Tổng hợp và xác định hoạt tính chống oxy hóa của các tryptophyllin l peptide bằng phương pháp mô hình thống kê định lượng 3dqsar kết hợp với thực nghiệm in vitro báo cáo tổng kết đề tài ngh

BỘ CÔNG THƯƠNG ĐẠI HỌC CÔNG NGHIỆP THÀNH PHỐ HỒ CHÍ MINH BÁO CÁO TỔNG KẾT ĐỀ TÀI KHOA HỌC KẾT QUẢ THỰC HIỆN ĐỀ TÀI NGHIÊN CỨU KHOA HỌC CẤP TRƯỜNG Tên đề tài Tổng hợp xác định hoạt tính chống oxy hóa Tryptophyllin L peptide phương pháp mơ hình thống kê định lượng 3DQSAR kết hợp với thực nghiệm in vitro Mã số đề tài: 21/1H03 Chủ nhiệm đề tài: TS TRẦN THỊ THANH NHÃ Đơn vị thực hiện: KHOA CƠNG NGHỆ HĨA HỌC TP HỒ CHÍ MINH, 11.2021 i Tp Hồ Chí Minh, … MỤC LỤC MỤC LỤC II DANH MỤC HÌNH ẢNH III DANH MỤC BẢNG V DANH MỤC PHỤ LỤC VI LỜI CÁM ƠN PHẦN I THÔNG TIN CHUNG CHƯƠNG : TỔNG QUAN 1.1 Tổng quan vấn đề nghiên cứu 1.2 Hiện trạng cơng trình nghiên cứu liên quan đến đề tài 10 1.2.1 Tình hình nghiên cứu quốc tế 10 1.2.2 Tình nghiên cứu nước: 11 1.2.3 Đánh giá kết cơng trình nghiên cứu cơng bố 11 1.3 Mơ hình mối quan hệ định lượng cấu trúc hoạt tính hợp chất (quantitative structure-activity relationship: QSAR) 12 1.3.1 Phương pháp So Sánh Trường Phân Tử (Comparative Molecular Field Analysis, CoMFA) 12 1.3.2 Phương pháp So Sánh Chỉ Số Tương Tự Phân Tử (Comparative Molecular Similarity Indices Analysis, CoMSIA) 15 1.4 Tổng hợp hữu chất mang pha rắn sử dụng kỹ thuật Fmoc 15 1.4.1 Chất (resin) nhóm liên kết (linker) 17 1.4.2 Các nhóm bảo vệ 18 1.4.3 Tác nhân hoạt hóa 20 CHƯƠNG : VẬT LIỆU VÀ PHƯƠNG PHÁP 22 2.1 Phương pháp tính tốn 22 2.1.1 Nguồn liệu, tối ưu hóa cấu trúc xếp chồng phân tử 22 2.1.2 Mơ hình Trường lực mơ hình Gaussian 3D-QSAR 23 2.1.3 Thẩm định mơ hình 24 2.2 Thực nghiệm 25 2.2.1 Vật liệu 25 2.2.2 Phương pháp thực nghiệm 25 CHƯƠNG : KẾT QUẢ VÀ THẢO LUẬN 27 3.1 Mơ hình Trường lực mơ hình Gaussian 27 3.2 Xác định cấu trúc độ tinh khiết peptide tổng hợp 31 3.2.1 Peptide Pro-Trp-Tyr (P-W-Y) 32 3.2.2 Peptide Pro-Trp-Tyr(NH2) (P-W-Y(NH2)) 34 3.2.3 Peptide Pro -Tyr-Trp (P -Y-W) 36 3.2.4 Peptide Pro -Tyr-Trp(NH2) (P -Y-W(NH2)) 38 3.2.5 Peptide Leu-Pro-Trp-Tyr(NH2) (L-P-W-Y (NH2): tryptophyllin L 4.1) 40 3.3 Hoạt động bắt gốc tự ABTS+ 42 3.4 Phản ứng khử sắt 45 CHƯƠNG : KẾT LUẬN VÀ KIẾN NGHỊ 49 ii DANH MỤC HÌNH ẢNH Hình Minh họa trường tương tác phân tử điểm nút mạng lưới 12 Hình Quy trình xây dựng mơ hình trường phân tử 3D-QSAR CoMFA 14 Hình Quy trình tổng hợp peptide pha rắn 16 Hình Các liên kết linker thường sử dụng 18 Hình Hệ bảo vệ Fmoc/tBu 19 Hình Sự phân tách nhóm Boc 19 Hình Sự phân tách nhóm Fmoc 19 Hình Phản ứng nhóm amine acid carboxylic hoạt hóa 20 Hình Xếp chồng a) 108 peptide từ liệu TEAC b) 16 tryptophyllin L hai tripeptide thiết kế 23 Hình 10 Đồ thị tương quan hoạt tính từ thực nghiệm dự đốn tập huấn luyện tập kiểm tra rút từ mơ hình a), b) Trường lực c), d) Gaussian 28 Hình 11 Bản đồ biên dạng (contour) năm trường bắt nguồn từ mơ hình Trường lực (a) trường khơng gian (dương: xanh lục, âm: vàng); (b) Trường tương tác tĩnh điện (xanh lam: dương; đỏ: âm) 29 Hình 12 Bản đồ đường biên dạng năm trường bắt nguồn từ mơ hình Gaussian: (a) trường khơng gian (dương: xanh lục, âm: vàng); (b) trường tương tác tĩnh điện (xanh lam: dương; đỏ: âm); (c) trường kỵ nước (cam: dương, trắng: âm), (d) trường chất nhận liên kết hydro (xanh lục: dương; hồng tươi: âm); (e) trường cho liên kết hydro (tím: dương; xanh lam: âm) 30 Hình 13 Cấu trúc peptide P-W-Y 32 Hình 14 Phổ khối dương [M+H]+ peptide P-W-Y 33 Hình 15 Phổ HPLC peptide P-W-Y 33 Hình 16 Phổ khối dương +ESI MS/MS peptide P-W-Y 34 Hình 17 Cấu trúc peptide P-W-Y(NH2) 34 Hình 18 Khối phổ dương [M+H]+ peptide P-W-Y(NH2) 35 Hình 19 Phổ HPLC peptide P-W-Y(NH2) 36 Hình 20 Phổ khối dương +ESI MS/MS peptide P-W-Y(NH2) 36 Hình 21 Cấu trúc peptide P -Y-W 36 Hình 22 Phổ khối dương [M+H]+ peptide P-Y-W 37 Hình 23 Phổ HPLC peptide P-Y -W 37 Hình 24 Phổ khối dương +ESI MS/MS peptide P- Y-W 38 Hình 25 Cấu trúc peptide P -Y-W(NH2) 38 Hình 26 Phổ khối dương [M+H]+ peptide P-Y-W(NH2) 39 Hình 27 Phổ HPLC peptide P-Y-W(NH2) 39 Hình 28 Phổ khối dương +ESI MS/MS peptide P- Y-W(NH2) 40 Hình 29 Cấu trúc peptide L-P-W-Y (NH2) (tryptophyllin L 4.1) 40 Hình 30 Khối phổ dương [M+H]+ peptide L-P-W-Y(NH2) 41 Hình 31 Phổ HPLC peptide L-P-W-Y(NH2) 41 Hình 32 Phổ khối dương +ESI MS/MS peptide L-P-W-Y(NH2) 42 iii Hình 33 Giá trị bắt gốc ABTS+ tương đương trolox peptide nồng độ 8,2 M thời gian phút 42 Hình 34 Tốc độ phản ứng bắt cation gốc 2,2’-azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS˙ +) peptide nồng độ (a) 2,5 M (b) 10 M 44 Hình 35 TEAC peptide glutathione khoảng nồng độ từ 2,5 đến 10 M 45 Hình 36 Hoạt tính khử sắt tryptophyllin L peptide Trolox nồng độ 25 M 30 phút 46 Hình 37 Sự thay đổi khả khử sắt peptide theo thời gian ba nồng độ khác a) 25, b) 50 c) 100 M 48 iv DANH MỤC BẢNG Bảng Năm peptide Tryptophyllin L dẫn xuất tổng hợp nghiên cứu Bảng Các Tryptophyllin L peptide phân lập từ loài ếch Litoria rubella [16, 18] Bảng Dữ liệu tập huấn mơ hình thử nghiệm mơ hình 22 Bảng Các thơng số thống kê mơ hình Trường lực (4 biến số chính) mơ hình Gaussian (5 biến số chính) 27 Bảng Đóng góp trường vào mơ hình Trường lực (4 biến số chính) mơ hình Gaussian (5 biến số chính) 28 Bảng Các giá trị TEAC dự đốn từ mơ hình Trường lực mơ hình Gaussian 31 Bảng Bảng phân tích phổ khối HPLC peptide tổng hợp 31 v DANH MỤC PHỤ LỤC Phụ lục Thiết bị thủy tinh dùng tổng hợp peptide 54 Phụ lục Điều kiện thời gian lưu HPLC sản phẩm tổng hợp PWY 55 Phụ lục Điều kiện thời gian lưu HPLC sản phẩm tổng hợp PWY(NH2) 56 Phụ lục Điều kiện thời gian lưu HPLC sản phẩm tổng hợp PYW 57 Phụ lục Điều kiện thời gian lưu HPLC sản phẩm tổng hợp PYW(NH2) 58 Phụ lục Điều kiện thời gian lưu HPLC sản phẩm tổng hợp LPWY (NH2) 59 vi DANH MỤC VIẾT TẮT 3D-QSAR: three dimensional quantitative structure activity relationship- mơ hình chiều định lượng quan hệ cấu trúc hoạt tính sinh học ABTS: 2,2'-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid Boc: t-butoxycarbonyl Bzl: benzyl BHA: butylhydroxyanisole BHT: butylated hydroxytoluene COMFA: Comparative molecular field analysis COMSIA: Comparative Molecular Similarity Indices Analysis CNN: convolutional neural networks- mạng thần kinh nơ-ron tích tụ DCM: dichloromethane DMF: dimethylformamide DPPH: 2,2-diphenyl-1-picryl-hydrazyl-hydrate FRAP: ferric reducing antioxidant power- khả chống oxi hóa thể qua khả khử sắt Fmoc: fluorenylmethyloxycarbonyl HATU: 1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate, Hexafluorophosphate Azabenzotriazole Tetramethyl Uronium HBTU: 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate, Hexafluorophosphate Benzotriazole Tetramethyl Uronium HPLC: high performance liquid chromatography- sắc ký lỏng hiệu cao IC50: half maximal inhibitory concentration- nồng độ ức chế tối đa 50% trình sinh học MLR: multiple linear regression- hồi quy tuyến tính đa biến PLS: partial least square- bình phương tối thiểu phần PyBroP RF: random forest- rừng ngẫu nhiên SPPS: Solid phase peptide synthesis- tổng hợp peptide pha rắn SVM: support vector machine- máy vectơ hỗ trợ tBu: tert-bytyl TEAC: trolox equivalent antioxidant capacity TFA: trifluoroacetic acid TPTZ: 2,4,6-tripyridyl-s-triazine vii LỜI CÁM ƠN Chúng xin chân thành cảm ơn Quỹ nghiên cứu khoa học IUH, Ban lãnh đạo IUH, Lãnh đạo khoa Cơng nghệ Hóa học, Phịng thí nghiệm Khoa Cơng nghệ Hóa học, thành viên đề tài giúp tơi hồn thành đề tài nghiên cứu khoa học Cảm ơn thành viên nhóm nghiên cứu: TS Trần Thị Diệu Thuần, TS Trần Đình Phiên thuộc Trung tâm nhiệt đới Việt Nga GS TS John Hamilton Bowie thuộc Trường Đại học Adelaide, đóng góp ý kiến ủng hộ tinh thần để tơi để hồn thành cơng trình nghiên cứu PHẦN I THƠNG TIN CHUNG I Thơng tin tổng quát 1.1 Tên đề tài: Tổng hợp xác định hoạt tính chống oxy hóa Tryptophyllin L peptide phương pháp mơ hình thống kê định lượng 3D-QSAR kết hợp với thực nghiệm in vitro 1.2 Mã số: 21/1H03 1.3 Danh sách chủ trì, thành viên tham gia thực đề tài Họ tên TT Đơn vị cơng tác (học hàm, học vị) Vai trị thực đề tài TS Trần Thị Thanh Nhã Khoa Cơng nghệ hóa học Chủ nhiệm TS Trần Thị Diệu Thuần Khoa Cơng nghệ hóa học Thành viên 1.4 Đơn vị chủ trì: Khoa Cơng nghệ Hóa học, Đại học Công nghiệp Tp.HCM 1.5 Thời gian thực hiện: 1.5.1 Theo hợp đồng: từ tháng 03 năm 2021 đến tháng 02 năm 2022 1.5.2 Gia hạn (nếu có): Khơng 1.5.3 Thực thực tế: từ tháng 03 năm 2020 đến tháng 11 năm 2020 1.6 Những thay đổi so với thuyết minh ban đầu (nếu có): 1.7 Tổng kinh phí phê duyệt đề tài: 55 triệu đồng II Kết nghiên cứu Đặt vấn đề Vai trò peptide ngày trở nên quan trong ngành cơng nghiệp nghiên cứu hóa dược, hóa mỹ phẩm gần cơng nghệ thực phẩm [1-4] Các cơng trình nghiên cứu cho thấy, peptide phân lập từ tự nhiên thể đa dạng hoạt tính sinh học bao gồm khả chống oxy hóa, chống tăng huyết áp, kháng vi sinh…[5, 6] Peptide chống oxy hóa đặc biệt ý chúng góp phần quan trọng vào việc ngăn ngừa chữa trị bệnh tim mạch, tiểu đường, ung thư viêm khớp [7-9] Việc khảo sát hoạt tính chống oxy hóa peptide tách chiết từ tự nhiên với việc nghiên cứu ảnh hưởng yếu tố cấu trúc đến khả chống oxy hóa chúng góp phần mở rộng ngân hàng peptide thuộc nhóm này, đồng thời làm sở cho việc dự đoán thiết kế peptide ứng dụng ngành công nghệ thực phẩm dược phẩm [10, 11] Chính mục đích mà nghiên cứu này, nhóm tiến hành sàng lọc tất peptide tryptophyllin L phân lập từ lồi ếch Litoria rubella sử dụng mơ hình ba chiều định lượng mối quan hệ hoạt tính cấu trúc (3D-QSAR: three dimensional quantitative structure activity relationship) dựa phương pháp CoMFA (Comparative Molecular Field Analysis) CoMSIA (Comparative Molecular Similarity Indices Analysis) Các peptide với dự đốn có giá trị bắt gốc tương đương trolox (trolox equivalent antioxidant capacity: TEAC) lớn tổng hợp pha rắn phương pháp Fmoc xác định hoạt tính chống oxy hóa thực nghiệm in vitro Mục tiêu 2.1 Mục tiêu tổng quát: Sử dụng phương pháp thống kê định lượng 3D-QSAR nhằm dự đoán sàng lọc peptide Tryptophyllin L dẫn xuất có hoạt tính chống oxi hóa cao Tổng hợp peptide sử dụng phương pháp tổng hợp pha rắn thử nghiệm đối chiếu hoạt tính sử dụng thí nghiệm bắt gốc tự Từ đó, đề xuất đặc điểm cấu trúc ảnh hưởng tích đến hoạt tính chống oxi hóa peptide này, đồng thời đề xuất cấu trúc tiến hành nghiên cứu ứng dụng xa công nghiệp thực phẩm dược phẩm 2.2 Mục tiêu cụ thể:  Xây dựng mơ hình 3D-QSAR CoMSIA hoạt tính chống oxi hóa hệ data tripeptide Sử dụng mơ hình để dự đốn phân tích mối liên hệ cấu trúc hoạt tính chống oxi hóa Tryptophyllin L tripeptide dẫn xuất  Tổng hợp pha rắn Tryptophyllin peptide dẫn xuất sử dụng kỹ thuật tổng hợp Fmoc  Tinh chế petide phương pháp sắc ký lỏng pha đảo, xác minh cấu trúc chúng phổ khối (MS) nuclear magnetic resornance (NMR) Received: 17 August 2021 Revised: October 2021 Accepted: 18 October 2021 DOI: 10.1002/psc.3380 RESEARCH ARTICLE Virtual screening and rational design of antioxidant peptides based on tryptophyllin L structures isolated from the Litoria rubella frog Thi Thanh Nha Tran1 | Dinh Phien Tran2 | Thi Minh Anh Nguyen1 | Thai Hoang Tran1 | Nu Ngọc Anh Phan1 | Van Cuong Nguyen1 | Van Trong Nguyen1 | John H Bowie3 Faculty of Chemical Engineering, Industrial University of Ho Chi Minh City, 12 Nguyen Van Bao, Ho Chi Minh City, 700000, Vietnam Department of Chemistry and Environment, Vietnam-Russia Tropical Centre, 63 Nguyen Van Huyen, Nghia Do, Cau Giay, Ha Noi, 11307, Vietnam Faculty of Science, The University of Adelaide, Adelaide, South Australia, 5000, Australia Correspondence Thi Thanh Nha Tran, Faculty of Chemical Engineering, Industrial University of Ho Chi Minh City, 12 Nguyen Van Bao, Ho Chi Minh City 700000, Vietnam Email: tranthithanhnha@iuh.edu.vn Funding information Industrial University of Ho Chi Minh City 64/HĐ-ĐHCN; Code No 21/1H03 Discovery of natural antioxidants has been carried out for decades relying mainly on experimental approaches that are commonly associated with time and cost demanding biochemical assays The maturation of quantitative structure activity relationship (QSAR) modelling has provided an alternative approach for searching and designing antioxidant compounds with alleviated costs As a contribution to this approach, this work aimed to establish a fragment-based 3D-QSAR procedure to discover and design potential antioxidants based on tryptophyllin L structures isolated from the red tree frog Litoria rubella A force field and a Gaussian 3D-QSAR model were built to screen for potential antioxidants from tripeptide fragments covering all sequences of tryptophyllin L database Among those, PWY(NH2) corresponding tryptophyllin L 4.1 was predicted to have the highest 2,20 -azino-bis(3-ethylbenzothiazoline6-sulphonic acid) radical cation (ABTS+) scavenging capability Two newly designed peptides PYW and PYW(NH2) together with PWY(NH2), tryptophyllin L 4.1, and the reference peptide PWY were synthesized and subjected to two antioxidant assays including ABTS scavenging and ferric reducing antioxidant power assays Although the experimental TEAC values of the five peptides were roughly similar to those from predictions, the activity order was not in agreement with the predictions The dissimilarities were accounted by the difference in the experimental procedures, the deviation of modelling regression, and the synergetic effect of structural and experimental features The ABTS radical scavenging assays revealed that all the tested peptides were strong ABTS+ scavengers with the antioxidant capabilities approximately twice as high as trolox and higher than glutathione The ferric reducing activities of the peptides were, on the other hand, much weaker than that of trolox suggesting different antioxidant mechanisms inserted by trolox and the peptides This work was a demonstration that 3D-QSAR methods can be employed in conjunction with experimental methods to effectively detect and design antioxidant peptides KEYWORDS ABTS radical scavenging, antioxidants, force field, Gaussian 3D-QSAR, virtual screening J Pep Sci 2021;e3380 https://doi.org/10.1002/psc.3380 wileyonlinelibrary.com/journal/psc © 2021 European Peptide Society and John Wiley & Sons, Ltd of 13 of 13 | TRAN ET AL I N T RO DU CT I O N linear regression (MLR), partial least square (PLS), random forest (RF), support vector machine (SVM), and recently convolutional neu- Reactive oxygen species (ROS) has been known to play dual roles in ral network (CNN) have been used to generate predictive patterns, biological systems At normal physiological concentrations, they are from which characteristics beneficial for antioxidant activity of pep- important regulators for various processes including cell cycle tides were extrapolated.28–32 The features chosen for modelling progression,1 intracellular signalling transduction,2 and gene expres- also varied from geometric, physicochemical, electronic to compara- sion or suppression In contrast, the excessive generation of ROS in tive molecular field analysis (CoMFA), and comparative molecular cells leads to oxidative stress, causing damages to biomolecules such similarity index analysis (CoMSIA) descriptors,24,28–35 from descrip- as proteins, lipids, and DNA, resulting in aging and a myriad of tors representing each peptide as a vector of individual amino pathologies.4 acids28,30,36 to descriptors representing the features of the whole Living aerobic organisms have evolved highly complex antioxidant peptide molecule.29,32 Most of the predictive models for antioxi- systems, which work synergistically to protect the cells and organs of dants were constructed with the first goal being to learn about the body against free radical damage.5 There are different ways to which independent variables or features affect antioxidant activities categorize antioxidants in these systems One of the most popular of the peptides in the studied databases, and the second goal was ways is to divide them to enzymatic and nonenzymatic antioxidants to employ the constructed model to predict and finally design The endogenous enzymes of glutathione peroxidase, catalase and effective antioxidant superoxide dismutase (SOD), and nonenzymatic antioxidants such as achieved as demonstrated by the QSAR statistics in these publica- glutathione, thioredoxin, ascorbate, tocopherol, flavonoids, and carot- tions, whereas the application of the models to search and design enoids work together to remove or interrupt free radical chains, new peptides was limited Most of them was used only once to maintaining the redox balance for the cells.5,6 However, this balance design several peptides, which have not appeared to be investi- may be disrupted under extreme environmental and pathological gated further in vitro and in vivo conditions, leading to the requirement for exogenous antioxidants to 7–10 peptides The former was undoubtedly On the other hand, although most peptides isolated from natural As a result, a sources have generally been tested with certain biochemical assays, a large number of studies has been conducted to discover new antioxi- number, if not a large number, of these peptides were not found to be dants providing a platform for further development of functional positive with the target assays As mentioned above, it would be too foods, health products, and food additives costly to continue carrying out more “blindfolded” tests and thus they complement the body's natural defence system The search for antioxidant peptides has shown an increasing remained to be explored in the database There should be an approach trend in recent years due to their ability to display very few side to make more use of the constructed QSAR models and available effects in humans.11,12 This trend has been growing even stronger databases of naturally occurring peptides, and as a contribution to this since reports on adverse effects of commercialized antioxidants such approach, the current study aimed to use 3D-QSAR models as a as butylhydroxyanisole (BHA) and butylated hydroxytoluene (BHT) screening tool to detect antioxidant peptides from tryptophyllin L on human enzyme systems and DNA have made their use family, a peptide group isolated from the Litoria rubella frog inhabiting restricted.13,14 In the last decades, the databases of discovered anti- over a large area of Australia.37 oxidant peptides have grown substantially with their major sources 15,16 being cereals, 22,23 able plants, marine protein by-products, 17,18 19–21 meat, edit- and recently some of them were isolated from frog's dorsal skins.24–26 In a recent study, we have investigated antioxidant potential of five major tryptophyllin L peptides.24 Two tryptophyllin L including FPWL(NH2) and SPWL showed comparable activities with those of glutathione in the three assays 2,20 -azino-bis(3-ethylbenzothiazoline- The standard procedure for discovering antioxidants peptides 6-sulphonic acid) (ABTS+), 1,1-diphenyl-2-picrylhydrazyl (DPPH.) normally started with protein hydrolysis by enzymes or microorgan- radical scavenging, and ferric reducing, suggesting that they may be a isms to obtain hydrolysates, which were then separated, purified, and part of the skin antioxidant system of the frog We were therefore identified using high performance liquid chromatography (HPLC) encouraged to search for more potential antioxidants in the whole coupled with mass spectrometry (MS) The isolated peptides were peptide family In the present study, we planned to explore potential finally subjected to a number of antioxidant assays to examine their antioxidant peptides from tryptophyllin L family using a fragment- antioxidant ability.11,12 Although such procedure undoubtedly con- based approach In particular, a force field and a Gaussian 3D-QSAR tributed significantly to expanding databases of antioxidant peptides, model were constructed and employed to screen all tripeptide frag- a major drawback of this routine is its “trial-and-error” characteristic ments derived from tryptophyllin L peptides (Table 1) The peptides as there is no certainty that this whole time-and-cost-demanding pro- corresponding to the fragments of highest predicted antioxidant cess would result in the discovery of any significant antioxidant pep- values, except for the five investigated in the previous paper, were tides.27 synthesized and subjected to the ABTS radical scavenging and ferric In recent years, the fields of bioinformatics and chemometrics reducing activity assays Based on the structure of the most potent have progressed rapidly, continuously providing powerful tools for tryptophyllin L peptide and feature analysis of the field-based examining the structure–activity relationships of biomolecules models, some peptides were rationally designed and tested further Among which, a variety of modelling algorithms including multiple in vitro of 13 TRAN ET AL software,42 with the template being PWY, the compound with highest MATERIALS AND METHODS | ABTS+ scavenging ability The peptides were aligned employing 2.1 | Computational method SMART and Maximum common substructure function jointly, followed by manual adjustment to maximize the overlap of all the 2.1.1 | Data source, structural optimization, and alignment peptide structures All tryptophyllin L fragments and newly designed tripeptides were also superimposed on the same template molecule following a similar procedure and presented in Figure 1B The trolox-equivalent-antioxidant-capability (TEAC) dataset of 108 tripeptides was collected from the literature40 (Table S1), 75% of which was assigned to the training set and the rest was used as the 2.1.2 | Force field and Gaussian 3D-QSAR models test set for modelling Optimization of the peptide structures was then carried out using semiempirical PM7 method from the Molecular Construction of force field and Gaussian 3D-QSAR models were Orbital Package (MOPAC) quantum chemistry program41 interfaced implemented based on CoMFA and CoMSIA43,44 methods, respec- with ChemDraw Professional 15.1 software tively, with a specific set of parameters as follows The Lennard–Jones The superimposition of the TEAC dataset was illustrated in steric potentials and charges for the electrostatic fields were calcu- Figure 1A The alignment was implemented using Maestro 11.5 lated from the OPLS_2005 force field Atom types and hydrophobic parameters were from Ghose et al.,45 whereas hydrogen bond acceptor and donor fields are based on Phase pharmacophore feature defi- T A B L E Sequences of tryptophyllins L isolated from the L rubella frog38,39 nitions, with projected points All the models were constructed using Peptide Sequence MW 1.1 Pro-Trp-Leu (NH2) 414 Maestro 11.5 software.42 To build force field models, each aligned molecule in the dataset was placed in a 3D cubic lattice with a grid spacing of 1.0 Å The size 1.1.1 Pro-Trp-Leu (OH) 415 of the cubic lattice was set to extend beyond the training set limits 1.2 Phe-Pro-Trp-Leu (NH2) 560 3.0 Å The values of steric and electrostatic fields were obtained by 1.2.1 Phe-Pro-Trp-Leu (OH) 561 calculating the interactions between the molecule and the carbon pro- 1.3 pGlu-Phe-Pro-Trp-Leu (NH2) 671 bes sitting at the intersections of the lattice The calculated fields sub- 1.4 Phe-Pro-Phe-Pro-Trp-Leu (NH2) 804 sequently served as independent variables in a partial least square 1.5 Ser-Pro-Trp-Leu (OH) 501 (PLS) fitting procedure46 to produce force field models All energy 2.1 Ile-Pro-Trp-Leu (NH2) 526 values were truncated at a cutoff 30 kcal/mol, and all variables with 3.1 Phe-Pro-Trp-Pro (NH2) 544 standard deviation less than 0.01, absolute t value smaller than 2, or 3.2 Phe-Pro-Trp-Pro (OH) 545 3.3 pGlu-Phe-Pro-Trp-Phe (NH2) 705 4.1 Leu-Pro-Trp-Tyr (NH2) 576 4.2 Phe-Leu-Pro-Trp-Tyr (NH2) 723 5.1 pGlu-Ile-Pro-Trp-Phe-His-Arg (NH2) 964 Note: pE: pyroglutamic acid (NH2): amidation at the C-terminal residue F I G U R E Superimposition of (A) 108 peptides from TEAC dataset and (B) 16 tryptophyllin L and two newly design tripeptides within 2.0 Å of any training set atom were eliminated For the Gaussian models, the molecular similarity indices derived from Steric and Electrostatic Alignment (SEAL) similarity fields47 served as independent variables These indices reflect indirectly the field similarity of molecules in a dataset by comparing the similarity of each molecule with a common probe atom (having a radius of Å, charge of +1 and hydrophobicity of +1) Five fields involved in the of 13 TRAN ET AL estimation of Gaussian based descriptors included steric, electrostatic, of diisopropylcarbodiimide/hydroxybenzotriazole to form the active hydrophobic, and hydrogen bonding properties All parameters were ester and coupling times of to h The peptides were cleaved from set as those for the force field models mentioned above the resin at the end of the synthesis by treatment with 95% trifluoroacetic acid/2.5% triisopropylsilane/2.5% water (5 ml) for h The peptides were precipitated from the cleavage reaction by the 2.1.3 | Model validation addition of 40 ml of cold diethyl ether The peptides were isolated by centrifugation, dissolved in 30% acetonitrile/water, and lyophilised For each method, 10 models were built with an increasing number of The purified peptides were >95% pure as confirmed by HPLC and PLS factors from to 10 The statistical parameters derived from each MS data model were examined and compared with each other in order to select the most reliable and robust models statistically These parame- ABTS radical scavenging assay ters included coefficient of determination (R2), cross-validation coeffi- The ABTS radical scavenging activity assay was performed to measure cient (R2CV) computed from predictions obtained by a leave-one-out the ability of a sample to scavenge ABTS radical cation (ABTS˙+), approach (LOO), correlation coefficient of external validation (Q2ext), monitored by spectrophotometric change of the ABTS˙+ solution in F value, root-mean-square error in the test set predictions (RMSE), the presence of sample.48 The radical was generated by mixing 7-mM and coefficient of determination obtained from randomization test in ABTS stock solution with 2.45-mM potassium persulfate, which was which the Gaussian descriptors were kept as they are but the activi- kept in the dark at room temperature for 12 to 16 h before use The ties of peptides were scrambled (R2scramble) While higher R2, R2CV, Q2 absorbance of the ABTS˙+ solution was adjusted to 0.70 ± 0.02 at ext, and F values correspond to the higher reliability and predictability 734 nm using 5-mM phosphate-buffered saline (PBS) (pH 7.4) and left of the models, large values of the rest of the aforementioned parame- to stabilize at 30 C Samples were dissolved in PBS (pH 7.4) to ters prove the opposite, and therefore, the optimal model from each achieve a range of concentration from 50 to 800 μM; 0.1 ml of each method was chosen in such a way to achieve the best compromise sample was then mixed with 3.9-ml ABTS˙+ solution and incubated at between the former and the latter 30 C prior to the UV-vis measurements carried out at different time Apart from the leave-one-out cross validation and scrambling vali- points TEAC, which is the micromolar concentration of Trolox having dation, the validity of the selected models was also examined through the antioxidant capacity equivalent to 1-μM solution of the substance bootstrapping tests in which the complete data set was randomly split under investigation, was extrapolated from the Trolox calibration several times into training and test sets by choosing different random curve seeds on the field-based QSAR panel The respective models were built and their basic LOO statistics (Q2bstr and R2bstr) were computed Radical scavenging activity ð%Þ ¼ 
 Acontrol  Asample =Acontrol  100%: and compared with those of the optimal model Ferric reducing activity 2.2 Experimental | The reducing activity was measured by the coloured ferrous-TPTZ complex formed upon reduction of ferric ions.49 FRAP working solu- 2.2.1 | Materials tion was freshly prepared by mixing 50 ml of 0.3-M acetate buffer, ml of 10-mM TPTZ in 40-mM HCl, and ml of 20-mM FeCl36H2O 6-Hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid (Trolox), L- (pH 3.6), while samples were dissolved in deionized water to achieve glutathione (reduced), ABTS, and TPTZ were purchased from Sigma- a range of concentrations The FRAP and the sample solutions were Aldrich (St Louis, MO, USA) All amino acids were provided in form of incubated separately at 30 C in 30 before being mixed in a ratio L-isomers The other chemicals and reagents were of analytical grade 5:1 (v/v) The reduction reaction was then monitored by measuring Thermo Scientific Genesys 20 served as the equipment in all the absorbance of the FRAP-sample mixture at 593 nm for up to ultraviolet–visible (UV-vis) measurements 150 All measurements were performed in triplicate to ensure the reliability of the results The total antioxidant capacity was calculated using a standard curve prepared by replacing the sample by FeSO4 2.2.2 | Methods standard solution with concentration ranging from 10 to 100 μM Peptide synthesis All syntheses were performed manually in a glass syringe fitted with a RESULTS AND DISCUSSION | polypropylene sinter and stopcock The syntheses were performed on Rink resin and 2-chlorotrityl chloride on a scale of 0.5 mmol All amino 3.1 | Force field and Gaussian models acids were supplied as the fluorenylmethyloxycarbonyl (Fmoc) derivative with Boc-protection on tryptophan and tBu-protection on Tyr The construction of both force field and Gaussian models were side chains The Fmoc amino acids were activated with one equivalent implemented on the aligned dataset of 108 tripeptides A force field of 13 TRAN ET AL model and a Gaussian model were selected from 10 models built by The contour maps of the Gaussian model provided a more com- each method according to the criteria mentioned in the Section The plicated picture of the field impacts on the activity with the most intri- summary of statistical parameters corresponding to the optimal cate field contributions occurring at the third residue To be more models are presented in Table 2, demonstrating the quality of these specific, expanding the third residues into different directions would models in terms of fitting and predictability In particular, the force result in completely different effects (Figure 4A) The presence of field model had an excellent linear fitting of PLS factors with R2 of charged groups at this position would decrease the activity if they are 0.916, F value of 207.3, and stability of 0.996, whereas the resided in the side chain but improve activity if located around the corresponding fitting parameters for the 5-factor FTC model were backbone and outside of the side chain (Figure 4B) The situations similar with the figures being 0.917, 165.2, and 0.993, respectively were similar for hydrophobic and hydrogen-bond acceptor interaction The prediction performance of the force field model (Q2ext of 0.904, with hydrophobic interaction located outside and hydrogen-bond Pearson r of 0.954 and RMSE of 0.19) were also comparable with acceptor groups located on top of the ring would be beneficial for the those of the Gaussian model (0.890, 0.966, and 0.20), and all the high activity but would be detrimental if distributed elsewhere within the values of Q2ext and Pearson r and low values of RMSE indicated that third residue (Figure 4C,D) On the other hand, the improvements both models are statistically reliable in terms of predictability could only be seen explicitly at the first residue with charge along the Internal validations in forms of LOO, scrambling, and boo- side chain, hydrophobic group on the second residue, and hydrogen- tstrapping tests resulted in significant statistics confirming the robust- bond donor at the third residue (Figure 4E), whereas the negative ness of the two models Particularly, R2LOO of 0.893 and 0.875 and R2 effect would occur if more charged groups were added to the second Scramble of 0.136 and 0.203 were computed from the force field and amino acid the Gaussian models, respectively With the aforementioned statistics, The predicted TEAC values of 16 tryptophyllin tripeptides frag- the selected models were qualified to be used as a tool for antioxidant ments were provided in Table Overall, the predictions of the two screening purpose models highly agreed with each other with the most active peptide It should be noted that the contribution of steric interaction field being PWY(NH2) coming from tryptophyllin L 4.1 (LPW(NH2)) It was to the models was the most significant explaining 81.6% and 59.4% of predicted to have a TEAC value of around 2.9, higher than the refer- variance in the force field and Gaussian models, whereas the least ence peptide in the modelling data Interestingly, LPW, FPW, and contributor to TEAC activity was electrostatic interaction, which only SPW were found to have relatively high TEAC values of approxi- contributed 18.4% and 4.8%, respectively Each of the three other mately 1.7 Among these, FPW and SPW are fragments from tryp- + fields accounted for approximately 10% of the ABTS  scavenging tophyllin L 1.2 and 1.5, which were investigated in the previous paper activity of the peptides in the Gaussian model (Table 3) The plots and would not be mentioned further in this study, whereas LPW is expressing the correlation between the experimental activities and the N-terminal fragment of tryptophyllin L 4.1 As both fragments of predicted activities in the training and the test sets derived from the the tryptophyllin L 4.1 had been suggested to be more active than two models were presented in Figure trolox with regard to the ability of scavenging ABTS radical, Looking at the steric contour of the force field model, a large PWY(NH2) and tryptphyllin L 4.1 were synthesized The most active green area appeared on the C-terminal (third) residue of the template peptide in the modelling dataset PWY was also synthesized for the peptide implying positive influence of steric interaction to the activity purpose of comparison at this position, while it caused a mixed effect on the second residue Based on the analysis of contour plots above and bearing in mind (Figure 3A) Electrostatic interaction was found to have more explicit the significant contribution of steric interaction to the activity, a effects as it improved the activity if located along the first residue QSAR chart was created and employed to guide the design of several and reduced the activity if located around the third amino acid tripeptides by altering residues at three different positions of the (Figure 3B) structure PWY(NH2) These peptides were subjected to the two TABLE Statistics of the selected force field and Gaussian models Model SD R2 R2LOO R2Scramble Stability F RMSE Q2 Pearson r 4-factor force field 0.203 0.916 0.893 0.136 0.996 207.3 0.19 0.904 0.954 5-factor Gaussian 0.204 0.917 0.875 0.203 0.993 165.2 0.20 0.890 0.966 T A B L E Field fractions corresponding to the force field and Gaussian models Model Steric Electrostatic Hydrophobic H-bond acceptor H-bond donor 4-factor force field 0.816 0.184 n/a n/a n/a 5-factor Gaussian 0.594 0.048 Note: n/a: not applicable 0.134 0.150 0.074 of 13 TRAN ET AL FIGURE Plots illustrating observed and predicted activities of the training and test sets derived from the TEAC and FTC models models for prediction Among the newly designed peptides, only two ABTS radical scavenging assay from which their relative antioxidant peptides PYW and PYW(NH2) were predicted to have the TEAC abilities to trolox were extrapolated as TEAC values In this study, values roughly twice as high as Trolox and higher than the second samples of final concentrations from 1.25 to 20 μM were measured most active tryptophyllin L fragment (PWF(NH2)) (Table 4), and there- for up to 60 to observe any change over the time and due to the fore, they were also synthesized for further biochemical testing difference in concentration; however, the concentration 8.2 μM and reaction time of were selected to compare the activities of the peptides as they were the experimental conditions applied to collect 3.2 | ABTS radical scavenging activity the modelling dataset (Figure 5) Although all the peptides had the ABTS scavenging capabilities approximately two times higher than The five peptides with the highest TEAC values including PWY, trolox, the activities are not in the same order as the predictions To PWY(NH2), LPWY(NH2), PYW, and PYW(NH2) were subjected to the be more specific, P-Y-W was the most active peptide whereas the TRAN ET AL F I G U R E Contour maps of the five fields derived from the force field model (A) Steric field (positive: green, negative: yellow) F I G U R E Contour maps of the five fields derived from the Gaussian model (A) Steric (green: positive; yellow: negative); (B) electrostastic (blue: positive; red: negative); (C) hydrophobic (orange: positive, white: negative), (D) hydrogen-bond acceptor (yellow-green: positive; magenta: negative); (E) hydrogen-bond donor (violet: positive; cyan: negative) of 13 of 13 TRAN ET AL No Title Prediction from the force field model Prediction from the Gaussian model FPW 1.66 1.67 pEFP 0.34 0.32 FPF 1.56 1.52 SPW 1.74 1.74 IPW 1.65 1.66 LPW 1.73 1.73 FLP 0.17 0.17 pEIP 0.22 0.20 PWF 1.89 1.89 10 PWF(NH2) 1.94 1.97 11 FHR(NH2) 0.26 0.34 12 PWP 0.54 0.53 13 PWP(NH2) 0.59 0.60 14 PWY(NH2) 2.94 2.93 15 PFP 0.338 0.304 16 WFH 0.258 0.251 17 PYW 2.00 2.11 18 PYW(NH2) 2.24 2.25 T A B L E Predicted TEAC values from the force field and Gaussian models F I G U R E ABTS+ scavenging activities of the five peptides lowest activity among the tested was L-P-W-Y(NH2) with the TEAC hydrogen bonding, and the shift in radical scavenging mechanisms values being 2.34 and 1.99, respectively P-W-Y(NH2) had an TEAC of due to the change in structures are such of the features contributing 2.11 similar to that of P-Y-W(NH2) (2.16) The amidated peptides had to the deviation of the observed activities from the predictions In lower activities than the acidic counterparts, but the difference was addition, every prediction by any model is always associated with a only in the range of 0.1–0.2 trolox equivalent standard deviation of the regression, also known as the RMSE in the The dissimilarity in the order of TEAC values between the experi- fitted activity values, distributed over n  m  degrees of freedom mental results and predictions was explainable as the difference in (n number of peptides, m PLS factors), which should be taken into TEAC values derived from 3D-QSAR models was solely due to the dif- account as a source of deviation from the experimental values ference in interaction fields of the peptides, the experimental trolox It should be noticed that the TEAC value of PWY extrapolated equivalents, on the other hand, reflected a synergetic effect of struc- from the ABTS assay in this work was 2.22 compared with 2.78 tural and experimental features of these peptides The influence of recorded in the TEAC data (Table S1) This discrepancy was antici- reaction medium to formal charge of individual peptide, intramolecular pated since the radical sources of the current procedure and the one of 13 TRAN ET AL producing the TEAC dataset were different In the current assay, as F-P-W-L(NH2) and P-W-L(NH2) displayed higher TEAC values than ABTS+ was generated from ABTS and persulfate salt before adding their acidic derivatives both experimentally and statistically.24 samples, whereas in the original assay, samples were added to the Therefore, it can be concluded that amidation does not affect ABTS+ reaction medium containing ABTS and metmyoglobin, and H2O2 was scavenging abilities of peptides significantly or other factors may finally included to form the radical This procedure has been criticized overrule this effect as the order of reagent addition may lead to an overestimation of antioxidant capability, 50 Figure illustrated the difference in ABTS+ scavenging rates of which had also been the reason for us to choose the peptides, trolox, and glutathione (GSH) at two different concen- a different ABTS+ scavenging procedure from the original, and there- trations over 60 It can be seen that the time taken for the sample fore, PWY was included to evaluate the relative activities of all sur- to reach maximum scavenging capacity and the percentage of ABTS+ veyed peptides It should be emphasized that all of the tested scavenging were concentration dependent The smaller the concen- peptides are more active than glutathione (GSH) (1.59), a well-known tration, the longer it took for the peptide to reach to its maximum as naturally occurring antioxidant peptide well as corresponded to lower maximum At the concentration of The effect of amidation was found inconsistent with that of the 2.5 μM, the reactions of peptides were not able to complete by previous study.24 While predictions from the present models indi- 60 min, whereas it took less than 30 minutes for the 10 μM to achieve + cated that amidation moderately increased ABTS  scavenging ability the maximum scavenging capability Additionally, the peptides and (Table 4), experimental results showed an opposite trend for two pairs GSH were found to react much slower than trolox, which can be of tripeptides (PWY, PYW, and their amidated counterparts; Figure 5) explained mainly by the difference in the structural characteristics of The results were not in agreement with those from the previous study the peptides and trolox F I G U R E The rates of 2,20 -azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) radical cation (ABTS˙+) scavenging at (A) 2.5 μM and (B) 10 μM 10 of 13 TRAN ET AL F I G U R E Trolox equivalents of the five peptides and glutathione over a concentration range from 2.5 to 10 μM F I G U R E Ferric reducing activity of tryptophyllin L peptides and trolox at concentration of 25 μM and 30 The TEAC values of the five peptides and GSH over a concentra- reducing agents compared with GSH with the figures being 0.13-, tion range from 2.5 to 10 μM at was compared in Figure It 0.11-, and 0.14-mM Fe2+/mM peptide, whereas the newly designed can be seen that as the concentrations of peptides increased, the peptide PYW corresponded to the highest Fe2+ equivalent (0.26-mM activity gaps between the peptides and trolox became smaller In Fe2+/mM peptide) and PWY corresponded the second highest Fe2+ other words, the peptides and GSH are more efficient ABTS+ scaven- equivalent (0.18-mM Fe2+/mM peptide) gers than trolox at low concentration This finding is in agreement In the previous study, we found that all of the investigated tryp- with the results conducted on five tryptophyllin L peptides in the pre- tophyllins L had considerately lower ferric reducing power than trolox vious paper.24 In order to test whether this tendency repeats for the currently studied peptides, the ferric reducing power of peptides and trolox were measured at the same conditions The results revealed that they were 3.3 | Ferric reducing power indeed much weaker ferric reducing agents than trolox, which had a reducing equivalent of 2.14-mM Fe2+/mM trolox Ferric reducing reactions were used to examine reducing abilities of Different from the ABTS+ scavenging reactions, the ferric peptides, expressed as the concentration of Fe(II) giving the same reducing reactions of the peptides were extremely slow and the absorbance in the presence of the Fe(III)-TPTZ complex The ferrous divergence of activities increased with time as demonstrated in equivalents of the five peptides and GSH at the concentration of Figure 9A Although 30 was chosen to compare the reducing 25 μM and reaction time of 30 were presented in Figure The power of peptides, the reactions continued long after that with the most striking feature of this figure was that the acidic C-terminal pep- ferrous equivalent of PYW(NH2) rising up to 0.77 at 150 min, tides have much higher reducing power than the amidated counter- whereas tryptophyllin L 4.1 and PWY(NH2) reached 0.44- and parts Tryptophyllin L 4.1 and its fragment were slightly weaker 0.38-mM Fe2+/mM peptide, respectively Surprisingly, the reducing 11 of 13 TRAN ET AL FIGURE The change of ferric reducing capability of peptides over time at three different concentrations (A) 25, (B) 50, and (C) 100 μM abilities of the five peptides are inversely proportional to the con- inconsistent across the tryptophyllin L peptides studied and this effect centrations as illustrated in Figure 9A–C, whereas the situation was may be overridden by other factors The ferric reducing activities of opposite for GSH Further investigation is needed to explain this the peptides, on the other hand, were much weaker than that of phenomenon, which is out of the scope of this study and thus trolox and inversely proportional to the concentrations of the pep- would not be discussed here tides employed in the assay The decrease of reducing activities of the peptides compared with the radical scavenging activities implied that trolox and the peptides express their antioxidant power by different | C O N CL U S I O N S mechanisms, in which trolox prefers transferring electrons to the oxidants while the peptides may act via a hydrogen-atom-transfer or In this work, we have employed a fragment-based 3D-QSAR approach electron accepting mechanism Noticeably, the newly design peptide to explore antioxidant potential of all tryptophyllin L peptides Force PYW was found to be more active than glutathione in both assays at field and Gaussian models were constructed and used to predict the lowest concentrations used in our assay (25 and 50 uM) This + ABTS  scavenging abilities of 16 tripeptide fragments covering all of study demonstrated that statistical tools such as 3D-QSAR modelling the sequences of tryptophyllin L peptides From which, the most methods can be employed in conjunction with biochemical assays to active peptide PWY, its corresponding tryptophyllin L 4.1 and two effectively screen and design antioxidant peptides from databases of newly designed peptides PYW and PYW(NH2) were synthesized to naturally occurring peptides + investigate experimentally by ABTS  scavenging and ferric reducing antioxidant power assays The ABTS radical scavenging assay revealed AC KNOW LEDG EME NT that these peptides were strong ABTS radical scavengers with TEAC This work is funded by Industrial University of Ho Chi Minh City values approximately twice as high as trolox and higher than GSH under Research Grant No 64/HĐ-ĐHCN; Code No 21/1H03 The activity order was not in agreement with the predictions due to the difference in the experimental procedures applied and other pos- OR CID sible causes The effect of amidation on ABTS+ scavenging is found Thi Thanh Nha Tran https://orcid.org/0000-0003-4792-6540 12 of 13 RE FE R ENC E S Verbon EH, Post JA, Boonstra J The influence of reactive oxygen species on cell cycle progression in mammalian cells Gene 2012; 511(1):1-6 Finkel T Signal transduction by reactive oxygen species J Cell Biol 2011;194(1):7-15 Turpaev K Reactive oxygen species and regulation of gene expression Biochemistry (Mosc) 2002;67(3):281-292 Schieber M, Chandel NS ROS function in redox signaling and oxidative stress Curr Biol 2014;24(10):R453-R462 Nimse SB, Pal D Free radicals, natural antioxidants, and their reaction mechanisms RSC Adv 2015;5(35):27986-28006 Sisein EA Biochemistry of free radicals and antioxidants Scholars Acad J Biosci 2014;2(2):110-118 Ali SS, Ahsan H, Zia MK, Siddiqui T, Khan FH Understanding oxidants and antioxidants: classical team with new players J Food Biochem 2020;44(3):e13145 Carr AC, McCall MR, Frei B Oxidation of LDL by myeloperoxidase and reactive nitrogen species: reaction pathways and antioxidant protection Atertio Thromb Vasc Biol 2000;20(7):1716-1723 Apel K, Hirt H Reactive oxygen species: metabolism, oxidative stress, and signal transduction Annu Rev Plant Biol 2004;55(1):373-399 10 Bouayed J, Bohn T Exogenous antioxidants—double-edged swords in cellular redox state: health beneficial effects at physiologic doses versus deleterious effects at high doses Oxid Med Cell Longev 2010;3(4): 228-237 11 Lorenzo JM, Munekata PE, Gomez B, et al Bioactive peptides as natural antioxidants in food products—a review Trends Food Sci Technol 2018;79:136-147 12 Sato K Structure, content, and bioactivity of food-derived peptides in the body J Agric Food Chem 2018;66(12):3082-3085 13 Shahidi F Antioxidants in food and food antioxidants Food Nahrung 2000;44(3):158-163 14 Qian Z-J, Jung W-K, Byun H-G, Kim S-K Protective effect of an antioxidative peptide purified from gastrointestinal digests of oyster, Crassostrea gigas against free radical induced DNA damage Bioresour Technol 2008;99(9):3365-3371 15 Esfandi R, Walters ME, Tsopmo A Antioxidant properties and potential mechanisms of hydrolyzed proteins and peptides from cereals Heliyon 2019;5(4):e01538 16 Zhuang H, Tang N, Yuan Y Purification and identification of antioxidant peptides from corn gluten meal J Funct Foods 2013;5(4): 1810-1821 17 Sila A, Bougatef A Antioxidant peptides from marine by-products: isolation, identification and application in food systems A review J Funct Foods 2016;21:10-26 18 Bougatef A, Nedjar-Arroume N, Manni L, et al Purification and identification of novel antioxidant peptides from enzymatic hydrolysates of sardinelle (Sardinella aurita) by-products proteins Food Chem 2010; 118(3):559-565 19 Kou X, Gao J, Xue Z, Zhang Z, Wang H, Wang X Purification and identification of antioxidant peptides from chickpea (Cicer arietinum L.) albumin hydrolysates LWT-Food Sci Technol 2013;50(2):591-598 20 Escudero E, Mora L, Fraser PD, Aristoy M-C, Toldrá F Identification of novel antioxidant peptides generated in Spanish dry-cured ham Food Chem 2013;138(2–3):1282-1288 21 Liu R, Xing L, Fu Q, Zhou G, Zhang W A review of antioxidant peptides derived from meat muscle and by-products Antioxidants 2016; 5(3):32 22 Sun J, He H, Xie BJ Novel antioxidant peptides from fermented mushroom Ganoderma lucidum J Agric Food Chem 2004;52(21): 6646-6652 23 Wong F-C, Xiao J, Wang S, Ee K-Y, Chai T-T Advances on the antioxidant peptides from edible plant sources Trends Food Sci Technol 2020;99:44-57 TRAN ET AL 24 Tran TTN, Tran DP, Nguyen VC, Tran TDT, Bui TTT, Bowie JH Antioxidant activities of major tryptophyllin L peptides: a joint investigation of Gaussian-based 3D-QSAR and radical scavenging experiments J Pept Sci 2021;27(4):e3295 25 Yang X, Wang Y, Zhang Y, Lee W-H, Zhang Y Rich diversity and potency of skin antioxidant peptides revealed a novel molecular basis for high-altitude adaptation of amphibians Sci Rep 2016;6(1):1-11 26 Wang X, Ren S, Guo C, et al Identification and functional analyses of novel antioxidant peptides and antimicrobial peptides from skin secretions of four East Asian frog species Acta Biochim Biophys Sin 2017;49(6):550-559 27 Zou T-B, He T-P, Li H-B, Tang H-W, Xia E-Q The structure-activity relationship of the antioxidant peptides from natural proteins Molecules 2016;21(1):72 28 Li Y-W, Li B Characterization of structure–antioxidant activity relationship of peptides in free radical systems using QSAR models: key sequence positions and their amino acid properties J Theor Biol 2013;318:29-43 29 Cheng Y, Luo F, Zeng Z, et al DFT-based quantitative structure– activity relationship studies for antioxidant peptides Struct Chem 2015;26(3):739-747 30 Tian M, Fang B, Jiang L, Guo H, Cui J, Ren F Structure-activity relationship of a series of antioxidant tripeptides derived from β-Lactoglobulin using QSAR modeling Dairy Sci Technol 2015;95(4): 451-463 31 Chen N, Chen J, Yao B, Li Z QSAR study on antioxidant tripeptides and the antioxidant activity of the designed tripeptides in free radical systems Molecules 2018;23(6):1407 32 Olsen TH, Yesiltas B, Marin FI, et al AnOxPePred: using deep learning for the prediction of antioxidative properties of peptides Sci Rep 2020;10(1):1-10 33 Guo H, Wang Y, He Q, et al In silico rational design and virtual screening of antixoidant tripeptides based on 3D-QSAR modeling J Mol Struct 2019;1193:223-230 34 Yan W, Lin G, Zhang R, Liang Z, Wu W Studies on the bioactivities and molecular mechanism of antioxidant peptides by 3D-QSAR, in vitro evaluation and molecular dynamic simulations Food Funct 2020;11(4):3043-3052 35 Liao W, Gu L, Zheng Y, et al Analysis of the quantitative structure– activity relationship of glutathione-derived peptides based on different free radical scavenging systems Med Chem Commun 2016;7(11): 2083-2093 36 Li Y-W, Li B, He J, Qian P Quantitative structure–activity relationship study of antioxidative peptide by using different sets of amino acids descriptors J Mol Struct 2011;998(1–3):53-61 37 Steinborner ST, Wabnitz PA, Waugh RJ, et al The structures of new peptides from the Australian red tree frog ‘Litoria rubella’ The skin peptide profile as a probe for the study of evolutionary trends of amphibians Aust J Chem 1996;49(9):955-963 38 Bowie JH, Tyler MJ Host defense peptides from Australian amphibians: caerulein and other neuropeptides In: Handbook of Biologically Active Peptides Elsevier; 2006:283-289 39 Tran TTN Structural and mechanistic studies of post-translationally modified peptides and proteins South Australia, 5000: The Unviersity of Adelaide; 2014 40 Saito K, Jin D-H, Ogawa T, et al Antioxidative properties of tripeptide libraries prepared by the combinatorial chemistry J Agric Food Chem 2003;51(12):3668-3674 41 James J MOPAC2016, Stewart Computational Chemistry Springs Colorado; 2016 42 Goodyear MD, Krleza-Jeric K, Lemmens T The declaration of Helsinki Br Med J 2007;335(7621):624-625 43 Klebe G, Abraham U, Mietzner T Molecular similarity indices in a comparative analysis (CoMSIA) of drug molecules to correlate and predict their biological activity J Med Chem 1994;37(24):4130-4146 13 of 13 TRAN ET AL 44 Klebe G, Abraham U Comparative molecular similarity index analysis (CoMSIA) to study hydrogen-bonding properties and to score combinatorial libraries J Comput Aided Mol Des 1999;13(1):1-10 45 Ghose AK, Viswanadhan VN, Wendoloski JJ Prediction of hydrophobic (lipophilic) properties of small organic molecules using fragmental methods: an analysis of ALOGP and CLOGP methods Chem A Eur J 1998;102(21):3762-3772 46 Geladi P, Kowalski BR Partial least-squares regression: a tutorial Anal Chim Acta 1986;185:1-17 47 Kearsley SK, Smith GM An alternative method for the alignment of molecular structures: maximizing electrostatic and steric overlap Tetrahedron Comput Methodol 1990;3(6):615-633 48 Re R, Pellegrini N, Proteggente A, Pannala A, Yang M, Rice-Evans C Antioxidant activity applying an improved ABTS radical cation decolorization assay Free Radic Biol Med 1999;26(9–10):1231-1237 49 Benzie IF, Strain JJ The ferric reducing ability of plasma (FRAP) as a measure of “antioxidant power”: the FRAP assay Anal Biochem 1996;239(1):70-76 50 Strljbe M, Haenen GR, Van Den Berg H, Bast A Pitfalls in a method for assessment of total antioxidant capacity Free Radic Res 1997; 26(6):515-521 SUPPORTING INF ORMATION Additional supporting information may be found in the online version of the article at the publisher's website How to cite this article: Tran TTN, Tran DP, Nguyen TMA, et al Virtual screening and rational design of antioxidant peptides based on tryptophyllin L structures isolated from the Litoria rubella frog J Pep Sci 2021;e3380 doi:10.1002/psc.3380