*3.4. Feeding 2-Keto-4-Methylpentanoic Acid to Banana Pulps to Produce Branched Alcohol, Aldehyde and Ester*

To test the production of branched chain alcohol, aldehyde and ester from keto acid, 2-keto-4-methylpentanoic acid was fed to banana pulps. As expected, 3-methylbutanol and 3-methylbutal were detected in the headspace (Figure 2a). However, there was no 3-methylbutyl 3-methybutanoate in the headspace, although 3-methylbutyl ethanoate was detected (Figure 7).

**Figure 7.** Formation of 3-methylbutanol, 3-methylbutanal and 3-methylbutyl ethanoate from exogenous 2-keto-4-methylpentanoic acid (50 nmol) after incubation with pulp slices (10 g) of bananas at the "yellow with green tips" stage. Vertical line at each marker shows average ± standard deviation (n = 3).

#### **4. Discussion**

Under anaerobic conditions, exogenous butanol fed to banana pulp was not effectively converted to butyl butanoate (Figure 3a), indicating that oxygen was required in the pathway from alcohols to esters (Figure 1). Due to ADH, ACS, and AAT all not directly requiring oxygen, aldehyde oxide (ALO) was suspected as the enzyme which catalyzes butanal to butanoic acid, although only several specific substrates were catalyzed by ALO in plants, some of which are precursors of important plant hormones (indole-3 acetate (IAA) and abscisic acid (ABA)) [23] (Figure 1). However, the butanal feeding experiment showed that oxygen was not directly required; under anaerobic conditions, banana pulp converted exogenous butanal to butanoic acid, meanwhile, NAD<sup>+</sup> was the cofactor (Figure 4c), indicating that ALDH, not ALO, is the enzyme. Another side-by-side experiment in Figure 4c showed that enzyme-deactivated pulp tissue by microwave heating lost the ability to convert butanal to butanoic acid. Nevertheless, there is not a clear answer why under anaerobic conditions, feeding with butanol did not produce butyl butanoate (Figure 3a). It seems that a deficit of NAD<sup>+</sup> may correlate to the cascade conversions from alcohols to carboxylate, and dehydrogenation of NADH to NAD+ was blocked under anaerobic conditions (Figure 1).

In vitro experiments showed that the optimum phosphate buffer concentration to ALDH was >100 mM (Figure 6a). This might be due to phosphate ion accelerating the activity of ALDH, which was observed in germinating peanut cotyledon [24]. NAD+ was a better coenzyme to ALDH and the efficiency of NADP<sup>+</sup> was only one third in comparison to NAD+ (data not shown), although NAD+ and NADP+ perform similar redox functions within the cell. The latter is also more confined to biosynthetic pathways and redox protective roles in general [25].

During ripening, ALDH activity in the homogenate increased continually until the "yellow with many brown flecks" stage, and then it remained unchanged or slightly decreased in the butanal feeding experiment (Figure 4b). However, when butanol was fed into banana slices at different ripening stages, the peak of butyl butanoate production appeared the "yellow with green tips" stage (Figure 3b), indicating that experiments using pulp slices and homogenate may lead to different results in enzyme activities: ALDH could have high activity at younger ripening stage, i.e., at the "yellow with green tips" stage (Figure 3b), but at that stage, polyphenol and tannin contents were much higher than at the "yellow with many brown flecks" stage, which were mixed with enzymes during homogenization, thus inhibiting the enzyme activity of ALDH (Figure 4b) [26].

In the substrate specificity experiments feeding C2–C6 branched or straight chain aldehydes, the ones with the highest affinity to ALDH were even and straight chain C6 and C4 aldehydes, and that with the lowest was C2 ethanal (Figure 4a). This means that ethanal formed during fermentation or senescence cannot leak to ethanoic acid, which usually is a key function in detoxification of exogenously and endogenously generated aldehydes in mammals [27]. Branched chain aldehydes were generally lower in affinity to ALDH in comparison to the straight chain aldehydes with the same carbon number (Figure 4a). The trends are similar to the affinity of the acyl-CoA/carboxylate to AAT [14]: the even and straight chain substrates had high affinity, but branched chain substrates had low affinity (Figure 4a). A conflicting observation was that even though the affinity of ALDH for hexanal was high (Figure 4a), very low conversion occurred from exogenous hexanol to hexyl hexanoate (Figure 3c). One of the possible reasons is due to slow partition of hexyl hexanoate in headspace, it did not build up enough vapor when direct headspace sampling was used, even though hexyl hexanoate was abundant in the pulps or solutions. For example, hexyl hexanoate was one of the most abundant esters in Gala apples when Tenax GC trap or solid-phase microextraction (SPME) trap sampling methods were used [28]. However, when direct headspace sampling method was used, hexyl hexanoate was not detectable [29].

TCA cycle and β-oxidation of fatty acids in fruit pulps continually provide ethanoyl-CoA (acetyl-CoA)/ethanoate to the background in the pulp. One of the consequences is

that when feeding ethanol, numerous ethyl ethanoates are produced, and it is difficult to differentiate whether the ethanoyl-CoA/ethanoate are from the fed ethanol or the background substrates. Thus, in Figure 3c, ethanol feeding was not shown. Thus, in bananas, there are rich sources of ethanoate/ethanoyl-CoA via TCA and β-oxidation, and ALDH is not the major way to produce ethanoate (Figure 4a).

It is well known that butanoate esters are the second most produced esters after ethanoate esters in banana fruits [30], and the high affinity of butanal to ALDH ensured sufficient butanoate (Figure 4a). However, ester profiles changed with fruit senescence [30,31], controlled/modified atmosphere [32–34], or other treatments which may extend shelf life with sacrifice of flavor quality [29]. Due to ester production being more dependent on alcohol and carboxylic acid substrate availability than substrate specificity of AAT [14,15,35], ALDH plays a key role in the ester profile and flavor quality of fresh fruit.

Feeding of 2-keto-4-methylpentanoic acid to banana pulps produced 3-methylbutanol, and 3-methylbutyl ethanoate, but no 3-methylbutanoate and the esters (Figure 7). The results confirmed that the affinity of ALDH for branched chain aldehydes is low. There are reports that some fruits such as apples and melons produce a relatively high amount of esters consisting of branched chain carboxylic acid [36,37]. It is worth continuing research on ALDH in branched chain carboxylic acid rich fruits.

The most critical question for this feeding model research was how to obtain high purity ALDH and minimize the effect of other enzymes, especially ADH, which is also determined by monitoring the change in NADH concentration. Furthermore, ADH facilitates the interconversion between alcohols and aldehydes with the redox between NAD<sup>+</sup> and NADH. In this study, two actions were taken to eliminate the potential effect of ADH. The first was to remove proteins that had high ADH activity but much less ALDH activity. We removed elutable proteins (protein-S) which contained 95% of ADH, but only about 40% of ALDH activity, and obtained proteins which had higher binding force to pellets—they were extracted under sonicator-assistance (Figure 5). In such pellet protein, there was little ADH activity (Figure 5). The second action involved using 4-methyl pyrazole, an ADH inhibitor, to block any potential ADH activity in the reaction mixture for ALDH activity determination (Figure 5).

#### **5. Conclusions**

ALDH, which converts aldehydes to carboxylic acids, was found in banana pulps and may play a key role in the conversion between alcohols, aldehydes, carboxylic acids and esters, and the formation of fruit aromas. Crude ALDH tests showed that the enzyme required NAD+ as a cofactor, and the optimum pH was 8.8. Lower molecular weight straight chain aldehydes, except ethanal, had high affinity to ALDH, while poor affinity was detected to branched chain aldehydes. Further research is needed to confirm whether ALDH is an enzyme in the routine pathway for volatile production associated with fruit ripening or just a consequence of aldehyde scavenging.

**Author Contributions:** Conceptualization, Y.U. and J.B.; methodology, Y.U., W.Z., H.I., Y.I., E.T., S.K.W., A.C. and J.B.; statistical analysis and the analysis of the results, Y.U. and J.B.; writing, Y.U., J.B., W.Z., H.I., Y.I., E.T., S.K.W. and A.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** Data is contained within the article.

**Acknowledgments:** This manuscript is dedicated to Yoshinori Ueda, in memory of his devotion in enzymology of fruit volatiles and postharvest plant physiology of horticultural crops. He will be greatly missed by those who were privileged to know him. We would thank Akira Wadano for his technical advice, and Hannah Clarke and Alice Bai for language improvement.

**Conflicts of Interest:** The authors declare no conflict of interest.
