Oil palm fiber extract as a fermentable functional ingredient: in vitro gastrointestinal carbohydrate hydrolysis, feline fecal fermentation, and microbiota-associated metabolite production

1 Introduction

Cats (Felis catus) are among the most popular companion animals worldwide, with increasing ownership driven by demographic changes, urban lifestyles, and the strengthening emotional bond between pets and their owners (González-Ramírez and Landero-Hernández, 2021). The growing role of pets as family members has contributed to an expanding pet food market, including the development of specialized diets formulated to address specific nutritional and health-related conditions. As a result, nutritional management has become a central strategy for maintaining feline health and wellbeing.

Among nutrition-related disorders, obesity is one of the most prevalent conditions affecting cats, with reported prevalence exceeding 63% in New Zealand and approximately 40% in the United States. Feline obesity primarily results from excessive energy intake combined with reduced physical activity and is associated with metabolic dysregulation and hormonal imbalance. Obesity is also considered a state of chronic low-grade inflammation, which may impair tissue and organ function over time (Saavedra et al., 2024). Furthermore, obese cats have a two- to fourfold increased risk of developing diabetes mellitus, as inflammatory processes disrupt insulin secretion and reduce insulin sensitivity (Clark and Hoenig, 2021). These challenges highlight the importance of nutritional strategies that support metabolic function and gastrointestinal health.

Cats are obligate carnivores and rely primarily on animal-derived protein; however, evidence indicates that dietary fiber and fiber-associated bioactive compounds can beneficially influence feline gut health. Previous studies have shown that supplementation with fiber-bound polyphenols can improve gastrointestinal function in cats (Jewell et al., 2022), while inclusion of fermentable carbohydrates such as short-chain fructooligosaccharides promotes beneficial bacterial populations, including Bifidobacterium spp. (Hall et al., 2020). These findings suggest that selected dietary fibers and polyphenols can exert prebiotic-like effects in the feline gastrointestinal tract, particularly through modulation of hindgut fermentation.

In humans and other animals, consumption of dietary fibers and polyphenols has been widely associated with positive effects on gut microbiota composition and microbial metabolic activity, including enhanced production of short-chain fatty acids (SCFAs) (He et al., 2022). Traditionally, these compounds have been sourced from fresh plant materials; however, increasing emphasis has been placed on the valorization of agro-industrial by-products to improve sustainability and resource efficiency. The palm oil industry generates substantial quantities of solid and liquid residues, with oil palm fiber representing a major lignocellulosic by-product. Globally, palm oil processing produces approximately 428 million tons of waste annually, and oil palm fiber contains measurable levels of phenolic compounds (3.59 mg gallic acid equivalents/g), indicating its potential as a functional feed resource (Alexandri et al., 2019). Various extraction approaches, including biological, chemical, physical, and combined methods, have been applied to disrupt plant cell walls and enhance the release of polyphenols from such materials (Gan et al., 2021).

Oil palm fiber (OPF) extract therefore represents a promising functional feed ingredient due to its dietary fiber content, resistance to upper gastrointestinal digestion, fermentability in the hindgut, and associated polyphenols. Nevertheless, information regarding its digestibility, fermentation characteristics, and effects on feline gut microbiota and metabolic outputs remains limited, particularly under obese conditions. Accordingly, the present study aimed to evaluate OPF as a functional feed ingredient using an in vitro obese feline gastrointestinal model, focusing on upper gastrointestinal digestibility, colonic fermentation, gut microbiota modulation assessed by next-generation sequencing, and microbial metabolite formation characterized by LC-MS/MS. The study provides preliminary evidence relevant to the potential valorization of oil palm by-products as fermentable ingredients for further investigation in feline nutrition.

2 Materials and methods

2.1 Sample preparation

Oil palm fiber extract (OPF) was prepared using a hot-water and ultrasound-assisted extraction procedure according to Anaya-Esparza et al. (2023) with slight modification. Briefly, oil palm fiber was extracted with distilled water at a solid-to-liquid ratio of 1:30 (w/v) at 120 °C for 15 min. The extraction suspension was subsequently subjected to ultrasound-assisted extraction for 30 min at 60 kHz. The resulting extract was concentrated by evaporation at 75 °C to obtain a solution containing approximately 30% solids and was subsequently freeze-dried to obtain OPF. Distilled water was used to dissolve the freeze-dried OPF prior to subsequent analyses.

For the in vitro gastrointestinal digestion and colonic fermentation experiments, a commercially manufactured feline diet (CFD) and a commercially manufactured feline weight-control diet (WCD) were used as dietary reference materials. The CFD represented a conventional feline diet and was used as the base diet for OPF supplementation, whereas the WCD was included as a commercially formulated weight-control comparator relevant to the potential application of OPF in diets intended to support weight management in felines.

Prior to the experiments, both commercial diets were ground into smaller particles to obtain a homogeneous sample and facilitate their use in the simulated gastrointestinal digestion procedure. OPF was incorporated into CFD at a ratio of 1:19 (w/w) to produce the OPF-supplemented commercial feline diet (OPF+CFD), corresponding to a 5% (w/w) OPF inclusion level in the final formulation. The inclusion level was selected to obtain a final phenolic concentration of approximately 0.1% based on the total phenolic content (TPC) of OPF. The CFD, WCD, and OPF+CFD formulations were subsequently subjected to the same simulated gastrointestinal digestion and in vitro colonic fermentation procedures for comparative evaluation.

2.2 Proximate composition analysis

The nutritional composition of the oil palm fibre extract was determined using validated analytical methods based on the Association of Official Analytical Chemists (AOAC, 2016). The analyses included moisture, crude protein, crude fat, carbohydrate, and ash. All determinations were performed in triplicate, and the results were expressed on a fresh weight basis.

Crude protein was determined using the Kjeldahl method (STM No. 03-017; AOAC Method 981.10), while crude fat was quantified using STM No. 03-184 based on AOAC Method 2008.06 and the Oxford Instruments Application Note 17 for the determination of fat and oil in food matrices. Moisture and ash contents were determined according to STM No. 03-115 (AOAC Method 925.23) and STM No. 03-114 (AOAC Method 923.03), respectively. Carbohydrate content was determined following the Method of Analysis for Nutrition Labeling (1993), while the metabolizable energy value was calculated using STM No. 03-010 in accordance with AOAC Method 994.12.

2.3 Polyphenol analysis

2.3.1 Polyphenol profiling

Phenolic compounds were profiled using liquid chromatography–quadrupole time-of-flight mass spectrometry (LC-QTOF-MS) with an Agilent 1290 Infinity II LC system coupled to an Agilent 6545 Quadrupole-TOF mass spectrometer (Agilent Technologies, United States of America). Chromatographic separation was performed using a ZORBAX Eclipse Plus C18 Rapid Resolution HD column (100 mm × 2.1 mm i.d., 1.8 µm particle size). The mobile phases consisted of 0.1% formic acid in LC-MS-grade water (A) and 0.1% formic acid in LC-MS-grade methanol (B). Gradient elution was performed at a flow rate of 0.2 mL/min as follows: 0–5 min, 95% A and 5% B; 5–35 min, 95%–0% A and 5%–100% B; 35–40 min, 0% A and 100% B; 40–45 min, 0%–95% A and 100%–5% B; and 45–50 min, 95% A and 5% B. The injection volume was 2 μL, and the column temperature was maintained at 25 °C.

Mass spectrometric detection was performed using a Dual Agilent Jet Stream electrospray ionization source operated in both positive and negative ionization modes. Data were acquired in AutoMS/MS mode using fixed collision energies of 10, 20, and 40 V. The instrument was operated with reference-mass correction, using m/z 121.0509 and 922.0098 in positive-ion mode and m/z 112.9856 and 1033.9881 in negative-ion mode. Source parameters included a gas temperature of 325 °C, gas flow of 13 L/min, nebulizer pressure of 35 psig, sheath-gas temperature of 275 °C, sheath-gas flow of 12 L/min, capillary voltage of 4000 V, nozzle voltage of 2000 V, fragmentor voltage of 175 V, skimmer voltage of 65 V, and octopole RF peak voltage of 750 V.

Data acquisition and qualitative analysis were performed using Agilent MassHunter Workstation software. Putative phenolic compounds were assigned by searching the MassHunter METLIN Personal Compound Database and Library (PCDL) and METLIN Metabolite PCDL, version 8. Because compound assignments were based on mass-spectral and library/database matching rather than confirmation with authentic analytical standards, the detected compounds were considered tentatively annotated. The LC-QTOF-MS workflow was also applied to characterize phenolic metabolites generated during the subsequent in vitro fecal fermentation.

2.3.2 Total phenolic and flavonoid contents

Total phenolic content (TPC) was determined by mixing 20 µL of sample with 40 µL of 10% Folin–Ciocalteu reagent and incubating for 6 min. Subsequently, 40 µL of 7.5% Na₂CO₃ and 100 µL of distilled water were added, followed by incubation for 60 min prior to absorbance measurement at 765 nm. Results were expressed as milligrams of gallic acid equivalents per Gram dry weight (mg GAE/g sample) (Teh et al., 2021). Total flavonoid content (TFC) was determined by adding 40 µL of 10% AlCl₃ to 20 µL of sample and incubating for 6 min. Then, 100 µL of 1 M NaOH was added, followed by incubation for 30 min. Absorbance was measured at 415 nm using a Varioskan LUX microplate reader (Thermo Fisher Scientific, Waltham, MA, United States of America). Results were expressed as milligrams of quercetin equivalents per Gram dry weight (mg QE/g sample) (Doloking et al., 2022).

2.4 Antioxidant activity assay

The antioxidant capacity of OPF was evaluated using 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging, 2,2′-azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) radical scavenging, ferric reducing antioxidant power (FRAP), and oxygen radical absorbance capacity (ORAC) assays. DPPH, ABTS, and FRAP results were expressed as millimolar Trolox equivalents per Gram dry weight (mM TE/g), whereas ORAC results were expressed as micromolar Trolox equivalents per Gram dry weight (µM TE/g). Trolox was used as the calibration standard for all assays, and sample responses were calculated from the corresponding calibration relationships and normalized to sample mass.

2.4.1 DPPH radical scavenging activity

A total of 180 µL of 0.2 mM DPPH prepared in methanol was mixed with 20 µL of OPF extract. The mixture was incubated in the dark for 30 min, and absorbance was measured at 517 nm using a Varioskan LUX microplate reader (Thermo Fisher Scientific, United States of America). Trolox standards ranging from 0 to 400 μM/mL were analyzed in triplicate to construct the calibration curve. The calibration relationship was y = 0.154237x − 0.590756 (R ² = 0.9988), where y represents DPPH radical-scavenging response and x represents Trolox concentration (µM/mL). OPF samples were evaluated at different concentrations, and the 2.5 mg/mL preparation was used for the reported antioxidant-capacity calculation. Results were normalized to sample mass and expressed as millimolar Trolox equivalents per Gram dry weight (mM TE/g sample) (Kedir et al., 2023).

2.4.2 ABTS radical scavenging activity

The ABTS radical solution was prepared by reacting 7 mM ABTS with 2.45 mM K₂S₂O₈ in acetate buffer (pH 4.5) and incubating the mixture in the dark for 12–16 h. The resulting solution was diluted to an absorbance of 0.74 ± 0.03 at 734 nm. Ten microliters of OPF extract were mixed with 200 µL of ABTS radical solution and incubated for 7 min. Absorbance was measured at 734 nm. Trolox standards ranging from 0 to 500 μM/mL were analyzed in triplicate. The experimental calibration relationship used for quantification was y = − 0.0008629x + 0.95099 (R ² = 0.9844), where y represents absorbance and x represents Trolox concentration (µM/mL). OPF samples were evaluated at different concentrations using four replicate measurements, and the 5 mg/mL preparation was used for the reported antioxidant-capacity calculation. Results were normalized to sample mass and expressed as millimolar Trolox equivalents per Gram dry weight (mM TE/g sample) (Xiao et al., 2020).

2.4.3 Ferric reducing antioxidant power (FRAP) assay

The FRAP working solution was prepared by mixing 0.3 M acetate buffer (pH 3.6), 10 mM 2,4,6-tripyridyl-s-triazine (TPTZ) in HCl, and 20 mM FeCl₃·6H₂O at a ratio of 10:1:1. The working solution was maintained at 37 °C for 30 min before use. Five microliters of OPF extract were mixed with 180 µL of FRAP reagent and incubated at 37 °C for 15 min in the dark. Absorbance was measured at 593 nm. Trolox standards ranging from 0 to 500 μM/mL were evaluated, and the linear calibration relationship used for quantification was y = 0.000849x + 0.14076 (R ² = 0.9989). OPF extracts were evaluated at concentrations ranging from 0.25 to 15 mg/mL using four replicate measurements. The 0.75 mg/mL preparation was selected for quantification, and the resulting antioxidant capacity was normalized to sample mass and expressed as millimolar Trolox equivalents per Gram dry weight (mM TE/g sample) (Xiao et al., 2020).

2.4.4 Oxygen radical absorbance capacity (ORAC) assay

Twenty-five microliters of OPF extract were mixed with 150 µL of fluorescein working solution (81.6 nM) in a 96-well black opaque microplate and incubated at 37 °C for 30 min. Subsequently, 25 µL of AAPH working solution (153 mM) were added. Fluorescence intensity was recorded every 2 min for 90 min at excitation and emission wavelengths of 485 and 530 nm, respectively, using a Varioskan LUX microplate reader (Thermo Fisher Scientific, United States of America). Reference wells without AAPH were included for baseline correction (Siripongvutikorn et al., 2024). Trolox standards ranging from 0 to 100 µM were used to construct the calibration curve. The calibration relationship was y = 0.107097x + 0.334304 (R ² = 0.9968), where y represents net AUC and x represents Trolox concentration (µM). Fluorescence decay was integrated over the 90-min measurement period using the trapezoidal method. Net AUC was calculated by subtracting the blank/reference AUC from the corresponding AUC. OPF samples were analyzed at the selected concentration, and the resulting antioxidant capacity was expressed as micromolar Trolox equivalents per Gram dry weight (µM TE/g sample).

2.4.5 Thiobarbituric acid reactive substances (TBARS) assay

The TBARS assay was conducted using monkey kidney epithelial (Vero) cells following the method adapted from Queiroz Junior et al. (2021). Vero cells (1 × 10⁵ cells) were suspended in 0.1 M phosphate-buffered saline (PBS, pH 7.4), with butylated hydroxytoluene (BHT) used as a control. Samples were added and incubated for 15 min on ice. Cell suspensions (50 µL) were mixed with 50 µL SDS and incubated for 5 min, followed by addition of 125 µL thiobarbituric acid (TBA) reagent and incubation at 95 °C for 60 min. Samples were cooled on ice, centrifuged, and absorbance was measured at 532 nm. Results were expressed as micromoles of malondialdehyde per Gram dry weight (µmol MDA/g), and percentage inhibition was calculated as described by Manhiani et al. (2023).

2.5 Biological activities in cell culture

2.5.1 Cytotoxicity by MTT assay

The murine macrophage cell line RAW 264.7 was cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) and 100 μg/mL penicillin–streptomycin at 37 °C in a humidified atmosphere containing 5% CO₂. When the cells reached approximately 80% confluence, the culture medium was removed, and the cells were detached using 0.25% trypsin–EDTA. Cells were counted and adjusted to 5 × 10⁵ cells/mL, after which 100 µL of cell suspension was seeded into each well of a 96-well plate, corresponding to approximately 5 × 10⁴ cells/well. The cells were incubated for 2 h to allow attachment.

The culture medium was subsequently replaced with 100 µL of OPF extract at the specified concentrations, while RPMI-1640 without OPF was used as the untreated control. Following 24 h of treatment, 10 µL of MTT solution (5 mg/mL) was added to each well and the cells were incubated at 37 °C for a further 2 h. The resulting formazan crystals were dissolved in 200 µL DMSO, and absorbance was measured at 570 nm. Cell viability was calculated using Equation 1 (Siripongvutikorn et al., 2024). where OD_sample and OD_control represent the absorbance values of sample and control, respectively.

Cell viability % = OD sample / OD control × 100 (1)

2.5.2 Nitric oxide inhibition

RAW 264.7 macrophages were cultured under the conditions described above until approximately 80% confluence. Cells were detached using 0.25% trypsin–EDTA, counted, and adjusted to 5 × 10⁵ cells/mL. A volume of 100 µL of cell suspension was seeded into each well of a 96-well plate (approximately 5 × 10⁴ cells/well) and incubated for 2 h. The cells were subsequently stimulated with 200 μg/mL lipopolysaccharide (LPS) and treated with OPF extract at the specified concentrations. Appropriate controls containing RPMI-1640 with or without LPS were included. Following 24 h of incubation at 37 °C in 5% CO₂, 100 µL of culture supernatant was transferred to a new 96-well plate and mixed with 100 µL of Griess reagent. Absorbance was measured at 570 nm. NO inhibition was calculated using Equation 2, with correction for sample and blank absorbance, according to Siripongvutikorn et al. (2024). where OD_C, OD_Bc, OD_S, and OD_Bs represented the absorbance of control (RPMI + LPS), blank (RPMI), sample (sample + LPS), and sample blank (sample + RPMI), respectively.

NO inhibition % = OD C – OD Bc – OD S – OD Bs / OD c – OD Bc × 100 (2)

2.6 In vitro feline gastrointestinal model

2.6.1 Simulated gastrointestinal digestion

The simulated gastrointestinal digestion was conducted with slight modifications from Liu et al. (2023). One liter of simulated gastric fluid (SGF; pH 2.5) consisted of NaCl (1.986 g), KCl (0.97 g), NaHCO₃ (0.116 g), KH₂PO₄ (0.25 g), MgCl₂ (0.004 g), CaCl₂ (0.134 g), and urea (0.546 g). A 10% (w/v) sample suspension was added to 200 mL of SGF in a digestion vessel and incubated at 39 °C for 2 h with agitation at 65 rpm. Pepsin (8 mg; P7000, Sigma-Aldrich, United States of America) was added prior to gastric digestion. Simulated intestinal fluid (SIF; pH 6.5; 750 mL) consisted of NaCl (1.314 g), KCl (0.615 g), NaHCO₃ (1.965 g), KH₂PO₄ (0.655 g), MgCl₂ (0.008 g), CaCl₂ (0.2 g), urea (0.885 g), taurine (0.534 g), and bile salts (0.134 g). Porcine pancreatin (0.07 g; 3× USP, P1625, Sigma-Aldrich, United States of America) was added to 100 mL of gastric digesta along with 30 mL of SIF, followed by simulated small intestinal digestion at 39 °C for 4 h. Following digestion, the digesta were transferred to a water bath at 80 °C for 2 min to deactivate the enzymes and then cooled on ice. The digestion solution was subsequently placed into a dialysis membrane (MWCO 1 kDa; Spectra/Por® 6, Repligen, United States of America), immersed in reverse-osmosis water, and stored at 4 °C for 24 h. The dialysis step was used to remove low-molecular-weight digestion products and obtain the retained non-digestible fraction intended for subsequent colonic fermentation. However, compounds below the membrane molecular-weight cutoff, including some low-molecular-weight phenolics and soluble digestion products, may diffuse into the dialysis medium during this process. Therefore, the retained fraction was considered the analyzed non-digestible fraction rather than a complete representation of all compounds originally present in the digested sample. The retentate was subsequently freeze-dried and used for colonic fermentation.

2.6.2 Gastrointestinal digesta analysis

Gastrointestinal digesta were analyzed for total phenolic content (TPC) and total flavonoid content (TFC) to assess the relative bioaccessibility of phenolic and flavonoid compounds following simulated gastrointestinal digestion. Total sugar and reducing sugar contents were measured to evaluate carbohydrate hydrolysis. Total sugar content was determined using the phenol–sulfuric acid method by mixing 0.5 mL of sample with 0.5 mL of 5% phenol, followed by the addition of 2.5 mL concentrated H₂SO₄. The mixture was incubated for 30 min before absorbance was measured at 490 nm (Bhuyar et al., 2022). Reducing sugar content was determined using the 3,5-dinitrosalicylic acid (DNS) method. The DNS working solution was prepared by mixing 5% DNS in 2 M NaOH, 60% sodium potassium tartrate in distilled water, and distilled water at a ratio of 1:2.5:1.5. DNS working solution (0.25 mL) was added to 0.25 mL of sample and boiled for 10 min, followed by immediate cooling. Distilled water (2.5 mL) was then added before absorbance measurement at 540 nm (de Borba et al., 2022). Carbohydrate hydrolysis was calculated using Equation 3 according to Keawyok et al. (2023). where RS_initial and RS_final represent the reducing sugar concentrations before and after the digestion process, respectively, and TS_final represents the total sugar concentration after digestion. The resulting value represents the proportion of the carbohydrate fraction hydrolyzed under the simulated gastrointestinal conditions. The remaining fraction was calculated as 100 − hydrolysis (%) and was reported as the non-hydrolyzed carbohydrate fraction.

Carbohydrate hydrolysis % = RS final – initial / TS final − RS initial × 100 (3)

TPC and TFC were determined using the Folin–Ciocalteu and aluminium chloride colorimetric methods, respectively. TPC was expressed as milligrams of gallic acid equivalents per Gram of sample on a dry-weight basis (mg GAE/g), while TFC was expressed as milligrams of quercetin equivalents per Gram of sample on a dry-weight basis (mg QE/g). Relative phenolic and flavonoid bioaccessibility was calculated according to Mihaylova et al. (2021) using Equation 4. where C_f represents the TPC or TFC measured in the analyzed digestion fraction after the specified gastrointestinal digestion phase and C₀ represents the corresponding TPC or TFC measured in the undigested sample. Values were expressed on a dry-weight basis. Because TPC and TFC were determined using colorimetric assays, the calculated percentage represents relative changes in assay-detectable phenolic or flavonoid equivalents rather than absolute recovery of individual polyphenols.

Relative phenolic / flavonoid bioaccessibility % = C f / C 0 × 100 (4)

Because TPC and TFC were determined using colorimetric assays, the results represent total phenolic and flavonoid equivalents expressed relative to gallic acid and quercetin standards, respectively, rather than concentrations of individual phenolic compounds. Consequently, values exceeding 100% were interpreted as an increase in assay response relative to the undigested reference and not as recovery of more than 100% of the original polyphenols. Such increases may result from gastrointestinal matrix disruption, release of compounds previously associated with the food matrix, changes in compound extractability, or differences in the response of digestion-derived compounds to the colorimetric assays. A complete mass balance of individual phenolic compounds across digestion, dialysis, and recovery was not established; therefore, the calculated values were interpreted as relative bioaccessibility responses in the analyzed digestion fraction.

2.6.3 Simulated feline colonic fermentation

Nine cats were selected as fecal donors based on predefined eligibility criteria, including age between 1 and 8 years, indoor housing, body condition score (BCS) ≥4 on a five-point scale, no consumption of diets containing prebiotics or probiotics, absence of digestive disorders, no history of gastrointestinal surgery, and no medication use. Individual donor characteristics, including breed, sex, sterilization status, age, body weight, BCS, and housing condition, are presented in Table 1. Fecal samples were collected separately from each donor and subsequently pooled to obtain a composite fecal inoculum. The pooled fecal material was homogenized to prepare a single fecal slurry, which was used as the common inoculum for all fermentation treatments.

TABLE 1. Characteristics of fecal donor cats.

The basal fermentation medium was prepared according to Zhang et al. (2023) and contained peptone water (2 g/L), yeast extract (2 g/L), NaCl (0.1 g/L), NaHCO₃ (2 g/L), L-cysteine HCl (0.5 g/L), KH₂PO₄ (0.04 g/L), K₂HPO₄ (0.04 g/L), bile salts (0.5 g/L), CaCl₂·6H₂O (0.01 g/L), hemin (0.025 g/L), MgSO₄·7H₂O (0.01 g/L), resazurin (0.01 g/L), Tween 80 (2 mL), and vitamin K₁ (10 µL). The basal medium was sterilized at 121 °C for 15 min before use. For each treatment, 8 mL of the pooled fecal slurry was combined with 72 mL of basal fermentation medium, corresponding to a 10% (v/v) fecal inoculum and a fecal slurry-to-medium ratio of 1:9.

The nondigestible fraction recovered after simulated gastrointestinal digestion was used as the fermentation substrate. A fixed amount of 0.8 g of freeze-dried nondigestible material, expressed on a dry-weight basis, was added to each fermentation vessel. The treatments consisted of OPF, OPF+CFD, CFD, and WCD, while basal-medium fermentation (BMF), containing the pooled fecal inoculum and basal fermentation medium without an added nondigestible substrate, served as the control. Thus, fermentation was standardized using an equal mass of the recovered nondigestible fraction rather than an equal mass of the original diets. Because the nondigestible fractions differed in their chemical composition according to the original diet, equal substrate mass did not represent equivalent amounts of dietary fiber, phenolic compounds, or other potentially fermentable constituents. Consequently, treatment responses were interpreted as effects of the respective nondigestible fractions rather than as responses to chemically equivalent amounts of a specific nutrient or bioactive compound.

One fermentation vessel was prepared for each treatment using the same pooled fecal inoculum, and all treatments were fermented simultaneously under identical conditions. Anaerobic conditions were established by flushing the fermentation vessels with nitrogen gas. Fermentation was conducted at 39 °C for 24 h, with the pH maintained at 6.8 ± 0.1 using 0.1 M NaOH or HCl as required, following the procedure described by Wichienchot and Keawyok (2021). Samples were sequentially withdrawn from the same fermentation vessel at 0, 6, 12, and 24 h for subsequent analyses. The withdrawn aliquots were not returned to the fermentation vessels. Thus, sampling at the different time points represented repeated sampling of the same fermentation vessel rather than independent fermentation replicates. Microbiome sequencing, SCFA/BCFA determination, and phenolic-metabolite analyses were conducted using triplicate analytical measurements from the respective fermentation samples. Because these triplicates originated from the same fermentation vessel and the same pooled fecal inoculum, they were considered analytical replicates rather than independent biological replicates. Accordingly, the pooled fecal inoculum and corresponding fermentation vessel constituted the experimental unit for the in vitro fermentation experiment.

2.7 Microbiome and metabolome analysis

2.7.1 Gut microbiome profiling using 16S rRNA gene sequencing

Fermentation samples collected at the designated time points were subjected to microbial DNA extraction using the PowerSoil® DNA Isolation Kit. DNA quality was assessed by gel electrophoresis. The V3–V4 regions of the bacterial 16S rRNA gene were amplified using the primers 338F (5′-CCTACGGGNGGCWGCAG-3′) and 806R (5′-GACTACHVGGGTATCTAATCC-3′) with TransGen DNA polymerase. PCR amplification was performed with an initial denaturation at 95 °C for 2 min, followed by 20 cycles of denaturation at 95 °C for 45 s, annealing at 55 °C for 30 s, and extension at 72 °C for 30 s, followed by a final extension at 72 °C for 5 min. The amplified products were purified using the AxyPrep DNA Gel Extraction Kit (AXYGEN) and indexed using Nextera XT index primers. The purified PCR products were pooled, diluted to the appropriate loading concentration, and sequenced on an Illumina MiSeq platform using 300-bp paired-end reads, as described by Chuaykarn et al. (2024).

Raw sequencing data were processed using the QIIME2 platform (version QIIME2–202006). Sequence quality filtering, denoising, chimera removal, and amplicon sequence variants (ASV) inference were performed using the DADA2 algorithm. Taxonomic classification was conducted using a naïve Bayes classifier trained on the SILVA 138.2 reference database through the QIIME2 classify-sklearn plugin. Microbial diversity within samples (alpha diversity) was characterized using the Observed ASVs, Chao1 richness, and Shannon diversity indices. Community composition was characterized using UniFrac and weighted UniFrac distance metrics, with the resulting distance relationships visualized by principal coordinate analysis (PCoA). Because fecal samples from the nine cats were pooled to generate a single composite inoculum and one fermentation vessel was used for each treatment, triplicate microbiome measurements were considered analytical replicates rather than independent biological replicates. Accordingly, the microbiome diversity and community-composition results were interpreted descriptively rather than as inferential tests of treatment effects.

The gut microbiota sequencing data generated from the feline fecal samples have been deposited in the NCBI BioProject database under BioProject accession number PRJNA1491343.

2.7.2 SCFA and BCFA analysis by GC-FID

Short-chain fatty acids (SCFAs) and branched-chain fatty acids (BCFAs) produced during in vitro fecal fermentation were analyzed at 0, 12, and 24 h using gas chromatography coupled with a flame-ionization detector (GC-FID; Agilent 7890A, Agilent Technologies, United States of America) according to Chuaykarn et al. (2024). Fermentation samples were centrifuged at 12,000 rpm for 15 min at 4 °C, and the resulting supernatants were filtered through 0.22 µm nylon membrane filters before dilution in acetone. Acetic acid, propionic acid, isobutyric acid, butyric acid, isovaleric acid, and valeric acid were used as analytical standards. Sample and standard solutions (1 µL) were injected into the GC system equipped with an HP-INNOWax capillary column (19091N-113I; 30 m × 0.320 mm i.d.; 0.25 µm film thickness). Helium was used as the carrier gas at a flow rate of 1.5 mL/min. The oven temperature was initially set at 60 °C and increased to 100 °C at 4 °C/min, held for 2 min, and subsequently increased to 230 °C at 15 °C/min and held for 5 min. The total run time was 25.67 min. The injection was performed using a split ratio of 20:1, with the injector temperature maintained at 250 °C.

The SCFAs and BCFAs were identified by comparing the retention times of sample peaks with those of the corresponding analytical standards. The measured retention times were 14.017 min for acetic acid, 15.465 min for propionic acid, 15.868 min for isobutyric acid, 16.628 min for butyric acid, 17.086 min for isovaleric acid, and 17.773 min for valeric acid. The measured retention times closely corresponded to the expected retention times, with deviations ranging from 0.046% to 0.229%.

Quantification was performed using external standard calibration curves relating peak area to standard concentration. The calibration equations were generated using the linear model y = mx + b. The correlation coefficients were 0.99213 for acetic acid, 0.99753 for propionic acid, 0.99870 for isobutyric acid, 0.99820 for butyric acid, 0.99850 for isovaleric acid, and 0.99750 for valeric acid. The corresponding calibration equations were used to calculate the concentrations of individual SCFAs and BCFAs, which were expressed as millimolar (mM) (Chuaykarn et al., 2024).

2.8 Statistical analysis

All analytical results were expressed as mean ± standard deviation (SD). Statistical analyses were conducted using SPSS software version 21.0 (SPSS Inc., Chicago, IL, United States of America), with p < 0.05 considered statistically significant for analyses in which inferential testing was appropriate.

For proximate composition, cell viability, nitric oxide inhibition, gastrointestinal digestion, and phenolic bioaccessibility, statistical comparisons were performed where the available replicate structure supported inferential analysis. Where applicable, differences among treatments were evaluated using the appropriate analysis of variance followed by a suitable post hoc comparison, with statistical significance set at p < 0.05.

Microbiome alpha-diversity indices were summarized descriptively. Beta diversity was evaluated using weighted UniFrac distances and visualized by principal coordinates analysis (PCoA). PERMANOVA and tests of multivariate dispersion were not performed because the fermentation experiment did not provide independent biological replicates at the treatment level. Consequently, microbiome community patterns were interpreted descriptively rather than as statistically confirmed treatment effects. Taxonomic profiles, heatmaps, and core-ASV distributions were likewise interpreted descriptively.

SCFA and BCFA concentrations were evaluated individually, with changes interpreted relative to the corresponding 0-h values and basal-medium control. Because the available measurements did not provide independent biological replication, formal treatment-by-time inferential analysis was not performed. SCFAs and BCFAs were therefore interpreted separately according to their temporal concentration patterns rather than being combined as a single measure of beneficial fermentation.

LC-QTOF-MS results were analyzed qualitatively based on metabolite detection and relative peak responses. Metabolites identified without confirmation using authentic standards were reported as tentatively annotated. Peak areas were interpreted as relative analytical responses and not as absolute metabolite concentrations.

3 Results and discussion

3.1 Nutritional composition

The nutritional composition of oil palm fiber extract (OPF), commercial feline diet (CFD), OPF-supplemented CFD (OPF+CFD), and commercial weight-control diet (WCD) is presented in Table 2. The four formulations exhibited distinct macronutrient profiles. OPF contained 10.38 ± 0.26 g/100 g protein, 22.70 ± 1.40 g/100 g crude fat, 7.87 ± 0.12 g/100 g moisture, and 12.56 ± 0.24 g/100 g ash, with total carbohydrate calculated by difference at 46.43 ± 0.89 g/100 g. In comparison, protein concentrations were higher in the complete feline diets, ranging from 35.15 ± 0.22 to 36.40 ± 0.33 g/100 g, whereas crude fat was lower, ranging from 7.01 ± 0.18 to 12.84 ± 0.75 g/100 g. The higher carbohydrate and crude-fat contents of OPF distinguish its nutritional profile from the commercial diets and indicate that incorporation of OPF can modify the overall nutrient composition of the resulting formulation.

TABLE 2. Proximate composition and calculated energy content of the experimental diet groups.

OPF: oil palm fiber extract; OPF+CFD: OPF, with commercial feline diet; CFD: commercial feline diet; WCD: commercial feline weight-control.

The OPF+CFD formulation contained 36.01 ± 1.18 g/100 g protein and 12.84 ± 0.75 g/100 g crude fat, compared with 35.15 ± 0.22 and 10.30 ± 0.35 g/100 g, respectively, in CFD. Thus, addition of OPF to CFD was associated with an increase in the measured crude-fat and total-carbohydrate contents and a small increase in protein content relative to CFD. The calculated carbohydrate content of OPF+CFD was 34.88 ± 1.58 g/100 g, compared with 39.02 ± 0.61 g/100 g for CFD and 42.18 ± 0.41 g/100 g for WCD. These differences reflect the distinct ingredient compositions of the experimental materials rather than differences attributable to a single nutrient component.

The moisture contents were relatively similar among the complete diets, ranging from 5.85 ± 0.09 to 6.40 ± 0.07 g/100 g, whereas OPF had a slightly higher moisture content of 7.87 ± 0.12 g/100 g. Ash content was also higher in OPF (12.56 ± 0.24 g/100 g) than in the complete diets, which contained 8.00–9.87 g/100 g. Because ash represents the inorganic residue remaining after combustion, the higher value in OPF indicates a greater mineral-containing fraction in the extract compared with the commercial diet formulations. However, the individual mineral elements were not determined in the present study; therefore, the ash result should not be interpreted as evidence of a specific mineral contribution.

The calculated energy values were 365.44 ± 11.95 kcal/100 g for OPF, 399.08 ± 3.59 kcal/100 g for OPF+CFD, 389.40 ± 1.50 kcal/100 g for CFD, and 377.43 ± 0.98 kcal/100 g for WCD. The relatively lower calculated energy value of OPF compared with the complete diets is consistent with its different macronutrient distribution. Importantly, these values were estimated using the Atwater conversion factors applied to protein, carbohydrate, and lipid contents and therefore represent calculated metabolizable-energy estimates rather than directly measured feline digestible or metabolizable energy.

Overall, the proximate composition demonstrates that OPF differs nutritionally from the commercial feline diets, particularly with respect to carbohydrate and crude-fat contents. The incorporation of OPF into CFD produced an intermediate nutritional profile between the OPF ingredient and the original commercial diet for several measured parameters. These compositional differences are relevant when interpreting the subsequent in vitro gastrointestinal digestion and fermentation results because differences in the chemical composition of the substrates may influence the amount and type of material available to the simulated colonic microbiota. Nevertheless, the present proximate analysis does not by itself establish effects on feline energy metabolism, weight control, or gastrointestinal health.

3.2 Bioactive compound analysis

The bioactive compounds identified in OPF using LC–QTOF–MS are summarized in Table 3, including their retention times, peak intensities, and library match scores.

TABLE 3. Phenolic compounds identified in OPF.

Referred to Phenol Explorer Ver. 3.6 database (Rothwell et al., 2013).

Chromatographic peak heights and corresponding library match scores were obtained using Agilent MassHunter software. Identified compounds were predominantly classified as phenolic acids. Phenolic acids detected in OPF included compounds belonging to the hydroxyphenylacetic acid, hydroxybenzoic acid, and hydroxycinnamic acid subclasses. Among the identified phenolic compounds, pyrocatechol, 3,4-dihydroxybenzoic acid, and caffeic acid were quantified by HPLC at concentrations of 0.30, 0.45, and 0.10 mg/g OPF, respectively. These phenolic acids have previously been associated with metabolic and anti-inflammatory benefits in obesity-related models. Pyrocatechol was reported to suppress lipopolysaccharide (LPS)–induced inflammatory responses in macrophages by inhibiting nitric oxide production and the expression of pro-inflammatory cytokines (Funakoshi et al., 2020). Similarly, 3,4-dihydroxybenzoic acid (protocatechuic acid) was shown to attenuate body weight gain and improve insulin sensitivity in high-fat diet–fed rodents (Nour et al., 2021), whereas caffeic acid supplementation was associated with reduced weight gain and enhanced energy expenditure in obese mouse models (Xu et al., 2020).

3.3 Total phenolic and flavonoid content

The total phenolic content (TPC) of OPF was determined using a gallic acid calibration curve (y = 0.0055x + 0.0405, R ² = 0.9964) and was measured at 20.13 ± 0.02 mg GAE/g. The total flavonoid content (TFC) was quantified using a quercetin calibration curve (y = 0.7621x + 0.0175, R ² = 0.9925), yielding a value of 9.86 ± 0.21 mg QE/g. Compared with unprocessed oil palm pressed fiber, which was reported to contain approximately 3.6 mg GAE/g (Alexandri et al., 2019), OPF exhibited a markedly higher total phenolic content (TPC), increasing by up to 460.72%, reflecting the effectiveness of water and ultrasonic-assisted extraction in liberating bound phenolic compounds.

3.4 Antioxidant activity

The antioxidant activity of OPF was evaluated using complementary assays, including DPPH, ABTS, FRAP, and ORAC to assess radical scavenging capacity, as well as the TBARS assay to evaluate lipid peroxidation inhibition, thereby providing a comprehensive assessment of its antioxidant potential.

OPF exhibited DPPH radical scavenging activity of 68.34 ± 1.99 mM TE/g and ABTS radical scavenging activity of 80.74 ± 0.38 mM TE/g. The ferric reducing antioxidant power (FRAP) of OPF was 51.80 ± 0.37 mM TE/g. The oxygen radical absorbance capacity (ORAC) value was 698.07 ± 58.62 μM TE/g. Lipid peroxidation inhibition was assessed using the TBARS assay. At a concentration of 200 μg/mL, OPF resulted in malondialdehyde (MDA) formation of 1.65 ± 0.13 µM, corresponding to 8.27 ± 0.63 µmol MDA/g. The percentage inhibition of lipid peroxidation was 29.74% ± 3.63% relative to the control, which showed an MDA level of 2.36 ± 0.06 µM.

3.5 Cytotoxicity and anti- inflammatory activity

The cytotoxicity of OPF toward RAW 264.7 macrophages were evaluated using the MTT assay at concentrations ranging from 1.5625 to 800 μg/mL. The LPS-treated control exhibited a cell viability of 87.65% ± 0.50%. As shown in Figure 1, OPF treatment maintained cell viability above the predefined 80% threshold at concentrations from 1.5625 to 100 μg/mL, with viability values ranging from 82.24% ± 1.26% to 93.16% ± 3.22%. In contrast, viability decreased to 75.53% ± 0.44%, 70.15% ± 0.08%, and 62.66% ± 1.16% at 200, 400, and 800 μg/mL, respectively. These results indicate that concentrations above 100 μg/mL were associated with appreciable reductions in cellular viability.

The effect of OPF on nitric oxide (NO) production was subsequently evaluated in LPS-stimulated RAW 264.7 macrophages. OPF produced measurable inhibition of NO production across the tested concentration range, although the response was not strictly linear with increasing concentration. NO inhibition increased from 10.40% ± 1.03% at 1.5625 μg/mL to 19.63% ± 0.47% at 3.125 μg/mL and 23.47% ± 0.47% at 6.25 μg/mL, decreased to 16.16% ± 0.19% at 12.5 μg/mL, and subsequently increased to 33.07% ± 0.39%, 19.50% ± 1.56%, 42.29% ± 1.11%, 40.31% ± 0.37%, 47.80% ± 0.32%, and 67.55% ± 1.19% at 25, 50, 100, 200, 400, and 800 μg/mL, respectively. The calculated concentration corresponding to 50% inhibition of NO production (IC₅₀) was 425.37 μg/mL.

However, the calculated NO-inhibition IC₅₀ occurred within a concentration range associated with reduced cellular viability. In particular, cell viability decreased to 70.15% ± 0.08% and 62.66% ± 1.16% at 400 and 800 μg/mL, respectively, while NO inhibition at these concentrations was 47.80% ± 0.32% and 67.55% ± 1.19%. Therefore, the greater inhibition of NO production observed at higher OPF concentrations cannot be attributed exclusively to a specific anti-inflammatory effect and may partly reflect reduced cellular viability. This consideration is important because decreases in NO production measured at cytotoxic concentrations may arise from loss or impaired metabolic activity of viable macrophages rather than selective suppression of inflammatory signaling.

Accordingly, interpretation of the NO-inhibitory response was restricted primarily to concentrations at which cell viability remained above 80%. At 100 μg/mL, OPF produced 42.29% ± 1.11% inhibition of NO production while maintaining cell viability at 82.24% ± 1.26%. Thus, rather than being designated as an “optimal” concentration, 100 μg/mL was considered a concentration at which measurable NO inhibition occurred while cellular viability remained above the predefined 80% threshold. The higher concentrations were interpreted cautiously because their increased NO-inhibitory responses coincided with reductions in cell viability. Similar concentration-dependent cytotoxic responses have been reported for polyphenol-rich extracts at relatively high in vitro concentrations, where excessive exposure to phenolic compounds may promote cellular stress and pro-oxidant effects (López-García et al., 2014; Jankowska-Kieltyka et al., 2021).

3.6 In vitro gastrointestinal digestion

3.6.1 Digestibility

Carbohydrate hydrolysis during the simulated feline gastrointestinal digestion is presented in Table 4 and Figure 2. The digestion model consisted of a 2 h gastric phase followed by a 4 h small-intestinal phase. Following the small-intestinal phase, OPF + CFD, CFD, and WCD exhibited greater carbohydrate hydrolysis than that observed after the gastric phase, indicating increased release of reducing sugars during the intestinal digestion stage. In OPF, total sugar concentration increased from 73.37 ± 5.71 mg/mL after gastric digestion to 173.43 ± 0.99 mg/mL following the small-intestinal phase, whereas reducing sugar concentration increased from 56.28 ± 0.93 to 58.28 ± 0.78 mg/mL.

TABLE 4. Carbohydrate hydrolysis during simulated gastrointestinal digestion.

OPF: oil palm fiber extract; OPF+CFD: OPF with commercial feline diet; CFD: commercial feline diet; WCD: commercial feline weight-control.

Based on the reducing- and total-sugar measurements, OPF exhibited a mean carbohydrate hydrolysis of 4.01% ± 0.54%, corresponding to 95.99% ± 0.54% of the carbohydrate fraction remaining non-hydrolyzed under the simulated gastrointestinal conditions. In comparison, carbohydrate hydrolysis was 70.33% ± 0.58% for OPF+CFD, 78.27% ± 3.09% for CFD, and 57.46% ± 0.99% for WCD, corresponding to non-hydrolyzed fractions of 29.67% ± 0.58%, 21.73% ± 3.09%, and 42.54% ± 0.99%, respectively. Incorporation of OPF into CFD at a ratio of 1:19 resulted in a numerically higher proportion of non-hydrolyzed carbohydrate than CFD alone under the simulated gastrointestinal conditions.

The relatively low carbohydrate hydrolysis observed for OPF indicates limited enzymatic breakdown of its carbohydrate fraction under the simulated gastrointestinal conditions used in this study. This may reflect the structural complexity and limited enzymatic accessibility of some plant-derived carbohydrate components under the simulated gastrointestinal conditions. Complex carbohydrate matrices can restrict enzyme accessibility and consequently reduce the release of soluble and reducing sugars during digestion (He et al., 2022). However, because the present assay was based on changes in total and reducing sugar concentrations, the results should be interpreted as carbohydrate hydrolysis rather than direct measurement of dietary-fiber digestibility. The non-hydrolyzed fraction likewise represents carbohydrates that remained unhydrolyzed under the experimental conditions and should not be interpreted as an absolute measure of physiological indigestibility.

3.6.2 Phenolic and flavonoid bioaccessibility

The bioaccessibility-related responses of phenolic and flavonoid compounds following in vitro upper gastrointestinal digestion are presented in Figure 3 and Table 5. TPC and TFC were determined after the gastric and small-intestinal digestion phases for all diet groups. The TPC and TFC responses were generally higher after the gastric phase than after the small-intestinal phase. Based on the calculated relative bioaccessibility, CFD exhibited the highest values for TPC (146.51% ± 1.87%) and TFC (115.22% ± 5.89%), followed by WCD (TPC: 134.98% ± 0.99%; TFC: 98.37% ± 2.81%) and OPF+CFD (TPC: 125.49% ± 0.92%; TFC: 88.40% ± 3.78%). OPF showed the lowest calculated TPC bioaccessibility (97.94% ± 0.32%) but a TFC bioaccessibility of 94.00% ± 1.91%. Despite its lower relative TPC bioaccessibility, OPF retained higher absolute TPC and TFC responses after small-intestinal digestion than the other diet groups.

TABLE 5. Effects of in vitro gastrointestinal digestion on phenolic and flavonoid contents.

OPF: oil palm fiber extract; OPF+CFD: OPF, with commercial feline diet; CFD: commercial feline diet; WCD: commercial feline weight-control diet. The results are expressed as mg GAE/g for TPC, and mg QE/g for TFC. A–C: statistically different between digestion phases within a sample, a–c: statistically different between samples for TPC, or TFC.

The greater TPC and TFC responses observed after gastric digestion may be related to the acidic conditions and structural disruption of the food matrix during this phase. Acidic gastric conditions can promote matrix swelling, disruption of interactions between phenolic compounds and macromolecular components, and release of compounds that are initially associated with the food matrix. In contrast, the near-neutral to mildly alkaline conditions encountered during small-intestinal digestion may increase the instability or degradation of some phenolic compounds, contributing to the lower TPC and TFC responses observed after the intestinal phase. Similar effects of gastrointestinal conditions on the release and retention of phenolic compounds from plant-derived matrices have been reported during in vitro digestion studies (Luo et al., 2022). Therefore, the present results are best interpreted as changes in phenolic- and flavonoid-equivalent assay responses across digestion phases rather than as direct measurements of the fate of individual phenolic compounds.

The differences among the diet groups further suggest that the overall food matrix influenced the apparent accessibility of phenolic and flavonoid compounds during digestion (Wang et al., 2022). CFD and WCD showed higher calculated relative TPC bioaccessibility than OPF, whereas OPF showed a relatively high TFC response despite its lower calculated TPC bioaccessibility. These differences may reflect variation in the composition and structural characteristics of the respective substrates, including interactions between phenolic compounds and proteins, carbohydrates, lipids, and other matrix components. During digestion, changes in pH, enzymatic hydrolysis, hydration, and matrix disintegration can modify such interactions and consequently influence the extractability of phenolic compounds.

The different food matrices may therefore influence the apparent accessibility of phenolic compounds during digestion. OPF may contain matrix components capable of interacting with phenolic compounds and influencing their extractability; however, the specific contribution of individual dietary-fiber fractions was not determined in the present study. Therefore, the observed differences in TPC and TFC responses should be interpreted in relation to the overall composition and structure of the respective substrates rather than attributed to specific soluble or insoluble dietary-fiber fractions. This interpretation is particularly important because the measured TPC and TFC responses reflect the combined effects of matrix composition, digestion conditions, compound extractability, and assay response rather than the contribution of a single food component.

Importantly, the calculated bioaccessibility values exceeding 100% for several samples should not be interpreted as recovery of more than 100% of the original phenolic or flavonoid compounds. The calculation was based on the ratio of the TPC or TFC response measured after digestion to the corresponding response before digestion. Because TPC and TFC were determined using colorimetric assays and expressed as gallic acid and quercetin equivalents, respectively, these measurements represent assay responses rather than absolute concentrations of individual phenolic compounds. Values above 100% may therefore result from disruption of the food matrix, liberation of previously bound or poorly extractable compounds, changes in compound extractability, and differences in the response of digestion-derived compounds to the Folin–Ciocalteu or aluminium-chloride assays. Accordingly, the reported values should be described as relative TPC and TFC bioaccessibility responses rather than absolute recovery percentages. A complete mass balance of individual phenolic compounds across digestion, dialysis, and recovery was not established.

This distinction is important when considering OPF as a potential functional ingredient. Polyphenol-rich agro-industrial by-products have been investigated as potential sources of bioactive compounds for functional-food applications. Di Nunzio et al. (2018), for example, investigated a polyphenol-rich extract derived from olive-oil industry by-products using an intestinal-cell model and demonstrated that such by-products can contain chemically complex phenolic mixtures capable of eliciting measurable biological responses. However, these findings were obtained using an intestinal-cell model and should not be extrapolated directly to feline physiological outcomes.

Similarly, the presence or apparent enrichment of phenolic compounds does not necessarily translate directly into biological efficacy. Di Nunzio et al. (2020) reported that the biological effects of an olive-oil by-product incorporated into bakery products were influenced by processing and food-matrix characteristics, highlighting the importance of considering phenolic accessibility and transformation rather than relying solely on total phenolic measurements. Thus, the TPC and TFC responses obtained for OPF should be regarded primarily as indicators of phenolic- and flavonoid-equivalent responses under the experimental conditions, rather than direct evidence of physiological efficacy. The functional relevance of individual compounds depends on their chemical identity, matrix interactions, gastrointestinal accessibility, subsequent microbial transformation, and the concentrations that become available to the relevant biological system.

The dialysis procedure used after simulated gastrointestinal digestion should also be considered when interpreting the phenolic bioaccessibility-related responses and subsequent fermentation results. The 1-kDa molecular-weight cutoff membrane and 24-h dialysis against water were used to remove low-molecular-weight digestion products and obtain the retained nondigestible fraction for subsequent colonic fermentation. However, this procedure may also remove some low-molecular-weight phenolic compounds and other small soluble constituents capable of passing through the membrane. Consequently, the freeze-dried retentate used for subsequent fermentation represents the retained nondigestible fraction after the simulated gastrointestinal digestion and dialysis procedures, rather than the complete phenolic fraction originally present in the diet. This methodological consideration limits direct mass-balance interpretation and should be considered when relating the gastrointestinal TPC/TFC responses to subsequent colonic fermentation outcomes.

Overall, the results indicate that OPF retained measurable phenolic- and flavonoid-equivalent responses following simulated gastrointestinal digestion. The differences observed among OPF, OPF+CFD, CFD, and WCD indicate that the composition and structural characteristics of the food matrix may influence the apparent accessibility and extractability of these compounds during digestion. However, the colorimetric TPC and TFC assays do not establish the bioaccessibility, bioavailability, or physiological efficacy of individual phenolic compounds. The findings therefore support further investigation of OPF as a phenolic-containing agro-industrial by-product and potential functional-ingredient substrate, but they do not provide evidence for a specific health benefit in cats.

3.7 In vitro feline colonic fermentation

3.7.1 Gut microbiota composition

Amplicon sequence variant (ASV) distributions before and after fermentation across different diet groups are shown in Figure 4. A total of 872 ASVs were identified in this study, including 115 core ASVs shared among all samples. Prior to fermentation, the numbers of unique ASVs detected in BMF, OPF, OPF+CFD, CFD, and WCD were 84, 91, 113, 100, and 93, respectively. After 24 h of fermentation, the numbers of unique ASVs were 67, 72, 34, 50, and 53, respectively. The relative abundance of gut microbiota at the phylum level before fermentation (BMF 0 h) is presented in Figure 5. The predominant phyla were Bacillota (Firmicutes), Actinomycetota (Actinobacteria), Bacteroidota (Bacteroidetes), and Pseudomonadota (Proteobacteria). At the family level, the most abundant taxa were Erysipelatoclostridiaceae, Coriobacteriaceae, Peptostreptococcaceae, and Lachnospiraceae. The dominant genera included Collinsella, Catenibacterium, Peptoclostridium, Olsenella, and Enterococcus. After 24 h of fermentation, differences in gut microbiota composition were observed among diet groups. In BMF at 24 h, Pseudomonadota (Proteobacteria) represented the dominant phylum, with Escherichia–Shigella as the predominant genus, accompanied by a reduction in Bacillota (Firmicutes). In contrast, the OPF-, OPF+CFD-, CFD-, and WCD-fermented samples showed higher relative abundances of Bacteroidota (Bacteroidetes) and Fusobacteriota, along with a higher proportion of Bacillota (Firmicutes) compared with BMF at 24 h. Hierarchical clustering analysis of the top 35 most abundant genera is illustrated in Figure 6.

Prior to fermentation (BMF, 0 h), the microbial community showed relatively lower abundances of Bacteroides, Solobacterium, Parabacteroides, Negativibacillus, Blautia, and Fusobacterium. Previous studies have reported associations between the abundance of some of these taxa and feline metabolic health. For example, reduced Bacteroides abundance has been reported in obese felines and associated with alterations in carbohydrate and fatty acid metabolism and branched-chain amino acid catabolism (Yoshida et al., 2021; Zhang et al., 2025). Blautia has also been associated with microbial metabolite production and maintenance of intestinal microbial balance, including contributions to short-chain and branched-chain fatty acid production (Lee et al., 2022). Similarly, variations in Fusobacterium abundance have been reported in relation to feline obesity and metabolic characteristics (Kieler et al., 2019). These previous observations provide biological context for the microbial profile observed before fermentation but should not be interpreted as evidence of metabolic dysfunction in the pooled fecal inoculum used in the present study.

Following 24 h of fermentation, changes in the relative abundance profiles of several bacterial taxa were observed across the fermentation conditions. In particular, the OPF-containing treatments showed changes in the relative abundances of Bacteroides, Blautia, and Fusobacterium relative to BMF control. These observations indicate that fermentation in the presence of OPF was associated with shifts in selected components of the microbial community. Such shifts may reflect the availability of fermentable dietary fiber and polyphenolic compounds in OPF, which can provide substrates or bioactive compounds capable of influencing microbial community composition. Dietary fibers that are accessible to intestinal microorganisms can serve as substrates for saccharolytic bacteria and thereby influence microbial community structure and metabolic activity during fermentation (Moreno et al., 2022). However, the observed taxonomic changes should be regarded as exploratory associations rather than evidence of a direct prebiotic or metabolic effect of OPF.

Alpha-diversity profiles changed following fermentation, with the magnitude and direction of change varying among treatments and diversity indices (Figure 7). Chao1 richness, Shannon diversity, and Simpson diversity showed different numerical patterns between the pre- and post-fermentation observations. These changes describe the behavior of the microbial community within the fermentation system; however, because the treatments were generated from a common pooled fecal inoculum and each treatment was represented by a single fermentation vessel, no inferential statistical comparisons were performed. The observed differences should therefore be interpreted as descriptive changes within the experimental system rather than population-level biological effects.

Beta diversity based on weighted UniFrac distances is presented in Figure 8. Principal coordinate analysis (PCoA) showed distinct ordination patterns among the fermentation conditions, while the replicate profiles within each treatment × time combination generally clustered together. These patterns provide an exploratory representation of differences in microbial community structure associated with the fermentation conditions. Given the pooled-inoculum design and absence of independent biological fermentation replicates, the PCoA patterns were interpreted descriptively rather than as evidence of statistically significant differences between treatments.

3.7.2 Short-chain and branched-chain fatty acid production

The concentrations of individual short-chain fatty acids (SCFAs) and branched-chain fatty acids (BCFAs) during in vitro colonic fermentation are presented in Figure 9 and Table 6. The initial concentrations measured at 0 h were relatively high, with the combined concentrations of the measured SCFAs and BCFAs ranging from approximately 88–91 mM across the fermentation treatments. These 0-h concentrations represent the initial metabolite pool present in the fermentation system before incubation and therefore should not be interpreted as newly produced microbial metabolites. The initial pool may reflect fatty acids already present in the pooled fecal inoculum, basal fermentation medium, and nondigestible fractions carried forward from the simulated gastrointestinal digestion. Accordingly, changes in fatty-acid concentrations during fermentation were interpreted relative to the corresponding 0-h values rather than solely on the basis of absolute concentrations.

TABLE 6. Concentrations of individual short-chain and branched-chain fatty acids during in vitro colonic fermentation.

BMF: basal fermentation medium with feline fecal inoculum (control); OPF: oil palm fiber extract; OPF+CFD: OPF, combined with commercial feline diet; CFD: commercial feline diet; WCD: commercial feline weight-control diet. Values are expressed as mM and represent triplicate analytical measurements from the same fermentation vessel. Acetic, propionic, butyric, and valeric acids were classified as SCFAs, whereas isobutyric and isovaleric acids were classified as BCFAs. Total measured fatty acids represent the sum of the six quantified acids. Because one fermentation vessel was used per treatment, the analytical replicates were not independent biological replicates; therefore, no inferential statistical comparisons were made among treatments or fermentation times.

The combined concentration of measured SCFAs and BCFAs increased during fermentation in all treatments. In the OPF treatment, the concentration increased from 89.39 mM at 0 h to 121.89 mM at 12 h and 144.51 mM at 24 h, corresponding to increases of 32.50 and 55.12 mM relative to the initial concentration, respectively. Similarly, OPF+CFD increased from 88.65 mM at 0 h to 98.18 and 129.45 mM at 12 and 24 h, respectively, while CFD increased from 87.57 mM to 92.38 and 120.29 mM. WCD increased from 91.39 mM at 0 h to 103.81 and 130.44 mM at 12 and 24 h, respectively. Importantly, the BMF control also showed an increase from 88.43 mM at 0 h to 94.35 mM at 12 h and 119.76 mM at 24 h, indicating that part of the temporal increase in fatty-acid concentrations occurred in the absence of an added nondigestible substrate. Therefore, absolute increases in the treatment groups should not be interpreted entirely as substrate-derived microbial production.

After accounting for the change observed in the BMF control, the 24-h increase in the combined measured fatty-acid concentration above the corresponding BMF change was approximately 23.79 mM for OPF, 9.47 mM for OPF+CFD, 1.38 mM for CFD, and 7.71 mM for WCD. The comparatively larger BMF-adjusted increase observed for OPF indicates a greater substrate-associated change in the measured fatty-acid pool under the conditions of the present fermentation model. However, because the fermentation experiment used one vessel per treatment, these values are interpreted descriptively rather than as statistically confirmed treatment effects.

The individual fatty acids were considered separately because SCFAs and BCFAs represent different aspects of microbial metabolism. Among the principal SCFAs, acetate, propionate, and butyrate showed increases during fermentation in the OPF treatment. Acetate increased from 33.02 mM at 0 h to 52.50 mM at 24 h, while propionate increased from 17.06 to 26.14 mM and butyrate from 13.48 to 23.59 mM. Valerate also increased from 9.09 to 14.47 mM over the same period. The increases in acetate, propionate, and butyrate are compatible with microbial fermentation of fermentable substrates present in the nondigestible fraction. Dietary fibers and other complex carbohydrates can provide substrates for saccharolytic microorganisms and contribute to SCFA formation during colonic fermentation (Rios-Covian et al., 2020; Shin et al., 2024). Nevertheless, the present experiment does not establish a direct causal relationship between changes in specific bacterial genera and the production of individual SCFAs.

BCFAs were evaluated separately from the principal SCFAs because isobutyrate and isovalerate are commonly associated with microbial degradation of branched-chain amino acids and therefore can reflect proteolytic fermentation. In the OPF treatment, isobutyrate increased from 8.54 mM at 0 h to 13.76 mM at 24 h, whereas isovalerate increased from 8.20 to 14.05 mM. Thus, the fermentation profile comprised both SCFA formation associated with carbohydrate utilization and BCFA formation associated, at least in part, with protein-derived substrates. The increases in isobutyrate and isovalerate were therefore not interpreted as equivalent to increases in beneficial saccharolytic SCFAs.

At 24 h, OPF+CFD exhibited higher concentrations of propionate, butyrate, isobutyrate, and isovalerate than CFD, while the combined concentrations of the measured SCFAs and BCFAs were 129.45 ± 1.04 mM for OPF+CFD and 120.29 ± 0.55 mM for CFD. OPF+CFD and WCD showed comparable combined concentrations at 24 h (129.45 ± 1.04 and 130.44 ± 0.38 mM, respectively). OPF+CFD also exhibited a higher butyrate concentration than WCD (21.66 ± 0.31 vs. 18.64 ± 0.03 mM), whereas differences in acetate, valerate, and isovalerate were less pronounced.

Because a single fermentation vessel was used for each treatment, with the same pooled fecal inoculum used across treatments, the triplicate measurements represented analytical replicates rather than independent biological replicates. Consequently, an inferential treatment × time model was not applied to the SCFA and BCFA data because treating the analytical replicates as independent biological observations would constitute pseudoreplication. Instead, the individual fatty-acid concentrations, temporal changes relative to 0 h, and BMF-adjusted changes were interpreted descriptively. These findings therefore describe changes in the fermentation metabolite profile within the common pooled fecal inoculum and should not be extrapolated to donor-level responses.

3.7.3 Phenolic metabolite production

The LC-QTOF-MS profiles obtained before and after fermentation are summarized in Table 7. At 0 h, several compounds were already detectable in the fermentation system, including amino acids, organic compounds, and other metabolites, indicating that the baseline inoculum contained a diverse range of detectable metabolites. Following 24 h of fermentation, several metabolites remained common across the OPF, OPF+CFD, and CFD treatments. In addition, a subset of compounds was tentatively annotated only in the OPF-containing treatments, including dihydroisoferulic acid, 3-(2-hydroxyphenyl) propionic acid, trans-2-hydroxycinnamic acid, 3,4-dihydrocoumarin, and 7-methoxycoumarin-4-acetic acid.

TABLE 7. Metabolites produced during in vitro colonic fermentation using obese feline fecal inocula.

All metabolites were detected in all samples (OPF, OPF+CFD, and CFD).

The metabolites were only found in OPF-added treatment after 24 h of colonic fermentation.

The appearance of these compounds in the OPF-containing treatments after 24 h, together with their absence from the CFD treatment under the analytical conditions used, suggests that OPF inclusion may have been associated with the presence of these phenolic-related compounds following fermentation. However, detection at 24 h does not by itself demonstrate microbial formation or establish a precursor–product relationship. Although caffeic acid was detected in the OPF extract and the detected hydroxyphenylpropionic acid- and coumarin-related compounds are chemically consistent with possible phenolic transformation pathways, direct conversion of caffeic acid into these compounds was not demonstrated in the present study. Therefore, this relationship is considered a proposed pathway rather than a confirmed microbial biotransformation.

Because the LC-QTOF-MS analysis relied on accurate-mass and library/database matching without confirmation using authentic standards, the compounds are reported as tentatively annotated. Furthermore, the present analysis was based on qualitative detection and peak area rather than absolute metabolite quantification. Thus, differences in peak area should not be interpreted as quantitative changes in metabolite concentration without appropriate calibration and validation. Future studies incorporating quantitative time-course measurements, authentic standards, appropriate substrate and inoculum controls, and isotope-tracing approaches would be required to confirm the formation and microbial origin of these metabolites.

Taken together, the findings of the present study provide preliminary in vitro evidence that OPF contains fermentable and phenolic-associated components capable of influencing microbial and metabolite responses under simulated gastrointestinal and colonic fermentation conditions. However, these findings should not be interpreted as evidence of weight-control efficacy, obesity management, or clinical gastrointestinal or metabolic benefits in cats. The present study did not directly assess body-weight change, food intake, adiposity, insulin sensitivity, systemic or intestinal inflammation, gastrointestinal health outcomes, palatability, feeding tolerance, long-term safety, contaminant exposure, or in vivo toxicological responses. In addition, the colonic fermentation experiment used a common pooled fecal inoculum and one fermentation vessel per treatment, with subsequent measurements representing analytical rather than independent biological replicates. Therefore, the observed microbial and metabolite responses should be interpreted within the context of the pooled in vitro fermentation model and should not be extrapolated directly to individual cats or clinical outcomes. Further controlled in vivo feeding studies, together with appropriate contaminant, toxicological, palatability, and long-term safety assessments, are required to determine whether the observed in vitro responses translate into meaningful physiological effects and to establish the practical suitability of OPF as a feline dietary ingredient.

4 Conclusion

This study provides preliminary in vitro evidence that OPF can withstand simulated gastrointestinal processing and subsequently undergo fecal fermentation, resulting in changes in fermentation-associated metabolites and microbial community profiles. OPF exhibited a distinct proximate composition from the commercial feline diets and showed measurable phenolic and flavonoid-equivalent responses following simulated gastrointestinal digestion. The relatively low carbohydrate hydrolysis observed during simulated digestion indicates that a substantial proportion of the carbohydrate fraction remained available for subsequent fermentation under the experimental conditions. During fermentation, OPF was associated with changes in individual SCFAs, including acetate, propionate, and butyrate, as well as BCFAs. These SCFAs and BCFAs were interpreted separately because BCFAs, particularly isobutyrate and isovalerate, may reflect branched-chain amino acid fermentation and proteolytic activity rather than being considered indicators of beneficial fermentation.

Microbiome profiling further revealed treatment-associated patterns in bacterial diversity and community composition. However, because fecal samples from nine cats were pooled into a single inoculum and each treatment was represented by one fermentation vessel, the analytical replicates were not considered independent biological replicates. Accordingly, microbiome, SCFA/BCFA, and fermentation findings were interpreted as exploratory patterns rather than statistically confirmed treatment effects. Formal PERMANOVA and other treatment-level inferential analyses were therefore not applied to the microbiome fermentation data to avoid pseudoreplication.

Several phenolic-related metabolites were detected in OPF-containing treatments following fermentation. However, these compounds were tentatively annotated based on mass-spectral/database evidence, and their appearance after 24 h does not establish microbial formation or direct conversion from caffeic acid or other specific OPF precursors. Therefore, the proposed phenolic transformation pathways remain hypothetical and require confirmation using authentic standards, quantitative time-course analyses, and preferably isotope-tracing approaches.

Overall, the findings support further investigation of OPF as a potential fermentable agro-industrial ingredient for feline nutrition but do not establish clinical efficacy for weight control, obesity management, metabolic health, or gastrointestinal health. The present study also does not establish long-term safety, palatability, feeding tolerance, or contaminant-related safety. Future studies should therefore employ independent feline donors and biological fermentation replicates, quantitative validation of fermentation and phenolic metabolites, and controlled in vivo feeding trials to determine the physiological relevance, efficacy, palatability, and safety of OPF before any health or weight-management claims can be made.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Shashi Kant Bhatia, Konkuk University, Republic of Korea

Reviewed by: Gianfranco Picone, University of Bologna, Italy

Kanwar Kumar Malhi, Northeast Agricultural University, China

Footnotes

OPF, Oil palm fiber; TPC, Total polyphenol content; GAE, Gallic acid equivalent; TFC, Total flavonoid content; QE, Quercetin equivalent; TE, Trolox equivalent; DPPH, 2,2-Diphenyl-1-picrylhydrazyl; ABTS, 2,2′-Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid); FRAP, Ferric reducing antioxidant power; ORAC, Oxygen radical absorbance capacity; MDA, Malondialdehyde; TBARS, Thiobarbituric acid reactive substances; NO, Nitric oxide; ASV, Amplicon sequence variant; BCFA, Branched-chain fatty acid; BCS, Body condition score; SCFA, Short-chain fatty acid; SGF, Simulated gastric fluid; SIF, Simulated intestinal fluid.