Eight types of pulses commonly consumed in Korea were selected for analysis as follows: soybeans (Glycine max, SB), kidney beans (Phaseolus vulgarius, KB), cowpeas (Vigna unguiculata, CB), chickpeas (Cicer arietinum, CP), black beans (Rhynchosia nulubilis, BB), mung beans (Vigna radiate, MB), Heunguseul (Vigna anaularis L. Heunguseul, HS), and red beans (Vigna anaularis L., RB). The samples were prepared in a consumable form by applying the most common cooking methods suggested by the National Standard Food Composition Table (provided by the Rural Development Administration, 10th version). Consequently, each sample was subjected to either boiling or stir-frying, with the specific methods outlined as follows: The SBs (Sangju, Gyeongsangbuk-do, Korea) were boiled for 40 min in water at three times their weight after a 4-hour soaking. CBs (Gurye, Jeollanam-do, Korea) were boiled for 50 min in water at 10 times their weight after soaking for 8 h. The KBs (Geochang, Gyeongsangnam-do, Korea), HSs (Wanju, Jeollabuk-do, Korea), MBs (Muan, Jeollanam-do, Korea), and RBs (Muan, Jeollanam-do, Korea) were boiled for 1 h in water at 10 times of their weight without prior soaking. The CP samples (USA) were boiled for 15 min in water at 10 times their weight, following a 6-hour soaking. The roasted BBs (Sangju, Gyeongsangbuk-do, Korea) were obtained from a commercial market.

A gastrointestinal model was constructed and simulated to examine the alteration of nutrient contents throughout the digestive process, which encompasses the oral, gastric, intestinal, and BBMV phases (Minekus et al., 2014). In the oral phase, cooked pulses were mixed with a 1:1 solution of simulated salivary fluid (KCl (Potassium chloride P9333, Sigma Aldrich Co, Seoul, Korea) 15.1 mmol/L, KH₂PO₄ (Potassium phosphate monobasic P0662, Sigma Aldrich Co.) 3.7 mmol/L, NaHCO₃ (Sodium bicarbonate S5761, Sigma Aldrich Co.) 13.6 mmol/L, MgCl₂(H₂O)₆ (Magnesium chloride hexahydrate 63,068, Sigma Aldrich Co.) 0.15 mmol/L, (NH₄)₂CO₃ (Ammonium carbonate 207,861, Sigma Aldrich Co.) 0.06 mmol/L, pH 7.0) and enzyme (75 U/mL salivary α-amylase solution (α-Amylase from human saliva A0521, Sigma Aldrich Co.), 0.3 M CaCl₂ (Calcium chloride C1016, Sigma Aldrich Co.), water). The solution was digested at 37°C for 2 min. The gastric phase entailed the mixture of the oral digested samples and a 1:1 solution of simulated gastric fluid (KCl 6.9 mmol/L, KH₂PO₄ 0.9 mmol/L, NaHCO₃ 25 mmol/L, NaCl (Sodium chloride S9888, Sigma Aldrich Co.) 47.2 mmol/L, MgCl₂(H₂O)₆ 0.1 mmol/L, (NH₄)₂CO₃ 0.5 mmol/L, pH 3.0) and enzyme (2000 U/mL porcine pepsin stock solution (Pepsin from porcine gastric mucosa P7000, Sigma Aldrich Co.), 0.3 M CaCl₂, 1 M HCl (Hydrochloric acid 4095–3705, Daejung, Seoul, Korea), water). The solution was incubated at 37°C for 120 min. Similarly, the intestinal phase was conducted through the mixing of gastric digested samples with a 1:1 solution of simulated intestinal fluid (KCl 6.8 mmol/L, KH₂PO₄ 0.8 mmol/L, NaHCO₃ 85 mmol/L, NaCl 38.4 mmol/L, MgCl₂(H₂O)₆ 0.33 mmol/L, (NH₄)₂CO₃ 0.6 mmol/L, pH 7.0) and enzyme (200 U/mL pancreatin from porcine pancreas [Pancreatin from porcine pancreas P7545, Sigma Aldrich Co.]), 160 mM bile salts (Bile extract porcine B8631, Sigma Aldrich Co.), 0.3 M CaCl₂, water and reacting at 37°C for 120 min. As the final digestive phase, 77.3 U/mL of BBMV was added for the digesta to react at 37°C for 180 min (Oku et al., 2011). Subsequently, the mixture was heated at 100°C for 10 min to inhibit further digestion.

The samples that underwent in vitro digestion were collected and placed in a 1 kDa molecular weight cut-off cellulose membrane (Spectra/Por^® 6 Standard RC Pre-Wetted Dialysis Tubing). They were then soaked in 10 mM NaCl at 37°C for 24 h to remove low-molecular-weight components (Shiowatana et al., 2006; Bouayed et al., 2011; Hettiarachchi et al., 2021). Once the dialysis process was complete, the digested material was transferred to the in vitro gut fermentation model.

The cooked pulses were subjected to oral, gastric, small intestinal digestion, and dialysis phases, and the major nutrient contents of each fully digested pulse sample were analyzed. A total of six types of saccharides, including oligosaccharides, disaccharides, and monosaccharides, were determined. In addition, 22 essential and non-essential amino acids and 6 fatty acids were quantified. These nutrients serve as energy sources for the gut microbiota in the simulated human gut, and the analysis data were used to examine the correlation between nutrient content and microbial growth.

All experiments were conducted in triplicate, and data entry, descriptive statistics, and the construction of bar graphs of the relative abundance of genera for each sample were performed using Microsoft Excel 2016 (Microsoft Corporation, Redmond, Washington, United States). All statistical tests, analyses, and visualizations, except those practiced by Excel were performed using RStudio (2023.12.1. +402 for Windows). To analyze differences in relative microbial abundances during colonic fermentation of digested pulse, a Wilcoxon rank-sum test (at the 95% confidence level) was used. For comparisons of α-diversity and β-diversity, the Shannon index (SI) was calculated for 0, 12, and 24 h fermentation samples of each pulse sample, and Bray-Curtis distances were used for multidimensional scaling (MDS) visualization. Significant differences in the nutrient contents and the SIs between samples were tested by ANOVA and Tukey’s method (p < 0.05) and presented in a box plot. Principal component analysis (PCA) was performed to identify the distribution of nutrient content—carbohydrates, amino acids, and fatty acids—within the digested pulse samples, selecting principal components with an accumulated variance greater than 80% and generating a biplot expressing loadings and scores. Primary variables for the network analysis were chosen from the factor analysis conducted on the carbohydrates, amino acid, and fatty acid content, as well as on the difference in absolute microbial abundance between 0 h and 12 h samples. Variables with a loading of ≥0.5 were selected, and only edges with a correlation >0.5 were included in the positive correlation network, while edges with a correlation < −0.5 were included in the negative correlation network. The centrality of each variable in the network analysis was calculated and represented by the size of the nodes. Communities within the network were identified using the Infomap method and displayed in different colors.

The degraded forms of carbohydrates in the pulses subjected to digestion were quantified, followed by the oral, gastric, small intestinal phases, and dialysis phases (Supplementary Table S1). During the digestive process, carbohydrates with high molecular weight were broken down into monosaccharides, and most of the low molecular weight compounds were eliminated by dialysis. However, some undigested oligosaccharides, such as stachyose, remained in the samples. The stachyose content in the digested samples of CB, CP, and BB was 63.84 ± 1.80 μg, 58.56 ± 0.99 μg, and 42.17 ± 0.45 μg, respectively, while 9.14 ± 0.57 μg of raffinose remained only in the digested BB. The monosaccharide glucose was detected in the digested samples of KB, CP, and CB at concentrations of 246.77 ± 1.93 μg, 228.95 ± 1.24 μg, and 206.96 ± 0.77 μg, respectively. Based on the PCA biplot (Figure 1A), samples other than KB, CP, CB, and BB contained relatively lower amounts of free saccharides.

Free amino acid content was also examined after proteins were degraded during the digestion process and subsequently filtered through dialysis phases. The consistent directionality of the loading vectors on the PCA biplot (Figure 1B) corresponding to each amino acid suggests a positive correlation in amino acid content across the samples. Notably, the loading vectors for glutamine, ornithine, and citrulline deviated slightly from the direction observed for other amino acids, indicating the possibility of differential levels of these amino acids in some pulses. Overall amino acid levels in eight pulses were high at glutamine, arginine, phenylalanine, and leucine (Supplementary Table S2). Specifically, glutamine showed particularly high levels, with contents of 92.05 ± 3.93 μg, 30.76 ± 0.13 μg, 23.80 ± 0.26 μg, 23.53 ± 0.36 μg, 20.54 ± 0.19 μg in the digested samples of BB, CP, CB, SB, and KB. Arginine also showed high concentrations of 27.60 ± 0.18 μg and 19.99 ± 0.23 μg in the digested CP and SB samples, while phenylalanine was present at levels of 21.76 ± 0.19 and 17.57 ± 0.73 in the CP and BB samples. Similarly, leucine was found in high concentrations of 27.62 ± 1.19 μg, and 23.39 ± 0.09 μg in BB and CP digesta, indicating that CP and BB digesta contained the highest overall levels of amino acids. As confirmed by the PCA plot (Figure 1B), relatively lower levels of amino acids were observed in samples from RB, KB, MB, and HS. Ornithine and citrulline were exclusively detected in the digested BB sample.

For free fatty acids, they were largely eliminated after dialysis, with only trace amounts remaining in the samples. According to the PCA plot (Figure 1C), the digested samples of BB, RB, and CP had the lowest fatty acid contents. In contrast, the digested SB sample contained the highest levels of oleic acid (10.02 ± 0.01 μg) and linoleic acid (25.27 ± 0.01 μg) (Supplementary Table S3). The digested KB sample also had a relatively high level of free fatty acids, with palmitic acid (3.67 ± 0.00 μg) and α-linolenic acid (5.51 ± 0.01 μg) present at higher levels compared to the other pulses.

The eight different digested pulses were cultured using an in vitro gut fermentation model, and products were collected at 0, 12, and 24 h in each pulse digesta. After collection, the genus-level 16S (V3-V4) rRNA sequencing and taxonomic profiling were performed, and the relative abundance of microbes was visualized in a bar plot chart (Figure 2). In the initial (0 h) colonic fermentation of the BB digesta, Enterobacter represented approximately 40% of the total gut microbiota, while the Prevotella 9 lineage accounted for around 15%. However, as the fermentation progressed, both Enterobacter and Prevotella 9 decreased in their portion, while Escherichia-Shigella, Fusobacterium, and Bacteroides increased, resulting in a simplified microbial community.

Besides, in the CB digesta, the initial microbial composition displayed a distribution of the diverse genus, including Prevotella 9, Faecalibacterium, Blautia, and Bacteroides, each representing about 15% of the community. But, over fermentation time, Escherichia-Shigella, Bifidobacterium, and Bacteroides dominated the community. This trend of simplification of the microbial composition was particularly pronounced in the CP digesta, with a remarkable increase in Escherichia-Shigella and Bacteroides, which came to account for a large portion of the microbial community. Similar patterns were observed during the fermentation of the KB digesta, where the relative abundance of Bifidobacterium increased.

Although the SB digesta initially contained a diverse community, with Alistipes, Akkermansia, Bacteroides, Aquamonas, and various Eubacterium groups comprising a substantial proportion, the Eubacterium groups increasingly constituted the community as fermentation progressed. In particular, Akkermansia gradually decreased during fermentation. Overall, a common trend was observed in the fermentation of BB, CB, CP, and KB digesta, characterized by an increase in the relative abundance of Escherichia-Shigella, Bacteroides, and Clostridium sensu stricto 1, along with a decrease in Blautia.

In the fermentation of HS digesta, a marked decrease in the relative abundance of Escherichia-Shigella, which comprised about 60% of the primary microbial community, was observed over time. Concurrently, the proportions of Bacteroides, Fusobacterium, and Bifidobacterium increased. A similar pattern was found in the fermentation of MB digesta, with a decline in the relative abundance of Bacillus and an increase in Bifidobacterium. An inverse relationship between the relative abundances of Escherichia-Shigella and Bacteroides was identified in most pulse digesta.

During the fermentation of RB digesta, the relative abundance of Escherichia-Shigella decreased significantly, while the portions of Bacteroides and the Eubacterium groups increased. This trend was consistent with the results seen in the SB digesta where Eubacterium oxidoreducens, Eubacterium hallii, and Eubacterium eligens groups collectively accounted for approximately 45% of the microbial community at the 12 h and 24 h time points. Furthermore, unlike the other pulses, RB digesta showed an increase in the relative abundance of Aquamonas, Alistipes, and Akkermansia. While Aquamonas is prevalent in environmental materials such as water and has not been studied in the context of human health or food samples, the positive health effects of Alistipes and Akkermansia are described in the discussions of Table 1.

Furthermore, diversity analyses were conducted to provide a quantitative framework the patterns of microbial variability among the samples that were illustrated in the aforementioned bar plot. The ɑ-diversity (SI) of the pulse samples at each in vitro fermentation time was calculated to assess the genus complexity of the fecal microbiota within each pulse (Figure 3). In most samples, there was a decrease in diversity at 12 h of colonic fermentation compared to 0 h, with a slight increase observed at 24 h. However, in RB, HS, and MB, diversity consistently increased over time. The mean SIs at 0 h and 12 h for each sample were as follows: SB 6.25 and 5.24, KB 6.57 and 5.22, CB 6.60 and 5.54, CP 6.63 and 4.86, and BB 6.24 and 5.72, with a similar range of decrease. In contrast, the diversity increased from 5.23 to 6.17 in RB, from 5.30 to 5.84 in HS, and from 5.89 to 5.95 in MB, indicating an upward trend in these three pulses.

The β-diversity of the microbial communities was analyzed using Bray-Curtis distances between samples (Figure 4). Based on the calculated distances, a general distribution was observed by the MDS method, which exerted significant shifts in the composition of the gut microbiota at either 12 or 24 h of fermentation compared to 0 h in all pulse samples. Since KB, CB, CP, BB, and SB showed a decrease also in α-diversity as fermentation progressed, it was suggested that the digestive residues of these samples reduced the richness of the microbial distribution in the colon, simultaneously altering the composition of the microbiota. However, RB, HS, and MB digesta positively influenced the distribution of a variety of microorganisms during colonic digestion.

Changes in gut microbiome genera were observed in three pulses where health-positive bacteria increased during colonic digestion (Table 1). As the relative abundances of Bacteroides, Bifidobacterium, and Fusobacterium were found to be increased by colonic fermentation in all three samples, the relative growth of Coprococcus and Lachnoclostridium was uniquely depicted in RB, while Odoribacter and Akkermansia were specifically increased in MB. The species-level analysis performed by Shotgun whole genome sequencing for these genera (Figure 5) showed that these genera contained only O. splanchnicus and A. muciniphila, respectively. Meanwhile, the proportions of Alistipes and Lachnoclostridium increased in both RB and MB. Alistipes belongs to a relatively recently discovered sub-branch genus within the phylum Bacteroidetes. The Shotgun sequencing identified the following four species for Alistipes: A. onderdonkii (69%), A. shahii (19%), A. senegalensis (5%), and A. finegoldii (4%).

The relative abundances of Sutterella and Akkermansia also increased in MB and HS, while the relative growth of Enterobacter was promoted in RB and HS. In the case of Enterobacter, only E. hormaechei constituted a minute fraction, comprising 0.003% of the entire Enterobacteriaceae family.

16S (V3-V4) rRNA sequencing and taxonomic profiling unveiled an enriched microbial community in RB, MB, and HS samples along with the fermentation time. Furthermore, there was a decrease in the relative abundance of health-negative genera such as Escherichia-Shigella, with an increase in beneficial bacteria including Bacteroides and Eubacterium groups. The term Escherichia-Shigella refers to both Escherichia coli and Shigella, which are genetically similar and difficult to be distinguished by 16S rRNA analysis. While most strains of E. coli are nonpathogenic and symbiotic in the human gut, Shigella is known to cause dysentery through intestinal infection (Wright et al., 2015; Xu et al., 2022). On the other hand, Bacteroides, although functionally diverse at the species level, plays a critical role in degrading complex carbohydrates in the gut, which supports intestinal immune function and homeostasis. Research has demonstrated that individuals with inflammatory bowel diseases (IBD), such as Crohn’s disease and ulcerative colitis, tend to have higher Escherichia-Shigella and lower Bacteroides levels (Wright et al., 2015).

Eubacterium oxidoreducens is known to participate in the oxidation–reduction reactions of various organic compounds in the gut and has been studied for its health-promoting role in cancer research, including esophageal (Li et al., 2024) and small cell lung cancer (Yang et al., 2024). Eubacterium hallii is an important gut bacterium that produces short-chain fatty acids (SCFAs), such as butyrate and propionate, which serve as an energy source for intestinal cells (Engels et al., 2016) and have been reported to contribute to anti-inflammatory effects and improved insulin sensitivity (Mukherjee et al., 2020; Seegers et al., 2021). Eubacterium eligens, known for its anti-inflammatory properties, regulates intestinal immune responses in the gut and may have protective effects against inflammatory bowel disease (IBD) (Chung et al., 2017; Olendzki et al., 2022). All of these species are known to metabolize complex carbohydrates and dietary fiber as energy sources, suggesting that measuring the dietary fiber and oligosaccharide content of RB may help identify the specific factors influencing the growth of these microorganisms.

While RB, HS, and MB positively influenced the distribution of a variety of microorganisms during colonic digestion, diversity analyses revealed that both α- and β-diversity varied among the initial fermentation samples (0 h). It has been previously reported that environments with high microbial diversity are known to be resilient to microbial perturbations, and ecological theory posits that environments with low microbial diversity are inherently less receptive to the colonization of new microbial species, as reduced diversity limits the availability of ecological niches and competitive opportunities (Lozupone et al., 2012; Bourdeau-Julien et al., 2023). Nevertheless, the present results, where samples with initially high diversity experienced a decrease while those with low diversity saw an increase, illustrated that RB, MB, and HS may have significant factors that promote a favorable gut environment compared to the others.

Not only was diversity enhanced in the three pulses, but there was also a recognized growth of multiple health-promoting microbiota including Bacteroides, Bifidobacterium, and Coprococcus. Bacteroides spp. protect the gut microbiome from pathogens and provide an energy source (Zafar and Saier, 2021). Approximately 50% of the carbohydrate in pulses is starch, which is converted to resistant starch (RS) during heat treatment. The metabolic capabilities of the gut microbiota vary depending on structural differences in RS, with RS4 consumption specifically known to increase the abundance of Bacteroides and Bifidobacterium (Kadyan et al., 2022).

Meanwhile, SCFA-producing Coprococcus (Nogal et al., 2021) has been found to grow with high quality diets. Since SCFAs are produced by the metabolism of anti-inflammatory bacteria that participate in energy synthesis in the human body, increasing the abundance of Coprococcus may help the growth of beneficial gut microbiomes (Laitinen and Mokkala, 2019). Coprococcus and Lachnoclostridium, along with Lachnospira, belong to the family Lachnospiraceae, which, according to NCBI, consists of 58 genera and several unclassified strains (Vacca et al., 2020). C. comes, consisting of 86% of Coprococcus genera, is a beneficial butyrate-producing bacterium found in the human gut. Butyrate, one of its major metabolic byproducts, plays a pivotal role in maintaining the intestinal barrier, reducing inflammation, and facilitating gut health. Studies have linked C. comes to positive outcomes in mental health, particularly in reducing symptoms of depression, due to its involvement in the gut-brain axis (Notting et al., 2023). Moreover, butyrate production by C. comes has been associated with anticancer properties, particularly in protecting against colorectal cancer by supporting the gut epithelium and regulating inflammatory processes (Liu et al., 2024). Clostridium symbiosum, the only species detected in the genus Lachnoclostridium, contributes to the fermentation of dietary fiber and aids in the production of SCFAs such as butyrate, which supports gut barrier function and reduces inflammation. By producing SCFAs, it can regulate metabolism and immune responses, playing a protective role in metabolic conditions like type 2 diabetes and colorectal cancer (Wang et al., 2012; Ren et al., 2024). However, overgrowth of C. symbiosum has been observed in conditions of gut dysbiosis, which may be associated with diseases such as obesity and inflammatory bowel disease (IBD) (Rodríguez et al., 2020).

The fermentation of MB digesta led to an enhanced relative abundance of Odoribacter, while HS digesta promoted the proliferation of Butrycimonas. Previous studies have indicated that Odoribacter, Butyricimonas, and Bacteroides are positively related with human cognitive function and contribute to the anti-inflammatory pathway and the modulation of the immune system (Zeng et al., 2021). The beneficial gut bacterium O. splanchnicus is known to utilize dietary fiber and resistant starch as its primary energy source, producing short-chain fatty acids (SCFAs) (Hiippala et al., 2020; Ren et al., 2024). During the fermentation of RB and MB, Alistipes showed a significant increase, with Alistipes onderdonkii emerging as the predominant species within this genus, and is recognized as a characteristic member of the gut microbiome with anti-inflammatory properties and involvement in the inhibition of tumor necrosis factor (TNF). It has been reported to play a beneficial role in cancer immunotherapy by modulating the tumor microenvironment (Parker et al., 2020; Li et al., 2023).

The genera Sutterella and Enterobacter, which showed increased abundance in HS and MB, and HS and RB respectively, are gut microbes with similar potential impacts on human health. Sutterella has been associated with inflammatory conditions such as irritable bowel syndrome (IBS) and autism spectrum disorders, suggesting that it may play a role in gut inflammation and immune modulation (Hiippala et al., 2016). On the other hand, Enterobacter species are opportunistic pathogens causing infection and dysbiosis, especially in immunocompromised individuals (Moreira de Gouveia et al., 2024). Both genera can affect gut homeostasis and contribute to disease states.

Besides, the relative abundance of Akkermansia significantly increased in HS and MB. Akkermansia muciniphila is one of the most abundant single species in the human gut microbiota, comprising 0.5–5% of the total bacteria. It degrades the mucin layer of the intestinal mucosa and produces beneficial metabolites such as acetate and propionate. Through this mechanism, A. muciniphila has been shown to reduce body fat mass and adipose tissue inflammation while improving insulin sensitivity, thereby enhancing gut barrier function (Cani and de Vos, 2017).

Protein degradation during pulse digestion led to a consistent increase in amino acid levels. The high interconnectivity between amino acids and sugars observed in the network analysis suggested that the pulses used in the experiment had proportional amounts of protein and carbohydrates. Fatty acids did not have direct positive networks with amino acids or sugars, indicating that the fat content of the pulses does not necessarily align with protein or carbohydrate level and may vary among the pulses.

In the positive correlation network of the eight pulses, few links were observed between microbes and nutrients. Fusicatenibacter, CAG-56, and Bifidobacterium were positively correlated with the fatty acid α-linolenic acid (ALA), while Anaerostipes was influenced by amino acids and associated with the growth of Blautia, Monoglobus, and Fusicatenibacter. Coakley et al. (2009) reported the ability of Bifidobacterium strains to convert linoleic acid to c9, t11-conjugated linoleic acid (CLA) and grow in the presence of anti-inflammatory ALA. Since the anti-inflammatory properties may create a favorable gut environment that is likely to foster the proliferation of SCFA-producing bacteria (Yang et al., 2023), the growth of SCFA-producing Fusicatenibacter (Pihelgas et al., 2024) might have been affected by ALA. Anaerostipes, a butyrate-producing genus, interacts with amino acids such as glutamate and lysine, which serve as substrates for SCFA production, especially butyrate, known for its anti-inflammatory effects (Singh et al., 2023) which can in turn promote the growth of Fusicatenibacter.

Cross-feeding interactions explain the correlations within this bacterial community, where one microbe’s byproducts support the growth of others. These interactions, several bacteria degrade dietary components into intermediate metabolites such as acetate or lactate, which are then sequentially used as nutrient substrates by butyrate-producing bacteria. Rivière et al. (2016) reported relationships between acetate- or lactate-producing Bifidobacterium and butyrate-producing colonic bacteria, such as Faecalibacterium prausnitzii (clostridial cluster IV) and Anaerostipes. Accordingly, positive associations of the SCFA metabolite content with microbial alteration were more frequent than those with nutrient content. This may stress that the composition and activity of the gut microbiota are of paramount importance in determining the production of key metabolites.

Therefore, optimizing microbial conditions may yield greater advantages to gut health than focusing solely on individual nutrient contents. Recent studies have highlighted the potential of dietary interventions in modulating the gut microbiome by enhancing beneficial microbial populations or increasing the production of SCFA. For instance, Purdel et al. (2023) reported that fasting and/or time-restricted regimens can achieve gut health. Time-restricted fasting, intermittent fasting, and Ramadan fasting protocols were shown to increase Akkermansia muciniphila and Bacteroidetes, associated with improved metabolic health and reduced gut inflammation. Caloric restriction and alternate day fasting resulted in increased microbial diversity and promoted Bacteroides and SCFA production. In addition, water-only fasting and fasting-mimicking diets led to a high abundance of beneficial microbes like Lactobacillus and Bifidobacterium, improving immune function and gut health.

The synergistic role of prebiotics―indigestible dietary fibers that provide food for beneficial bacteria―with probiotics in promoting a diverse and stable gut microbiome has also been emphasized (Shang et al., 2024). Combining probiotic-rich meals with prebiotic fibers enhances intestinal permeability defenses through SCFA production, maintaining stability in gut flora despite dietary or environmental changes. A previous study by Starke et al. (2023) reported that amino acid-cross-feeding bacteria contribute to the compositional stability of the gut microbiome, with no evidence that dietary amino acid intake affects the frequency of auxotrophy, suggesting that simple manipulation of nutrient content may not be sufficient to regulate microbial-nutrient metabolic pathways.

In the negative network analysis of the eight pulses, a community centered around Bacteroides, Fusobacterium, Sutterella, and Enterobacter exhibited the greatest centrality and had strong correlations with amino acids. These findings were on the opposite side with previous studies that have identified various taxa involved in amino acid metabolism, including nitrogen metabolism and the tricarboxylic acid (TCA) cycle (Kanehisa et al., 2023), implying the positive relationships between those microbes and amino acids. For example, some Enterobacter and Sutterella species are engaged in the glutamate metabolism of the TCA cycle, producing glutamine as a byproduct. The pathway also includes the conversion of glutamate to GABA, suggesting that a decrease in glutamate might lead to a reduced GABA level (Kanehisa et al., 2023). Given that no relationships were recognized for these bacteria with amino acids even in the positive network of present study, it was discerned that an imbalance in the overall nutritional state provoked by amino acids may have created an unfavorable gut environment for bacterial growth. Despite their reliance on amino acids for energy, their metabolism can produce toxic byproducts such as ammonia and hydrogen sulfide. These substances can disrupt gut homeostasis, contribute to dysbiosis, and negatively affect the balance of microbial populations (Lin et al., 2017).

The positive and negative networks for RB, MB, and HS had higher connectivity between communities, with a broader range of microbes interacting directly or indirectly with fatty acids and amino acids compared to the complete network. This could interpret the improved microbial diversity observed in these three pulses. Since the three pulses comprising the subset network were not identified for their nutrient abundance, it was proposed that the overall nutritional status in RB, MB, and HS influenced the nutrient-microbe interactions.

Evidence from this study insists that facilitating a gut environment driven by the comprehensive nutrient balance, rather than individual nutrients, promotes eubiosis, thereby enhancing microbial diversity and stimulating the growth of beneficial microorganisms. In fact, the linear regression demonstrated the quantitative relationship between nutrient content and microbial diversity. Excess sugars and amino acids were found to negatively affect the gut environment, with a weak correlation between fatty acids and the SI. However, fatty acid content as a ratio to sugars or amino acids significantly influenced gut microbial diversity. Consequently, the nutritional context-dependence for maintaining gut health was highlighted by elucidating the importance of avoiding overconsumption of sugars and amino acids while maintaining a balance with fatty acids. In the human colon, an overload of certain nutrients can promote the overgrowth of specific microorganisms, which can disrupt the environment for others and reduce overall microbial diversity. Amino acids, in particular, can facilitate the growth of pathogenic amino-acid-degrading bacteria, such as Clostridium, and produce metabolites such as ammonia, phenols, and amines that can alter gut pH.

A previous opinion by Rios-Covian et al. (2017) suggested the metabolic remodeling of the host and Bacteroides in response to dietary organic nitrogen/carbohydrate combinations. In addition, a study in mice by Holmes et al. (2017) described that the pattern of protein and carbohydrate intake drives diet-microbiome interactions. The interdependence of sugars and amino acids in the digested pulses used in this study obscured the association of the ratio of carbohydrates to amino acids with microbiota, which also emphasized the relevance of the context of the food matrices in determining microbiome metabolism. Although this study concluded that balancing fatty acids with sugars or amino acids is relevant to a healthy microbiome, the context-based effect may not be distinguished when examining food groups in which carbohydrates are predominant, such as fruits and vegetables. Therefore, elucidating the multidirectional connections between nutrients and microbes using a variety of food ingredients might notify the way to combine foods that complement each ingredient’s nutritional benefits for gut microbiota health.

Several limitations need to be discussed in further studies. Although the primary energy source of microorganisms – saccharides, amino acids, and fatty acids—was quantified in this study, the content of complex carbohydrates such as dietary fiber, resistant starch, or a variety of oligosaccharides also needs to be analyzed. Functional food components including isoflavones could also be the other candidates, as their effects on human microbiome changes were reported (Aboushanab et al., 2022; Narduzzi et al., 2022). Furthermore, since a single food group, pulses, was studied in the present research, other food groups with diverse matrices containing various proportions of macronutrients should be examined to demonstrate the multifaceted interactions between nutrient components and microbial diversity.