Our project aims to enhance skincare formulations through the utilization of isoenzymes derived from papain, a cysteine protease renowned for its gentle exfoliating and moisturizing properties. Papain functions by breaking down keratin proteins in the stratum corneum, thereby facilitating the removal of dead skin cells and promoting skin regeneration. Compared to conventional acid-based peels, papain offers a far gentler and more biocompatible alternative for exfoliation.

To develop superior enzymatic skincare agents, our research focused on identifying papain isoenzymes that exhibit optimal activity across diverse skin types and pH conditions. Using NCBI BLAST, genes with 50%-80% sequence similarity to papain were selected as candidate isoenzymes. Structural modeling of these candidates was then performed via AlphaFold to evaluate and prioritize enzymes with favorable active site characteristics. Ultimately, five cysteine proteases were chosen for their promising suitability in skincare applications: those derived from Abrus precatorius, Actinidia eriantha, Cicer arietinum, Cajanus cajan, and papain itself. These enzymes are expected to act synergistically to enhance the efficacy, stability, and safety of enzymatic skincare products.

Figure 1. Overview of Experimental Steps

The DNA sequence was amplified using PCR to produce adequate quantities for downstream cloning in E.coli DH5α later on. The E.coli DH5α containing the recombinant plasmids are also amplified. Figure 2 shows the agarose gel electrophoresis results of PCR for five clones. We can find out that the DNA fragment of Papain is approximately 678 bp in length, Abrus precatorius is 1083 bp, Actinidia eriantha is 1089 bp, Cajanus cajan is 1083 bp and Cicer arietinum is 1095 bp, which is consistent with the predicted size. This indicates that the target gene was successfully amplified.

Figure 2. Agarose gel electrophoresis results of PCR for five clones

In this study, double digestion was employed to cut pET28a plasmid vector with two restriction enzymes NdeI and HindIII, producing specific overhangs for the insertion of the target gene. The gene fragment was designed with approximately 15-20 bp of homologous sequences at both ends, matching the terminal regions of the linearized plasmid. These homologous overlaps facilitated precise insertion through Exnase-mediated homologous recombination. The Exnase enzyme catalyzed the recombination between the plasmid ends and the corresponding homologous sequences of the gene, resulting in seamless and directional integration without the need for ligase.

After homologous recombination, the recombinant plasmids were transformed into chemically competent E. coli DH5α cells via the heat shock method. Following overnight incubation on LB agar plates containing the appropriate antibiotic, several well-isolated colonies were observed (Figure 3), indicating successful transformation.

Figure 3. Agar plate showing DH5α colonies after transformation

To verify the presence of the target gene, five individual colonies were randomly selected for colony PCR. As shown in the gel electrophoresis results (Figure 4.), each colony produced a DNA band of the expected size, confirming successful plasmid construction and correct insertion of the target gene.

Figure 4. Agarose gel electrophoresis results of Colony PCR identification for five clones

To confirm the accuracy of plasmid construction, recombinant pET28a plasmids from five positive clones were sent to a commercial sequencing facility for Sanger sequencing. Sequence alignment (Figure 5.) showed that the inserted gene fragments matched the designed sequences exactly, with no frameshift mutations or base substitutions, indicating successful and accurate cloning of the target genes.

Figure 5. DNA sequencing results of Recombinant plasmid

Each of the five recombinant plasmids encoding papain, Abrus, Actinidia, Cajanus, and Cicer proteases was independently transformed into E. coli BL21(DE3) competent cells using the heat shock method. The competent cells were first incubated with plasmid DNA on ice and then exposed to a brief heat shock at 42 °C to facilitate DNA uptake. Following recovery in LB medium, the cells were plated on kanamycin-selective LB agar plates and incubated overnight at 37 °C.

As shown in Figure 6, well-isolated single colonies appeared for each transformation, indicating successful uptake of the recombinant plasmids. One colony from each transformation was randomly selected for colony PCR. The agarose gel electrophoresis results confirmed the presence of the target genes, verifying successful transformation and plasmid integrity in all five constructs.

Figure 6. Agar plate showing BL21 colonies after transformation

Each PCR-verified colony was inoculated into liquid LB medium and cultured overnight, followed by scale-up fermentation. After cell harvest via centrifugation, bacterial pellets were subjected to ultrasonic lysis to release intracellular contents. Crude protein was extracted from the lysates without further purification at this stage.

Protein concentration was determined for each lysate using the Bradford assay. As shown in Figure 7, the total protein concentrations varied slightly among strains but were all within an acceptable range for downstream purification and activity assays.

Figure 7. Concentration of five lysate protein

To verify the expression of recombinant proteins and evaluate their approximate molecular weights, SDS-PAGE analysis was performed on both crude and purified samples. Proteins were separated on a 15% polyacrylamide gel and stained with Coomassie Brilliant Blue.

As shown in Figure 8, distinct protein bands were observed in all five lysate samples. A ~24 kDa band was detected in the papain-expressing strain, consistent with the expected molecular weight of papain. For Abrus, Actinidia, Cajanus, and Cicer proteases, bands were observed around ~39 kDa, matching their predicted sizes. These results confirmed that all five recombinant enzymes were successfully expressed in E. coli.

Figure 8. SDS-PAGE Analysis Before and After Protein Elution

To preliminarily assess proteolytic activity, the recombinant enzyme papain was tested for its ability to dissolve acid-induced casein precipitates in milk.

As shown in Figure 9, fresh milk containing 6% (w/v) casein appeared as a uniform liquid (Figure 9A). Upon mixing with dilute hydrochloric acid, a visible white precipitate formed, representing denatured casein aggregates (Figure 9B). After treatment with papain for 1 hour at 37 °C, the precipitate significantly reduced, and the solution became clearer (Figure 9C), suggesting successful hydrolysis of the protein clot.

This observation demonstrates that papain possesses caseinolytic activity and can cleave aggregated casein chains. The experiment provides qualitative evidence of protease activity but does not differentiate between the relative activity levels of different enzymes.

Figure 9. Results of casein hydrolysis

To evaluate whether the recombinant enzymes possess proteolytic activity toward physiologically relevant substrates, fibronectin-1 (FN1) was selected for hydrolysis assays.

As shown in Figure 10, the FN1 protein appeared as a single intact band at approximately 47 kDa in the blank group, indicating structural stability in the absence of active enzymes. In contrast, all five enzyme-treated samples (Papain, Abrus, Actinidia, Cajanus, and Cicer) displayed a significant reduction in the 47 kDa FN1 band intensity. Simultaneously, distinct new bands appeared around 24 kDa, suggesting enzymatic cleavage of FN1 into smaller peptide fragments.

The consistent presence of ~24 kDa fragments across all treatment groups implies that the enzymes—despite their different sources—may hydrolyze FN1 at similar cleavage sites. This result confirms that all five recombinant proteases exhibit biological activity against FN1, supporting their potential for applications involving ECM protein remodeling.

Figure 10. Results of FN1 proteolytic digestion

To quantitatively evaluate the enzymatic activity of the crude lysates, an ELISA assay was performed. Protein concentrations were first normalized using NanoDrop, and all lysates were diluted to different concentrations.

Figure 11. The standard table of papain concentration

A standard curve was generated based on serial dilutions of the ELISA kit standard (Figure 12), yielding a linear regression model for activity estimation.

Figure 12. The standard curve of papain concentration

We used ELISA kits to individually test and analyze the activity of several enzymes. Based on the above standard curves, we calculated the activity of each enzyme and performed significance analysis using Graphpad to calculate the P-value. As shown in Figureure 13, the enzyme activity was Papain > Act > Abr > Caj > Cic.

Figure 13. Single enzyme activity detection data

Single-enzyme activity results (Figure 14) showed that Papain had the highest protease activity among all five enzymes, consistent with the qualitative results from the casein assay. No statistically significant differences were observed between Cajanus and Cicer.

Figure 14. Graphs of single-enzyme activity comparison

We used an ELISA kit to analyze the activity of the dual-enzyme combinations and verify whether there was a synergistic effect. Based on the standard curve, we calculated the activity of the dual-enzyme combinations and performed a significance analysis using Graphpad to calculate the P-value. As shown in Figure 15, the optimal combination was Caj+Cic.

Figure 15. Dual-enzyme activity detection data

In the dual-enzyme combinations (Figure 16), the mixture of Cajanus and Cicer displayed the highest activity. Other combinations, such as papain with Cajanus or Cicer, and Abrus with Actinidia, showed moderate activity but no significant enhancement compared to single enzymes.

Figure 16. Graphs of dual-enzyme combination comparison

We used an ELISA kit to analyze the activity of the Three-enzyme combination and verify whether there was a synergistic effect. Based on the standard curve, we calculated the activity of the Three-enzyme combination and performed a significance analysis using Graphpad to calculate the P value. As shown in Figure 17, the optimal combinations were Papain + Abrus + Actinidia and Abrus + Actinidia + Cajanus.

Figure 17. Three-enzyme activity detection data

Three-enzyme combinations (Figure 18) revealed that the groups Papain + Abrus + Actinidia and Abrus + Actinidia + Cajanus produced the strongest activity levels, although their differences were not statistically significant. These combinations outperformed most others in the same group.

Figure 18. Graphs of Three-enzyme combinations comparison

We used an ELISA kit to analyze the activity of the four-enzyme combination and verify whether there was a synergistic effect. Based on the standard curve, we calculated the activity of the four-enzyme combination and performed a significance analysis using Graphpad to calculate the P value. As shown in Figure 19, the optimal combination was Abrus + Actinidia + Cajanus + Cicer.

Figure 19. Four-enzyme activity detection data

In the four-enzyme combinations (Figure 20), the mixture of Abrus + Actinidia + Cajanus + Cicer showed the highest value, indicating the strongest activity among all tested combinations of four enzymes.

Figure 20. Graphs of four-enzyme combinations comparison

We used an ELISA kit to analyze the activity of the five-enzyme combination and verify whether there was a synergistic effect. Based on the standard curve, we calculated the activity of the five-enzyme combination and performed a significance analysis using Graphpad to calculate the P value, as shown in Figure 21.Subsequently, we compared the combinations with the strongest enzyme activity under different combinations. The results showed that the dual enzyme combination had the strongest effect.

Figure 21. Different combinations activity detection data

As shown in Figure 22, the dual combination of Cajanus + Cicer displayed the highest overall activity, even exceeding that of the full five-enzyme mixture. This suggests that specific pairings may exhibit synergistic effects stronger than broader mixtures.

Figure 22. Graphs of optimal combinations comparison

To explore structural similarities between papain and other potential isoenzymes, a NCBI BLAST search was performed using the amino acid sequence of Carica papaya papain as the query. As shown in Figure 23, we identified 30 homologous sequences with more than 50% sequence identity and less than 80%.

Figure 23. Results of NCBI BLAST analysis

The amino acid sequences of these homologous enzymes were successfully downloaded (Figure 24) and subsequently subjected to three-dimensional structure prediction using AlphaFold2. The predicted structures provide a solid foundation for downstream molecular docking simulations, enabling a structure-based analysis of binding behavior and functional conservation among papain-like proteases.

Figure 24. Results of sequence alignment analysis

To evaluate binding potential, we conducted molecular docking simulations between the predicted structures and the synthetic tripeptide substrate Pglu-Phe-Leu-pNA, commonly used in protease activity assays.

As shown in Figure 25, Papain exhibited the strongest binding affinity, as indicated by the lowest docking energy score. In addition, isoenzymes from Actinidia eriantha, Abrus precatorius, Cicer arietinum, and Cajanus cajan also showed relatively strong binding scores, suggesting their ability to form stable enzyme-substrate complexes. These findings imply that the candidate enzymes may possess similar substrate recognition and catalytic properties as papain.

Figure 25. Indication of the binding affinity between five enzymes and the substrate

Detailed analysis of the docked complexes revealed that all five high-affinity enzymes formed dense hydrogen bonding networks with the substrate (Figure 26). Notably, papain formed both hydrogen bonds and additional π–π stacking interactions, a unique feature not observed in the other isoenzymes.

These π–π interactions are likely to contribute to enhanced binding stability, potentially explaining why papain displayed the strongest affinity in docking simulations. While the other isoenzymes also formed stable complexes, their lack of π–π interactions and subtle differences in binding pocket geometry likely account for the variation in binding strength.

This binding affinity pattern is consistent with the enzyme activity results obtained from ELISA assays, further validating the structural-functional correlation. Together, the docking and binding affinity analyses strongly support the conclusion that microbially synthesized papain retains functional features of plant-derived papain, laying the foundation for future application-oriented studies.

Figure 26. Results of molecular interaction analysis between five enzymes and the substrate

These results not only confirm the proteolytic activity and substrate-binding potential of the recombinant papain and its isoenzymes, but also provide key structural and functional data as raw material for further studies on multi-enzyme synergy. For example, whether the enzymes act simultaneously or sequentially during substrate degradation remains unknown and could be clarified through substrate competition assays or reaction time-course studies.

Furthermore, the demonstrated activity of microbially synthesized papain lays the foundation for its potential large-scale replacement of plant-derived papain, particularly through alternative microbial hosts such as Bacillus subtilis or yeast, which may offer higher yields and better post-translational modifications than E. coli.

To ensure applicability in biomedical or cosmetic fields, toxicity, allergenicity, and safety assessments of the recombinant enzymes are also warranted. Collectively, the current findings provide experimental and computational groundwork for optimizing enzyme formulation, improving production strategies, and evaluating biosafety for future industrial applications.