II. LITERATURE REVIEW
2.3 Banana
Banana (Musa spp.) is a tropical fruit belonging in the Musaceae family. It is one of the most produced fruits globally with nearly 120 tonnes of production in 2020. Banana is one of the most widely produced, traded, and consumed fruits in the world. There are over 1000 varieties of bananas in the world, which provide essential nutrients to populations in both producing and importing countries (FAO, 2022). Nutrient content in banana varies depending on the cultivar, harvesting age, soil, climate and weather, etc. In average, chemical composition in banana is as follows. 70.6% water, 0.89% ash, 26.86% carbohydrate, 1.67%
protein, and 0.06% fat. Aside from that, bananas also contain 114 kcal energy per 100 gr, making them a great source of energy (Hapsari & Lestari, 2016).
Not only its fruit, banana plants are arguably versatile. Almost every part of banana plants can be used in some aspects. Banana leaves for wrapper and food containers, flowers as vegetables, shoots for textiles, and trunks for paper. Bananas, including their fruit peel, also serve as medicine in the traditional use because not only nutritious in macronutrients, but
also in micronutrients in terms of vitamins and minerals. Aside from that, they also high in phytochemicals, including phenolic compounds, which have been shown to exhibit health promoting benefit (Lai et al., 2017).
With its huge production size, bananas also come with a large amount of fruit waste.
Banana peel is one of the major fruit wastes from banana processing. It comprises nearly 40%
of the banana fruit by mass (Sharma et al., 2016). This causes huge amounts of waste to end up in the landfill or incinerator, which would later cause further environmental problems if managed inappropriately, such as greenhouse gasses emission and toxic incomplete combustion generation. On the other side, banana peel provides a good source of fiber, protein, and phytochemicals, including phenolic compounds (Acevedo et al., 2021). Numerous studies have also found the health benefit of banana peel, notably its antioxidant, antimicrobial, and antidiabetic activity, especially with regards to their fiber and phytochemical contents, including phenolic compounds (Hikal et al., 2022; Wang et al., 2022).
Hence, there is a need to find alternative utilization of banana peel to reduce fruit waste and exploit its benefit.
2.2.1. Phenolic Compounds in Banana Peel
Phenolic compounds have piqued the interest of researchers in recent years, owing to their abundance in plants as secondary metabolites, high antioxidant capacity, and health-promoting effects. Antioxidants, anti-cancer, anti-diabetes, inhibiting adipogenesis, decreasing blood pressure, and suppressing inflammatory genes are among the health benefits (Gutiérrez-Grijalva et al, 2016). The anti-diabetic properties of phenolic compounds have been widely investigated. In vivo studies in animal models and limited human models have shown that phenolic compounds can reduce hyperglycemia while also improving insulin secretion and sensitivity (Aryaeian et al., 2017; Naz et al., 2019). Furthermore, studies of several phenolic acids have suggested an anti-diabetic mechanism via PPAR activation, GLUT4 activation, PI3K
activation, NF-B inhibition, and insulin secretion stimulation. In vitro studies revealed that phenolic extracts of some plants have anti-diabetic properties by inhibiting α-amylase and α-glucosidase, with the suggested phenolic compounds being polyphenols, phenolic acids, anthocyanins, and proanthocyanidins (Asgar, 2012;
Wojdyo et al., 2016; Praparatana et al., 2022).
The phenolic compounds are found relatively high in banana, including its peel (Acevedo et al., 2021). Total phenolic compounds in banana peel varies depending on the cultivar, ranging from 4.95 – 47 mg gallic acid equivalent (GAE)/ g dry matter (Vu et al., 2018). A study that compared total phenolic compounds in different fruit revealed that total phenolic compounds in banana peel was comparably higher than apricot, kiwi, dragon fruit, melon, pear, papaya, peach, pineapple, plum, pomegranate, and passion fruit peel, even though it was still lower than citrus, mango, and apple peel (Suleria et al., 2020). The phenolic compounds detected in banana peel varied across their cultivars and are summarized in Table 4 below.
Table 4. Phenolic compounds found in banana peel. Summarized from Suleria et al. (2020), Bashmil et al. (2021), and Aboul-Enein et al. (2016).
Class Compounds Species Reference
Phenolic Acids Caffeic acid MP Aboul-Enein et al. (2016) Chlorogenic acid MAC Suleria et al. (2020) Ferulic acid MAC Suleria et al. (2020) Gallic acid MAC Suleria et al. (2020) Hydroxybenzoic Acid Protocatechuic acid
4-O-glucoside
MAC Suleria et al. (2020)
2-Hydroxybenzoic acid MAC Suleria et al. (2020) 3,4-O-Dimethylgallic acid MAL Bashmil et al. (2021) Hydroxycinnamic Acid Caffeoyl glucose MAC Suleria et al. (2020)
Cinnamic acid MP, MAC Aboul-Enein et al.
(2016), Suleria et al.
(2020)
m-coumaric acid MAC Suleria et al. (2020) Hydroxyphenylpropanoic
Acids
3-Hydroxyphenylpropionic acid
MAC Bashmil et al. (2021)
p-Coumaroyl glycolic acid MAC, MAD, MAL, MAR, MP,
Bashmil et al. (2021)
Hydroxyphenylacetic Acids
3,4-Dihydroxyphenylacetic acid
MAC, MAR
Bashmil et al. (2021)
Hydroxycoumarins Urolithin A MAC Bashmil et al. (2021) Scopoletin MP Bashmil et al. (2021) Umbelliferone MAM Bashmil et al. (2021) Anthocyanins Cyanidin 3,5-O-diglucoside MAR Bashmil et al. (2021)
Delphinidin 3-O-(6”-acetyl-galactoside)
MP, MAC, MAM
Bashmil et al. (2021)
Malvidin 3-O-(6”-acetyl-glucoside)
MAR Bashmil et al. (2021)
Flavones Chrysin MP Aboul-Enein et al. (2016)
Gardenin B MAC Suleria et al. (2020) Cirsilineol MAC Suleria et al. (2020) Chrysoeriol 7-O-glucoside MAC Bashmil et al. (2021) 6-Hydroxyluteolin
7-rhamnoside
MAC Suleria et al. (2020)
Flavanones Hesperetin 3'-O-glucuronide
MAC Suleria et al. (2020)
Naringenin MAC Suleria et al. (2020) Neoeriocitrin MAR Bashmil et al. (2021) Flavonols 3-Methoxysinensetin MAC Suleria et al. (2020)
Isorhamnetin 3-O-glucoside 7-O-rhamnoside
MAC Bashmil et al. (2021)
Myricetin 3-O-galactoside MAC Suleria et al. (2020) Myricetin 3-O-rhamnoside MAC Suleria et al. (2020) Myricetin 3-O-rutinoside MAC Bashmil et al. (2021) Patuletin
3-O-glucosyl-(1->6)- [apiosyl(1->2)]-glucoside
MAR Bashmil et al. (2021)
Quercetin 3-O-xylosyl-glucuronide
MAC Bashmil et al. (2021)
Rutin MAC Suleria et al. (2020) Isoflavonoids
5,6,7,3',4'-Pentahydroxyisoflavone
MAC Suleria et al. (2020)
Isoquercitrin MAC Suleria et al. (2020) Hydroxybenzaldehydes 4-Hydroxybenzaldehyde MAC Suleria et al. (2020) Curcuminoids Demethoxycurcumin MAC Suleria et al. (2020) Furanocoumarins Isopimpinellin MAC Suleria et al. (2020) Phenolic Terpenes Carnosic acid MAC Suleria et al. (2020)
Lignans Schisantherin A MAC Suleria et al. (2020)
Other Polyphenols Salvianolic acid B MAC Suleria et al. (2020) Note: MP: Musa paradisiaca, MAC: Musa acuminata Canvendish, MAD: Musa acuminata Ducasse, MAL:
Musa acuminata Ladyfinger, MAR: Musa acuminata Red Dacca, MAM: Musa acuminata Monkey
2.4. In Silico Approach of Antidiabetic Study
The prevalence of DM keeps increasing while the current treatment is still limited. With that in mind, research all over the world is still trying to find a better, more effective, yet safe means for the treatment of T2DM. In silico screening is of the current interest for drug discovery and design as it is cost- and time-effective. In silico approaches have been shown to be useful in estimating the biological activities of chemical compounds against a target.
Furthermore, it has been utilized to investigate binding affinities toward the target and to predict physicochemical attributes of a wide range of chemical compounds based on their molecular and structural aspects (Jabalia et al., 2021).
Some phenolic compounds in plants have shown to exhibit anti-diabetic effect via in vivo and in vitro (Aryaeian et al., 2017; Naz et al., 2019; Asgar, 2012; Wojdyo et al., 2016;
Praparatana et al., 2022). However, the underlying molecular mechanisms is yet to be elucidated. Furthermore, in silico studies of phytochemicals in plants targeting diabetic protein have been done through molecular docking and molecular dynamic. Study by Khanal et al. (2019) who conducted docking study of phytochemicals in Tinospora towards 11 target proteins, revealed that the effect of magnoflorine was potential and comparable to sitagliptin
and repaglinide. Sharma et al. (2020) docked phytochemicals in Phyllanthus emblica towards GLP1, SGLT2, and PPARγ, and showed that 7 compounds were potential. Mechchate et al.
(2021) specifically docked gentisic acid, a phenolic acid, towards PTP1B, PPARγ, and other six proteins, and their study showed that it moderately interacted with most targets. On the broader spectrum but still focus in phenolic compounds, Damián-Medina et al. (2020) did in silico study of antidiabetic activity of phenolic compounds from blue corn and black bean towards 13 proteins and revealed that four compounds highly corresponded with the proteins, mainly 11βHS, PTP1B, GFAT, PPARγ, and tyrosine kinase insulin receptor. There were lots of other works using in silico approach, indicating its robustness and reliability. To note, many of the phytochemicals indicated by those study were phenolic compounds and their interactions were indicated positive, including towards PTP1B. Nonetheless, in silico study regarding the molecular mechanism of phenolic compounds in banana peels is still lacking. Since the target proteins in the DM and their structures, as well as the ligands (phenolic compounds in banana peels) are well discovered, in silico approach to search potential treatment for T2DM is best used.
In silico study provides a critical and useful step in drug discovery, that is to screen and
identify potential drug candidates through a cost-effective means (Brogi et al., 2020). On the other hand, this rationale will also decrease the unnecessary use of animal model (in vivo) and in vitro studies by limiting the number of compounds to be tested to only those deemed as
potential candidates. Albeit robust and accurate, these in silico approaches are still predictions using computational chemistry and bioinformatics which should be proven experimentally in the next step by in vitro and in vivo studies
2.3.1. QSAR Analysis
QSAR analysis is a bioinformatics approach to quantitatively correlate the chemical structure of a compound with its biological activity or chemical reactivity (Bustamam et al., 2021). One of the online services that uses this approach to screen
compounds is PASS SERVER (Filimonov et al., 2014; Widodo et al., 2018). Chemical structure of the compound can be submitted to the software, then Pa (probability of activity) and Pi (probability of inactivity) will be returned in the score between 0.000 to 1.000. The results of PASS SERVER are interpreted as (i) activities with Pa > Pi are deemed possible, (ii) Pa > 0.7 means that the chance to experimentally find the activity is high, (iii) 0.5 < Pa < 0.7 means that the structure of the compounds is quite dissimilar with known pharmaceutical agents and the chance to experimentally find the activity is less, (iv) Pa < 0.5 means that the chance to experimentally find the activity is less, but the chance to find a new structurally active compound is more (Goel et al., 2011 as cited in Hussain et al., 2016).
2.3.2. Molecular Docking
Molecular docking is a technique to study the interaction between two or more molecules (e.g., protein and ligand). The ability of protein to have interaction with ligand to form a complex changes the structure and dynamics of the protein, which would inhibit or enhance its biological function. This method aims to identify correct ligand conformation in a binding pocket of protein and predict the affinity between the them. Docking mechanism consists of two fundamental steps: predicting ligand conformation as well as its location and orientation in protein active sites, and determining binding affinity or score. In the first step, it is related to the sampling approaches that is used for the docking. Depending on the method that is used, conformation of the ligand and protein flexibility can be limited. On the other hand, the chosen scoring algorithm will determine the binding affinity predicted in the docking simulation (Roy et al., 2015).
Table 5. Types of binding interaction and their energy Types of Interactions Energy (kcal/mol)
Covalent C=O 165
C=C 143
C-H 103
C-C 86
C-O 81
Non-covalent
Hydrogen bond 2 – 30 Electrostatic 1 – 20 π-π stacking 0 – 10 Hydrophobic <10 Van der Waals 0.1 – 1
The optimal binding pose of the ligand can be predicted using molecular docking.
By retrieving this information, subsequently the interaction with the amino acid residues of the protein where the ligand binds can be analyzed as well. According to Chen and Kurgan (2009), the most significant and commonly found interaction in ligand-protein binding included hydrogen bond, covalent bond, Van der Waals interaction, electrostatic force, and coordination bond, in which, strong interactions occurred in covalent and coordination bond. Additionally, Biovia has put their classification of bonds into covalent and non-covalent interaction (Table 5) (Biovia, 2019). In their Discovery Studio software, they have included the possible interactions based on this.
The interactions include favorable and unfavorable interaction, shown in Figure 1.
Favorable Unfavorable
• Charge
o Attractive charges o π-cation
o π-anion
• Halogen o Fluorine o Chlorine o Bromine o Iodine
• Hydrophobic o π -π stacked
• Hydrogen bond o Conventional o C-H bond
o π donor hydrogen bond
o Water mediated hydrogen bond o Salt bridge
• Others
o Metal-acceptor o π-sulfur o Sulfur-halogen
• Steric bumps
• Charge repulsion
• Acceptor-acceptor clashes
• Donor-donor clashes
o π -π T-shaped o Amine-π stacked o Alkyl
o π -σ o π -alkyl
o π-lone pair
Figure 1. Binding interaction calculated in Discovery Studio 2.3.3. Drug-likeness and ADME-Tox Prediction
Drug-likeness provides a useful consideration in early-stage screening of drug discovery.
This notion comes from a similar distribution of some key physicochemical properties of approved drugs, including but not limited to molecular weight, hydrophobicity, polarity, solubility, and permeability (Oprea, 2000 as cited in Bickerton et al., 2012). The rationale behind this drug-likeness prediction is that physicochemical properties of a compound exhibit a molecular behavior in vivo. That being said, compounds that fall within this distribution are considered as likely to be drug candidates. Practically, the assessment of drug-likeness goes under set of rules, most notably and the first of which is Lipinski’s rule of five (Ro5) (Lipinski et al., 1997), especially for oral bioavailability. Ro5 states that a poor absorption is most likely to occur when two or more of the following criteria are violated: (i) calculated logP (lipophilicity) < 5, (ii) MW < 500 g/mol, (iii) less than 5 hydrogen bond donors, or (iv) less than 10 hydrogen bond acceptors exist.
Furthermore, ADMET analysis is preferred and developed by ADMETLab 2.0, by including 17 physicochemical properties, 23 ADME-related properties, 27 toxicity endpoints, 8 toxicophore rules, and 13 medicinal properties, in addition to only considering Lipinski’s Ro5, in order to achieve a more comprehensive and accurate predictions (Xiong et al., 2021).
2.3.4. Molecular Dynamics Simulation
Molecular dynamics is a computational prediction of the movement of atoms in the protein complex in a dynamic model. Important biomolecular processes, such as conformational change, ligand binding, and protein folding can be captured through this simulation (Hollingsworth et al., 2018). Such information is critical for guiding drug
discovery and design as it reveals the understanding on the structure-function relationship of the ligand-protein interaction. Therefore, molecular dynamics has been widely applied in the drug discovery and design process, especially after obtaining high affinity of ligand-protein interaction using molecular docking (Liu et al., 2017). The simulation is carried out using Newtonian dynamic equation