Loop 1

Design

The bcs operon of cellulose-producing Komagataeibacter sucrofermentans species comprises four structural genes, bcsA, bcsB, bcsC and bcsD, that act in concert to assemble and export BC (Figure 1). Among them, bcsA encodes a membrane-associated glycosyltransferase that, once activated by the second messenger c-di-GMP, transfers glucose from UDP-glucose to a growing β-1,4-glucan chain. Its periplasmic partner BcsB binds c-di-GMP, stabilizes BcsA, and directs the nascent polymer across the inner membrane; the BcsAB complex is therefore the essential catalytic core. BcsC forms an outer-membrane channel for polymer extrusion, while BcsD promotes periplasmic crystallization into ordered microfibrils.

Figure 1. Genetic organization of bcs operon^[1].
When BcsA (green) is activated by c-di-GMP, it incorporates glucose units into a cellulose chain in the cytoplasm using UDP glucose as a substrate. BcsB (blue) guides the glucan chain through the periplasm; BcsD (orange) crystallizes four glucan chains in the periplasm, and finally, BcsC (gray) exports the BC micro-fibrils into the extracellular space.

Through literature review^[2], we learned that there are differences between E. coli DH5α and BL21(DE3) when used as host strains for expressing bcsABCD. The selection of host strains is crucial. Since the "one-step" system is designed for bacterial cellulose (BC) synthesis at 25°C, we first investigated which strain-E. coli DH5α or BL21(DE3)-is more suitable for BC expression at this temperature.

Build

Given the excessively long length of the full, bcsABCD gene cluster (9,126 bp), we divided these four core genes into two segments, namely bcsAB and bcsCD, to facilitate their expression in E. coli. Specifically, the bcsA and bcsB genes were assembled into the pACYCDuet-1 vector to construct the plasmid pACYCDuet-1-bcsAB (Figure 2a), while bcsC and bcsD were inserted into the pQE-60 vector to construct pQE-60-bcsCD (Figure 2b). However, the original pACYCDuet-1 vector harbors a T7 promoter-a regulatory element incapable of driving gene expression in E. coli DH5α, as this strain lacks T7 RNA polymerase. To address this limitation, we first cloned bcsAB into the pQE-60 vector to construct an intermediate plasmid. Subsequently, we cloned the bcsAB fragment (along with the T5 promoter and lambda t0 terminator derived from pQE-60) into the pACYCDuet-1 vector to complete the construction of pACYCDuet-1-bcsAB. The bcsAB gene fragment (Figure 2c), bcsCD gene fragment (Figure 2c), pQE-60 vector fragment (Figure 2d), T5 promoter-bcsAB-lambda t0 terminator gene fragment (Figure 2e) and pACYCDuet-1 vector fragment (Figure 2e) were amplified by PCR, and the resulting components were subjected to In-fusion Cloning.

Figure 2. Plasmid map, PCR verification and sequencing identification of recombinant plasmids pACYCDuet-1-bcsAB and pQE-60-bcsCD.
a. Plasmid map of pACYCDuet-1-bcsAB. b. Plasmid map of pQE-60-bcsCD. c. 1% agarose gel electrophoresis of PCR-amplified bcsAB (4,675 bp) and bcsCD (4,451 bp). M: DNA marker; Lane 1: bcsAB; Lane 2: bcsCD. d. 1% agarose gel electrophoresis of PCR-amplified pQE-60 vector (3,423 bp). M: DNA marker. e. 1% agarose gel electrophoresis of PCR-amplified T5 promoter-bcsAB-lambda t0-terminator (4,915 bp) and pACYCDuet-1 vector (2,276 bp). M: DNA marker; Lane 1: T5 promoter-bcsAB-lambda t0-terminator; Lane 2: pACYCDuet-1 vector. f. Sequencing results of the recombinant plasmid pACYCDuet-1-bcsAB. g. Sequencing results of the recombinant plasmid pQE-60-bcsCD.

Next, the recombinant plasmids were transformed into E. coli DH5α. Based on the sequencing results (Figure 2f and 2g), we successfully obtained the plasmids pACYCDuet-1-bcsAB and pQE-60-bcsCD. These two recombinant plasmids were then co-transformed into E. coli DH5α and BL21(DE3) strains, respectively.

Test

Both engineered strains were cultured and induced for expression in LB medium. When the OD₆₀₀ value reached 0.8, IPTG was added to a final concentration of 0.05 mmol/L to induce expression under the conditions of 25°C and 180 rpm, with a control group set up simultaneously. After 6 h, visible gel-like dispersed structures were observed in the medium containing the DH5α strain (Figure 3a), while no such phenomenon was detected in the medium with the BL21(DE3) strain. After another 12 h, 1 mL of induced cells was harvested via centrifugation at 10,000 × g for 10 min at 4°C. The harvested cells were first washed twice with nuclease-free water, then mixed with 2× sample buffer, and finally boiled for 5 min to denature the proteins. These processed samples were subsequently used for SDS-PAGE and Coomassie Brilliant Blue staining analysis (Figure 3b).

Figure 3. Expression of BcsABCD
a. BC was detected as white gel-like dispersed structures in DH5α 6 h after 0.05 mM IPTG induction. b. SDS-PAGE of expression products of BcsABCD. Lane 1: Protein ladder; Lane 2-3: DH5α protein samples obtained from induction with 0.05 mmol/L IPTG and from the uninduced group, respectively. Lane 4-5: BL21 protein samples obtained from induction with 0.05 mmol/L IPTG and from the uninduced group, respectively.

The results showed that all four key proteins were detected as the expected molecular weights in both DH5α and BL21 strains. However, visible BC was only formed in DH5α. In contrast, although BL21 (DE3) also expressed the bcsABCD operon under the same conditions, it failed to produce visible BC.

Learn

Based on the result above, we reviewed relevant literature and identified the potential reasons for the differences in BC expression between the DH5α and BL21(DE3): first, the membrane-associated BcsABCD required for BC synthesis exhibit toxicity when overexpressed in BL21(DE3), disrupting critical physiological functions of the host cell membrane and interfering with the normal cellular environment necessary for BC synthesis; second, although SDS-PAGE analysis showed that BL21(DE3) could express recombinant bcs family proteins after induction, these proteins may form inactive inclusion bodies due to abnormal folding, preventing their assembly into functional BC synthesis complexes with "catalysis-transport" capabilities^[2]. Therefore, we selected DH5α for subsequent BcsABCD expression experiments.

Loop 2

Design

The overexpression of the BcsABCD requires careful balancing. Insufficient induction leads to low-level BC production, whereas excessive induction risks protein misfolding, inclusion body formation, and metabolic burden. To address this, we needed to determine the optimal induction conditions for the BcsABCD in DH5α.

Build

To determine the optimal IPTG concentration for inducing the expression of the BcsABCD, we designed the following concentration gradient: 0, 0.05, 0.1, 0.5, and 1.0 mmol/L. The expression conditions were the same as before (25°C, 180 rpm, 18 h).

Test

After 6 h, gel-like dispersed structures were first observed in the medium with a final IPTG concentration of 0.05 mmol/L, followed by those in the media containing 0, 0.1, 0.5 mM IPTG. After another 12 h, denatured proteins were obtained following the previous methods for SDS-PAGE and Coomassie Brilliant Blue staining analysis.

Figure 4. SDS-PAGE of expression products of BcsABCD under different induction conditions in DH5α.
Lane 1: Protein ladder; Lane 2-6: DH5α protein samples derived from induction with 0, 0.05, 0.1, 0.5 and 1.0 mmol/L IPTG, respectively.

It could be observed that the expression levels of BcsABCD were relatively higher in the groups induced with 0, 0.05, and 0.1 mM IPTG, but the overall difference among all groups was not significant (Figure 4). Given that bcsA catalyzes BC synthesis using UDP-glucose as a direct substrate, DH5α first converts glucose in the medium into UDP-glucose. Therefore, changes in the glucose concentration in the culture medium should be correlated with BC production. To further evaluate the optimal induction conditions, we quantified the temporal changes in glucose concentration in the culture medium during the cultivation of the DH5α strain. We sampled at 0, 6, 18, and 24 h respectively, determined the glucose concentration in the medium, and plotted a glucose concentration curve based on these data (Figure 5), aiming to indirectly calculate the BC synthesis rate.

Figure 5. Changes in glucose concentration during cultivation of DH5α strain under induction with different IPTG concentrations.
a. Schematic diagram of glucose concentration changes in 96-well plates at different incubation times and IPTG concentrations. b. Time-dependent curve of glucose concentration in medium during cultivation of DH5α strain induced with different IPTG concentrations

Notably, for the groups with 0 mmol/L and 0.05 mmol/L IPTG, the glucose concentration first increased and then decreased over time, and the glucose in the medium barely changed after approximately 18 h. We used the following formulas to determine the synthesis rate and consumption rate:
$$\text{Glucose synthesis rate (mg/(L}\cdot\text{h))} = \frac{\text{Glucose concentration|}_\text{0 h}^\text{6 h} }{\text{Time interval|}_\text{0 h}^\text{6 h} }$$
$$\text{Glucose synthesis rate (mg/(L}\cdot\text{h))} = \frac{\text{Glucose concentration|}_\text{0 h}^\text{18 h} }{\text{Time interval|}_\text{6 h}^\text{18 h} }$$ The results showed that the group induced with 0.05 mmol/L IPTG exhibited the highest glucose utilization efficiency, with a synthesis rate of 28.7 mg/(L·h) and a consumption rate of 13.7 mg/(L·h). This was followed by the 0 mmol/L IPTG group, where the synthesis rate and consumption rate were 11.6 mg/(L·h) and 6.43 mg/(L·h) respectively, which also indicates that the leaky expression of the T5 promoter is relatively severe. In contrast, the groups induced with higher IPTG concentrations (≥0.1 mmol/L) showed lower and relatively similar glucose utilization efficiencies, with almost no change in glucose concentration in their culture media.

Learn

At an IPTG induction concentration of 0.05 mmol/L, the engineered bacteria balanced the expression of BcsABCD and the host metabolic burden, thereby achieving the highest glucose utilization efficiency to support efficient BC production. In contrast, higher IPTG concentrations might cause metabolic disturbance, making it difficult for the engineered bacteria to synthesize BC. Next, we planned to observe BC using a scanning electron microscope (SEM).

Loop 1

Design

Bacterial cellulose (BC) is naturally colorless and lacks visual diversity, which limits its direct application in textiles and functional materials. To overcome this, we aimed to endow BC with stable coloration by introducing natural pigments. We chose eumelanin, a biopolymer pigment with high stability and biocompatibility. Its synthesis requires the oxidation of L-tyrosine, and tyrosinase (Tyr) is the key catalyst for this step (Figure 6)^[3]. We selected tyrosinase (TyrBm) derived from Bacillus megaterium as the catalyst^[4]. This enzyme has been proven to achieve efficient expression in E. coli BL21(DE3) cultured in TB medium and catalyze the oxidation reaction of L-tyrosine. Since the subsequent "one-step" system needs to be implemented in LB medium, we first verified whether TyrBm could achieve soluble expression in LB medium.

Figure 6. Eumelanin biosynthetic pathway.

Build

We constructed plasmid pET-28a(+)-TyrBm (Figure 7a). The TyrBm gene fragment (Figure 7b) and pET-28a(+) vector fragment (Figure 7c) were amplified by PCR, and the resulting components were subjected to In-fusion Cloning.

Figure 7. Plasmid map of pET-28a(+)-TyrBm and 1% agarose gel electrophoresis of PCR-amplified pET-28a(+)-TyrBm components.
a. Plasmid map of pET-28a(+)-TyrBm. b. 1% agarose gel electrophoresis of PCR-amplified pET-28a(+) vector (5,326 bp). M: DNA marker. c. 1% agarose gel electrophoresis of PCR-amplified TyrBm (909 bp). M: DNA marker.

Next, the recombinant plasmid was transformed into E. coli DH5α. Based on the sequencing results (Figure 8), we successfully obtained the plasmid pET-28a(+)-TyrBm. Then, the plasmid was transformed into E. coli BL21(DE3).

Figure 8. Sequencing results of the recombinant plasmid pET-28a(+)-TyrBm.

Test

We expressed TyrBm in LB and TB media respectively. When the OD₆₀₀ reached 0.6, the expression was induced by IPTG under the conditions of 37°C and 150 rpm. After 12 h, the color changes of the cultures were observed (Figure 9a). After harvesting the cell pellet, it was resuspended in binding buffer (20 mmol/L sodium phosphate buffer, pH 7.5, 500 mmol/L NaCl; and 20 mmol/L imidazole), ultrasonic cell disruption (power: 50 W, work time: 1.0 s, interval: 2.5 s, total time: 40 min, ice bath) was performed to separate the supernatant and precipitate (16,000 × g, 15°C, 20 min), followed by SDS-PAGE and Coomassie Brilliant Blue staining analysis (Figure 9b). The results showed that TyrBm (39.26 kDa) could be expressed in both LB and TB media. However, there were two problems: first, in both media, the expressed TyrBm was mainly distributed in the precipitate with low soluble expression level; second, the TB medium turned significantly black after induction, while no obvious blackening was observed in LB medium.

Figure 9. Expression of recombinant TyrBm under TB/LB culture conditions.
a. Colors of each group of cultures after 12 h of TyrBm expression. -: No addition of IPTG; +: Addition of IPTG. b. SDS-PAGE of expression products of TyrBm. Lane 1: Protein ladder; Lane 2-4: Whole-cell lysate, supernatant and pellet from uninduced cells in TB medium, respectively; Lane 5-7: Whole-cell lysate, supernatant and pellet from uninduced cells in LB medium, respectively; Lane 8-10: Whole-cell lysate, supernatant and pellet from IPTG induced cells in TB medium, respectively; Lane 11-13: Whole-cell lysate, supernatant and pellet from IPTG induced cells in LB medium, respectively.

Learn

The above experiments confirmed that LB medium could be used as the subsequent optimization medium. Regarding the issue of low soluble expression levels, we hypothesized that induction intensity might be an influencing factor. For the problem that TyrBm was expressed in LB medium but no blackening was observed, we compared the components of TB and LB media and proposed that the blackening phenomenon in TB medium might be attributed to its higher content of tryptone and yeast extract (which are rich in amino acids and trace metal ions). Therefore, we hypothesized that the presence of substrates/cofactors (L-tyrosine and Cu²⁺) in the medium might affect the formation of blackened products.

Loop 2

Design

Literature reports indicate that the substrate for the tyrosinase-catalyzed reaction is L-tyrosine, and the activity/assembly of the enzyme depends on Cu²⁺; common systems are supplemented with 0.5 g/L L-tyrosine and 10 μmol/L CuSO₄. To investigate the effects of substrates/cofactors on the soluble expression of TyrBm and product formation, we designed an IPTG gradient experiment and set up controls with or without substrates and cofactors in the medium.

Build

Six experimental combinations were designed (Table 1), including three levels of IPTG (0, 0.5, 1 mmol/L) and controls with or without substrates and cofactors (0.5 g/L L-tyrosine + 10 μmol/L CuSO₄) in the medium.

Table 1. Detailed information of experimental combinations

Number	IPTG（mmol/L）	Substrates and cofactors addition (0.5 g/L L-tyrosine + 10 μmol/L CuSO4)
1	0	-
2	0	+
3	0.5	-
4	0.5	+
5	1	-
6	1	+

-: No addition of 0.5 g/L L-tyrosine + 10 μmol/L CuSO₄; +: Addition of 0.5 g/L L-tyrosine + 10 μmol/L CuSO₄.

The colors of each group before expression were shown in Figure 10a. The expression conditions were the same as before (37°C, 150 rpm, 12 h).

Test

The color changes of the cultures were observed (Figure 10a), and each group of samples was subjected to ultrasonic cell disruption, supernatant-precipitate separation, SDS-PAGE, and Coomassie brilliant blue staining analysis (Figure 10b).

Figure 10. Expression of recombinant TyrBm among the 6 experimental combinations.
a. Color changes of cultures in each group (Table 1) after 12 h of TyrBm expression. b. SDS-PAGE of expression products of TyrBm. Lane 1: Protein ladder; Lanes 2–3: Supernatant and pellet from uninduced cells in LB medium without substrates and cofactors, respectively; Lanes 4–5: Supernatant and pellet from uninduced cells in LB medium with substrates and cofactors, respectively; Lanes 6–7: Supernatant and pellet from 0.5 mmol/L IPTG induced cells in LB medium without substrates and cofactors, respectively; Lanes 8–9: Supernatant and pellet from 0.5 mmol/L IPTG induced cells in LB medium with substrates and cofactors, respectively; Lanes 10–11: Supernatant and pellet from 1 mmol/L IPTG induced cells in LB medium without substrates and cofactors, respectively; Lanes 12–13: Supernatant and pellet from 1 mmol/L IPTG induced cells in LB medium with substrates and cofactors, respectively.

The results showed that the soluble expression level of TyrBm was relatively high under two conditions: 0.5 mmol/L IPTG with the addition of L-tyrosine and Cu²⁺, and 1 mmol/L IPTG without the addition of L-tyrosine and Cu²⁺ (Figure 5c). Given that our "one-step" system requires the addition of substrates and cofactors at the beginning of the dyeing module to accelerate the dyeing process, we selected 0.5 mmol/L IPTG as the appropriate induction concentration. Notably, under the condition of LB with substrates and cofactors, the culture medium showed obvious blackening even without IPTG addition (Figure 5b). This suggested that in LB medium, tyrosine oxidation/polymerization or spontaneous discoloration triggered by medium components occurs even without induction.

Learn

We confirmed that 0.5 mmol/L IPTG was the appropriate induction concentration. However, non-induced blackening occurred when substrates and cofactors were added to LB medium. We hypothesized that components in the medium might be involved in the spontaneous oxidation of L-tyrosine.

Loop 3

Design

It was necessary to clarify the cause of blackening of TyrBm under the condition of LB with substrates and cofactors. We knew from previous studies that TB medium was commonly used as the conventional expression system for TyrBm. Therefore, we first needed to confirm whether the same blackening phenomenon occurred under the classic TB condition, so as to determine whether this phenomenon was specific to the target protein or caused by the LB medium.

Build

Eight experimental combinations were designed (Table 2): LB or TB medium with ±IPTG (0 or 0.5 mmol/L) and ±substrates and cofactors (0.5 g/L L-tyrosine + 10 μmol/L CuSO₄).

Table 2. Detailed information of experimental combinations.

Number	Medium type	IPTG (mmol/L)	Substrates and cofactors addition (0.5 g/L L-tyrosine + 10 μmol/L CuSO4)
1	LB	0	-
2	LB	0	+
3	LB	0.5	-
4	LB	0.5	+
5	TB	0	-
6	TB	0	+
7	TB	0.5	-
8	TB	0.5	+

-: No addition of 0.5 g/L L-tyrosine + 10 μmol/L CuSO₄; +: Addition of 0.5 g/L L-tyrosine + 10 μmol/L CuSO₄.

The expression conditions were the same as before (37°C, 150 rpm, 12 h).

Test

The colors of the cultures were observed (Figure 11).

Figure 11. Colors of each group (Table 2) of cultures after 12 h of TyrBm expression.

The results showed that under the non-induced condition (0 mmol/L IPTG) with substrates and cofactors addition, the LB culture medium turned significantly black, while no blackening was observed in TB culture.

Learn

The experiment indicated that the difference in composition between the two media might be a key factor contributing to the non-induced blackening of substrates in LB medium.

Loop4

Design

The main difference between TB medium and LB medium lies in that TB medium contains an additional 0.4% (v/v) glycerol and 89 mmol/L sodium phosphate buffer. We hypothesized that either one of these two components or their synergistic effect inhibited the non-induced oxidation of L-tyrosine, thereby reducing blackening.

Build

Using LB medium as the base, we employed the controlled variable method to add 0.4% (v/v) glycerol or 89 mmol/L sodium phosphate buffer. Against a background containing substrates and cofactors, we set up six experimental groups (Table 3) with IPTG concentrations of 0 or 0.5 mmol/L.

Table 3. Detailed information of experimental combinations.

Number	Medium type	IPTG (mmol/L)	89 mmol/L sodium phosphate buffer	0.4% (v/v) glycerol	Substrates and cofactors addition (0.5 g/L L-tyrosine + 10 μmol/L CuSOCuSO4)
1	LB	0.5	-	-	+
2	LB	0	-	-	+
3	LB	0.5	-	+	+
4	LB	0	-	+	+
5	LB	0.5	+	-	+
6	LB	0	+	-	+

-: No addition of the corresponding component; +: Addition of the corresponding component.

The expression conditions were the same as before (37°C, 150 rpm, 12 h).

Test

The colors changes of the media of the were observed (Figure 12).

Figure 12. Culture supernatant obtained without bacteria after induction for 12 h of each group (Table 3).

The results showed that after adding glycerol or sodium phosphate buffer to the LB medium, the non-induced blackening of the LB medium was significantly reduced, indicating that both could inhibit this phenomenon to a certain extent. Among the two, the addition of glycerol exhibited the best effect.

Learn

Based on the above results, we found that adding glycerol to the medium could inhibit the non-enzymatic darkening phenomenon in LB medium. For the subsequent large-scale expression of TyrBm, we will adopt LB supplemented with 0.4% (v/v) glycerol, 0.5 mmol/L IPTG, 0.5 g/L L-tyrosine, and 10 μmol/L CuSO₄ as the standard conditions.

Loop5

Design

After completing the expression, we needed to determine the catalytic activity and kinetic characteristics of TyrBm, clarify its monophenol oxidation activity (L-tyrosine -> L-DOPA) and diphenol oxidation activity (L-DOPA -> dopaquinone), and obtain its kinetic parameters.

Build

Under the optimized culture conditions, TyrBm was induced for expression at 37°C and 150 rpm for 12 h, with the group without substrates and cofactors addition used as a control. After expression, the cells were disrupted by ultrasonication, and the supernatant and precipitate were separated. Subsequently, Immobilized Metal Affinity Chromatography (IMAC) purification was performed using a HypurT Ni-NTA 6FF (His-Tag) prepacked Gravity column (1 mL), with elution carried out using 20 mmol/L sodium phosphate buffer (pH 7.5) containing 500 mmol/L NaCl and either 300 mmol/L or 500 mmol/L imidazole. The purified enzyme was subjected to dialysis treatment against 50 mmol/L sodium phosphate buffer (pH 6.5) with 0.01 mmol/L CuSO₄ at 4°C for 24 h.

Test

According to SDS-PAGE analysis, TyrBm achieved partial soluble expression under both sets of conditions (Figure 13a). After IMAC elution, the main target band appeared in the 500 mmol/L imidazole buffer fraction (Figure 13b).

Figure 13. SDS-PAGE analysis of TyrBm expression products and IMAC-purified fractions.
a. SDS-PAGE of expression products of TyrBm. Lane 1: Protein ladder; Lanes 2-3: Supernatant and pellet from 0.5 mmol/L IPTG induced cells in LB medium with 0.5 g/L L-tyrosine and 10 µmol/L CuSO₄, respectively; Lanes 4-5: Supernatant and pellet from 0.5 mmol/L IPTG induced cells in LB medium, respectively. b. SDS-PAGE analysis of protein fractions eluted from the Ni-NTA column. Lane 1: Protein ladder; Lane 2-5: TyrBm supernate obtained from LB medium supplemented with with 0.5 g/L L-tyrosine and 10 µmol/L CuSO₄ after being bound to Ni-NTA resin, eluted with 20, 300 and 500 mmol/L imidazole, respectively ; Lane 6-9: TyrBm supernate obtained from LB medium after being bound to Ni-NTA resin, TyrBm supernate eluted with 20, 300 and 500 mmol/L imidazole, respectively.

In a 96-well plate, L-tyrosine and L-DOPA were used as substrates to determine the monophenolase and diphenolase activities of TyrBm, respectively. Besides, an increasingly obvious colour occurred in experimental group compared to control group along with the increase of concentration (Figure 14). The reaction system was incubated at 37°C, and the change in absorbance at 475 nm was recorded (ε₄₇₅=3,600/(mol·cm)). The amount of dopachrome produced was calculated using the formula ΔA = A₁ − A₀, and the enzyme activity unit was converted based on this (1 U is defined as the amount of enzyme required to produce 1 μmol of dopachrome per minute). Subsequently, the specific activity was calculated according to the protein concentration (determined by the Bicinchoninic Acid Assay method).

Figure 14. The 96 Well Cell Culture Plates of tyrosinase TyrBm used for activity＆kinetic assay and ΔA of each well.
a. TyrBm purified from LB medium supplemented with L-tyrosine and CuSO₄. b. TyrBm purified from LB medium. Reaction system incubated at 37°C for 30 min, absorbance measured at 475 nm (ε₄₇₅=3,600/(mol·cm))

The results showed that the monophenolase activity without IPTG addition was 1,461 U/mg and the diphenolase activity was 4,032 U/mg; with IPTG addition, the monophenolase activity was 35,648 U/mg and the diphenolase activity was 199,346 U/mg. These results indicated that substrates and cofactors significantly enhanced the catalytic activity of TyrBm. Subsequently, based on the group with the highest specific activity, Michaelis-Menten and Lineweaver-Burk curves were plotted (Figure 15), yielding the following Michaelis constant (K_m) and maximum velocity (v_max) results: for monophenol oxidation, K_m = 34.37 μmol/L and v_max = 2.487 μmol/(L·min); for diphenol oxidation, K_m = 540.84 μmol/L and v_max = 2.274 μmol/(L·min).

Figure 15. The kinetic assay results of TyrBm.
a-b. Michaelis-Menten plot of enzymatic reaction from tyrosine to dopachrome experiments and from L-DOPA to dopachrome. c-d. Lineweaver-Burk double reciprocal plot of enzymatic reaction from L-tyrosine to dopachrome experiments and from L-DOPA to dopachrome.

Dyeing Time Simulation via PDE Model
To quantitatively determine the specific dyeing duration, a combined approach of mathematical modeling and biological simulation was conducted. In the mathematical simulation, a partial differential equation (PDE) model was established to describe the transport and diffusion of pigment molecules within the cellulose matrix. Using the finite element method (FEV) for numerical computation, we obtained the temporal and spatial evolution of pigment concentration across different regions. The simulation results revealed that the dyeing rate varied significantly among local zones, which can be attributed to the porous and heterogeneous nature of the cellulose structure. Due to the presence of pores with varying diameters, pigment penetration occurred at different rates, leading to non-uniform color development across the material (Figure 16):

Figure 16. Simulation slicing of the staining time calculation process using the PDE algorithm.

Based on the simulation outcomes, the effective dyeing duration was determined to be approximately 37,825 s (≈10.5 h), corresponding to the point at which the pigment concentration within the cellulose matrix reached equilibrium (Figure 17). Beyond this time, the diffusion flux approached zero, and the color intensity exhibited minimal further change. This result provided a quantitative reference for optimizing dyeing parameters in experimental and industrial applications, ensuring sufficient coloration while avoiding unnecessary processing time.

Figure 17. Computational visualization of the PDE algorithm for the quantitative analysis of staining time dynamics.

If you are interested to learn more, please visit the Model page.

The Model page

Learn

It could be concluded from the experimental results that the supplementation of L-tyrosine and Cu²⁺ significantly promoted the activity and stability of TyrBm. TyrBm exhibited a high affinity for L-tyrosine, while the oxidation rate of L-DOPA was slightly faster, which was consistent with its natural catalytic order. A dual-pore PDE model was also built to characterize cellulose dyeing; simulations showed most nodes reach dye equilibrium at approximately 37,825 s, some unsaturated.

Loop1

Design

The temperature regulation module played a crucial role in our weaving and dyeing platform. It contained three main parts: CI857, FourU riboswitch and PhlF repressor. To better demonstrate the module's performance, we designed a testing circuit (Figure 18). Once induced by IPTG, CI857 expresses. Under 25°C, CI857 forms dimers that bind with R promoter to confine the combination of RNA polymerase with the promoter, thus blocking the expression of PhlF and mRFP1. Meanwhile, FourU stem loop comes into being inhibiting the RBS and again weakens the repressor and the fluorescence protein. The restriction of PhlF initiates the translation of sfGFP guided by PhlF promoter that is inhibited by the repressor. Under 37°C, however, CI857 dimers disassociate, which eliminates the inhibition of R promoter while FourU exposed the RBS binded with it, making ribosome possible to bind with RBS^{[5, 6]}. As a result, mRFP1 and PhlF emerge smoothly, the latter integrated with PhlF promoter, decreasing the synthesis of sfGFP. In short, mRFP1 expresses more at higher temperature and sfGFP possesses stronger fluorescence intensity at lower temperature.

Figure 18. The testing circuit of the temperature regulation circuit that can express proteins with different fluorescence at distinct temperatures.

Build and Test

To ensure the feasibility of our temperature regulation module, mathematical modeling was processed to predict the structure transformation of FourU. The simulation mainly focused on the binding of UUUU sequence of the riboswitch with the SD sequence of RBS (AGGAGG). Firstly, we leveraged RNA Structure developed by Mathews Laboratory to to forecast secondary structures of FourU at distinct temperatures and discovered stable base pairing between AGGAGG and UUUU at 310 K (36.85°C), partially disrupted pairing at 430 K (156.85°C) and completely disassembled pairing at 520 K (246.85°C) (Figure 19).

Figure 19. Predicted secondary structures from RNA Structure of FourU at different temperatures.

The secondary structure prediction demonstrated the potential of FourU to regulate genes behind them, but the precise transition temperature should only be forecasted with a quantitative analysis. Hence, a translation initiation potential σ was introduced: $$\sigma=\exp (-\frac{1}{RT}(E_{total}))$$ To be specific, R = 1.987 × 10⁻³ kcal/(mol⋅K) is the Boltzmann constant, T is the absolute temperature, and E_total could be calculated by the formula: $$E_{total}=E_{SD}+E_{tRNA}+E_{stem}+E_{4U{binding}}+E_{coop}$$ Whereupon, we used Python to plot a normalized translation potential-temperature curve, finding out that 38°C was the unchaining temperature for the riboswitch, similar to the expected transition temperature (37°C) (Figure 20). As for the subtle error, it was suspected that compared to the real circumstances, our model could not cover all situation variates. For instance, the ionic environment was simplified to sodium. More information about our modeling was provided at Model page.

The Model page

In a word, 37°C was a suitable transition temperature for our project, which could be further authenticated with experiment.

Figure 20. Temperature response curve of FourU. After the primary verification of FourU, the plasmid construction took place.

At first, we planned to use single fragment In-Fusion cloning to link the gene fragment and the vector (Figure 21).

Figure 21. The scheme of single fragment In-Fusion cloning for pET-28a(+)-CI857-FourU-PhlF-mRFP1-sfGFP. SfGFP was amplified from mRFP1-sfGFP (Figure 22).

Figure 22. 1% agarose gel electrophoresis of the PCR amplified sfGFP (799 bp).

Whereafter, we amplified the 7,587 bp pET-28a(+)-CI857-FourU-PhlF-mRFP1 vector by TCI-F (5'-GATCCGGCTGCTAACAAAGCC-3') and TCI-R (5'-ACCTTAACGATACGGTACGTTTCGTAT-3') with the following procedure but failed (Table 4).

Table 4. PCR procedure for linearized pET-28a(+)-CI857-FourU-PhlF-mRFP1 vector.

Temperature(°C)	Time	Cycles
95	30 s	×1
95	15 s	×30
56	15 s	×30
72	4 min	×30
72	5 min	×1

Hence, primers were inspected carefully whose base pair amount, annealing temperature and GC content were all able to successfully amplify the target fragment. Given the correctness of primer design, we began to think whether our procedure had made amplification difficult. Through discussion and searching for alternative procedures from articles, specifications and the Internet, two optimized strategies were carried out. For one, we adjusted the original extension time span to provide DNA polymerase with more time to form the DNA strand complementary to template. Meanwhile, cycles were decreased from 30 to 28 so as to reduce the opportunity of non-specific binding (Table 5).

Table 5. Optimized PCR procedure for linearized pET-28a(+)-CI857-FourU-PhlF-mRFP1 vector.

Temperature(°C)	Time	Cycles
95	30 s	×1
95	15 s	×28
56	15 s	×28
72	8 min	×28
72	5 min	×1

For another, Touch Down two-step PCR approach was given a trial, which started at a relatively high temperature, confining non-specific binding whereas at a lower temperature, correct products with a higher concentration made the amplification more accurate (Table 6).

Table 6. Touch Down PCR procedure for linearized pET-28a(+)-CI857-FourU-PhlF-mRFP1 vector.

Temperature(°C)	Time	Cycles
95	3 min	×1
95	15 s	×5
74	8 min	×5
95	15 s	×5
72	8 min	×5
95	15 s	×5
70	8 min	×5
95	15 s	×25
68	8 min	×25
68	5 min	×1

Unfortunately, the two new approaches was unable to acquire the linearized vector. Therefore, we gave up the plotted In-Fusion coloning strategy and chose to adopt a multi-fragment In-Fusion cloning way (Figure 23).

Figure 23. The scheme of multi-fragment In-Fusion cloning for pET-28a(+)-CI857-FourU-PhlF-mRFP1-sfGFP.

Primers were designed to break the vector into two parts of 3,298 and 4,310 bp respectively, making the amplification possible (Figure 24). With the final catalysis of Exnase MultiS, the construction of pET-28a(+)-CI857-FourU-PhlF-mRFP1-sfGFP was eventually accomplished.

Figure 24. 1% agarose gel electrophoresis of the PCR amplified pET-28a(+)-CI857-FourU-PhlF-mRFP1 fragment 1 (4,310 bp) and pET-28a(+)-CI857-FourU-PhlF-mRFP1 fragment 2 (3,298 bp).

We transformed the plasmid into E. coli DH5α and ensured its correctness with Sanger sequencing (Figure 25).

Figure 25. Sanger sequencing confirmed the correct assembly of the CI857-FourU-PhlF-mRFP1-sfGFP plasmid.

Learn

Long fragment PCR has long been an obstacle in gene amplification since primers are easier to miss-prime during long extending time whilst GC-rich region will stop DNA polymerizing^[7]. Thus, PCR procedures should be designed carefully to meet the demand of distinct target products. For instance, a relatively long extending time guided by the specification is beneficial for the binding of DNA polymerase with single string sequence whereas the reduction in cycles enhances the fidelity during the amplification. If all changes fail to obtain the target sequence, different methods are worth a shot. Touch Down behaves better at the acquisition of complicated or long fragments and the shift from single fragment In-Fusion to multi-fragment In-Fusion cloning strategy acts as a much easier way if the construction of recombinant plasmid is demanded later. Additionally, optimization of codons themselves at synthesis is able to relieve the challenges at the following PCR especially after the decrease in GC content. In this condition, however, restriction endonuclease is also an option since we attempted to acquire the whole linearized vector, meaning that it was of necessity to add cutting sites when synthesising plasmids.

Loop2

Design

Due to E. coli BL21(DE3)'s efficient protein expression capacity, it was an ideal choice for relative fluorescence testing.

Build

Samples were collected periodically to measure OD₆₀₀ as well as sfGFP and mRFP1 intensity using a microplate reader. The mRFP1 intensity was determined by using an excitation wavelength of 584 nm and an emission wavelength of 666 nm while the sfGFP intensity was measured with an excitation wavelength of 480 nm and an emission wavelength of 520 nm. Relative fluorescence intensity over OD₆₀₀ was leveraged calculated using the following formula^[8]: $${(\frac{\text{FL}}{\text{OD}})}_\text{corrected}=\frac{\text{FL}-\text{FL}_\text{bg}}{\text{OD}-\text{OD}_\text{bg}}$$

Test

The tested right plasmid was transformed into E. coli BL21(DE3) to express fluorescent protein. Colonies were then inoculated into 5 mL of LB medium and incubated at 37°C with rotation at 250 rpm for 16 h so as to establish the seed cultures. Afterwards, we shifted 50 μL of the cultures into 5 mL of of LB medium at 37°C, 250 rpm until their OD₆₀₀ reached 0.9, where they were then induced by 1 mmol/L IPTG and the cultures were incubated at 25°C, 30°C and 37°C for 34 h. Fluorescence emitted by sfGFP and OD₆₀₀ at each temperature were measured at 6, 9, 12 and 30 h postinoculation by the microplate reader with three repetitions (Figure 26).

Figure 26. Figure 26. Relative fluorescence intensity over OD₆₀₀ at different temperatures.

The unstable growth at the few hours after the addition of IPTG might lead to the relatively high standard error. However, since the line tended to go smoother, the reliability of our detection was guaranteened. As was depicted, with time passing, the difference of sfGFP signal at different temperatures enlarged. Since the sfGFP was placed at the same place of the Bcs operon anticipated to produce bacterial cellulose at 25℃ where the relative fluorescence intensity mounted to the highest, the strongest expression of the operon came into being at the temperature. Meanwhile, our platform initiated the dyeing module at 37℃. Under such temperature, the sfGFP expressed weakly from which it could be inferred that the Bcs operon expressed less enzymes making TyrBm and CYP102A dye without disturbance.

Learn

As for the expression of fluorescence proteins, these proteins were of no differnce with normal proteins, hence where they are placed attaching crucial significance to its translation. To be specific, in our experiment design, behind the PhlF repressor lied mRFP1 regulated by the same RBS. Therefore, we failed to achieve the correct expression level of it, which was possibly because the second gene behind the RBS expressed relatively poorer than the first. Hence, the next time we plan to conduct verification with the usage of fluorescence protein, it is reasonable to add an RBS before it to enhance its expression.

Loop 1

Design

Hydrophobicity of our bacterial cellulose (BC) was achieved with the employment of biofilm-surface layer protein A (BslA) from Bacillus subtilis with a double cellulose binding module (dCBM). After a comprehensive review of corellated research literature, it is certificated that the recombinant protein can bind effectively to the BC, thus forming a waterproof coat (Figure 27)^[9].

Figure 27. Ideal interaction of BslA-CBMs with BC fibres^[10].

Build

First, we amplified the bslA-dcbm gene and the linear pET-28a(+) vector by the usage of PCR. Then, In-fusion cloning was leveraged to construct pET-28a(+)-BslA-dCBM recombinant plasmid that was transferred into E. coli DH5α (Figure 28).

Figure 28. The plasmid map of pET-28a(+)-BslA-dCBM and 1% agarose gel electrophoresis of the PCR amplified pET-28a(+)-BslA-dCBM parts.
a. The plasmid map of pET-28a(+)-BslA-dCBM. b. 1% agarose gel electrophoresis of the PCR amplified BslA-dCBM (918 bp) and pET-28a(+) vector (5,323 bp).

After colony PCR and Sanger sequencing, we ensured the correctness of the plasmid we built (Figure 29).

Figure 29. Verification of recombinant plasmid pET-28a(+)-BslA-dCBM.
a. 1% agarose gel electrophoresis of colony PCR of using T7 and T7 ter primers. b. The result of sequencing the BslA-dCBM of the recombinant plasmid.

Next, the plasmid was transformed into E. coli BL21(DE3) to express protein in LB medium at 30°C for 16 h induced by 0, 0.2, 0.5 and 1 mmol/L IPTG.

Test

Fusion protein as BslA-dCBM, AlphaFold and GROMACS were leveraged to evaluate its stability. The mathematical simulation started with the prediction of the structure of BslA and CBM respectively by AlphaFold (Figure 30). BslA structure prediction yielded ipTM = pTM = 0.69, indicating high confidence. BslA adopts a β-barrel fold, supporting hydrophobic interactions consistent with its biological role. CBM structure prediction yielded ipTM = pTM = 0.7, indicating high confidence. CBM adopts a β-sandwich fold, typical of carbohydrate-binding modules, enabling cellulose recognition and adhesive function.

Figure 30. BslA and CBM predicted result by AlphaFold.

AlphaFold was also used to forecast the recombinant protein's strycture, but found that confidence drops in the flexible linker region instead of functional domains, which indicated that their structures had been altered. Nevertheless, the relevant articles wrote that the fusion protein could serve expected function (Figure 31). Hence, it was not appropriate to leverage Alphafold to forecast the fusion protein's structure.

Figure 31. Diagonal matrix comparison of BslA-dCBM showing the confidence pridicted by Alphafold.
a. BslA predicted result by AlphaFold. b. CBM predicted result by AlphaFold.

With the usage of Kabsch Comparison, it could be told that CBM domains (cbm1, cbm2) showed minimal displacement (RMSD < 1.1 Å) and BslA deviated more (4.264 Å) but remained acceptable thanks to linker flexibility indicating that the fusion protein structure's rationality since all domains' functions remained stable. Meanwhile, we leveraged GROMACS simulation which also confirmed stable BslA and CBM conformations as the result of β-structures maintenance and no unfolding or disorder (Figure 32). Other information of the fusion protein's structure prediction is available in Model page.

The Model page

Figure 32. GROMACS simulation results of BslA-dCBM for 3 ns.

We examined the translation of BslA-dCBM by SDS-PAGE. As is shown in Figure 33, no stronger bonds were found at 34.68 kDa, the relative molecular mass of BslA-dCBM.

Figure 33. SDS-PAGE of expression products of BslA-dCBM.
Lane 1: Protein ladder; Lane 2-4: Whole-cell lysate, supernatant and pellet from uninduced cells, respectively; Lane 5-7: Whole-cell lysate, supernatant and pellet from 0.2 mmol/L IPTG induced cells, respectively; Lane 8-10: Whole-cell lysate, supernatant and pellet from 0.5 mmol/L IPTG induced cells, respectively; Lane 11-13: Whole-cell lysate, supernatant and pellet from 1 mmol/L IPTG induced cells, respectively.

Hence, we managed to distinguish the target protein from others by deleting the 6 × His tag before it and to optimize the expression condition guided by corresponding articles. After the verification with SDS-Page, 1 mmol/L IPTG was considered the most appropriate concentration for our engineered bacterial to express BslA-dCBM (Figure 34).

Figure 34. SDS-PAGE of expression products of BslA-dCBM after deleting the His tag before.
Lane 1: Protein ladder; Lane 2-4: Whole-cell lysate, supernatant and pellet from uninduced cells, respectively; Lane 5-7: Whole-cell lysate, supernatant and pellet from 0.2 mmol/L IPTG induced cells, respectively; Lane 8-10: Whole-cell lysate, supernatant and pellet from 0.5 mmol/L IPTG induced cells, respectively; Lane 11-13: Whole-cell lysate, supernatant and pellet from 1 mmol/L IPTG induced cells, respectively.

Afterwards, we decided to purify the hydrophobic protein by IMAC with common methods guided by the specification whose SDS-PAGE, however, demonstrated relatively shallow bonds after the elution of buffer with high imidazole concentration (Figure 35).

Figure 35. SDS-PAGE of expression products of BslA-dCBM purified by IMAC using traditional methods.
Lane 1: Protein marker; Lane 2: BslA-dCBM supernate after being bound to Ni-NTA resin; Lane 3: BslA-dCBM supernate eluted with Buffer A; Lane 4-7: BslA-dCBM supernate eluted with 50, 150, 300 and 500 mmol/L imidazoles.

Learn

Successful as the construction of pET-28a(+)-BslA-dCBM-Kan was, the fusion protein's production efficiency was far below our expectation. Such phenomenon reminded us of the significance of expression conditions. During the protein expression experiment mentioned above, we added IPTG when OD₆₀₀ reached 0.6, incubated the engineered E. coli for 16 h as well as resuspened with 10mL Buffer A (100 mmol/L Tris, 0.5 mmol/L NaCl and 20 mmol/L imidazole), which was effective for regular protein but obviously unsuitable for our fusion protein. Then, by studying relevant literature, we ultimately confirmed our final expression protocal, leading to the success of the acquisition of BslA-dCBM. Meanwhile, since intracellular proteins was suspected to disturb the normal bonds formed by the recombinant protein, we learnt that the size of the target protein should be under considerartion while designing a plasmid, which was able to save a great deal of time in the following experiment. With the ultimate success in protein expression, we attempted to depurate BslA-dCBM but failed since no strong bonds were found at high immidazole concentration. Hence, new protocols were carried out.

Loop 2

Design

As the functional testing of BslA-dCBM was scheduled, the pure protein solution was of tremendous necessity. Hence, we attempted to adopt Immobilised Metal Affinity Chromatography (IMAC) to realize the purification process with the help of Ni-NTA resin and dialysis to remove immidazole impurity from the elution solution.

Build

The usage of traditional IMAC ways to purify the recombinant protein had proved invalid, from which we suspected that the His tag of BslA-dCBM failed to integrate with nickel. Therefore, from scientists' experience and depuration approaches mentioned in assays, we discussed and summarized two strategies to enhance the binding of our protein and nickel. For one, before the elution of imidazole, protein samples could be incubated at 4°C for several hours. For another, we let the supernatant interact with the Ni-NTA resin 5 times.

Test

As depicted in Figure 31, without optimition, there emerged two bands at 50 mmol/L imidazole buffer which could be explained by following theories. To begin with, we supposed that another protein was also equipped with structures similar to His tag that enabled it to flow through the resin with imidazole solution. Plus, as 50 mmol/L was not a proper concentration for the elution of regular proteins, the poor combination of the protein and nickel in the Ni-NTA resin could also lead to the generation of multiple bands in SDS-PAGE. Thus, an inspection was conducted to find whether there existed protein impurity and ended with a negative result. From this, we designed a novel scheme to let BslA-dCBM integrate well with nickel. After leveraging new protocols to depurate, we conducted an SDS-PAGE experiment, from which a band near 36 kDa was discovered when BslA-dCBM was eluted with 300 mmol/L imidazoles (Figure 36).

Figure 36. SDS-PAGE of expression products of BslA-dCBM purified by IMAC.
a. BslA-dCBM purified after 4 hours of incubation. Lane 1: Protein ladder; Lane 2-3: Supernatant and pellet from 1 mmol/L IPTG induced cells, respectively; Lane 2: BslA-dCBM supernate after being bound to Ni-NTA resin; Lane 3: BslA-dCBM supernate eluted with Buffer A (100 mmol/L Tris, 0.5 mmol/L NaCl and 20 mmol/L imidazole); Lane 4-7:BslA-dCBM supernate eluted with 50, 150, 300 and 500 mmol/L imidazoles. b. BslA-dCBM flowing 5 times through the Ni-NTA resin. Lane 1: Protein ladder; Lane 2-4: Whole-cell lysate, supernatant and pellet from 1 mmol/L IPTG induced cells, respectively; Lane 5: BslA-dCBM supernate after being bound to Ni-NTA resin; Lane 6: BslA-dCBM supernate eluted with Buffer A; Lane 7-10: BslA-dCBM supernate eluted with 50, 150, 300 and 500 mmol/L imidazoles.

Nonetheless, BslA-dCBM possessed a relative molecular mass of 30.63 kDa twice. Such errors made it difficult to judge whether the purified protein was the target one. Therefore, we conducted a Western Blot (Figure 37), demonstrating that we had acquired BslA-dCBM engineered with 6 × His Tag, thus initiating the following large-scale depuration procedure.

Figure 37. Western Blot of expression products of BslA-dCBM purified by IMAC.
Lane 1-4: BslA-dCBM supernate eluted with 500, 300, 150 and 50 mmol/L imidazoles; Lane 5: BslA-dCBM supernate eluted with Buffer A (100 mmol/L Tris, 0.5 mmol/L NaCl and 20 mmol/L imidazole); Lane 6: BslA-dCBM supernate after being bound to Ni-NTA resin.

Eventually, dialysis of the pure proteins acquired from two depuration approaches took place, where both stategies showcased stong bonds in SDS-PAGE testing (Figure 38), after which we leveraged BCA assay to measure its concentrationand and was found to be 0.025 mg/mL.

Figure 38. SDS-PAGE of expression products of BslA-dCBM after dialysis overnight.
Lane 1: Protein ladder; Lane 2: BslA-dCBM purified after 4 hours of incubation after dialysis overnight; Lane 3: BslA-dCBM flowing 5 times through the Ni-NTA resin after dialysis overnight.

Learn

The complicated purification process suggested that distinct samples should be disposed differently while discussion was necessary for a complete protocal. Additionally, repetitive errors may not be the result of operation mistakes, but could be attributed to objective causes such as the lack of precision of protein ladders at low relative molecular mass, which would be resolved with another method to assist, hence verifying the scientificity of the experiment. The obtainment of BslA-dCBM led us to the final testing of its waterproof capacity.

Loop 3

Design

After the acquisition and purification of our recombinant protein, we discovered a convenient way to verify its hydrophobicity through browsing a great number of articles with the usage of red ink and filter paper^[10]. Similar to our BC, filter paper is composed of cellulose with a mass of hydroxyls, a hydrophilic radical binding well with water and red ink. Meanwhile, the capillary action makes red ink possible to permeate the paper, whereas the repulsive force generated by BslA-dCBM resists ink diffusion, which means that the rate of areas without or with the protein is able to test its hydrophobicity.

Build

500 μL of dialysis buffer, TyrBm solution and BslA-dCBM solution were pipetted in the centre of the circular filter paper (110 mm diameter) and 1 mL of red ink was used to flood the filter paper. Afterwards, we took pictures of the dyed filter paper and calculated its hydrophobicity ratio: $$\text{Hydrophobicity Ratio}=\frac{\text{Red ink coloured area without the target protein}}{\text{Red ink coloured area with the target protein}}$$

Test

As was planned, the testing procedure came off but ended with a meaningless consequence since the whole filter paper was stained red. In our following experiments, we increased our yield, diluted the red ink with deionized water and decreased the volume of dyed water we leveraged, leading to the final success (Figure 39). Details of our new protocol are elaborated in Results.

Figure 39. Waterproof ability test of BslA-dCBM with controls.
a. Red ink dripped on the boundary of dialysis buffer. b. Red ink dripped on the boundary of 0.079 mg/mL TyrBm. c. Red ink dripped on the boundary of 0.079 mg/mL BslA-dCBM. d. Histogram of different treatments' hydrophobicity ratio.

Although the functional testing received success, we were still curious whether such quantity of CBM could achieve the best effect, thus conducting another round of mathematical modeling. This time, we used GROMACS molecular dynamics simulation so as to find out the number of CBMs that achieved the strongest binding of the recombinant protein to BC. The binding force was assessed with three parameters: hydrophobic area, RMSD and structural analysis. Hydrophobic area analysis demonstrated that BslA with 2 or 3 CBMs possessed relatively more area, meaning they were equipped with better waterproof ability (Table 7).

Table 7. Proportions of hydrophobic area of BslA-CBMs.

Quantity of CBM	Number of binding sites	Proportion of hydrophobic area
1	37	55.97%
2	47	58.18%
3	50	58.22%
4	23	58.76%

In addition, RMSD was also calculated. The lower RMSD is, the steadier the structure of the protein is. In Figure 40, we could tell that when the quantity of CBM was 2 or 3, the fusion protein possessed good stability.

Figure 40. RMSD Values of BslA-CBMs during 3 ns molecular dynamics simulation.

Eventually, structural analysis with PyMOL demonstrated that three CBMs spanned both sides of the cellulose plane, which collided with our goal of covering only one side of the cellulose but the dCBM matched well with our goal (Figure 41). If you are interested to learn more, please visit the Model page.

The Model page

Figure 41. Structure diagram of two and three CBMs.
a. Two CBMs bind to cellulose. b. Three CBMs bind to cellulose.

Learn

The failure of the first trial suggested two drawbacks in our waterproof ability testing protocol. Firstly, the low production made the fusion protein hard to perform extraordinary hydrophobicity, which was also proved in articles as the well-performed waterproof coat they experimented on reached 10 mg/mL, far larger than BslA-dCBM we produced. Secondly, the red ink we used possesses a powerful dyeing capability, thus exerting stronger force than the repulsion generated by waterproof fusion protein.

• [1]Augimeri R V, Strap J L. The phytohormone ethylene enhances cellulose production, regulates CRP/FNR_Kx transcription and causes differential gene expression within the bacterial cellulose synthesis operon of Komagataeibacter (Gluconacetobacter) xylinus ATCC 53582[J]. Frontiers in Microbiology, 2015, 6: 1459.
• [2]Al-Janabi S S, Shawky H, El-Waseif A A, et al. Stable, efficient, and cost-effective system for the biosynthesis of recombinant bacterial cellulose in Escherichia coli DH5α platform[J]. Journal of Genetic Engineering and Biotechnology, 2022a, 20(1): 90.
• [3]Walker K T, Li I S, Keane J, et al. Self-pigmenting textiles grown from cellulose-producing bacteria with engineered tyrosinase expression[J]. Nature Biotechnology, 2025, 43(3): 345-354.
• [4]Shuster V, Fishman A. Isolation, cloning and characterization of a tyrosinase with improved activity in organic solvents from Bacillus megaterium[J]. Journal of Molecular Microbiology and Biotechnology, 2009, 17(4): 188-200.
• [5]Jiang S, Yuan J F, et al. Genetically encoded biosensors for constrained biological functions in probiotic Escherichia coli Nissle[J]. ACS Synthetic Biology, 2025, 14: 296-303.
• [6]Wang X, Chen G Q, et al. Reversible thermal regulation for bifunctional dynamic control of gene expression in Escherichia coli[J]. Nature Communications, 2021, 12: 1411.
• [7]Wang S, Xiang S L, et al. Restriction-based multiple-fragment assembly strategy to avoid random mutation during long cDNA cloning[J]. Journal of Cancer, 2015, 6: 632-635.
• [8]Chen H F, Yuan J F, et al. Design of thermoresponsive genetic controls with minimal heat-shock response[J]. ACS Synthetic Biology, 2024, 13: 3032-3040.
• [9]Gilmour K A, Zhang M, et al. Environmentally conscious hydrophobic spray coatings on bacterial cellulose for sustainable and reusable textiles[J]. Journal of Cleaner Production, 2025, 514: 145775.
• [10]Gilmour K A, Zhang M, et al. Biofilm inspired fabrication of functional bacterial cellulose through ex-situ and in-situ approaches[J]. Carbohydrate Polymers, 2023, 304: 120482.