The 29th Special CU-af Seminar 2021

The 29th Special CU-af Seminar 2021 August 25, 2021 Figure 5: Plasma flux enhanced wound healing process in the mouse models. Characteristics of wounds in mouse models with burn injury with or without bacterial infection after treatment by plasma flux (Plasma) or Argon gas alone (Untreat) as determined by wound area and wound rank score (see method) with the representative pictures (A-C), serum cytokines (IL-6 and TNF-α) (D, E) and gut permeability measurement (FITC-dextran assay) (F) (n = 5-7/ time-point or group) are demonstrated. 36

The 29th Special CU-af Seminar 2021 August 25, 2021 Figure 6: Plasma flux attenuated local inflammation in the mouse models. Characteristics of wounds in mouse models with burn injury with or without bacterial infection after treatment by plasma flux (Plasma) or Argon gas alone (Un-treat) as determined by cytokines from skin tissue (TNF-α, IL-6 and IL-10) (A-C), bacterial burdens (D) and collagen in the lesions with the representative Masson’s trichrome stained histological pictures (E) (n = 5-7/ time-point or group) are demonstrated. 37

The 29th Special CU-af Seminar 2021 August 25, 2021 2. Plasma flux promoted wound healing in burn wound of mice, regardless of infection The in vivo experiments with 2 mouse models of burn wound with and without infection by Staphylococcus aureus, the most common secondary infection in burn wound[42], were used. Notably, susceptibility against plasma flux of S. aureus and methicillin resistant S. aureus (MRSA) were non-different as approximately 30% reduction on colony count (by culture at 24 h) in both strains after a 30s plasma flux expo-sure (data not shown). As such, the macroscopic 7 days wound monitoring demonstrated the higher area of wound and inflammation (wound rank score, see method) in burn wound with bacterial infection compared with non-infection (Figure 5A-C). With plasma flux treatment, wound area was smaller with the less prominent inflammatory signs in the burn wound model either with or without bacterial infection (Figure 5A-C). However, our models were not severe enough to demonstrate the systemic effect of burn wound or infected burn wound as indicated by the non-difference in serum cytokines and gut permeability defect when compared with control group (Figure 5D-F) different from other publications[30,43]. At 7 days of the models, plasma flux attenuated tissue pro-inflammatory cytokines (TNF-α and IL-6), but not anti-inflammatory IL-10, in the bacterial infected burn wound, but not in the non-infected burn wound (Figure 6A-C). These data support an anti-inflammatory effect of plasma during the wound healing process. On the other hand, bacterial burdens in skin of the infected wound model was higher than the wound without infection; however, plasma flux did not reduce bacterial burdens in wound of both models (Figure 6D), despite the bactericidal activity of plasma flux in vitro[44]. Moreover, the accumulation of collagen, an indicator of the better wound healing ,[45] as determined by Masson’s trichrome staining was higher in plasma-treated non-infected burn wound compared with control group (Figure 6E). However, the collagen deposition was not different between plasma treatment and control group in the infected burn wound model (Figure 6E). These data implied that plasma flux promoted wound healing in infected burn wound partly through an-ti-inflammation (skin cytokines) (Figure 6A, B), while improved wound condition in the non-infected burn wound, at least in part, through the accelerated collagen deposition (Figure 6E). Discussion Argon-sourced non-thermal plasma induced anti-inflammatory macrophages that promoted wound healing in mouse burn wound models regardless of the infection. 1. Non-thermal plasma flux induced anti-inflammatory macrophages and fibroblast migration Non-thermal atmospheric pressure plasma might be beneficial in several health topics, including dentistry[46], hematology (coagulation)[47], microbiology (microbial eradication)[48], surgery (wound healing)[49] and oncology[50]. Interestingly, plasma flux could augment wound healing (in burn) or suppress angiogenesis (in cancer) with different energy levels of the ionized gas[51] that induces different intensities of bio-active molecules[52]. Indeed, plasma flux increased macrophage ROS, a mediator of several cell activities (apoptosis, proliferation and inflammation), leading to inhibit NF-κB[53,54] in several cells, including epithelial cells (H9C2)[55] and T cells (Jurkat cell)[56]. Here, plasma flux enhanced ROS and reduced NF-κB abundance in macro-phages that might, at least in part, attenuate macrophage inflammatory responses. However, the reduced pro-inflammatory status of plasma-treated macrophages was not associated with the cell energy status as the abundance of AMPK, a sensor of cell energy[39], 38

The 29th Special CU-af Seminar 2021 August 25, 2021 was not different from control group. Because i: M2 polarized macrophages promote several processes of wound healing[25], ii) excessive pro-inflammatory macrophages decelerated wound healing[57] and iii) plasma flux could alter macrophage inflammatory responses[58], severe genes are interesting to explored. Without LPS (an inflammatory activation), macrophages in a neutral state did not produce inflammatory cytokines and plasma flux showed only a slight impact on macrophage as indicated by the upregulation only Fizz, a biomarker of M2 macrophage polarization, but not other M2 polarized genes of macrophages. With LPS stimulation, plasma flux demonstrated the higher anti-inflammatory effect as indicated by reduced IL-6 production (at 24 h post-LPS), IL-1β down-regulation (6 h post-LPS) and IL-10 upregulation (3 and 6 h post-LPS). However, the expression M2 associated genes was not different between plasma flux treatment and the control group implying the plasma flux induced anti-inflammatory macrophages but did not profoundly induce M2 macrophage polarization. Nevertheless, the anti-inflammatory state could promote wound healing process[25] and the wound healing promotion in our mouse models might partly due to the plasma flux anti-inflammatory effect. Due to the influence of fibroblasts in wound healing processes (fibrin clot lysis and production of extra cellular matrix and collagen[45]), impact of plasma flux on fibroblasts was tested. Indeed, plasma flux induced fibroblast migration without the enhanced fibroblast proliferation. Despite the promotion on both proliferation and migration in human fibroblast-like cells by helium plasma flux[38], our Argon-based plasma flux could not induce fibroblast proliferation possibly due to the difference on gas source, power intensities or the active molecules. Nevertheless, the enhanced fibroblast migration is one of the wound healing promotion factors[59]. Notably, the non-difference proliferation between plasma-treated group and control implied a safety of plasma flux on fibroblasts. 2. Non-thermal plasma flux promoted healing process of burn wound, regardless of infection Amongst the proposed plasma flux application, wound healing promotion is one of the most interesting treatment indications[60], mostly through the anti-inflammation effect on epithelial cells[38,61,62]. However, other cell types, including immune cells and fibroblasts are also important in wound healing process[63]. Unfortunately, there are still limited data on the impact of plasma flux on non-epithelial cells. Because burn wound is associated with the high mortality rate[64] partly through immune dysregulation-induced opportunistic infection[16,65], burn wound models with and without S. aureus infection are used[66]. Although MRSA-infected burn wound is the most important problem in clinical practice, a standard ATCC strain of S. aureus, but not clinical isolated MRSA, were used due to a concern on model reproducibility in this level of a proof of concept experiment. Indeed, plasma flux treatment attenuated burn wound with and without infection as evaluated by wound area and wound inflammatory score. However, plasma flux might promote wound healing differently between burn wound and burn infected wound. In non-infected burn wound, there was non-different dermal inflammatory cytokines between plasma-treated and control group. However, plasma promoted wound healing in non-infected burn wound possibly through the enhanced fibroblast migration and collagen production. As such, the effective migration accelerated a condition for fibroblasts to effectively produce collagen[59]. On the other hand, in the infected burn wound, plasma flux reduced pro-inflammatory cytokines in dermal tissue of without an effect on collagen production. These data implied a promotion of wound healing process through an anti-inflammatory effect in the infected burn wound. 39

The 29th Special CU-af Seminar 2021 August 25, 2021 In-deed, the pro-inflammatory macrophages might be too high in the infected burn wound due to the activation by pathogen molecules[67] and plasma flux adjusted balance of the responses. However, bacterial burdens from culture of the whole dermal tissue of wound was not different between plasma-treated and control group. Perhaps, the bactericidal effect might be limited only on surface of the lesions due to the limited depth of penetration by energy flux. More proper method to determine bacterial bur-dens on the surface of the lesions is needed. Nevertheless, our experiments support the utilization of plasma flux in burn wound that could be easily used in the real clinical practice. More studies in patients are interesting. Conclusion In conclusion, non-thermal atmospheric pressure Argon-based plasma induced anti-inflammatory macrophages, possibly through i) the reduced NF-κB by ROS and ii) the enhanced fibroblast migration, that were responsible for wound healing promotion in murine burn wound models. References 1. Izadjoo, M.; Zack, S.; Kim, H.; Skiba, J. Medical applications of cold atmospheric plasma: state of the science. J Wound Care 2018, 27, S4-S10, doi:10.12968/jowc.2018.27.Sup9.S4. 2. Bernhardt, T.; Semmler, M.L.; Schafer, M.; Bekeschus, S.; Emmert, S.; Boeckmann, L. Plasma Medicine: Applications of Cold Atmospheric Pressure Plasma in Dermatology. Oxid Med Cell Longev 2019, 2019, 3873928, doi:10.1155/2019/3873928. 3. Rehman, M.U.; Jawaid, P.; Uchiyama, H.; Kondo, T. Comparison of free radicals formation induced by cold atmospheric plasma, ultrasound, and ionizing radiation. Arch Biochem Biophys 2016, 605, 19-25, doi:10.1016/j.abb.2016.04.005. 4. Davies, K.J. The broad spectrum of responses to oxidants in proliferating cells: a new paradigm for oxidative stress. IUBMB Life 1999, 48, 41-47, doi:10.1080/713803463. 5. Jacobson, M.D. Reactive oxygen species and programmed cell death. Trends Biochem Sci 1996, 21, 83-86. 6. Fridman, A.A. Plasma chemistry; Cambridge University Press: Cambridge ; New York, 2008; pp. xlii, 978 p. 7. Kalghatgi, S.; Kelly, C.M.; Cerchar, E.; Torabi, B.; Alekseev, O.; Fridman, A.; Friedman, G.; Azizkhan-Clifford, J. Effects of non-thermal plasma on mammalian cells. PLoS One 2011, 6, e16270, doi:10.1371/journal.pone.0016270. 8. Zhou, Y.; Hileman, E.O.; Plunkett, W.; Keating, M.J.; Huang, P. Free radical stress in chronic lymphocytic leukemia cells and its role in cellular sensitivity to ROS-generating anticancer agents. Blood 2003, 101, 4098-4104, doi:10.1182/blood-2002-08-2512. 9. Dunnill, C.; Patton, T.; Brennan, J.; Barrett, J.; Dryden, M.; Cooke, J.; Leaper, D.; Georgopoulos, N.T. Reactive oxygen species (ROS) and wound healing: the functional role of ROS and emerging ROS-modulating technologies for augmentation of the healing process. Int Wound J 2017, 14, 89-96, doi:10.1111/iwj.12557. 10. Trachootham, D.;Alexandre, J.; Huang, P. Targeting cancer cells by ROS-mediated mechanisms: a radical therapeutic approach? Nat Rev Drug Discov 2009, 8, 579-591, doi:10.1038/nrd2803. 11. Kvam, E.; Davis, B.; Mondello, F.; Garner, A.L. Nonthermal atmospheric plasma rapidly disinfects multidrug-resistant microbes by inducing cell surface damage. Antimicrob Agents 40

The 29th Special CU-af Seminar 2021 August 25, 2021 Chemother 2012, 56, 2028-2036, doi:10.1128/AAC.05642-11. 12. Schmidt, A.; Bekeschus, S. Redox for Repair: Cold Physical Plasmas and Nrf2 Signaling Promoting Wound Healing. Antioxidants (Basel) 2018, 7, doi:10.3390/antiox7100146. 13. Haertel, B.; von Woedtke, T.; Weltmann, K.D.; Lindequist, U. Non-thermal atmospheric-pressure plasma possible application in wound healing. Biomol Ther (Seoul) 2014, 22, 477-490, doi:10.4062/biomolther.2014.105. 14. Fathollah, S.; Mirpour, S.; Mansouri, P.; Dehpour, A.R.; Ghoranneviss, M.; Rahimi, N.; Safaie Naraghi, Z.; Chalangari, R.; Chalangari, K.M. Investigation on the effects of the atmospheric pressure plasma on wound healing in diabetic rats. Sci Rep 2016, 6, 19144, doi:10.1038/srep19144. 15. Duchesne, C.; Banzet, S.; Lataillade, J.J.; Rousseau, A.; Frescaline, N. Cold atmospheric plasma modulates endothelial nitric oxide synthase signalling and enhances burn wound neovascularisation. J Pathol 2019, 249, 368-380, doi:10.1002/path.5323. 16. Jeschke, M.G.; van Baar, M.E.; Choudhry, M.A.; Chung, K.K.; Gibran, N.S.; Logsetty, S. Burn injury. Nature Reviews Disease Primers 2020, 6, doi:ARTN 11 10.1038/s41572-020-0145-5. 17. Smolle, C.; Cambiaso-Daniel, J.; Forbes, A.A.; Wurzer, P.; Hundeshagen, G.; Branski, L.K.; Huss, F.; Kamolz, L.P. Recent trends in burn epidemiology worldwide: A systematic review. Burns 2017, 43, 249-257, doi:10.1016/j.burns.2016.08.013. 18. Lachiewicz, A.M.; Hauck, C.G.; Weber, D.J.; Cairns, B.A.; van Duin, D. Bacterial Infections After Burn Injuries: Impact of Multidrug Resistance. Clin Infect Dis 2017, 65, 2130-2136, doi:10.1093/cid/cix682. 19. Dai, T.; Vrahas, M.S.; Murray, C.K.; Hamblin, M.R. Ultraviolet C irradiation: an alternative antimicrobial approach to localized infections? Expert Rev Anti Infect Ther 2012, 10, 185-195, doi:10.1586/eri.11.166. 20. Nicol, M.J.; Brubaker, T.R.; Honish, B.J., 2nd; Simmons, A.N.; Kazemi, A.; Geissel, M.A.; Whalen, C.T.; Siedlecki, C.A.; Bilen, S.G.; Knecht, S.D.; et al. Antibacterial effects of low-temperature plasma generated by atmospheric-pressure plasma jet are mediated by reactive oxygen species. Sci Rep 2020, 10, 3066, doi:10.1038/s41598-020-59652-6. 21. Schmidt, A.; Bekeschus, S.; Wende, K.; Vollmar, B.; von Woedtke, T. A cold plasma jet accelerates wound healing in a murine model of full-thickness skin wounds. Exp Dermatol 2017, 26, 156-162, doi:10.1111/exd.13156. 22. Kaushik, N.K.; Kaushik, N.; Adhikari, M.; Ghimire, B.; Linh, N.N.; Mishra, Y.K.; Lee, S.J.; Choi, E.H. Preventing the Solid Cancer Progression via Release of Anticancer-Cytokines in Co-Culture with Cold Plasma-Stimulated Macrophages. Cancers (Basel) 2019, 11, doi:10.3390/cancers11060842. 23. Dang, C.P.; Leelahavanichkul, A. Over-expression of miR-223 induces M2 macrophage through glycolysis alteration and attenuates LPS-induced sepsis mouse model, the cell-based therapy in sepsis. PLoS One 2020, 15, e0236038, doi:10.1371/journal.pone.0236038. 24. Taratummarat, S.; Sangphech, N.; Vu, C.T.B.; Palaga, T.; Ondee, T.; Surawut, S.; Sereemaspun, A.; Ritprajak, P.; Leelahavanichkul, A. Gold nanoparticles attenuates bacterial sepsis in cecal ligation and puncture mouse model through the induction of M2 macrophage polarization. BMC Microbiol 2018, 18, 85, doi:10.1186/s12866-018-1227-3. 25. Krzyszczyk, P.; Schloss, R.; Palmer, A.; Berthiaume, F. The Role of Macrophages in Acute and Chronic Wound Healing and Interventions to Promote Pro-wound Healing Phenotypes. Front Physiol 2018, 9, 419, doi:10.3389/fphys.2018.00419. 26. Cai, E.Z.; Ang, C.H.; Raju, A.; Tan, K.B.; Hing, E.C.; Loo, Y.; Wong, Y.C.; Lee, H.; Lim, 41

The 29th Special CU-af Seminar 2021 August 25, 2021 J.; Moochhala, S.M.; et al. Creation of consistent burn wounds: a rat model. Arch Plast Surg 2014, 41, 317-324, doi:10.5999/aps.2014.41.4.317. 27. Kim, H.K.; Missiakas, D.; Schneewind, O. Mouse models for infectious diseases caused by Staphylococcus aureus. J Immunol Methods 2014, 410, 88-99, doi:10.1016/j.jim.2014.04.007. 28. Lu, X.; Naidis, G.V.; Laroussi, M.; Reuter, S.; Graves, D.B.; Ostrikov, K. Reactive species in non-equilibrium atmospheric-pressure plasmas: Generation, transport, and biological effects. Phys Rep 2016, 630, 1-84, doi:10.1016/j.physrep.2016.03.003. 29. Jang, S.I.; Mok, J.Y.; Jeon, I.H.; Park, K.H.; Nguyen, T.T.; Park, J.S.; Hwang, H.M.; Song, M.S.; Lee, D.; Chai, K.Y. Effect of electrospun non-woven mats of dibutyryl chitin/poly(lactic acid) blends on wound healing in hairless mice. Molecules 2012, 17, 2992-3007, doi:10.3390/ molecules17032992. 30. Baron, P.; Traber, L.D.; Traber, D.L.; Nguyen, T.; Hollyoak, M.; Heggers, J.P.; Herndon, D.N. Gut failure and translocation following burn and sepsis. J Surg Res 1994, 57, 197-204, doi:10.1006/jsre.1994.1131. 31. Earley, Z.M.; Akhtar, S.; Green, S.J.; Naqib, A.; Khan, O.; Cannon, A.R.; Hammer, A.M.; Morris, N.L.; Li, X.; Eberhardt, J.M.; et al. Burn Injury Alters the Intestinal Microbiome and Increases Gut Permeability and Bacterial Translocation. PLoS One 2015, 10, e0129996, doi:10.1371/journal.pone.0129996. 32. Panpetch, W.; Sawaswong, V.; Chanchaem, P.; Ondee, T.; Dang, C.P.; Payungporn, S.; Leelahavanichkul, A. Candida Administration Worsens Cecal Ligation and Puncture-Induced Sepsis in Obese Mice Through Gut Dysbiosis Enhanced Systemic Inflammation, Impact of Pathogen-Associated Molecules From Gut Translocation and Saturated Fatty Acid. Front Immunol 2020, 11, 561652, doi:10.3389/fimmu.2020.561652. 33. Ondee, T.; Gillen, J.; Visitchanakun, P.; Somparn, P.; Issara-Amphorn, J.; Dang Phi, C.; Chancharoenthana, W.; Gurusamy, D.; Nita-Lazar, A.; Leelahavanichkul, A. Lipocalin-2 (Lcn-2) Attenuates Polymicrobial Sepsis with LPS Preconditioning (LPS Tolerance) in FcGRIIb Deficient Lupus Mice. Cells 2019, 8, doi:10.3390/cells8091064. 34. Visitchanakun, P.; Saisorn, W.; Wongphoom, J.; Chatthanathon, P.; Somboonna, N.; Svasti, S.; Fucharoen, S.; Leelahavanichkul, A. Gut leakage enhances sepsis susceptibility in iron-overloaded beta-thalassemia mice through macrophage hyperinflammatory responses. Am J Physiol Gastrointest Liver Physiol 2020, 318, G966-G979, doi:10.1152/ajpgi.00337.2019. 35. Issara-Amphorn, J.; Chancharoenthana, W.; Visitchanakun, P.; Leelahavanichkul, A. Syk InhibitorAttenuates Polymicrobial Sepsis in FcgRIIb-Deficient Lupus Mouse Model, the Impact of Lupus Characteristics in Sepsis. J Innate Immun 2020, 12, 461-479, doi:10.1159/000509111. 36. Panpetch, W.; Kullapanich, C.; Dang, C.P.; Visitchanakun, P.; Saisorn, W.; Wongphoom, J.; Wannigama, D.L.; Thim-Uam,A.; Patarakul, K.; Somboonna, N.; et al. CandidaAdministration Worsens Uremia-Induced Gut Leakage in Bilateral Nephrectomy Mice, an Impact of Gut Fungi and Organismal Molecules in Uremia. mSystems 2021, 6, doi:10.1128/mSystems.01187-20. 37. Kanazawa, S.; Fujiwara, T.; Matsuzaki, S.; Shingaki, K.; Taniguchi, M.; Miyata, S.; Tohyama, M.; Sakai, Y.; Yano, K.; Hosokawa, K.; et al. bFGF regulates PI3-kinase-Rac1-JNK pathway and promotes fibroblast migration in wound healing. PLoS One 2010, 5, e12228, doi:10.1371/ journal.pone.0012228. 38. Brun, P.; Pathak, S.; Castagliuolo, I.; Palu, G.; Brun, P.; Zuin, M.; Cavazzana, R.; Martines, E. Helium generated cold plasma finely regulates activation of human fibroblast-like primary cells. PLoS One 2014, 9, e104397, doi:10.1371/journal.pone.0104397. 39. Jaroonwitchawan, T.; Visitchanakun, P.; Dang, P.C.; Ritprajak, P.; Palaga, T.; Leelahavanichkul, 42

The 29th Special CU-af Seminar 2021 August 25, 2021 A. Dysregulation of Lipid Metabolism in Macrophages Is Responsible for Severe Endotoxin Tolerance in FcgRIIB-Deficient Lupus Mice. Front Immunol 2020, 11, 959, doi:10.3389/ fimmu.2020.00959. 40. Udompornpitak, K.; Bhunyakarnjanarat, T.; Charoensappakit, A.; Dang, C.P.; Saisorn, W.; Leelahavanichkul, A. Lipopolysaccharide-Enhanced Responses against Aryl Hydrocarbon Receptor in FcgRIIb-Deficient Macrophages, a Profound Impact of an Environmental Toxin on a Lupus-Like Mouse Model. 2021, 22, 4199. 41. Issara-Amphorn, J.; Surawut, S.; Worasilchai, N.; Thim-Uam, A.; Finkelman, M.; Chindamporn, A.; Palaga, T.; Hirankarn, N.; Pisitkun, P.; Leelahavanichkul, A. The Synergy of Endotoxin and (1-->3)-beta-D-Glucan, from Gut Translocation, Worsens Sepsis Severity in a Lupus Model of Fc Gamma Receptor IIb-Deficient Mice. J Innate Immun 2018, 10, 189-201, doi:10.1159/000486321. 42. Church, D.; Elsayed, S.; Reid, O.; Winston, B.; Lindsay, R. Burn wound infections. Clin Microbiol Rev 2006, 19, 403-434, doi:10.1128/CMR.19.2.403-434.2006. 43. Herndon, D.N.; Zeigler, S.T. Bacterial translocation after thermal injury. Crit Care Med 1993, 21, S50-54, doi:10.1097/00003246-199302001-00010. 44. Bae, S.; Lim, D.; Kim, D.; Jeon, J.; Oh, T. In vitro antibacterial effects of non-thermal atmospheric plasma irradiation on Staphylococcus pseudintermedius and Pseudomonas aeruginosa. Pol J Vet Sci 2020, 23, 13-19, doi:10.24425/pjvs.2019.131414. 45. Bainbridge, P. Wound healing and the role of fibroblasts. J Wound Care 2013, 22, 407-408, 410-412, doi:10.12968/jowc.2013.22.8.407. 46. Sung, S.J.; Huh, J.B.; Yun, M.J.; Chang, B.M.; Jeong, C.M.; Jeon, Y.C. Sterilization effect of atmospheric pressure non-thermal air plasma on dental instruments. J Adv Prosthodont 2013, 5, 2-8, doi:10.4047/jap.2013.5.1.2. 47. Kalghatgi, S.U.; Fridman, G.; Cooper, M.; Nagaraj, G.; Peddinghaus, M.; Balasubramanian, M.; Vasilets, V.N.; Gutsol, A.F.; Fridman, A.; Friedman, G. Mechanism of blood coagulation by nonthermal atmospheric pressure dielectric barrier discharge plasma. Ieee T Plasma Sci 2007, 35, 1559-1566, doi:10.1109/Tps.2007.905953. 48. Gilmore, B.F.; Flynn, P.B.; O’Brien, S.; Hickok, N.; Freeman, T.; Bourke, P. Cold Plasmas for Biofilm Control: Opportunities and Challenges. Trends Biotechnol 2018, 36, 627-638, doi:10.1016/j.tibtech.2018.03.007. 49. Kubinova, S.; Zaviskova, K.; Uherkova, L.; Zablotskii, V.; Churpita, O.; Lunov, O.; Dejneka, A. Non-thermal air plasma promotes the healing of acute skin wounds in rats. Sci Rep 2017, 7, 45183, doi:10.1038/srep45183. 50. Wolff, C.M.; Kolb, J.F.; Weltmann, K.D.; von Woedtke, T.; Bekeschus, S. Combination Treatment with Cold Physical Plasma and Pulsed Electric Fields Augments ROS Production and Cytotoxicity in Lymphoma. Cancers (Basel) 2020, 12, doi:10.3390/cancers12040845. 51. Boeckmann, L.; Schafer, M.; Bernhardt, T.; Semmler, M.L.; Jung, O.; Ojak, G.; Fischer, T.; Peters, K.; Nebe, B.; Muller-Hilke, B.; et al. Cold Atmospheric Pressure Plasma in Wound Healing and Cancer Treatment. Appl Sci-Basel 2020, 10, doi:ARTN 6898 10.3390/ app10196898. 52. Smolkova, B.; Frtus, A.; Uzhytchak, M.; Lunova, M.; Kubinova, S.; Dejneka, A.; Lunov, O. Critical Analysis of Non-Thermal Plasma-Driven Modulation of Immune Cells from Clinical Perspective. Int J Mol Sci 2020, 21, doi:10.3390/ijms21176226. 53. Gloire, G.; Legrand-Poels, S.; Piette, J. NF-kappaB activation by reactive oxygen species: fifteen years later. Biochem Pharmacol 2006, 72, 1493-1505, doi:10.1016/j.bcp.2006.04.011. 43

The 29th Special CU-af Seminar 2021 August 25, 2021 54. Morgan, M.J.; Liu, Z.G. Crosstalk of reactive oxygen species and NF-kappaB signaling. Cell Res 2011, 21, 103-115, doi:10.1038/cr.2010.178. 55. Levrand, S.; Pesse, B.; Feihl, F.; Waeber, B.; Pacher, P.; Rolli, J.; Schaller, M.D.; Liaudet, L. Peroxynitrite is a potent inhibitor of NF-{kappa}B activation triggered by inflammatory stimuli in cardiac and endothelial cell lines. J Biol Chem 2005, 280, 34878-34887, doi:10.1074/ jbc.M501977200. 56. Ogino, T.; Hosako, M.; Hiramatsu, K.; Omori, M.; Ozaki, M.; Okada, S. Oxidative modification of IkappaB by monochloramine inhibits tumor necrosis factor alpha-induced NF-kappaB activation. Biochim Biophys Acta 2005, 1746, 135-142, doi:10.1016/j.bbamcr.2005.10.005. 57. Sindrilaru, A.; Peters, T.; Wieschalka, S.; Baican, C.; Baican, A.; Peter, H.; Hainzl, A.; Schatz, S.; Qi, Y.; Schlecht, A.; et al. An unrestrained proinflammatory M1 macrophage population induced by iron impairs wound healing in humans and mice. J Clin Invest 2011, 121, 985-997, doi:10.1172/JCI44490. 58. Bekeschus, S.; Scherwietes, L.; Freund, E.; Liedtke, K.R.; Hackbarth, C.; von Woedtke, T.; Partecke, L.I. Plasma-treated medium tunes the inflammatory profile in murine bone marrow-derived macrophages. Clin Plasma Med 2018, 11, 1-9, doi:10.1016/ j.cpme.2018.06.001. 59. Addis, R.; Cruciani, S.; Santaniello, S.; Bellu, E.; Sarais, G.; Ventura, C.; Maioli, M.; Pintore, G. Fibroblast Proliferation and Migration in Wound Healing by Phytochemicals: Evidence for a Novel Synergic Outcome. Int J Med Sci 2020, 17, 1030-1042, doi:10.7150/ijms.43986. 60. Weltmann, K.D.; von Woedtke, T. Plasma medicine-current state of research and medical application. Plasma Phys Contr F 2017, 59, doi:Artn 014031 10.1088/0741-3335/59/1/014031. 61. Jeong, W.S.; Kwon, J.S.; Choi, E.H.; Kim, K.M. The Effects of Non-Thermal Atmospheric Pressure Plasma treated Titanium Surface on Behaviors of Oral Soft Tissue Cells. Sci Rep 2018, 8, 15963, doi:10.1038/s41598-018-34402-x. 62. Choi, J.H.; Song, Y.S.; Lee, H.J.; Hong, J.W.; Kim, G.C. Inhibition of inflammatory reactions in 2,4-Dinitrochlorobenzene induced Nc/Nga atopic dermatitis mice by non-thermal plasma. Sci Rep 2016, 6, 27376, doi:10.1038/srep27376. 63. Arturson, G. Pathophysiology of the burn wound. Ann Chir Gynaecol 1980, 69, 178-190. 64. 64. Gomez, R.; Murray, C.K.; Hospenthal, D.R.; Cancio, L.C.; Renz, E.M.; Holcomb, J.B.; Wade, C.E.; Wolf, S.E. Causes of mortality by autopsy findings of combat casualties and civilian patients admitted to a burn unit. J Am Coll Surg 2009, 208, 348-354, doi:10.1016/ j.jamcollsurg.2008.11.012. 65. Ballard, J.; Edelman, L.; Saffle, J.; Sheridan, R.; Kagan, R.; Bracco, D.; Cancio, L.; Cairns, B.; Baker, R.; Fillari, P.; et al. Positive fungal cultures in burn patients: a multicenter review. J Burn Care Res 2008, 29, 213-221, doi:10.1097/BCR.0b013e31815f6ecb. 66. Abdullahi, A.; Amini-Nik, S.; Jeschke, M.G. Animal models in burn research. Cell Mol Life Sci 2014, 71, 3241-3255, doi:10.1007/s00018-014-1612-5. 67. van de Goot, F.; Krijnen, P.A.J.; Begieneman, M.P.V.; Uhich, M.M.W.; Middelkoop, E.; Niessen, H.W.M. Acute Inflammation is Persistent Locally in Burn Wounds: A Pivotal Role for Complement and C-Reactive Protein. Journal of Burn Care & Research 2009, 30, 274-280, doi:10.1097/BCR.0b013e318198a252. 44



Early Detection of Anthracnose on Mango Fruit Using Hyperspectral Imaging Ubonrat SIRIPATRAWAN and Yoshio MAKINO

The 29th Special CU-af Seminar 2021 August 25, 2021 Early Detection of Anthracnose on Mango Fruit Using Hyperspectral Imaging Ubonrat SIRIPATRAWAN 1* and Yoshio MAKINO 2* Abstract Mango is one of the most important economic fruit in Thailand. However, mango is susceptible to anthracnose disease causing quality loss and reduced market value. This research developed a rapid, nondestructive and chemical-free method based on hyperspectral imaging spectroscopy (HIS) coupled with chemometrics and pseudo-color image processing for early detection of anthracnose caused by Colletotrichum gloeosporioides on mango. The HIS system covering the wavelength range of 400-1000 nm was used to acquire the hypercube data of the samples. The HIS integrated with chemometrics including principal component analysis (PCA) and discriminant factor analysis (DFA) was able to detect the infection at the early stage of disease symptoms and classify different levels of infection. Anthracnose symptom distribution maps were constructed using pseudo-color image processing to facilitate visualization of the anthracnose symptom. The HIS coupled with chemometrics and pseuso-color image processing can be used to rapidly identify and classify samples with anthracnose infection. 1Department of Food Technology, Faculty of Science, Chulalongkorn University, Bangkok THAILAND 2Graduate School of Agricultural and Life Science, The University of Tokyo, Tokyo JAPAN 47

The 29th Special CU-af Seminar 2021 August 25, 2021 Introduction and Objectives Mango is one of the most popular and important economic fruit in Thailand and is exported worldwide due to its good taste and flavor. However, mango fruit is susceptible to various postharvest diseases such as anthracnose causing quality deterioration and reduced market value (Office of Agricultural Economics, 2561; Zhang et al., 2013). Anthracnose caused by Colletotrichum species is recognized as the most important field and postharvest disease of mango. Anthracnose infection often occurs in the field and remains in quiescence (latent) stage. When the fruit enters the ripening stage, postharvest anthracnose symptoms progressively develops as dark brown to black circular lesions cover the fruit skin. The task of anthracnose detection is challenging since the fungal infection is usually not visible at early stage of infection and the appearance of the infected fruit is similar to healthy fruit (Li et al., 2016; Sankaran et al., 2010). Monitoring of health and detection of early stage of fungal infection in mango fruit is thus crucial for effective postharvest management. Early information on disease detection can facilitate the control of diseases and can reduce economic losses in postharvest and food industry (Sankaran et al., 2010). Visual scales have traditionally been used to observe the severity of the disease on infected fruit (Koomen and Jeffries, 1993). Brodrick (1978) use the method which is based on the percentages of the area affected on the fruit by the fungal pathogen. Corkidi et al. (2006) developed an accurate image-analysis method by using camera to assess mango fruit with lesions from anthracnose infection. A technique based on a visible yellow fluorescence under UV light irradiation is also employed recently for visual inspection of fungal decay of fruit (Momin et al., 2013). However, these techniques are effective only when the anthracnose symptoms clearly appear on the fruit skin (Li et al., 2016; Nagle et al., 2016). Thus, the infection at early stage of fungal growth is still challenging. There is still a demand for a rapid, sensitive, and selective method for early detection of anthracnose on mango fruit to facilitate advancements in postharvest management. The current technologies that have potential to assist in monitoring quality and diseases in fruit include spectroscopic-based and volatile profiling-based detection methods. Among them, hyperspectral imaging spectroscopy (HIS) offers many advantages over conventional spectroscopic methods. HIS combines spectroscopy and computer vision to obtain both spectral and spatial information from an analyzed object. The acquired multicomponent data, called hypercube, contain hundreds of reflectance images, of which each pixel contains the spectral information over a certain wavelength range (Kandpal et al., 2016; Siripatrawan andHarte, 2015). The main advantage of hyperspectral imaging is in the spatial feature, because it can provide information about the distribution of the components on the samples (ElMasry and Nakauchi, 2016; Piqueras et al., 2013). With HIS analysis, a large area on the sample is feasible and consequently often generates multicomponent information. HIS is a non-contact, nondestructive method which can provide rapid and accurate detection of fruit diseases at early stages(ElMasry and Nakauchi, 2016; Kandpal et al., 2016). Hence, this research proposes the HIS technique coupled with chemometrics and pseudo-color image process for early detection and classification of anthracnose on mango fruit. To our knowledge, this is the first attempt to employ HIS as a rapid, nondestructive, and chemical-free approach with pseudo-color image processing for detection of anthracnose infection on mango. The objective of this research was, hence, to develop a rapid, simple, and effective method based on HIS spectroscopic analysis coupled with chemometrics and image analysis to early identify anthracnose infection on mango fruit. 48

The 29th Special CU-af Seminar 2021 August 25, 2021 Methods Mango treatment and preparation of fungal inoculum Before use, Colletotrichum gloeosporioides was maintained on potato dextrose agar (PDA) and incubated at 25 °C as recommended by Sarkhosh et al. (2017). Inoculum suspension was prepared from fresh, mature 10-day-old cultures grown on PDA. The specific inoculum concentration (spores/mL) was adjusted by microscopic enumeration with a haemocytometer. Before use, Nam Dok Mai mango fruit (Mangifera indica L.) with uniform size and free of physical defect was treated with 200 ppm Sodium hypochlorite, rinsed twice with tap water and air dried. The fruit was then wounded using a sterile syringe at the equator. Approximately 10 μL of the conidial suspension of C. gloeosporioides containing 105 to 106 spores mL−1 was pipetted to each wound (Deng, Zhou and Zeng, 2015). Each fruit was then individually packaged in a mesh produce bag and stored at 20 °C and 85-90 % RH. Hyperspectral imaging system A pushbroom line-scanning hyperspectral imaging system (JFE, Techno-Research Corporation, Tokyo, Japan) consisting of a moving table, light source, objective lens, and a charged couple device (CCD) camera (Figure 1), was used to produce full contiguous spectral and spatial information. The system is attached to a stage control unit (Model SGSP 26-200, Sigma-Kaki Co. Ltd., Tokyo, Japan). The ImSpector spectrograph (Model V10, Spectral Imaging Ltd., Oulu, Finland) operating in the wavelength range of 400 to 1000 nm with a spectral resolution of 5 nm was used to align the imaging system, acquire images, and store hyperspectral image data in a 12-bit binary file. A hyperspectral image cube was created by scanning in the direction perpendicular to the spatial plane of the ImSpector. The Spectrum Analyzer (Version 1.8.5, JFE, Techno-Research Corporation, Tokyo, Japan) software was used to construct the hyperspectral images. Figure 1: Hyperspectral imaging system Mathematical pretreatment Data were made up of 240 samples from 6 subgroups, including mango inoculated with C. gloeosporioides on initial day of inoculation (d0) and inoculated mango after storage at 20 °C and 85 % RH for 2, 4, 6, 8 and 10 days. The hypercube data of the ROI obtained one sample scanned comprised spatial data of 100*50 pixels and reflectance spectral of 121 distinct wavelengths. Reflectance spectra of each sample were mathematically pretreated in order to reduce the spectral variability due to the morphological effects of the fruit, to 49

The 29th Special CU-af Seminar 2021 August 25, 2021 attenuate the intrinsic noise of the hyperspectral data. The optimal mathematical pretreatment was selected by exploring from different approaches including Adjacent-averaging, Standard normal variate (SNV), Savitzky-Golay filter for smoothing, Savitzky-Golay 1st derivatives and Mean normalization to compensate the scatter-induced baseline offsets and intensity variations from path length differences. Data exploration using PCA Principal component analysis (PCA) was used to explore the relationship between all variables. The variables which contribute most to sample spoilage were determined by compressing the wavelength variables. PCA removes multi-collinearity and finds a lower dimensional representative of a dataset of which as much information as possible about the original data is preserved. Sample classification using DFA Following the data reduction, linear discriminant factor analysis (DFA) was used to classify samples with different levels of infection. DFA was used to visualize class separation between samples using the data obtained from PCA. DFA is a method that differentiates between the within- and between-class scatters to derive class-specific feature spaces. Classification was performed by assigning a pattern vector to the class with the closest Mahalanobis distance metric (Siripatrawan and Makino, 2015). Pseudo-color image processing of symptom distribution map In order to facilitate a rapid and easy interpretation of the hyperspectral data, the pseudo-color image processing of the distribution map of anthracnose symptom was performed on each pixel of the selected ROI on the spatial plane of a hyperspectral image. The pseudo-color image was used to facilitate visualization and interpretation of the severity of anthracnose symptoms. The image analysis was applied for detection and classification of different state of anthracnose symptom. The image processing was performed by firstly unfolding the hypercube data, determining the optimum PC score image which can best describe the samples and then transforming them into the pseudo-color image. All matrix calculations and image processing were performed using the routines MATLAB (Mathworks Inc., Natick, MA, USA) routines written by the authors. Results and Discussion Growth of fungi on mango during storage A digital photograph of mango fruit was taken to observe the lesion of anthracnose symptom area around the inoculation point on the fruit skin. Figure 2 displays the mango fruit infected with C. gloeosporioides and stored at 20 °C for 0, 2, 4, 6, 8, and 10 days (marked as d0, d2, d4, d6, d8 and d10, respectively). The discolored lesion area was not observed during the early days of storage, but appeared after 4 days of storage and became larger with the storage time. Figure 2: Mango fruit infected with C. gloeosporioides and stored at 20 °C for 0, 2, 4, 6, 8, and 10 days. 50

The 29th Special CU-af Seminar 2021 August 25, 2021 Hypercube of reflectance data Figure 3 shows signal intensity profiles of healthy skin, infected skin of mango and background of hyperspectral data. The detector signal intensity counts were transformed into reflectance units by comparing with spectra of dark current and dividing by similarly corrected total reflectance spectrum. Figure 4 (a) presents average raw signals of each sample subgroup. The detector signal intensity counts were transformed into reflectance units (Figure 4(b)) by comparing with spectra of dark current and dividing by similarly corrected total reflectance spectrum. Figure 3: Signal intensity profiles of healthy skin, infected skin of mango and background of HIS data. Figure 4: Signal intensity profiles (a) and reflectance profiles (b) of healthy skin, infected skin of mango and background of spectral data. The hyperspectral data were made a three-dimensional array (hypercube), comprising contiguous wavebands for each spatial position of a target studied and each pixel of the image contained the spectrum of that specific position. The reflectance spectra of each sample were obtained from a region of interest (ROI) on the sample surface. The selected rectangular ROI with the size of 5000 pixels obtained from the spatial data of 100 x 50 pixels around the center of the inoculation mark of each sample. The resulting 5000 pixel spectra, each with 121 data points, can be thought of as a fingerprint which can be used to characterize the alterations of biochemical components secreted by C. gloeosporioides during the growth or the rot of fruit tissue during spore germination. Figure 5 displays the concatenation (6 sample groups*40 samples in each group) of hyperspectroscopic reflectance data of infected mango on the first day and after storage for 2- 10 days. Changes the reflectance values of infected mango fruit with different storage time are possibly due to the present of the fungal secreted proteins such as glycoprotein which is found during the biotrophic phase of C. gloeosporiodes development (Gong et al., 2020; Naveen et al., 2021), the rot of fruit tissue during spores germination, the intracellular primary hyphae extending through host tissue, and the cell wall-degrading enzymes secreted by C. gloeosporiodes during the growth (De Silva, 2017). 51

The 29th Special CU-af Seminar 2021 August 25, 2021 The differences in reflectance with different levels of fungal infection was possibly due to an invasion of the anthracnose causing an alteration of biochemical compositions caused by fungal colonization (Naveen et al., 2020), the formation of fungal structure (hyphae, mycelium, and spores), and the type and concentration of metabolic products from fungal activities as affected by culture age (Singh et al., 2010). Figure 5: Three dimensional HIS data of mango fruit infected with C. gloeosporioides and stored at 20 °C for 0, 2, 4, 6, 8, and 10 days. 52

The 29th Special CU-af Seminar 2021 August 25, 2021 Mathematical pretreatment of hypercube data Since the morphology of mango fruit which is of agricultural base often presents a variability caused by spatial position of curved shape samples (Jia et al., 2020; Esquerre et al., 2012), as well as variations due to light scattering. The optimal mathematical pretreatment was selected by exploring from different approaches including Adjacent- averaging, Standard normal variate, Savitzky-Golay filter for smoothing, Savitzky-Golay 1st derivatives and Mean normalization. It was found that Mean normalization was optimal to compensate the scatter-induced baseline offsets and intensity variations from path length differences. Principal component analysis PCA was performed to transform a number of correlated variables into a smaller number of uncorrelated variables called principal components and the results are displayed in Figure 6. The PCA decomposed the data matrix (240 samples× 121wavelengths) into two smaller matrices of PC loading and PC scores. From PCA, the 121 correlated wavelength variables were linearly transformed into a relatively small set of 3 uncorrelated variables (PCs). The first PC accounts for a major fraction (91.83%) of the total variance of the data. The second and third axes are orthogonal to the first eigenvector and accounts for 6.8 % and 0.34 %, respectively, of the variation not accounted for by the first factor. These three PCs together represent as high as 98.97 % of the information in the overall data set. The loading in Figure 6(a) is an indication of the importance of particular variables for each PC, i.e., the highlighting of variables which have the highest positive or negative loadings on a PC. Wavelengths ranging from 960–1000 nm had high positive loadings on PC1. Wavelengths around 850-900 and 560-570 nm had high positive and high negative loadings, respectively, on PC2. It should be noted that PC1and PC2 are important because of their high loadings. The score plot in Figure 6 (b) shows sample clusters ascribed to positive and negative scores on all 3 PCs. By plotting the points on the new coordinates, significant effects within the data can be better visualized. The PCA score plot did not show clear clustering among 6 subgroups. Therefore, following the PCA data reduction, discriminant factor analysis (DFA) was used to classify samples with different levels of anthracnose symptom as a result of storage time. Figure 6: PCA loading plot (a) and score plot (b) of HIS spectra of mango fruit infected with C. gloeosporioides stored at 20 °C for 0-10 days. Classification of mango with different symptom DFA was used to visualize class separation between samples using the data obtained from PCA. For DFA classification, the first variable (Factor 1) was the linear combination of the original variables that best discriminate among the groups, the second variable (Factor 2) was orthogonal to the first and was the next best combination of variables. Sample factor scores were plotted on a DFA graph, and the separation of class-labeled samples is shown in Figure 7. From the DFA pattern, the data were correctly classified into 6 groups. However, some samples 53

The 29th Special CU-af Seminar 2021 August 25, 2021 from the adjacent storage time overlapped. This indicated that the HIS could identify infected from healthy samples and classify different levels of infection, which may be attributed to the metabolic products secreted by C. gloeosporioides during the growth and the rot of fruit tissue due to spore germination. From the classificatory discriminant factor analysis, the percent of correct classification ranged from 88-100% with a mean value of 95.3%. The mango fruit in d8 was falsely identified as d10 and the other way around, probably because the infection of d8 and d10 was in the similar stage of which the symptom appears as a prominent dark circular sunken due to the abundant sporulation of the anthracnose totally covered the area around the center of infection. Sample d0 was analyzed immediately after the inoculation and thus was used as control (healthy sample). After 2 days of storage, the infected fruit samples (d2) showed no sign of anthracnose symptom and is thus considered in the quiescence stage. During the quiescent stage, Colletotrichum species have suspended activity and almost no growth occurs, but the fungi remain dormant inside the plant tissue (Prusky et al., 2013). From our results, the HIS was able to separate the healthy samples (d0) from d2 samples and thus the proposed technique can detect the C. gloeosporiodes infection at the early stage of infection when the symptom is not visible to the naked eye. Our results demonstrated that differences in the levels of fungal infection and extent of colonization of fungi were reflected by the HIS spectra. Therefore, the levels of fungal development in correspondence to the storage time could be monitored by the HIS. Most importantly, the HIS is proved to be useful for early detection of anthracnose infection on mango fruit even when the symptom is not visible to the naked eye. Figure 7: DFA of HIS spectra of mango fruit infected with C. gloeosporioides and stored at 20 °C for 0, 2, 4, 6, 8, and 10 days. Pseudo-color image of symptom distribution map Hyperspectral data obtained from the sample comprised contiguous wavebands for each spatial position of a target studied and each pixel of the image contained the spectrum of that specific position. The resulting pixel spectra can be thought of as a fingerprint which can be used to characterize the biochemical compositions and rot of fruit tissue due to the growth of C. gloeosporioides , and consequently indicate the anthacnose symptoms. Therefore, an individual hyperspectral image can be mapped to provide spatial information of anthracnose symptoms. Each single pixel in the image of ROI around the area of inoculation point contains a reflectance spectrum in relation to metabolic components secreted by C. gloeosporioides or the rot of fruit tissue during the fungal growth. To facilitate rapid and easy visualization and interpretation of fruit with different levels of infection, the symptom distribution map was constructed on each pixel of the selected ROI on the spatial plane of a hyperspectral image using the pseudo-color image processing. 54

The 29th Special CU-af Seminar 2021 August 25, 2021 Figure 8 (a) presents the mango fruit infected with C. gloeosporioides and stored at 20 °C and 85% RH for 0, 2, 4, 6, 8, and 10 days, and Figure 8 (b) demonstrates the RGB color image of the ROI (5000 pixels) around the area of infection center of the infected mango shown in Figure 8 (a). The RGB images were constructed using wavelength at 720 nm (R), 580 nm (G), and 460 nm (B). According to the RGB image of mango fruit in Figures 8(a), the infected mango on day 2 (d2) showed no symptoms of anthracnose infection. PCA was used to determine the best PC which can be used to visualize reflectance profile of each pixel. In this study, 3 PCs were extracted with a cumulative variance of 98%. PC1 accounts for a major fraction (91.83%) of the total variance of the data. Therefore, only PC1 score image was used for further construction of the pseudo-color image processing of the samples. Figure 8 (c) displays the psudo-color images with jet color criterion of the infection distribution map of the samples which was constructed on the original spatial locations of the source pixel spectra. Each pixel was colored with respect to the severity of the symptom, as shown on the color bar on the right of the infection distribution map. By applying pseudo-color to each pixel in the tested image, a color distribution map was produced. Pseudo-color images can facilitate the visualization and interpretation of the level of anthracnose infection on the mango fruit by presenting complicated information in a single uncomplicated image. The results recognized in this study have confirmed that the differences in distribution maps between samples could be used as an indication to evaluate the infection severity of the samples. This technique enables early sorting of healthy and infected fruit without additional laborious chemical analysis. Figure 8: (a) Mango fruit infected with C. gloeosporioides and stored at 20 °C for 0, 2, 4, 6, 8, and 10 days, (b) RGB color image, and (c) Distribution map of psedo-color PC1-score image of the ROI of the infected mango fruit (color scale on the right indicate the severity of the symptom). Conclusion This study developed a rapid, nondestructive and chemical-free method based on HIS coupled with chemometrics and pseudo-color image processing for early detection of anthracnose on mango. When integrated with chemometrics, the HIS was able to detect and classify the infection at the early stage before the onset of disease symptoms. The pseudo-color images processing can be used to construct the symptom distribution map to facilitate visualization and interpretation of different levels of anthracnose infection on the mango fruit by presenting complicated information in a single uncomplicated image. The results in this study have emphasized the capability of the developed HIS technique as a rapid and nondestructive method to identify and classify healthy and infected fruit, as well as the differences in severity of the symptom. The results suggested that HIS coupled with 55

The 29th Special CU-af Seminar 2021 August 25, 2021 chemometrics and pseuso-color image processing can be used to rapidly identify and classify samples with anthracnose infection. Success of this research would prevent anthracnose-infected mango from entering the food chain which can consequently reduce economic losses through effective postharvest management. Acknowledgments This research was funded by the Asahi Glass Foundation and we are also indebted to the Thailand Toray Science Foundation for the complimentary financial support granted to Ubonrat Siripatrawan. References 1. De Silva, D. D., Crous, P. W., Ades, P. K., Hyde, K. D., Taylor, P.W. Life styles of Colletotrichum species and implications for plant biosecurity. Fungal Biol Rev, 31, 2017, 155-168. 2. Del Fiore, A., Reverberi, M., Ricelli, A., Pinzari, F., Serranti, S., Fabbri, A.A., Bonifazi, G., Fanelli, C., 2010. Early detection of toxigenic fungi on maize by hyperspectral imaging analysis. Int J Food Microbiol 144, 64-71. 3. Deng, L., Zhou, Y., Zeng, K. Pre-harvest spray of oligochitosan induced the resistance of harvested navel oranges to anthracnose during ambient temperature storage, Crop Protection, 70, 2015, 70-76. 4. ElMasry, G.M., Nakauchi, S. Image analysis operations applied to hyperspectral images for non-invasive sensing of food quality–a comprehensive review, Biosyst Eng, 142, 2016, 53–82. 5. Gong, A., Jing, Z., Liu, W., Zhang, K., Tan, Q., Wang, G., Liu W. Bioinformatic analysis and functional characterization of the CFEM proteins in maize anthracnose fungus Colletotrichum graminicola. J Integr Agric, 19, 2020, 541-550. 6. Jackowiak, H., Packa, D., Wiwart, M., Perkowski, J., 2005. Scanning electron microscopy of Fusarium damaged kernels of spring wheat. Inter J Food Microbiol 98, 113-123. 7. Jia B., Wang W.,Ni X.Lawrence K.C., Zhuang H., Yoon, S., Gao, Z. Essential processing methods of hyperspectral images of agricultural and food products. Chemom Intell Lab Syst,198, 2020, 103936. 8. Kandpal, L.M., Lohumi, S., Kim, M.S., Kang, J.S., Cho, B.K. Near-infrared hyperspectral imaging system coupled with multivariate methods to predict viability and vigor in muskmelon seeds, Sensor Actuat B-Chem, 229, 2016, 534–544. 9. Li, J., Ye, X., Wang, Q., Zhang, C., He, Y., 2014. Development of prediction models for determining N content in citrus leaves based on hyperspectral imaging technology. Spectrosc Spect Anal, 34, 212-216. 10. Li, T., Wang, Y., Chang, C., Hu, N., Zheng, Y. Color-appearance-model based fusion of gray and pseudo-color images for medical applications. Inf Fusion, 19, 2014, 103-114. 11. Li, J., Huang, W., Tian, X., Wang, C., Zhao, C. Fast detection and visualization of early decay in citrus using Vis-NIR hyperspectral imaging, Comput Electron Agric, 127, 2016, 582-592. 12. Liu, F., Li, G., Yang, S., Yan, W., He, G., Lin, L. Detection of heterogeneity on multi-spectral transmission image based on multiple types of pseudo-color maps. Infrared Physics Technol, 106, 2020, 103285. 56

The 29th Special CU-af Seminar 2021 August 25, 2021 13. Nagle, M., Intani, K., Romano, G., Mahayothee, B., Müller, J. Determination of surface color of ‘all yellow’ mango cultivars using computer vision. Int J Agric Biol Eng, 9, 2016, 42-50. 14. Naveen, J., Navya, H. M., Hithamani, G., Hariprasad, P., Niranjana, S. R. Pathological, biochemical and molecular variability of Colletotrichum truncatum incitant of anthracnose disease in chili (Capsicum annuum L.). Microb Pathog, 152, 2021, 104611. 15. Office of Agricultural Economics, Ministry of Agriculture and Cooperatives http://www. phtnet.org/phtic-research/view-article.aspID=10 [Access: September, 2018]. 16. Prusky, D., Alkan, N., Mengiste, T., Fluhr, R., 2013. Quiescent and necrotrophic life style choice during postharvest disease development. Annu Rev Phytopathol, 51, 155e176. 17. Sankaran, S., Mishra, A., Ehsani, R., Davis C., A review of advanced techniques for detecting plant diseases, Comput Electron Agric, 72, 2010, 1-13 18. Sarkhosh, A., Vargas, A.I., Schaffer, B., Palmateer, A.J., Farzaneh, M. Postharvest management of anthracnose in avocado (Persea americana Mill.) fruit with plant-extracted oils, Food Packag Shelf, 12, 2017, 16-22. 19. Singh,C. B., Jayas, D. S., Paliwal, J., White,N. D.G. Identification of insect-damaged wheat kernels using short- wave near-infrared hyperspectral and digital colour imaging. Comput Electron Agric, 73, 2010, 118-125. 20. Siripatrawan U., Harte, B. Data visualization of Salmonella Typhimurium contamination in packaged fresh alfalfa sprouts using a Kohonen network, Talanta, 136, 2015, 128-135. 21. Siripatrawan, U., Makino, Y., Kawagoe, Y., Oshita, S. Rapid detection of Escherichia coli contamination in packaged fresh spinach using hyperspectral imaging. Talanta, 85, 2011, 276-281. 22. Yeh, Y, Chung, W, Liao, J, Chung, C, Kuo, Y, Lin, T. Strawberry foliar anthracnose assessment by hyperspectral imaging. Comput Electron Agric, 122, 2016, 1-9. 57

Functional Identification of Dof Transcription Factors Controlling Auxin Biosynthesis and Starch Degradation in Durian Fruit Ripening Supaart Sirikantaramas Gholamreza Khaksar and Pinnapat Pinsorn

The 29th Special CU-af Seminar 2021 August 25, 2021 Functional Identification of Dof Transcription Factors Controlling Auxin Biosynthesis and Starch Degradation in Durian Fruit Ripening Supaart Sirikantaramas1* Gholamreza Khaksar1* and Pinnapat Pinsorn1* Abstract DNAbinding with one finger (Dof) transcription factor family regulates diverse biological processes, including fruit ripening. Here, comprehensive transcriptome analysis revealed 24 Dofs in durian pulps (DzDofs), out of which 15 were expressed in the fruit pulp. Some of these DzDofs harboured a differential expression, but durian cyclic Dof2 (DzcDof2) and DzDof2.1 exhibited marked ripening-associated expression patterns during post-harvest ripening of two commercial durian cultivars from Thailand, Monthong and Phuangmanee and were selected as candidate Dofs. Correlation network analysis of DzcDof2 with target ripening-related genes revealed a strong positive correlation between DzcDof2 and ethylene biosynthetic genes, followed by DzcDof2 and auxin biosynthesis, and DzcDof2 and beta-amylase, involved in starch degradation. Hence, we suggest a ripening-associated role of DzcDof2 possibly through transcriptional regulation of ethylene and auxin biosynthesis and starch degradation. Moreover, dual-luciferase reporter assay determined that DzDof2.1 mediates durian fruit ripening via trans-activating methionine gamma lyase, involved in aroma formation. 1Molecular Crop Research Unit, Department of Biochemistry, Faculty of Science, Chulalongkorn University, Bangkok, Thailand 59

The 29th Special CU-af Seminar 2021 August 25, 2021 Introduction and Objectives Transcription factors (TFs) are the main regulators of gene expression networks which control various biological processes in plants, including plant growth, development, and responses to environmental stimuli. The functional characterization of TFs could provide a deeper understanding of how plants grow and respond to various environmental stressors. The DNA binding with one finger (Dof) proteins constitute a ubiquitous plant-specific TF family. A highly conserved domain of about 50-52 amino acids (known as Dof domain), uniformly found at the N-terminus, is the signature characteristic of members of the DofTF family. The Dof domain has a Cys2/Cys2 Zn2+ finger structure which specifically binds to the cis-element of AAAG or its reverse-oriented sequence CTTT in the promoter of its targeted genes. The C terminus of Dofs is highly variable and acts as a transcriptional activator or repressor of its responsive genes. Dof TFs regulate a wide range of biological processes, including plant hormonal signaling, light responses, seed germination, carbon metabolism, and responses to both biotic and abiotic stressors. To date, the identification and functional characterization of the members of the Dof TF family have been extensively documented for numerous plant species, such as Arabidopsis (Yanagisawa, 2002) and several non-model crop species, including tomato (Cai et al., 2013), rice (Lijavetzky et al., 2003), Chinese cabbage (Ma et al., 2015), cucumber (Wen et al., 2016), pepper (Wu et al., 2016) and banana (Feng et al., 2016). Recent studies have well documented that an increasing number of TFs have been identified with ripening-associated functions in both climacteric and non-climacteric fruit, such as grape (Nicolas et al., 2014), sweet cherry (Shen et al., 2014), strawberry (Medina-Puche et al., 2014), banana (Shan et al., 2012) and tomato (Ma et al., 2014). However, our actual knowledge regarding the role of Dof TFs during ripening of economic fruit crops is almost scarce. Notably, Feng et al. (2016) identified four ethylene-inducible Dofs which were expressed increasingly during banana ripening. This was the first report of a possible role for Dof TFs in climacteric ripening. Notably, they investigated the interactions of banana Dofs (MaDofs) with a ripening-associated ethylene response factor (MaERF9), as well as the involvement of MaDofs in the transcriptional regulation of some ripening-associated genes, such as genes related to fruit softening and aroma. In addition, Zhang et al. (2018) investigated the role of kiwifruit Dof (AdDof3) in starch degradation. AdDof3 was shown to interact with AdBAM3L promoter, a key gene in starch degradation. Durian (Durio zibethinus L.) is an edible tropical fruit endemic to Southeast Asia. In the region, durian is known as the ‘king of fruits’ for its formidable spiny husk, overpowering flavor, and unique odor (Teh et al., 2017). Durian plays a vital role in this region as an important export commodity. Given that durian is a climacteric fruit, its shell life is restricted once ripening process starts. This phenomenon could negatively affect the handling and transportation of durian and cause major economic losses. In Thailand, two durian cultivars, known as Phuangmanee and Monthong, have been widely cultivated in many regions. These two cultivars exhibit different characteristics including odor strength and postharvest ripening duration. Therefore, gaining a deeper understanding of ripening process in durian is of utmost importance. Despite the importance of durian as an economic fruit crop, durian-related genomics research is almost non-existent, especially with regard to the Dof TFs and their potent roles during post-harvest ripening of durian fruit. The draft genome of durian was previously released (Teh et al., 2017), which enabled further studies on the identification of TFs regulating fruit ripening on a genome-wide scale. Previously, we conducted a genome-wide analysis of the Dof (DNA binding with one 60

The 29th Special CU-af Seminar 2021 August 25, 2021 finger) TF family and identified 24 durian Dofs (DzDofs), of which 15 were expressed in the fruit pulp. The functional characterization of DzDof2.2 suggested a role during fruit ripening by regulating auxin biosynthesis and auxin–ethylene crosstalk (Khaksar et al., 2019). In another study, we identified a member of the auxin response factor (ARF) TF family, DzARF2A, which mediates durian fruit ripening through the transcriptional regulation of ethylene biosynthetic genes (Khaksar and Sirikantaramas, 2020). Using metabolome and transcriptome analyses, Sangpong et al. (2021) investigated dynamic changes in the contents of flavor-related metabolites during the post-harvest ripening of durian fruit and identified key genes involved in their biosynthetic pathways. These reports provide us with a better understanding of the transcriptional and hormonal regulatory networks involved in durian fruit ripening. However, our actual knowledge of the molecular mechanisms through which Dof TFs regulate post-harvest ripening of durian fruit is still lacking. The different expression levels of ripening-associated and cultivar-dependent DzDofs could contribute to different sensory characteristics and faster ripening of fast-ripening cultivars, such as Phuangmanee compared to slow-ripening ones, such as Monthong. Functional characterization of these Dofs can be exploited as valuable information towards molecular marker development for breeding new cultivars. Herein, to address this, we conducted a transcriptome-wide analysis to confirm our previous study and rediscovered 24 DzDofs. We then profiled their expression levels during post-harvest ripening of durian fruit cv. Monthong and Phuangmanee. We then investigated and visualized the gene expression correlations of our candidate ripening-associated and cultivar-dependent DzDof with some ripening-related genes. Finally, we examined the in vivo transcriptional activity of a candidate ripening-associated DzDof. Methods Plant materials Durian (Durio zibethinus L.) fruit samples cv. Phuangmanee and Monthong were harvested from a commercial durian orchard located in Trat province in the eastern part of Thailand. Fruit samples harbouring similar size and weight (~1-2 kg each) were collected at mature stage which is at 90 days after anthesis (DAA) (for Phuangmanee) and 105 DAA (for Monthong). Three types of samples (unripe, midripe, and ripe) of these cultivars were used in our study. Fruit harvested at mature stage were kept at room temperature (30 °C) for one day and then peeled (unripe samples). To obtain midripe fruit samples, fruits harvested at the mature stage were kept at room temperature for two days (for Phuangmanee) and three days (for Monthong) and then peeled. For ripe samples, fruit samples harvested at the mature stage were kept at room temperature for three days (for Phuangmanee) and five days (for Monthong) and then peeled. After peeling, two central pulps were collected and processed following the method described in Pinsorn et al. (2018). A texture analyser was used to measure fruit softness of the first pulp to ensure samples of the two cultivars were compared at the same ripening stage (Khaksar et al., 2019). Thereafter, the second fruit pulp was collected without a seed, immediately frozen in liquid nitrogen, and stored at -80 °C for further analysis (RNA extraction). Transcriptome analysis To obtain transcriptome data for durian fruit cv. Monthong at the three stages of post-harvest ripening (unripe, midripe, and ripe), sequencing reads from the RNA-Seq study of durian fruit cv. Monthong (generated by our group) were retrieved from a public repository database with 61

The 29th Special CU-af Seminar 2021 August 25, 2021 the following accession number: PRJNA683229 (Sangpong et al., 2021). For the Phuangmanee cultivar, total RNA was extracted from durian samples at unripe, midripe, and ripe stages (at least three biological replicates) using PureLink Plant RNA Reagent (Invitrogen™, USA) following the manufacturer’s instruction. After removing the genomic DNA with DNase I (Thermo Fisher Scientific™), the quality and quantity of RNA samples were examined using agarose gel electrophoresis and an Eppendorf BioPhotometer D30. For each biological replicate, one paired-end library with an approximate 300 bp insert size was prepared using an in-house protocol at the Beijing Genomics Institute (BGI-Shenzhen, China). Libraries were sequenced on BGISEQ-500 platform. More than 30 million reads were generated for each sample. Mapping the reads to the D. zibethinus reference genome and expression analysis We used the OmicsBox program (Biobam, Spain) for transcriptome data analysis. Raw reads were filtered to obtain high-quality clean reads by removing adapters, reads shorter than 60 bp, and low-quality reads with a Q-value ≤ 30 using FastQC and Trimmomatic. Then, a gene-level analysis was performed by aligning the reads against the reference genome of durian cv. Musang King (Teh et al., 2017) using STAR (Spliced Transcripts Alignment to a Reference). Counting of reads and expression analysis were performed with HTSeq-count using default parameters. Transcripts with normalized reads <1 reads per kilobase of exon per million fragments (RPKM) were considered not expressed. Transcriptome-wide identification and expression profiling of durian Dofs Based on gene annotation and bioinformatics analysis, we identified 24 genes harboring the Dof domain as Dof family genes in durian (DzDofs), termed following the previously annotated DzDofs in the durian genome (Teh et al., 2017). First, we downloaded the hidden Markov model (HMM) file corresponding to the Dof domain (PF02701) from the Pfam protein family database (http://pfam.xfam.org/). We then used it as a query to search against the transcriptome database of durian fruit cv. Monthong and Phuangmanee, using HMMER software (http://hmmer.org/) with the parameters of score (bits) >200 and e-value cut-off ≤1e−5. The amino acid sequences of DzDofs were further confirmed in the SMART database for the presence of the conserved Dof domain. We then profiled the expression levels of these DzDofs at three stages (unripe, midripe, and ripe) during post-harvest ripening of both Monthong and Phuangmanee cultivars. The transcripts were represented as the mean of the RPKM value at each ripening stage. Expression data are presented as fold-change. This approach enabled us to identify 15 fruit pulp-expressed DzDofs. Gene network visualization To investigate and visualize the gene expression correlations of our candidate ripening- associated and cultivar-dependent DzDof with some ripening-related genes previously identified by Teh et al. (2017) [from an RNA-Seq study of durian cv. Musang King, including ACS, ACO, methionine gamma lyase (MGL), pectinesterase (PME40), S-adenosylmethionine (SAM) synthase, β-D-xylosidase 1 (BXL1), cytochrome P450 71B34 (CYP71B34), sulfur deficiency- induced 1 (SDI1), and DPNPH] and Khaksar et al. (2019) [L-tryptophan aminotransferase 1 (TAA1) and indole-3-pyruvate monooxygenase (YUCCA4)], along with alpha-amylase, beta-amylase, expansin, sulfate transporter, and polygalacturonase (PG), the conserved domain of each enzyme (based on an HMM) was first obtained from the Pfam protein database 62

The 29th Special CU-af Seminar 2021 August 25, 2021 (http://pfam.xfam.org/). This sequence was used as a query to search against the transcriptome database of durian fruit cv. Monthong and Phuangmanee and the Musang King genome (i.e., ACS (XM_022901720.1), ACO (XM_022903266.1), TAA1 (XM_022878297.1), YUCCA4 (XM_022900772.1), MGL (XM_022917834.1), PME40 (XM_022875865.1), SAM synthase (XM_022915017.1), BXL1 (XM_022866549.1), CYP71B34 (XM_022919875.1), SDI1 (XM_022914153), alpha-amylase (XM_022877573.1), beta-amylase (XM_022904686.1) expansin (XM_022886892), sulphate transporter (XM_022886687.1), and PG (XM_022892391.1)). The network of TFs and candidate target genes was visualized using Cytoscape (v3.7.1, USA). A correlation heatmap was generated using MetaboAnalyst 4.0. Dual-luciferase reporter assay To examine the binding activity of DzDof2.1 to the promoter of MGL, dual-luciferase assay was performed following the method described in Khaksar and Sirikantaramas (2020) with modifications. Briefly, genomic DNA was extracted from durian leaves using the cetyltrimethyl ammonium bromide (CTAB)-based method according to Abdel-Latif and Osman (2017). The 2000-bp promoter regions of MGL of durian; DzMGL (XM_022917834.1) was amplified and cloned into the pGreenII 0800-LUC double-reporter vector (Hellens et al., 2005). For generating the effector construct, the coding sequence of DzDof2.1 was cloned into the pGreen62-SK expression vector. The resulting effector and reporter plasmids were used for Arabidopsis protoplast transfection. After incubation, the protoplasts were harvested, frozen in liquid nitrogen and stored at –80 °C. The frozen protoplasts were re-suspended in 100 μL Passive lysis buffer (Promega, USA). Then, firefly (LUC) and Renilla (REN) luciferase activities were measured by a dual-luciferase reporter assay system (Promega, E1910) according to the manufacturer’s instructions on a Tecan Infinite 200 PRO microplate reader (TECAN) and the relative LUC/REN ratios were calculated. Three independent biological replicates were used. Statistical analyses All experiments were performed using three independent biological replicates. In the figures, data are plotted as means ± standard deviations. Statistical comparisons of the mean values were carried out using one-way ANOVA, followed by Duncan’s multiple range test or a Student’s t-test at the significance level of 0.05, using the statistical package for social sciences (SPSS) software, version 20. Results and Discussion Differential expression patterns of DzDofs during post-harvest ripening of the Monthong cultivar We examined the expression levels of 24 DzDofs at three different stages (unripe, midripe, and ripe) during post-harvest ripening of durian fruit cv. Monthong. Notably, we observed that 15 of the 24 DzDofs were expressed in the fruit pulp, which were: durian cyclic Dof1 (DzcDof1), DzcDof2, DzcDof3, DzDof1.4, DzDof1.5, DzDof2.1, DzDof2.2, DzDof2.4, DzDof2.5, DzDof3.1, DzDof4.6, DzDof5.1, DzDof5.3, DzDof5.4, and DzDof5.6 (Figure 1). This finding was similar to the fruit pulp-expressed DzDofs in Musang King cultivar (Khaksar et al., 2019). Notably, these 15 fruit pulp-expressed DzDofs showed different expression patterns during post-harvest ripening of durian fruit. Out of the 15 fruit pulp-expressed DzDofs, the expression patterns of nine varied over the course of post-harvest ripening. Six DzDofs, 63

The 29th Special CU-af Seminar 2021 August 25, 2021 including DzcDof1, DzcDof2, DzDof2.1, DzDof3.1, DzDof5.3, and DzDof5.4 were expressed at increasingly high levels as fruit ripening progressed in the Monthong cultivar. Moreover, three DzDofs, including DzDof1.4, DzDof2.4, and DzDof2.5 were expressed at decreasing levels over the course of fruit ripening (Figure 1). Taken together, these nine DzDofs showed ripening-associated expression patterns, suggesting a role during post-harvest ripening of durian fruit. In contrast, the expression of DzcDof3, DzDof1.5, DzDof2.2, DzDof4.6, DzDof5.1, and DzDof5.6 did not vary significantly during post-harvest ripening, and thus these were not considered to be ripening-associated TFs. Notably, in our previous study, we examined the expression levels of 24 DzDofs at three different stages (unripe, midripe, and ripe) during post-harvest ripening of durian fruit cv. Monthong using reverse transcription quantitative polymerase chain reaction (RT-qPCR) (Khaksar et al., 2019). The transcript accumulation patterns of most DzDofs observed herein are consistent with the data obtained through RT-qPCR. Among the ripening-associated DzDofs, DzcDof2 harboured the greatest fold change during post-harvest ripening (at midripe and ripe stages) and was selected as the candidate ripening-associated DzDof for further analysis. In addition, taking into account the transcript level during ripening, DzDof2.1 showed the highest expression level at the ripe stage. Therefore, this Dof was also considered as a candidate one. Figure 1: Fold changes in expression levels of fruit pulp-expressed DzDofs at three different stages (unripe, midripe, and ripe) during post-harvest ripening of durian fruit (Monthong cultivar). The relative transcript levels (RPKM values) were normalized by the unripe stage as control. Three independent biological replicates were used. Bars with different letters show significant differences (P < 0.05). 64

The 29th Special CU-af Seminar 2021 August 25, 2021 Differential expression levels of fruit pulp-expressed DzDofs during post-harvest ripening of the Phuangmanee cultivar To examine the expression levels of DzDofs in another cultivar with different ripening behaviour and sensory characteristics, their expression levels were also measured during post-harvest ripening in the Phuangmanee cultivar. Of particular note, we observed that 15 of the 24 DzDofs were expressed in the fruit pulp (Figure 2), similar to our observation in Monthong. However, we observed some differences in the expression levels of some fruit-expressed DzDofs. Out of the 15 fruit pulp-expressed DzDofs, the expression patterns of 12 varied over the course of post-harvest ripening. Nine DzDofs, including DzcDof1, DzcDof2, DzcDof3, DzDof2.1, DzDof3.1, DzcDof4.6, DzcDof5.3, DzDof5.4, and DzDof5.6 were expressed increasingly whereas three DzDofs, including DzDof1.4, DzDof2.4, and DzDof2.5 were expressed decreasingly over the course of fruit ripening (Figure 2). Figure 2: Fold changes in expression levels of fruit pulp-expressed DzDofs at three different stages (unripe, midripe, and ripe) during post-harvest ripening of durian fruit (Phuangmanee cultivar). The relative transcript levels (RPKM values) were normalized by the unripe stage as control. Three independent biological replicates were used. Bars with different letters show significant differences (P < 0.05). 65

The 29th Special CU-af Seminar 2021 August 25, 2021 The differences in the expression patterns and/or levels of some DzDofs between the two cultivars might be responsible for the different ripening characteristic of Phuangmanee and Monthong. Similarly, we observed some differences in the expression levels of some DzDofs between Monthong and Chanee cultivars (Khaksar et al., 2019). Comparisons of the expression levels of fruit pulp-expressed DzDofs between those cultivars revealed a total of 10 potential cultivar-dependent Dofs, among which, the expression level of DzDof2.2 underwent a significantly greater fold change at each ripening stage in the Chanee than in the Monthong cultivar. Accordingly, Khaksar et al. (2019) suggested that higher expression levels of DzDof2.2 in Chanee could enhance its auxin levels during ripening through transcriptional regulation of auxin biosynthetic genes. Higher auxin levels in Chanee would upregulate ethylene biosynthesis by transcriptionally activating the ACC synthase gene family, and lead to an earlier ethylene response (auxin-ethylene crosstalk), and thus faster post-harvest ripening compared to that in the Monthong cultivar. Similar to our finding in Monthong, DzcDof2 was the candidate ripening-associated Dof with the highest fold increase during post-harvest ripening of durian fruit cv. Phuangmanee. We also compared the expression levels of DzcDof2 between the two cultivars and found that DzcDof2 harboured a significantly higher transcript level (RPKM value) during ripening (at midripe and ripe stages) in Phuangmanee compared to that of the Monthong cultivar (Figure 3). This observation suggested that DzcDof2 could also act as a cultivar-dependent transcription factor. Taken together, the marked ripening-associated expression pattern of DzcDof2 during ripening of durian fruit cv. Monthong and Phuangmanee and its potential cultivar-dependent role prompted its further analysis. Notably, according to the phylogenetic analysis of DzcDof2, a member of the Dof TF family from grape (VvDof8) and from tomato (SlDof3) were found to be the closest orthologues of DzcDof2 which were clustered together in subclade A1 (Khaksar et al., 2019). Characterized orthologue identification is widely used as a strong tool for predicting the function of genes (Chen and Jeong, 2000). However, to the best of our knowledge, there are no published studies of the functional characterization of these orthologues. Further studies regarding the functional characterization of our candidate DzcDof2 could provide more insights into its role during fruit ripening. 66

The 29th Special CU-af Seminar 2021 August 25, 2021 Figure 3: Comparison of the expression level of our candidate DzcDof2 between the Monthong and Phuangmanee cultivars. Transcript levels of DzcDof2 during ripening were compared at three different stages (unripe, midripe, and ripe) between the Monthong and Phuangmanee cultivars. A comparison was made between the two cultivars at each stage. Three independent biological replicates were used. An asterisk (*) above the bars indicates a significant difference between the two cultivars (Student’s t-test, P < 0.05). Ethylene plays a major role in climacteric fruit ripening. In our previous study, we validated the ripening-associated role of candidate DzDofs, including DzcDof2 and DzDof2.1 through profiling their expression levels under three different ripening treatments, including natural, ethephon-induced, and 1-methylcyclopropene (1-MCP)-delayed ripening. Notably, the transcript accumulation of both Dofs significantly increased under ethephon treatment and dramatically decreased with 1-MCP relative to that in the control (natural ripening) (Khaksar et al., 2019). These ethylene-inducible DzDofs were expressed increasingly during ripening, and thus could act as either putative transcriptional activators or repressors of ripening. Zhang et al. (2018) identified an ethylene-induced Dof TF in kiwifruit (AdDof3) that acted as a ripening-activator TF and enhanced starch degradation during fruit ripening. On the other hand, an ethylene-induced banana Dof (MaDof23), which was expressed at increasing levels during ripening, was a transcriptional repressor. This Dof acted antagonistically to a ripening-activator ethylene response factor (MaERF9) to regulate banana fruit ripening (Feng et al., 2016). Regulatory effects of ripening-associated and cultivar-dependent DzcDof2 on some target ripening-related genes Gene expression correlations of DzcDof2 with some previously identified ripening- related genes in durian fruit (SDI1, DPNPH, and sulfate transporter: sulfur metabolism; SAM synthase, ACS, and ACO: ethylene biosynthesis; MGL and expansin: aroma formation; PME40, BXL1, and PG: cell wall modification; CYP71B34: fruit ripening; TAA1 and YUCCA4: auxin biosynthesis; alpha- and beta-amylase: starch degradation) were investigated and visualized as a clustered heatmap (Figure 4A) and a correlation network (Figure 4B). As revealed by hierarchical clustering of Pearson’s correlations, DzcDof2 and SAM synthase, ACS, ACO, TAA1, YUCCA4, beta-amylase, SDI1, and MGL were clustered together and were positively correlated (Figure 4A). Notably, as shown in Figure 4B, the highest positive correlation was observed between DzcDof2 and ethylene biosynthetic genes (SAM synthase, ACS, and ACO), followed by DzcDof2 and beta-amylase, and DzcDof2 and auxin biosynthetic genes (TAA1 and YUCCA4). However, DzcDof2 was negatively correlated with CYP71B34, alpha-amylase, BXL1, PME40, expansin, sulfate transporter, PG, and DPNPH (Figure 4A). The highest negative correlation was found between DzcDof2 and alpha-amylase (Figure 4B). 67

The 29th Special CU-af Seminar 2021 August 25, 2021 Figure 4: Gene expression correlation of ripening-associated and cultivar-dependent DzcDof2. (A) Heatmap of hierarchical clustering of Pearson’s correlations (R) for DzcDof2 and previously identified ripening-related genes. Genes with a normalized expression level (RPKM) > 1 were log2 transformed before analysis and were designated as expressed. The DzcDof2 and its positively correlated genes are highlighted with a red frame. The genes which are negatively correlated to the DzcDof2 are highlighted with a blue frame. (B) Correlation network analysis of DzcDof2 and previously identified ripening-related genes. The thickness of the line corresponds to the correlation strength. Red lines represent positive correlations, whereas blue lines indicate negative correlations. Our findings herein suggested a potential ripening-associated role of DzcDof2 during post-harvest ripening of durian fruit mainly via transcriptional regulation of ethylene and auxin biosynthesis. In addition, it might possibly regulate starch degradation during fruit ripening. Consistently, our in silico analysis of the 2-kb promoter region located upstream of the translation start site of beta-amylase of durian revealed the existence of cis-regulatory binding sites (AAAG/ CTTT) for Dof TFs (Figure 5). The possible role of a member of Dof TF family in regulating kiwifruit ripening via starch degradation has been previously documented (Zhang et al., 2018). Fruit ripening, a genetically programmed and coordinated developmental process, is controlled by a hormonal and transcriptional regulatory network involved in starch degradation, cell wall metabolism, and hormone metabolism. Starch degradation is considered as a critical step for both fruit initial ripening and the fruit flavor attributed to the soluble sugars. Further studies to confirm the potential role of DzcDof2 in transcriptional regulation of starch degradation during durian fruit ripening is of utmost importance. 68

The 29th Special CU-af Seminar 2021 August 25, 2021 Figure 5: Nucleotide sequence of the 2-kb promoter region of durian beta-amylase. Dof binding sites (AAAG and CTTT) are highlighted in yellow. The translational start site (ATG) is underlined. 69

The 29th Special CU-af Seminar 2021 August 25, 2021 Transcriptional activity of DzDof2.1 DzDof2.1 harboured the highest expression level at the ripe stage and was selected as a candidate ripening-associated Dof to evaluate its binding activity to the promoter of MGL, a key gene involved in aroma formation during durian fruit ripening. Accordingly, dual-luciferase reporter assay was performed to examine the in vivo transcriptional activity of DzDof2.1. As shown in Figure. 6, the co-expression of DzDof2.1 with the promoter of MGL significantly enhanced the relative LUC/REN ratio when compared to the control (empty pGreen62-SK), suggesting that DzDof2.1 binds to the promoter of MGL and transcriptionally activates it. Figure 6: In vivo transcriptional activity of DzDof2.1 in Arabidopsis protoplasts. Firefly luciferase (LUC) and Renilla luciferase (REN) activities were measured and presented as relative LUC/REN ratio. The LUC/REN ratio in the control (empty pGreen62-SK) was set as 1 and was used as a calibrator. Error bars represent means ± standard deviations (SD) among four independent biological replicates. An asterisk (*) above the bars indicates a significant difference compared to the control (Student’s t-test, P < 0.05). The distinctive odor of durian fruit has been biochemically investigated and characterized as odor-active compounds including volatile sulfur compounds (VSCs) with MGL as a key gene involved in VSC production (Teh et al., 2017). Our finding strongly suggests a potential role of DzDof2.1 in regulating post-harvest ripening of durian fruit via transcriptional regulation of MGL. Notably, a positive correlation was observed between DzcDof2 and MGL (Figure 4B), suggesting a potential role of DzcDof2 in transcriptional regulation of MGL. Often, different TFs regulate the expression of a particular gene through the formation of enhanceosome or repressosome complexes (Martinez and Rao, 2012). The possible interaction between DzcDof2 and DzDof2.1 in transcriptional regulation of MGL during fruit ripening in durian could be the subject of a further investigation. 70

The 29th Special CU-af Seminar 2021 August 25, 2021 Conclusion In conclusion, transcriptome-wide identification revealed 24 members of the Dof family in durian. Among these, 15 were expressed in the fruit pulps of durian cv. Monthong and Phuangmanee, out of which some DzDofs were identified as being ripening-associated for both cultivars. The marked ripening-associated expression patterns of DzcDof2 and its strong correlation with ethylene and auxin biosynthetic genes and a starch degrading gene suggest its potential role during post-harvest ripening of durian fruit via transcriptional regulation of ethylene and auxin biosynthesis and starch degradation. Our findings obtained herein should provide an important foundation for further studies regarding the functional characterization of ripening-associated and cultivar-dependent DzDofs. References 1. Abdel-Latif, A., and Osman, G. (2017). Comparison of three genomic DNA extraction methods to obtain high DNA quality from maize. Plant Methods. 13: 1. 2. Arnone, M.I., and Davidson, E.H. (1997). The hardwiring of development: organization and function of genomic regulatory systems. Development. 124, 1851-1864. 3. Cai, X., Zhang, Y., Zhang, C., Zhang, T., Hu, T., Ye, J., Zhang, J., Wang, T., Li, H., and Ye, Z. (2013). Genome-wide analysis of plant-specific Dof transcription factor family in tomato. J. Integr. Plant Biol. 55, 552–566. 4. Chen, R., and Jeong, S. (2000). Functional prediction: Identification of protein orthologs and paralogs. Protein Sci. 9, 2344–2353. 5. Chin, S.T., Nazimah, S.A.H., Quek, S.Y., Che Man, Y.B., Abdul Rahman, R., and Mat Hashim, D. (2007). Analysis of volatile compounds from Malaysian durians (Durio zibethinus) using headspace SPME coupled to fast GC–MS. J. Food Compost. Anal. 20, 31–44. 6. Feng, B., Han, Y., Xiao, Y., Kuang, J., Fan, Z., Chen, J., and Lu, W. (2016). The banana fruit Dof transcription factor MaDof23 acts as a repressor and interacts with MaERF9 in regulating ripening-related genes. J. Exp Bot. 67, 2263–2275. 7. Hellens, R., Allan, A., Friel, E., Bolitho, K., Grafton, K., Templeton, M., et al. (2005). Transient expression vectors for functional genomics, quantification of promoter activity and RNA silencing in plants. Plant Methods. 1: 13. 8. Jaswir, I., Che Man, Y.B., Selamat, J., Ahmad, F., and Sugisawa, H. (2008). Retention of volatile components of durian fruit leather during processing and storage. J. Food Process. Pres. 32, 740–750. 9. Khaksar, G., Sangchay, W., Pinsorn, P., Sangpong, L., and Sirikantaramas, S. (2019). Genome-wide analysis of the Dof gene family in durian reveals fruit ripening-associated and cultivar-dependent Dof transcription factors. Sci rep. 9: 12109. 10. Khaksar, G., and Sirikantaramas, S. (2020). Auxin response factor 2A is part of the regulatory network mediating fruit ripening through auxin-ethylene crosstalk in durian. Front. Plant Sci. 11: 543747. 11. Lehmann, M., Schwarzlander, M., Obata, T., Sirikantaramas, S., Burow, M., Olsen, C. E., Tohge, T., Fricker, M. D., Moller, B. L., Fernie, A. R., Sweetlove, L. J., and Laxa, M. (2009). The metabolic response of Arabidopsis roots to oxidative stress is distinct from that of heterotrophic cells in culture and highlights a complex relationship between the levels of transcripts, metabolites, and flux. Mol. Plant. 2, 390–406. 12. Lijavetzky, D., Carbonero, P., and Vicente-Carbajosa, J. (2003). Genome-wide comparative phylogenetic analysis of the rice and Arabidopsis Dof gene families. BMC Evol. Biol. 3, 17. 71

The 29th Special CU-af Seminar 2021 August 25, 2021 13. Liu, K., Yuan, C., Feng, S., Zhong, S., Li, H., Zhong, J., Shen, C., and Liu, J. (2017). Genome-wide analysis and characterization of Aux/IAA family genes related to fruit ripening in papaya (Carica papaya L.). BMC Genomics. 18, 351. 14. Ma, N., Feng, H., Meng, X., Li, D., Yang, D., Wu, C., et al. (2014). Overexpression of tomato SlNAC1 transcription fac- tor alters fruit pigmentation and softening. BMC Plant Biol. 14, 351. 15. Ma, J., Li, M. Y., Wang, F., Tang, J., and Xiong, A.S. (2015). Genome-wide analysis of Dof family transcription factors and their responses to abiotic stresses in Chinese cabbage. BMC Genomics 16, 33. 16. Martinez, G. J., and Rao, A. (2012). Immunology. Cooperative transcription factor complexes in control. Science. 338, 891–892. 17. Mayer, F., Takeoka, G.R., Buttery, R.G., Whitehand, L.C., Naim, M., and Rabinowitch, H.D. (2008). Studies on the aroma of five fresh tomato cultivars and the precursors of cis- and trans-4, 5-epoxy-(E)-2-decenals and methional. J. Agric. Food Chem. 56, 3749–57. 18. Medina-Puche, L., Cumplido-Laso, G., Amil-Ruiz, F., Hoffmann, T., Ring, L., Rodriguez-Franco, A., et al. (2014). MYB10 plays a major role in the regulation of flavonoid/phenylpropanoid metabolism during ripening of Fragaria x ananassa fruits. J Exp Bot. 65, 401–417. 19. Nicolas, P., Lecourieux, D., Kappel, C., Cluzet, S., Cramer, G., Delrot, S., et al. (2014). The basic leucine zipper transcription factor ABSCISIC ACID RESPONSE ELEMENT-BINDING FACTOR2 is an important transcriptional regulator of abscisic acid-dependent grape berry ripening processes. Plant Physiol. 164, 365–383. 20. Park, M., Park, S., Cho, S., and Kim, K. (2009). Nicotiana benthamiana protein, NbPCIP1, interacting with Potato virus X coat protein plays a role as susceptible factor for viral infection. Virology. 386, 257–269. 21. Sangpong, L., Khaksar, G., Pinsorn, P., Oikawa, A., Sasaki, R., Erban, A., Watanabe, M., Wangpaiboon, K., Tohge, T., Kopka, J., Hoefgen, R., Saito, K., and Sirikantaramas, S. (2021). Assessing Dynamic Changes of Taste-Related Primary Metabolism During Ripening of Durian Pulp Using Metabolomic and Transcriptomic Analyses. Front. Plant Sci. 12: 687799. 22. Shan, W., Kuang, J.F., Chen, L., Xie, H., Peng, H.H., Xiao, Y.Y., et al. (2012). Molecular characterization of banana NAC transcription factors and their interactions with ethylene signaling component EIL during fruit ripening. J Exp Bot. 63, 5171–87. 23. Shen, X., Zhao, K., Liu, L., Zhang, K., Yuan, H., Liao, X., et al. (2014). A role for PacMYBA in ABA-regulated anthocyanin biosynthesis in red-colored sweet cherry cv. Hong Deng (Prunus avium L.). Plant Cell Physiol. 55, 862–80. 24. Teh, B.T., Lim, K., Yong, C.H., Young Ng, C.C., Ramesh Rao, S., Rajasegaran, V., et al. (2017). The draft genome of tropical fruit durian (Durio zibethinus). Nat. Genet. 49, 1633–1644. 25. Trainotti, L., Tadiello, A., and Casadoro, G. (2007). The involvement of auxin in the ripening of climacteric fruits comes of age: the hormone plays a role of its own and has an intense interplay with ethylene in ripening peaches. J. Exp Bot, 3299–3308. 26. Wen, C.L., Cheng, Q., Zhao, L., Mao, A., Yang, J., Yu, S., et al. (2016). Identification and characterization of Dof transcription factors in the cucumber genome. Sci. Rep. 6, 23072. 27. Wu, Z., Cheng, J., Cui, J., Xu, X., Liang, G., Luo, X., Chen, X., Tang, X., Hu, K., and Qin, C. (2016). Genome-Wide Identification and Expression Profile of Dof, Transcription Factor Gene Family in Pepper (Capsicum annuum L.). Front. Plant Sci. 7, 574. 28. Yanagisawa, S. (2002). The Dof family of plant transcription factors. Trends. Plant Sci. 7, 555–560. 72

The 29th Special CU-af Seminar 2021 August 25, 2021 29. Zhang, A., Wang, W., Tong, Y., Li, M., Grierson, D., Ferguson, I., Chen, K., and Yin, X. (2018). Transcriptome analysis identifies a zinc finger protein regulating starch degradation in kiwifruit. Plant. Physiol. 178, 850–863. 73

Development of a Yeast-Based Assay and Screening for Compounds that can alleviate the Toxicity of Human Alpha-Synuclein, a Neurodegenerative Disease Associated Protein Anyaporn SANGKAEW Thanaporn KOJORNNA Asamaporn KLINKAJORN Ryoya TANAHASHI Apichart SUKSAMRARN Hiroshi TAKAGI and Chulee YOMPAKDEE

The 29th Special CU-af Seminar 2021 August 25, 2021 Development of a Yeast-Based Assay and Screening for Compounds that can alleviate the Toxicity of Human Alpha-Synuclein, a Neurodegenerative Disease Associated Protein Anyaporn SANGKAEW1*, Thanaporn KOJORNNA1*, Asamaporn KLINKAJORN1*, Ryoya TANAHASHI2*, Apichart SUKSAMRARN3*, Hiroshi TAKAGI2* and Chulee YOMPAKDEE1* Abstract Alpha-synuclein (α-syn) aggregation, is a hallmark and a therapeutic target in Parkinson’s disease (PD). Since current therapeutics has not yet been successful, it is necessary to develop a novel therapeutic strategy that either prevent or delay the disease progression. Inhibition of α-syn aggregation including promoting ubiquitin-mediated degradation of α-syn is one of effective disease modifications. The ubiquitin ligase Rsp5 is involved in the degradation of abnormal or unfavorable proteins in Saccharomyces cerevisiae. This study aimed to establish a yeast-based assay and screen for compounds alleviating α-syn toxicity. We established a powerful yeast-based assay using rsp5A401E mutant, which is hypersensitive to α-syn aggregation, and further overexpressed GFP-fused α-syn leading to growth inhibition of the resultant strain. Upon screening using the developed yeast-based assay, ASCY130, an active compound isolated from Stephania suberosa Forman, was the most potent compound with minimal effective concentration at 1.56 mg/ml. It may be a potential lead compound for the design of therapeutic agent for PD treatment. 1Department of Microbiology, Faculty of Science, Chulalongkorn University, Bangkok, Thailand. 2Division of Biological Science, Graduate School of Science and Technology, Nara Institute of Science and Technology, Nara, Japan 3Department of Microbiology, Faculty of Science, Ramkamhaeng University, Bangkok, Thailand 75

The 29th Special CU-af Seminar 2021 August 25, 2021 Introduction and Objective Parkinson’s disease (PD) is an age-related neurodegenerative disorder. PD is characterized by progressive and selective loss of dopaminergic neurons in the substantia nigra that contributes to the cardinal motor symptoms of the disease: bradykinesia, rigidity, resting tremor and postural instability[1]. PD can sometimes be associated with depression and withdrawal from participation in social activity[2]. Nowadays, there are no disease-modifying treatments for PD. Medical management is predominantly focused on controlling the motor symptoms using drugs, mostly using dopamine replacement by administration of dopamine precursor (L-DOPA), combined with carbidopa, a L-DOPA decarboxylase inhibitor or catechol-O-methyl transferase inhibitor[3]. However, the long-term duration of disease leads to sophisticated medication regimes aimed at controlling the motor symptoms in patients, with a likelihood of problematic side effects[4]. Therefore, there is a great requirement to develop a new drug for the management of PD, which are not only therapeutic but can also prevent the initiation or delay or stop progression of the disease. Neuropathological hallmark of PD is the accumulation of misfolded and aggregated α-synuclein (α-syn) that causes the formation of proteinaceous cytoplasmic inclusions termed Lewy-bodies and Lewy neurites (LBs/LNs)[5, 6]. Human α-synuclein (α-syn) is an intrinsically unfolded protein consisting of 140 amino acids that has the propensity to self-assemble and to form oligomeric protofibrils that can further mature into different type of fibers and aggregates[7, 8]. Although the normal function of α-syn in neuron is still unknown, it is believed that α-syn plays a role in synaptic transmission, specifically in the recycling of synaptic vesicles[9]. Multiple lines of evidence suggested that α-syn is involved in the initiation and progression of neurodegeneration in PD pathogenesis[6, .10-12] Therefore, the modulation of α-syn aggregation oligomerization, fibrillation and propagation to reduce its toxicity emerged as an important therapeutic target for slowing or halting disease progression[5, 13]. In human, most cellular proteins are selectively targeted for degradation after conjugation to ubiquitin by the E3 ubiquitin-protein ligase Nedd4. The overexpression of Nedd4 enhanced α-syn ubiquitination, leading to α-syn clearance. Therefore, the mechanism of α-syn aggregate clearance is a central question in understanding the PD pathology[14, 15]. In this way, the agents stabilizing, promoting clearance, degrading misfolded proteins, solubilizing oligomers or inhibiting the propagation of α-syn aggregates are pharmacologically appropriate and a clinically relevant therapeutic strategy for PD[16, 17]. For drug discovery, the budding yeast Saccharomyces cerevisiae is one of the organisms used. S. cerevisiae can be highly useful in the first-line screening of potential active compounds[18]. Previous researches suggested that α-syn-related effects, such as proteasome impairment, vesicle trafficking dysfunction and reactive oxygen species generation, can be efficiently mimicked in S. cerevisiae cells, which do not endogenously express the α-syn ortholog. Additionally, the heterologous expression of human α-syn results in intracellular α-syn accumulation and growth defect in the α-syn dose-dependent manner[19-21]. These have made S. cerevisiae as a valuable cell model to study physiological and pathological features of α-syn and to screen active compounds alleviating α-syn toxicity. Wijayanti, Watanabe[22] showed that the overexpression of α-syn in S. cerevisiae cells lead to growth inhibition, especially in the rsp5A401E mutant, indicating that the rsp5A401E mutant is also hypersensitive to α-syn accumulation. Therefore, biological/chemical agents or compounds that could complement the stress sensitivity of the rsp5A401E mutant might be promising as drug candidates. 76

The 29th Special CU-af Seminar 2021 August 25, 2021 In this work, we developed a novel yeast-based assay system to search for molecules that could alleviate human α-syn-induced cytotoxicity of the multidrug-sensitive strain with rsp5A401E background. Yeast-integrating plasmid carrying the rsp5A401E gene was introduced into the drug sensitive strain and then the sensitivity to various stresses, such as high temperature, L-azetidine-2-carboxylic acid (AZC) and α-syn, of the resultant strains was observed. Subsequently, the yeast-based assay system with resazurin was optimized to be used as a high-throughput platform and was further validated using two known compounds, baicalein and ampicillin, for sensitivity and specificity. Methods The Yeast strains used in this study (Table 1) Table 1: List of yeast strains used in this study Construction of yeast rsp5A401E in BY25929 background overexpressing human α-syn To construct strain TK01 (the rsp5A401E mutant in BY25929 background; trp1-1 leu2-3, 112 his3-11, 15 ura3-1 ade2-1 can1-100 yrs1::HIS3 yrr1::TRP1 pdr1::hisG pdr3::hisG), the yeast integrating plasmid pRS406 (URA3) was used for two-step replacements of the wild-type RSP5 with rsp5A401E mutant allele[25]. For overexpression of human α-syn, a galactose-inducible multicopy plasmid pYES2-α-syn GFP[22] was transformed with lithium acetate method[26]. Determination of yeast rsp5A401E phenotypes For assay of AZC sensitivity, the cell suspensions were spotted onto SD medium supplemented with 1mM AZC, and the plates were incubated at 30 ⁰C for 2-3 d. In parallel, high temperature sensitivity of yeast cells was tested on YPD medium. After spotting cell suspensions, the plates were incubated at 37 ⁰C for 30 h and further incubated at 30 ⁰C for 2-3 d. The sensitivity against α-syn was determined using yeast cells overexpressing α-syn. Cell suspensions were spotted onto SG-Ura and SC-Ura for α-syn overexpression and α-syn shutoff, respectively. After 2-3 d incubation at 30 ⁰C, the plates were photographed. 77

The 29th Special CU-af Seminar 2021 August 25, 2021 Setting-up resazurin yeast-based cell viability assay Resazurin yeast-based assay was performed using black polystyrene, non-tissue cultured treated, sterile, low binding 96-well microtiter plate (Thermo Scientific, USA) and fluorescent signal of resorufin was measured using Ensight Multimode Microplate reader (PerkinElmer, USA). Excited wavelength was set on 540 nm with 590 nm as emission wavelength. The experimental parameters such as resazurin concentration, initial cell densities and length of incubation of yeast indicator strain with resazurin were optimized to determine cell viability of the yeast strain. The resazurin reduction test was performed. Briefly, 50 µL of yeast cultured SR-Ura was added to wells of 96-well black microtiter plate. Subsequently, 50 µL of inducible SG-Ura medium supplemented with or without test compound was added to each well for a final volume of 100 µL. The assay plate was incubated at 30 ⁰C for 24 h. To determine cell viability, 10 µL of resazurin at desired concentration was added to each well and measured in relative fluorescence unit (RFU) with top scanning mode. RFU was recorded immediately after the resazurin dosing to all wells, and then again in 30 min period until RFU in well without test compound decreased. Two known compounds, baicalein and ampicillin, were used to validate our system. The minimal effective concentration (MEC) was defined as the lowest concentration of test compound that could alleviate human α-syn toxicity and so accelerated the resazurin metabolism. Screening of compounds alleviating human α-syn toxicity For screening of compounds alleviating α-syn toxicity, 32 natural compounds isolated from Thai medicinal plants were obtained from Laboratory of Prof. Dr. Apichart Suksamrarn, Department of Chemistry, Ramkhamhaeng University. Test compounds were prepared in 100% DMSO and further dissolved in liquid medium to desired final concentrations. The screening was performed as described in resazurin yeast-based cell viability assay. The results were expressed as percentage of α-syn toxicity reduction as follows: • % Reduction of toxicity = (RFU/min (treated) – RFU/min (negative control)) x100 • Treated: Test compound treated TK01(pYES2-a-syn-GFP) cells • Negative control: Only solvent treated TK01(pYES2-a-syn-GFP) cells Results and Discussion Effect of AZC and high temperature on the growth phenotype of the rsp5A401E mutant yeast In the previous study, the rsp5A401E mutant strain showed hypersensitivity to various stresses, such as high temperature and toxic amino acid analogues AZC. To enhance sensitivity to test compounds, the yeast strain BY25929 (obtained from the Yeast Genetic Resources Center, Japan) in which multidrug resistant genes were removed was used for replacing rsp5A401E mutant allele. The resultant strain was designated as TK01 strain (BY25929 rsp5A401E). To examine the growth phenotype, both BY25929 and TK01 strains were grown in test conditions. For the sensitivity to AZC, high growth sensitive to 1 mM AZC on assay medium was clearly observed in the TK01 strain in comparison with the wild type RSP5 BY25929 strain (Figure 1A). For high temperature sensitivity, the result showed that TK01 strain showed high temperature growth defect compared to the yeast RSP5 wild-type strain (Figure 1B). Both of growth phenotypes observed indicated the characteristics of rsp5A401E mutant yeast[27]. 78

The 29th Special CU-af Seminar 2021 August 25, 2021 Figure 1: Effect of the A401E variant Rsp5 on yeast BY25929 growth phenotypes. Serial dilutions of yeast strain were spotted and incubated on SD+Ade+Ura +Leu medium containing 1 mM AZC at 30 ⁰C for 2 d (A) and on YPD medium at following conditions, at 30 ⁰C for 2 d or at 37 ⁰C for 30 h and then at 30 ⁰C for 2 d. Effect of α-syn on the growth phenotype of rsp5A401E mutant yeast In order to evaluate whether the rsp5A401E mutants are more sensitive to α-syn than RSP5 wild-types, the mutant TK01 including the BY25929 strain were transformed with either empty vector pYES2 or pYES2-α-syn GFP. The transformants were grown onto either SC-Ura (no α-syn expression) or SG-Ura (induction of α-syn expression). Under induced expression of α-syn condition as shown in Figure 2 (right panel), it demonstrated that the overexpression of α-syn led to a severe growth defect. As expected, rsp5A401E cells showed more sensitivity to α-syn than wild-type cells. Figure 2: Effect of α-syn expression on the growth of yeast strains. Cell suspension of 5-fold serial dilutions of the yeast transformants harboring plasmid either empty vector or α-syn- GFP were spotted onto selective medium containing 2% glucose or 2% galactose, and the plates were incubated at 30 ⁰C for 2 d. Setting up the resazurin yeast-based assay to search for compounds alleviating human α-syn toxicity Metabolic activity may differ among yeast strains, thus optimizing the concentration of resazurin is important. TK01(pYES2-α-syn GFP) were plated in 96 well-plates at various cell concentrations. Different concentrations of resazurin were added, starting from 0.05 mM to 0.4 mM. Figure 3 showed that the resazurin reduction was positively proportional to the resazurin concentration (from 0.05 mM until 0.2 mM). However, the resazurin reduction at 0.4 mM showed significant decrease compared to 0.2 mM. This data indicated that the optimal resazurin concentration of these test strains was 0.2 mM. 79

The 29th Special CU-af Seminar 2021 August 25, 2021 Figure 3: Effects of resazurin and cell concentrations on fluorescence intensity in TK01 (pYES2 α-syn GFP). Various cell concentrations of yeast strains were added into wells of a 96-well plate. Different resazurin concentrations were then added, ranging from 0.05 to 0.4 mM. After 90 min incubation, fluorescence intensity was measured and expressed as relative fluorescence unit (RFU). The values are means ± standard deviations (SD) of results from three independent experiments. Statistical significances are indicated by the P-value (***P ˂ 0.001, 0.2 mM vs. 0.05, 0.1, 0.4 mM). Resazurin is an oxidation-reduction dye that turns into a fluorescent compound (resorufin) after a reduction process caused by viable cell, and further reduced to nonfluorescent compound (hydroresorufin)[28, 29]. Because resazurin reduction is affected by time and cell concentration, it is important to determine the initial cell concentration and resazurin incubation time. The TK01strains carrying either pYES2-GFP or pYES2-α-syn GFP were cultured in SG-Ura at different initial cell concentrations for 16-18 h. Resazurin solution was added according to the optimal concentration obtained. The kinetic of resazurin transformation to resorufin in cell suspension was determined. In this experiment, TK01(pYES2- α-syn GFP) was used as assay strains to determine the effect of compound on alleviating α-syn toxicity. As yeast strain harboring empty vector, which showed faster metabolic kinetic than those of the assay strains, because of no cytotoxicity from α-syn accumulation, was served as positive growth control. As shown in Figure 4A, initial cell concentrations of the TK01 background strains between 5x105 and 1x106 cells mL-1 were almost immediately metabolized the resazurin to resorufin and hydroresorufin with a steep increase and decrease in fluorescence intensity, respectively. While in lower cell concentrations started to increase the amount of resorufin with a slower metabolic kinetic however, starting cell concentration of less than 5x104 cells mL-1 display significant lags in growth. While starting cell concentration between 5x104 to 1x105 cells mL-1 yielded significant increase in RFU over time as compared to those other cell concentrations, thus were chosen as the optimal range of cell concentration. Our results showed that resazurin reduction rate was accelerated with increasing amount of the yeast strain. When the amount of resorufin reached its maximum and entered a stationary phase, fluorescence became stable for a certain time-period, while in the final part of each curve RFU values decreased. The decline of the curve was probably attributed to a secondary reduction of resorufin to colorless and non-fluorescent product hydro-resorufin[30]. According to the optimal starting cell concentrations chosen, the incubation time with resazurin should be monitored at least up to 240 min. 80

The 29th Special CU-af Seminar 2021 August 25, 2021 Figure 4: Kinetics of resazurin metabolization of different strains in varying cell concentrations. Strains TK01 (pYES2-α-syn GFP) (A), and TK01 (pYES2-GFP) (B) at varying initiate cell concentrations were grown in SG-Ura at 30 oC for 16-18 h. The kinetics were determined immediately after adding resazurin at every 30 min for 300 min. The relationship between RFU values and times was plotted. The values are means ± SD of replicate cultures (n = 3) for each cell concentration tested. Validation of the resazurin yeast-based assay system Usually test compounds were dissolved in 100% DMSO solvent, therefore DMSO concentration must be optimized so that no toxic harmful effect to the yeast assay strains. Our optimization was demonstrated that final concentration of DMSO at 0.5% v/v had not the cytotoxicity (data not shown). To determine the reliability of this resazurin yeast-based assay for drug screening, we determined the effect of baicalein, a known α-syn oligomer inhibiting compound[31]. Varying concentrations of baicalein (25 – 0.04 µM) was tested. The strain TK01 (pYES2- α-syn GFP) treated with 5 µM baicalein showed a significant increase (P <0.001) in RFU values, as compared to that of the untreated control (0 µM) (Figure 5C). However, the RFU values from cells treated with lower concentrations (< 5 µM) of baicalein did not exhibit statistically different (P >0.05) (Figure 5C). On the other hand, the effect of baicalein could not be observed in HT01 (pYES2- α-syn GFP) strain (Figure 5A). The strain harboring empty vector treated with any concentration of baicalein served as indicator cells for cytotoxicity of the compound at concentration tested. At 25 µM, baicalein caused toxicity to all tested strains (Figure 5A-D). These results demonstrated that TK01(pYES2- α-syn GFP) showed more sensitive to baicalein than HT01(pYES2- α-syn GFP). Therefore, the more sensitive yeast strain, TK01(pYES2- α-syn GFP), was chosen to further determine specificity of the assay. Irrelevant compound, ampicillin, a β-lactam antibiotic that inhibit bacterial cell wall synthesis[32], was also used to test for specificity of the assay system. The TK01 carrying either pYES2- α-syn GFP or empty vector were incubated in assay medium containing ampicillin at concentration ranging from 1.56 – 25 mM for 24 h. The results showed that the ampicillin treatments did not exhibit any statistically significant difference (P >0.05) from that of the untreated control, indicating no role on alleviating α-syn toxicity of ampicillin. However, at high concentrations of ampicillin showed cytotoxicity with significant lower RFU signal (Figure 5E-F). These results assured the specificity of the developed assay system. 81

The 29th Special CU-af Seminar 2021 August 25, 2021 Figure 5: Specificity of resazurin yeast-based assay using baicalein, a known compound inhibiting α-syn oligomer and irrelevance compound, ampicillin. Strains HT01 (pYES2 α-syn GFP) (A), HT01 (pYES2-GFP) (B), TK01 (pYES2 α-syn GFP) (C) and TK01 (pYES2-GFP) (D) were grown in SG-Ura supplemented with baicalein at final concentrations 0-25 µM. Strains TK01 (pYES2 α-syn GFP) (E) and TK01 (pYES2-GFP) (F) were grown in SG-Ura supplemented with ampicillin at final concentrations 0-25 mM. The 96-well plates were incubated at 30 oC for 16-18 h. Resazurin was added to a final concentration at 0.2 mM. Fluorescence intensity was measured over 100 min. The values are means ± standard deviations (SD) of results from three independent experiments. To compare sensitivity to α-syn toxicity in BY25929 background strain between rsp5A401E mutant and RSP5WT for using in the screening system, both yeast strains were treated with 2-fold serially diluted concentrations of baicalein. The results in Figure 6A and 6C indicated that at 3.13 µM of baicalein was the minimal effective concentration with significantly higher RFU values compared to untreated control (P<0.01 and P<0.001 in RSP5WT and rsp5A401E strain, respectively). Therefore, S. cerevisiae TK01(pYES2- α-syn GFP) showed more sensitive and will be used as yeast indicator strain for screening of compounds alleviating α-syn toxicity. Furthermore, to compare sensitivity of our developed assay system with that of the conventional growth monitoring method using optical density measurement at 660 nm, the 96-well clear microplate containing test strains treated with varying baicalein concentrations were incubated at 30 ⁰C for 24 h. Then, growth of yeast cells in each well was measured at OD660 nm. No statistically significant difference (P >0.05) of growth between the cells treated with 0.39-6.25 µM baicalein and the untreated control was observed (Figure 6B and 6D). Meanwhile, a significant decrease in the growth of cells treated with 12.5 and 25 µM baicalein was observed (P< 0.05 and P< 0.01), indicating the cytotoxic effect of baicalein at such concentrations. As our resazurin yeast-based screening system could detect baicalein as low as 3.13 µM. These results showed that the assay based on fluorescent signal of resorufin was much more sensitive as compared to the cell turbidity determination (Figure 6). 82

The 29th Special CU-af Seminar 2021 August 25, 2021 Figure 6: Comparison of sensitivity to various baicalein concentrations in the assay systems. Strains BY25929 (pYES2 α-syn GFP)) (A and B) and TK01 (pYES2 α-syn GFP) (C and D) were grown in SG-Ura supplemented with baicalein at final concentrations 0-25 µM. The cultures were incubated at 30 oC for 16-18 h. For (b) and (d), their growth was assessed by measuring the optical density at 660 nm (OD660). For (A) and (C), resazurin was added to a final concentration at 0.2 mM. Fluorescence intensity was measured over 100 min. Results represent the mean ± standard deviations (SD) of three independent experiments. For (A) and (C), statistical significances are indicated by the P-values (**P ˂ 0.01; ***P ˂ 0.001, 3.13 µM vs 0 µM). For (B) and (D), statistical significances are indicated by the P-values (*P ˂ 0.05, 12.5 µM vs. 0 µM; **P ˂ 0.01, 25 µM vs 0 µM). These findings altogether supported that the resazurin yeast-based assay system using TK01 (pYES2- α-syn GFP) strain was the effective assay system with high sensitivity and specificity. Furthermore, the toxicity of test compound could be also confirmed during screening. Screening of compounds alleviating human α-syn toxicity Thirty-three compounds isolated from medicinal plants were tested for the ability to alleviating α-syn toxicity using our developed yeast-based assay. The results were expressed as percentage of α-syn toxicity reduction. In primary screening, using final concentrations of each test compound at 200, 20 and 2 µg/ml, 8 out of 32 test compounds treated yeast TK01 (pYES2-α-syn-GFP) cells showed alleviation in α-syn toxicity as indicated by significantly higher in % reduction of toxicity than those of the untreated control (defined as 0% reduction) (Table. 2). 83

The 29th Special CU-af Seminar 2021 August 25, 2021 Table 2: Percentage of human α-syn toxicity reduction of candidate compounds isolated from primary screening using the developed yeast-based assay To determine the minimal effective concentration of these candidate compounds, they were subjected to serial dilution to get final concentrations between 200 – 0.4 mg/ml and were treated to the yeast indicator cells. Our result showed ASCY130 was the most potent compound with minimal effective concentration at 1.56 µg/ml which could significantly reduce human a- syn toxicity of 31% from the untreated control (Figure 7 and data not shown for the other candidate compounds). In addition, test concentrations up to 200 µg/ml did not exhibit cytotoxicity. Therefore, ASCY130 is a potential candidate to be used as a lead compound for the design of therapeutic agent for the treatment of neurodegenerative diseases especially PD. Figure 7: Determination of minimal effective concentration of ASCY130. The yeast strain TK01 (pYES2-α- syn GFP) was incubated in SG-Ura supplemented with ASCY130 at final concentrations ranging from 0.4-200 µg/ml as indicated. The cultures were incubated at 30 oC for 16-18 h. Their growth was assessed by measuring fluorescence intensity of resorufin and further calculated to % reduction of human α-syn toxicity. Results represent mean ± standard deviations (SD) of three independent experiments. Statistical significances (compared to the untreated control defined as 0% reduction) are indicated by the P-values: **P ˂ 0.01 and ***P ˂ 0.0001. 84

The 29th Special CU-af Seminar 2021 August 25, 2021 Conclusion Alpha-synuclein is pathological hallmark of PD. It has been a therapeutic target to slow or stop the progression of disease. In this study, we have successfully established and set up a novel resazurin yeast-based screening system with high sensitivity and specificity for screening of compounds alleviating human α-syn toxicity. Since our assay system was based on restoration of cell viability, therefore the cytotoxicity of test compounds could be simultanouesly identified during screening. Besides the other properties of test compounds such as solubility, stability, cell permeability would also be obtained. By using the developed yeast-based assay to screen 32 compounds isolated from medicinal plants, 8 were detected as active compound in the primary screening. On comparison of minimal effective concentration of each candidate compound, the ASCY130, an active compound isolated from the Stephania suberosa Forman, was detected as the most potent compound with the lowest minimal effective concentration at 1.56 µg/ml and the compound showed no cytotoxicity at least at concentration of 200 µg/ml. Therefore, ASCY130 may be a potential lead compound for the design of therapeutic agent for the treatment of neurodegenerative diseases including PD. In the future, the mode of action of ASCY130 will be investigated. This compound may be either efficiently suppress α-syn aggregation or enhance ubiquitin-dependent degradation. Acknowledgements This work was supported by the Asahi Glass Foundation to CY, Ratchadapisek Somphot Fund for Postdoctoral Fellowship, Chulalongkorn University to AS, and by Global Collaborative Program (2019-2020), Nara Institute of Science and Technology, Nara, Japan, to HT. We are grateful to Boon-ek Yingyongnarongkul, Ph.D., Department of Chemistry, Faculty of Science, Ramkhamhaeng University for providing baicalein. We also thank the Yeast Genetics Resource Center, Japan for providing strain BY25929. References 1. Davis TL, Rafferty M, Lyons K, Ramirez-Zamora A, Gao H, Wu S, et al. Movement Disord. 2019:S246-S. 2. Soh S-E, McGinley J, Watts J, Iansek R, Murphy A, Menz H, et al. Quality of life research : an international journal of quality of life aspects of treatment, care and rehabilitation. 2012. 3. Dexter D, Jenner P. Free Radic Biol Med. 2013(62) 4. Zahoor I, Shafi A, Haq E. In: Stoker TB, Greenland JC, editors. Parkinson’s Disease: Pathogenesis and Clinical Aspects. Brisbane (AU)2018. 5. Kalia L. J Neurochem. 2019(150):35. 6. Stefanis L. Cold Spring Harb Perspect Med. 2012(2):a009399. 7. Kim WS, Kagedal K, Halliday GM. Alzheimers Res Ther. 2014(6):73. 8. Lashuel HA, Overk CR, Oueslati A, Masliah E. Nat Rev Neurosci. 2013(14):38-48. 9. Cheng F, Vivacqua G, Yu S. J Chem Neuroanat. 2010(42):242-8. 10. Vekrellis K, Stefanis L. Expert Opin Ther Targets. 2012(16):421-32. 11. Fields CR, Bengoa-Vergniory N, Wade-Martins R. Front Mol Neurosci. 2019(12). 12. Singh SK, Dutta A, Modi G. Future Med Chem. 2017(9):1039-53. 13. Brundin P, Dave KD, Kordower JEH. Exp Neurol. 2017(298):225-35. 85