Skin-penetrating peptides derived from computational simulation improve transdermal absorption and facilitate topical treatment of melanoma

We are sharing a study led by Gang Wei's team at Fudan University, titled "Skin-penetrating peptides derived from computational simulation improve transdermal absorption and facilitate topical treatment of melanoma."​ This work aims to address a critical bottleneck in transdermal drug delivery—the insufficient skin permeability of drugs. The research team first systematically screened cell-penetrating peptides (CPPs) from various sources and identified penetratin as the "seed" peptide with optimal transdermal efficiency and safety. Utilizing AlphaFold2 computational simulation and structural analysis, they revealed for the first time that the positively charged surface patch area (PSPA) is the core physicochemical parameter determining the skin permeability of peptides. Based on this finding, the team established a quantitative predictive model and rationally redesigned the amino acid sequence of penetratin accordingly, successfully obtaining a derivative peptide, 589WP, with significantly optimized performance. Finally, the researchers covalently linked the anticancer drug floxuridine (FUdR) to 589WP, formulated it into a gel, and validated its outstanding local therapeutic efficacy and high safety in a melanoma mouse model. This work not only developed an efficient transdermal delivery tool but also established a new paradigm of "computational simulation-guided rational peptide design."

01 Research Background

Transdermal drug delivery systems offer significant advantages such as avoiding hepatic first-pass metabolism and improving patient compliance. However, the stratum corneum, the outermost layer of the skin, constitutes a formidable permeability barrier that limits the delivery of most drugs. As novel penetration enhancers, peptides (such as cell-penetrating peptides, CPPs) have attracted considerable attention due to their biodegradability and good biocompatibility. However, current peptide-based transdermal delivery strategies face two core challenges: first, a lack of systematic understanding of peptide structure-permeability relationships and reliable predictive models, with design largely reliant on trial-and-error methods; second, the delivery efficiency of existing strategies still has significant room for improvement. Although previous studies have attempted to elucidate the transdermal mechanisms of peptides (e.g., interaction with keratin, increasing lipid fluidity), due to the vast differences in physicochemical properties among different peptides, universal governing principles have not been fully revealed. Computational simulation technologies (e.g., AlphaFold2) provide powerful tools for analyzing peptide structures at the atomic level and predicting their properties, but their application in guiding the rational design of transdermal peptides remains at the cutting edge. This study aims to integrate experimental screening and computational simulation to systematically clarify the key factors affecting peptide transdermal capability, and based on this, design novel skin-penetrating peptides with excellent performance, ultimately for application in the local treatment of melanoma.

02 Innovative Highlights

First definitive identification of "Positively Charged Surface Patch Area" as the core determinant of peptide transdermal permeability.

Through systematic parallel comparison of the transdermal performance of 20 known CPPs, combined with AlphaFold2-predicted structures and calculation of multiple physicochemical parameters, the research moved beyond traditional analysis of simple parameters (e.g., net charge, helicity). Correlation analysis revealed for the first time that PSPA showed a significant positive correlation with the transdermal permeability of different categories of CPPs (cationic, amphipathic, hydrophobic), while the correlation of other parameters (e.g., hydrogen bond donor area, net charge) was either not universal or not statistically significant, thereby precisely pinpointing the key influencing factor.

Establishment of the first two-factor quantitative predictive model for peptide transdermal permeability based on surface properties.

After discovering that the single-factor PSPA prediction had deviations (e.g., failure in predicting 359R-penetratin), the study introduced hydrophobic surface patch area (HSPA) as a synergistic variable and proposed an "Amphiphilicity Score" model. This model (SOA = PSPA + 1/3 HSPA) successfully fitted the transdermal data of penetratin and its derivatives, with prediction correlation coefficients exceeding 0.85 for both rat and pig skin, achieving a leap from qualitative experience to quantitative prediction and providing a reliable tool for rational design.

Acquisition of a derivative peptide with significantly enhanced performance through "computational simulation-guided site-directed mutagenesis."

Based on the SOA model, the research team performed systematic virtual screening of single-point, double-point, and triple-point mutations on the penetratin sequence, rationally selecting the optimal combination from over 1,500 candidate sequences. The final derivative peptide, 589WP (sequence: RQIKWWFWWRRMKWKK), simultaneously increased both PSPA and HSPA by replacing specific hydrophobic residues with tryptophan and methionine, without introducing additional positively charged amino acids. Its permeability coefficients for rat and pig skin in vitro were increased by 41% and 62%, respectively, compared to wild-type penetratin.

Construction of a "prodrug-peptide" conjugate gel enabling highly efficient and low-toxicity local chemotherapy.

589WP was covalently linked to the clinical antimetabolite drug floxuridine (FUdR) via an esterase-sensitive ester bond, constructing the FUdR-589WP prodrug. This prodrug can specifically release the active drug under the action of esterases highly expressed at tumor sites. Formulated as an anhydrous gel for topical administration, in a melanoma mouse model, the FUdR-589WP gel at only 1/5 of the clinical dose (0.5%) showed significantly superior antitumor efficacy compared to a 5% high-dose free FUdR gel, and caused no skin irritation, achieving a dual breakthrough in efficacy and safety.

03 Results and Discussion

3.1 Screening of Efficient Transdermal Peptides and Safety Assessment

Among 20 FAM-labeled CPPs, penetratin and transportan demonstrated the best transdermal permeability, with apparent permeability coefficients approximately 56 times that of free FAM. However, transportan exhibited cytotoxicity towards human immortalized keratinocytes (HaCaT), while penetratin showed no toxicity even after 72 hours of incubation. Considering the cost and synthesis advantages brought by its shorter peptide chain (16 vs. 27 amino acids), penetratin was selected as the template for subsequent optimization.

3.2 Identification of Key Physicochemical Parameters and Establishment of Predictive Model

By calculating multiple parameters of the 20 peptides and performing correlation analysis with transdermal permeability, PSPA was found to be the most universally and significantly correlated single factor. However, the simple PSPA model failed in predicting the triple-arginine mutant (359R-penetratin), whose predicted high PSPA did not translate to high permeability; instead, its structure became loose. To address this, the researchers synthesized four tryptophan-mutated derivatives, introduced HSPA as a synergistic variable, and established the SOA two-factor model. This model perfectly fitted all known data points and possessed high statistical significance.

3.3 Rational Design and Performance Validation of the Optimized Peptide 589WP

Based on the SOA model, the optimal peptide 589WP was obtained through virtual screening and experimental validation. Compared to wild-type penetratin, the Papp values of 589WP for rat and pig skin increased from 1.75 and 1.11 (×10⁻⁶ cm/s) to 2.47 and 1.80 (×10⁻⁶ cm/s), respectively. In vivo fluorescence imaging further confirmed that strong fluorescent signals were still detectable in mouse skin and body 8 hours after treatment with Cy5-589WP gel, whereas signals in the free Cy5 group had largely disappeared, visually demonstrating the superior penetration-promoting ability of 589WP. Structural analysis showed that the mutations optimized the distribution of surface patches without disrupting the overall folding.

3.4 Construction, Release, and In Vitro Activity of the FUdR-589WP Prodrug

The FUdR-589WP conjugate was successfully synthesized. Release experiments showed that FUdR release exhibited dual pH- and esterase-dependence: slow release in the weakly acidic skin microenvironment, but significantly accelerated in the weakly alkaline, esterase-rich tumor-simulating environment. Although covalent linkage slightly reduced the in vitro cytotoxicity of FUdR (IC50 increased from ~5 µM to ~15 µM), Calcein-AM/PI live/dead cell staining showed that at 15 µM, the FUdR-589WP treatment group could still kill approximately 50% of B16F10 melanoma cells, indicating a significant effect.

3.5 Outstanding In Vivo Antitumor Efficacy and Safety

In a B16F10 cell xenograft mouse model, topical application of 0.5% FUdR-589WP gel demonstrated significantly superior tumor inhibition compared to 2.5% high-dose free FUdR gel, and also outperformed a 0.5% FUdR + 589WP physical mixture gel. This indicates that covalent linkage is crucial for achieving efficient synergistic drug delivery. Histopathological analysis showed that FUdR-589WP gel treatment caused no toxicity to major organs. Skin irritation assessment (IL-1β immunohistochemical staining) indicated that 0.5% FUdR-589WP gel, similar to the blank gel, did not induce significant inflammation, whereas 2.5% free FUdR gel elicited notable immunopositive reactions, demonstrating the great advantage of the peptide-based prodrug strategy in reducing local irritation.

04 Conclusion and Future Perspectives

This study successfully established a complete workflow from mechanism elucidation, model development, rational design to application validation for developing high-performance skin-penetrating peptides. Its core achievements include: 1) discovering PSPA as the key structural parameter determining peptide transdermal permeability; 2) establishing a quantitative SOA-based predictive model, providing a powerful tool for rational peptide design; 3) obtaining the derivative peptide 589WP with significantly enhanced transdermal performance; 4) constructing a 589WP-based prodrug-gel system that achieved highly efficient, low-toxicity local chemotherapy in a melanoma model, demonstrating tremendous clinical translation potential.

The profound significance of this work lies in the deep integration of artificial intelligence structure prediction (AlphaFold2) with drug delivery research, advancing the development of transdermal peptides from an "experience-driven" to a "computation-driven" new stage. The established SOA model is not only applicable to penetratin derivatives; the relationship it reveals between "surface patch area" and permeability provides a new perspective for understanding the transmembrane mechanisms across other biological barriers (e.g., blood-brain barrier, intestinal mucosa). In the future, this research could be further explored in the following directions: 1) Extending the SOA model to peptide libraries with more structural types to validate its universality; 2) Delving deeper into the specific cellular and molecular pathways by which 589WP promotes transdermal permeation (e.g., interactions with skin lipids and proteins); 3) Promoting the FUdR-589WP gel towards clinical research and exploring this platform for delivering other drugs (e.g., macromolecular drugs, nucleic acids) to treat a broader range of skin diseases. In summary, this study provides an innovative solution with both scientific and practical value for revolutionizing transdermal drug delivery systems and enabling local, precise treatment of skin diseases.