Microbial-derived metabolites as a risk factor of age-related cognitive decline and dementia

Age is the predominant risk factor for cognitive decline. Whilst some decline in cognition is considered an inevitable part of healthy ageing, deleterious changes in cognition, including mild cognitive impairment (MCI) and age-related dementias (e.g., Alzheimer’s disease, AD), are estimated to impact approximately 15% and 11% of the population over 65 years respectively [1, 2]. By 2050, the global elderly population is expected to increase by 21% [2], increasing incidences of cognitive decline [3]. Cognitive decline exacerbates broad social and economic issues, including depression, social withdrawal, difficulties performing everyday tasks, drastic reductions in quality of life and greater reliance on others (social care) [4]. Understanding how to promote healthy ageing whilst resisting aberrant changes in cognition is therefore becoming a priority.

Addressing modifiable risk factors can delay the onset, or even ameliorate cognitive decline [5], whilst assisting with the identification of asymptomatic individuals with an increased chance of developing the condition in the future [6]. Currently, hypertension [7], diabetes mellitus [8, 9], arteriosclerosis [10], obesity [11] and hypercholesterolemia [12] are the most significant risk factors associated with age-related cognitive decline among others [13]. Given the connection between cognition and these metabolic diseases, it is perhaps unsurprising that dietary factors can elicit a substantial influence upon cognitive function [14] through the modulation of a microbiota-gut-brain axis [15]. The microbiota-gut-brain axis is a complex communication system bridging the gut, liver and the central nervous system (CNS) that is modulated by the microbiome, a collection of 10¹⁴ microorganisms with an extensive functional gene repertoire [16]. These microorganisms predominantly reside in the gut, metabolising dietary compounds into a vast range of metabolites. Metabolites can cross the intestinal epithelial barrier; a structure connected by tight junction proteins, lamina propria and reinforced by mucosal secretions [17], primarily via active transport, and enter systemic circulation. From here, metabolites can directly initiate physiological responses by crossing the blood–brain barrier (BBB) and influencing the CNS [18], or indirectly via vagus nerve stimulation (Fig. 1) [19].

The capability of microbial-derived bioactive metabolites to influence the CNS provides a novel mechanistic pathway for cognitive decline, warranting its further exploration. Within the gastrointestinal (GI) tract, microbiota populations are in part reflective of their local physiological conditions. The small intestine, due to its proximity with the stomach, contains high concentrations of acids, oxygen and antimicrobials, thereby restricting bacterial growth to predominantly fast-growing anaerobes that can adhere to epithelia or mucus [20]. Conversely, colonic regions promote much denser bacterial communities, dominated by anaerobes such as Prevotellaceae and Lachnospiraceae, that can digest complex carbohydrates [21]. Numerous intrinsic factors (e.g., genetics, immune response, metabolites) and extrinsic factors (e.g., diet, lifestyle) also impact gut microbial composition, making it an attractive therapeutic intervention target [22]. The composition of these microbial communities determines the concentration of neurotransmitters or neuromodulators (including microbial-derived metabolites) released into circulation. Broad deviations in these microbial compositions, often referred to as “dysbiosis”, condition distinctly different metabolic profiles that may contribute to cognitive decline [23, 24]. Gut microbial composition is known to be significantly altered in patients with MCI, a transitional stage preceding AD, suggesting microbial changes may occur in the early stages of cognitive decline and influence its progression [25‐29].

Intestinal microbiota possess the capacity to produce hundreds of metabolites [30, 31], yet the influence of these compounds on cognitive health have not been uncovered. The present review details the roles of newly emerging microbial-derived metabolites that are less-explored in the current literature in the context of cognitive health and provide an additional in-depth discussion of their use as potential indicative factors of cognitive decline.

As we age, some of our cognitive abilities decline. Cognitive capabilities such as verbal skills, remain largely unaffected by brain ageing and can even improve over time [32]. Other essential capabilities, including mental reasoning, memory (in particular episodic, working and recognition memory) and processing speed, steadily deteriorate with age (See review [33] for further information). During ageing, the brain undergoes various structural and functional changes. The most apparent being a gradual shrinkage of the brain, alongside an increase in ventricular space and cerebrospinal fluid (CSF) [34, 35]. Brain atrophy increases in the elderly in an anterior–posterior gradient, with the most severe consequences taking place in the prefrontal regions [36, 37]. A reduction in white matter (the nerve fibres connecting different brain regions) integrity has been linked with normal cognitive ageing, impairing the transfer of information between cortical regions [38], an essential process for higher cognitive functioning [39].

Structural neuroimaging highlights differing trends in the neurobiology of pathological ageing and detrimental cognitive decline. Here, individuals are more likely to experience reductions in gray matter in the dorsolateral and medial prefrontal, parietal, and lateral temporal regions [40, 41], alongside a loss of white matter integrity in the cingulum, corpus callosum, and superior longitudinal fasciculus [42‐44]. This is instead of a decline in the frontal regions that typically occurs in healthy ageing.

AD is also associated with volume loss in the medial temporal lobe, a brain region highly associated with memory functions. Reduction typically starts in the anterolateral entorhinal cortex and advances medially across the remaining entorhinal cortex to the hippocampus [45, 46], with atrophy occurring at rates of 4.9–8.2%. In healthy ageing, atrophy in these regions occurs at a lower rate, diverging from pathological ageing, at 0.2–3.8% [46]. More recently, using longitudinal MRI and PET data, a similar divergence in volume loss has been noted in the locus coeruleus [47].

Finally, the default mode network (DMN), a resting-state network associated with cognitive processes of oneself (e.g., autobiographical memory), demonstrates connectivity patterns that distinguish healthy ageing from AD. Results from a task free-fMRI suggest AD patients have an accelerated ageing pattern of connectivity [48] and decreased resting-state activity in the posterior cingulate and hippocampus when compared with age-matched controls [49]. However, the biological mechanisms behind the heterogeneity of age-related cognitive decline are complex and not well understood.

The human gut microbiome represents a complex community of microbes that live in a mutualistic relationship with their host. Initially, these microorganisms were considered to be solely responsible for intestinal processes (fermentation of carbohydrates, synthesis of vitamins and xenobiotic metabolism) [50]. However, over the last 15 years, this notion has been revised, owing to increasing evidence of a bidirectional communication system between the CNS and the GI tract, more commonly referred to as the ‘gut-brain axis’.

The gut-brain axis encompasses the CNS, the autonomic and enteric nervous system, and peripheral nerves and is vital for maintaining homeostasis. Signals from the brain control the secretory and sensory function of the gut, whilst the brain and gut communicate via physiological channels including the neuroendocrine, autonomic nervous system, neuroimmune pathways and molecules synthesised from gut microbes [51]. Since the gut microbiota is integral to the modulation of this communication at different levels (from the gut lumen to the CNS) and chronologically as we age, many have broadened the term to ‘microbiota-gut-brain axis’ [52]. Indeed, the existence of the microbiota-gut-brain axis is supported by substantial preclinical and human evidence, highlighting its effect on different cognitive domains. Firstly, germ-free (GF) mice show that the brain is markedly affected by the absence of microbiota, exhibiting deficiencies in learning, memory recognition and emotional behaviours [53‐56]. These behavioural changes were accompanied by altered brain-derived neurotrophic factor (BDNF) expression in the hippocampus [54, 57, 58], a molecule inherently linked with synaptic plasticity and cognitive function [59‐61], and significant microbiota-associated changes in the quantity of dopamine and activation of serotonin synthesis pathways [62‐65], suggesting an important role of microbiota in memory, brain health and behaviour. Secondly, chronic antibiotic depletion of microbiota populations alters tryptophan metabolism and the expression of key cognitive signalling molecules in the brain such as BDNF, N-methyl-d-aspartate receptor subunit 2B (NR2B), serotonin transporter, neuropeptide Y system, oxytocin and vasopressin [66, 67]. These changes are associated with long-lasting effects on cognition and increases in anxiety-related behaviours [66, 68]. Finally, administering specific prebiotics/probiotics modulates behaviour in both rodents and humans, including changes in depression, anxiety and stress [57, 69‐73], alongside changes in immune markers, hippocampal synaptic efficacy and tryptophan metabolism [74, 75].

As we age, microbiota composition and function changes [76]. In humans, this has been associated with a decrease in species diversity, a reduction in Clostridiales and Bifidobacterium and a rise in Proteobacteria and pathobionts such as Enterobacteriaceae [76, 77]. However, abnormal alterations in intestinal microbiota composition, as seen in early cognitive decline and AD [25, 78], are associated with local and systemic inflammation, and dysregulation of the microbiota-gut-brain axis [79]. Advances in sequencing technologies have enabled us to investigate the association between cognitive decline and gut dysbiosis at the phylum level. These studies have highlighted differences in taxonomic levels of Bacteroides, Firmicutes, Actinobacteria, Ruminococcus, Lachnospiraceae, and Selenomonadales between AD patients in comparison to controls [25, 78, 80‐82].Such dysregulation has been associated with an increase in inflammatory markers, cytokines and the permeability of the gut epithelial barrier (‘leaky gut’), resulting in excessive leakage of bioactivate molecules, such as short-chain fatty acids (SCFAs), kynurenines, melatonin, histamine, bile acids, and neurotransmitters, into the blood. The resulting increase in neuroactive products can no longer efficiently be removed by the body’s next barrier; the liver, and therefore can cause a variety of physiological changes directly and indirectly affecting the CNS, including further decrease in BBB function. In the elderly population, this dysregulation becomes particularly relevant, as the BBB becomes more permeable with age [83]. A more permeable BBB allows an increased influx of harmful blood components, including microbial metabolites, into the brain; a feature seen in AD patients (reviewed by [84]). This process promotes neuroinflammation and macrophage dysfunction, leading to neural injury and ultimately a reduction in cognitive function [85, 86].

As previously outlined, the gut can also influence the brain indirectly through the vagus nerve activation. The vagus nerve consists of 80% afferent and 20% efferent fibres [87]. Afferent fibres connect to all four layers of the digestive tract, but do not cross the epithelial layer and therefore are not in direct contact with gut microbiota. As a result, the microbiota activates these fibres indirectly via the release of metabolites or bacterial products. Enteroendocrine cells (ECCs) make up approximately 1% of intestinal epithelial cells and can detect signals from the microbiota through toll-like receptors (TLR), capable of identifying bacterial compound such as lipopolysaccharides (LPS) [88], or through receptors activated by microbiota-derived metabolites such as SCFAs [89]. ECCs can subsequently interact with vagal afferent fibres through the release of serotonin or gut hormones [90, 91]. This indirect signalling between the gut microbiota and the brain via the vagus nerve can modulate certain cognitive functions. For example, rodents fed with the probiotic L. rhamnosus for 28 days had a decrease in anxiety-related behaviour, whilst inducing region-dependent alterations in γ-aminobutyric acid (GABA) receptor [92]. Importantly, this result only occurred with an intact vagus nerve, as mice undergoing a vagotomy did not display these behavioural and neurochemical changes. Similarly, in a colitis model, the normalisation of anxiety-like behaviours by the probiotic Bifidobacterium longum NCC3001 was found to be vagally dependent [93]. However, the total effects of the microbiome are not solely dependent on the vagus nerve stimulation, as mice orally receiving a mixed antimicrobial treatment had altered exploratory behaviour and hippocampal BDNF, independently of vagal integrity [94]. Together, these data emphasise that while the vagus nerve provides a crucial bridge allowing communication between the gut, its microbiome and brain, there are also other essential routes of communication comprising the microbiota-gut-brain axis, indicating its complex connectivity.

The relationship between the gut, liver and brain has increasingly been highlighted in recent years due to a high prevalence of liver disease, which is commonly accompanied by clear and global cognitive impairment (hepatic encephalopathy) [95, 96]. The gut and liver are linked by the portal vein, biliary tract and systemic circulation, allowing microbial and host-derived metabolites to influence liver function. Conversely, the liver acts as a vital barrier, removing potentially harmful compounds from the blood using a range of hepatic immune cells, including Kupffer cells, hepatic stellate cells and natural killer cells [97], modulating the concentration of metabolites directly and indirectly influencing the CNS. The liver also controls unrestricted bacterial growth in the gut, maintaining gut eubiosis, through the transport of bile salts through the biliary tract into the intestinal lumen leading to the secretion of antimicrobial compounds [98, 99]. For example, bile acids can bind to the FXR receptor in enterocytes, initiating the production of antimicrobial peptides such as angiogenin 1 and RNAse family member 4, which can inhibit bacterial overgrowth in the gut and intestinal barrier dysfunction [100]. Gut dysbiosis causes an imbalance of microbial and host-derived products, reducing the epithelial barrier function and causing increased leakage in the system. Long-term, this process can initiate metabolic disorders in the liver, promoting liver damage (reviewed by [101]). As such, liver damage has been found to correlate with the severity of gut dysbiosis [102]. Since a diseased liver cannot effectively remove harmful products from the blood or inhibit the overgrowth of bacteria [103‐105], this process can accelerate microbiota-gut-brain axis dysregulation and ultimately cognitive decline. Thus, when considering the occurrence of cognitive decline associated with microbial-derived metabolites, the role and function of the liver cannot be ignored.

Trimethylamine N-oxide (TMAO) is a microbial-dependent metabolite generated by the breakdown of dietary fish, meat and fat [147, 155]. Trimethylamine (TMA), the precursor to TMAO, is produced from the metabolism of choline, L-carnitine and phosphatidylcholine by anaerobic microbes in the gut, predominantly located in the small intestine (Fig. 2) [144, 145]. TMA subsequently travels through the portal vein to the liver where it is oxidised by flavin-containing monooxygenase 1 and 3 (FMO1 and FMO3) to form TMAO [146]. Once formed, TMAO can enter the systemic circulation, hence TMAO plasma levels (typically 3 μmol/L in healthy individuals [148]) have been found to correlate with the gut microbial composition [149].

During digestion, nutrients and bioactives (proteins, amino acids, polysaccharides, fibres, fats, polyphenols, etc.) are catabolised into host-derived and bacterial metabolites that have the ability to interact with the host’s cells and the resident gut microbiome [256, 274]. This continuous process results in the production of a wide array of chemicals representing a wealth of chemical classes. Dietary proteins are broken down into potentially active peptides [256] that are further transformed into bacterial products such as neurotransmitter amino acids like glutamate, glycine, aspartate, serine and GABA or polyamines [250, 275, 276]. Aromatic amino acids (tryptophan, tyrosine, and phenylalanine) and polyphenols yield a myriad of compounds during catabolism leading to the formation of simpler structures containing at least one phenol ring (phenols) which can then be further transformed by the host (sulfation, glucuronidation) before re-entering circulation [256, 274]. Additionally, dietary choline and niacin are substrates for the synthesis of molecules essential for cellular function in the brain namely acetylcholine and nicotinamide adenine dinucleotide (NAD +) precursors [277‐279], some of which have recently been shown to be synthesised by the gut microbiota [250, 254, 275, 278, 280]. These recent developments provide further evidence of the microbiota’s role in the production of beneficial signalling molecules that contribute to the maintenance of homeostasis during the ageing process.

As described earlier in the review, the BBB selectively allows circulating solutes to enter the CNS. Polyamines, polyphenols and some of their products (3-(3’-hydroxyphenyl)propionate and 3-hydroxybenzoate) [281] have been shown to cross the barrier even though the transfer seems somewhat limited [282‐286]. Meanwhile, nicotinamide and niacin, both precursors for NAD + , a coenzyme essential for the maintenance of the CNS, have the capacity to freely cross the BBB [278, 287]. More research on the topic is needed as the knowledge regarding their transport across BBB is in its infancy and partly based on in vitro models [282, 286, 288]. The question of whether the potential activity of those molecules on brain functions is either direct or based on interactions with peripheral systems remains open [289].

The concept of microbial-derived metabolites influencing cognitive decline is gaining traction, with implications in the field of neuroscience, metabolomics and hepatology. However, due to the complexity of this relationship, the specific myriad of mechanisms responsible remain largely unknown, whilst defined roles of individual metabolites are only characterised for a select few (for a summary see Fig. 3). Therefore, amid this ambiguity, there remains a real need for additional research to highlight and validate key pathways, metabolites and mechanisms to further elucidate the influence of the microbiota-gut-brain axis on cognition [189].

There remain many challenges facing this growing field. Firstly, a lack of specificity limits our ability to distinguish between host vs microbiota-derived metabolite contribution as particularly if there is known co-metabolism, true microbial involvement may be masked or exaggerated. Secondly, as demonstrated by TMAO [362], some inconsistencies still exist among certain metabolites under context-specific vs dose-specific conditions. This may in part be due to heterogeneity between studies, with variations in study designs, methods of assessing cognitive performance and/or quantifying metabolites. As a result, further research ought to be collated via a more standardised methodology to increase comparability. Thirdly, the influence of the microbiome on cognition is not the totality of microbial metabolites produced in the gut as the varying capabilities of these metabolites to penetrate the BBB play a key role [79]. Consequently, the mechanisms used by many metabolites to cross the BBB are still unknown and some may even be synthesised de novo. Fourthly, from a translational perspective, the described research has largely been conducted in animals. Establishing whether these findings translate to humans will be crucial yet challenging due to the greater complexity and environmental exposure humans encounter, in turn shaping each individual’s microbiome [369]. Finally, understanding these highly complex systems, particularly as we move more towards human studies, requires the continued advancement of computational and statistical methods to obtain and implement multi-omics and longitudinal data necessary for a comprehensive approach [330]. Together, these challenges render it difficult to outline specific host-microbiota interactions in a mechanistic manner, which is needed to advance the field past associations towards implementable microbiota-driven targets.

Nevertheless, the wealth of association studies highlight a positive future for the use of microbiota-derived metabolites as risk factors of cognitive decline [370]. Future studies should progress using robust and replicable metabolic phenotyping across various stages of cognitive decline in humans. Recent advancements using this approach are underway, utilising metabolic phenotyping of urine [371] and blood [153, 372] to predict incipient AD with high degrees of accuracy. However, several studies using comparable approaches have not been able to replicate these findings [373, 374]. This may be due to intrinsic difficulties surrounding the heterogeneity of cognitive decline seen in neurodegenerative diseases and the variety of analytical methods used in metabolic profiling, ranging from ¹H-NMR, LC–MS/MS, GC–MS, UHPLC-MS and CE-MS [371]. Hence, currently, the literature is too scarce to support the implementation of metabolite-derived risk factors in clinical practice.

In conclusion, although significant work remains to fully understand the role of microbial-derived metabolites as key mediators of cognitive decline, identifying modifiable factors that promote healthy ageing and cognition will have vital clinical implications in today’s growing elderly population, whilst also helping to identify novel underlying mechanism.