Comparative metabolomic assessment of bioconverted traditional and black wheat via the in vitro piglet digestion model

Abstract

Colored wheat is gaining significant attention in the food and feed industry due to its components’ potential health benefits. As weaning piglets experience major discomfort in their early life, the quality and bioavailability of the feed nutrients become crucial for their survival. Fermentation of the feed aims to be a cheap and efficient stage of the feed production, as it is able to overcome major shortcomings of the cereals. The aim of this study was to investigate the differences in metabolites between traditional (variety “Herkus”) and black (variety “DS8558-1”) (DS) wheat cereals (both non-pre-treated and fermented with Lactiplantibacillus plantarum strain) using an in vitro swine digestion model with weaning piglets’ faecal microbiota. The following in vitro digestion and fermentation parameters were evaluated: free and total amino acids concentration in the digesta; total amino acids in the undigested residue and bioconverted/control wheat wholemeal substrate; short-chain fatty acids (SCFAs), volatile organic compounds (VOCs), free amino acids (FAAs), biogenic amines (BAs) and other small non-volatile metabolites in the in vitro fermentation samples. The in vitro digestibility and fermentability were significantly altered by bioconversion, while the substrate factor was significant only for the in vitro fermentability. The in vitro digestion of some of the non-essential and essential (valine, phenylalanine, leucine/isoleucine, histidine) AAs was significantly affected by both factors. Lysine digestibility was affected only by bioconversion. The bioconversion significantly affected the content of SCFAs, such as 2-methylpropanoic, butanoic, 3-methylbutanoic and pentanoic acids in in vitro fermentation (IVF) samples. The substrate was significant for all the analysed SCFAs, except propanoic and hexanoic acids. Benzenacetic, dihydrocinnamic and 4-hydroxyhydrocinnamic acids were significantly upregulated in DS IVF samples. P-cresol was upregulated in the bioconverted IVF samples. Both of the factors were significant for p-cresol concentrations in the IVF samples. The γ-aminobutyric acid (GABA) production in the IVF samples was significantly affected by both factors. Lactate concentrations in the IVF samples were affected only by the substrate factor. These results highlight the significance of bioconversion for the overall digestibility of wheat wholemeal substrates as well as changes in the production of biologically relevant metabolites for piglets.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-48313-9.

Introduction

Wheat continues to be a major cereal grain for human and animal consumption, as various products and byproducts, such as wheat bran, wheat middlings, millrun or red dogs, are used in livestock feeding^1–3. Coloured wheat is gaining significant attention in nutrition, as anthocyanins, which lie inside the pericarp and aleurone layers, provide additional antioxidative and antimicrobial properties, help in the prevention of cardiovascular diseases, inflammation, diabetes, obesity and improve cognitive functionality^4–7. Moreover, it has been established that polyphenols, including anthocyanins, can modulate bacterial communities in the gut⁸. Anthocyanins can also impair the digestibility of starch in the food matrix, which can reliably reduce the blood glucose levels of the host⁹. Often, anthocyanins are associated with ocular improvement by myopia inhibition, reduction of eye fatigue, improving vision adaptation and increasing ocular blood flow, which prevents retinal degeneration^6,10. The majority of these anthocyanins are various derivatives of cyanidin, delphinidin, malvidin, pelargonidin, peonidin and petunidin, which include singular or multiple glycosylations and acylations^6,11. Many in vitro studies report significant degradation of anthocyanins in the food matrix, mostly in the intestinal digestion phase^12–14. As the gastric digestion phase allows rapid absorption of anthocyanins without significant hydrolysis due to favourable conditions, a serious degradation of those compounds in the intestinal phase is observed due to a relatively high pH and high expression of metabolic enzymes that degrade anthocyanins^14,15.

Also, wheat cereal contains various antinutritional factors, such as inhibitors of protease enzymes, tannins and phytates that affect the digestibility of feed, intestinal mucus secretion, and bioavailability of important minerals^16–18. Evidently, fermentation helps to solve these issues by reducing antinutritional factor concentrations, improving digestion via breakdown of the feed matrix, and providing probiotics for the host’s gut^16,19,20. Solid state fermentation (SSF) is proven to be an attractive method for bioprocessing feed, as it allows the use of any nutrient-rich substrate, doesn’t require sterile conditions, demands low consumption of water, allows for easy recovery of products and requires smaller and less complex equipment for processing^21,22.

Lactic acid bacteria (LAB) are a diverse group of gram-positive microorganisms that are widely applied in industrial processing, as they are equipped with a wide range of carbohydrate-degrading enzymes, can improve organoleptic properties, produce a wide range of bioactive metabolites, suppress food spoilage, and improve shelf-life^23,24. Lactiplantibacillus plantarum is able to ferment a wide range of carbohydrates, inhibit pathogens, tolerate very low pH, and produce significant amounts of bioactive compounds, such as neurotransmitters and bacteriocins^25–28. Lp. plantarum is recognised as “generally recognised as safe” (GRAS) by the Food and Drug Administration and has the status of “qualified presumption of safety” (QPS) by the European Food Safety Authorities²⁴.

As genetic selection and modern intensive farming systems significantly increased the average litter size of sows, piglets nowadays suffer from weaning stress, which is associated with certain changes in microbiota as piglets go through a gradient change from liquid to solid feed^29–31. As such, dietary changes drastically modify the gut microbiome, which can eventually lead to impaired digestion, post-weaning diarrhea, reduced gut mucosal integrity, or even increase the risk of gastrointestinal tract diseases^32–34. Wheat-based products are widely used in swine feeding due to their high starch content and noteworthy abundance of essential amino acids³.

The aim of this study was to investigate the differences in metabolites (amino acids (AAs), GABA, biogenic amines (BAs), volatile organic compounds (VOCs), short-chain fatty acids (SCFAs) and other metabolites) between traditional (variety “Herkus”) and black (variety “DS8558-1”) wheat cereals (both non-pre-treated and fermented with Lactiplantibacillus plantarum strain) using an in vitro swine digestion model with weaning piglets’ faecal microbiota.

Materials and methods

Wheat cereals and lactic acid bacteria strain used for fermentation

The winter wheat (Triticum aestivum L.) cereal varieties ‘Herkus DS’ (traditional wheat), and new bred varieties DS8558-1 (black wheat) were provided by the Institute of Agriculture, Lithuanian Research Centre for Agriculture and Forestry (LAMMC, Akademija, Kėdainiai distr., Lithuania). A detailed description of the growing conditions is described in the supplementary methods file.

The LAB strain Lactiplantibacillus plantarum was obtained from the Lithuanian University of Health Sciences (Kaunas, Lithuania) and used for the wholemeal wheat flour fermentation. The Lp. plantarum characteristics are described by Bartkiene et al.³⁵. Before the experiments, the LAB strains stored at −80 °C (Microbank system, Pro-Lab Diagnostics, Birkenhead, UK) were activated and multiplied in de Man–Rogosa–Sharpe (MRS) broth (Oxoid Ltd., Hampshire, UK) at (30 ± 2) °C for 24 h, under aerobiosis, before use in the wholemeal flour fermentation. Wheat cereal was milled with a Laboratory Mill 120 (Perten Instruments AB, Stockholm, Sweden) until the particle size was 1–2 mm. The wholemeal wheat flours, water (1:1 w/w) and a suspension of LAB strain (3% of dry matter of the wholemeal flour) containing 8.7 log₁₀ CFU/mL were fermented at (30 ± 2) °C for 24 h. The non-fermented wholemeal samples were analysed as a control. The experimental scheme is provided in Fig. 1.

Fig. 1.

Schematic representation of the experiment. AA - amino acid, FAA - free amino acid, BA - biogenic amine, VOC - volatile organic compound, SCFA - short-chain fatty acid, TMS - trimethylsilyl.

Evaluation of acidity and microbiological characteristics of wheat wholemeal and the in vitro fermentation samples

The analysis of total titratable acidity, pH and microbiological characteristics was performed directly after fermentation, while the rest of the samples were vacuum packed and frozen at − 18 °C until further analysis of other compounds. Detailed protocol is provided in the supplementary methods file.

In vitro digestion protocol of wheat wholemeal samples

The procedure of the in vitro digestion protocol was based on the INFOGEST static in vitro digestion protocol with some modifications³⁶. Full procedure with modifications is described in the supplementary methods file. The resulting undigested residues were filtered, washed with ethanol and acetone, and dried in a laboratory oven until a stable mass was achieved.

In vitro fermentation protocol of the in vitro digestion residues

The residue from the in vitro digestion phase was subsequently subjected to colonic fermentation, using an inoculum of fresh faecal samples from healthy piglet donors of 42 days of age. The protocol used for the experiment was based on the protocol described in Burillo et al. (2021)³⁷ with some modifications. Detailed procedure is described in the supplementary methods file. Resulting unfermented residues were filtered, washed with water, and dried in a laboratory oven until a stable mass was achieved.

2.5 Free and Hydrolysed Amino Acid Analysis of the Bioconverted/Control Wholemeal, Digested Wholemeal Supernatant and Residue Samples.

Analysis protocol of free amino acids and hydrolysates of corresponding samples is described in the supplementary methods file. In brief, free amino acids were extracted with 0.1 M HCl solution, while the hydrolysis of samples was carried out using 6 M HCl solution, containing 0.1% phenol, at 110 ℃ for 23 h. The derivatization and chromatographic analysis protocol were carried out according to Hua-Lin Cai et al. (2010)³⁸ with some modifications.

Volatile organic acid analysis of the in vitro fermentation samples

The VOCs of the IVF samples were analysed using gas chromatography–mass spectrometry. A solid-phase microextraction (SPME) device with Stableflex™ fibre, coated with a 50 μm-ϕ Carboxen™- PDMS layer (Supelco, USA), was used for headspace sampling. The prepared samples were analysed with a GCMS-QP2010 (Shimadzu, Kyoto, Japan) gas chromatograph and mass spectrometer. A detailed description of the analysis is provided in the supplementary methods file.

Analysis of free amino acids and biogenic amines in the in vitro fermentation samples

Concentrations of amino acids were determined using the GCMS-QP2010 (Shimadzu, Kyoto, Japan) gas chromatograph with a mass spectrometer. Samples were diluted and derivatized using isobutylchloroformate. A detailed description is provided in the supplementary methods file.

Analysis of short-chain fatty acids in the in vitro fermentation samples

For SCFA analysis in IVF samples, 100 µL the of sample was spiked with 100 µL 90 mM 3-methylvaleric acid and diluted to 1 mL with 0.1 M HCl solution. The resulting mixture was centrifuged at 13,200 RPM for 5 min, and the resulting supernatant was used for the analysis. The prepared samples were analysed with a GCMS-QP2010 (Shimadzu, Kyoto, Japan) gas chromatograph and mass spectrometer. The following conditions were used for analysis: injector temperature 150 °C; injection volume 0.2 µL; ion source temperature 200 °C; and interface temperature 230 °C. The mass spectrometer was operating in single ion monitoring mode. Helium was used as a carrier gas at a 0.86 mL/min flow-rate. For the separation of VOCs, Stabilwax-Da column (Restek, USA) (length 30 m, coating thickness 0.25 μm-ϕ, inner diameter of 0.25 mm-ϕ) was used. The temperature gradient was programmed from starting at 100 °C (0.5 min hold) to 200 °C (5 °C/min, 3 min hold) up to 250 °C (15°/min, 3 min hold).

Untargeted analysis of small non-volatile and semi volatile metabolites in the in vitro fermentation samples

Untargeted analysis of small non-volatile and semi-volatile metabolites was accomplished by diluting samples with methanol and performing derivatization using methoxyamine hydrochloride in pyridine and subsequent silylation with N, O-bis(trimethylsilyl)trifluoroacetamide (BSTFA), containing 1% of trimethylchlorosilane. The prepared samples were analysed with a GCMS-QP2010 (Shimadzu, Kyoto, Japan) gas chromatograph and mass spectrometer. A detailed protocol is provided in the supplementary methods file.

Statistical analysis

Microbiological and physicochemical results were expressed as the mean (n = 3) ± standard deviations. Statistical calculations were carried out using R statistical programming software packages. In order to evaluate the effects of the different factors, the data were analysed using analysis of variance (ANOVA), and Tukey HSD (Tukey’s honest significant difference) post hoc test was conducted to detect differences among sample groups. To determine the normality and homoskedasticity, the Shapiro–Wilk test and Levene test were performed, using “stats” (version 4.4.2) and “car” (version 3.1-3.1) packages. The bioconversion (fermented/nonfermented) and substrate (Herkus/DS) were chosen for factor analysis. The linear Pearson’s correlation coefficients were calculated and visualized using “psych” (version 2.5.6) and “corrplot” (version 0.95) packages. The results were recognized as statistically significant at p < 0.05. The heatmap visualization and analysis were carried out using the R statistical programming software package “ComplexHeatmap” (version 2.25.2). The principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA) and variable importance in projection (VIP) were carried out using the “mixOmics” package (version 6.26.0). Volcano plot analysis and visualizations were carried out using the “EnhancedVolcano” package (version 1.24.0). Missing values for the volcano plot analysis were filled using the half minimum method.

Results

Acidity and microbiological characteristics of wheat wholemeal samples

The results of the acidity and microbiological characteristics are presented in Table 1. The LAB count in fermented wheat wholemeal samples (Herkus_F and DS_F) was significantly higher, while the pH was significantly lower than in control samples (Herkus_NF and DS_NF). Moreover, titratable acidity was significantly higher in fermented wheat wholemeal samples. In addition, Herkus_F titratable acidity was the highest of all sample groups.

Table 1.

LAB count, pH and titratable acidity of wheat wholemeal samples.

Measurement Sample	pH	TTA°	LAB Count, log10 CFU/g
DS_F	3.943 ± 0.09 c	9.067 ± 0.153 b	8.503 ± 0.015 a
DS_NF	5.643 ± 0.04 b	2.383 ± 0.076 c	4.407 ± 0.035 b
Herkus_F	3.950 ± 0.03 c	9.900 ± 0.10 a	8.483 ± 0.07 a
Herkus_NF	5.943 ± 0.045 a	1.917 ± 0.076 d	4.423 ± 0.042 b

DS – wheat variety DS8558-1, F - sample fermented with Lactiplantibacillus plantarum, NF – control sample, TTA° - total titratable acidity, LAB – lactic acid bacteria. a-d Mean values denoted with different letters indicate significantly different values between the different sample groups. Results are statistically significant when p < 0.05.

Total amino acid composition of wheat wholemeal samples

The results of the total AA composition of wheat wholemeal samples are presented in Table 2. Overall, most AAs and total AA concentrations were not significantly different between fermented samples and corresponding control samples. Leucine/isoleucine was significantly higher in Herkus_F samples in comparison to Herkus_NF. DS samples contained significantly higher concentrations of arginine (16.5%), glutamic acid (38.9%), threonine (37.5%), glycine (29.3%), proline (49.0%), phenylalanine (45.4%), leucine/isoleucine (41.5%), cystine (31.1%), and total AA (35.7%), in comparison to Herkus sample groups. The bioconversion was a significant factor for leucine/isoleucine, cystine, and lysine concentrations, while the substrate was a significant factor only for leucine/isoleucine (supplementary file results Table 1). Moreover, the bioconversion and substrate factor interaction was significant for all amino acids (supplementary file results Table 1).

Table 2.

Total amino acids (g/100 g) in wheat wholemeal samples.

Sample	DS_F	DS_NF	Herkus_F	Herkus_NF
Arginine	0.406 ± 0.021 a	0.408 ± 0.001 a	0.348 ± 0.014 b	0.351 ± 0.0003 b
Glutamic acid	2.539 ± 0.024 a	2.616 ± 0.08 a	1.820 ± 0.11 b	1.892 ± 0.067 b
Serine	0.354 ± 0.029 a	0.363 ± 0.004 a	0.291 ± 0.043 a	0.288 ± 0.009 a
Aspartic acid	0.313 ± 0.028 a	0.309 ± 0.004 a	0.259 ± 0.035 a	0.258 ± 0.01 a
Threonine	0.225 ± 0.007 a	0.222 ± 0.001 a	0.167 ± 0.001 b	0.158 ± 0.008 b
Glycine	0.287 ± 0.017 a	0.300 ± 0.004 a	0.221 ± 0.017 b	0.233 ± 0.009 b
Alanine	0.281 ± 0.019 a	0.268 ± 0.018 ab	0.227 ± 0.004 ab	0.217 ± 0.01 b
Proline	1.657 ± 0.032 a	1.746 ± 0.133 a	1.166 ± 0.028 b	1.118 ± 0.035 b
Valine	0.285 ± 0.018 a	0.298 ± 0.022 a	0.231 ± 0.009 a	0.224 ± 0.024 a
Methionine	0.068 ± 0.01 a	0.074 ± 0.011 a	0.043 ± 0.005 a	0.054 ± 0.008 a
Phenylalanine	0.411 ± 0.009 a	0.415 ± 0.001 a	0.312 ± 0.039 b	0.256 ± 0.013 b
Leucine/Isoleucine	0.961 ± 0.01 a	0.951 ± 0.039 a	0.741 ± 0.01 b	0.610 ± 0.045 c
Cystine	0.100 ± 0.002 a	0.115 ± 0.009 a	0.072 ± 0.008 b	0.092 ± 0.001 ab
Lysine	0.219 ± 0.005 a	0.207 ± 0.007 a	0.197 ± 0.013 ab	0.164 ± 0.007 b
Histidine	0.158 ± 0.0004 a	0.154 ± 0.008 a	0.128 ± 0.011 ab	0.110 ± 0.015 b
Tyrosine	0.305 ± 0.005 ab	0.325 ± 0.027 a	0.268 ± 0.003 ab	0.265 ± 0.002 b
Total amino acids	8.569 ± 0.162 a	8.771 ± 0.204 a	6.492 ± 0.161 b	6.289 ± 0.149 b

DS – wheat variety DS8558-1, F - sample fermented with Lactiplantibacillus plantarum, NF – control sample. a-c Mean values denoted with different letters indicate significantly different values between the different sample groups. Results (n = 2) are statistically significant when p < 0.05.

Total and free amino acids in the in vitro digestion samples

Table 3 describes the analysis results of the FAAs in the in vitro digestion (IVD) samples. The substrate-containing samples (DS_F, DS_NF, Herkus_F, Herkus_NF) showed significantly higher concentrations of arginine (51.2%), glutamine (82.2%), GABA (84.2%), proline (55.6%), alanine (38.3%), threonine (37.2%), valine (51.5%), methionine (66.6%), phenylalanine (57.5%), leucine/isoleucine (53.5%), histidine (45.5%), and total AA (46.5%). Only the fermented samples contained significantly higher amounts of glycine (20.3%), in comparison to the blank IVD samples. Lysine concentrations were significantly higher in all the substrate IVD samples, except DS_NF, in comparison to the blank IVD samples. Glutamic acid was significantly higher only in the Herkus_F samples. The fermented sample groups contained significantly higher concentrations of GABA (90.0%) and proline (124.5%), and significantly lower concentrations of glutamine (15.6%), in comparison to the nonfermented sample groups. The bioconversion was a significant factor for glutamine, glutamic acid, glycine, GABA, and proline, while the substrate was significant for glutamine, GABA, proline, tryptophan, and cystine (supplementary file results Table 2).

Table 3.

Free amino acids concentrations (mg/L) in the in vitro digestion samples.

Sample	Blank	DS_F	DS_NF	Herkus_F	Herkus_NF
Arginine	256.07 ± 45.47 b	516.38 ± 21.86 a	555.04 ± 52.03 a	498.5 ± 38.1 a	511.08 ± 50.52 a
Glutamine	125.42 ± 14.98 d	620.58 ± 38.21 bc	743.76 ± 33.13 a	550.15 ± 44.78 c	643.03 ± 12.1 b
Asparagine	91.68 ± 10.66 a	108.02 ± 17.88 a	119.33 ± 15.03 a	107.33 ± 14.43 a	116.27 ± 4.34 a
Glutamic acid	148.27 ± 20.21 b	179.17 ± 31.51 ab	154.32 ± 14.26 ab	200.29 ± 6.42 a	176.21 ± 6.96 ab
Serine	86.67 ± 19.04 b	105.98 ± 16.73 ab	115.58 ± 17.1 ab	121.84 ± 7.11 ab	131.76 ± 12.31 a
Aspartic acid	80.78 ± 15.16 a	100.41 ± 13.68 a	96.32 ± 4.46 a	97.8 ± 25.32 a	92.25 ± 27.5 a
Threonine	75.48 ± 15.36 b	113.36 ± 1.05 a	126.39 ± 3.07 a	111.17 ± 12.71 a	119.71 ± 14.81 a
Glycine	119.53 ± 8.71 b	142.96 ± 9.39 a	131.19 ± 5.11 ab	144.6 ± 6.3 a	131.21 ± 5.55 ab
Alanine	107.99 ± 10.55 b	182.25 ± 17.75 a	170.87 ± 13.84 a	188.12 ± 6.62 a	169.92 ± 11.81 a
GABA	2.35 ± 0.38 d	20.04 ± 2.61 b	9.84 ± 0.97 c	30.31 ± 2.13 a	16.67 ± 0.75 b
Proline	5.23 ± 0.53 d	43.18 ± 2.28 a	19.75 ± 0.74 c	37.67 ± 1.21 b	16.27 ± 1.16 c
Tryptophan	109.84 ± 19.42 b	139.03 ± 22.56 ab	139.77 ± 20.77 ab	177.04 ± 11.48 a	180.3 ± 10.41 a
Valine	111.74 ± 21.08 b	194.45 ± 10.56 a	216.43 ± 19.32 a	205.53 ± 27.18 a	231.06 ± 8.66 a
Methionine	29.61 ± 7.21 b	90.62 ± 9.74 a	88.88 ± 19.66 a	88.64 ± 13.94 a	91.62 ± 1.87 a
Phenylalanine	146.36 ± 11 b	361.03 ± 68.06 a	371.05 ± 75.86 a	341.43 ± 16.29 a	341.82 ± 21.89 a
Leucine/Isoleucine	368.89 ± 36.36 b	811.09 ± 143.11 a	770.37 ± 68.86 a	778.27 ± 80.87 a	849.98 ± 8.59 a
Cystine	26.9 ± 6.47 a	22.49 ± 1.24 a	24.76 ± 2.67 a	27.53 ± 3.04 a	29.19 ± 0.99 a
Lysine	275.72 ± 27.84 b	389.8 ± 55.09 a	372.03 ± 58.1 ab	390.22 ± 16.74 a	385.61 ± 15.83 a
Histidine	64.39 ± 8.3 b	108.36 ± 17.97 a	116.67 ± 15.02 a	109.4 ± 3.97 a	120.37 ± 7.66 a
Tyrosine	225.97 ± 26.93 a	344.35 ± 62.79 a	323.46 ± 65.94 a	253.47 ± 34.53 a	302.83 ± 55.96 a
Total AAs	2458.9 ± 304.92 b	4593.52 ± 492.63 a	4665.82 ± 360.1 a	4459.33 ± 176.09 a	4657.16 ± 93.63 a

DS – wheat variety DS8558-1, F - sample fermented with Lactiplantibacillus plantarum, NF – control sample, GABA – γ-aminobutyric acid. a-d Mean values denoted with different letters indicate significantly different values between the different sample groups. Results are statistically significant when p < 0.05.

The results of total AA in IVD samples are presented in Table 4. The DS samples contained significantly higher concentrations of leucine/isoleucine (26.8%) and tyrosine (36.1%), in comparison to the Herkus sample groups. The bioconversion was a significant factor only for concentrations of glycine, while the substrate was a significant factor for alanine, glutamic acid, threonine, proline, valine, methionine, phenylalanine, leucine/isoleucine, tyrosine, and the total AAs in the IVD samples (supplementary file results Table 3).

Table 4.

Total amino acids (mg/L) in the in vitro digestion samples.

Sample	Blank	DS_F	DS_NF	Herkus_F	Herkus_NF
Arginine	409.06 ± 7.97 b	885.5 ± 14.4 a	916.4 ± 50.8 a	847.6 ± 116.5 a	718 ± 84.8 a
Glutamic acid	739.93 ± 54.82 b	3538.6 ± 93.7 a	3544.2 ± 257.1 a	2901.4 ± 289.8 a	2882.2 ± 363.5 a
Serine	405.67 ± 39.43 b	828.8 ± 80.9 a	821.8 ± 125.8 a	731.2 ± 89.9 a	820.3 ± 131.6 a
Aspartic acid	770.47 ± 1.61 b	1130.7 ± 92.7 a	1074.8 ± 75.4 a	1066.7 ± 90.3 a	1073 ± 127.9 a
Threonine	294.74 ± 32.06 b	594.4 ± 28 a	555.9 ± 33.6 a	521 ± 50.7 a	517.5 ± 9.7 a
Glycine	531.04 ± 9.81 c	904.6 ± 46 a	850.8 ± 45.4 ab	863.3 ± 63.9 ab	780.7 ± 11.9 b
Alanine	284.86 ± 9.84 b	580.3 ± 34.6 a	582.4 ± 21.1 a	565.4 ± 54.9 a	524.4 ± 32.4 a
Proline	557.23 ± 84.24 c	2633.8 ± 71.6 ab	2786.5 ± 147.3 a	1953.2 ± 158.1 ab	2009 ± 44.7 ab
Valine	323.4 ± 41.59 b	801.1 ± 77.6 a	776.2 ± 29.8 a	657.8 ± 58.6 a	697.4 ± 69.8 a
Methionine	46.17 ± 14.99 b	182.8 ± 2.9 a	178.4 ± 11.2 a	131.7 ± 35.5 a	133.0 ± 10.0 a
Phenylalanine	238.39 ± 30.83 d	699.8 ± 18 ab	757.8 ± 79.5 a	554.3 ± 64.9 c	596.9 ± 29.5 bc
Leucine/Isoleucine	677.95 ± 17.53 d	1813.9 ± 73.1 ab	1910.7 ± 205.9 a	1458 ± 131.1 c	1479.8 ± 40.4 bc
Cystine	128.95 ± 17.88 b	238.2 ± 7.5 a	237.4 ± 10.2 a	240.2 ± 19.8 a	225.5 ± 22.3 a
Lysine	383.14 ± 41.34 b	642.3 ± 40.4 a	693.5 ± 55.4 a	666.1 ± 25.3 a	664.7 ± 51.1 a
Histidine	126.63 ± 5.06 b	345.1 ± 16.9 a	355.6 ± 35.8 a	306.9 ± 31.8 a	326 ± 31.4 a
Tyrosine	359.5 ± 28.31 c	771.7 ± 28.9 a	789.9 ± 22.8 a	506 ± 109 bc	641.2 ± 81.6 ab
Total AAs	6304.3 ± 436.89 c	16667.3 ± 244.8 a	16893.7 ± 586.7 a	14010.4 ± 1183.7 b	14156.7 ± 886.7 b

DS – wheat variety DS8558-1, F - sample fermented with Lactiplantibacillus plantarum, NF – control sample. a-d Mean values denoted with different letters indicate significantly different values between the different sample groups. Results are statistically significant when p < 0.05.

The total AA analysis results of the undigested IVD residue hydrolysates in the undigested IVD residues are presented in Table 5. Significantly higher concentrations of threonine were found in the fermented sample groups, in comparison to the corresponding control sample groups. While glutamic acid (37.2%), glycine (28.6%), alanine (40.0%), proline (43.3%) and total AA (32.5%) concentrations were significantly higher in the Herkus_F samples, in comparison to the Herkus_NF samples. The bioconversion factor was significant for all amino acid concentrations, except serine, valine, phenylalanine, and leucine/isoleucine, while the substrate was significant for glutamic and aspartic acids, threonine, proline, histidine and total AAs (supplementary file results Table 4). The interaction of the factors was significant for arginine, proline, and histidine concentrations in the IVD residue samples.

Table 5.

Total amino acids (g/100 g) in the in vitro digestion undigested samples.

Sample	DS_F	DS_NF	Herkus_F	Herkus_NF
Arginine	0.524 ± 0.036 a	0.527 ± 0.029 a	0.549 ± 0.039 a	0.420 ± 0.025 b
Glutamic acid	1.452 ± 0.128 a	1.33 ± 0.065 a	1.332 ± 0.086 a	0.971 ± 0.144 b
Serine	0.495 ± 0.114 a	0.415 ± 0.035 a	0.430 ± 0.049 a	0.337 ± 0.051 a
Aspartic acid	0.931 ± 0.136 a	0.677 ± 0.038 b	0.782 ± 0.089 ab	0.560 ± 0.076 b
Threonine	0.333 ± 0.036 a	0.216 ± 0.02 bc	0.241 ± 0.007 b	0.154 ± 0.029 c
Glycine	0.493 ± 0.039 ab	0.435 ± 0.046 b	0.526 ± 0.018 a	0.409 ± 0.028 b
Alanine	0.432 ± 0.019 a	0.345 ± 0.023 ab	0.434 ± 0.054 a	0.310 ± 0.042 b
Proline	0.960 ± 0.045 a	0.880 ± 0.008 a	0.934 ± 0.052 a	0.652 ± 0.058 b
Valine	0.429 ± 0.036 a	0.390 ± 0.033 a	0.406 ± 0.051 a	0.390 ± 0.046 a
Methionine	ND	ND	ND	ND
Phenylalanine	0.361 ± 0.026 a	0.322 ± 0.024 a	0.336 ± 0.059 a	0.312 ± 0.029 a
Leucine/Isoleucine	0.860 ± 0.078 a	0.746 ± 0.065 a	0.850 ± 0.19 a	0.684 ± 0.032 a
Cystine	ND	ND	ND	ND
Lysine	0.502 ± 0.084 a	0.392 ± 0.031 ab	0.437 ± 0.031 ab	0.331 ± 0.051 b
Histidine	0.126 ± 0.026 a	0.092 ± 0.011 a	0.131 ± 0.018 a	ND b
Tyrosine	0.572 ± 0.017 a	0.507 ± 0.058 a	0.597 ± 0.047 a	0.503 ± 0.044 a
Total AA	8.477 ± 0.146 a	7.274 ± 0.182 ab	7.984 ± 0.728 a	6.025 ± 0.635 b

DS – wheat variety DS8558-1, F - sample fermented with Lactiplantibacillus plantarum, NF – control sample, ND – not detected. a-c Mean values denoted with different letters indicate significantly different values between the different sample groups. Results are statistically significant when p < 0.05.

In vitro digestibility and fermentability parameters of wheat wholemeal samples

Table 6 contains digestibility and fermentability parameters of the IVD/IVF samples. The in vitro digestibility in the fermented samples was significantly higher than in the respective control samples. Moreover, the highest AA digestibility was observed in the DS_F samples, while the lowest digestion was observed in the Herkus_NF sample groups. The in vitro fermentability in the fermented sample groups was significantly lower than in the corresponding control sample groups. Moreover, Herkus_NF samples had the highest in vitro fermentability results, while the DS_F samples had the lowest in vitro fermentability results. Almost all of the individual AA digestibility was not significantly different in the bioconverted samples, in comparison to the control sample groups (supplementary results file Table 5). Only arginine in the DS samples and valine, phenylalanine, cystine, histidine and tyrosine in the Herkus samples were significantly different between the respective control and the bioconverted sample groups. The bioconversion was a significant factor for all the in vitro digestibility and fermentability parameters, except glutamic acid, serine, aspartic acid, threonine, glycine, and alanine (supplementary results file Table 6). While substrate was a significant factor for all the parameters, except for the in vitro digestibility, arginine, glutamic acid, serine, aspartic acid, threonine, and lysine. The factors’ interaction was significant for the in vitro digestibility, valine, phenylalanine, and histidine.

Table 6.

In vitro digestibility and fermentability of wheat wholemeal samples.

Sample	DS_F	DS_NF	Herkus_F	Herkus_NF
In vitro digestibility, % DM	83.91 ± 0.49 a	78.79 ± 0.65 b	84.79 ± 0.53 a	77.59 ± 0.49 b
Total AA digested, %	91.88 ± 0.18 a	91.07 ± 0.49 ab	90.57 ± 0.61 ab	88.96 ± 1.41 b
In vitro fermentability, % DM	42.70 ± 0.24 c	48.11 ± 1.27 b	49.56 ± 1.60 b	57.51 ± 0.83 a
Total in vitro digestibility, % DM	90.78 ± 0.31 b	89.00 ± 0.22 c	92.33 ± 0.25 a	90.48 ± 0.21 b

DS – wheat variety DS8558-1, F - sample fermented with Lactiplantibacillus plantarum, NF – control sample, AA – amino acids, DM – dry matter. a-c Mean values denoted with different letters indicate significantly different values between the different sample groups. Results are statistically significant when p < 0.05.

Free amino acids and biogenic amines analysis results of the in vitro fermentation samples

The FAA and BA analysis results are presented in Table 7. The Herkus samples contained significantly higher concentrations of glycine, valine, leucine, serine, proline, aspartic acid, glutamic acid, methionine, phenylalanine, histidine, tryptophan and tyrosine in comparison to the blank samples after 24 h and the DS sample groups. The Herkus samples contained significantly higher amounts of glycine (124.1%), valine (1652.9%), leucine (1481.3%), serine, proline (1073.9%), aspartic acid (265.7%), glutamic acid (2190.9%), methionine (191.6%), phenylalanine (389.3%), histidine, tryptophan (562.9%) and tyrosine (279.7%), in comparison to the DS sample groups. The bioconversion was significant for almost all the FAAs and BAs, except for alanine, valine, leucine, threonine, aspartic acid, phenylalanine, tyramine, histidine and tryptophan, while the substrate was significant for all analytes, except for lysine (supplementary results file Table 7). The interaction of those factors was significant for all analytes, except for alanine, valine, leucine, GABA, threonine, aspartic acid, methionine, lysine, histidine, and tyrosine.

Table 7.

Free amino acids and biogenic amines analysis results (mg/L) in the in vitro fermentation samples.

Sample	B0	Bl	DS_F	DS_NF	Herkus_F	Herkus_NF
Alanine	111.88 ± 4.66 bc	1.67 ± 0.83 d	109.08 ± 15.95 c	153.72 ± 18.52 abc	175.08 ± 21.55 ab	186.69 ± 39.09 a
Glycine	38.5 ± 2.58 b	5.33 ± 0.71 c	8.21 ± 3.10 c	56.14 ± 9.18 ab	71.79 ± 2.72 a	72.43 ± 15.44 a
Valine	102.54 ± 7.4 a	9.19 ± 1.76 cd	2.38 ± 0.15 d	1.48 ± 0.09 d	27.5 ± 3.73 bc	40.16 ± 19.76 b
Leucine	231.41 ± 9.75 a	9.92 ± 1.61 c	2.30 ± 0.35 c	2.02 ± 0.33 c	30.89 ± 2.23 b	37.42 ± 15.89 b
Isoleucine	74.52 ± 1.76 a	8.09 ± 1.16 c	5.18 ± 0.05 c	4.21 ± 0.67 c	5.96 ± 0.63 c	41.65 ± 23.42 b
GABA	ND c	ND c	43.99 ± 4.89 c	303.61 ± 105.18 b	385.28 ± 50.94 b	627.96 ± 76.96 a
Serine	64.98 ± 18.96 a	ND c	ND c	ND c	27.74 ± 0.61 b	33.89 ± 3.34 b
Threonine	46.97 ± 14.42 a	ND c	14.62 ± 1.06 bc	11.61 ± 0.44 bc	25.63 ± 1.47 ab	39.74 ± 16.41 a
Proline	24.72 ± 0.05 bc	25.3 ± 5.04 bc	2.66 ± 0.81 c	55.02 ± 13.02 b	331.75 ± 13.89 a	345.38 ± 17.43 a
Asparagine	ND	ND	ND	ND	ND	ND
Aspartic acid	26.41 ± 3.31 a	5.2 ± 0.2 b	8.85 ± 0.12 b	7.86 ± 4.50 b	35.14 ± 8.24 a	25.97 ± 8.73 a
Phenylethylamine	1.82 ± 0.13 a	ND b	ND b	ND b	ND b	ND b
Glutamic acid	55.44 ± 19.23 c	23.65 ± 10.04 c	13.83 ± 0.48 c	9.9 ± 2.31 c	358.27 ± 50.2 a	185.35 ± 13.91 b
Methionine	42.06 ± 11.43 ab	9.25 ± 0.24 c	11.22 ± 0.26 c	16.21 ± 5.24 c	31.9 ± 3.88 b	48.07 ± 7.06 a
Phenylalanine	81.00 ± 30.34 bc	9.47 ± 3.27 c	6.40 ± 0.78 c	104.99 ± 32.53 b	290.37 ± 19.63 a	254.68 ± 53.52 a
Putrescine	ND c	ND c	116.45 ± 6.63 b	107.92 ± 12.68 b	160.29 ± 16.51 a	107.16 ± 11.54 b
Cadaverine	5.38 ± 0.21 c	ND c	63.20 ± 6.41 b	60.03 ± 5.48 b	89.72 ± 4.15 a	60.04 ± 8.06 b
Histamine	ND c	ND c	23.30 ± 1.97 b	23.95 ± 1.15 b	37.95 ± 4.78 a	26.25 ± 5.58 b
Tryptamine	ND c	ND c	5.69 ± 0.06 b	5.76 ± 0.75 b	18.13 ± 3.29 a	6.39 ± 0.63 b
Lysine	57.6 ± 11.12 a	ND d	39.63 ± 0.06 bc	27.11 ± 0.29 c	42.86 ± 3.88 ab	33.09 ± 8.83 bc
Tyramine	ND b	ND b	4.25 ± 0.04 b	7.42 ± 3.10 b	39.39 ± 19.1 a	10.89 ± 4.31 b
Histidine	374.98 ± 34.88 a	ND c	ND c	ND c	336.51 ± 6.93 ab	291.12 ± 52.2 b
Tryptophan	31.93 ± 2.72 b	ND d	ND d	11.50 ± 3.50 c	42.99 ± 2.92 a	33.24 ± 4.17 b
Tyrosine	74.11 ± 25.18 a	5.96 ± 0.34 b	28.54 ± 0.17 b	11.93 ± 5.00 b	87.42 ± 18.64 a	66.24 ± 4.56 a

DS – wheat variety DS8558-1, F - sample fermented with Lactiplantibacillus plantarum, NF – control sample, Bl – fermentation blank at 48 h, B0 – fermentation blank at 0 h. a-d Mean values denoted with different letters indicate significantly different values between the different sample groups. Results are statistically significant when p < 0.05.

Volatile organic compounds analysis results of the in vitro fermentation samples

The VOCs analysis results are presented in the supplementary results file Table 8 and depicted in the heatmap in Fig. 2. VOCs data were Pareto-scaled, and hierarchical clustering was performed for the individual analytes. Various alkanes (tridecane, octadecane, hexadecane, dodecane, tetradecane), aromatic and heterocyclic compounds (benzothiazole, 2-phenoxyethanol, naphthalene, skatole, 4-ethylphenol, and 2,4-ditertbutylphenol), organic acids and alcohols (nonanoic acid, 4-methylpentanoic acid, and 3-hexadecanol) and several aldehydes (hexadecanal, nonanal) were significantly lower in almost all the IVF DS and Herkus samples, in comparison to the blank after 24 h IVF samples. Moreover, various aliphatic and aromatic esters (propyl pentanoate, hexyl n-valerate, pentyl levulinate), aldehydes (oct-(2E)-enal, pentadecanal, and tetradecanal), various aromatic compounds (p-cresol, indole, hydrocinnamic acid, and benzeneacetic acid), and hexanoic acid were the highest in the DS_F samples. The bioconversion factor was significant for 31 out of 53 VOCs, while the substrate was significant for 36 out of 53 VOCs in the IVF samples (supplementary results Table 9). While the factors’ interaction was significant for 6 organic acids (propanoic, 2-methylpropanoic, butanoic, 3-methylbutanoic, pentanoic and hexanoic acids), 3 aromatic compounds (indole, hydrocinnamic acid, and p-cresol), pentyl levulinate, tetradecanal, and octadecane in the IVF samples.

Fig. 2.

Heatmap of volatile organic compounds in the in vitro fermentation samples.

PLS-DA and VIP analysis results are depicted in Fig. 3. Statistical analysis was performed on Pareto-scaled data. PLS-DA plots show clear separation between the fermented and the nonfermented samples (Fig. 3-a), as well as the different substrates (Fig. 3-c). For the first model, VIP analysis showed that isophorone, mesityl oxide, acetone, pentyl levulinate, methyl thioacetate, aliphatic organic acids (pentanoic, butanoic, nonanoic, octanoic, 2-methylpropanoic, and 3-methylbutanoic acids), alkanes (hexadecane, tridecane, octadecane, and dodecane), aldehydes (pentadecanal, hexadecanal, dodecanal, and decanal), and aromatic compounds (p-cresol, isocreosol, 2-phenoxyethanol, and 2,4-ditertbutylphenol) were contributing the most to the differences between the fermented and nonfermented samples. While various aromatic compounds (acetophenone, benzeneacetic acid, hydrocinnamic acid, naphthalene, 4-ethylphenol, phenylethyl alcohol, 2-phenoxyethanol, and p-cresol), indoles (skatole and indole), aliphatic organic acids (acetic, 4-methylpentanoic, nonanoic, pentanoic, octanoic, 2-methylbutyric, and heptanoic acids), alkanes and aliphatic aldehydes (nonanal, hexadecanal, octadecane, tetradecane, dodecane, and oct-(2E)-enal), γ – decalactone, benzothiazole, hexyl n-valerate, and ethyl palmitate had the highest VIP values for the Herkus – DS PLS-DA model.

Fig. 3.

PLS-DA and VIP plots of the volatile organic compounds of the in vitro fermentation samples. a, c – PLS-DA analysis plots, where categorical values are bioconversion and substrate; b, d – VIP plots of the respective PLS-DA analyses.

Additionally, volcano plots of the VOCs in IVF samples were constructed to depict differences between fermented – nonfermented (Fig. 4-a) and Herkus – DS sample groups (Fig. 4-b). The analysis results depict significantly higher concentrations of p-cresol, mesityl oxide, isophorone, and pentyl levulinate in fermented IVF samples. While the comparison between the Herkus and DS sample groups outlines significantly higher concentrations of benzenacetic acid, hexyl n-valerate, hydrocinnamic acid, indole, p-cresol, oct-(2E)-enal, and pentyl pentanoate in the DS samples.

Fig. 4.

Volcano plots of the volatile organic compounds of the in vitro fermentation samples.

Untargeted metabolomics analysis results of the in vitro fermentation samples

The IVF untargeted metabolomics analysis results are presented in the supplementary results file Table 11. The heatmap visualization and hierarchical clustering analysis are depicted in Fig. 5. The IVF untargeted metabolomics data were Pareto-scaled and hierarchical clustering was performed for the individual analytes. Almost all of the identified metabolites’ concentrations were significantly higher in the substrate-fermented (DS_F, DS_NF, Herkus_NF, and Herkus_F) samples, in comparison to the blank after 24 h IVF samples. The bioconversion was a significant factor for succinic, glyceric, dihydrocinnamic, 2-deoxytetronic, pyroglutamic, α-hydroxyglutaric, 4-hydroxyphenylacetic, 3-hydroxyphenylpropionic and indole-3-acetic acid concentrations (supplementary results, Table 12). While the substrate was a significant factor for lactic acid, glycolic acid, 4-hydroxybutyric acid, ethanolamine, benzeneacetic acid, succinic acid, glyceric acid, dihydrocinnamic acid, 2-deoxytetronic acid, pyroglutamic acid, α-hydroxyglutaric acid, β-phenyllactic acid, 5-aminopentanoic acid, 3-hydroxyphenylpropionic acid, 4-hydroxyhydrocinnamic acid, dihydroferulic acid, and indole-3-acetic acid in the IVF samples. The interaction of the aforementioned factors was significant for succinic acid, glyceric acid, uracil, dihydrocinnamic acid, α-hydroxyglutaric acid, β-phenyllactic acid, 5-aminopentanoic acid, 4-hydroxyphenylacetic acid, and 3-hydroxyphenylpropionic acid in the IVF samples.

Fig. 5.

Heatmap of in vitro fermentation untargeted analysis results.

The PLS-DA and VIP analysis results are presented in Fig. 6. The IVF metabolites between the Herkus and the DS sample groups (Fig. 6-c) show clear separation of sample groups, while the fermented and nonfermented sample groups (Fig. 6-a) demonstrate some overlap of the sample groups. The VIP analysis results depict that 4-hydroxyphenylacetic acid, 2-deoxytetronic acid, 3-hydroxyphenylpropionic acid, pyroglutamic acid, propane-1,3-diol, succinic acid, dihydrocinnamic acid, ethanolamine, and α-hydroxyglutaric acid were the most contributing compounds for the fermented – nonfermented PLS-DA model. While 5-aminopentanoic acid, 4-hydroxybutyric acid, indole-3-acetic acid, benzeneacetic acid, glyceric acid, β-phenyllactic acid, 4-hydroxyhydrocinnamic acid, lactic acid, dihydroferulic acid, α-hydroxyglutaric acid, dihydrocinnamic acid, glycolic acid, and succinic acid had the most contribution to the Herkus – DS PLS-DA model.

Fig. 6.

PLS-DA and VIP plots of the untargeted metabolites of the in vitro fermentation samples. A, c – PLS-DA analysis plots, where categorical values are bioconversion and substrate; b, d – VIP plots of the respective PLS-DA analyses.

The volcano plots of the fermented-nonfermented sample groups (Fig. 7-a) show significantly lower concentrations of 4-hydroxyphenylacetic acid, while having significantly higher concentrations of 3-hydroxyphenylpropionic acid in the fermented IVF samples. Glyceric, β-phenyllactic, 4-hydroxybutyric, succinic, lactic, and α-hydroxyglutaric acids were significantly higher in the Herkus sample groups, in comparison to the DS sample groups (Fig. 7-b). While 5-aminopentanoic acid, indole-3-acetic acid, benzenacetic acid, 4-hydroxycinnamic acid, dihydroferulic acid, dihydrocinnamic acid and 3-hydroxyphenylpropionic acid concentrations were significantly higher in the DS sample groups.

Fig. 7.

Volcano plots of the untargeted metabolites of the in vitro fermentation samples.

The PCA analysis results of the VOCs and untargeted metabolites in the IVF samples are presented in Fig. 8. PCA results of the VOCs in the IVF samples show clear separation between the fermented vs. nonfermented and the Herkus vs. DS sample groups (Fig. 8-a, b). While PCA of the untargeted metabolites showed clear separation between the Herkus and DS sample groups, there was no clear distinction between the fermented and nonfermented samples of untargeted metabolites (Fig. 8-c, d).

Fig. 8.

PCA plots of the in vitro fermentation samples. a, b - PCA plots of volatile organic compounds of the in vitro fermentation samples; c, d - PCA plots of untargeted metabolomics results of the in vitro fermentation samples.

Short-chain fatty acids analysis results of the in vitro fermentation samples

The analysis results of SCFA are presented in Table 8. Acetic acid concentrations were the highest in the DS sample groups. The highest concentrations of butanoic and pentanoic acids were observed in the Herkus_NF sample group. The bioconversion factor was significant for the concentrations of 2-methylpropanoic, butanoic, 3-methylbutanoic and pentanoic acids, while the substrate was a significant factor for almost all of the SCFAs, except for propanoic and hexanoic acids (supplementary results file Table 9). The interaction of both factors was significant for propanoic, butanoic, pentanoic, and hexanoic acids in the IVF samples.

Table 8.

Short-chain Fatty Acid concentrations (mmol/L) in the in vitro fermentation samples.

Sample	B0	Bl	DS_F	DS_NF	Herkus_F	Herkus_NF
Acetic acid	6.161 ± 0.142 c	34.44 ± 10.526 b	86.466 ± 7.905 a	74.829 ± 8.17 a	47.43 ± 12.029 b	50.394 ± 0.989 b
Propanoic acid	2.06 ± 0.094 c	8.083 ± 0.167 c	39.326 ± 2.663 b	51.636 ± 3.226 a	51.274 ± 7.01 a	46.858 ± 1.619 ab
2-methylpropanoic acid	0.232 ± 0.032 d	2.609 ± 0.063 b	3.261 ± 0.135 a	2.777 ± 0.18 ab	2.616 ± 0.195 b	1.951 ± 0.295 c
Butanoic acid	1.307 ± 0.041 d	11.167 ± 0.266 c	23.707 ± 0.585 b	23.462 ± 2.133 b	23.829 ± 4.016 b	38.35 ± 3.545 a
3-methylbutanoic acid	0.253 ± 0.025 c	3.823 ± 0.078 ab	4.495 ± 0.102 a	4.154 ± 0.168 a	3.982 ± 0.237 ab	3.253 ± 0.661 b
Pentanoic acid	0.376 ± 0.03 e	6.458 ± 0.187 d	12.311 ± 0.726 c	16.404 ± 0.194 b	16.6 ± 1.744 b	26.023 ± 1.878 a
4-methylpentanoic acid	< 0.1	< 0.1	< 0.1	< 0.1	< 0.1	< 0.1
Hexanoic acid	< 0.1 c	0.085 ± 0.02 c	0.752 ± 0.064 a	0.525 ± 0.155 ab	0.342 ± 0.116 bc	0.711 ± 0.166 a
Heptanoic acid	< 0.1	< 0.1	< 0.1	< 0.1	< 0.1	< 0.1

DS – wheat variety DS8558-1, F - sample fermented with Lactiplantibacillus plantarum, NF – control sample, Bl – fermentation blank at 48 h, B0 – fermentation blank at 0 h. a-e Mean values denoted with different letters indicate significantly different values between the different sample groups. Results are statistically significant when p < 0.05.

Discussion

All of the bioconverted wheat wholemeal samples had pH levels between 3.9 and 4.0. While its optimal range was exceeded, various LAB strains can tolerate pH up to 3.7³⁹. Furthermore, L. plantarum showed good survivability at low pH, reaching over 8 log₁₀ CFU/g values⁴⁰.

Most of the AAs and total AAs in the bioconverted wheat wholemeal samples weren’t significantly altered. Similar trends in the crude protein content were observed in spontaneous SSF fermentations of wheat wholemeal as well⁴¹. While significant increases in crude protein in the SSF fermented samples are also observed whenever the fermentation process exceeds 24 h⁴².

While the total FAA concentrations in all the IVD samples did not differ significantly between substrate digested sample groups, certain amino acids, such as glutamine, GABA and proline, were significantly different. Those differences can be attributed to Lp. plantarum activity in the wheat wholemeal samples as these bacteria produce various protein-degrading enzymes (proteases and peptidases) as well as the GABA-producing enzyme glutamic acid decarboxylase, which converts glutamic acid into GABA^43,44. The results of the IVD hydrolysates mostly did not differ significantly between the bioconverted and control samples. While we found significant differences in the essential (threonine, methionine, valine, phenylalanine, leucine/isoleucine), non-essential (alanine, glutamic acid, proline, tyrosine) and total amino acid concentrations between the Herkus and DS samples. The IVD residue analysis reveals that fermentation significantly altered the digestibility of some of the non-essential (arginine, proline, tyrosine) and essential (valine, phenylalanine, leucine/isoleucine, lysine, histidine) amino acids. The in vitro digestibility and total AA digestibility were significantly affected by the bioconversion as well. However, the substrate was a significant factor for the digestion of the essential amino acids, such as valine, phenylalanine, isoleucine/leucine and histidine, some of the non-essential amino acids (glycine, proline, tyrosine), as well as total AA digestibility, the in vitro fermentability and the total IVD digestibility. The differences between the in vitro digestibility parameters due to the bioconversion can be attributed to the increased proteolytic activity in the fermented samples, as bacterial proteolytic enzymes are excreted into the sample matrix^45,46. Certain differences between the substrate digestibility are expected, as cultivars do have an effect on the ileal digestibility of protein in vitro and in vivo^47,48. The in vitro digestibility values of the individual AAs varied between 75.0 and 94.5%, as similar values(77–94.7%) were reported in arious in vivo and in vitro experiments^48–50. Interestingly, lysine had one of the lowest digestibility as reported in previous studies, as most of it lies in the aleurone fraction^48,51. Thus, bioconversion could be used to improve its bioavailability as well as other essential amino acids, such as phenylalanine, histidine, leucine/isoleucine, and valine, due to bacterial proteolytic activity.

Biogenic amines are products of amino acid degradation in the gut microbiome. Most of the BAs, such as histamine, tyramine and phenylethylamine, are produced by a single-step decarboxylation of amino acids via respective decarboxylase enzymes^52,53. Putrescine can be produced via the ornithine decarboxylase (single step) or arginine decarboxylase and agmatine deiminase (multiple steps using arginine)^52,53. We found that all of the factors and their interactions were significant for putrescine, cadaverine, histamine and tryptamine, while only substrate and both factors’ interaction were significant for tyramine concentration in the IVF samples (supplementary results file Table 7). Certainly, BAs are often produced by bacteria to regulate the pH or even have immunoregulation properties for the host^53–55. While BAs play a significant role in the animal gut and even other organs, such as the brain, acting as neurotransmitters, the putrefactive fermentation inside the gut can lead to a significant increase in those compounds, leading to serious complications^53,56,57. Some of those complications include injury of the intestinal mucosa, headaches, and digestive disorders^58,59. Certainly, the elevated levels of BAs are associated with such problems; they are often a clinical expression of pathogens, especially when the majority of the BAs in the piglet colon are of bacterial origin^57,60. Moreover, significant negative correlations between putrescine, cadaverine, histamine and total bacteria count (TBC) were found (supplementary results file Fig. 5), confirming the link between the faecal microbiota and its products. Though no significant correlations between the BAs and LAB count were found. This could suggest some mechanism of bacterial community modulation via the BAs, as certain negative correlations with other bacterial communities were observed in the literature⁶¹.

We found significant differences in propanoic, pentanoic, nonanoic and 2-methylpropanoic acids between the fermented and nonfermented samples, which were consistent between the methods (VOC analysis and SCFA analysis). Also, these compounds were among the most important compounds in the VIP analysis results. Additionally, most of the SCFAs, such as acetic, 4-methylpentanoic, nonanoic, pentanoic, octanoic, 2-methylbutyric and heptanoic acids were significantly different between the different substrates, as we found in the VIP analysis from the VOCs results, and from the factor analysis of the SCFAs and the VOCs. The SCFAs are important products of the indigestible polysaccharide fermentation, which can regulate gut metabolism, provide food for intestinal epithelial cells and maintain a favourable environment for the beneficial bacteria by controlling the population of pathogens^56,62. The branched-chain SCFAs, on the other hand, are produced by the amino acid fermentation in the colon^56,62. Thus, significant negative correlations between certain amino acids (valine, leucine, isoleucine, serine, threonine and histidine) and 2-methylpropanoic, 2-methylbutyric and 3-methylbutanoic acids in the IVF samples were established as well (supplementary results file Fig. 3). Also, the possibility of certain modulation of bacterial colonies was observed, as significant negative correlations were established between the SCFAs and TBC and LAB count values (supplementary results file Fig. 2). Interestingly, significant negative correlations were found between the non-branched SCFAs (butanoic, propanoic, pentanoic and hexanoic acids) and TBC, while branched-chain SCFAs (2-methylpropanoic and 3-methylbutanoic acids) and acetic acid were significantly negatively correlated with the LAB count. Branched-chain SCFAs production is often associated with high protein diet and is mainly produced by Bacteroides and Clostridium⁶³. These correlations might indicate certain microbiota modulation, as various correlations between different bacterial genera, especially Bacteroides and Clostridium, were observed in the literature^64,65.

Certain aliphatic aldehydes and alkanes were significantly contributing to the differences between the fermented vs. nonfermented (pentadecanal, hexadecanal, dodecanal, hexadecane, tridecane, octadecane, dodecane) and the Herkus vs. DS (nonanal, hexadecanal, octadecane, tetradecane, dodecane, oct-(2E)-enal) groups, which can be associated with differences in lipid metabolism in the sample groups^66–68. Indoles were significant for distinguishing between the Herkus and DS IVF samples, as we found indole from the VOCs and indole-3-acetic acid from the untargeted metabolites downregulated in the Herkus sample group. We found significant negative correlations between the indoles and tryptophan (supplementary results file Figs. 3 and 4), as indoles are associated with bacterial degradation of tryptophan⁶⁹.

P-cresol is a product of tyrosine and phenylalanine degradation in the gut, mostly by Escherichia coli, Clostridium sporogenes and Clostridium difficile, which is a protein-binding uremic toxin⁷⁰. Other phenolic compounds, such as 4-ethylphenol, are also associated with tyrosine and phenylalanine catabolism by Clostridium species⁷¹. While it’s found in various fermented food products, such as wine, its metabolites at high concentrations in blood can cross the blood-brain barrier and cause serious damage to the host’s organism^72,73. Other products of bacterial phenolic amino acid degradation include 3-hydroxyphenylpropionic acid via the deamination of tyrosine, benzenacetic acid via the oxidative or non-oxidative decarboxylation of phenylalanine, 4-hydroxyhydrocinnamic acid via the catabolism of tyrosine, β-phenyllactic acid via the metabolism of phenylalanine^74–77. While many of those fermentation products, such as 4-hydroxyhydrocinnamic acid, benzenacetic acid, 3-hydroxyphenylpropionic acid, hydrocinnamic acid (aka dihydrocinnamic acid or benzenepropanoic acid), dihydroferulic acid, can also be attributed to phenolic acids, such as ferulic and caffeic acid, catabolised by the gut microbiome^78–80. We found benzenacetic, dihydrocinnamic, dihydroferulic and 4-hydroxyhydrocinnamic acids upregulated in the DS sample group (Fig. 7). Significant negative correlations between benzenacetic acid, dihydrocinnamic acid, 4-hydroxyhydrocinnamic acid and phenolic amino acids (tyrosine and phenylalanine) (supplementary results file Figs. 3 and 4), suggesting their origin to be of the amino acid catabolism. While 3-hydroxyphenylpropionic acid was significantly upregulated in the fermented samples (Fig. 7-a) and in the DS samples (Fig. 7-b) as well. It may be a product of phenolic acid degradation, as we failed to find significant correlations with phenolic amino acids (supplementary results file Fig. 4). The sole product of phenolic amino acids, β-phenyllactic acid, was upregulated in the Herkus sample group (Fig. 7). It was also significantly correlated with the TBC (supplementary results file Fig. 2), possibly modulating the bacterial community, as its modulating properties were established in the literature⁸¹. We found that cresol was upregulated in the fermented sample group and downregulated in the Herkus sample group (Fig. 4). We found significant upregulation of 4-hydroxyphenylacetic acid, a precursor to p-cresol, in the non-fermented sample group (Fig. 7). This suggests certain inhibition of its conversion into p-cresol in the non-fermented samples. The factors were significant for both p-cresol and 4-ethylphenol, while factor interaction was significant only for cresol. The significant negative correlations were established with phenylalanine and tyrosine, as well as other amino acids (supplementary results file Fig. 3). Moreover, the potential modulation of specific bacterial communities was indicated by the significant correlation observed between the p-cresol and LAB counts, a relationship that has also been reported in the literature⁷⁰.

While 2,4-ditertbutylphenol is considered an exogenous compound, several species of bacteria do produce it in significant amounts^82,83. Other possible exogenous compounds include 2-phenoxyethanol, as a preservative, naphthalene, as a polycyclic aromatic compound^84,85. Phenylethyl alcohol could be exogenous, as it is used as a preservative or in perfumes due to its pleasant aroma, while its origins can be linked to yeast activity⁸⁶. Other phenolic metabolites, such as isocreosol, were detected in urine⁸⁷. Acetophenone is associated with the bacterial catabolism of cinnamic acid, and γ-decalactone can be produced by LAB via fatty acid metabolism^88,89.

Acetone production can be attributed to solventogenic clostridia, producing other compounds such as butanol, acetic acid and lactic acid^90,91. VIP analysis results showed that acetone was significant for the fermented – nonfermented PLS-DA model, as bioconversion of the substrate can reduce complex carbohydrates into smaller fragments, possibly so that other bacteria can more easily compete with clostridia as, it is able to ferment complex carbohydrates^92,93.

Aliphatic organic acids, such as 2-deoxytetronic acid, succinic acid (the intermediate of carbohydrate fermentation to propionate and lysine degradation), α-hydroxyglutaric acid, 5-aminopentanoic acid (metabolites of lysine degradation by Escherichia coli), and glyceric acid are produced by bacteria in the gut^94–97. Pyroglutamic acid is another typical product of the LAB fermentation⁹⁸. Propane-1,3-diol production is associated with carbohydrate fermentation of various Clostridium, Salmonella, Bacteroides strains, and others⁹⁹. Lactic acid is a classic metabolite of carbohydrate fermentation by the LAB^56,100. Interestingly, succinic acid and α-hydroxyglutaric acid were upregulated in the Herkus samples, while 5-aminopentanoic acid was upregulated in the DS samples. This could suggest possible inhibition of 5-aminopentanoic acid conversion to glutarate semialdehyde via 4-aminobutyrate-2-oxoglutarate transaminase and subsequently to succinate⁹⁵. While the same enzyme can be used in E. Coli to catabolize GABA via the GABA shunt mechanism^101–104. We found positive correlations between the GABA and lysine metabolites (supplementary results file Fig. 4), suggesting that the abundance of substrate reduces the GABA shunt activity^101–104. Glyceric and lactic acid were also upregulated in the Herkus samples (Fig. 7). While certainly lactic acid is mainly produced by carbohydrate fermentation in the gut microbiome, it is also produced via branched-chain amino acid catabolism to pyruvate and subsequently to lactate by auxotrophic bacteria¹⁰⁵. We found significant positive correlations between valine, leucine and isoleucine and lactic acid (supplementary results file Fig. 4), suggesting the absence of auxotrophic bacterial activity. Glyceric acid, although a common product of carbohydrate and glycerol fermentation, may exert detrimental effects at elevated concentrations, as it can act as a metabolic toxin¹⁰⁶.

Conclusions

Lp. plantarum strain is effective for the SSF of wheat cereals, while the total AA composition of bioconverted substrates was not significantly altered for the most part; the in vitro digestibility parameters were altered, notably increasing in vitro total digestibility and some essential and non-essential AAs as well. The differences between different wheat substrates for the in vitro digestibility were also established. The significantly higher concentrations of GABA were observed in the bioconverted IVD samples. The production of most of the BAs was significantly affected by both factors (bioconversion and substrate) and their interaction. The majority of the SCFA production was significantly different due to the substrate factor, while the bioconversion of the substrate affected all of the SCFAs, except acetic acid. Also, propionic acid production was affected by both of the analysed factors. Indole compounds were downregulated in the Herkus samples. We found significant differences in aromatic amino acids and other phenolic acid degradation products between the substrates. Acetone production was significantly affected by the bioconversion of the substrate. Both factors were significant for GABA production in the IVF samples. Substrate was a significant factor for the LAB count. The PCA of the VOCs showed separation between groups, while the untargeted metabolomics results only showed separation between the bioconverted sample groups. The bioconversion and the substrate had significant effects on the major in vitro digestibility and fermentability characteristics, as well as the overall metabolome in the simulated colon fermentation. Therefore, noting substrate selection and its bioprocessing importance for piglets’ health and nutrition.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

DS

DS8558-1

SCFA

short-chain fatty acid

VOC

volatile organic compound

FAA

free amino acid

BA

biogenic amine

IVF

in vitro fermentation

SSF

solid state fermentation

LAB

lactic acid bacteria

GRAS

“generally recognised as safe”

QPS

“qualified presumption of safety”

GABA

γ-aminobutyric acid

MRS

Man–Rogosa–Sharpe broth

SPME

solid-phase microextraction

BSTFA

N, O-bis(trimethylsilyl)trifluoroacetamide)

ANOVA

analysis of variance

Tukey HSD

Tukey’s honest significant difference

PCA

principle component analysis

PLS-DA

partial least square discriminant analysis

VIP

variable importance in projection

DS_F

DS8558-1 sample fermented with Lactiplantibacillus plantarum

DS_NF

DS8558-1 control sample

Herkus_F

Herkus sample fermented with Lactiplantibacillus plantarum

Herkus_NF

Herkus control sample

AA

amino acid

IVD

in vitro digestion

TBC

total bacteria count

Author contributions

Conceptualization, E.B., E.M; methodology, E.M., V.S., Z.L.; software, E.M.; validation, E.M, D.K. and V.S.; formal analysis, E.M., V.S.; investigation, E.M., D.K., E.B; resources, E.B., Z.L., V.R. S.B., M.R.; data curation, E.M., V.S., E.B; writing—original draft preparation, E.M., E.B., D.K; writing—review and editing, E.M., V.S., D.K., S.B., M.R., Z.L., V.R., E.B.; visualization, E.M.; supervision, E.B.; project administration, E.B. All authors have read and agreed to the published version of the manuscript.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.