Preliminary Characterization of the Protocatechuic Acid Operon and Bile Salt Sensing Device

Design

We selected protocatechuic acid (PCA), a natural metabolite of green tea, as the inducer for GLP-1 synthesis, aiming to leverage the daily consumption of green tea as a healthy routine to assist in glucose management and weight loss for people. Since the PCA-responsive operon [1] ( BBa_25N5FWAK) is a relatively new iGEM part, systematic characterization was necessary before its integration into the circuit. Additionally, we chose sodium cholate, a bile salt present at characteristic concentrations in the gut, as the second input signal. We identified a bile salt sensing device ( BBa_K1962010) from the iGEM registry for this purpose.

For initial characterization, we used GFP (BBa_K2717024) as a reporter in place of the downstream split intein-transcription factor halves. This allowed us to assess the expression profiles of both operons under preliminary induction conditions derived from literature [2], with results quantified as normalized fluorescence intensity.

Figure 1. Schematic diagram of the initial operon characterization in Cycle 1.

Build

We commissioned the synthesis of codon-optimized sequences for both the PCA operon and the bile salt sensing device. The synthesized fragments and the downstream GFP reporter were amplified by PCR and assembled into the pSB3K3 plasmid backbone. After verifying the fragments by gel electrophoresis, the constructs were assembled via Gibson assembly, transformed into Escherichia coli TOP10 competent cells, and positive clones were selected on kanamycin-containing LB plates. Successful construction of the target plasmids was confirmed by DNA sequencing.

Figure 2. Construction of expression plasmids for the two operons. (A) Results of electrophoretic bands of GFP (881 bp) and the pSB3K3 plasmid backbone harboring the PCA-responsive operon (~3.3 kb) after PCR. (B) Plasmid map of pSB3K3-J23117-PcaV-P_LV-GFP. (C) Sequencing results of pSB3K3-J23117-PcaV-P_LV-GFP. (D) Results of electrophoretic bands of pSB3K3-J23101-RamA-P_acrRA-GFP (~5 kb) after PCR. (E) Plasmid map of pSB3K3-J23101-RamA-P_acrRA-GFP. (F) Sequencing results of t pSB3K3-J23101-RamA-P_acrRA-GFP.

Test

The sequence-verified plasmids were expressed in TOP10. For the PCA operon, we tested PCA concentrations at 0, 100, 200, 300, 400, 500, 550, 600, 650, 700, 800, and 1000 μmol/L based on literature. For the cholate-sensing device, sodium cholate was applied at 0, 25, 50, 100, and 150 μmol/L. Inducers were added when cultures reached an initial OD₆₀₀ of 0.02, followed by 6-hour incubation at 37 °C with shaking at 220 rpm. Cells were then harvested, washed with 1× PBS, and analyzed for fluorescence (excitation: 485 nm, emission: 520 nm) and OD₆₀₀. Finally, we evaluated induction efficiency as normalized fluorescence intensity.

Figure 3. Characterization of the two operons after 6 hours of induction. (A) Normalized fluorescence intensity under different PCA concentrations. (B) Normalized fluorescence intensity under different sodium cholate concentrations.

Learn

For the PCA operon, fluorescence intensity increased with PCA concentration but saturated rapidly, with significant leakage observed in the absence of PCA and a relatively low maximum fluorescence level. We hypothesize that insufficient expression of the PcaV repressor led to weak repression of the P_LV promoter, combined with excessively high sensitivity [3]. Therefore, we planned to replace the upstream constitutive promoter with a stronger version to elevate repressor levels.

In the bile salt induction experiment, the fluorescence/OD₆₀₀ ratio remained very low and showed no significant response to increasing sodium cholate concentrations, indicating minimal induction of the device. This is likely due to the weak basal activity of the P_acrRA promoter, resulting in negligible background GFP expression. Alternatively, the device may be incompatible with the E. coli expression system. So optimizing expression by modifying the RBS sequence and reassess its performance could be a helpful strategy.

Optimizing Expression Levels of Operon Elements

Design

To address the issues identified in Cycle 1, we optimized the expression strength of the original systems by modifying the constitutive promoters and RBS sequences, thereby further validating operon functionality. For the PCA-responsive operon, we replaced the upstream promoter of the PcaV repressor with higher-strength promoters J23101 and J23105 (strength: J23101>J23105>J23117) to enhance repression of the downstream P_LV promoter and reduce leakage. We also optimized the induction time for the PCA operon to assess system responsiveness and provide reference for in vivo applications.

For the bile salt sensing operon, we optimized the RBS sequences upstream of both the transcription factor and the reporter gene, designing three additional RBS combinations to enhance downstream expression levels, improving the dynamic response range and signal-to-noise ratio under sodium cholate gradient induction.

Figure 4. Schematic diagram of the operon optimization in Cycle 2.

Build

Using the successfully constructed plasmids from previous cycles as templates, we performed site-directed replacement of promoters and RBS sequences via primer design and circular PCR. The modified constructs were assembled, transformed, and screened for positive clones as described in Cycle 1. This yielded three PCA-responsive operon variants with different expression strengths and two bile salt sensing device variants. Despite repeated attempts including Gibson assembly with re-amplified fragments, using high-fidelity PCR enzymes, and multiple rounds of circular PCR with point-mutation templates, we were unable to successfully replace the RBS32 sequence upstream of the transcription factor with RBS30. Consequently, further characterization based on this specific modification was discontinued.

Test

Since the sodium cholate operon primarily functions as a “localizer” and the intestine provides a sustained supply of cholate, while PCA serves as the key input signal for limited induction, we first optimized the induction time for the PCA operon. Preliminary tests led us to select TOP10 strains carrying the J23105 promoter for a 9-hour induction experiment with PCA concentrations of 0, 250, and 400 μM, and take samples at regular intervals to measure fluorescence and OD₆₀₀. Based on the results, we selected 4 hours as the final induction time to ensure sufficient bacterial growth for subsequent processing steps such as centrifugation, pelleting, and washing. We then evaluated the induction performance of the five plasmid variants under this 4-hour induction period, using PCA and sodium cholate concentration gradients similar to those in Cycle 1.

We found that normalized fluorescence intensity gradually increased with PCA concentration before reaching saturation. Higher expression levels of the PcaV repressor resulted in lower leakage, but this was accompanied by reduced sensitivity to changes in PCA concentration, further validating our conclusions from the first iteration. For the cholate-sensing device, modifying the RBS sequences downstream increased leakage expression but did not yield significant cholate-induced activation. Subsequent tests with higher cholate concentrations still showed no notable increase in fluorescence signal and even negatively affected bacterial growth.

Figure 5. Characterization of the two optimized operons. (A) Preliminary experiment on PCA induction time of the operon regulated by promoter J23105 over 0-9 hours. (B) Measurement of PCA-induced fluorescence intensity with constitutive promoters of various strengths at 4-hour induction time. (C) Measurement of BS-induced fluorescence intensity with different RBS combinations at 4-hour induction time for the constructed plasmids.

Learn

Based on these results, we selected the J23105 promoter to regulate PcaV repressor expression and retained the original RBS combination for the cholate operon in subsequent AND-gate construction. This strategy aims to minimize leakage in the absence of inducers, which could interfere with validation of the AND-gate logic. The PCA-responsive operon performed satisfactorily, whereas the cholate-sensing device continued to underperform. We suspect that the issue may stem from toxicity caused by overexpression of the heterologous transcription activator RamA in E. coli, or from insufficient tolerance mechanisms in the host toward cholate, affecting its metabolic regulation and induction [4]. Future work may require deeper investigation into the cholate regulatory mechanism or substitution with alternative gut-specific molecular sensors with improved performance.

Construction of the Final AND-gate Circuit with the Characterized Operons

Design

Following the individual characterization of the operons, we proceeded to construct the final AND-gate circuit ( BBa_2581ND93) by linking them to orthogonal split intein-transcription factor halves. In accordance with the original reference, we cloned the upstream AND-gate sensing module and the downstream reporter module into two separate plasmids, pSB3K3 and pSB4A3 respectively. These plasmids were subsequently co-transformed into E. coli TOP10 for functional characterization of the AND-gate circuit.

Figure 6. Schematic diagram of the AND-gate sensing circuit.

Build

As illustrated in Figure 6, we placed the two operons upstream of the genes encoding the N-terminal ( BBa_259H9ZPU) and C-terminal ( BBa_25YZWH1A) split halves respectively. Concurrently, the P₁₆ promoter and GFP reporter were cloned into a separate pSB4A3 plasmid carrying ampicillin resistance ( BBa_25MR74ON). After successful assembly, transformation, and sequence verification, both plasmids were co-transformed into TOP10 for functional assessment.

Figure 7. Construction of the AND-gate plasmids in Cycle 3. A. Plasmid map of pSB3K3-J23105-PcaV-P_LV-SspGyrB^C-ECF16^C-J23101-RamA-PacrRA-SspGyrBN-ECF16N. B. Plasmid map of pSB4A3- P₁₆-GFP. C, D. Sequencing results of target AND-gate plasmids.

Test

We cultured the engineered E. coli strain harboring both plasmids overnight, selected using double-antibiotic medium. Then we added inducers at an initial OD₆₀₀ of 0.02 and cultured for 10 hours. The sodium cholate concentration was fixed at 25 µM, while PCA was applied in a gradient from 0 to 1000 µM, based on literature references [2]. The normalized GFP fluorescence intensity was measured to validate AND-gate functionality and identify the optimal PCA concentration. Results indicated that fluorescence saturation occurred at approximately 400 µM PCA, which was consequently selected for subsequent experiments.

We then fixed the 400 µM PCA and 25 µM sodium cholate concentrations to characterize and optimize the induction time, simulating the duration from PCA induction to GLP-1 production by the gut flora. Negative controls included bacteria carrying the AND-gate plasmids without PCA induction, and bacteria carrying empty plasmids with both inducers. Results demonstrated an approximately 5-fold ON/OFF signal amplitude after 5 hours of induction, confirming proper AND-gate logic function with a relatively short overall induction time.

Figure 8. Characterization of the AND-gate genetic circuit in Cycle 3. (A) Measurement of GFP unit fluorescence intensity under 0–1000 µM PCA and fixed 25 µM sodium cholate over 10 hours. (B) Measurement of the unit fluorescence intensity of GFP for verification of the AND-gate logic during 3.5–6.0 hours of induction.

Learn

Through these iterative design and validation cycles, we successfully constructed an AND-gate sensing circuit capable of dynamically responding to PCA and cholate concentrations. This system shows potential for application in probiotic bacteria to sense green tea metabolites and trigger target gene expression upon gut colonization. However, several challenges remain, including the in vivo response time to PCA, optimization of the cholate-sensing operon, plasmid stability during bacterial propagation, and residual gene leakage. Addressing these issues will be essential for practical therapeutic development of the circuit.

This iterative process has deepened our understanding of how promoter and RBS sequences influence operon sensitivity, leakage, and dynamic response profiles. By systematically optimizing component expression levels, we enhanced target gene output appropriately. Furthermore, constructing this AND-gate logic has expanded our knowledge and application of split intein splicing systems [5], providing a foundation for future research and insights.

Extracellular and Intracellular Expression of the fusion GLP-1(Z)

Design

To enable the GLP-1 produced by engineered EcN in the intestinal environment to traverse cellular membranes into the bloodstream, while also facilitating in vitro experimental validation, we employed a modular structural design for two sets of GLP-1-based fusion proteins intended for extracellular and intracellular expression, respectively. These designs incorporate key components including:

• a signal peptide to direct secretory proteins into the periplasm.
• modified GLP-1 functional domains (GLP-1Z) [6] and codon-optimized human-derived GLP-1(7-36) [7] as a characterization control.
• a cell-penetrating peptide (CPP), rich in positively charged arginine and lysine residues, to mediate transmembrane transport [6].
• a C-terminal His tag for nickel-column affinity purification.

The intracellularly expressed GLP-1 fusion protein omits the signal peptide and CPP that serve as the functional domains for achieving secretory expression compared to its secreted counterpart (S-GLP1Z-CPP-H, BBa_25FJQXDY).

For the expression vector, we selected the pET-28a(+) plasmid, a widely used and reliable backbone in prokaryotic systems, particularly suitable for efficient expression of recombinant protein production in E. coli BL21(DE3). Expression is driven by the T7 promoter, which enables high-level inducible transcription of the target gene upon addition of isopropyl-β-D-thiogalactoside (IPTG), meeting the experimental demand for efficient protein expression.

Build

We commissioned a commercial vendor to synthesize the DNA sequences for two designed fusion proteins: the signal peptide–GLP-1Z–cell-penetrating peptide–His-tag (hereafter referred to as S-GLP1Z-CPP-H) and a codon-optimized GLP-1(7–36) sequence. These were subsequently cloned into a pET-28a(+) plasmid backbone. Using Gibson assembly and circular PCR, we then removed both the cell-penetrating peptide and the signal peptide sequences from the S-GLP1Z-CPP-H construct to generate an intracellular expression version, thereby successfully constructing two sets of recombinant plasmids after sequencing verification.

Extracellular expression: pET-signal peptide-GLP-1-cell-penetrating peptide-His-tag (pET-S-GLP1-CPP-H) and pET-signal peptide-GLP-1Z-cell-penetrating peptide-His-tag (pET-S-GLP1Z-CPP-H).

Intracellular expression: pET-GLP-1-His-tag and pET-GLP-1Z-His-tag.

Figure 9. Plasmid maps of GLP-1(Z) expression in pET-28a(+). (A) pET-S-GLP1-CPP-H. (B) pET-S-GLP1Z-CPP-H. (C) pET-GLP-1-His-tag. (D) pET-GLP-1Z-His-tag.

Test

The sequencing-verified plasmids were transformed into E. coli BL21(DE3), and positive clones were selected by plating on LB agar containing kanamycin, followed by overnight incubation at 37 ℃. Selected single colonies were then preserved and incubated in LB medium for later use. When the bacterial culture reached an OD₆₀₀ of 0.6–0.8, indicating the logarithmic growth phase with active proliferation and stable metabolism, we induced the protein expression by adding IPTG to a final concentration of 0.5 mM and cultured overnight.

For extracellularly expressed proteins:
The S-GLP1-CPP-H and S-GLP1Z-CPP-H fusion proteins contain an N-terminal signal peptide, a cell-penetrating peptide domain, and a C-terminal His-tag, which confer transmembrane activity and nickel-affinity binding capability, respectively. The purification workflow for extracellular protein samples included the following steps:

Figure 10. Schematic workflow of protein purification and concentration in Cycle 1.

1. Pre-treatment: The IPTG-induced bacterial culture medium was filtered through a membrane to remove large particulate impurities, and the filtrate was collected.

2. Ultrafiltration concentration: The filtrate was concentrated using a 3 kDa molecular weight cut-off (MWCO) ultrafiltration membrane (the molecular weight of the S-GLP1Z-CPP-H fusion protein is approximately 9.8 kDa) and washed with 1× phosphate-buffered saline (PBS) to remove small-molecule impurities while enriching the target protein.

3. Nickel-affinity chromatography: The concentrated sample was loaded onto a nickel-ion affinity column. The specific interaction between the C-terminal 6×His tag and the nickel ions enabled the separation and purification of the target protein. The process included sample loading, washing with 50 mM imidazole to remove nonspecific impurities, and elution of the target protein using a 250 mM imidazole gradient. The eluate was collected and further washed and concentrated via ultrafiltration.

SDS-PAGE analysis of samples collected throughout the purification process revealed that, even with the inclusion of a signal peptide and a cell-penetrating peptide, the band at the expected molecular weight for GLP-1(Z) was rarely observed in the extracellular fraction. This suggests that the plasmid system designed for secretory expression was largely ineffective. The lack of stable extracellular expression could be attributed to the relatively low molecular weight of the fusion peptide, making it susceptible to proteolytic degradation within the cell, combined with the inherent inefficiency of the E. coli secretory expression for this specific protein.

Figure 11. SDS-PAGE analysis of samples from each step of extracellular protein purification.

For intracellularly expressed proteins:
For intracellular protein extraction, we utilized ultrasonic cell disruption to rupture cell walls and membranes, releasing the intracellular contents. The lysate was then centrifuged to pellet cell debris, and the supernatant containing soluble proteins was collected. Similarly, this supernatant was subsequently applied to a nickel-ion affinity chromatography column where target protein separation and purification were achieved through the specific binding of the His-tag to the immobilized nickel ions. The procedure mirrored that used for extracellular proteins, comprising sample loading, impurity washing, gradient elution, and final sample concentration. Samples from each stage were collected and analyzed by SDS-PAGE to evaluate purification efficiency.

We also examined whether the S-GLP1Z-CPP-H protein was present intracellularly. SDS-PAGE results showed that visible bands corresponding to both S-GLP1Z-CPP-H and S-GLP1-CPP-H were observed at approximately 10 kDa after concentration and purification of the proteins, indicating their intracellular expression. This confirms successful protein expression but suggests inefficient extracellular secretion.

Figure 12. SDS-PAGE analysis of samples from each step of protein purification for S-GLP1Z-CPP-H and GLP-1Z-His-tag.

For GLP-1 expressed without the signal peptide and CPP, obtaining a distinct target band from the intracellular fraction was challenging. This is likely due to the protein's small size. We hypothesize that significant non-specific adsorption occurred during separation, the protein itself may be inherently unstable inside the cells, and leakage through the ultrafiltration membrane due to an inappropriately large molecular weight cut-off could result in the loss.

Figure 13. SDS-PAGE analysis of samples from each step of protein purification for GLP-1-His-tag.

Learn

Theoretically, intracellular expression offers advantages over extracellular expression by avoiding losses associated with transmembrane transport and medium dilution, while also simplifying purification by eliminating the need for large-volume medium pretreatment and enabling direct extraction from the cell pellet. Our initial attempt to increase protein yield by scaling up the culture volume was unsuccessful and failed to provide sufficient material for subsequent bioactivity assays, structural analysis, and other characterization experiments.

Regarding protein characteristics, all GLP-1(Z) proteins are small peptides, regardless of the presence of a cell-penetrating peptide, which are particularly susceptible to loss via adsorption and membrane permeation.

According to the theoretical guidelines for ultrafiltration, it’s recommended to use a membrane molecular weight cut-off (MWCO) 3–5 times smaller than the target protein's molecular weight to ensure effective retention [8], and our use of a 3 kDa MWCO membrane was incompatible with the molecular weights of our target proteins. This mismatch likely resulted in significant loss of the target proteins through the membrane.

We therefore managed to select more efficient chromatography resins or nanofiltration membranes with higher retention efficiency, as well as optimize the affinity chromatography elution by fine-tuning the imidazole gradient and increasing wash steps to reduce non-specific binding and ultimately obtain high-purity samples for subsequent characterization.

Improved Protein Purification Strategy Based on Nanofiltration

Design

Given the unsatisfactory protein purification results obtained with 3 kDa molecular weight cut-off (MWCO) ultrafiltration tubes which are nearly half the molecular weight of our ~7 kDa target protein GLP‑1(Z), we confirmed that the membrane pore size was inappropriate. Theoretical guidelines recommend an MWCO 3–5 times smaller than the target protein to ensure effective retention. Since the 3 kDa membrane likely permitted significant protein loss through permeation, we consulted Dr. Zhikan Yao, a researcher specializing in membrane separation (read our Human Practices Page for details), to design a nanofiltration‑based purification strategy [9] that offers stricter molecular sieving and potentially higher protein recovery.

Figure 14. Nanofiltration setup for protein purification.

Build

We established a nanofiltration system and iteratively optimized its operating parameters. Initial trials were conducted at moderate pressure (0.2–0.3 MPa) to assess baseline performance. After observing unacceptably low flux, we systematically improved the process by:
(1) increasing the operating pressure to 0.3–0.4 MPa to enhance transmembrane flow;
(2) introducing magnetic stirring in the filtration cell to minimize concentration polarization and maintain uniform protein distribution;
(3) gently resuspending the retained concentrate after filtration to minimize membrane damage and recover protein adsorbed on the membrane surface.

Test

We compared two operating regimes. Under low-pressure conditions (0.2–0.3 MPa), permeate flux and processing efficiency were poor, largely due to severe concentration polarization. The optimized high-pressure regime (0.3–0.4 MPa with stirring) yielded improved hydrodynamic performance. SDS-PAGE was used to analyze pre- and post-filtration samples, including controls, retentate, and permeate fractions, to evaluate protein retention and recovery. Unexpectedly, across multiple replicates, no protein bands were detected in either the retentate or permeate fractions.

Figure 15. SDS-PAGE analysis of samples from each step of protein purification for GLP-1(Z)-His-tag via nanofiltration.

Figure 16. SDS-PAGE analysis of samples from each step of protein purification for S-GLP1Z-CPP-H via nanofiltration.

Learn

We hypothesize that the nanofiltration membrane still exhibited substantial nonspecific adsorption toward the small protein, causing it to adhere to the membrane surface or pores and resist elution [10]. Another possible explanation is that the extremely slow filtration rate resulted in an excessively large final retentate volume and an insufficient concentration factor, thereby leading to an inadequate sample loading volume for SDS-PAGE. Although nanofiltration is theoretically well-suited for size-based protein separation, our tests indicated that it did not fully resolve the issues of filtration efficiency and protein loss in this application. In future work, we may explore fusion‑tag strategies to increase the overall molecular weight, perform initial enrichment using conventional efficient ultrafiltration, and subsequently cleave and purify the target GLP‑1(Z) product.

Protein Purification and Concentration Using His-SUMO Tag

Design

Nanofiltration proved to be an excessively time-consuming method for protein concentration and required constant cooling with ice packs to maintain low temperatures, thereby preserving the stability and native activity of GLP‑1(Z). Experimental results further indicated unsatisfactory filtration performance of nanofiltration, with limited capacity for large-volume enrichment and purification, likely due to nonspecific adsorption of the protein to the nanofiltration membrane. Upon the recommendation of our Primary PI Dr. Yu, we introduced an N-terminal His‑SUMO fusion tag to facilitate the purification and recovery of GLP‑1(Z). The SUMO (Small Ubiquitin-like Modifier) tag is a post-translational modifier structurally similar to ubiquitin, known to enhance solubility, expression levels, and protein stability. Moreover, following His‑tag purification, the SUMO tag can be removed by SUMO protease cleavage, yielding functionally intact GLP‑1(Z) for subsequent characterization [11].

Figure 17. Schematic diagram of the SUMO protease cleavage site.

Build

Starting with the original pET‑S‑GLP1‑CPP‑H and pET‑S‑GLP1Z‑CPP‑H plasmids, we first removed the C‑terminal His‑tag via circular PCR. Then, we obtained a plasmid containing the His‑SUMO tag from Dr. Yu’s lab, and amplified the corresponding gene fragment via PCR, assembling to replace the N‑terminal signal peptide sequence. This yielded two target fusion constructs in the pET‑28a(+) backbone: His-SUMO tag-GLP-1 (H-SUMO-GLP-1, BBa_25DRMK5X ) and His-SUMO tag-GLP-1Z-cell-penetrating peptide (H-SUMO-GLP1Z-CPP, BBa_2504KELH ). Sequencing results confirmed that both plasmid sequences were as expected.

Figure 18. Construction of expression plasmids for GLP-1(Z) purification and concentration using a His-SUMO fusion tag. (A) Plasmid map of pET-H-SUMO-GLP-1. (B) Plasmid map of pET-H-SUMO-GLP1Z-CPP. (C) Sequencing result of pET-H-SUMO-GLP-1. (D) Sequencing result of pET-H-SUMO-GLP1Z-CPP.

Test

Following a procedure similar to earlier cycles, we induced expression in BL21(DE3) harboring the sequence-verified plasmids with IPTG overnight, then harvested the cultures for protein recovery and purification. The fusion proteins H‑SUMO‑GLP‑1 and H‑SUMO‑GLP1Z‑CPP were purified using the same affinity chromatography method as in Cycle 1 and subsequently incubated with SUMO protease at 4 °C overnight.

Given the absence of clear bands in the gel images from Cycles 1 and 2, we attributed this in part to insufficient protein loading during electrophoresis, as supported by literature. We therefore optimized the SUMO protease cleavage system and increased the sample loading concentration.

Table 1. H-SUMO-GLP-1/H-SUMO-GLP1Z-CPP and SUMO protease cleavage incubation system

Component	Volume (μl)
H2O	X
10 × Reaction Buffer + Salt	5
SUMO-GLP-1	Y
SUMO Protease (10 U/µl)	0.25
Total	50

Note: If the concentration of the SUMO‑tagged target protein is 5 µg/µl, then Y = 20/5 = 4, meaning 4 µl of target protein should be used.

The cleavage mixture was subsequently purified by nickel‑affinity chromatography. The 6×His‑tagged SUMO tag bound to the nickel column, while the target proteins GLP‑1 and GLP1Z‑CPP were collected in the flow‑through, enabling effective separation. The eluate was concentrated using a 2 kDa MWCO ultrafiltration tube and analyzed by SDS‑PAGE to verify cleavage and purification efficiency.

Samples from each purification stage were analyzed by SDS-PAGE to assess protein expression levels and purification efficiency. SDS-PAGE results indicated the following: The native GLP-1 sequence (3.37 kDa) appeared as a diffuse band. The SUMO tag (13.7 kDa) was visible in Box I. The uncut SUMO-GLP-1 fusion protein (16.96 kDa) was present at high concentration, suggesting the cleavage reaction required optimization. The target GLP-1Z-CPP protein (5.70 kDa), after SUMO tag removal, also showed a diffuse band with lower intensity than GLP-1 as shown in Box II, potentially due to structural constraints hindering efficient protease cleavage. Furthermore, low sample concentration contributed to faint band visibility.

Figure 19. SDS-PAGE results of GLP-1(Z) samples during purification. A. Enzymatic cleavage results of His-SUMO-GLP-1. (1: Unincubated SUMO-GLP-1 protein sample; 2: His-SUMO-GLP-1 sample incubated with SUMO protease; 3: Effluent sample after nickel-affinity purification; 4: Impurity-washing sample with 50 mM imidazole; 5: Eluted sample with 250 mM imidazole; 6: Supernatant from ultrafiltration of Sample 3; 7: Precipitate from ultrafiltration of Sample 3; 8: Supernatant from ultrafiltration of Sample 4; 9: Precipitate from ultrafiltration of Sample 4; 10: Supernatant from ultrafiltration of Sample 5; Bands corresponding to the cleaved SUMO tag are present in Samples 8 and 10.) B. Enzymatic cleavage results of His-SUMO-GLP-1Z-CPP. (1: Unincubated His-SUMO-GLP-1Z-CPP sample; 2: His-SUMO-GLP-1Z-CPP sample incubated with SUMO protease; 3: Effluent sample after nickel-affinity purification; 4: Impurity-washing sample with 50mM imidazole; 5: Eluted sample with 250mM imidazole; 6: Supernatant of Sample 4 after ultrafiltration; 7: Precipitate of Sample 5 after ultrafiltration.)

Learn

Through three cycles of engineering iteration, we have established an in vitro purification and enrichment process for the target protein GLP‑1(Z), enabling subsequent activity characterization. However, limited by the project timeline, the concentration efficiency and purity of the current process still require improvement. Additionally, the SUMO protease removes both the SUMO tag and the adjacent His‑tag, preventing further His‑tag‑based purification or characterization of the cleaved target protein. To address this, we propose an optimized design in which the His‑tag is placed between the SUMO tag and the GLP‑1(Z) sequence, creating SUMO-His‑GLP‑1(Z) fusion constructs. Theoretically, after cleavage, GLP‑1 would retain the His‑tag, allowing separation from the SUMO protease via nickel‑affinity chromatography, with final isolation by ultrafiltration based on molecular weight. Further work may also focus on optimizing secretory expression conditions, including screening for more efficient signal peptides and cell-penetrating peptides, to achieve robust extracellular production of GLP‑1(Z).

A Simplified OR-Gate Biosafety Circuit Based on Dual Arabinose and Temperature Regulation

Design

In the initial circuit design, we constructed an OR-gate responsive genetic circuit by incorporating a temperature-regulated operon [12] and an arabinose operon elements [13], following established literature. We first built a temperature-regulated self-repressive system (BBa_2575N2V6) using the temperature-responsive P_tlpA promoter ( BBa_25RFWS4Y) and an improved repressor protein TlpA* ( BBa_25XF4HZ6), to regulate the expression of the arabinose repressor protein AraC. Subsequently, the arabinose promoter (BBa_25NOQJ8F) controls the expression of a downstream toxin gene for suicide regulation. For preliminary characterization and testing, we replaced the ccdB gene with the reporter gene GFP ( BBa_K2365043) and temporarily omitted ccdA and its upstream constitutive promoter.

This circuit is designed to block AraC repressor expression at low temperatures and, in the presence of arabinose, to relieve repression by AraC, thereby enabling expression of the target gene downstream of the arabinose operon and ultimately inducing bacterial suicide. Experimentally, this should manifest as: (1) increased fluorescence intensity with rising arabinose concentration; and (2) significantly higher fluorescence intensity at 30 °C (in vitro) compared to 37 °C (in vivo).

Figure 20. Schematic diagram of the Cycle 1 biosafety circuit design

Build

We obtained the E222 plasmid containing the arabinose operon from Professor Baojun Wang’s lab and subsequently commissioned a company to synthesize the sequence of the temperature-regulated self-repressive system for site-directed replacement. We then amplified the OR-gate circuit elements along with the downstream GFP reporter gene fragment via PCR and assembled them into the pSB3K3 plasmid backbone using Gibson assembly. The assembled product was transformed into E. coli TOP10 competent cells, and positive clones were selected on LB plates containing kanamycin for sequencing verification. The results confirmed the successful construction of the characterization plasmid.

Figure 21. Construction of expression plasmid for Cycle 1. (A) Results of electrophoretic bands of pSB3K3-P_tlpA-tlpA^*-araC-P_BAD gene fragment (~5 kb) after PCR. (B) Plasmid map of pSB3K3-P_tlpA-tlpA^*-araC-P_BAD-GFP. (C) Sequencing results of target plasmid for the Cycle 1 biosafety genetic circuit.

Test

We selected the sequence-verified plasmid for transformation into TOP10, inoculating them into 4 mL of LB medium with kanamycin for overnight culture. The culture was then diluted to an OD₆₀₀ of 0.02 and supplemented with arabinose at final concentrations of 0, 0.003, 0.03, and 0.3 mM. Each induction condition was divided into two groups and incubated at 30 °C and 37 °C, respectively, for 6 hours. GFP fluorescence intensity was measured using a microplate reader with excitation at 485 nm and emission at 520 nm, and used to calculate normalized fluorescence units. The experimental results are shown below.

Figure 22. Measurement results of unit fluorescence intensities of the Cycle 1 circuit after 6-hour induction with arabinose at different temperatures.

Learn

The trend in fluorescence intensity with increasing arabinose concentration generally aligned with the expected behavior. However, the temperature sensor exhibited a narrow dynamic range (approximately 2.5-fold), and absolute fluorescence intensity across all groups remained below 1.5. This may be attributed to significant leakage expression of the P_tlpA promoter or incomplete activation of P_BAD in the circuit design. Therefore, we planned to prioritize separate characterization and optimization of the temperature self-repressive system and the induction time.

RBS Expression Tuning and Optimization of the Self-repressive System

Design

To investigate the failure of temperature regulation in the Cycle 1 circuit, we replaced the native RBS sequence upstream of the AraC repressor protein in the original design. We aimed to determine whether repressor expression levels affected the function of the arabinose operon, thereby constructing a Cycle 2 biosafety genetic circuit and reasonably extended the induction time to 8 hours. To isolate contributing factors, we also characterized and optimized the temperature-responsive self-repressive system over various induction durations to inform subsequent circuit refinement.

Figure 23. (A) Schematic diagram of the Cycle 2 biosafety circuit design. (B) Schematic diagram of the genetic circuit for characterizing the temperature-responsive self-repressive system.

Build

Based on the Cycle 1 circuit, we designed primers with homologous arms and performed circular PCR to replace the RBS sequence preceding the AraC repressor gene. The PCR product was transformed into E. coli TOP10 to construct the complete plasmid, and positive clones were selected on LB plates with kanamycin and verified by sequencing. We then amplified the temperature-regulated self-repressive elements harboring different RBS sequences via PCR and assembled to the GFP reporter gene for characterization. These fragments were assembled via Gibson assembly and validated using methods similar to those previously described, thus obtaining the series of plasmids required for the second iteration.

Figure 24. Construction of expression plasmids of the Cycle 2 biosafety design and temperature-responsive self-repressive system. (A-B) Plasmid maps of the Cycle 2 biosafety circuit with various RBS sequences. (C) Results of electrophoretic bands of P_tlpA-tlpA^* with various RBS sequences (~1.2 kb) and the pSB3K3 plasmid backbone fragments harboring GFP (~3.6 kb) after PCR. (D-F) Plasmid maps of the temperature-responsive self-repressive system with GFP reporter and various RBS sequences (strength: RBS34>RBS30>RBS32).

Test

We transformed the successfully constructed plasmids into TOP10 for characterization. In addition to testing at 30 °C and 37 °C with the same arabinose concentration gradients, we extended the induction time for the Cycle 2 circuit to 8 hours. Unit fluorescence intensity was measured using a microplate reader similar to Cycle 1, with bacteria carrying the empty pSB3K3 plasmid serving as a negative control. Results showed that arabinose regulation still functioned normally at 37 °C, whereas at 30 °C, the induction effect was nearly absent or even slightly decreased. In the absence of arabinose, temperature variation produced almost no signal difference, indicating that the OR-gate regulatory function failed again.

Figure 25. Measurement results of unit fluorescence intensities of the Cycle 2 circuit after 8-hour induction with arabinose at different temperatures. (A) Original RBS. (B) RBS30. (C) RBS32.

We next evaluated the performance of the temperature self-repressive system. Cultures with an initial OD₆₀₀ of 0.02 were incubated at 30 °C and 37 °C respectively, and samples were taken periodically to measure unit fluorescence intensity, thereby assessing system performance and optimizing the induction time parameter.

Results showed that fluorescence intensity at 30 °C remained low and stable, with consistently low leakage expression. In contrast, the unit fluorescence intensity at 37 °C initially increased and then decreased, peaking at approximately 4 hours. Therefore, we selected 4 hours as the induction time for temperature-regulated signaling. We further compared the 37 °C/30 °C signal ratios corresponding to the three RBS sequences at 4 hours and found that they could all achieve a switching effect of one order of magnitude.

Figure 26. Characterization and optimization of expression plasmids for the temperature-responsive self-repressive circuit. (A–C) Measurement results of unit fluorescence intensity over 9 hours at different temperatures for TOP10 carrying the self-repressive circuit with different RBS sequences. (D) Temperature-regulated switching effects of the three characterization circuits at 4 hours.

Learn

Even with extended induction time and RBS replacement, the original genetic circuit failed to achieve effective OR-gate regulation. We speculated that the regulatory mechanism of the arabinose operon is complex and cannot be resolved simply by reducing AraC repressor expression. Subsequent literature review confirmed that the arabinose operon requires both arabinose and the AraC protein for proper downstream gene expression. We further discovered that AraC exhibits dual functionality in its interaction with P_BAD, extending beyond mere repression [13]. Constitutive expression of the repressor or adoption of alternative operon elements with simpler regulatory mechanism may be necessary.

Meanwhile, we confirmed that the P_tlpA-tlpA^* self-repressive system itself exhibited low leakage and could reliably achieve an order-of-magnitude signal switching. Considering both the leakage level at 30 °C and downstream gene expression efficiency, we selected the composite element containing RBS30 and a 4-hour induction time for further optimization of the OR-gate genetic circuit.

OR-Gate Circuit Optimization Using the Tetracycline Operon and Constitutively Expressed Repressor

Design

Our experiments revealed that the previous OR-gate circuit failed to achieve temperature regulation in the absence of arabinose, as well as arabinose-induced expression at 35 °C. We hypothesized that the AraC repressor may actually play a dual role, cooperatively facilitating downstream expression when binding to the P_BAD promoter. In this iteration, we reconstructed the OR-gate circuit by constitutively expressing AraC and incorporated the tetracycline (Tet) operon [14] with simpler regulatory mechanism for logic optimization.

To ensure reliable target gene expression, we further designed a tandem promoter construct. We initially postulated that this composite element would implement OR-gate logic, wherein transcription initiation from either promoter would be sufficient to drive downstream gene expression. Additionally, we introduced the RiboJ insulator element [15] (BBa_25XUNXCE ) to enhance downstream gene expression and minimize interference from extended upstream leader sequences.

In our circuit design, both AraC and TlpA^* are constitutively expressed. At temperatures below 35 °C, the conformational change in TlpA^* enables its specific binding to the P_tlpA promoter, repressing downstream tetR expression. Low TetR levels fail to repress its cognate promoter P_tet, thereby activating GFP expression. When arabinose is present, it binds to AraC, activating the P_BAD promoter and similarly inducing downstream GFP expression. The overall circuit schematic diagram is shown below.

Figure 27. Schematic diagram of the Cycle 3 biosafety circuit design

Build

We first obtained the RiboJ insulator and tet operon elements from Professor Wang’s lab. Given the large number of components and the low efficiency of Gibson assembly with more than four DNA fragments, we split the circuit into two modular plasmids, which were later integrated via PCR and Gibson assembly into the final circuit plasmid. We designed primers to amplify each element for assembly into both Plasmid 1 and 2. After transforming E. coli TOP10, we only confirmed successful construction of Plasmid 1 but not Plasmid 2, thereby redesigned the primers to use Plasmid 1 as the backbone for integrating the remaining elements, and ultimately assembling the final OR-gate circuit.

Table 2. DNA fragments for Gibson assembly of each plasmid.

Plasmid 1	Plasmid 2	Circuit 3
RiboJ-GFP	PBAD	PBAD
PtlpA	J23105-Ptet	J23105-Ptet
tetR-J23105	tlpA*	tlpA*
pSB3K3-araC	pSB3K3	Plasmid 1

Note: The last row indicates the plasmid backbone vector.

Figure 28. Construction of expression plasmid for Cycle 3. (A) Results of electrophoretic bands of P_BAD (120 bp), pSB3K3-araC (~3.7 kb), J23105 (130 bp), P_tlpA (181 bp) and tlpA^* (~1.2 kb) gene fragments after PCR. (B) Plasmid map of Plasmid 1 for constructing the Cycle 3 circuit. (C) Plasmid map of the Cycle 3 biosafety circuit. (D) Sequencing result of Plasmid 1 for constructing the Cycle 3 circuit. (E) Sequencing result of the Cycle 3 biosafety circuit.

Test

We transformed the sequence-verified final circuit plasmid into TOP10 for characterization. Cultures were inoculated at an initial OD₆₀₀ of 0.02 and induced with arabinose at final concentrations of 0, 0.01, 0.039, 0.153, and 0.6 mM, followed by incubation at 30 °C, 35 °C, and 37 °C in groups for 4 hours. We then measured normalized GFP fluorescence intensity using a microplate reader and compiled the results. It turned out that the arabinose operon functioned as expected, with high fluorescence intensity at elevated arabinose concentrations. However, in the absence of arabinose, although fluorescence was highest at 30 °C, it was not distinctly separated from expression levels at other temperatures, indicating limited temperature-dependent regulation.

Figure 29. Measurement results of unit fluorescence intensities of the Cycle 3 circuit after 4-hour induction with arabinose at different temperatures.

Learn

As observed, fluorescence intensity showed a strong positive correlation with arabinose concentration, fully consistent with expectations. However, at high arabinose concentrations, fluorescence was lowest at 30 °C, contrary to the anticipated additive effect of the tandem promoters. We preliminarily attribute this to metabolic burden imposed by the circuit on the host strain.

More critically, in the absence of arabinose, although fluorescence was highest at 30 °C, the difference compared to other temperatures was not pronounced. In designing this circuit, we hypothesized that downstream gene expression would occur whenever either promoter in the tandem construct was activated. Based on the observed results, we suspected that when AraC binds to P_BAD, downstream GFP expression may be suppressed regardless of P_tet activation status. Future efforts could explore tuning the constitutive expression level of AraC, though synchronously optimizing expression strengths of multiple components in such a complex circuit remains challenging.

Replacing the Arabinose Operon with the Fucose Operon

Design

Previously, we found that the conformational regulation mechanism of the arabinose operon is relatively complex, so we planned to select other operons with simpler regulation principles. Through literature and patent searches, we identified a type of fucose operon [16] that not only possesses the safe and edible property but also seems to be simpler. Based on this, we replaced the original arabinose operon, constructed the Cycle 4 biosafety genetic circuit using the fucose operon, and conducted corresponding characterization experiments.

Figure 30. Schematic diagram of the Cycle 4 biosafety circuit design

Build

We selected this component primarily because the arabinose operon fails to drive downstream gene expression in the absence of its cognate repressor AraC. To address this, we first needed to verify whether the P_fuc promoter could constitutively express genes without its repressor FucR. We commissioned a company to synthesize the fucose operon elements. Interestingly, despite codon optimization, only the linear DNA fragment of the operon could be obtained. Multiple attempts using various commercial competent cell strains failed to integrate the construct into a plasmid expression system without point mutations. Nevertheless, we successfully amplified the P_fuc promoter via PCR purification, assembled it with the GFP reporter gene into the pSB3K3 plasmid, and confirmed its expression.

Figure 31. Construction of the repressor-free verification plasmid. (A) Schematic diagram of the verification circuit. (B) Results of electrophoretic bands of P_fuc (243 bp) and FucR (~1.5 kb) gene fragments after PCR. (C) Plasmid map of pSB3k3-Pfuc-GFP. (D) Sequencing result of pSB3k3- P_fuc-GFP.

Test

Figure 32. Measurement results of unit fluorescence intensities of the verification circuit after overnight induction.

Learn

It can be observed that, compared to the negative control, the fluorescence intensity in the experimental group increased only marginally to approximately 0.4, indicating it’s unable to drive constitutive downstream expression, rendering the system unsuitable for OR-gate circuit construction.

Furthermore, repeated assembly and transformation attempts failed to yield characterization circuits containing the functional FucR repressor. Even when the synthesis company employed specialized EPI400 competent cells, they only obtained plasmids with a nonsense point mutation introducing a premature stop codon within the repressor coding sequence. We suspected that the codon-optimized element may exert toxic effects on E. coli, and finally gave up carrying on further work involving the fucose operon for characterization and application.

Figure 33. Sequencing result of the fucose operon transformed using EPI400 competent cells by the synthesis company.

Tuning the Constitutive Expression of AraC in Cycle 3

Design

In the third cycle, we observed that the dual-promoter system failed to activate under arabinose-free conditions, even when TetR repressor expression was successfully regulated by temperature. We hypothesized that the AraC repressor protein persists in blocking transcriptional initiation from the P_tet promoter, consequently preventing downstream gene expression. To address this, we aimed to alleviate the excessive repression exerted by AraC on the dual-promoter system in the absence of arabinose via attenuating its constitutive expression level. Specifically, the original constitutive promoter J23105 was replaced with the weaker variants J23117 and J23115, thereby enhancing the system's sensitivity to temperature signals.

Figure 34. Schematic diagram of the Cycle 5 biosafety circuit design.

Build

During our plasmid construction attempts, we encountered significant challenges in performing in-situ replacement of the promoter sequence. The presence of identical J23105 sequences within the composite plasmid led to multiple non-specific bands in PCR products, which subsequently compromised the efficiency of Gibson assembly and transformation. When selecting positive clones on kanamycin-containing LB plates, the majority were either the original template plasmid or contained various point mutations in the promoter region. Despite repeating the experiment for two additional rounds, we were unable to successfully construct the target plasmid and therefore discontinued further assembly and testing.

Learn

We suspected that the construction failure stemmed from the high sequence similarity or homology within the template plasmid, which hindered precise promoter replacement via primer-based PCR. Reconstructing the plasmid from individual components would have been time-consuming, and due to time constraints, we did not pursue this alternative. Furthermore, theoretically, modulating the constitutive expression level of AraC may not effectively enhance temperature sensitivity under arabinose-free conditions. This is because even basal expression of AraC could allow it to bind its cognate P_BAD promoter and exert repression, regardless of expression level. Consequently, we planned to optimize the circuit by redesigning the original tandem promoter composite element in subsequent work.

Integration of the Temperature-responsive Self-repressive System with a Tandem Dual-Promoter Element

Design

Building on insights from previous iterations, we integrated and optimized core functional components from earlier circuits to construct our Cycle 6 OR-gate genetic circuit. As mentioned above, AraC is essential for activating the P_BAD promoter. We therefore explored whether modifying the upstream promoter construct could bypass or compensate for P_BAD regulation in the absence of the repressor. To this end, we designed a composite element ( BBa_2552PNC9 ) by placing the constitutive promoter J23101 upstream of P_BAD, forming a tandem construct. This design allows transcriptional read-through of the downstream gene even in the absence of AraC. We also incorporated the RiboJ element to minimize the impact of the upstream leader sequence on gene expression, thereby enhancing output signal levels.

Figure 35. Schematic diagram of the Cycle 6 biosafety circuit design

Build

We first constructed a repressor-free validation circuit as well to confirm that downstream gene expression could be activated without AraC. Accordingly, we assembled the plasmid pSB3K3-J23101-P_BAD-RiboJ-GFP. The J23101 promoter sequence was introduced via PCR primers designed with homologous arms, and the resulting fragments were assembled into the pSB3K3 backbone using Gibson Assembly. The constructed plasmid was transformed into E. coli TOP10, and positive clones were selected on kanamycin-containing LB plates, followed by sequencing verification.

Figure 36. A tandem dual-promoter strategy adopted to construct the Cycle 6 OR-gate circuit. (A) Schematic diagram of the tandem dual-promoter system. (B) Results of electrophoretic bands of plasmid backbone vector (~2.7 kb) and the RiboJ-GFP fragment harboring the J23101 promoter sequence (~1 kb) after PCR. (C) Plasmid map of pSB3K3-J23101- P_BAD-RiboJ-GFP. (D) Sequencing results of the target plasmid harboring the dual-promoter element.

Building on the characterization of the basic and composite elements, we proceeded to construct the complete biosafety OR-gate circuit by incorporating the arabinose repressor AraC and the temperature-regulated self-repressive system. Following the aforementioned procedure, the plasmid backbone, araC gene fragment, self-repressive composite element, and dual-promoter expression unit were amplified by PCR with specific primers. These fragments were assembled via Gibson Assembly, transformed into TOP10, and positive clones were selected with verification by DNA sequencing. Ultimately, we successfully assembled the Cycle 6 biosafety OR-gate circuit with GFP as the reporter gene.

Figure 37. Construction of the Cycle 6 biosafety OR-gate genetic circuit. A. Results of electrophoretic bands of the pSB3K3 plasmid backbone fragment harboring J23101- P_BAD-RiboJ-GFP (~3.7 kb) and P_tlpA-tlpA^*-araC fragments with different RBS sequences (~2.2 kb) after PCR. B. Plasmid map of pSB3K3- P_tlpA-tlpA^*-araC-J23101- P_BAD-RiboJ-GFP. C. Sequencing results of the Cycle 6 OR-gate biosafety plasmid.

Test

After confirming the plasmid sequence, TOP10 cells carrying the validation plasmid were cultured overnight at 37 °C. The sequence-verified Cycle 6 plasmid was also transformed into TOP10 for functional characterization. Cultures were incubated in shakers at 37 °C and 30 °C with arabinose gradients (0, 0.01, 0.039, 0.153, and 0.6 mM), and fluorescence intensity was measured using a microplate reader. Cells transformed with the empty pSB3K3 plasmid served as the negative control.

The results demonstrated that the engineered strain lacking the araC gene exhibited significant fluorescence expression, confirming that the tandem dual-promoter strategy effectively overcame the inherent constraint of the arabinose operon and enabled the intended circuit function. For Cycle 6 circuit characterization, a notable level of leaky expression persisted at 37 °C even without arabinose. However, the overall circuit response aligned with our design principles. Normalized fluorescence intensity increased with higher arabinose concentrations, and the expression level was substantially greater at 30 °C compared to 37 °C.

Figure 38. Characterization of the genetic circuits in Cycle 6. (A) Functional validation of the tandem dual-promoter circuit. (B) Results of temperature and arabinose response curves of the Cycle 6 biosafety circuit. (C) Switching effects of the Cycle 6 biosafety circuit under four input states.

Learn

By integrating the aforementioned elements and design strategies, we successfully constructed a temperature-arabinose dual-input OR-gate genetic circuit, achieving an ON/OFF signal amplitude of approximately 5-fold. Furthermore, the response threshold can be dynamically adjusted by regulating the expression intensity of subsequent toxin-antitoxin genes, thereby preventing unintended cell death in the "OFF" state and ensuring the strain's ability to produce GLP-1 in the human intestine.

However, during actual construction, we encountered difficulties in effectively integrating the ccdA/ccdB toxin-antitoxin elements [17] into this biosafety circuit, likely due to signal intensity mismatch. Subsequent work should involve sequence optimization of this system or selection of native E. coli toxin-antitoxin systems. Due to time constraints, we may not be able to conduct subsequent experiments on optimizing culture and characterization conditions, but we have preliminarily verified its feasibility.

Microspheres Based on Sodium Alginate–Calcium Chloride Hydrogel with Chitosan Coating

Design

Following our Human Practices research, we selected the highly biocompatible hydrogel as the carrier for the oral bacterial agents. Through literature and patent surveys, we found a preparation method for sodium alginate–calcium chloride hydrogel microspheres coated with chitosan [18]. This delivery system relies on the fact that the carboxyl groups of alginate remain unprotonated under acidic conditions, reducing intermolecular electrostatic repulsion and forming a relatively dense protective matrix for the bacteria. Under neutral conditions, increased electrostatic repulsion causes the hydrogel network to swell, facilitating bacterial release. The chitosan coating provides a protective layer of protonated amino groups and steric hindrance in acidic environments, shielding the microspheres from hydrogen ions and macromolecules like proteases, thereby creating a favorable microenvironment to maintain bacterial viability [19].

We followed the protocol using 1% (w/v) sodium alginate and 0.1 M calcium chloride solution to prepare hydrogel microspheres, which were subsequently coated with 0.5% (w/v) chitosan. The protective and release performance was preliminarily assessed by OD₆₀₀ measurement.

Figure 39. "Egg-box" model of sodium alginate–calcium chloride cross-linking. [20]

Build

We first prepared sterile solutions of 1% (w/v) sodium alginate, 0.1 mol/L CaCl2, and 0.5% (w/v) chitosan with 1% (v/v) acetic acid in deionized water, followed by filter sterilization using syringes and sterile membrane filters. Overnight cultures of engineered EcN were then added in gradients of 10, 20, 50, 100, and 200 µL to the sterile sodium alginate solution. The mixtures were manually dispensed dropwise from syringes into the calcium chloride solution. After standing for 10 minutes, calcium alginate microcapsules gradually formed. The microspheres were then collected with a sterilized spoon and transferred into the chitosan solution, incubated for approximately 1.5 hours to obtain chitosan-coated hydrogel microspheres.

Figure 40. Preparation of sodium alginate–calcium chloride hydrogel microspheres. (A) Hydrogel microspheres prepared with different bacterial culture volumes. (B) Microstructure of hydrogels observed under optical microscopy at 400× magnification.

Test

We successfully prepared chitosan-coated sodium alginate–calcium chloride hydrogel microspheres. Optical microscopy revealed relatively uniform morphology, with color intensity increasing with higher bacterial loadings. The microspheres were incubated separately in simulated gastric fluid (SGF) and simulated intestinal fluid (SIF) (turn to Experiments page for preparation details), washed thoroughly with PBS buffer, and then transferred into liquid LB medium for 6 hours of culture. The protective effect was evaluated by measuring the OD₆₀₀ of the cultures. Tubes containing microspheres treated with SIF showed significant turbidity. In contrast, tubes with microspheres treated with SGF exhibited almost no increase in OD₆₀₀ (close to blank control), indicating minimal resistance of the hydrogel system to simulated gastric conditions and an inability to fulfill the oral delivery function.

Learn

The sodium alginate–calcium chloride hydrogel prepared according to the patented method failed to protect the bacteria effectively from gastric acid and digestive enzymes as intended. We hypothesized that the cross-linked alginate-calcium network and the chitosan coating were too loose, insufficiently dense, or too thin, allowing gastric acid to penetrate the microspheres and kill the encapsulated E. coli [20]. The inherent cationic nature of chitosan, attributable to its positive surface charge, may confer toxicity toward the indigenous gut microbiota, necessitating strict control over its dosage [21]. Consequently, this system proved inadequate for oral delivery. Future efforts will focus on designing hydrogel systems with better encapsulation efficiency or smaller pore sizes to improve protection.

Interfacial Engineering of Living Bacteria Using Carboxymethyl Starch Sodium

Design

In the first cycle, we suspected that the spatial grid structure of the alginate‑calcium chloride coating was too loose, allowing gastric acid to penetrate the hydrogel and compromise bacterial viability. To address this, we investigated strategies to reduce diffusion and permeation within the gel. One approach to improve protection is to increase the gel’s coverage ratio per unit surface area, or to directly form a protective layer over the bacterial surface. Accordingly, we adopted a surface engineering strategy based on carboxymethyl starch sodium (CMS-Na).

CMS-Na is commonly used as a natural excipient and, similar to alginate, can resist gastric acid and enzyme penetration via protonation of its carboxyl groups. Upon reaching the intestine, it swells and dissociates, releasing the bacteria. By cross‑linking CMS-Na with calcium ions on the bacterial surface, a hydrogel coating can be formed to enable intestinal delivery and release.

Figure 41. Schematic diagram of the interfacial engineering for living drugs[22].

Build

We pre-cultured a 4 mL tube of EcN overnight at 37 °C and 220 rpm as the encapsulation stock. The culture was then centrifuged at 6000 rpm for 5 min and washed twice with 1× PBS. The pellets were resuspended in 1 mL of sterile PBS containing 12.5 mM CaCl2 and incubated for 5 min. After that, 50 µL of sterile 10 mg/mL CMS‑Na solution was added, and the mixture was vortexed for another 5 min. Finally, the sample was centrifuged and washed twice with PBS to obtain the gel‑coated living drug formulation, denoted LD(P). A negative control (LD) was prepared via only centrifugation and PBS washing.

Test

We adjusted the OD₆₀₀ of both suspensions to 1 and incubated them in SGF and SIF for 0–4 hours. After incubation, the samples were centrifuged, washed with PBS, and approximately 100 µL of each was spread on LB plates for overnight culture. Colony counts were compared to evaluate the acid‑ and enzyme‑resistance of the hydrogel. Visually, after centrifugation, the LD control treated with SGF showed almost no pellet, whereas LD(P) remained in hydrogel form. Plate results qualitatively indicated that the CMS‑Na coating improved the acid tolerance of EcN.

Figure 42. Protective effect of LD(P) against SGF. (A) Centrifuged pellets after PBS washing. (B) Colony growth on plates from samples incubated in SIF and SGF for 2 hours.

Learn

Our experiments preliminarily verified that CMS‑Na‑based surface engineering can enhance bacterial resistance to gastric acid. However, several practical issues arose: 12.5 mM CaCl2 in PBS readily formed gelatinous precipitates due to reaction between calcium and phosphate ions; the coating was uneven, leading to poor batch-to-batch reproducibility and difficulty in tracking protection dynamics over 0–4 hours; LD(P) storage stability was uncertain; and the quality of the CMS‑Na used was variable, necessitating stricter control of post‑preparation pH and other parameters. Therefore, a more reproducible and quantitatively reliable protocol is needed to accurately evaluate the effects of the coating on bacterial viability and release, while minimizing systematic bias.

Combined Strategy of Enteric Capsules and Sodium Alginate–Calcium Chloride Hydrogel Microspheres

Design

To improve batch-to-batch reproducibility and achieve precise control over the release profile, enabling targeted intestinal delivery and prolonged colonization, we selected a combination of enteric capsules and sodium alginate (SA)-calcium chloride hydrogel microspheres as the carrier [23]. This combination strategy enhances gastric acid resistance while leveraging the sustained-release properties of hydrogels to deliver probiotics to specific intestinal regions. Moreover, it also facilitates standardized preparation of live bacterial formulations based on gel quality and encapsulation efficiency.

Figure 43. Schematic diagram of our hydrogel microsphere delivery system with gelatin capsules.

Build

EcN was first cultured overnight at 37 °C. Bacterial cells were then harvested and diluted to approximately 1×10⁹ CFU/mL, washed twice with 1× PBS buffer, and resuspended in 1% (w/v) sodium alginate (SA) solution. The mixture was extruded dropwise using syringes into a 0.1 M calcium chloride solution to form hydrogel microspheres. After 10 minutes of crosslinking, the microspheres were collected and encapsulated into gelatin enteric capsules to form the final oral delivery formulation.

To further optimize controlled release, we prepared a series of microspheres by varying the concentrations of SA and calcium chloride to adjust crosslinking density.

Figure 44. Preparation of enteric capsule-encapsulated hydrogel microspheres for oral delivery. (A) Sodium alginate–calcium chloride hydrogel microspheres. (B) Microspheres encapsulated within a gelatin enteric capsule.

Test

We evaluated the protective and release performance of the formulation by incubating samples in SGF for 1 hour, then transferring the microspheres to SIF for extended release. Samples were taken hourly, centrifuged, washed with PBS, serially diluted, and plated on LB agar. After 24-hour incubation at 37 °C, colonies were carefully counted. Microspheres without capsule protection served as the negative control. Results indicated that enteric capsules maintained their structural integrity in SGF but dissolved almost completely within 10 minutes upon transfer to SIF at 37 °C. Release profiles showed that capsule-protected microspheres released probiotics more slowly.

For microspheres with different crosslinking densities, minimal release was observed within the first 4 hours, followed by rapid release of encapsulated bacteria, demonstrating a clear delayed-release effect. Varying sodium alginate and calcium chloride concentrations did not significantly alter this release pattern.

Figure 45. Characterization of the protective and release properties of capsule-encapsulated hydrogel microspheres. (A) Differences of the capsule structural integrity in SGF and SIF. (B) Effect of enteric capsule encapsulation on the release profile of hydrogel microspheres. (C) Viability and accumulation of E. coli released from microspheres with different crosslinker concentrations. (D) Release profiles of probiotic carriers with different cross-linking degrees. (E) CFU measurement during the first 4 hours with 10^-3 dilution.

Learn

After 7 hours in SIF, capsule-encapsulated microspheres had released only 54.2% of their load relative to controls, suggesting the release had not yet reached a plateau. We hypothesized that gelatin degradation products may engage in weak electrostatic interactions or hydrogen bonding with carboxyl groups on the alginate microsphere surface, thereby altering release kinetics.

All crosslinking concentrations tested exhibited similar release profiles, with a consistent delay of approximately 4 hours before the onset of rapid release. This might be due to insufficient variation in concentration, which did not significantly alter hydrogel porosity or crosslinking topology. Future work could explore alternative crosslinkers to further optimize release kinetics.

In summary, combining hydrogel microspheres with enteric capsules provides effective protection against gastric conditions and enables sustained intestinal release, offering a promising platform for oral probiotic delivery. Further optimization of release kinetics and adhesive properties may be required to enhance the long-term colonization and drug production of engineered bacteria.

Thiol-Modified Hydrogels for Enhanced Bacterial Adhesion and Colonization

Design

To extend colonization and reduce dosing frequency, we sought to improve the adhesive properties of the hydrogel carrier. Literature indicates that probiotic adhesion relies on surface protein interactions and specific receptors. The intestinal mucus layer is rich in cysteine residues [24], and thiolated hydrogels can form strong disulfide bonds with both bacterial pili as well as mucosal cysteines, prolonging retention of engineered probiotics [25]. So we also preliminarily performed thiolation modification on carboxymethyl cellulose (CMC), which is a hydrogel material similar to sodium alginate and carboxymethyl starch (CMS), and characterized the adhesion properties of the resulting CMC-SH, laying a foundation for the development of long-acting hydrogel-based oral bacterial formulations.

Figure 46. Schematic diagram of thiol-modified hydrogels for enhanced bacterial adhesion and intestinal colonization.

Build

We fortunately obtained approximately 0.1 g of CMC-SH from Professor Zhuxian Zhou’s lab (synthesis methods as detailed in the literature [26]) and combined it with sodium alginate to form composite films for evaluating bacterial adhesion. Due to the complexity of CMC-SH synthesis, we prepared a limited batch by dissolving 0.07 g CMC-SH and 0.04 g sodium alginate in 1.4 mL distilled water, incubating at 25 °C for 4 hours, adding 21 mg glycerol, and sonicating to mix thoroughly [27]. The gel was then spread on one half of a split LB plate, leaving the other half uncoated. After solidification, both sides were streaked with pre-cultured EcN using an inoculating loop. Plates were inverted and incubated overnight at 37 °C and unmodified CMC served as a control.

Test

We assessed the adhesion property using a simple rinsing assay (see our experimental protocol. In both types of plates, bacteria on uncoated agar were easily rinsed off. In contrast, CMC-SH-coated areas and the bacteria adhering to its surface were much harder to be rinsed off compared with the CMC coating, showing little difference from the unwashed ones. We preliminarily conclude that thiolated CMC is capable of enhancing its adhesive capacity, even for surface-associated cells, which is conducive to the long-term colonization of engineered bacteria in the intestine.

Figure 47. Adhesion performance of CMC and CMC-SH hydrogel coatings for EcN on LB agar based on the running water rinsing experiment.

Learn

After establishing a delivery system that protects probiotics and enables intestinal release, we further explored thiolation as a strategy to prolong colonization. Owing to time constraints and experimental conditions, while we have not conducted in vivo experimental verification, and simple water flow impact experiments cannot achieve quantitative evaluation of hydrogel adhesion properties, this strategy has been fully characterized in the literature [27] and exhibits practical feasibility. Subsequently, we could conduct further verification by performing bacterial encapsulation experiments based on modified hydrogel carriers, or experiments designed for the thiolation modification of components such as sodium alginate.

Through these studies, we have systematically optimized and combined commonly used hydrogel materials to develop an oral bacterial formulation that offers gastric protection, targeted release, sustained colonization, storage stability, and reproducible production. Our project aims to provide preliminary validation of an oral live drug delivery platform, bringing us closer to a user-acceptable, compliance-friendly solution for weight management and glucose control.