cells. The IPTG-induced expression of CBM3-SUMO-defensins was validated in the same manner as the defensins. SDS-PAGE analysis revealed the presence of the correct bands (Fig. 4b). Subsequently, we scaled up the fermentation to 400 mL. Once the optical density (OD) reached 0.6–0.8, we induced expression with 0.1 mM IPTG at 20°C for 12 hours. Afterward, we lysed the cells and purified the target protein from the cell lysate using a Ni-NTA affinity chromatography column.

Verification of CBM3-SUMO-Defensins Function

4 types of CBM3-SUMO-Defensins were subjected to salt removal by gradient dialysis, followed by cleavage with recombinant Ulp1 (Beyotime, P2312S). The results from SDS-PAGE electrophoresis showed a slight decrease in the molecular weight of the target protein (Fig. 5 A), indicating successful removal of ~4 kDa defensin peptides. Since CBM3-SUMO-Defensins would ultimately be incorporated into wound dressing products in a domain-bound form rather than as individual defensins, we did not further purify the defensins. Instead, we utilized the enzyme-cleaved CBM3-SUMO-Defensins (designated as CBM3-SUMO↓Defensins) for antimicrobial assays. As depicted in Fig. 5b, Escherichia coli and Staphylococcus aureus were selected as representatives of Gram-positive and Gram-negative bacteria, respectively. The CBM3-SUMO↓Defensins cleaved by Ulp1 enzyme exhibited antimicrobial activity against both strains, while the uncleaved CBM3-SUMO-Defensins showed no antimicrobial activity. This suggests that we successfully produced active defensin molecules using the fusion protein cleavage approach.

Subsequently, during our interview with Dr. Dai, we learned that the market approval of antimicrobial drugs requires the determination of Minimum Inhibitory Concentration (MIC). Therefore, we utilized the microdilution method to determine the MIC values of four types of CBM3-SUMO-Defensins. For specific details, please refer to our measurement section. Initially, we examined the 24-hour growth curves of Staphylococcus aureus with the addition of CBM3-SUMO↓Defensins. Within the 0–8 hour range, all four types of CBM3-SUMO↓Defensins exhibited antimicrobial activity (Fig. 6A). We selected the 8-hour time point to define the MIC values against Staphylococcus aureus. At this point, the MIC50 values for CBM3-SUMO↓HNP1, CBM3-SUMO↓HNP4, CBM3-SUMO↓HD5, and CBM3-SUMO↓HBD3 were 0.74 μM, 0.368 μM, 1.475 μM, and 1.001 μM, respectively. Additionally, the MIC90 values for CBM3-SUMO↓HNP4/HD5 were 0.735 μM and 1.475 μM, respectively. These values are close to the MIC values reported in the literature for the four defensins (Wei et al., 2009).

It is worth mentioning that when the concentrations of the four types of CBM3-SUMO↓Defensins were reduced to 185 nM, 184 nM, 184 nM, and 125 nM, they were able to promote the growth of Staphylococcus aureus (Fig. 6 B). This suggests that the non-defensin portion of CBM3-SUMO↓Defensins may serve as a nutrient for bacteria, providing amino acids upon hydrolysis. After 8 hours, high concentrations of CBM3-SUMO↓Defensins were able to promote the growth of Staphylococcus aureus (Fig. 6 A). We speculate that this is due to the short peptide nature of defensins, which makes them susceptible to degradation by proteases, resulting in a shorter effective period. After defensins become inactive after 8 hours, CBM3-SUMO↓Defensins act as nutrients that promote bacterial growth. Therefore, in our antibacterial dressings, a higher concentration is not necessarily better. We believe that in the future, we can choose smaller Binding domains or optimize the sequence of natural Binding domains to increase the proportion of defensin molecules as much as possible while keeping the molar concentration of the fusion protein constant, thereby reducing the non-defensin portion to avoid providing nutrients to bacteria and improving the MIC. When we consulted Dr. Ren, he informed us that all antibacterial dressings need to be changed within 12 hours. This is not a defect, but it should be stated in the product instructions.

After separately validating the effectiveness of the binding domain and defensins, we expanded CBM3-SUMO↓Defensins to other binding domains by replacing CBM3 with the functionally validated α2, CBM5, and CBMxx. This allows for binding to collagen, chitosan, and alginate, thereby creating a more versatile collection of antibacterial dressings, which enhances the potential of our first aid kit to address more complex situations.

Construction of binding domain-SUMO-HNP1 strain

We assembled the binding domain fragments from DBTL Cycle I with the SUMO and HNP1 fragments from DBTL Cycle 2 using Golden Gate assembly, to obtain SHuffle T7 strains that express α2-SUMO-HNP1, CBM2-SUMO-HNP1, CBM5-SUMO-HNP1, and CBMxx-SUMO-HNP1 (Fig. 7a). We then used the same fermentation method as for CBM3-SUMO-HNP1 to produce the corresponding fusion proteins (Fig. 7b). Among these, CBM5-SUMO-HNP1 and α2-SUMO-HNP1 were successfully expressed in the supernatant, but the yield was relatively low. In contrast, CBMxx-SUMO-HNP1 did not express, which we speculate may be due to contamination. Due to time constraints, we only tested the successfully expressed CBM5-SUMO-HNP1 and α2-SUMO-HNP1.

Verification of binding domain-SUMO-HNP1

We used the same cell lysis purification method as CBM3-SUMO-HNP1 to purify CBM5-SUMO-HNP1 and α2-SUMO-HNP1. Due to their poor solubility, we increased the amount of the upper column by two-fold. Subsequently, we performed SUMO cleavage and antibacterial tests using the same procedure. Both CBM5-SUMO↓HNP1 and α2-SUMO↓HNP1 demonstrated inhibitory effects on Escherichia coli and Staphylococcus aureus. Therefore, we believe that the designed Binding domain-SUMO-Defensins are scalable.

After validation by DBTL III and IV, the feasibility of our Binding domain-SUMO-Defensins in antimicrobial activity has been demonstrated. Next, we need to move it into practical application by attaching the defensin to wound dressings through the binding domain and releasing the defensin for antimicrobial activity through SUMO cleavage.

However, we will face a practical challenge, that is how to use the SUMO enzyme? In preliminary experiments, we used commercially purchased recombinant yeast Ulp1, where 1 uL (10 U) can be used to cleave 20 ug of fusion protein. Based on our MIC determination results, it is estimated that 1-2 enzyme cleavage reactions are needed to produce one piece of antimicrobial dressing, while the cost of the SUMO enzyme is approximately 10-20 RMB. The cost of such antimicrobial dressings is too high that no consumer will choose them, especially in war zones and disaster areas that require humanitarian aid.

Expression and enzyme digestion verification of rtUlp1

To reduce costs, we decided to produce recombinant Ulp1 from Escherichia coli to lower costs. Based on this, we constructed a truncated Ulp1 strain (rtUlp1) that retains the enzyme's active center (Wang et al. 2016) and achieved soluble expression in a 400 mL flask (Fig.9A).

Subsequently, we performed the enzyme digestion reaction under the same conditions as the purchased Ulp1 enzyme. This experiment was carried out simultaneously with the optimization experiment of HNP1 (link to modeling), so we chose CBM3-SUMO-HNP1AWK as the substrate and conducted antibacterial experiments on E. coli. The results are shown in Fig. 9B. After our expressed rtUlp1 processed CBM3-SUMO-HNP1AWK, it displayed the same inhibition zone as the CBM3-SUMO↓HNP1AWK processed by Ulp1, indicating that our expressed rtUlp1 has SUMO enzyme cutting activity.

In this experiment, we took 200 mL of the fermentation broth from the 400 mL shaker flask for purification, and obtained 3 mL of rtUlp1 (concentration not tested). We then added 1 μL to successfully achieve the enzyme-mediated release of the defensin. Therefore, we believe that the cost of producing our own Ulp1 is extremely low, and it can be mass-produced for wound dressings. To extend the shelf life of rtUlp1, we decided to store it in powder form with water, together with the wound dressing that connects the Binding domain-SUMO-Defensins, as described in the description.

Binding domain-rtUlp1 design

After showcasing the above results and the product to Mr. Dawa again, he was highly impressed and provided valuable feedback: the current usage is too cumbersome, as the injured person or medical staff would need to carefully mix the water with the rtUlp1 powder, and then combine the rtUlp1 solution with the wound dressing. This would be difficult for injured hands or in urgent situations, making it less applicable than traditional dressings and antibiotics. Therefore, we decided to optimize the design.

Our solution is to adopt the same immobilization strategy as the defensins, by fusing rtUlp1 with the Binding domain. The rtUlp1 can be placed at either the N-terminus or C-terminus of the fusion protein, and then combined with the Binding domain-SUMO-Defensins on the wound dressing through freeze-drying. By immobilizing the rtUlp1 via the binding domain, the distance between rtUlp1 and the SUMO tag will be reduced, which may enhance the enzyme cutting activity of rtUlp1, further reducing the production cost.