To develop a 7-MX (7-methylxanthine)-dependent auxotrophic strain, we first analyzed the relevant metabolic pathways. Through querying the Kyoto Encyclopedia of Genes and Genomes (KEGG) database, we identified xanthine as a downstream product of 7-MX, which is closely associated with the synthesis of DNA and RNA precursors. The survival of the strain depends on two metabolic pathways (Fig. 1): the primary pathway regulated by the guaB gene, and the secondary pathway utilizing xanthine as a substrate, mediated by the gpt gene. To render the strain dependent on 7-MX for survival, we designed a strategy to disrupt the primary metabolic pathway by knocking out the guaB gene, thereby activating the secondary pathway. This would force the strain to rely on xanthine metabolism, ultimately achieving the construction of the target strain.

Fig. 1 Xanthine metabolism chart

To implement this design, we opted to use the RED homologous recombination technology to knock out the guaB gene and simultaneously replace it with a kanamycin resistance gene (kana). The introduction of the kana not only facilitates subsequent screening and validation, but also serves as a selectable marker, providing convenience for subsequent experiments.

First, we designed and constructed the targeting fragment for homologous recombination. By retrieving genomic information, we obtained the complete sequence of the guaB gene. Subsequently, we amplified approximately 500 bp homologous arms upstream and downstream of the guaB gene using PCR. Next, we used overlap PCR to ligate the upstream and downstream homologous arms with the kana fragment, generatingthe targeting construct. Finally, we co-transformed this construct with the pKD46 plasmid into Escherichia coli (E. coli) competent cells, utilizing the RED homologous recombination system to guide the targeting fragment to recombine with the guaB gene locus, replacing the original guaB gene by kana gene (Fig. 2).

Fig. 2 The guaB gene was knocked out and replaced by kana gene using the RED homologous recombination

To screen for clones with successful guaB gene knockout, we performed colony PCR on the transformed bacteria using the primers located in the upstream and downstream homologous arms (Fig. 3). If the PCR amplification shows a 1600 bp band (corresponding to the replaced kana gene), it indicates successful knockout; if the result matches the negative control (i.e., a 2000 bp band, corresponding to the length of the guaB gene), the knockout does not occur (Fig. 3).

Fig. 3 Colony PCR to screen for the knockout clones using primers located in the upper and lower homology arms.

After successfully obtaining the knockout clones, we picked up colony-PCR positive strains that were the xanthine-dependent auxotrophic strain, designated as E. coli BW25113-ΔguaB. These bacteria could survive dependently on xanthine, and this could be used as a foundation for subsequent experiments.

Through this round of experiments, we successfully constructed the xanthine-dependent strain BW25113-ΔguaB. This demonstrates that by knocking out the guaB gene, the secondary metabolic pathway was activated, making the strain depend on exogenously supplied xanthine to survive. Next, based on the caffeine metabolism pathways reported in the literature, we plan to further modify the metabolic pathways to shift the strain's dependency from xanthine to 7-MX, aiming to achieve our ultimate goal of constructing a 7-MX biosensor.

Based on our review of previous literature^[1][2][3][4], we selected the NdmDCE protein as the target enzyme for conversion. NdmDCE can effectively convert 7-MX, enabling the strain to respond to the presence of 7-MX. Therefore, we plan to introduce the exogenous NdmDCE metabolic pathway into the previously constructed BW25113-ΔguaB auxotrophic strain, allowing it to detect the presence and concentration of 7-MX through the expression of the NdmDCE protein. To achieve this, we designed and constructed the plasmid pYB1s-ndmDCE (pDCE).

To construct pDCE, we first obtained the sequences of the ndmD, ndmC, and ndmE genes from the NCBI database and made them synthesized by a company. Next, we employed the Golden Gate assembly to subclone these gene fragments into the pYB1s vector (Fig. 4). After completing the plasmid construction, we introduced pDCE into the previously constructed BW25113-ΔguaB competent strain via chemical transformation.

Fig. 4 Plasmid construction of pYB1s-ndmDCE

To verify the successful construction of pDCE, we performed colony PCR on the transformed strains using the primers ndmC-F and ndmE-R. The experimental results showed that the length of the amplified fragment was consistent with the expected size (Fig. 5), indicating that the plasmid was successfully constructed and inserted into the target strain.

Fig. 5 Results of colony PCR screening

Through further validation experiments, we confirmed the successful construction of pDCE (Fig. 6), and the target biosensor was successfully established.

Fig. 6 Plasmid pYB1s-ndmDCE

Through this round of component assembly and genetic engineering, we successfully constructed a biosensor capable of detecting 7-MX. The sensor is sensitive to 7-MX concentration under optimal conditions: the strain exhibits normal growth in the presence of high concentrations of 7-MX, while its growth is inhibited under low concentration or absence 7-MX conditions. This provides a critical foundation for achieving a 7-MX-dependent sensor.

To test the specificity and sensitivity of the biosensor, we need to use a large amount of 7-MX as the substrate. However, the high cost of commercially available 7-MX standards may exceed our budget. Inspired by the 2023 NEFU-China team’s project, we decided to produce 7-MX by ourselves using synthetic biology approaches ^[5][6].

We designed and constructed a plasmid, pYB1s-ndmDtBA (Fig. 7), which features a truncated NdmD to enhance the catalytic efficiency. This plasmid can sequentially demethylate caffeine to ultimately produce 7-MX. Additionally, we constructed a cofactor regeneration plasmid, pSB1c-frmAB-FDH (Fig. 8), which improves the production efficiency of 7-MX by reducing the accumulation of toxic by-products, which also eliminates environmental pollution. Finally, we co-transformed these two plasmids into E. coli BW25113 competent cells, creating a strain capable of producing 7-MX.

Fig. 7 Plasmid construction of pYB1s-ndmDtBA

Fig. 8 Plasmid construction of pSB1c-frmAB-FDH

Construction of the 7-MX Production Pathway

We obtained the ndmA, ndmB, and ndmD genes through gene synthesis and amplified these fragments using PCR. We truncated the ndmD gene by removing the coding region for the N-terminal 266 amino acids and obtained the ndmDt gene fragment. Using Golden Gate technology, we replaced the original ndmD gene with ndmDt, constructing the improved pYB1s-ndmDtBA plasmid.

During the Golden Gate assembly process, the ndmDt gene was driven by the strong promoter pBAD, while the ndmA and ndmB genes were controlled by the J23107 promoter. Ultimately, we constructed the pYB1s-ndmDtBA plasmid (Fig. 9).

Fig. 9 Plasmid pYB1s-ndmDtBA

Construction of the Cofactor Regeneration System

We amplified the frmA, frmB, and FDH genes via PCR and used DpnⅠ digestion to eliminate the template DNA, which reduced false positive clones. Subsequently, we utilized Golden Gate assembly to ligate these fragments into a plasmid, constructing the pSB1c-frmAB-FDH plasmid (Fig. 10).

Fig. 10 Plasmid pSB1c-frmAB-FDH

Finally, we co-transformed the pSB1c-frmAB-FDH and pYB1s-ndmDtBA plasmids into E. coli BW25113 competent cells, successfully obtaining a strain capable of 7-MX production.

First, we activated and induced the obtained 7-MX production strain. Then, following the designed system (Fig. 11), we added the substrate to initiate the whole-cell catalytic reaction.

Fig. 11 Whole Cell Catalytic System Diagram

Due to the low solubility and high density of 7-MX in water, after 25 h of catalysis, we separated the bacterial cells by centrifugation at 4200 rpm for 10 min. The cells were then washed 2-3 times to extract the 7-MX precipitate. The 7-MX precipitate was subsequently dried in an oven at 60°C for 3 h, and finally ground to obtain 16 mg of 7-MX product (Fig. 12).

Fig. 12 Images of a and b (after 7-MX purification)

Through the careful design of the metabolic pathway and optimization of reaction conditions, we successfully enhanced the yield and production efficiency of 7-MX. Moreover, our production process is not only cost-effective but also environmentally friendly, which significantly reduced the generation of harmful by-products, and thereby provided a feasible solution for large-scale 7-MX production.

After successfully constructing the sensor, we needed to validate its functionality to ensure it could be effectively used in screening mutant strains. Since the sensor is based on an auxotrophic strain, its growth is significantly restricted in the absence of 7-MX or xanthine. Therefore, we designed an experimental protocol using M9 medium without xanthine, ensuring that the strain maintains limited growth due to the lack of essential substrates. We then tested the sensor's substrate specificity by adding different substrates in a 96-well plate. The growth curves of the strain were measured using a microplate reader, and the sensor strain's growth (OD₆₀₀) was used to reflect its ability to utilize different substrates, thereby assessing its specificity.

We selected a small quantity of commercially available methylxanthines and xanthine standards, along with the previously prepared 7-MX, as test substrates. According to the literature, most methylxanthines are water-soluble, so we dissolved seven types of methylxanthines and xanthine in water. The induced bacterial cells were washed to remove the xanthine-containing yeast extract from the ZY medium and resuspended in water. We used double-strength yeast ZY medium and LB medium to induce the sensor strain.

Next, we prepared M9 medium without yeast extract as the growth environment for the cells. A mixture of 100 μL of M9 medium, 100 μL of methylxanthine substrate solution, and 10 μL of bacterial suspension was added to a 96-well plate, and the growth curves were measured using a microplate reader.

The figure below shows the changes in strain activity measured by the microplate reader over 36 h (Fig. 13).

Fig. 13 Growth curves of the experimental bacteria in the enzyme labeling apparatus over a period of 36 h

We initially anticipated that only the strains with added 7-MX and xanthine substrates would grow well, while the growth of strains under other substrate conditions would be inhibited. However, all tested strains exhibited some degree of growth, indicating that the sensor has low specificity for the target substrate, which significantly deviates from our expected outcome.

In this round of experiments, we found that the antibiotic selection concentration may have been insufficient, leading to incomplete selection of resistant strains and, consequently, affecting the specificity performance of the sensor. After discussions with our advisor and thorough analysis, we hypothesized that the lack of strict antibiotic selection during transformation, activation, and induction may have resulted in contamination by non-target strains. This not only weakened the growth advantage of the target sensor strain but also prevented it from displaying a significant specificity advantage when cultured in the presence of target substrates (7-MX and xanthine).

To address this issue, we planned to increase the intensity of antibiotic selection in subsequent experiments and optimize the transformation and induction processes to minimize contamination. These improvements would help enhance the specificity performance of the sensor strain. We were confident that with these adjustments, the sensor's sensitivity and specificity would be significantly improved, providing a solid foundation for the subsequent screening of mutant strains.

To enhance the effectiveness of antibiotic selection and ensure the acquisition of more target strains, we increased the antibiotic concentration during the experiment based on the kanamycin resistance gene in the plasmid to improve selection efficiency.

We used the previously constructed 7-MX auxotrophic mutant competent strain BW25113-ΔguaB and transformed it with the NdmDCE plasmid. During subsequent plating, activation, and induction processes, the antibiotic concentrations were adjusted to k₃₀ and s₅₀(Final concentration of 30 and 50 μg/mL), thereby constructing the dual-antibiotic-resistant target strain as the improved sensor strain.

To verify the feasibility of the sensor, we introduced the strain into the microplate reader along with different types of methylxanthines and xanthine, and measured the growth curves over 36 h (Fig. 14).

Fig. 14 Growth curves of the experimental bacteria in the media containing different types of methylxanthines

After improving the selection method, the sensor's response to different types of methylxanthines and xanthine was as expected: the strain exhibited normal growth with the addition of 7-MX, while growth was inhibited with other substrates. However, we observed significant data variation in the results, which we speculate may be due to suboptimal experimental conditions, leading to a wide range of fluctuations. We plan to optimize the experimental protocol in the next round to obtain more accurate data.

Based on previous experiments, we hypothesized that dual antibiotic selection may have resulted in excessively low OD values for the target strain. Additionally, we suspected that the low purity of the previously produced 7-MX may have contributed to significant errors. Therefore, in this round of experiments, we improved the culture medium formulation and purchased new 7-MX standards for testing.

We dissolved seven types of methylxanthines and xanthine in M9 medium. During cell induction, we used double-strength yeast ZY medium and LB medium to ensure better cell quality. When transferring the cells, we washed the induced cells with water to remove the xanthine-containing yeast extract from the ZY medium, followed by washing and resuspending the cells in M9 medium. Finally, 200 μL of substrate solution and 10 μL of bacterial suspension were added to a 96-well plate for detection in the microplate reader.

The experimental results are shown in the figure below (Fig. 15). To more clearly present the outcomes, we selected several key substrates related to caffeine metabolism and plotted a bar chart of the OD values of the strains over 36 h (Fig. 16).

Fig. 15 Growth curves of the experimental bacteria change it according to Fig 14

Fig. 16 OD value of the strain at 36 h

By comparing the results of this round with those from the third round of experiments, we observed a significant reduction in error. The sensor was able to specifically recognize the 7-MX substrate, aligning with the experimental design objectives.

In previous experiments, we validated the specificity of the pYB1s-ndmDCE (ΔguaB) sensor strain for the 7-MX substrate. This round of experiments aimed to further assess the sensor's response to different concentrations of 7-MX to test its sensitivity. We plan to vary the concentration of 7-MX and observe whether the strain's growth correlates proportionally with the substrate concentration, thereby evaluating the sensor's sensitivity.

We used an improved formulation by preparing a 1 mM/L stock solution of 7-MX in M9 medium and diluting it in a gradient to concentrations ranging from 0.1 to 0.5 mM/L. After induction of the sensor strain in double-strength yeast ZY medium and LB medium, the cells were washed and resuspended to remove residual xanthine. Finally, 200 μL of each substrate concentration and 10 μL of bacterial suspension were added to a 96-well plate for detection using a microplate reader.

The experimental results are shown in Fig. 17, illustrating the changes in strain growth activity over 36 h in response to varying 7-MX concentrations.

Fig. 17 Growth curves of experimental bacteria in different concentrations of 7-MX

The experimental results indicated that the strain's growth was proportional to the 7-MX concentration, with the sensor showing a strong response to higher concentrations of 7-MX. Although sensitivity decreased under low-concentration conditions, the overall performance met expectations. We plan to further optimize the sensor design to enhance its sensitivity at lower concentrations.

After screening mutants with the biosensor, we initially identified NdmB^Q289A as the strain with the highest 7-MX production capability. However, considering that the screening was performed in a small-scale 96-well plate system with relatively high errors, we decided to verify its 7-MX production ability in a larger-scale experiment. We started by using a 50 OD bacterial culture for a substrate concentration gradient experiment to explore the maximum 7-MX yield.

Additionally, based on the reports that cofactor regeneration systems (CRS) can enhance the yield of wild-type strains in 7-MX production, we also investigated the functionality of the CRS in this context.

According to a previous study ^[6][10], we determined that the optimal conditions for 7-MX production should be 25°C induction for 18 h and 20°C whole-cell catalysis for 18 h, as these conditions could best maintain enzyme activity. Therefore, our experiments to test the substrate concentration gradient was conducted as follows:

• Use a 50 OD bacterial culture.
• Centrifuge at 4200 rpm and 4°C, discard the supernatant.
• Resuspend the bacterial pellet in 1 mL of fermentation solution (Tris-HCl, pH=9) with substrate concentrations of 1 mM, 2 mM, 4 mM, 8 mM, and 10 mM.
• Perform whole-cell catalysis in a 100 mL shaking flask at 20°C on a shaker.

Resuspend the bacterial pellet in 1 mL of fermentation liquid (Tris-HCl, pH=9) with substrate concentrations of 1 mM, 2 mM, 4 mM, 8 mM, and 10 mM.

Fig. 25 Bar plot of the effect of different substrate concentrations on the production of experimental strain 7-MX at 50 OD without CRS system

Fig. 26 Bar chart of the effects of different substrate concentrations on 7-MX production with experimental strains under 50 OD conditions

The results of our experiments to test the substrate concentration gradient are shown in Figure 25 and Figure 26. Based on these data, the strain without the CRS produced less 7-MX compared to the strain with CRS, demonstrating that the CRS still plays a crucial role in the mutant ndmB strain. Additionally, in the no-CRS group, the 7-MX yield decreased at a 10 mM substrate concentration compared to 8 mM, which may be due to formaldehyde accumulation, which was not efficiently removed by CRS.

Therefore, in subsequent experiments, only strains with CRS were used.

When the substrate concentration was below 4 mM, the 7-MX yield positively correlated with substrate concentrations. However, when the substrate concentration exceeded 4 mM, the 7-MX yield plateaued around the 4 mM level, while the accumulation of the intermediate product PX continued to increase.

It is evident that, with a 50 OD cell density, the maximum yield of 7-MX reaches only around 4 mM and does not increase further with higher substrate concentrations. Meanwhile, caffeine is still nearly completely converted, and the accumulation of PX continues to rise. We cannot rule out the possibility that relatively low cell density is a limiting factor. Therefore, the next step will be to conduct a cell density gradient experiment to investigate whether increasing the cell density can break the bottleneck of 7-MX production.

Based on the results from the first round of optimization, we decided to conduct a cell density gradient experiment using the NdmB^Q289A mutant under the highest substrate concentration (10 mM) to explore the condition of maximum 7-MX production.

Based on existing research and the results from the first round, we decided to induce the bacteria at 25°C for 18 h.

For the whole-cell catalysis, the gradient experiment groups were set as follows: 10 OD, 20 OD, 40 OD, 80 OD, and 100 OD.

All other conditions were consistent: After centrifugation at 4200 rpm and 4°C for 10 min, the supernatant was discarded. The cell pellets were resuspended in 1 mL of fermentation solution (Tris-HCl, pH=9, 10 mM substrate) in a 10 mL centrifuge tube.

Finally, the whole-cell catalysis was conducted in a 100 mL shaking flask at 20°C and 200 rpm for 18 h.

Fig. 27 The bar chart of the effects of different OD value on 7-MX production with experimental strains cultured in medium containing 10 mM caffeine

The experiment of gradient cell biomass yielded the results in Figure 27. Based on the data, with increased cell biomass, especially beyond 50 OD, the yield of 7-MX was significantly improved. During this process, the consumption of the substrate caffeine gradually reached 100%. Under the 100 OD condition, the concentration of the intermediate product PX was approximately 1 mM, and the molar conversion rate of 7-MX was enhanced to about 90%.

As shown by the experimental results, under the conditions of 100 OD cell biomass and 10 mM substrate concentration, the yield of 7-MX reached a peak of 9 mM, representing a significant breakthrough. Caffeine was completely converted, and the accumulation of PX was minimized under optimal conditions. From the results of these two rounds of experiments, it is evident that even at a 50 OD cell biomass, caffeine consistently achieves a relatively high molar conversion rate, while the intermediate product PX continues to accumulate. This likely indicates that the conversion of intermediate PX to 7-MX remains a rate-limiting step even when using NdmB Q298A for 7-MX production. To confirm this hypothesis, we plan to perform time-course sampling during the whole-cell catalysis process.

Based on previous research results and the observed accumulation of intermediate product PX, we hypothesize that the conversion of intermediate PX to the product 7-MX is the primary rate-limiting step. We have decided to perform time-course sampling during the whole-cell catalysis to validate this hypothesis.

Based on existing research and the results from the first two rounds, we have determined that the highest conversion rate can be achieved with a biomass of 100 OD at a substrate concentration of 10 mM. Therefore, we decided to perform time-course sampling under these conditions.

Other conditions remain consistent: induction at 25°C for 18 h. After centrifugation at 4200 rpm and 4°C, the supernatant was discarded, and the bacterial pellet was resuspended in 1 mL fermentation broth (Tris-HCl, pH=9, 10 mM substrate) in a 10 mL centrifuge tube. Finally, whole-cell catalysis was carried out in a 100 mL shaker flask at 20°C and 200 rpm.

During the whole-cell catalysis, samples were collected every hour.

Fig. 28 Time curve of 7-MX production with the experimental strains cultured in medium containing 10 mM caffeine with bacteria in 50 OD

The relationship between catalysis time and the amounts of each substance is shown in Figure 28. During the 18-h whole-cell catalysis, the substrate is nearly completely converted within the first 4 h. The remaining time is for the slow conversion of the intermediate product PX to 7-MX, with the conversion rate decreasing over time.

By plotting the catalytic time curve, we confirmed the previous hypothesis that the conversion efficiency of PX to 7-MX is relatively slow, and the overall conversion rate decreased over time. Nevertheless, under conditions of 100 OD cell density and 10 mM substrate concentration, we achieved the optimal yield of 7-MX, laying the foundation for further practical production applications. Next, we will utilize these optimized conditions to attempt caffeine extraction from spent coffee grounds and produce 7-MX, aiming to recycle food waste and explore more economically viable production strategies.

Before converting caffeine in coffee grounds to 7-MX, we plan to first determine the caffeine content in the coffee grounds and preliminarily estimate the potential yield of 7-MX.

We first collected sufficient coffee grounds from four coffee shops: Luckin, Starbucks, Kudi, and 818, and then dried them in an oven. After that, we performed gradient extraction of caffeine from the coffee grounds obtained from Luckin and Starbucks. We took 5 g of dry coffee grounds and dissolved them in 20 mL, 40 mL, 80 mL, and 100 mL of distilled water. The flasks were then placed in an ultrasonic bath for 10 min to enhance caffeine dissolution. After dissolution, 1 mL of the liquid was transferred to a 1.5 mL EP tube and centrifuged at 15,000 g for 5 min. The supernatant was collected, centrifuged again, and then passed through a filter using a 1 mL syringe. The filtered liquid was transferred to a brown liquid phase bottle. The prepared samples were then subjected to HPLC analysis to determine the caffeine content.

Fig. 29 Different brands of collected coffee grounds

Fig. 30 Bar chart of the concentration of caffeine extracted from Luckin and Starbucks coffee grounds

Our first-round testing results showed that the caffeine content in coffee grounds from both brands ranged between 90 mg and 100 mg per gram, which is significantly different from the theoretical values of 0.5% to 1.5% found in existing literature^[11]. Therefore, we conclude that this test was not successful.

We suspect that the high caffeine content measured by us may be due to the use of ultrasonic dissolution, which could have caused the breakdown of certain substances in the coffee (such as lipids and proteins), leading to inflated measurements. Therefore, we decided to switch to heating and boiling to facilitate caffeine dissolution. Additionally, considering that a low dilution factor might result in insufficient dissolution of caffeine from the coffee grounds, we plan to increase the dilution factor to further enhance the dissolution capability.

Based on the results from the previous round, ultrasonic-assisted caffeine dissolution was ineffective. Therefore, we changed our method and opted for heating and boiling to promote the dissolution of caffeine.

We took 5 g of dry coffee grounds from each store and added them to 100 mL of distilled water. The mixture was heated in a microwave oven to boil 5 times, allowing the caffeine to fully dissolve into the distilled water. After cooling, the solution was filtered using a funnel and filter paper to remove the residue, leaving the filtrate. We then transferred 1 mL of the filtrate into a 1.5 mL EP tube, centrifuged it at 15,000 g for 5 min, and collected the supernatant with a 1 mL syringe. The supernatant was filtered twice through membranes and then transferred to a brown liquid chromatography bottle. The sample was prepared for subsequent analysis by HPLC to determine the caffeine content.

Fig. 31 The overall process of extracting caffeine from coffee grounds

Fig. 32 Bar chart of the concentration of caffeine extracted from 4 different brands

With the improved method, the caffeine content extracted from the coffee grounds of the four brands ranged from 5 mg/g to 10 mg/g, corresponding to 0.5% to 1% of the dry weight of the coffee grounds. This is consistent with the previously reported values of 0.5% to 1.5%. Therefore, we have successfully determined the caffeine content in the coffee grounds from the four brands, and our testing results are accurate.

We have recognized the drawbacks of ultrasonic extraction. In future experiments, we will first consider whether the method might interfere with the substances being tested before deciding to use it.

Attempt to Use Coffee Grounds as a Low-Cost Raw Material: We plan to explore the use of coffee grounds as a raw material for producing 7-MX, as this approach reduces the costs and offers a new method for recycling coffee grounds. By utilizing our optimized strains, we aim to produce 7-MX from the caffeine in coffee grounds while evaluating both the environmental and economic benefits. We will use the optimized strains to perform whole-cell catalysis with conversion solutions derived from coffee grounds of various brands and plot the corresponding time curves.

We plan to use the optimized NdmB^Q289A strain for the production of 7-MX. We will first prepare conversion solutions by extracting caffeine. For each brand of coffee grounds, 2 g will be weighed and dissolved in 5 mL of Tris buffer. The mixture will be heated to boiling at 100°C in a water bath for 10 min to ensure complete dissolution of caffeine. After cooling, the solution will be centrifuged at 4000 rpm for 10 min at 4°C. The supernatant will be transferred to a 2 mL EP tube and centrifuged again at 15000 g for 10 min. The supernatant will be used as the conversion solution.

The induced NdmB^Q289A strain will be used for whole-cell catalysis. We will use 100 OD of the induced strain, resuspend it in 1 mL of the conversion solution, and place it in a 100 mL Erlenmeyer flask. Whole-cell catalysis will be conducted under the optimized conditions for 18 h. Starting from 0 h, a 50 μL aliquots will be taken every 2 h, diluted with 50 μL of Tris buffer, and centrifuged at 15000 g for 3 min. The supernatant will be sampled and analyzed by HPLC to determine the 7-MX content and calculate the conversion rate.

Fig. 33 Procedures before 18 h fermentation

Fig. 34 Mix the NdmB^Q289A strain with transformation liquid containing caffeine

Fig. 35 The fermentation curves with 100 OD NdmB^Q289A over 18 h

The results show that at 0 h, only caffeine was present, with no 7-MX detected. As the fermentation progressed, the caffeine content gradually decreased, while the levels of 7-MX and PX increased until the caffeine was depleted. We conclude that caffeine was successfully converted to the target product, 7-MX, indicating that our test was successful. However, compared to fermentation with pure caffeine, the conversion rate of 7-MX was not ideal. We speculate that some substances in the coffee grounds may have interfered with the enzyme's activity, reducing the conversion rate of 7-MX.

As shown by the experimental results， we can successfully utilize caffeine from spent coffee grounds for the efficient production of 7-MX, making a significant advancement in the project. The experiment not only showed successful conversion of caffeine into 7-MX but also highlighted the robust conversion capability of our optimized NdmB^Q289A strain in practical applications. Although the conversion rate of 7-MX from spent coffee grounds was slightly lower compared to pure caffeine, this result confirmed that we have achieved the goal of producing high-value compounds from waste materials.