Proteomic potential of East African Highland Bananas (EAHBs) for banana juice extraction: comparison between juice-producing and cooking cultivars

ABSTRACT This study investigated the proteomic difference between juice-producing and cooking EAHBs to clarify the role of protein in the production of banana juice. A comparative study was carried out to determine protein content, molecular weight distributions, and amino acid profile of the pulps of ten different (five juice-producing and five cooking) banana cultivars. There was low variability in crude protein content of banana cultivars, the level fell within the range of 0.80 g/100 g to 1.02 g/100 g. SDS-PAGE results visualised that the cultivars had similar molecular weights, ranging between 10 kDa and 76 kDa. The HPLC analysis showed that the relative compositions of amino acids differed significantly (p ≤ .05) within and between juice-producing and cooking cultivars. Both banana cultivars had a significantly higher amount of glutamic and aspartic acids, but significantly lower concentrations of tyrosine, methionine. The results suggest that protein content, molecular weight, and amino acid composition of banana cultivars are not the major factors in determining a banana’s ability to release juice.


Introduction
East African highland bananas (EAHBs) are common banana cultivars that are used as a staple food by more than 25 million people in East Africa. They are mostly utilised for cooking and juice production (Karamura et al., 1998). Bananas are highly perishable, which makes them susceptible to post-harvest deterioration, and significant amounts are discarded, leading to huge losses of valuable food resources (Pessu et al., 2011). The introduction of different post-harvest processing techniques, e.g. processing bananas into juice, is an important step towards more responsible food production.
In East Africa, banana juice extraction is conducted using a traditional technology that involves the use of wooden boat-like containers with grass, mainly Imperata cylindrica. The bananas are peeled, grasses are added and the mixture is kneaded for 15-25 min to squeeze the juice from the bananas (Kyamuhangire et al., 2002). This traditional technology has existed for centuries; however, the commercialisation of this banana juice technology is challenging due to occasional juice extraction failures (Kibazohi et al., 2017;Kyamuhangire et al., 2002;Kyamuhangire & Pehrson, 1999).
The mechanism of banana juice release has been linked with protein-tannin interactions (Kibazohi et al., 2017;Kyamuhangire et al., 2006). A recent study has shown that prolonged blending of juice-producing bananas can be used to facilitate protein-tannin interactions to obtain banana juice and without the addition of grass (Kibazohi et al., 2017). Interactions between protein and tannins involve hydrophobic effects (Ozdal et al., 2013), resulting in the formation of insoluble aggregates in an aqueous solution, and separation of the liquid part (Prigent, 2005;Prigent et al., 2009;Rawel & Rohn, 2010). It is likely that the same phenomenon is occurring during banana juice release, with the assumption that the dissociated complex forms the hydrophobic aggregates that facilitate juice release (Kyamuhangire et al., 2006). A number of studies have been conducted to understand the mechanisms behind traditional banana juice production. Characterisation of tannins in juiceproducing and cooking bananas has revealed that juice-producing bananas are more likely to produce banana juice, which has been suggested to be due to the high amount of tannins (Kyamuhangire et al., 2006). Further, Kibazohi et al. (2017) reported that banana juice was able to be recovered from bananas with the condensed tannins above 0.68% (w/w) of a peeled banana. However, there is a lack of comprehensive proteomic data with regard to different EAHBs.
In this study protein content, molecular weight, and amino acid composition of different EAH banana cultivars were analysed in an attempt to clarify the role of protein during mechanical juice extraction.

Protein extraction
The extraction of protein from the banana pulp was conducted following Wang et al. (2006). Homogenised freeze-dried sample (0.5 g) was weighed into a 2 mL test tube and suspended in (1.5 mL) 10% trichloroacetic acid (TCA)-acetone, mixed, vortexed, centrifuged at 5311 g force (Multifuge I.S.R Heraeus, Kandro laboratory products, Germany) for 3 min at 4°C. The supernatant was removed and obtained residue was treated with 80% methanol in 0.1 M ammonium acetate (1.5 mL). Centrifugation was repeated as for the previous step and the supernatant was discarded. In the next step, 1 mL of 80% acetone was added to the tube to wash the sample, and the same procedures as described above were applied, and the final supernatant was discarded. The pellets were allowed to air-dry at room temperature for 10 min, 1 mL of Phenol: SDS (1:1) was added, mixed, incubated for 5 min, and centrifuged. The upper phenol layer was put in another 2 mL test tube and methanol containing 0.1 M ammonium acetate was added to the test tube, stored at 20°C for 10 min, and thereafter centrifuged for 5 min at 4°C. The supernatant was discarded after centrifugation and the protein pellets were collected and stored at −20°C until analysis.

Extraction of amino acids and derivatisation
Approximately 0.5 g of lyophilized sample was weighed into hydrolysis tubes in triplicate, 5 mL of 6 M hydrochloric acid (HCl) was added, and the tubes were flushed with N 2 , quickly cupped, and placed in an oven at 110°C for 24 h. Following hydrolysis, the tubes were opened and flushed again with nitrogen gas under vacuum to remove HCl. The dry residues were dissolved in buffer (MEOH-H2O-TEA, 2:2:1 [v/v]) and excess solvents were removed under vacuum at 65°C. Next, 30 μL of the derivatising reagent (MeOH-H2O-TEA-PITC, 7:1:1:1 [v/v]) was added, and removed under vacuum conditions at 65°C. After derivatisation, the tubes were sealed and stored at 4°C until analysis. Prior to HPLC injection, 150 µL of 5 mM sodium phosphate buffer with 5% acetonitrile was added to each tube to dilute the derivatised samples. The amino acid standard H solution was derivatised by the addition of 10 µL of standard H (amino acid hydrolysate standard) to 30 µL of PITC reagent, and the mixture was processed in the same way as the sample. The sample extracts and standards were filtered through 0.22 µm syringe filters (Sigma-Aldrich, USA) before HPLC analysis.

Crude protein determination
The protein contents of the banana cultivars were determined by the Lowry method (Waterborg, 2009). The following reagents were used for the analysis: reagent A, 2% sodium carbonate in 0.1 N sodium hydroxide, reagent B, 0.5% copper sulfate solution in 1% sodium potassium, and reagent C, a mixture of reagent A and B in the ratio 5:1 (v/v). A 0.1 mL aliquot of each sample extract was put in a 2 mL test tube. To a 0.1 mL sample, 4 mL of reagent C was added, before incubation for 10 min at room temperature. Subsequently, 0.5 mL of Folin-Ciocalteu reagent (FCR) was added to each test tube, mixed, and allowed to stand in darkness for 30 minutes. The absorbance of the resulting solution was read at 660 nm using a UV-spectrophotometer (Safire II plate reader Tecan, Austria). A standard curve made from bovine serum albumin at the concentrations of 0.2, 0.4, 0.6, 0.8, and 1.0 mg/mL was used for quantification.

SDS-PAGE analysis
Protein separation by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis was performed based on the method described by Zheng et al. (2007). Protein samples were digested with Laemmli buffer (30 µL) and β-Mercaptoethanol (30 µL) and heated at 95°C for 5 min. Samples and standards (20 µL) were loaded on pre-casted gels (Bio-Rad Laboratories, Inc, USA) and run at 100 V for 2 h. The SDS-PAGE run was stopped when the bands reached the end of the gel, A staining solution (0.025% Coomassie brilliant blue R-250, 45% methanol, 10% acetic acid) was added and remained for 1 h while shaking. Lastly, the staining solution was removed and a destaining solution (40% methanol, 10% acetic acid) was added to remove excess dye and obtain clear protein bands. Molecular mass was determined by comparison with broad range molecular weight (10-250 kDa) standard markers (Bio-Rad Laboratories, Inc).

HPLC analysis of amino acids
An HPLC system (1260 Affinity II Agilent, Germany) was used to quantify amino acids (AAs) according to the method by González-Castro et al. (1997). Amino acids do not have chromophores for UV detection, thus pre-column derivatisation with PITC was employed to obtain PITC derivatives. Subsequently, the phenylthiocarbamide (PTC) amino acids were auto-injected, and the separation of the derivatised amino acids was achieved using a Poroshell C8 column (5 µm, 150 × 4.6 mm, Agilent, Germany) at 25°C, and wavelength 254 nm. Mobile phase A was aqueous sodium acetate buffer (0.14 M) with 0.5 ml/L of TEA, mobile phase B consisted of acetonitrile-water (60:40). For the separation a flow rate of 1.0 mL/min with a gradient elution employed at the following order (0-12) min 10% B, (12-20) min 30% B, (20-22) min 48% B, (22-24) min 100% B, (24-30) min 10% B, was used. The amino acid stock solution mixture was prepared in (0.1) M hydrochloric acids and contained 0.25 mg L −1 of each amino acid. Quantification of individual AAs was performed by using an external standard calibration curve drawn by linear regression analysis.

Statistical analysis
All experiments were carried out in triplicate and the results were expressed as mean values with a standard deviation (± SD). Significant differences in the amino acid composition of different cultivars were verified by one-way analysis of variance (ANOVA). Tukey's multiple comparisons were used to compare means of statistically significant difference between samples at a 95% confidence level using IBM ® SPSS ® statistics ver.24 software.

Crude protein
The results from the determination of crude protein in the banana cultivars are shown in Table 1. The protein content varied slightly between juice-producing banana and cooking banana cultivars, ranging between 0.80 and 0.97 g/100 g FW, but without significant (p ≥ .05) difference. These results Los valores son las medias de tres repeticiones ± desviación estándar. las medias con letras distintas en cada fila son significativamente diferentes (Tukey, 0.05). J se refiere a los cultivares de plátano que producen jugo y c a los cultivares de plátano para cocinar. are in close agreement with previous results (John & Marchal, 1995;Zanariah & Noor Rehan, 1987), of the protein content of 1.09 g/100 g FW for Musa acuminata ssp and 0.8 g/100 g FW for Pisang awak, respectively. However, our results were much lower than the protein content (2.2 g/ 100 g DW) found in Cameroon banana pulp (Adisa & Okey, 1987). Furthermore, the protein content of the different parts of the banana pulp has previously been studied. Forster et al. (2003) observed no significant difference in protein content in samples taken from the external, medium, and centre positions of the pulp (0.95, 0.95, and 1.27 g/100 g DW), respectively. In support of the above findings, a slight variation in protein content between cultivars might be attributed to the genetic variation and different agro-ecological conditions (Akaninwor & Sodge, 2005;Lustre et al., 1976;Pareek, 2016). It is also suggested that the variation in protein content across the cultivars might be explained by the difference in solubility of proteins in the extraction solvents (Chan & Phillips, 1994).

Protein molecular weight distributions
The ten fractions of proteins from banana pulp cultivars were separated by SDS-PAGE. The patterns of banana proteins obtained were highly similar for all cultivars (Figure 1), indicating the presence of the same types of proteins in the juice-producing and cooking cultivars. These results showed that the protein aggregates from both juice-producing and cooking cultivars were primarily composed of monomer subunits at 13, 17, 23, 33, 42, 53, and 76 kDa respectively. Thus, the protein molecular weight distributions in juice-producing and cooking banana cultivars were similar. Consequently, molecular weight distribution cannot be used as a marker to distinguish between the two banana cultivars and their potential ability to release banana juice. Identification of these banana proteins has been reported by other researchers, chitin-binding proteins monomers at 30 kDa, 32 kDa, and 33 kDa were isolated from banana fruits respectively (Ho & Ng, 2007;Mikkola et al., 1998). Moreover, the monomer at 22 kDa was identified as a thaumatin-like protein in the banana of Musa acuminata (Nikolic et al., 2014;Toledo et al., 2012). These proteins are produced as the response to biotic and abiotic stresses in plants (Ishige et al., 1993;De Jesús-pires et al., 2020;Mathivanan et al., 1998;Mehta et al., 1991). Furthermore, the band identified at 42 kDa was similar to the results reported by Yang et al. (2000), who found a similar protein in the banana pulp of Musa sapientum L. (Musaceae).

Quantification of amino acids (HPLC)
The binding affinity of protein to the phenolic compound is a function of amino acid composition (Rawel et al., 2005;Sęczyk et al., 2019). Moreover, protein-rich in proline amino acid has shown a high affinity to phenolic compounds (Luck et al., 1994). Previous studies suggest that a high affinity of salivary proline-rich proteins (PRPs) results in the formation of an insoluble complex and hence an astringency sensation when eating tannin-rich diets (Asquith et al., 1987;Makkar & Becker, 1998;Shimada, 2006). Considering the influence of PRPs on tannins, it was hypothesised that high proline content in a banana is the main characteristic of juice cultivars and it may be linked to their ability to release juice. In the present study, the amino acid concentrations of juice-producing and cooking Values are means of three replicates ± standard deviation.
Means with different letters in each row are significantly different (Tukey, 0.05). n.d: not detected; J refers to juice-producing banana cultivars and c refers to cooking banana cultivars.
cultivars were analysed and calculated using 17 commercial amino acid standards. To reveal cultivar-specific differences between different samples, the Tukey test (p ≤ .05) of all amino acids data was subsequently performed based on the mean values of all cultivars ( Table 2). Chromatograms of the detected amino acids in the standard solution and samples are shown in Figures 2-4, respectively. Table 2 shows the amino acid composition in the analysed cultivars. In general, significant differences (p ≤ .05) in AAs were found among the banana cultivars, however, there were no specific AAs that were found in a juice-producing or cooking cultivar. Many of the observed AAs have been previously identified in plantains  ; 5, glycine; 6, histidine; 7, arginine; 8, threonine; 9, alanine; 10, proline; 11, tyrosine; 12, valine; 13, methionine; 14, isoleucine; 15, leucine; 16, phenylalanine; 17, lysine. Figura 2. Cromatograma de HPLC de la mezcla estándar de aminoácidos. Picos: 1, ácido aspártico; 2, ácido glutámico; 3, hidroxiprolina; 4, serina; 5, glicina; 6, histidina; 7, arginina; 8, treonina; 9, alanina; 10, prolina; 11, tirosina; 12, valina; 13, metionina; 14, isoleucina; 15, leucina; 16, fenilalanina; 17, lisina.    as well as other cooking bananas (Alsmairat et al., 2018;Emaga et al., 2007;Sharaf et al., 1979). Further, glutamic and aspartic acids were detected as the most prominent amino acids in the majority of the banana pulps. Similar values of glutamic and aspartic acids have been reported in other banana cultivars such as plantains; however, the values obtained in this study were higher than those reported in the literature (Emaga et al., 2007). Only minor amounts of tyrosine and methionine were detected in the banana cultivars in the current study (Table 2). Methionine and threonine were not detected in most of the cultivars and the remaining AAs such as hydroxyproline, isoleucine, serine, and arginine were found in small or trace amounts. Earlier reports suggest that the low contents of tyrosine and methionine are likely due to the degradation caused by oxidation of these AAs prior to HCL hydrolysis (Bergès et al., 2011). In addition, cooking cultivars had a higher content of the essential AAs valine and lysine than juice cultivars ( Table 2). The content of valine in cooking cultivars was more than double the content of valine in juice banana cultivars, the highest content was found in the Nshakala KG (101.5 ± 0.19 mg/100 g DW), Malindi (99.1 ± 2.10 mg/100 g DW), and Nshakala KL (81.2 ± 0.56 mg/ 100 g DW) cultivars. These results are in close agreement with those previously reported by Sheng et al. (2010). The variation in valine content across banana cultivars might be explained by the differential uptake or consumption of valine during the formation of volatile compounds in banana (Musa sp., AA group, Cavendish) pulp (Alsmairat et al., 2018).
Nshakala KG contained the highest amounts of alanine (119.3 ± 0.34 mg/100 g DW), glycine (122.2 ± 1.32 mg/100 g DW), and histidine (144.1 ± 2.31 mg/100 g DW). Furthermore, lysine was high in Malindi (128.0 ± 3.10 mg/100 g DW) and Nshanshambile (103.6 ± 2.41 mg/100 g DW). A significant amount of alanine, glycine, and histidine have been reported in banana of Musa Cavendish "Hindi" variety from Egypt (Sharaf et al., 1979), however, the values obtained in this study were lower than data found in the literature using the Nitrogen analyser technique. This difference was mostly explained by variation in genetic factors and analytical methods (Hall & Schönfeldt, 2013).
In considering the proline contents of cultivars, the values varied from 31.1 ± 0.71 mg/100 g DW to 72.6 ± 1.21 mg/ 100 g DW. The proline content is related to the affinity of protein to phenolic compounds (Hagerman, 2012). With respect to the two groups (Juice-producing and cooking cultivars), the values obtained in this study did not distinguish the two groups in terms of proline content since the values were significantly different (p ≥ .05) across all cultivars. Surprisingly, cooking cultivars showed both high and low proline levels. Ndeshi laini (cooking) banana cultivar had the highest proline content (72.6 ± 1.21 mg/100 g DW), and Malindi (cooking) cultivar had the lowest (31.1 ± 0.71 mg/ 100 g DW) ( Table 2). The different AAs' composition in the banana cultivars might be attributed to varietal differences, different environmental conditions, and biochemical changes during growth and ripening (Drawert et al., 1972;Florent et al., 2015;Joshi & Kumar, 1989).

Principle component Analysis (PCA)
Principal component analysis (PCA) was applied to determine the number of significant variables and to establish the correlation between the measured amino acids (aspartic acid, glutamic acid, hydroxyproline, serine, glycine, histidine, arginine, threonine, alanine, proline, tyrosine, valine, methionine, isoleucine, leucine, phenylalanine, and lysine) in the studied cultivars ( Figure 5). The results from the PCA revealed that the first principal component PC1 scored 42.6%, and the second principal component PC2 scored 15.1% of the total variation in the data set. As shown in the PCA plot, the cultivars with the smallest distance between each other are considered to be more closely related to each other. Moreover, the two principal components grouped banana cultivars into four different clusters. The first cluster consisted of FIA 23, Ndeshi laini, Mtwishe, and Pisang awak cultivars, the second cluster of Nyengele and Nshakala KG, the third cluster of Toke Uganda (Nshakala KL) and Malindi, and the fourth cluster of Nshanshambile and KM5. The cultivars within the 1st and the 4th clusters are capable of producing juice, while cultivars within the 2nd and 3rd clusters are primarily used for cooking purposes.

Conclusion
The present study reveals that there was no significant difference (p ≥ .05) in protein content among the studied banana cultivars. Moreover, the electrophoretic patterns of all banana cultivars resembled each other with the MW range of 10-76 kDa. Furthermore, all cultivars showed the predominance of glutamic and aspartic acid. The varietal differences in the amino acid composition between the juice-producing and cooking cultivars did not follow a definite pattern, hence the AAs are not a key decisive factor in the banana juice release, though they may participate in the protein-tannin interaction. Further studies are however needed, to understand the mechanism of interactions between proteins and tannins in the banana pulp during mechanical juice extraction.