Chapter 12
Artificial Intelligence and Quantum Computing for the Design of Novel Kisspeptin-10 Analogues and Their Experimental Evaluation in Cervical Cancer Cells
Deisy Yurley Rodríguez-Sarmiento
Faculty of Health Sciences, Universidad Autnoma de Bucaramanga, Bucaramanga, 680003, Colombia
Abstract
Cervical cancer remains a leading cause of morbidity and mortality in women, underscoring the need for innovative therapies. Kisspeptin-10 (KP10), a decapeptide that activates the G-protein-coupled receptor (GPCR) Kiss1R, shows antitumour potential. Using artificial intelligence (AI) based modelling, a focused library of KP10 analogues was designed; 10 top candidates were synthesised and evaluated. Molecular docking predicted enhanced Kiss1R affinity, which was confirmed in vitro by MTT and wound-healing assays in cervical cancer cells. Several analogues showed superior cytotoxicity and migration inhibition versus KP10. These results highlight AI-assisted peptide design as an efficient approach for developing GPCR-targeted therapeutics in cervical cancer.
Keywords: Kisspeptin-10, Cervical cancer, GPCR, Artificial intelligence, Peptide design, Antitumour activity
Introduction
Cervical cancer remains a major global health concern and one of the leading causes of cancer-related death among women, particularly in low- and middle-income countries. Despite progress in screening, vaccination, and treatment, many cases are diagnosed at advanced stages with poor outcomes, highlighting the need for new targeted therapies1,2.
The kisspeptin system has recently gained attention beyond its reproductive role3. Kisspeptins, derived from the KISS1 gene, include the potent fragment Kisspeptin-10 (KP10), which binds the G-protein-coupled receptor (GPCR) Kiss1R4. Initially identified for its antimetastatic properties in melanoma and breast cancer5, kisspeptin signalling is now linked to key oncogenic processes such as proliferation, migration, angiogenesis, and apoptosis6. These findings support the exploration of kisspeptin analogues as therapeutic modulators, particularly in hormone-responsive and reproductive cancers like cervical cancer.
Artificial intelligence (AI) and machine learning have transformed peptide-based drug discovery by enabling rational design, predictive modelling, and efficient screening of large peptide libraries7. These tools accelerate the identification of bioactive sequences with favourable pharmacological profiles and guide experimental validation8.
This study applies AI-driven prediction and molecular docking to design and evaluate KP10 analogues with potential antitumour activity. Ten top-scoring peptides were synthesised and tested for cytotoxic and anti-migratory effects in cervical cancer cells, establishing a framework for AI guided GPCR targeted peptide discovery in oncology.
Results and Discussion
AI-based prediction of KP10 analogues
KP10 analogues were rationally designed through an AI-driven workflow to identify variants with improved antitumour potential. Starting from the native sequence (YNWNSFGLRF), over 5000 analogues were generated via Python-based systematic mutations and extensions (1012 amino acids). Each sequence was encoded with 1477 physicochemical descriptors from Propy 1.09, and filtered by charge, hydrophobicity, molecular weight, and complexity, yielding 70 biologically plausible candidates. These were evaluated using three machine-learning classifiers: SVM, Random Forest, and ANN, trained on balanced anticancer peptide datasets. Consensus predictions identified the most robust candidates for subsequent structural modelling and biological testing.
In silico prediction of antitumour potential
The antitumour potential of 70 KP10 analogues was assessed using the MLACP 2.0 server10, which applies physicochemical descriptors and machine learning algorithms (LightGBM, Extra Trees) trained on validated anticancer peptide datasets. Peptides with high-confidence anticancer predictions were prioritised. This two-tiered strategy (AI classifiers optimised for HeLa cytotoxicity followed by external ACP validation) strengthened candidate selection for molecular docking and experimental evaluation. See Table 1.
Table 1: List of designed analogues, including MLACP 2.0 prediction scores and consensus annotations.
|
Peptide |
Sequence |
Length (aa) |
Net Charge |
MW (Da) |
MCLAP Score |
|
KP10 |
YNWNSFGLRF |
10 |
+2.0 |
1302.45 |
0.833 |
|
KP10-01 |
YNWNTFGLRF |
10 |
+2.0 |
1316.48 |
0.848 |
|
KP10-02 |
YNWNSFSLRF |
10 |
+2.0 |
1332.48 |
0.893 |
|
KP10-03 |
YNWNSFGIRF |
10 |
+2.0 |
1302.48 |
0.887 |
|
KP10-04 |
YNFWNSFGLRF |
11 |
+2.0 |
1449.63 |
0.880 |
|
KP10-05 |
YDWNSFGLRF |
10 |
+1.0 |
1303.44 |
0.877 |
|
KP10-06 |
YNWNSWGLRF |
10 |
+2.0 |
1341.49 |
0.843 |
|
KP10-07 |
YNFNSFGLRF |
10 |
+2.0 |
1263.42 |
0.836 |
|
KP10-08 |
YNWNSFGLRW |
10 |
+2.0 |
1341.49 |
0.828 |
|
KP10-09 |
YNWNSFGQLRF |
11 |
+2.0 |
1430.58 |
0.819 |
|
KP10-10 |
YNWNSFGVRF |
10 |
+2.0 |
1288.43 |
0.815 |
Molecular docking simulations and binding affinity estimations
Molecular docking simulations were conducted using HPEPDOCK 2.011 to assess receptor binding of KP10 analogues. All peptides showed negative docking scores, indicating favourable interactions. Native KP10 exhibited a score of 298.2, while most analogues ranged from 257.1 to 287.7; KP10-01 and KP10-06 showed comparable affinities. Binding free energies (ΔG) and dissociation constants (Kd) estimated with PRODIGY12 supported these results, with KP10-05, KP10-06, KP10-09, and KP10-10 displaying lower ΔG values (9.3 to 8.8 kcal/mol) and submicromolar Kd (≈10⁻⁷ M) versus KP10 (ΔG = 7.5 kcal/mol; Kd = 4.810⁻⁶ M) (See Table 2). These findings indicate that specific sequence modifications improve peptide receptor interactions and guided analogue prioritisation for synthesis and biological testing.
Table 2: Predicted binding affinity of KP10 analogues to the receptor target based on molecular docking simulations.
|
Peptide |
|
Docking Score |
|
∆G (kcal/mol) |
|
Kd (M) |
|
KP10 |
|
-298.2 |
|
-7.5 |
|
4.8x10-6 |
|
KP10-01 |
|
-287.7 |
|
-6.2 |
|
4.2x10-5 |
|
KP10-02 |
|
-275.1 |
|
-6.8 |
|
1.6x10-5 |
|
KP10-03 |
|
-265.9 |
|
-7.8 |
|
3.1x10-6 |
|
KP10-04 |
|
-257.1 |
|
-8.3 |
|
1.3x10-6 |
|
KP10-05 |
|
-264.6 |
|
-9.3 |
|
2.9x10-7 |
|
KP10-06 |
|
-286.3 |
|
-9.3 |
|
2.8x10-7 |
|
KP10-07 |
|
-273.1 |
|
-5.8 |
|
8.7x10-5 |
|
KP10-08 |
|
-257.6 |
|
-6.2 |
|
4.1x10-5 |
|
KP10-09 |
|
-267.2 |
|
-8.8 |
|
6.4x10-7 |
|
KP10-10 |
|
-282.1 |
|
-8.8 |
|
6.4x10-7 |
Chemical synthesis of selected peptides
The 10 selected KP10 analogues were synthesised by solid-phase peptide synthesis (SPPS) using standard Fmoc chemistry on Rink amide resin13. Coupling was performed with HBTU/HOBt and DIPEA, and peptides were cleaved from the resin with a TFA-based cocktail. Crude products were purified on Sep-Pak C18 cartridges and confirmed by electrospray ionisation mass spectrometry (ESI-MS). Purified peptides were lyophilised and stored at 20 C until biological testing.
Cytotoxicity and migration assays in cervical cancer cells
The antiproliferative activity of KP10 and its 10 analogues were evaluated in HeLa cells by MTT assay14 after 48 hours of treatment at 10500 nM. Most peptides induced a dose-dependent decrease in cell viability. KP10-09 and KP10-10 showed the strongest effects, reducing viability to 1.6% and 7.7% at 500 nM, compared with 74.1% for KP10. KP10-05 also showed high cytotoxicity (23.7% viability) (Figure 1). Some analogues displayed non-linear profiles, suggesting compensatory cellular mechanisms. These results demonstrate that sequence modifications can enhance KP10 cytotoxicity against cervical cancer cells. Calculated IC₅₀ values were 1618 nM for KP10, 0.2466 nM for KP10-09, and 0.5364 nM for KP10-10, confirming their markedly greater potency relative to the native peptide.
To further assess functional activity, KP10-09 and KP10-10 were selected for migration assays based on their strong cytotoxic effects. A wound healing assay at 100 nM evaluated their influence on HeLa cell migration. The mean open wound area was quantified at 0, 24, 48, and 72 hours for untreated controls, KP10, and the analogues (Figure 2). At 24 hours, KP10-09 and KP10-10 maintained larger wound areas (63.37% and 64.01%) than the control (40.29%) and KP10 (55.26%), a trend persisting through 72 hours (18.71% and 14.62% vs. 1.12% for control). Both analogues significantly delayed wound closure compared with native KP10, demonstrating sustained inhibition of cell migration and supporting their potential to limit metastatic progression in cervical cancer.
|
Figure 1: Cell viability of HeLa cells treated with KP10 and its analogues. HeLa cells were treated with increasing concentrations (10, 100, 250, and 500 nM) of KP10 and 10 synthesised analogues for 24 hours. Cell viability was measured using the MTT assay and is expressed as a percentage relative to untreated control cells (set at 100%). Data represent the mean standard deviation of three independent experiments. A dose-dependent reduction in viability was observed for most analogues, with KP10-09 and KP10-10 showing the strongest cytotoxic effects at 500 nM. |
|
Figure 2: Inhibitory effects of KP10, KP10-09, and KP10-10 on HeLa cell migration assessed by wound healing assay. Cells were treated with 100 nM of each peptide, and the mean percentage of open wound area was measured at 0, 24, 48, and 72 hours for untreated controls, native KP10, and its analogues. Bars represent the average wound area (%) remaining over time, indicating delayed wound closure in cells exposed to the analogues. |
Conclusion
This study demonstrates the value of integrating AI-driven peptide design, molecular modelling, and experimental validation to identify KP10 analogues with enhanced antitumour activity. From the computational library, 10 peptides were synthesised and tested in vitro; KP10-09 and KP10-10 showed the strongest cytotoxicity and migration inhibition in HeLa cells, consistent with in silico predictions. These results highlight the efficiency of computational pipelines for early peptide discovery and support the potential of rationally engineered kisspeptin analogues as leads for cervical cancer therapy.
References
Arbyn, Marc, Elisabete Weiderpass, Laia Bruni, Silvia de Sanjos, Mona Saraiya, Jacques Ferlay, and Freddie Bray. Estimates of Incidence and Mortality of Cervical Cancer in 2018: A Worldwide Analysis. The Lancet Global Health 8, no. 2 (2020): e191203. https://doi.org/10.1016/S2214-109X(19)30482-6
Phan, Le Thi, Hyun Woo Park, Thejkiran Pitti, Thirumurthy Madhavan, Young-Jun Jeon, and Balachandran Manavalan. MLACP 2.0: An Updated Machine Learning Tool for Anticancer Peptide Prediction. Computational and Structural Biotechnology Journal 20 (2022): 447380. https://doi.org/10.1016/j.csbj.2022.07.043
Xue, Li C., Joo Pglm Rodrigues, Panagiotis L. Kastritis, Alexandre Mjj Bonvin, and Anna Vangone. PRODIGY: A Web Server for Predicting the Binding Affinity of Proteinprotein Complexes, Bioinformatics 32, no. 23 (2016): 36768. https://doi.org/10.1093/bioinformatics/btw514
Cao, Dong-Sheng, Qing-Song Xu, and Yi-Zeng Liang. Propy: A Tool to Generate Various Modes of Chous PseAAC. Bioinformatics 29, no. 7 (2013): 9602. https://doi.org/10.1093/ bioinformatics/btt072
Jabeen, Saima, Muhammad Zahid Qureshi, Zeeshan Javed, Muhammad Javed Iqbal, Muhammad Ismail, and Ammad Ahmad Farooqi. Kisspeptin Mediated Signaling in Cancer. Current Topics in Medicinal Chemistry 16, no. 22 (2016): 24716. https://doi.org/10.2174/ 1568026616666160212123309
Isaac, Kathy Sharon, Michelle Combe, Greg Potter, and Stanislav Sokolenko. Machine Learning Tools for Peptide Bioactivity Evaluation Implications for Cell Culture Media Optimization and the Broader Cultivated Meat Industry. Current Research in Food Science 9 (2024): 100842. https://doi.org/10.1016/j.crfs.2024.100842
Kotani, Masato, Michel Detheux, Ann Vandenbogaerde, David Communi, Jean-Marie Vanderwinden, Emmanuel Le Poul, Stphanie Brzillon, et al. The Metastasis Suppressor Gene KiSS-1 Encodes Kisspeptins, the Natural Ligands of the Orphan G Protein-coupled Receptor GPR54. Journal of Biological Chemistry 276, no. 37 (2001): 346316. https://doi.org/10.1074/ jbc.M104847200
Lee, Jeong-Hying, Mary E. Miele, Deana J. Hicks, Karen K. Phillips, Jeffery M. Trent, Bernard E. Weissman, and Danny R. Welch. KiSS-1, a Novel Human Malignant Melanoma Metastasis-suppressor Gene. Journal of the National Cancer Institute 88, no. 23 (1996): 17317. https://doi.org/10.1093/jnci/88.23.1731
Liang, Chun-Chi, Ann Y. Park, and Jun-Lin Guan. In Vitro Scratch Assay: A Convenient and Inexpensive Method for Analysis of Cell Migration in Vitro. Nature Protocols 2 (2007): 32933. https://doi.org/10.1038/nprot.2007.30
Zhou, Pei, Bowen Jin, Hao Li, and Sheng-You Huang. HPEPDOCK: A Web Server for Blind Peptide-protein Docking Based on a Hierarchical Algorithm. Nucleic Acids Research 46, no. W1, no. W1 (2018): W44350. https://doi.org/10.1093/nar/gky357
Popa, Simina M., Donald K. Clifton, and Robert A. Steiner. The Role of Kisspeptins and GPR54 in the Neuroendocrine Regulation of Reproduction. Annual Review of Physiology 70 (2008): 21338. https://doi.org/10.1146/annurev.physiol.70.113006.100540
Merrifield, R. B. Solid Phase Peptide Synthesis. I. The Synthesis of a Tetrapeptide. Journal of the American Chemical Society 85, no. 14 (1963): 214954. https://doi.org/10.1021/ja00897a025
Zhai, Silong, Tiantao Liu, Shaolong Lin, Dan Li, Huanxiang Liu, Xiaojun Yao, and Tingjun Hou. Artificial Intelligence in Peptide-based Drug Design. Drug Discovery Today 30, no. 2 (2025): 104300. https://doi.org/10.1016/j.drudis.2025.104300
Mosmann, Tim. Rapid Colorimetric Assay for Cellular Growth and Survival: Application to Proliferation and Cytotoxicity Assays. Journal of Immunological Methods 65, no. 12 (1983): 5563. https://doi.org/10.1016/0022-1759(83)90303-4
World Health Organization: WHO. Cervical Cancer, December 2, 2025. https://www.who.int/news-room/fact-sheets/detail/cervical-cancer