Introduction

With the exception of its role in food sensitivity and food allergy, dietary protein is seldom viewed as having anything other than positive effects on health. Consequently, dietary guidelines encouraging populations to increase protein intake by up to 100%, in a bid to combat the effects of sarcopenia [1], have been made with little consideration given to safety in terms of potential adverse effects on gut, immune and metabolic health. Until recently, digestion and absorption of dietary protein was considered to be highly efficient. However, recent evidence suggests that up to 10% of consumed protein reaches the colon undigested in humans [2]. The colon is host to the greatest density of commensal bacteria in mammals [3], and is where the highest level of bacterial fermentation of undigested food products occurs. Undigested colonic dietary protein is fermented by specific components of the resident microbiota [4] and thus has the potential to drive the expansion of proteolytic populations at the expense of more beneficial groups such as bifidobacteria and lactobacilli. Work in this area is limited, but intervention studies using varying amounts of increased dietary protein and time have reported changes in colonic bacterial metabolism [5,6,7], with increased dietary protein intake being linked to adverse shifts in bacterial populations [8]. Such changes can result in increases in the production of microbial-derived negative end-point metabolites, including ammonia and phenolic compounds, which have been shown to reduce gut barrier function in vitro [9,10,11,12]. The gut wall provides a physical, selective barrier preventing potentially harmful products of gut bacteria and food digestion from entering blood circulation [13]. Reductions in this barrier function through inflammation, disruption, and impaired integrity of the gut lining, a condition often referred to as ‘leaky gut’, can result in increased passage of these products, such as lipopolysaccharide (LPS), into blood [13, 14]. This can cause chronic low-grade systemic inflammation and thus promote the development of chronic degenerative diseases of the liver and cardiovascular system [14, 15]. While leaky gut may have multifactorial causes, fermentation of excess dietary protein by colonic microbes could be an important contributor to the development of these chronic diseases.

Healthy adult females have been shown to express lower and more variable permeability in the gut barrier than males. Additionally, adult male gut barrier function appears more stable and is less sensitive to impairment caused by non-steroidal anti-inflammatories (NSAIDs) than females [16]. However, female gut barrier function is more resilient to shock states such as acidosis or hypoxia than that of males [17], but becomes less stable with age [18]. The cause of these differences is multifactorial and is thought to involve sex hormones, immunity and the gut microbiota [19,20,21]. For example, female microbiotas are significantly more diverse than those in males, [22,23,24] and tend to have a lower abundance of Bacteroidetes, but higher proportions of Bacillota (Firmicutes) and Actinobacteria than male microbiotas [25, 26]. A greater diversity and high proportions of Bacillota can be indicative of a healthy microbiota [27], which might contribute to the greater efficiency observed in female barrier functionality. Microbiota sex differences also extend to lower taxonomic levels, with populations of Prevotellaceae, Ruminococcaceae [28], Clostridiales [22], and Escherichia [29] being lower in females than males. Specific microbial genera and species have been demonstrated to correlate with components of gastrointestinal immunity, including Clostridium-driven increases in intestinal mucosal regulatory T-cell (Treg) populations [30]. Higher levels of Clostridium populations in the male gut and the association between these microbes and increased levels of Tregs, may contribute to the disproportionally lower incidence of inflammatory gut conditions observed in males compared to females.

Complex interactions occur between gut barrier integrity, immune function and the gut microbiota. Since there is sexual dimorphism in all three of these systems in healthy adults, it is likely that the detrimental effects of excess dietary proteins on microbial-derived metabolites associated with reduced barrier function could also be sexually dimorphic. To the best of our knowledge, this has not yet been explored and possible mechanisms remain undefined. However, significant sex-dependent responses to dietary interventions have previously been identified, including inulin, starch and the probiotic Bifidobacteria lactis in immune-associated protein expression in the gut of 28-day-old piglets [31]. This supports potential links between nutrition and sex-based differences in physiological responses.

Consumption of proteins originating from different sources has previously been shown to have differential effects on metabolic end-product production by bacterial populations in the gut [32, 54,55,56].

Increased protein availability was linked with significant shifts in the composition of the gut microbiota, with 7 out of the 11 functional groups quantified, differing significantly from the controls. Two of the most notable bacterial functional groups that were less prevalent under increased protein availability were Bifidobacterium and Lactobacillus spp genera. These groups include the most commonly used bacteria for probiotic supplementation, due to the large body of evidence that supports their positive effects on health [57,58,59], including reductions in intestinal permeability [60, 61]. In our trial, four phenolic metabolites (phenol, p-cresol, indole and ammonia), which have been shown to reduce barrier function in vitro [12], were significantly elevated under high protein conditions compared to the low protein conditions of the controls. These results are in accordance with a previously published study which also used in-vitro gut model systems [62]. However, most studies exploring the effects of high protein availability on microbiota composition and/or metabolic activity used hydrolysed protein in order to obtain high purity (∼ 95%), for example, casein hydrolysates [9, 62]. These ultra-concentrated sources of protein, limit levels of the non-protein substrates usually associated with these proteins when consumed as part of a normal human diet. These additional substrates may include non-digestible oligosaccharides, which are preferentially fermented by some gut bacteria before the proteins are utilised, and thus affect the composition of the gut microbiota. Therefore, using less pure proteins, as were used here (75–81%), is more reflective of normal high-protein diets and will generate results with higher translational potential than using ultra-pure protein substrates [9, 62]. Furthermore, hydrolysed proteins are absorbed absolutely in the upper intestinal tract [63]. Consequently, due to increased digestibility, it is highly unlikely that hydrolysed proteins reach the human colon where the majority of bacterial fermentation of dietary proteins occurs. This could, in part, explain the limited translation to in-vivo systems where increased dietary casein hydrolysate did not result in significant shifts in bacterial populations in human stools [9].

Although limited in number, the majority of published studies which explore the impact of increased dietary protein on microbiota composition and/or metabolic output do so by using protein derived from single sources [62, 9, 64,65,66]. However, here we demonstrate that bacterial fermentation of dietary proteins from different sources had differential impacts on colonic bacterial populations and metabolic end-products in vitro. The basis of this probably arises from disparities in the proteolytic capacity of different bacterial groups. For example, bacteria which secrete different proteases and peptidases, including specific species of Bacteroides, Clostridium and Fusobacterium [4, 67], have growth advantages over other bacterial groups in relation to availability of proteins from different sources. This is a result of varying capacities to cleave exogenous proteins into constituent amino acids, which are utilised more efficiently than larger peptides [68,69,70,71,72,73]. Subsequently, proteins from different sources are preferentially utilised by different proteolytic groups and thus differential population expansion would occur, as we observed. Linked to this, each of the tested proteins had unique amino acid compositions which may also provide growth advantages to bacterial groups that preferentially utilise specific amino acids [74, 75]. For example, when screened for amino acid content, Staphylococcus aureus had significantly higher levels of alanine than other Gram-positive bacteria. These bacteria can eithersynthesise alanine de novo, or have increased capacity to obtain alanine from extracellular sources [74]. Since all the proteins used in our trial were non-hydrolyzed, bacteria with lower proteolytic and higher saccharolytic capacity, such as Roseburia, were likely to have preferentially utilised any residual carbohydrates present in the protein additives [76]. This could explain why total numbers of Roseburia were significantly higher than controls in the non-animal proteins which are likely to contain additional fermentable carbohydrate compared to the animal proteins [4]. Since increased dietary protein is rarely a result of the consumption of an ultra-pure hydrolysed protein from a single source, our results better reflect the effects of high-protein diets on the composition and metabolic output of healthy human gut microbiotas.

Variation in amino acid compositions between the proteins assessed is highly likely to have contributed to the observed differential production of nitrogenous metabolites. The deamination of all three aromatic amino acids (phenylalanine, tyrosine and tryptophan) results in the production of phenol [77]. Similarly high phenylalanine and tryptophan content of milk, whey, soya and mycoprotein [78, 79] resulted in microbial fermentation of these proteins, producing significantly more phenol than fermentation of inulin. Furthermore, decarboxylation of tyrosine results in p-cresol production; since mycoprotein typically has low levels of tyrosine, it might be expected that it would produce less p-cresol than milk, fish and whey proteins which have relatively high levels of tyrosine [79]. However, we found that mycoprotein fermentation produced the highest concentration of p-cresol. While the reason for this is currently unclear, it could be due to the different changes in microbial composition between the proteins leading to elevated levels of p-cresol producers under mycoprotein conditions. Both phenol and p-cresol have been shown to increase permeability of colonocyte monolayers in vitro [21]. Taken together, our results demonstrate that the source of dietary protein should be taken into account when exploring the impact of high protein diets on microbiota composition and subsequent metabolic activity in relation to gut barrier functionality.

Our results are consistent with the growing body of evidence supporting sexual dimorphism in gut microbiotas [26, 80,81,82,83,84,85]. As previously noted, female gut microbiotas are often more diverse and have higher overall cell counts than male microbiotas [81, 85, 86]. However, the bacterial species reported to be different between the sexes is inconsistent. For example, males have been reported to have a greater abundance of the Bacteroides-Prevotella genera18,25,26, and certain species within the Bacteroides genus, than females [87], whilst, females have been found to have significantly more lactic acid producing bacteria than males [24]. Although we did observe higher overall bacterial cell counts in females, our study found no differences in the Bacteroidetes-Prevotellaceae (BAC) or Lactobacillus-Enterococcus (LAB) groups between the sexes at baseline. These discrepancies suggest there are as yet unknown caveats perhaps age could play a key role. Alternatively, our FISHflow methodology fully quantifies functional groups while deeper sequencing techniques may well have identified more subtle differences between the sexes at baseline. Additionally, reports of sex-based differences in gut microbiotas are often based on human trials where many additional influential factors are present. For example, bile acids [88] and components of immunity [89, 90], which have considerable impacts on gut microbiotas and are sexually dimorphic, but were absent from our in vitro models [91,92,93]. Regardless of this, we did observe significantly lower levels of Clostridium cluster IX in females than males, which is consistent with previous reports [81, 85, 86]. Interestingly, we report protein × sex interactions regarding changes in microbiota composition in response to increased fermentation of proteins from different sources. Here, Clostridium cluster IX (PRO) and Lactobacillus spp. (LAB) were present in greater numbers in males following fermentation of mycoprotein and whey protein than in females. Diet-dependent sex differences have previously been reported in animals and humans [82, 84, 94,95,96], with human males, in general, being more susceptible to dietary-driven changes in microbiota composition than females [82]. However, the underlying mechanisms have yet to be elucidated. During fermentation, differences in microbiotas may be exacerbated by dietary substrate availability, which could impact on bacterial metabolic activity. Although there were only limited sex-dependent microbiota compositional differences in response to fermentation of proteins from different sources, we did observe significant differences between the sexes in microbial metabolite production in response to differential dietary proteins. This is consistent with the metabolic potential of gut microbiotas varying between the sexes. This is of particular interest with regard to the microbial production of phenol in response to dietary fish protein, which was significantly higher in females compared to males. This finding is consistent with females being more susceptible to perturbations in gut barrier function in response to specific stimuli[17]. However, we also observed increases in propionate production by the gut microbiota in females in response to fish protein fermentation. Previously, propionate has been demonstrated to ameliorate dextran sodium sulphate-induced colitis in murine models by increasing expression of tight cell junction (TCJ)-associated proteins (e.g. Zona occludin, E-cadherin) [97]. Propionate has also been shown to promote expression of TCJ-associated proteins (claudins and occludins) in healthy pig jejunual mucosa [98].

In conclusion, our results describe strong correlations between increased dietary protein availability, shifts in microbial populations towards more proteolytic phenotypes and subsequent increases in bacterial metabolic end-products linked with increased intestinal permeability In vitro. Our results are consistent with our hypotheses and we demonstrate, for the first time, considerable sex-dependent influences on these interactions. Importantly, we show that both the source of protein is highly influential for microbiota composition and microbial metabolic outcomes. Perhaps caution should be exercised regarding blanket recommendations to increase protein consumption in the wake of the results reported here.