Macromolecular and microstructural evolution of enzymatically purified banana albedo dietary fiber during ripening: Targeted valorization for functional and prebiotic applications

Abstract

The macromolecular and microstructural properties of enzymatically purified dietary fiber (EPDF) from banana albedo were affected by ripening. EPDF samples at mature green (MG), yellow green (YG), and fully yellow (FY) stages were compared. Enzymatic purification increased dietary fiber content 2.09-fold to 57.13%. Fourier Transform Infrared Spectroscopy (FTIR) showed changes in cellulose-, pectic-, and lignin-related bands at 1023, 1735, 1604, and 1513 cm⁻¹. X-ray diffraction showed increased relative crystallinity from 14.8% in MG to 17.1% in YG and 26.3% in FY-EPDF. Scanning Electron Microscopy (SEM) showed MG-EPDF had pores, a heterogeneous surface, the highest hydration and adsorption capacities (WHC 5.00 g/g; GAC 38.53 mg/g; CAC 8.46 mg/g). YG-EPDF showed the highest fermentation responses (ΔpH 2.42, TTA 0.577%, acetate 0.483 g/l). FY-EPDF had the highest SDF/IDF ratio (0.078) and relative crystallinity but lower hydration-related functionality. Ripening altered EPDF structural, techno-functional, and in vitro fermentation properties, indicating distinct functional profiles among stages.

Graphical abstract

Highlights

• •Enzymatic purification enriched banana albedo TDF by 2.09-fold.
• •Ripening increased EPDF relative crystallinity from 14.8% to 26.3%.
• •MG-EPDF showed greater water-holding and glucose/cholesterol adsorption.
• •YG-EPDF showed the highest fermentation responses (ΔpH, TTA, and acetate).
• •EPDF supported growth of L. plantarum and L. casei.

1. Introduction

Long-term solutions are needed to add value and advance the circular bioeconomy because numerous byproducts are generated globally through agricultural processing. Bananas are widely cultivated and consumed in tropical and subtropical regions, with global production reaching major peaks and generating substantial agricultural waste (Zaini et al., 2022). Specifically, banana peel accounts for 35%–50% of the total fruit mass, producing approximately 36 million tons of peel waste annually that is typically dumped into the environment (Zaini et al., 2022). Expanding on this global perspective, Asia dominates worldwide banana production, contributing approximately 51% to the total output, with Southeast Asia serving as a crucial hub for its cultivation and industrial processing (Zaini et al., 2025). Consequently, the banana sector currently yields over 57.6 million metric tons of banana peel waste annually (Zaini et al., 2025). Banana peel is often discarded after processing. During decomposition, it can produce unpleasant odors and gases that contribute to the greenhouse effect (Srivastava, 2023). Banana peel has also been used in value-added food products, such as biscuits, cookies, muffins, and pizza base (Srivastava, 2023).

Recognized as “the seventh nutrient” due to their essential roles in regulating physiological health (Zhang et al., 2024), dietary fibers have attracted growing interest as novel, health-promoting ingredients. Banana peel is one of the most common agricultural byproducts, accounting for 30–40% of the fruit weight (Saeed et al., 2024). Banana peels are becoming increasingly popular as a source of raw material for producing useful dietary fiber such as cellulose (7.6–9.6%), hemicellulose (6.4–9.4%), lignin (6–12%), and pectin (10–21%) (Hong et al., 2023). It has gained increasing attention as a low-cost source of dietary fiber and bioactive compounds, making it a promising candidate for various applications in the nutraceutical, pharmaceutical, and functional food industries (Anjum et al., 2014; Campos et al., 2024). Saeed et al. (2024) reported that dried banana peels contain up to 51.25% carbohydrates and 29.52% crude fiber. Ijaz et al. (2023) reported a total dietary fiber content (TDF) of 45.3%, comprising 34.23% insoluble dietary fiber and 15.63% soluble dietary fiber. Dietary fiber provides health benefits, including improving colon health and blood sugar control, as well as lowering blood LDL cholesterol levels through the effective removal of bile salts from the intestines (Ijaz et al., 2023). In addition, the extracts from banana peels show significant free radical scavenging activity, with up to 82.03% DPPH inhibition and high total phenolic contents of around 19.83 mg GAE/g (Ijaz et al., 2023). Because of their rich, quantifiable compositional profile, banana peels are recognized as a safe, low-cost raw material for extracting highly functional dietary fibers for the food industry.

Commercially, ‘Hom Thong’ (Musa acuminata, AAA) is important and widely consumed in Southeast Asia. The peel of this banana cultivar is distinguished by an albedo layer, a comparatively thick interior white pulp rich in complex carbohydrates. The banana albedo is the white inner layer of the peel. When dried and processed into banana peel albedo powder (BPAP), it is a high-fiber ingredient rich in insoluble dietary fiber, with moderate soluble fiber, minerals such as potassium, and phenolic antioxidants. The BPAP included 25.19 g of insoluble dietary fiber per 100 g. Furthermore, the soluble dietary fiber content was measured to be 6.20 g/100 g (Unal et al., 2025). The albedo portion is structurally more spongy and parenchymatous, with less pigment and slightly lower total fiber and mineral density than the whole peel. However, it is still dominated by insoluble polysaccharide cell-wall material (cellulose, hemicellulose, pectin with some lignin) plus associated phenolics. Although much research has been conducted on dietary fibers from whole banana peels, investigations concentrating just on the isolated albedo layer are limited. The albedo has a unique spongy, parenchymatous structure that contrasts with the epidermal layer. It potentially offers improved microstructural and functional advantages.

Dietary fibers (DF) denote carbohydrate polymers with a minimum of ten monomeric units that cannot be digested by the endogenous enzymes found in the small intestine. DFs are categorized into two types: soluble dietary fiber (SDF) and insoluble dietary fiber (IDF), which have major structural and functional distinctions. Both SDF and IDF contribute to maintaining physiological health. Plant cell walls are the primary source of SDF, which has good water solubility and gelling qualities that help regulate blood glucose, lipids, and inflammatory responses. SDF enhances prebiotic activity due to its fermentability by gut microbes in the colon, producing short-chain fatty acids (SCFAs) with prebiotic qualities (Zou et al., 2022). The SCFAs produced regulate appetite, influence insulin secretion, and control blood glucose levels (Hou et al., 2021). IDF is predominantly composed of cellulose, hemicellulose, and lignin. Cellulose is a linear glucose polymer with β-1,4-glycosidic linkages, hemicellulose is a heteropolysaccharide with branching units, and lignin is a three-dimensional network polymer composed of phenylpropane derivatives (Bocker & Silva, 2025; Chen et al., 2019). The IDF has a benefit glucose adsorption capacity (GAC), and this ability might be related to the porosity, surface area, and hydration properties (Elleuch et al., 2011; Niu et al., 2023; Jiang et al., 2022).

When establishing an appropriate extraction technique, several critical elements must be considered, including cost, time, environmental sustainability, purity, and yielding. Traditional acid or alkaline solvent extraction takes time and effort (Karimi et al., 2018; Margareta & Nyman, 2002). The enzymatic extraction produces more fiber from rice bran than the chemical method (33.97 vs. 67.53%) (Salehifar & Fadaei, 2011). The dietary fiber microstructure and content are influenced by processing conditions, which may have an impact on their physicochemical and functional characteristics (Kaur et al., 2021; Peerajit et al., 2012). Kaur et al. (2024) discovered that mango peels extracted using alkali and enzymatic methods had 42.4% (13.2% SDF and 29.4% IDF) and 66% (27.2% SDF and 38.8% IDF) total dietary fiber. Furthermore, the enzymatically extracted DF exhibited superior functional properties, such as higher water holding capacity (WHC), swelling capacity, and oil holding capacity (OHC), which are important attributes for food applications. Salehifar and Fadaei (2011) reported that the extraction method of rice bran dietary fiber greatly affected its fat binding capacity (FBC) (p < 0.05). The enzymatically extracted fiber yielded an FBC of 3.50 ml/g, notably greater than the 3 ml/g obtained from the chemically extracted fiber. The enzymatic method not only improved extraction efficiency but also enhanced the functional qualities of dietary fiber, making it more suitable for use as a functional food ingredient.

Fruit ripening is one of the many variables that affect changes in polysaccharides. The major components of banana peels (BP) were carbohydrates and fibers, which accounted for 61.34% and 16.56%, respectively (Khamsaw et al., 2024). According to Happi Emaga et al. (2007), the primary components identified in BP included dietary fiber (43–50%), fat (3–10%), protein (6–9%), and starch (3%). The ripening process, a biological process that occurs in bananas and is initiated by ethylene (Adi, 2024; Ng et al., 2013). Fruit development alters the structures and content of cell walls, resulting in visible changes in texture, color, taste, and hardness (Zhu et al., 2021; Sharif, Su, Chen and Qi, 2022a, Sharif, Su, Chen and Qi, 2022b). Cellulose, the major structural component of plant cell walls, is a linear chain of beta-1,4-linked glucan, stabilized by many intermolecular and intramolecular hydrogen bonding interactions that give rise to a rigid and crystalline microfibrillar network (Tibolla et al., 2014; Zuluaga et al., 2009). Hemicellulose is also a heterogenic polysaccharide. The structural polysaccharides greatly influence fruit ripening (Mohapatra et al., 2010). The hemicellulose matrix is composed of several monosaccharides structurally connected to cellulose microfibrils (Mohapatra et al., 2010). These polysaccharides interact strongly with the amorphous lignin components present in the natural banana albedo matrix. It is composed of a very dense and physically strong lignocellulosic structure (Tarrés et al., 2017). Besides their structural functions, the physicochemical properties of dietary fiber, such as hydration, swelling, and glucose adsorption capacity, are related to changes in glycemic responses (Hong et al., 2023). The starch content in green banana peels decreases from 14.0% to 2.6% as the fruit ripens. Free sugar levels increased from 1.4% to 33.2%. Hemicellulose content decreased from 5.3 to 0.4% as the fruit matured. On the other hand, lignin content increased dramatically from 7.3% to 13.3%. Cellulose levels remained rather steady throughout ripening (Happi Emaga et al., 2007).

Furthermore, the impact of ripening on the macroscopic composition of the whole peel is well-documented, whereas the microstructural and physicochemical alterations of EPDF obtained exclusively from the albedo during ripening remain largely unexamined. This unexplored influence on techno-functionality and in vitro physiological functions represents a significant research gap. Therefore, the aim of this study was to assess the impact of three different ripening stages (mature green, yellow-green, and fully yellow) on the structural, crystalline, physicochemical, techno-functional, and in vitro physiological properties of EPDF derived from the Hom Thong banana albedo. It was hypothesized that enzymatic treatment would enhance total dietary fiber content, whereas ripening would modify cell wall structure and insoluble-to-soluble dietary fiber distribution in banana peel. During ripening, insoluble dietary fiber was expected to decrease while soluble dietary fiber was expected to increase, consequently affecting the structural properties, techno-functionality, and fermentation behavior of the fiber.

2. Materials and methods

2.1. Raw materials and chemicals

The Hom Thong bananas (Musa acuminata, AAA group) came from a local market in Pathum Thani, Thailand. To evaluate the influence of ripening, the fruits were classified into three ripening stages based on established peel color and TSS criterias: mature green (MG, L* = 50 to 53, a* = −18 to −15 and b* = 30 to 35; TSS = 2 to 3.5°Brix), yellow green (YG, L* = 58 to 63, a* = −10 to −5 and b* = 40 to 45; TSS = 14 to 16.5°Brix), and fully yellow (FY, L* = 65 to 70, a* = −2 to +5 and b* = 48 to 55; TSS = 19 to 23°Brix). The food-grade enzymes of neutral protease (EC 3.4.21.62, enzyme activity of 80,000 U/ml, glucoamylase (EC 3.2.1.3, enzyme activity of 200,000 U/ml), and alpha-amylase (EC 3.2.1.1, enzyme activity of 40,000 U/ml) from Reach Biotechnology Co., Ltd. in Thailand were supplied. All other chemical reagents utilized in this study were of analytical grade.

2.2. Enzymatically purified dietary fiber (EPDF) preparation

The efficiency of enzymatic purification was evaluated in the preliminary EPDF extraction from a representative commercial banana albedo sample. After validation of the purification efficiency, EPDF was extracted from each banana ripening stage (MG, YG and FY) in three independent extraction batches (n = 3). For each extraction batch, approximately 3.5–4.0 kg of fresh banana peels were manually separated to yield 2.0 kg of fresh albedo layer as the starting material. Briefly, manually separated banana albedo was homogenized in distilled water at the ratio of 1:5 (w/v). The pH of the slurry was adjusted to 5.8 with 0.325 M HCl, and 0.05% (w/v) α-amylase was added. After 30 min in an 80 °C water bath, the mixture was allowed to cool to room temperature. The digest was centrifuged (4000 ×g, 15 min) and filtered to obtain the insoluble residue, which was washed with warm distilled water to remove the hydrolyzed starch by-products. The collected residue was resuspended in distilled water; the pH was adjusted to 4.0 with 0.325 M HCl and 0.05% (w/v) glucoamylase was added. The suspension was incubated at 60 °C for 30 min and then cooled. The mixture was centrifuged, filtered and washed with warm distilled water. To remove the residual proteins, the residue was resuspended in distilled water, the pH was adjusted to 7.0 with 5% NaOH and 1.0% (w/v) protease was added. After 30 min at 50 °C, the mixture was heated at 95 °C for 10 min to stop the enzymatic processes. Finally, the hydrolyzed mixture was centrifuged and filtered to isolate the insoluble fiber cake. The residue was sequentially washed three times with warm distilled water (60 °C) and four volumes of 95% ethanol to remove residual soluble sugars, peptides, and enzymatic by-products. The EPDF was dehydrated in a hot air oven at 60 ± 5 °C until a constant moisture content (<10%) was achieved. A control sample of manually separated raw albedo layer was dehydrated under 60 ± 5 °C without enzymatic treatment. All extraction yields and dietary fiber compositions were calculated on a dry weight basis (db). The raw albedo and EPDF samples were then milled into fine powders for further analysis.

2.3. Composition of EPDF

The fiber composition of EPDF was characterized through crude fiber and dietary fiber profile analyses.

2.3.1. Crude fiber determination

Crude fiber determination was performed following the AOAC official method using a Fibertec™ 8000 system (FOSS, Hillerød, Denmark) following the dietary fiber analytical approach described by McCleary (2023). In brief, 1.0 g of EPDF (W₀) was exposed to sequential digestion using 150 ml of 1.25% H₂SO₄ and 1.25% NaOH for 30 min each, with intermediate boiling distilled water washes. To neutralize the pH, the dietary fibers were washed with acetone before being dried in a hot-air oven at 105 °C for 3 h. The crucibles were cooled in a desiccator for 30 min and weighed to obtain the dried residue weight after digestion (W₁). The samples were then ashed at 550 °C for 3 h in a muffle furnace, cooled in a desiccator for 40 min, and weighed again to obtain the ash weight (W₂). The percentage of crude fiber was computed using the following Eq. (1).

$$\mathit{Crude\; fiber}\left(\%\right)=\frac{W_1-W_2}{W_0}\times 100$$ (1)

where W₀ is the initial sample weight (g), W₁ is the weight of the dried residue after acid and alkali digestion (g), and W₂ is the ash weight after incineration (g).

2.3.2. Dietary fiber profile (TDF, IDF, and SDF)

The total dietary fiber (TDF), insoluble dietary fiber (IDF), and soluble dietary fiber (SDF) fractions of the EPDF were determined using the enzymatic–gravimetric method according to AOAC Official Method 991.43 (AOAC International, 2023; McCleary, 2023). The EPDF conducted sequential enzymatic digestion with heat-stable alpha-amylase, protease, and amyloglucosidase. After digestion and filtering, the residue was the IDF fraction, which was washed with distilled water, 95% ethanol, and acetone, then dried at 105 °C and weighed. To precipitate the SDF fraction, 4 volumes of preheated 95% ethanol (60 °C) were added to the filtration. The resultant precipitation was filtered, washed, dried, and weighed. Both IDF and SDF values were quantitatively adjusted for residual protein and ash content, and TDF was determined by adding the corrected fractions.

2.4. Physicochemical and structural characterization

The physicochemical and structural character of EPDF were comprehensively evaluated by determining water activity, color parameters, microstructural features, and molecular functional groups.

2.4.1. Water activity (a_w) and the color

Water activity (a_w) was measured using an AQUALAB 4TE water activity meter (METER Group, Pullman, WA, USA). The color was evaluated using the benchtop spectrophotometer (ColorFlex EZ 45/0 LAV, HunterLab, Reston, VA, USA), and the results were expressed according to the CIELAB color space (L*, a*, b*).

2.4.2. Scanning electron microscopy (SEM)

The SEM was utilized to investigate the microstructural morphology of EPDF following the method described by Wei et al. (2024). Samples were bonded to carbon double-sided tape, sputter-coated with a gold‑palladium alloy for 90 s prior to observation. Microstructural characterization was conducted using a compact tabletop field emission scanning electron microscope (FE-SEM; Hitachi, Tokyo, Japan) operated at accelerating voltages between 3.0 and 15 kV, with micrographs acquired at magnifications ranging from 50× to 5000×.

2.4.3. Fourier transform infrared (FTIR) spectroscopy

The functional groups and molecular conformation of the EPDF samples were characterized using a Fourier Transform Infrared (FTIR) spectrometer (INVENIO S, Bruker, Germany) equipped with an Attenuated Total Reflectance (ATR) accessory featuring a diamond crystal. Briefly, the dried EPDF samples were placed directly onto the ATR crystal and uniformly pressed. The spectra were recorded in absorbance mode over a mid-infrared wavenumber range of 4000 to 400 cm⁻¹. To ensure a high signal-to-noise ratio, spectra were produced using an aggregation of 32 co-added scans at a spectral resolution of 4 cm⁻¹. Background spectra of the clean air were collected prior to each measurement under identical conditions. The resulting spectra were qualitatively analyzed to identify critical functional group transformations, particularly focusing on the absorption bands corresponding to the structural polysaccharides (cellulose and hemicellulose), esterified pectins, and aromatic vibrations of lignin associated with the ripening stages.

2.4.4. X-ray diffraction (XRD) analysis

The crystalline structures of the raw albedo layer (RA, control) and enzymatically purified dietary fibers from different ripening stages (MG-EPDF, YG-EPDF, and FY-EPDF) were investigated using an X-ray diffractometer (Bruker D8 Advance, Bruker AXS GmbH, Karlsruhe, Germany) equipped with Cu Kα radiation (λ = 1.5406 Å). The experiments were carried out at room temperature using a generator voltage of 40 kV and a current of 40 mA. The samples were scanned across a 2θ range of 5° to 80° using a step size of 0.02° and a scan speed of 1°/min. X-ray diffractograms were processed using the DIFFRAC.EVA evaluation program (Bruker AXS, Version 7.0). The built-in Crystallinity function of DIFFRAC.EVA was used to automatically calculate relative crystallinity (RC) based on the integrated diffraction areas of each pattern (Greasley et al., 2022; Pilz et al., 2024). The program calculates RC using the ratio of the Reduced Area (R) to the Global Area (G), as expressed in Eq. (2).

$$\mathit{RC}\left(\%\right)=\left(\frac{R}{G}\right)\times 100$$ (2)

where R is the Reduced Area and G is the Global Area obtained from the DIFFRAC.EVA crystallinity analysis. The RC values obtained directly from the software output were reported as relative crystallinity percentages.

2.5. Techno-functional properties

The techno-functional properties of EPDF were investigated through hydration, oil-holding, and swelling capacity analyses.

2.5.1. Water-holding capacity and oil-holding capacity

The water-holding capacity (WHC) and oil-holding capacity (OHC) of EPDF were determined using the methods described by Jia et al. (2019) and Abdul-Hamid and Luan (2000) with slightly modification. Briefly, 1.0 g of EPDF (M₀) was mixed with 30 ml distilled water for WHC or 40 ml soybean oil for OHC and equilibrated at room temperature for 18 h. For WHC determination, excess water was removed by filtration, whereas OHC samples were centrifuged at 3000 rpm for 20 min before decanting the supernatant oil. The hydrated or oil-retained residues were weighed (M₂). The values of WHC and OHC were expressed as g/g and calculated using the following Eq. (3).

$$\mathit{WHC}\mathit{or}\mathit{OHC}=\frac{M_2-M_0}{M_0}$$ (3)

where M₀ is the initial dry sample weight (g) and M₂ is the weight of the hydrated or oil-retained residue (g).

2.5.2. Swelling capacity

The swelling capacity (SC) was determined according to the method described by Qi et al. (2015) with slight modifications. Briefly, 1.0 g of EPDF was mixed with 10 ml of distilled water in a graduated cylinder. They were allowed to equilibrate for 18 h. The final swollen volume (V) was recorded. The SC was expressed as ml/g and calculated using the following Eq. (4).

$$\mathit{SC}=\frac{V}{M}$$ (4)

where V is the final swollen volume (ml) and M is the initial dry sample weight (g).

2.6. In vitro physiological properties

The in vitro physiological properties of EPDF associated with glucose and cholesterol adsorption were evaluated to assess its potential hypoglycemic and hypocholesterolemic functionalities.

2.6.1. Cholesterol adsorption capacity

The cholesterol adsorption capacity (CAC) was determined according to the methods described by Jiang, Zhang, et al. (2022) and Feng, Ameer, Jiang, and Ramachandraiah (2024) with minor modifications. A new egg yolk was mixed with nine parts of distilled water and blended by stirring. The 1.0 g of EPDF (M₀) was combined with 30 ml of diluted fresh egg yolk emulsion, and the pH was adjusted to 2.0. The combinations were blended uniformly and incubated at 37 °C for 2 h, thereafter being subjected to centrifugation at 6000 rpm for 15 min. A control egg-yolk emulsion devoid of EPDF was subjected to identical conditions. The control and post-adsorption supernatants were subjected to a five-fold dilution before cholesterol assessment by the o-phthalaldehyde (OPA) colorimetric technique. Standard solutions of cholesterol from 0.025 to 0.125 mg/ml were used to generate a standard curve for cholesterol, and absorbance was read at 550 nm. The regression formula obtained was y = 15.88x + 0.0008 (R² = 0.9992). Cholesterol concentrations of the control and post-adsorption supernatants were calculated using the standard curve. CAC was measured in mg/g and calculated using the following Eq. (5).

$$\mathit{CAC}=\frac{(C_0-C_{e)\times \mathit{DF}\times V}}{M_0}$$ (5)

where $C_0$ is the cholesterol concentration in the five-fold diluted control egg-yolk emulsion without EPDF (mg/ml), $C_e$ is the residual cholesterol concentration in the five-fold diluted post-adsorption supernatant (mg/ml), $\mathit{DF}$ is the dilution factor (5), $V$ is the volume of diluted egg-yolk emulsion used in the adsorption assay (30 ml), and $M_0$ is the dry sample weight (g).

2.6.2. Glucose absorption capacity

Glucose adsorption capacity (GAC) was determined according to the method described by Tan et al. (2024) with slight modifications. Briefly, 1.00 g of EPDF (W_S) was mixed with 30 ml of 18.02 mg/ml glucose solution (V_S) and the initial glucose concentration was designated as G₁. The mixture was incubated at 37 °C for 6 h and then centrifuged at 4000 rpm for 20 min. The supernatant was diluted 20-fold with distilled water prior to DNS color development. The residual glucose concentration was determined at 540 nm using the DNS colorimetric method. A glucose calibration curve was prepared over a concentration range of 0–1.00 mg/ml, yielding the equation $y=0.5601x+0.0014$ ($R^2=0.999$). The residual glucose concentration in the diluted supernatant was corrected using the dilution factor (DF = 20) in Eq. (6).

$$\mathit{GAC}=\frac{\left(G_1-(G_2\times \mathit{DF}\right))\times V_S}{W_S}$$ (6)

where $G_1$ and $G_2$ are the initial glucose concentration and the residual glucose concentration determined from the diluted supernatant, respectively; $\mathit{DF}$ is the dilution factor (20); $V_S$ is the volume of glucose solution; and $W_S$ is the dry sample weight.

2.7. In vitro bacterial growth and fermentation responses

The in vitro bacterial growth assay was adapted from the methods of Huebner et al. (2007) and Kustyawati et al. (2022) to evaluate the bacterial growth and fermentation response of EPDF. Two probiotic strains of L. plantarum TISTR 541 and L. casei TISTR 390 were used and E. coli ATCC 25922 was used as a non-beneficial reference bacterium. Carbohydrate-free MRS broth with 1% (w/v) sterilized EPDF was used as the fermentation medium. The 1% inulin (w/v) was used as a positive prebiotic control, while carbohydrate-free medium without carbohydrate supplementation served as a negative control. During the experiment, all bacterial strains were grown separately. The same carbohydrate-free MRS basal medium was applied for all strains to maintain the carbohydrate background consistent for comparison of their growth responses to EPDF. The 10⁶ CFU/ml of each bacterial culture was inoculated and incubated at 37 °C for 48 h (under anaerobic conditions for probiotic strains and aerobic conditions for E. coli). After fermentation, viable bacterial plate counts were determined and expressed as log₁₀ CFU/ml. The bacterial cells and insoluble materials were separated from the fermentation broth by centrifuging at 5000 ×g for 10 min at 4 °C. The supernatant was used for the determination of pH, total titratable acidity (TTA) and acetate concentration. TTA was determined by titration with 0.1 N NaOH and expressed as % lactic acid equivalents with a conversion factor of 9 for lactic acid (Mantzourani et al., 2023; Samtiya et al., 2024). Acetate concentration was determined enzymatically using a commercially available acetic acid assay kit (Megazyme, Bray, Ireland) according to the manufacturer's protocol and expressed as g/l (Lukova et al., 2020).

A modified prebiotic activity score (PAS) was used to summarize the differential growth response of the probiotic strains relative to E. coli. The score was calculated using the following Eq. (7).

$$\mathit{PAS}=\left(\text{ΔLog}\left(\text{probiotics}\right)-\text{ΔLog}\left(E.\mathit{coli}\right)\right)/\mathrm{Log}\text{inoculum}$$ (7)

where ΔLog(probiotics) represents the mean change in log₁₀ CFU/ml of L. plantarum and L. casei, ΔLog(E. coli) represents the change in log₁₀ CFU/ml of E. coli, and Log inoculum represents the initial inoculum level (6.0 log₁₀ CFU/ml).

2.8. Statistical analysis

Three independent extraction batches (n = 3) were used for the chemical analyses, and color parameters were measured with fifteen independent readings to obtain a representative average. Results are expressed as the mean ± standard deviation (SD). Statistical analysis was conducted using SPSS software. An independent samples t-test was used to evaluate the efficacy of enzymatic purification. Differences among the three ripening stages were analyzed using a one-way analysis of variance (ANOVA) followed by Duncan's multiple range test (p < 0.05). Finally, Principal Component Analysis (PCA) was applied to assess the multidimensional relationships among the physicochemical, microstructural, and techno-functional parameters.

3. Results and discussion

3.1. Efficacy of enzymatic treatment on EPDF

To extract biomolecules from natural sources using enzyme extraction is a green and efficient method (Nadar et al., 2018). To verify the efficacy of the enzymatic purification method, an independent samples t-test was performed in this study. The initial TDF of the raw albedo layer was 27.43% on a dry weight basis. The complex matrix of banana albedo, which has intracellular proteins, resistant starches, and other digestible carbohydrates, interferes with the amount of TDF. As can be seen in Fig. 1, enzyme purification had a significant effect on the content of dietary fiber (p < 0.05).

Fig. 1.

Effect of enzymatic purification on the dietary fiber composition of Hom Thong banana albedo. Note: Values are expressed as mean ± standard deviation (n = 3) on a dry weight (db) basis. Asterisks (*) indicate a significant difference between the raw albedo layer and the EPDF within the same fiber category, as determined by an independent samples t-test (p < 0.05). The values above the bars represent the fold increase in fiber content relative to the raw material. SDF = soluble dietary fiber; IDF = insoluble dietary fiber; TDF = total dietary fiber.

It climbed to 57.13%, representing a 2.09-fold increase for TDF. This was clearly represented in both IDF and SDF, which were 2.20-fold (from 22.69% to 50.01%) and 1.53-fold (from 4.67% to 7.12%), respectively. The enzymatic method surpasses traditional extraction procedures by enhancing efficiency and selectivity, apart from operating under moderate conditions, hence causing less structural damage. Nadar et al. (2018) suggested that enzyme protocol is also useful for the extraction of polysaccharides including dietary fibers since it improves extraction efficiency through breaking down plant cell walls and releasing internal components such as hemicelluloses and pectin. The study undertaken by Kaur et al. (2023) reported that the combination of enzymatic and physical treatments effectively disrupted the non-cellulosic matrix. Therefore, a significant recovery of up to 24.47% of orange peels fiber was observed. Moreover, Feng, Ameer, Jiang and Ramachandraiah (2024) observed that the enzymatic extraction method was able to solubilize 80.77% of dietary fiber from papaya peel. The acid extraction method was able to solubilize 77.87% of dietary fiber. Additionally, the fiber extracted using the enzymatic method showed better thermal stability, maintaining its structural integrity without significant heat-induced deterioration.

3.2. Extraction yield and composition of EPDF across ripening stages

Table 1 presented the extraction yield and composition of the EPDF obtained from Hom Thong banana albedo at three distinct MG, YG and FY.

Table 1.

The EPDF yield and composition at different ripening stages.

Ripening Stages	Yield (%)ns	Crude Fiber (%)	EPDF composition (%)			SDF/IDF Ratio
			SDF	IDF	TDF
MG	4.33 ± 0.08	9.22 ± 0.03a	2.96 ± 0.12c	55.19 ± 0.04ᵃ	58.15 ± 0.10a	0.054
YG	4.64 ± 0.10	7.27 ± 0.11ᵇ	3.32 ± 0.11ᵇ	45.34 ± 0.13ᵇ	48.66 ± 0.12b	0.073
FY	4.51 ± 0.01	4.82 ± 0.16ᶜ	3.39 ± 0.10a	43.55 ± 0.27ᶜ	46.94 ± 0.19c	0.078

Note: ns = non-significant; different superscript letters within columns indicate significant differences (p < 0.05). TDF is calculated as SDF + IDF. Values represent mean ± SD (n = 3).

No significant difference in the percentage extraction yield across ripening was observed (p > 0.05). They were in the ranges of 4.33 to 4.64%. Previous studies undertaken by Abou-Arab and Abu-Salem (2018), Saeed et al. (2024) and Khamsaw et al. (2024) reported a higher level of 13–16% extraction yield of dried whole banana peel fibers. This variance is primarily attributable to changes in the part of raw material used and the extraction methods. The previous study produced a crude powder, preserving intact proteins, digestible carbohydrates, and other non-fiber dry matter. In addition, enzymatic extraction promotes the degradation of non-fiber components, thereby enhancing the isolation of the target dietary fibers. This is supported by Kaur et al. (2023), who reported that enzymatic treatment increased the dietary fiber content of kinnow (Citrus reticulata) peels. The content of TDF can be significantly enhanced by 12.45–24.47% compared to conventional ultrasound-assisted extraction, probably due to the degradation of starch and protein contaminants by enzymes. The recovered EPDF contained 46.94–58.15% TDF, as directly determined in Table 1. The proximate composition of the recovered EPDF was additionally characterized. Protein, fat, ash, and moisture contents were 4.80, 3.50, 11.50, and 5.20% in MG, 4.20, 3.10, 12.20, and 5.40% in YG, and 3.70, 2.80, 12.80, and 5.60% in FY, respectively. The remaining carbohydrate fraction, calculated by difference after accounting for TDF, protein, fat, ash, and moisture, was 16.85%, 26.44%, and 28.16% in MG, YG, and FY, respectively.

Additional XRD analysis of the unpurified raw albedo showed a higher relative crystallinity of 48.6% (data not shown) than the EPDF samples (14.8–26.3%, as discussed in Section 3.3.4). The higher relative crystallinity of the raw albedo could be due to the presence of crystalline constituents such as native starch and ordered cellulosic domains. The lower relative crystallinity observed after enzymatic purification is consistent with a change in the relative amounts of crystalline and amorphous components during processing but does not directly indicate the removal of starch. Changes in the DF profile may be due to endogenous enzymatic restructuring of the cell wall during fruit ripening. As crude fiber and IDF are mainly composed of hard structural polysaccharides such as cellulose and hemicellulose, their decrease may be due to ripening-related modification of the cell wall polysaccharide matrix. As maturation progressed from the MG to the FY stages, a significant reduction in crude fiber (from 9.22 to 4.82%), TDF (from 58.15 to 46.94%), and IDF (from 55.19 to 43.55%) was observed, accompanied by a concurrent elevation in the ratio of SDF/IDF (Table 1). The sharp decrease in insoluble fraction is consistent with Yi et al. (2024). It was found that increased xylanase activity (4.6 folds) in ripened banana peel was accompanied by a decrease in xylan content to 20.80 mg/g dw. The results showed a sharp decrease in hemicellulose content (from 160.36 to 121.11 mg/g dw) in banana peels with ripening. Conversely, the portion of SDF significantly increased from 2.96 to 3.39% from MG to FY stages, as shown in Table 1. Wang et al. (2021) found an increase in water-soluble pectin content by 113.0% and a decrease in water-insoluble precursor and hemicellulose by 46.5% and 33.5%, respectively. Activities of pectin methylesterase (PME) and polygalacturonase (PG) associated with ripening have been associated with modification and solubilization of pectic polysaccharides (Wang et al., 2021). Similarly, Liu et al. (2025) reported that cellulase-producing fermentation increased SDF yield and was accompanied by changes of the physicochemical structure of the resulting SDF. These results indicate that the ripening stage is associated with significant changes in the dietary fiber composition of the banana albedo. It is an important factor for its potential food applications. Akter et al. (2024) have successfully developed a functional food with dynamic structural properties for Agri by-products. Prebiotic yogurt quality and functionality were improved by the addition of green banana peel. The changes in fiber types during ripening might affect the specific application of extracted fibers. The fecal bulking effects of the mature green fiber are significant due to its thickness and stiffness and could potentially be beneficial for maintenance of intestinal regularity. However, the higher relative amount of SDF in later ripening stages may lead to higher solubility and fermentability, which are important properties for its potential use as a fermentable dietary fiber component (Das et al., 2026).

3.3. Physicochemical and structural characterization

3.3.1. Color profile and water activity (a_w)

As shown in Fig. 2, the EPDF of the MG, YG and FY stages exhibited dark reddish-brown colors. The CIE L*, a* and b* color parameters showed substantial changes across maturity (p < 0.05), as can be seen in Table 2.

Fig. 2.

Visual appearance of the EPDF powders derived from Hom Thong banana albedo at different ripening stages. Note: MG: mature green; YG: yellow green; FY: fully yellow.

Table 2.

The color (L*, a*and b*) and water activity (a_w) EPDF derived from Hom Thong banana albedo at different ripening stages.

Ripening Stages	Color			awns
	L*	a*	b*
MG	34.19 ± 0.62ᵃ	6.75 ± 0.15ᵃ	14.65 ± 0.07ᵃ	0.22 ± 0.02
YG	26.02 ± 0.03ᵇ	7.61 ± 0.24ᵇ	13.31 ± 0.06b	0.22 ± 0.01
FY	26.17 ± 0.03ᵇ	7.80 ± 0.04ᵇ	12.66 ± 0.08c	0.23 ± 0.03

Note: ns = non-significant; different superscript letters within columns indicate significant differences (p < 0.05). L*, a* and b* refer to lightness, redness and yellowness. Values represent mean ± SD (n = 15 for L*, a* and b*; n = 3 for a_w).

The significantly highest L* = 34.19 was observed in MG with a gradual decline to 26.02 and 26.17, for YG and FY, respectively. Concurrently, the a* significantly increased from 6.75 in MG to 7.80 in FY stages. The yellowness parameter (b*) progressively declined from 14.65 to 12.66 during ripening process.

This was an indicator of advanced non-enzymatic browning, predominantly driven by the Maillard reaction, accelerating melanoidin formation (Silva et al., 2020; Starowicz & Zieliński, 2019). Water activity (a_w) is a key physicochemical property that affects powder product shelf life and microbiological stability. According to Tapia et al. (2020), the growth of pathogenic microorganisms, osmophilic yeasts, and xerophilic molds is inhibited at an a_w below 0.60. As shown in Table 2, the ripening stages had no significant effect on the a_w of EPDF (p > 0.05). They were consistently low within a narrow margin of 0.22 to 0.23 across all stages. Similarly, Hosseininejad et al. (2024) reported similarly low a_w values (a_w = 0.21) of functional powder persimmon peel. This a_w value is a point where microbial spoilage and chemical degradative reactions are greatly reduced. In addition, it is critical to produce free-flowing powder.

3.3.2. Microstructural analysis

To investigate the microstructural characteristics of DF, the surface morphology of DF was examined. The scanning electron micrographs of the EPDF at 1000× and 5000× magnifications show pronounced morphological variations across the physiological maturation stages, as shown in Fig. 3. The EPDF isolated from the MG stage had a very heterogeneous and rough surface, with many visible pores and micro-fissures (Fig. 3, A1 and A2). MG-EPDF had a high IDF content (55.19%). Goulao and Oliveira (2008) reported that the main rigid structural polysaccharides of green fruits are insoluble cellulose and hemicellulose. In agreement with this, the high IDF content in MG-EPDF may contribute to the heterogeneous and open surface morphology observed after dehydration. Therefore, the high number of visible pores, folds, and micro-fissures may favor fiber matrix–water interactions. This is in agreement with the water-holding and swelling properties reported for more open fiber structures (Ramaswamy et al., 2013).

Fig. 3.

Microstructural morphology of the EPDF derived from Hom Thong banana albedo at different ripening stages. Note: MG (A1-A2), YG (B1-B2), and FY (C1-C2), determined by scanning electron microscopy at 1000× and 5000× magnifications.

The surface morphology of the YG-EPDF was intermediate between the MG and FY stages (Fig. 3, B1 and B2). The surface was still irregular and heterogeneous, but the large visible pore-like openings identified in MG-EPDF were less evident. At higher magnification, the YG surface appeared more aggregated and compact while retaining a rough surface and small visible voids. These observations suggest a gradual change in the surface morphology from the mature green to the fully yellow stage. The transition to the FY stage, however, was characterized by a dense, smoother and more continuous surface with fewer visible pores and surface openings (Fig. 3, C1 and C2). During fruit ripening, the cell wall is exposed to endogenous enzymatic remodeling, which may contribute to gradual changes in the structural organization of the fiber matrix. The IDF content significantly decreased from 55.19% to 43.55% as fruit maturation advanced from the MG to FY stages, while the SDF increased. This compositional change is consistent with modifications of hemicellulose and pectin networks during ripening. Yi et al. (2024) reported a substantial decrease in hemicellulose content in banana peel during ripening, accompanied by increased xylanase activity. In addition, the activities of pectin methylesterase (PME) and polygalacturonase (PG) increased during fruit ripening and were associated with the modification and solubilization of insoluble pectic polysaccharides (Wang et al., 2021). Such ripening-associated modifications may reduce the mechanical strength of the cell wall. Such changes after extraction and drying may contribute to the dense and fused surface appearance observed in FY.

3.3.3. Fourier-transform infrared (FTIR) spectroscopy

To aid in understanding the qualitative structural changes of dietary fiber during ripening, FTIR spectroscopy was performed on the corresponding dietary fiber fractions obtained from each of the MG, YG, and FY stages. As shown in the FTIR spectra (Fig. 4), all EPDF samples exhibited relatively similar characteristic absorption patterns, suggesting the presence of a lignocellulosic framework. However, the distinct spectral changes that occurred during the ripening stages indicated structural changes, especially in spectral regions associated with polysaccharide and pectic components within the DF matrix.

Fig. 4.

FTIR spectra of EPDF derived from Hom Thong banana albedo at different ripening stages. Note: The spectra were vertically offset for clarity; characteristic absorption bands are indicated by their corresponding wavenumbers.

The MG-EPDF exhibited a well-organized lignocellulosic structure at the MG stage. The FTIR spectra revealed a broad absorption band at around 3340 cm⁻¹, which corresponds to O—H intramolecular and intermolecular stretching vibrations in lignin, cellulose, and hemicellulose (Javier-Astete et al., 2021). The carbohydrate fingerprint region (1200–890 cm⁻¹) showed a prominent signal at 1023 cm⁻¹ and a peak at 890–893 cm⁻¹, associated with polysaccharide and cellulose structures, respectively (Javier-Astete et al., 2021).The YG-EPDF showed the main absorption bands observed in the MG- and FY-EPDFs, indicating an intermediate spectral profile during ripening. The fingerprint polysaccharide bands, such as those at 1078, 1041, and 1023 cm⁻¹, were still present at the YG stage. However, changes in the overall spectral profile were already noticeable in comparison with MG-EPDF. YG-EPDF showed pectic-associated bands at 1735 and 1604 cm⁻¹. These findings are consistent with continuing changes in the cell-wall polysaccharide matrix throughout the intermediate ripening stage. The retention of the major characteristic bands suggests that the overall lignocellulosic framework remained present.

As ripening advanced from YG to FY, structural alterations in the cell-wall polysaccharide matrix may have occurred. Sharp aliphatic C—H stretching vibrations at 2923 cm⁻¹ (asymmetric stretching) and 2855 cm⁻¹ (symmetric stretching) may indicate ripening-related changes in the polysaccharide matrix. These spectral changes may reflect changes in the molecular environment of the methyl and methylene groups during ripening. Changes in the spectrum in the C—O and C—O—C stretching regions around 1023 cm⁻¹ may be associated with endogenous hydrolytic enzymes that modify complex carbohydrates (Skolik et al., 2019). This important carbohydrate fingerprint region is also affected by simultaneous modifications of some hemicellulosic and pectic components. Skolik et al. (2019) reported that spectral changes at 1078 cm⁻¹ and 1041 cm⁻¹ have been associated with changes in xyloglucan and arabinogalactan, respectively. Changes in xyloglucan have been linked with xyloglucan endotransglucosylase/hydrolase (XTH) activity and may contribute to cell wall loosening (Skolik et al., 2019; Wang et al., 2021). Changes in complex pectic side chains may be linked to arabinogalactan modification. The O—H stretching band in the 3600–3200 cm⁻¹ region was broadened during the FY stage. This change may be associated with changes in hydrogen-bonding interactions during advanced ripening (Hong et al., 2021; Sinanoglou et al., 2023).

Structural changes were also observed in the pectic components. Bands at 1735 cm⁻¹ and 1604 cm⁻¹ were assigned to esterified carbonyl (C O stretching) and carboxylate groups (COO⁻ stretching), respectively. These bands, representing different pectic-associated functional groups, were interpreted separately rather than as a direct spectral shift from 1735 to 1604 cm⁻¹. Their presence throughout the ripening stages is consistent with structural modification of pectic components. These bands correspond to functional groups commonly found in galacturonic acid-rich pectic polysaccharides. Galacturonic acid is a key component of the pectin backbone (Duggal et al., 2024; Sinanoglou et al., 2023). These spectral features are consistent with pectic modification and may be associated with the increase in the SDF/IDF ratio in FY samples. The progressive modification of the pectic and hemicellulosic networks suggested by the FTIR spectra was consistent with the X-ray diffraction (XRD) analysis. The XRD analysis showed an increase in the relative contribution of ordered domains during ripening (Section 3.3.4). Taken together, these findings are consistent with structural modification and possible changes in polysaccharide organization during ripening. The FY spectrum contained lignin-associated absorption bands, which are consistent with changes in the surrounding polysaccharide matrix. The band at 1513 cm⁻¹ corresponds to aromatic skeletal vibrations of lignin and extractives. The band at 1242 cm⁻¹ corresponds to C—O stretching and O—H in-plane vibrations of polysaccharides and lignin, respectively. These findings indicate that ripening was associated with structural rearrangement and modification of cell-wall polysaccharides rather than complete breakdown of lignocellulosic components.

3.3.4. X-ray diffraction (XRD) analysis and relative crystallinity

The X-ray diffraction (XRD) analysis was performed to evaluate the structural organization of EPDF in the ripening phases MG, YG and FY. As shown in Fig. 5, the diffraction patterns of the three EPDF samples were mostly similar, showing broad cellulose-associated diffraction characteristics in the ranges of around 15–18° and 21–23° (2θ) and a minor characteristic at 34–35° at (2θ). Similar diffraction areas were reported by Mishra et al. (2023) in banana peel-derived cellulosic materials, corresponding to the specific reflections of cellulose I-type structures. The stability of these diffraction features through the three ripening phases indicates that the main organized cellulose domains remained in EPDF during ripening.

Fig. 5.

X-ray diffractograms and relative crystallinity (RC) of the enzymatically purified dietary fiber (EPDF) derived from Hom Thong banana albedo at different ripening stages. Note: The shaded regions indicate the computationally separated amorphous background; (A) MG (RC = 14.8%); (B) YG (RC = 17.1%); and (C) FY (RC = 26.3%).

DIFFRAC.EVA area-based analysis of the XRD patterns showed differences in the relative proportions of crystalline contributions in the three EPDF samples. The relative crystallinity increased from 14.8% in MG to 17.1% in YG and 26.3% in FY-EPDF (Fig. 5). Thus, YG-EPDF was in an intermediate structural state compared with MG- and FY-EPDF, consistent with the progressive changes across the three ripening stages. The increase in relative crystallinity with ripening may instead reflect changes in the relative contribution of ordered and amorphous components of the cell-wall matrix. Banana peel undergoes substantial hemicellulose remodeling during fruit ripening. Yi et al. (2024) observed that ripened banana peel contained lower amounts of hemicellulose and xylan, and that the expression and activity of enzymes involved in hemicellulose metabolism were increased. Such changes suggest that the non-cellulosic polysaccharide matrix surrounding cellulose is progressively modified during banana ripening. Wang et al. (2021) reported that during kiwifruit ripening, water-insoluble pectin and hemicellulose decreased whereas water-soluble pectin increased, while the cellulose nanostructure remained relatively stable. These observations are consistent with the possibility that modification and solubilization of relatively amorphous pectic and hemicellulosic components, consistent with the distinct spectral changes observed in the FTIR analysis, proportionally increase the relative contribution of the remaining ordered domains within the EPDF matrix. This interpretation is also supported by structural studies of banana-peel-derived cellulosic materials. Mishra et al. (2023) noted that extraction of lignin and hemicellulose from banana peel was correlated with an increase in crystallinity, while the cellulose I structure was retained. The increased relative crystallinity measured in FY-EPDF may therefore be, at least partially, a consequence of ripening-related modification of amorphous cell wall components and not necessarily a reflection of formation of additional crystalline cellulose.

The XRD results provide further structural information that complements the SEM and FTIR observations. The relative crystallinity of YG-EPDF was intermediate, and FY-EPDF showed a higher value, concurrent with changes in surface morphology and FTIR spectral features during ripening. Taken together, these results are consistent with the ripening-associated reorganization of the EPDF cell-wall matrix involving modification of amorphous polysaccharide components and increased relative contribution of ordered domains at advanced ripening.

3.4. Techno-functional, in vitro physiological properties and PCA analysis of EPDF

The techno-functional and in vitro physiological properties of EPDF (MG, YG, and FY) derived from Hom Thong banana albedo were presented in Table 3. The properties of WHC, OHC, CAC and GAC were significantly different (p < 0.05) among ripening stages while SC was not significantly different. In general, the highest WHC (5.00 g/g), OHC (2.84 g/g), CAC (8.46 mg/g), and GAC (38.53 mg/g) were observed at the MG stage, followed by the YG and FY stages. In contrast, the SDF/IDF ratio gradually increased during ripening (Table 1), indicating gradual structural modification and partial solubilization of cell wall polysaccharides.

Table 3.

Techno-functional and in vitro physiological properties of EPDF derived from Hom Thong banana albedo at different ripening stages.

Ripening Stages	WHC (g/g)	OHC (g/g)	SC (ml/g)ns	CAC (mg/g)	GAC (mg/g)
MG	5.00 ± 0.06ᵃ	2.84 ± 0.12ᵃ	4.47 ± 0.12	8.46 ± 0.37ᵃ	38.53 ± 0.31ᵃ
YG	3.97 ± 0.06ᵇ	2.62 ± 0.03ᵃᵇ	4.47 ± 0.06	7.27 ± 0.39ᵇ	29.34 ± 0.07b
FY	3.59 ± 0.07ᵇ	2.41 ± 0.15ᵇ	4.30 ± 0.17	5.72 ± 0.26ᶜ	24.15 ± 0.17c

Note: ns = non-significant; different superscript letters within columns indicate significant differences (p < 0.05). Values represent mean ± SD (n = 3).

At the MG stage, the EPDF exhibited a rough and heterogeneous surface morphology that may contribute to its higher hydration and adsorption capacities, consistent with the SEM observations (Fig. 3, A1–A2). The rough, heterogeneous surface with visible pores, folds, and micro-fissures may facilitate interactions between the fiber matrix and water, contributing to water retention. The neutral sugar residues inside the polysaccharide framework may be linked to the rough and uneven surface shape, which may enhance surface reactivity and functional performance (Duggal et al., 2024). Dhar et al. (2025) showed that the pineapple waste dietary fibers treated enzymatically exhibit greater surface roughness and porosity, which enhanced hydration and adsorption. Likewise, Wachirasiri et al. (2009) reported that maintaining porous structures and surface area in banana peel dietary fiber was associated with higher WHC and OHC. These properties may facilitate water accessibility and retention of hydrophilic polysaccharides. The WHC (5.00 g/g) and OHC (2.84 g/g) obtained in the MG in the present study were also comparable to that report by Feng, Ameer, Jiang and Ramachandraiah (2024). The values of WHC (∼5.0 g/g) and OHC (∼2.8 g/g) were observed for enzymatically extracted papaya peel dietary fibers. The elevated WHC and OHC values observed in the MG stage may also be associated with the predominance of IDF. Similarly, Liu et al. (2025) reported that cellulase-producing fermentation modified the physicochemical structure of whole-wheat SDF and was accompanied by improvements in WHC, OHC, and cholesterol adsorption capacity. Although the modification process differed from physiological ripening, these findings further support an association between dietary fiber organization and its hydration- and adsorption-related functionalities. IDF-rich fractions generally exhibit stronger water- and oil-binding capacities, probably due to their multilayered fibrous structures and greater physical entrapment ability (Zhu et al., 2010). Fayaz et al. (2024) similarly established a strong positive correlation between OHC, porosity, and IDF content. They suggested that capillary structures support predominantly oil entrapment. In the present study, MG samples exhibited the highest IDF and TDF contents, which corresponded well with the strongest hydration properties. As ripening progressed toward the FY stage, the reductions in WHC and OHC were observed. This decline was consistent with the denser, more compact, and smoother surface morphology observed during fruit ripening (Fig. 3, C1–C2). Such morphological changes may be associated with the endogenous enzymatic modification of cellulose, hemicellulose, and pectic substances (Ali et al., 2004; Brummell, 2006). Consequently, the denser surface morphology may have reduced the accessibility of water- and oil-binding sites. This change may have contributed to the lower water and oil retention capacities observed in the FY samples. Similarly, Wen et al. (2016) reported reductions in hydration-related functionality according to the structural disruption of insoluble fiber matrices in enzymatically modified rice bran dietary fiber. The SC showed no significant difference among ripening stages, remaining within a relatively narrow range of 4.30–4.47 ml/g. Wen et al. (2016) proposed that swelling behavior is strongly associated with the structural organization and physicochemical characteristics of the fiber matrix. The ripening process was associated with partial modification of pectic substances, while the remaining rigid framework, particularly the cellulose matrix and lignin-associated components, retained sufficient structural integrity to maintain stable swelling capacity. The present FTIR analysis supported this interpretation by showing the persistence of distinct lignin-associated functional group absorption bands even at the advanced FY stage. Because lignin is relatively resistant to degradation, the persistence of lignin-associated structures may contribute to maintaining the structural integrity of the fiber matrix and thus the swelling performance across the ripening stages. In vitro physiological properties also followed a similar declining trend during ripening. The rough and heterogeneous structure of MG may contribute to its higher CAC and GAC. This morphology may provide greater accessibility to binding sites involved in molecular adsorption. Similarly, Dhar et al. (2025) reported that increased surface roughness and exposure of hydrophilic groups enhanced glucose adsorption and cholesterol-binding functionality of enzymatically modified queen pineapple dietary fibers. Moreover, Wen et al. (2016) presented that the adsorption of target molecules by enzymatically modified rice bran dietary fibers was controlled by the active binding domains. On the other hand, the smoother and denser morphology in FY might have limited adsorption accessibility and decreased the efficiency of physical entrapment.

To evaluate the multivariate interrelationships between techno-functional and in vitro physiological properties across ripening stages, a PCA biplot was generated, as can be seen in Fig. 6. The first two main components explained 99.05% of the total variance, with PC1 and PC2 explaining 85.85% and 13.20%, respectively. A distinct sequential segregation was observed along the PC1 axis, with MG on the positive side, FY on the negative side, and YG in the intermediate transitional region. According to PC1, it was strongly associated with WHC, OHC, CAC, and GAC vectors, all of which pointed toward the MG cluster. In particular, the WHC and GAC vectors appeared nearly superimposed. This positioning closely matches that of Fayaz et al. (2024), who showed that after viscosity increases, glucose absorption is intrinsically associated with fiber hydration. Structurally, sugar molecules can act as spacers, trapping physically or establishing hydrogen bonds with water in the swollen fiber network. The gradual modification and solubilization of pectic components during ripening were consistent with the movement of FY samples toward the negative region of PC1, where the overall functional performance was lower. These favorable hydration properties and structural characteristics of MG-EPDF highlight its potential as a functional texturizing and moisture-retaining ingredient in baked goods, which aligns with recent findings on the successful incorporation of fiber-rich banana peel into cookie and biscuit formulations to enhance nutritional and functional profiles (Shafi et al., 2022).

Fig. 6.

PCA biplot showing the multivariate relationships among techno-functional and in vitro physiological properties of EPDF derived from Hom Thong banana albedo at different ripening stages.

Note: The cumulative variance of the first two principal components is 99.05% (85.85% for PC1 and 13.20% for PC2).

3.5. In vitro bacterial growth and fermentation responses

Table 4 presents the in vitro bacterial growth and fermentation responses of EPDF produced from Hom Thong banana albedo at different ripening stages. Probiotic bacterial growth, PAS, ΔpH, and TTA were significantly higher in all EPDF samples than in the negative control (p < 0.05), while acetate was detected in all EPDF samples but not in the negative control.

Table 4.

In vitro bacterial growth, prebiotic activity score (PAS), and fermentation responses of enzymatically purified dietary fiber (EPDF).

Substrate/Fiber Sample	Viable Bacterial Count (log10 CFU/ml)			Prebiotic Activity Score (PAS)	Fermentation responses
					ΔpH (48 h)	TTA (% lactic acid equivalents)	Acetate (g/l)
	L. plantarum	L. casei	E. coli
Negative Control	6.28 ± 0.21b	6.15 ± 0.28b	7.85 ± 0.18a	−0.27 ± 0.04c	0.12 ± 0.03c	0.012 ± 0.003d	ND
MG	9.62 ± 0.19a	9.31 ± 0.22a	5.87 ± 0.25b	0.60 ± 0.05b	2.18 ± 0.08b	0.417 ± 0.006c	0.345 ± 0.005c
YG	9.95 ± 0.14a	9.48 ± 0.18a	5.52 ± 0.21b	0.70 ± 0.04a	2.42 ± 0.11a	0.577 ± 0.025a	0.483 ± 0.006a
FY	9.74 ± 0.23a	9.37 ± 0.25a	5.69 ± 0.28b	0.64 ± 0.05ab	2.31 ± 0.09ab	0.507 ± 0.012b	0.413 ± 0.012b
Inulin (Positive Control)	9.82 ± 0.17a	9.57 ± 0.15a	5.64 ± 0.19b	0.68 ± 0.04ab	2.28 ± 0.10ab	0.510 ± 0.010b	0.410 ± 0.010b

Note: Different superscript letters within columns indicate significant differences (p < 0.05). Values represent mean ± SD (n = 3). PAS = prebiotic activity score; TTA = total titratable acidity; ND = not detected under the assay conditions. For acetate, statistical comparisons were performed only among treatments with quantifiable values. L. plantarum = Lactobacillus plantarum; L. casei = Lactobacillus casei; E. coli = Escherichia coli.

Significant promotion compared to the negative control of growth of L. plantarum and L. casei, together with lower viable counts of E. coli, was observed in all EPDF. Moreover, the bacterial growth responses and PAS values of EPDF samples were generally comparable to those of commercial inulin. Comparable PAS values ranging from 0.60 to 0.70 were observed across all EPDF samples. YG-EPDF showed the highest mean PAS (0.70), which was significantly higher than that of MG-EPDF but not significantly different from FY-EPDF or the inulin control (p > 0.05). For instance, YG-EPDF supported the growth of L. plantarum and L. casei to 9.95 and 9.48 log₁₀ CFU/ml, respectively, which were not significantly different from the commercial inulin control. The growth of 2-strain probiotics in the inulin control was 9.82 and 9.57 log₁₀ CFU/ml. The YG-EPDF contained 48.66% TDF, providing a substantial dietary fiber substrate for bacterial fermentation. A comparable ability of banana-peel-derived fiber to support L. plantarum and L. casei growth has also been reported for partially hydrolyzed cellulose obtained from banana peel (Phirom-on & Apiraksakorn, 2021). It was consistent with the findings of Zahid et al. (2021), who reported that 42.39% total dietary fiber in 2% supplementation with freeze-dried banana peel powder increased Lactobacillus populations by approximately 1.25 log CFU/ml. Likewise, Ayele (2026) reported that green banana peel flour supported the proliferation of Lactobacillus rhamnosus GR-1 and Bifidobacterium longum BB536 up to 8.99 log CFU/ml after fermentation, further supporting the fermentability of banana-derived dietary fibers. More recently, green banana flour has also been shown to support the growth of Bifidobacterium animalis subsp. lactis BB-12 during fermentation (Thakur et al., 2026), further supporting the fermentability of banana-derived dietary fibers.

As reported by Qin et al. (2024), the increase of the SDF content of modified corn straw dietary fiber from 2.64% to 17.15% could improve fermentability, probiotic proliferation, and short-chain fatty acid synthesis during in vitro fermentation. A similar relationship between dietary fiber structure and fermentability may contribute to the responses observed in the ripened EPDF samples. The previous compositional investigation demonstrated an increase in the SDF/IDF ratio from 0.054 in MG to 0.073 in YG and 0.078 in FY. Dietary fiber modification has been associated with changes in fermentability and fermentation metabolite production (Liu et al., 2025; Qin et al., 2024). Similar findings were observed in the present study, where YG-EPDF showed a layered surface morphology with visible micro-fissures as observed by SEM. These structural characteristics occurred concurrently with the higher mean fermentation responses observed for YG-EPDF. Significantly, although FY-EPDF exhibited the highest SDF/IDF ratio (0.078), its PAS value (0.64) was comparable to that of YG-EPDF (0.70). This result suggests that the fermentation responses may not depend solely on the level of SDF but may also be associated with differences in fiber composition and structural characteristics. The more compact surface morphology observed in FY-EPDF coincided with fermentation responses that were generally comparable to those of the inulin control. In contrast, MG-EPDF, which contained the highest IDF content, showed the lowest mean PAS, ΔpH, TTA, and acetate values among the three EPDF samples, although it exhibited improved techno-functional and adsorption properties.

Fermentation-associated acid production was reflected by the reduction in pH in all EPDF samples. A notable pH decline (ΔpH = 2.42) was observed in YG-EPDF, which also exhibited high viable counts of L. plantarum and L. casei. The final pH of all EPDF samples ranged between approximately 3.43 and 3.65. This is consistent with the results of Tan et al. (2024), who reported that banana peel-supplemented fermentation media reached a final pH of approximately 3.43, along with exceeding 11 log CFU/ml of Lactobacillus growth. Carbohydrate fermentation generates short-chain fatty acids (SCFAs) and other acidic metabolites related to low pH levels. For further characterization of the fermentation response and metabolite production, total titratable acidity (TTA, expressed as % lactic acid equivalents) and acetate concentrations were quantified. YG-EPDF had significantly higher TTA (0.577%) and acetate concentration (0.483 g/l) than MG-EPDF, FY-EPDF, and the inulin control (p < 0.05). FY-EPDF had TTA and acetate values of 0.507% and 0.413 g/l, respectively, which were not significantly different from the inulin control (0.510% and 0.410 g/l, respectively; p > 0.05). The corresponding values for MG-EPDF were significantly lower, at 0.417% TTA and 0.345 g/l acetate. The negative control showed significantly lower TTA (0.012%) than all supplemented treatments (p < 0.05), while acetate was not detected under the assay conditions. The significantly greater TTA and acetate formation in YG-EPDF occurred together with the largest ΔpH, showing a consistent pattern among the measured fermentation responses.

The acidification observed in the present study is also consistent with recent fermentation of green banana flour with B. animalis subsp. lactis BB-12, in which bacterial growth was accompanied by increased titratable acidity and reduced pH (Thakur et al., 2026). The significant pH decreases, increased bacterial growth, and elevated organic acid levels are consistent with fermentation activity in the EPDF-supplemented treatments. In vitro colonic fermentation reported by Li et al. (2025) showed that SDF fractions produced more SCFA than IDF. An increase in fermentation products, such as propionic acid, accounting for up to 53.10% of total SCFAs, was observed. In addition, Bashmil et al. (2024) reported the formation of SCFAs during in vitro colonic fermentation of green banana, supporting the ability of banana-derived carbohydrate fractions to serve as substrates for microbial fermentation. These results are consistent with the higher fermentation responses for YG-EPDF, although the present study quantified acetate as a representative SCFA rather than a complete SCFA profile.

A reduction in viable counts from 7.85 log₁₀ CFU/ml in the negative control to 5.52–5.87 log₁₀ CFU/ml in EPDF samples was observed. Similar results were observed by Powthong et al. (2020) who reported that the indigestible carbohydrates from banana enhanced the activity of probiotics and simultaneously restricted the growth of pathogenic bacteria. In the present study, however, the lower E. coli counts are described as a differential bacterial growth response under the assay conditions rather than as direct inhibition by EPDF.

4. Conclusions

The physiological ripening of Hom Thong banana albedo was associated with distinct changes in the composition, structure, techno-functional properties, and fermentation responses of EPDF. MG-EPDF retained a heterogeneous surface with numerous visible pores, a relatively preserved lignocellulosic spectral profile, and the lowest relative crystallinity among the three ripening stages. These structural characteristics coincided with the greatest hydration- and adsorption-related functionality, suggesting its potential as a functional fiber ingredient in bakery products and other formulated foods in which moisture retention and texture are important. YG-EPDF represented an intermediate stage in both surface morphology and relative crystallinity, together with ongoing spectral changes associated with cell-wall polysaccharides. This stage showed the strongest measured fermentation responses among the three EPDF samples, as reflected by greater acidification, titratable acidity, and acetate formation, while PAS remained comparable across the EPDF samples and inulin. These characteristics potentially enable its use as a fermentable fiber ingredient in fermented dairy or plant-based food formulations. FY-EPDF showed the highest SDF/IDF ratio, a more compact and smoother surface with fewer visible openings, and the highest relative crystallinity consistent with more extensive reorganization of the cell-wall matrix during ripening. FY-EPDF also retained measurable fermentation responses, although its hydration-related functionality was lower than that of MG-EPDF. The higher proportion of soluble fiber indicates its possible application in beverages, soups, sauces, or other soft-textured food formulations where soluble dietary fiber is desirable. Overall, the progressive structural and functional differences observed from MG to YG and FY indicate that the maturity of banana albedo could be considered when producing EPDF with different functional profiles. Such stage-dependent utilization may allow a more efficient and differentiated valorization of banana-processing by-products for specific food applications.

Declaration of generative AI and AI-assisted technologies in the manuscript preparation process

During the preparation of this manuscript, the authors used Gemini for English language refinement and polishing, Chat GPT and Canva for graphical abstract design assistance. The authors reviewed and edited the content and took full responsibility for the final content of the published article.

CRediT authorship contribution statement

Sarinya Sangkasanya: Writing – review & editing, Writing – original draft, Visualization, Validation, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization. John Joven Baluyot Ramirez: Investigation, Formal analysis, Data curation. Chriseah Denise Gabriel Inocencio: Investigation, Formal analysis, Data curation. Phisut Naknaen: Writing – review & editing, Supervision.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Sarinya Sangkasanya reports financial support was provided by Thailand Science Research and Innovation. SARINYA SANGKASANYA reports a relationship with Thailand Science Research and Innovation that includes: funding grants. SARINYA SANGKASANYA has patent #Thai Petty Patent (Utility Model) Application No. 2403001699: Process for extracting functional fiber from mature green’Hom Thong’ banana peel pending to Rajamangala University of Technology Thanyaburi. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.