The role of serotonin within the microbiota-gut-brain axis in the development of Alzheimer’s disease: A narrative review

Alzheimer's disease (AD) is the most common cause of dementia, accounting for more than 50 million patients worldwide. Current evidence suggests the exact mechanism behind this devastating disease to be of multifactorial origin, which seriously complicates the quest for an effective disease-modifying therapy, as well as impedes the search for strategic preventative measures. Of interest, preclinical studies point to serotonergic alterations, either induced via selective serotonin reuptake inhibitors or serotonin receptor (ant)agonists, in mitigating AD brain neuropathology next to its clinical symptoms, the latter being supported by a handful of human intervention trials. Additionally, a substantial amount of preclinical trials highlight the potential of diet, fecal microbiota transplantations, as well as pre- and probiotics in modulating the brain's serotonergic neurotransmitter system, starting from the gut. Whether such interventions could truly prevent, reverse or slow down AD progression likewise, should be initially tested in preclinical studies with AD mouse models, including sufficient analytical measurements both in gut and brain. Thereafter, its potential therapeutic effect could be confirmed in rigorously randomized controlled trials in humans, preferentially across the Alzheimer's continuum, but especially from the prodromal up to the mild stages, where both high adherence to such therapies, as well as sufficient room for noticeable enhancement are feasible still. In the end, such studies might aid in the development of a comprehensive approach to tackle this complex multifactorial disease, since serotonin and its derivatives across the microbiota-gut-brain axis might serve as possible biomarkers of disease progression, next to forming a valuable target in AD drug development. In this narrative review, the available evidence concerning the orchestrating role of serotonin within the microbiota-gut-brain axis in the development of AD is summarized and discussed, and general considerations for future studies are highlighted.


About Alzheimer's disease
Dementia roughly affects 50 million people worldwide, and numbers are expected close to double every 20 years (Prince et al., 2015). Dementia is a broader term for the decline in cognitive function, including memory, learning and thinking, in a more drastic manner than is expected from normal aging (WHO, 2020). This can be caused by a range of conditions, yet, the most common one is Alzheimer's disease (AD) which accounts for 60-70% of all cases as stated by the WHO. Although AD is considered a disease of the elderly, Zhu et al. (2015) estimated that early-onset AD (< 65 years) accounts for 6% of cases. Regardless of the age of onset, the course of the disease extends over a period of about 15-25 years as a continuum (Scheltens et al., 2021). At onset of the pathology, the patient may be asymptomatic or experience mild cognitive impairment (MCI). Over time, however, symptoms gradually become worse in function of the progressive neuronal loss (Duyckaerts et al., 2009;Förstl and Kurz, 1999). During all disease stages, a change in mood and behavior is often experienced (Lyketsos et al., 2002). These common neuropsychiatric symptoms include anxiety, depression, irritability, reduced appetite, stereotyped behavior, psychosis and aggression (Craig et al., 2005). Given the cognitive and behavioral alterations, the dementia syndrome forms a burden both on the individual suffering from the disease, as well as on family, caregivers, friends in addition to the entire society. As an indication, the global socioeconomic costs for dementia were calculated to be about 670 billion euros in 2015 (Prince et al., 2015).
Although many gene polymorphisms have been linked to AD, genetics give a far from complete explanation, with an exception for the rare familial (often early-onset) forms of AD (Gatz et al., 2006;Vogrinc et al., 2021). Nevertheless, related genes may give an indication of the possible pathophysiological mechanisms, such as with the apolipoprotein-E (APOE) determined allelic risk variation (Scheltens et al., 2021). The general picture of AD consists of the progressive topographic decline in cholinergic, catecholaminergic (dopamine, (nor) adrenalin) and indoleaminergic (serotonergic) neuronal functioning and loss (for review : Š imić et al., 2017), preceded by neurotoxic amyloid-beta (Aβ) plaque aggregation extraneuronally, and, intraneuronally deposited neurofibrillary tangles (NFT) of phosphorylated tau (P-tau), both being histological hallmarks of AD (Braak and Braak, 1991). Other factors are suspected to be equally involved, such as a blood-brain barrier disintegrity, oxidative stress and mitochondrial dysfunction (Vidal and Zhang, 2021). Another etiological factor is the glycosylation of lipids and proteins, giving rise to advanced glycation end products (Haukedal and Freude, 2021). Furthermore, a substantial amount of evidence suggests that neuroinflammation plays a contributing role in AD development by accelerating the abovementioned processes (Kinney et al., 2018). Especially microglia seem to be involved (Hansen et al., 2017). Finally, the microbiota-gut-brain axis may be involved in the development of AD as well (Bonfili et al., 2021;Doifode et al., 2021;Generoso et al., 2020;Kesika et al., 2021). All in all, these suspected disease modulators are current targets in the ongoing search for an effective cure . At the same time, Livingston et al. (2020) identified 12 potentially modifiable risk factors across the lifespan, accounting for around 40% of worldwide dementias, aiding the development of public health prevention strategies. Of these risk factors, lifestyle in general plays a prominent role.

Serotonergic neurotransmitter system alterations in Alzheimer's disease: gut involvement
Prominent changes in AD brain expand far beyond Aβ and tau, with a disturbed serotonergic neurotransmitter system as one of the most prominent neurochemical alterations, which is involved in but not restricted to emotional and cognitive dysfunction (Ciranna, 2006). Firstly, a decrease in total brain serotonin content, particularly in the temporal and frontal cortex, has earlier been identified (Aral et al., 1984;Palmer et al., 1987), next to alterations of cerebrospinal fluid (CSF) serotonin levels (Tohgi et al., 1992). Secondly, Cross et al. (1984) found a substantial loss of serotonin (5-hydroxytryptamine, 5-HT) type 1 and 2 receptors in the amygdala, neocortex and hippocampus in post-mortem brains of Alzheimer's patients, and, more recently, Solas et al. (2021) examined involvement of 5-HT7 receptors in psychotic symptoms in AD. A correlation has also been observed between aggressive as well as depressive symptoms and serotonin levels (and its metabolite Fig. 1. Serotonin and kynurenine biosynthetic and metabolic pathways starting from tryptophan. The essential amino acid tryptophan forms the basis for the synthesis of serotonin (5-HT). Its deductive metabolic pathway consists of the kynurenine pathway. Serotonergic brain circuitry all start from the raphe nuclei, a collection of serotonin-producing neurons, located in the brainstem at the height of the pons, and have efferents connecting with the entire neocortex, limbic system (of which amygdala and hippocampus), diencephalon, cerebellum and peripheral/autonomous nervous system (e.g. spinal cord, vagus nerve). The mechanism of action of an SSRI is to block SERT, thus preventing the reuptake of serotonin after its release from the synaptic cleft back into the presynaptic neuron. The kynurenine pathway elicits the formation of both neurotoxic and neuroprotective metabolites. Kynurenic acid is mainly formed in astrocytes (purplish) and is an effective NMDA receptor antagonist, preventing abundant intracellular release of calcium, and, consequently, excitoxicity. Contrariwise, 3-hydroxykynurenine is known as a potent oxidative stressor and free radical donor, leading to mitochondrial damage and the creation of reactive oxygen species. A similar neurotoxic function has been ascribed to quinolinic acid, an NMDA receptor agonist, with an opposite function compared to kynurenic acid. Quinolinic acid is mainly formed in microglia (pinkish). Abbreviations: 3-HAO: 3-hydroxyanthranilate 3,4-dioxygenase; 5-HIAA: 5-hydroxyindoleacetic acid; 5-HT: 5-hydroxytryptamine (serotonin); AADC: aromatic L-amino acid decarboxylase; ACMSD: 2-amino-3-carboxymuconate semialdehyde decarboxylase; ASMT: acetylserotonin O-methyltransferase; Ca 2+ : calcium; IDO: indoleamine 2,3-dioxygenase; KAT: kynurenine aminotransferase; KMO: kynurenine 3-monooxygenase; KYNU: kynureninase; MAO: monoamine oxidase; NAD + : nicotinamide adenine dinucleotide; NAT: N-acetyltransferase; NMDA: N-methyl-D-aspartate; SERT: serotonin transporter; SSRI: selective serotonin reuptake inhibitor; TDO: tryptophan 2,3-dioxygenase; TPH: tryptophan hydroxylase; QPRT: quinolinate phosphoribosyltransferase. Created with BioRender.com. 5-hydroxyindoleacetic acid (5-HIAA)) in specific brain areas, among which the hippocampus (Vermeiren et al., 2014). Multiple studies also revealed that selective serotonin reuptake inhibitors (SSRI), which act on the serotonin transporter (SERT) (Fig. 1), relief both behavioral and cognitive phenomena in AD patients, among which aggression and anxiety . Additionally, CSF Aβ-concentrations were shown to be associated with SSRI treatment (Cirrito et al., 2011;Sheline et al., 2014) and the long-term use of antidepressants, such as SSRI, seems to lower the elevated risk on developing dementia in depressed individuals (Kessing et al., 2009). These findings hypothesize (in)direct involvement of serotonergic system alterations and AD development, making it a valuable target both in terms of prevention and (symptomatic) treatment.
On the whole, the total serotonin content in brain is far less than that in gut tissue (Erspamer, 1966;Vermeiren et al., 2016Vermeiren et al., , 2015, and, even more less compared to concentrations in the intestinal lumen. Fecal concentrations give an indication of the latter (Hirabayashi et al., 2020). Total estimates range from 5% to 10% of its production solely in the brain, compared to 90-95% in the gut. The essential aromatic amino acid tryptophan (Bender, 1983;Udenfriend et al., 1956) is the main precursor of serotonin synthesis. After dietary or supplemental ingestion, the amino acid can be converted through a chain of reactions into several products of which serotonin, or, more specifically, 5-hydroxytryptamine (5-HT), is one example. An intermediate in the formation of the neurotransmitter is 5-hydroxytryptophan (5-HTP) (Udenfriend et al., 1956). Following its synthesis, serotonin can in turn be converted into other metabolic products, such as 5-HIAA via the action of monoamine oxidase (MAO) (Fig. 1). Nevertheless, tryptophan can also be metabolized via the kynurenine pathway, which requires the enzyme indoleamine-2,3-dioxygenase (IDO) (for review: Wichers and Maes, 2004). An essential enzyme required for the synthesis of serotonin itself is tryptophan hydroxylase (TPH), which plays a role in the rate-limiting step (Bender, 1983). Both neurons and enterochromaffin cells (ECC) of the gut comprise this enzyme, although slightly different variants exist (Côté et al., 2003;Walther et al., 2003). TPH1 and TPH2 are the most abundant in gut and brain, respectively. Beside EEC, several microorganisms in the gut are also able to produce hormones and neurotransmitters, including serotonin (for review: Clarke et al., 2014). Escherichia coli k12 and Lactobacillus plantarum, for instance, are examples of bacteria that possess this ability, at least in vitro.

Its potential importance in Alzheimer's disease
In short, the brain, gut and microbiota all produce serotonin. However, serotonin itself, unlike its intermediates, is hardly able to cross the blood-brain barrier, as evidenced in rats by the use of radiolabelling techniques (Oldendorf, 1971). This points out the existence of distinct pools of serotonin, which, on the contrary, may be able to interact with one another (Clarke et al., 2013). This notion is supported by the fact that gut and brain are bidirectionally connected via metabolic, hormonal and neural routes as reviewed by Wang and Wang (2016). Short-chain fatty acids (SCFA), metabolites produced by gut microbiota following (mainly) dietary fiber intake, are considered important mediators in this communication with an effect on cognitive function (Dalile et al., 2019). This might be through the impact on gene expression, since SCFA stimulate the transcription of TPH1 (Reigstad et al., 2015). As a logical consequence, interfering with the microbiota (composition) in the gut, either by means of nutrition, fecal microbiota transplantations (FMT) or a combination of pre-and probiotics, has become an emerging potential modulator of brain health, and is likely to affect both distantly related serotonin pools (Liu et al., 2015a).
As an indication of the importance of the indoleamine neurotransmitter serotonin within the gut-brain axis, enrichment of diet with tryptophan has previously been evidenced to enhance learning and memory abilities in aged rats (Musumeci et al., 2017) while decreasing hippocampal apoptosis and intraneuronal Aβ load in transgenic AD mice . Musumeci et al. (2015) claim these effects to be the consequence of changes in serotonin and brain-derived neurotrophic factor expression in both frontal cortex and hippocampus. Additionally, FMT is able to modulate Aβ content in the hippocampus as shown in a senescence accelerated mouse model (Cui et al., 2018).

Research question
In general, AD is a complex multifactorial disease of which the mechanisms remain incompletely understood. There is mostly preclinical evidence that serotonin may play a role in AD-related cognitive decline and neuropathological aspects, and that this might be indirectly modulated through the microbiota-gut-brain axis, both in terms of development and onset. In this narrative review, multiple relevant studies will be discussed aiming to answer the question 'what is the role of serotonin within the microbiota-gut-brain axis in the development of AD?'. Since there are no studies to date yet that have tackled this issue as a whole, the research question will be subdivided into two subquestions. First, 'are the alterations in the brain's serotonergic system implicated in the development of AD?', followed by 'is it possible to alter the brain's serotonergic system through modulation of the microbiota-gut-brain axis?'.

Methods
In order to gather literature for this narrative review, two databases were searched. The search was performed in PubMed and Scopus using a set of queries. For each subquestion, specific queries were used. These were: SSRI AND alzheimer * AND (plaque * OR (amyloid AND beta) OR tau OR tangle * OR learning OR memory OR cognit * OR atrophy OR neurodegeneration), ("serotonin receptor" AND (agonist OR antagonist)) AND alzheimer * AND (plaque * OR (amyloid AND beta) OR tau OR tangle * OR learning OR memory OR cognit * OR atrophy OR neurodegeneration), (probiotic * OR prebiotic*) AND (serotonin OR serotonergic) AND brain, (nutrition OR diet*) AND (microbiota OR microbiome) AND (serotonin OR serotonergic) AND brain and ((fecal OR fecal) AND transplant*) AND (serotonin OR serotonergic) AND brain. Occasionally, the techniques forward and backward snowballing were used. Duplicate findings were excluded. The remaining acquired articles were screened by looking at the title and abstract, after which relevant articles were read more thoroughly. Both human and animal in vivo designs, as long as it were intervention studies, were considered eligible for the purpose of this review. In general, review articles were excluded. Exceptions were made for reviews that summarized trials otherwise excluded in this review. For the first subquestion, preclinical randomized controlled trials that included an AD mouse or rat model as well as human clinical trials from the last two decades were found eligible, at least, if they specifically manipulated the brain's serotonergic system. Trials with subjects that had pre-existing mental disorders (such as depression) were excluded. This was also the case for trials that focused on one specific behavioral symptom (such as agitation or depression), instead of a variety of behavioral symptoms, cognition and/ or underlying pathology. For the second subquestion, studies involving either healthy subjects or a(n) (induced) disease state related to AD pathology or symptoms that looked at serotonin (related) enzymes, receptors, transporters or concentrations in the brain were included, provided that the studies intervened through prebiotics, probiotics, FMT or nutrition. Furthermore, articles written in another language than English were not considered. Eventually, 67 articles were considered relevant for inclusion in this review, as can be seen in the overview (Fig. 2).

Serotonergic alterations in Alzheimer's mouse models and patients
Although a body of evidence supports the existence of serotonergic changes in AD, it does not necessarily imply a causal relationship. Therefore, intervention studies that manipulate serotonin synthesis, metabolization or transport and meanwhile assess its effect on AD brain pathology or clinical symptomatology are required. One well-studied way to manipulate brain serotonin concentrations is the administration SSRI, known for its widespread use as antidepressants. Multiple studies report their effect on Aβ plaques in mouse models of AD, as summarized in Table 1.
Different types of SSRI, including fluoxetine (Chao et al., 2020;Huang et al., 2018;Jin et al., 2017;Ma et al., 2017;Qiao et al., 2016), escitalopram , citalopram (Reddy et al., 2021;Sheline et al., 2014;Zhang et al., 2018) and paroxetine Olesen et al., 2017), induce a decrease in Aβ levels and/or plaques in either the whole brain, cortex or hippocampus, of which the latter region is the most studied one. The effect might be region specific, since Olesen et al. (2016), Severino et al. (2018) and Von Linstow et al. (2017) failed to replicate the effect for the neocortex. Another neuropathological hallmark within the AD brain, P-tau depositions, has been studied by Ai et al. (2020) and Jin et al. (2017). However, no significant effect was found of paroxetine and fluoxetine in their used AD mouse models. On the other hand, the findings of Halliday et al. (2017) did reveal improvement in tau burden in Tau P301L positive mice after administering trazodone. A preventive effect on neuronal loss has been repeatedly observed as well in either the hippocampus (Halliday et al., 2017) or nearby regions, such as the dentate gyrus (Jin et al., 2017;Ma et al., 2017). Other AD-model induced abnormalities seem to be improved likewise, such as microgliosis  and mitochondrial deficits (Reddy et al., 2021). Importantly, most of the mentioned preclinical studies also reported protective effects on cognitive function, such as for learning and memory, next to the alterations concerning plaques, tangles and neuronal loss Chao et al., 2020;Halliday et al., 2017;Huang et al., 2018;Jin et al., 2017;Ma et al., 2017;Olesen et al., 2016;Qiao et al., 2016;Reddy et al., 2021;Zhang et al., 2018). The positive effect on cognition has been confirmed by the randomized controlled trial of Torrisi et al. (2019). However, this is contradicted by the findings of Olesen et al. (2017) and Severino et al. (2018). The latter also found a decreased survival rate in APP/PS1 mice, which in part questions its utility and safety on the long term, even though chronic SSRI administration via repeated intraperitoneal injections in this particular study may have created a stressful living condition for these mice on the whole. In line with this assumption, one emerged theory suggests that the effects of SSRI on the depression-like phenotype are not determined by the drug per se, but may be induced by the drug, and, driven by the environment. Especially, mice that were administered fluoxetine in an enriched condition overall improved their depression-like phenotype compared to their control littermates, whereas those treated in a stressful living environment showed a distinct worsening (Alboni et al., 2017).
Furthermore, human intervention studies have been executed. This is exemplified by a placebo-controlled trial in cognitively healthy older adults (n = 114) with escitalopram by Sheline et al. (2020). Dosages ranged from 20 to 30 mg and duration from two to eight weeks. A 9.4% larger reduction in CSF Aβ1-42 levels was found in the overall treated groups compared to the non-treated group. Additionally, the small single-dose placebo-controlled crossover trial of Klaassens et al. (2019) showed a protective effect of citalopram (30 mg) on (characteristic) connectivity loss in the precuneus and posterior cingulate cortex, while it failed to show effects on cognition. Cognitive functioning was measured using the NeuroCart test battery, in both mild AD patients (n = 12) and controls (n = 12). Contradictory, a meta-analysis of 14 randomized placebo-controlled trials suggests a beneficial effect of SSRI on cognitive performance in AD patients (Xie et al., 2019). On the whole, the majority of evidence both from animal and human intervention studies support the notion that SSRI are able to alter Alzheimer's neuropathology and symptoms.
Besides SSRI, a variety of other compounds are able to modulate the serotonergic system, such as serotonin (5-HT) receptor antagonists and agonists. Firstly, it has been shown that 5-HT6 receptor antagonists have positive effects on cognition in preclinical trials Shahidi et al., 2019), however, clinical trials with actual AD patients fail to prove significant effects (for review: Andrews et al., 2018;Khoury et al., 2018). Not only the 5-HT6 receptor, but also the 5-HT1A receptor has gained interest. The partial 5-HT1A receptor agonist tandospirone has been shown to improve anxiety, depression, agitation, irritability and delusion in AD and vascular dementia patients, as assessed with the Neuropsychiatric Inventory and Mini-Mental State Examination score (Sato et al., 2007). Similarly, NAD-299, a high affinity 5-HT1A receptor antagonist, has been shown to produce numerous effects in a streptozotocin-induced AD rat model, both on Aβ plaque load in the cortex and hippocampus (Afshar et al., 2018), as well as on hippocampal oxidative stress, damage and neuronal connections (Afshar et al., 2019). Remarkably, similar effects were observed by Afshar et al. (2019) after administration of the 5-HT2A receptor agonist TCB-2. On the contrary, pimavanserin, an inverse 5-HT2A receptor agonist, has been shown to reduce Aβ in the cortex, hippocampus and CSF, accompanied by improvement in cognitive function in APP/PS1 mice (Yuede et al., 2021). Cognition was improved as well in a trial with the 5-HT2A receptor antagonist desloratadine, using the same type of transgenic mice . Additional findings include improved microglial phagocytosis, microglial-plaque interaction and neuronal plasticity, accompanied by reduced neuroinflammation, and, decreased Aβ plaques in the CA1 region of the hippocampus. Furthermore, two types of 5-HT4 receptor agonists (RS 67333 and SSP-002392) showed promising effects on learning and memory, in combination with decreased Aβ plaques in several brain regions of transgenic AD mice (Giannoni et al., 2013;Tesseur et al., 2013). However, the effect on plaques seems dependent on treatment duration and onset, as Tesseur et al. (2013) and Giannoni et al. (2013) failed to replicate these effects in some of the intervention arms. The decrease in Aβ plaques in the hippocampus combined with

Fig. 2. Flowchart of the inclusion and exclusion process for the review.
Created with Lucidchart.com.

Table 1
Preclinical studies in AD mouse models investigating SSRI administration on Aβ plaque and tau tangle load and/or related cognitive and/or behavioral functioning. improved cognition was also reported after administration of a 5-HT7 receptor agonist named AS19 . This agonist has also shown to decrease hippocampal apoptosis and improve plasticity in an AD model of male Wistar rats (Hashemi-Firouzi et al., 2017;Shahidi et al., 2018). Finally, a clinical trial has been conducted with the 5-HT3 receptor antagonist ondansetron, which failed to show any effect on cognitive parameters (Dysken et al., 2002). Overall, these studies suggest that 5-HT1A/2A/4/6/7 receptor (ant)agonists exert varying effects related to AD pathology and clinical symptomatology (Table 2).

Brain serotonergic alterations in response to fecal microbiota transplantation
The role and manipulability of microbiota in brain serotonergic alterations in AD can be studied using FMT. Unfortunately, such studies are currently lacking in both AD mouse models, as well as patients. Nevertheless, Hata et al. (2019) conducted such a transfer from four anorexia nervosa patients, as well as four healthy age-matched individuals, to four-week old germ-free female mice (n = 72). A decrease in serotonin content of the brainstem was significantly observed afterwards, in addition to a trend of decreased serotonin and increased 5-HIAA content in other brain regions. Behavioral testing (i.e. open field and marble burying) indicated promising alterations. More specifically, mice receiving FMT from the anorexia nervosa patients showed more anxiety-like and compulsive behavior. Correspondingly, a study in which FMT was conducted from 11 schizophrenia patients to five-week old antibiotics-treated (pathogen-free) mice in comparison with FMT from ten control individuals, showed an increase in hippocampal and striatal serotonin, prefrontal cortex and striatal kynurenine and hippocampal TPH-1 expression . These findings were accompanied by an increase in learning and memory impairment as assessed with the elevated plus maze, reciprocal social interaction, forced swim test, open field test, Barnes maze, three chamber sociability test and novel object recognition test. Both studies thus evidenced that FMT has the ability to affect the serotonergic neurotransmission in addition to clinical functioning, at least in germ-free mice. Moreover, a human intervention study including Caucasians (n = 24) aged 50-70 with treatment-naïve metabolic-syndrome showed a positive trend in both hypothalamic and thalamic SERT binding after FMT from post-gastric bypass patients compared to oral butyrate supplementation (Hartstra et al., 2020). The serotonin transporter was visualized in both regions with region of interest analysis using single photon emission computed tomography (SPECT) after injection of 123 I-ioflupane as the radioligand. Additionally, significant differences in microbiota composition between the two groups were measured in the fecal microbiota analysis. In conclusion, these handful of studies indicate that FMT is able to exert serotonergic changes in the brain and may even have profound subsequent effects on both cognitive and behavioral aspects.

Brain serotonergic alterations as a result of dietary interventions
Less drastic, but, at the same time, more difficult to control for, is the dietary approach. Firstly, a randomized controlled preclinical trial focussed on the western diet, defined by its high fat content, compared to a standard diet as a possible modulator of the gut-brain axis (Ohland et al., 2016). Composition of the diets were 28% and 29% protein, 49% and 55% refined carbohydrates, and, 33% and 13% fat, respectively. The study had a small sample size of only three to four male mice (6 weeks old) per group. After the three-week intervention period, behavioral tests such as the elevated Barnes maze and latency to step down were performed. The diet group showed a decrease in anxiety-like and exploratory behavior. Also, neurotransmitter analyses of the brain revealed an enhancing effect on tryptophan levels in the hippocampus. Nevertheless, hippocampal serotonin levels and TPH2 expression remained unchanged. The larger study of Beilharz et al. (2018) also reported on the effects of the western diet compared to a standard diet, although for a total of about four weeks in male rats (n = 60). Importantly, the diet increased 5-HT1A while it decreased 5-HT2C receptor gene expression in the hippocampus. These effects were absent in the perirhinal cortex. Behavioral tests (elevated plus maze, object recognition task and place recognition task) revealed negative effects on spatial memory, but not anxiety. Additional findings were the decreased microbial diversity. Remarkably, the observed effects on spatial memory and microbial diversity could be prevented by a two-week treatment of the probiotic containing Bifidobacterium longum, infantis and breve, Lactobacillus acidophilus, paracasei, bulgaricus and plantarum, as well as  Streptococcus salivarius. In another randomized controlled trial male rats (n = 12) were also fed a high fat diet for four weeks, although in this case, until obesity. These findings even revealed a decrease in whole brain serotonin, accompanied by the overgrowth of Bacteroides as assessed from fecal samples (Labban et al., 2020). Secondly, Egerton et al. (2020) investigated the effect of a specific dietary component, namely fatty fish oil, combined with fluoxetine added to a standard diet for two weeks in maternally separated male rats (n = 58). Behavioral tests, such as the elevated plus maze, open field test and forced swim test, revealed improvement in depression and anxiety. In contrast, the subsequent biochemical analysis showed no significant difference in brainstem serotonin levels. Unfortunately, no other brain regions were investigated, complicating the interpretation of findings. However, both fatty fish oil and fluoxetine, separately or combined, did lower the level of serotonin's main metabolite, 5-HIAA, in the brainstem. The change of gut microbiota composition and SCFA production thus suggest a potential modulatory role in this effect. For instance, increased prevalence of Bacteroidetes and Prevotellaceae in combination with reduced levels of butyrate seemed characteristic for the fatty fish oil group.
Finally, the most compelling evidence so far, even though it was not a whole diet approach, comes from Musumeci et al. (2015Musumeci et al. ( , 2017 and Noristani et al. (2012), as previously mentioned. Noristani and colleagues examined reduced CA1 hippocampal intraneuronal Aβ in the triple transgenic AD mouse model following an acute one month increase of dietary tryptophan intake (0.40 g tryptophan/100 g), whilst Musumeci et al. (2015) provided direct evidence that an alike diet increased the serotonergic neurotransmission, particularly in the hippocampus of aged rats. In the same way, tryptophan-deprived (non-AD) mice recently showed significant reduction in 5-HT and 5-HIAA levels in a brain region-specific manner, namely in hippocampus, brainstem, cortex and striatum (Zhang et al., 2022).

Brain serotonergic alterations induced by pre-and probiotics
A variety of studies highlight the effect of pre-and/or probiotics on the brain's serotonergic system, often in combination with behavioral and cognitive changes. Details of all the included studies can be found in Table 3.
Firstly, nine randomized controlled preclinical trials reported the effects of a probiotic containing Lactobacillus plantarum. An anxiolytic effect of the probiotic was confirmed by several studies (Davis et al., 2016;Liu et al., 2016Liu et al., , 2015bMorshedi et al., 2018;Zaydi et al., 2020). The same holds true for improvement in learning (Morshedi et al., 2020) and memory (Zaydi et al., 2020). Besides cognitive alterations, Zaydi et al. (2020) showed the enhancing effect of the probiotic on serotonin-related enzymes, in this case TPH1, in a D-galactose-induced rat model of aging. Additionally, the probiotic increased serotonin in the whole brain (Mustafa et al., 2020), with the hippocampus in specific (Morshedi et al., 2020), next to the amygdala (Morshedi et al., 2018). Also, an increase in the expression of the serotonin transporter (5-HTT or SLC6A4) in healthy rats has been reported (Reza et al., 2019). This is confirmed in stressed Zebrafishes, specifically for the serotonin transporter subtype SLC6A4a (Davis et al., 2016). Findings seem contradicting in the case of its metabolite 5-HIAA. Liu et al. (2016) showed an increase in the striatum, but not prefrontal cortex or hippocampus of male germ-free mice, while Liu et al. (2015b) showed an overall decrease in male mice with early life stress.
Other Lactobacillus strains show similar effects. Borrelli et al. (2016) and Xie et al. (2020) reported an increased DNA expression of serotonin-related enzymes in the brain, as well as the serotonin transporter, in zebrafish and male adult mice, respectively. Notably, Xie et al. (2020) found only effects of the probiotic when stress was induced. Furthermore, related preclinical trials observed enhancement of brain serotonin levels, either in a specific region such as the hippocampus, or, the whole brain (Chen et  Four randomized placebo-controlled preclinical trials with Bifidobacterium are contradicting. For instance, Tian et al. (2020) found an increase in hippocampus, but not prefrontal cortex, of serotonin levels in chronically-stressed adult male C57BL/6 mice fed with the infantis strain. With the same strain, Desbonnet et al. (2008) found decreased 5-HIAA levels in the frontal cortex, albeit in rats, while the 5-HIAA/5-HT ratio, as a measure of catabolic turnover, and, overall serotonin content remained unaffected. On the other hand, Tian et al. (2019b) reported an increase in serotonin levels in the prefrontal cortex, but not the brainstem, of chronically stressed adult male C57BL/6J mice after administration of the Breve strain. Furthermore, Engevik and colleagues showed that administration of the dentium strain to germ-free mice enhanced the expression of the 5-HT2A receptor primarily in the CA1 subregion of the hippocampus. Changes were accompanied by SCFA composition changes in feces in some cases. Acetate was found to be increased in the trial with the dentium strain (Engevik et al., 2021), while acetate, n-butyrate, propionate and isobutyrate were found to be decreased with the infantis strain (Tian et al., 2019b).
Four other types of probiotics were also found to enhance serotonin levels in the brain. Pyrroloquinoline quinone-producing Escherichia coli affected whole brain serotonin levels in subcutaneously 1,2-dimethylhydrazine-injected rats (Pandey et al., 2015), while Clostridium  and Akkermansia (Yaghoubfar et al., 2020) both affected hippocampal levels in male stressed and non-stressed mice, respectively. In addition, Clostridium also decreased MAO, SLC6A4 and 5-HT1A/2A/5/6 receptor expression, while simultaneously increasing TPH2 expression in the hippocampus. Next, Clostridium also improved depressive-like behavior . On the other hand, Enterococcus faecium had the opposite effect on whole brain serotonin content in stressed goslings (Ibrahim et al., 2018).
Combinations of several probiotics improved age-related cognitive decline (Corpuz et al., 2018) and possibly depression in mice or rats (Tian et al., 2019a;Tillmann et al., 2018). No effect was found on anxiety, social behavior or memory, as reported by Tillmann et al.

Table 3
Preclinical intervention studies in animal models investigating the effect of pre-and probiotics on brain serotonin levels, receptors, transporters, enzymes and related gene expression. -The probiotic increased brain 5-HT levels and downregulated MAO -It reduced age-related cognitive decline Davis et al. (2016) Wild-type zebrafish (adult) Unclear ± 30 CMS Lactobacillus plantarum -The probiotic induced the upregulation of 5-HT transporter SLC6A4a, but not SLC6A4b -It also had an anxiolytic effect *the effect of the probiotic on healthy non-stress induced zebrafish was not studied Desbonnet et al. (2008) Male Sprague-Dawley rats (adult) 20 14 -Bifidobacterium infantis 35624 -The probiotic increased plasma concentrations of tryptophan and decreased 5-HIAA in frontal cortex -No effect on 5-HT/5-HIAA ratios in the frontal cortex; no effect on brain 5-HT levels Engevik et al. (2021) Swiss Webster germ-free mice (adult) 50 14 -Bifidobacterium dentium -The probiotic increased 5-HT2A expression in CA1 region of HC -It increased fecal acetate levels Fleming et al. (2019) Naturally farrowed intact male pigs (newborn) 24 31 -polydextrose and galactooligosaccharide -The prebiotic decreased hippocampal and striatal 5-HT levels -The probiotic improved recognition memory and exploratory behavior -Hippocampal 5-HT correlated with exploratory behavior, but not with recognition memory Ibrahim et al. (2018) Chicks (1 week) 12 42 Induced stress (with a higher density of chicks per square meter) Enterococcus faecium -A higher density of chick per square meter increased brain 5-HT levels -The probiotic lowered brain 5-HT levels, however, only in the high density group Kao et al. (2018) Female Sprague-Dawley rats (6-8 weeks) 24 7 -galacto-oligosaccharides -The prebiotic had no effect on brain 5-HT2A receptor protein and mRNA levels Li et al. (2019) Male Wistar rats (age not mentioned) 50 28 CMS Bifidobacterium longum, Lactobacillus rhamnosus (probiotics) and fructooligosaccharide and galactooligosaccharide (prebiotics) -CMS decreased TPH2 and 5-HT, while it increased IDO levels in HC and frontal cortex -Pre-and probiotics enhanced TPH2 and 5-HT, but decreased IDO levels in HC and frontal cortex *the effect of the pro-and prebiotics without CMS was not studied Liang et al. (2015) Male specific-pathogenfree  -5-HT levels were decreased, without alterations in 5-HIAA levels, in stressed mice vs. nonstressed -The probiotic reduced 5-HIAA in stressed mice; naïve mice had reduced 5-HIAA but increased 5-HT -The probiotic increased locomotor activity and decreased anxiety in both naïve and stressed mice -It decreased depressive-like behavior in stressed mice only Liu et al. (2016) Male germ-free C57BL/ 6JNarl mice (6 weeks) 18 16 -Lactobacillus plantarum PS128 -The probiotic increased 5-HT and 5-HIAA levels in striatum, however, not in prefrontal cortex or HC -It decreased anxiety-like behavior and increased locomotor activity Liu et al. (2019) Male Wistar rats (7-8 weeks) 24 28 CMS Lactobacillus fermentum PS150 -CMS induced memory and learning deficits and a drop in whole brain 5-HT -Probiotic prevented the memory and learning deficits and the drop in whole brain 5-HT *the effect of the probiotic without CMS was not studied Luo et al. (2014) Specific-pathogen-free male Sprague-Dawley rats ( Escherichia coli CFR 16 -The induced oxidative stress decreased brain 5-HT levels -The probiotic enhanced brain 5-HT levels *the effect of the probiotic in rats without induced oxidative stress was not studied Savignac et al. (2016) Male CD1 mice ( Sun et al. (2018) Male C57BL/6 mice (6-8 weeks) 30 28 CMS-induced depression Clostridium butyricum WZMC1018 -CMS reduced 5-HT levels in HC and induced depressive-like behavior -The probiotic elevated the hippocampal 5-HT levels and improved depressive-like behavior *the effect of the probiotic in healthy mice without CMS was not tested Szklany et al. (2020) Male BALB/c mice (newborn) and their mothers (adult) 20 pups, 11 dams 77 -Galacto-oligosaccharides and longchain fructo-oligosaccharide -The prebiotic decreased tryptophan and 5-HT levels in the prefrontal cortex and enhanced 5-HT/5-HIAA ratio in the somatosensory cortex; no differences were measured in the amygdala or HC -It decreased the expression of 5-HT1A receptor mRNA in the prefrontal cortex, but not amygdala or HC -TPH2 was unaffected by the prebiotic -It decreased anxiety-like and repetitive behaviors and enhanced social behavior in adulthood -It enhanced acetate, propionate and butyrate levels in fecal samples Tian et al. (2019a) Male C57BL/6J mice ( (2018). In general, brain serotonin levels were increased in the whole brain of senescence-accelerated mice (Corpuz et al., 2018). In rats, the same effect was found specifically in the hippocampus (Li et al., 2019;Tian et al., 2019a) and frontal cortex (Li et al., 2019). However, Tillmann et al. (2018) found no effect in the prefrontal cortex and hippocampus in a genetic rat model of depression. Nevertheless, the probiotic mixture, enriched with prebiotics, showed an increase in TPH2 and a decrease in IDO in both hippocampus and (pre)frontal cortex of male Wistar rats (Li et al., 2019). Apart from TPH2, MAO levels were observed to be downregulated in an alike probiotic mixture intervention too (Corpuz et al., 2018). Finally, the use of only prebiotics has been previously looked at with regards to brain serotonin levels albeit by few studies so far. Firstly, it affected cognitive function in pigs (Fleming et al., 2019), as well as anxiety and behavior in new-born mice and their mothers (Szklany et al., 2020). Secondly, findings on the effect on expression of serotonin receptors 5-HT2A, 5-HT2C and 5-HT1A are rather contradictive between prebiotic type and/or study (Kao et al., 2018;Mika et al., 2017Mika et al., , 2018Savignac et al., 2016;Szklany et al., 2020). Thirdly, serotonin levels were found to be decreased in specific areas of the brain in pigs and male BALB/c mice (Fleming et al., 2019;Szklany et al., 2020).

Discussion
The involvement of the microbiota-gut-brain axis in AD with possible implications for prevention and treatment have been highlighted previously (Arora et al., 2020;Kesika et al., 2021;Liu et al., 2020). Additionally, the suggestion of a(n) (in)direct link between the axis and AD development due to neurotransmitter alterations (serotonin, gamma aminobutyric acid) has recently been raised by a Mendelian randomization analysis (Zhuang et al., 2020). Notably, there is also a phase three trial ongoing with GV-971, a pharmaceutical drug derived from seaweed extracts (sodium oligomannate), targeting the gut microbiota (NCT04520412; for review . These recent developments highlight the importance of the axis in the search for disease-modifying therapies, apart from the modifying role of serotonin and its derivatives within the microbiota-gut-brain axis in the development of AD in particular.

Modulatory effects of the serotonergic system in Alzheimer's disease
The literature search was aimed at finding out if and how serotonergic alterations and AD development are related. In this regard, the majority of enlisted studies, mainly preclinical but also a few human intervention trials, show that SSRI and serotonin receptor (ant)agonists may very well modify the underlying neuropathology, with inclusion of clinical symptoms. Though the effectiveness of treatment might be dependent on the disease stage, as already highlighted by the trial of Giannoni et al. (2013). The suggested modulatory involvement of the serotonergic system is further strengthened by mechanistic in vitro studies. This is exemplified by the work undertaken by Hornedo-Ortega et al. (2018), who showed that serotonin is able to prevent destabilization of Aβ oligomers and fibrils and thus insoluble plaque formation. This effect could be established by disruption of salt bridges between and within Aβ42 protofibrils, as well as beta-sheet structure . Apart from the effect on plaque burden, serotonin might also exert neuroprotective effects mediated via its action on heat shock protein 70, SIRT-1 and SIRT-2 gene expression, as evidenced in rat cells (Hornedo-Ortega et al., 2018).
On the contrary, few trials, failed to replicate the effects of SSRI on Aβ plaque reduction in the hippocampus (Von Linstow et al., 2017) alongside mitigating the cognitive dysfunction (Klaassens et al., 2019;Olesen et al., 2017;Severino et al., 2018). In the case of Klaassens et al. (2019), this could be due to the small sample size and the single administration dosage. The unexpected findings of the three animal trials might be partially explained by administration route, since the trials belong to the minority that administered the SSRI orally. Other factors, such as sample size, intervention duration and type of AD model, do not seem to be crucially different as compared to the other animal trials that did find an effect. Furthermore, the effects of SSRI on P-tau remain ambiguous, since the findings of Halliday et al. (2017) and Jin et al. (2017) contradict each other even though both trials involved mice that overexpressed the human tau mutation. Finally, the effect of SSRI on the neocortex plaque load, and, 5-HT3 receptor antagonists on cognition, seem absent, although again this might be due to the administration route. Nevertheless, the majority of included studies supports the overall hypothesis that brain serotonergic neurotransmitter system alterations are intrinsically involved in AD pathophysiology, Lactobacillus plantarum DR7 -D-galactose injections affected cognitive function, memory, anxiety and TPH1 expression -The probiotic increased TPH1 expression and improved cognitive function, memory and anxiety *the effect of the probiotic was not studied in healthy rats Abbreviations: 5-HIAA: 5-hydroxyindoleacetic acid; 5-HT: 5-hydroxytryptamine (serotonin); 5-HTP: 5-hydroxytryptophan; Aβ: amyloid-beta; AD: Alzheimer's disease; APP: amyloid-precursor protein; CMS: chronic mild stress; DG: dentate gyrus; DRN: dorsal raphe nucleus; HC: hippocampus; IDO: indoleamine 2,3-dioxygenase; MAO: monoamine oxidase; PS1: presenilin 1; P-tau: phosphorylated tau; SSRI: selective serotonin reuptake inhibitors; SCFA: short-chain fatty acid; TPH: tryptophan hydroxylase.
thereby suggesting that interfering with its evolution from the earliest stages onwards could be a viable target for prevention, and, possibly (symptomatic) treatment. This notion is consistent with the review of Joshi et al. (2020) concerning multiscale and multilevel serotonergic modeling approaches for AD.

The potential of the microbiota-gut-brain axis in modulating the brain's serotonergic system
The question hereafter remained whether these serotonergic alterations could be prevented or prohibited by modulating the microbiotagut-brain axis. Indeed, nutrition, probiotics, prebiotics and FMT seem to affect serotonin levels, serotonin receptors, related enzyme expression (TPH1, TPH2, MAO, IDO) and serotonin transporter (SLC6A4) expression in the brain. This observed connection between gut and brain with respect to serotonin, might be the consequence of multiple different interactions as visualized in Fig. 3. One example could be modulation of vagus nerve activity similarly as is the case with SSRI (McVey Neufeld et al., 2019). In this study, for instance, oral SSRI administration enhanced vagus nerve activity, and, vagotomy subsequently removed its antidepressant effect. More recently, bacterial tryptophan metabolites have even been linked to vagus nerve stimulation, through the activation of epithelial sensory enteroendocrine cells of the intestine (Ye et al., 2021). In addition, serotonergic changes in the included studies were often accompanied by cognitive or behavioral changes. As an example, the prebiotic trial of Liu et al. (2019) revealed both cognitive improvement as well as whole brain serotonin level enhancement. Although improvement in brain function might be a direct effect of the enhanced serotonin level (Fig. 3), it could also be related to the alternative fate of dietary tryptophan, namely the kynurenine pathway. Metabolites of this pathway, such as quinolinic acid (Fig. 1), have shown to be neurotoxic through a variety of mechanisms like agonizing the N-methyl-D-aspartate receptor (for review: Lugo-Huitrón et al., 2013). Thus, inhibiting the formation of such compounds through the action of microbiota, for instance by lowering the activity of IDO or availability of tryptophan, might reduce neurotoxicity which is logically beneficial for overall brain health. Accordingly, Yu et al. (2015) showed the positive effect of IDO inhibition on cognition, Aβ formation and neuronal loss, while Parrott et al. (2012) highlighted a preventive effect on anxietyand depressive-like symptoms, both in AD mouse models. Notably, not all metabolites of the kynurenic pathway are necessarily detrimental for the brain. Nicotinamide adenine dinucleotide (NAD + ), which is an essential cofactor important for mitochondrial function, gained interest as a possible modulator of age-related diseases (for review: Castro-Portuguez and Sutphin, 2020;Verdin, 2015).
Observed (beneficial) serotonergic alterations, however, are not consistent in all included intervention trials. For example, the impairment of learning and memory is accompanied by an increase in hippocampal serotonin levels in one of the FMT trials. This seems contradictory to the hypothesis that decreased (hippocampal) serotonin levels in AD affect cognition in addition to the previously described positive effects of SSRI in AD animal models. As for two of the included dietary interventions, serotonin itself seemed unaffected, although serotonin-related changes (e.g. tryptophan or 5-HIAA levels) were observed next to effectively modulated behavior. However, it should be noted that only serotonin levels in the brainstem were measured in one trial, compared to a very small sample size in the other. Moreover, there were only a few dietary trials available, investigating a limited variety of analytical neurochemical measures. Dietary enrichment with high levels of tryptophan, however, revealed to reduce Aβ load in a transgenic AD mouse model . Unfortunately, the researchers could not reveal direct serotonergic changes in hippocampus nor raphe nuclei related to the higher tryptophan intake, apart from increased sprouting of hippocampal serotonergic fibers in the transgenic AD mouse model (irrespective of diet) as a potential defense mechanism against Aβ accumulation . One proposed neuroprotective mechanism by which serotonergic sprouting in the vicinity of plaques in AD brain might exert its effects, is via the hyperpolarization of nearby neurons through activation of 5-HT1A/B receptors, and subsequent opening of K + channels. Hyperpolarisation in turn limits Ca 2+ entry, and, hence, excitoxicity, since voltage-gated calcium channels then will remain closed and the Mg 2+ block of NMDA receptors becomes favored .
Finally, pre-and probiotics seem to differ widely in their impact on the brain, which highlights the importance of strain choice. Lactobacillus plantarum is one of the most studied strains and seems potent in modulating brain serotonin with beneficial effects on both cognition and behavior. On the other hand, Enterococcus faecium lowered brain serotonin levels in density-stressed goslings. This effect was also observed in two prebiotics trials in the hippocampus of pigs, and the prefrontal cortex of mice. The largest limitation for most included strains is the lack of replication studies, as well as the inconsistency in study endpoints. Some studies focus on enzymes and receptors, while others focus exclusively on serotonin and its metabolite levels. Another variation can be found in the targeted region for measurements: some were done in whole brains, several others in only a few regions. On the whole, the mentioned limitations make it difficult to draw final conclusions, apart from the general observation that diet, FMT, pre-and probiotics, as well as bacterial strain choice seem intrinsically linked with serotonergic changes, irrespective of the direction of change (increase or decrease) and measured analyte (whether whole levels, or, receptors, enzymes or transporters).

Meanings of the above findings for possible clinical applications
The implications of the discussed interventions for AD development remain to be determined, since neither of the included trials similarly assessed (i) serotonin levels both in gut and brain (or associated biofluid), (ii) entry route (e.g. pre-/probiotics, diet, FMT), and, (iii) cognitive and/or behavioral outcome in involved (iv) AD mouse models or patients at (v) different disease stages. Furthermore, it should be kept in mind that everyone has their own individual microbiota composition, which is likely to impact their personal response to FMT, diet, pre-and probiotics. This is not represented in the included FMT trials, since effects on the brain's serotonergic system were measured in subjects that underwent antibiotic treatment prior to FMT. Consequently, it remains unclear whether the microbiota from the transfer could colonize the gut sufficiently, and, consequently, alter gut-brain communication in actual humans with their own initial microbiota composition. In clinical studies with FMT transfers from healthy subjects to (irritable bowel disorder or depressive) patients, psychiatric symptoms did improve, however, this benefit lasted only for about three to six months (Chinna Meyyappan et al., 2020). Noteworthy, one case study observed rapid improvement of cognitive and behavioral symptoms following FMT in an AD patient that suffered from an infection with Clostridioides difficile. Stool from the patient's 85-year-old wife as a donor was used. The improvement was noticeable up to six months post-intervention (no further data provided) (Hazan, 2020). The same question about effectiveness in humans could be raised for the pre-and probiotic trials, especially since several included trials used gnotobiotic or germ-free mice. Meanwhile, a randomized controlled trial that investigates the effect of Bifidobacterium (three months administration) on microbiota composition, brain networks and cognition in individuals with amnestic MCI is ongoing (NCT03991195), next to an alike trial in which both Bifidobacterium and Lactobacillus strains will be supplemented to AD patients for 12 weeks (NCT05145881). Brain serotonin measurements are unfortunately not part of the outcome measures in neither studies. Finally, external factors in preclinical studies should also be vigorously investigated and controlled for before actual translation to the human situation, since SSRI treatment efficacy, for instance, has been hypothesized to be largely dependent on environmental influences, with even a chance of significant worsening rather than improvement if under Fig. 3. Serotonergic functions mediated by gut-brain crosstalk in relation to potential enhancers. Serotonin (5-HT) is a critical modulator of microbiota-gutbrain axis signaling and exerts multiple functions throughout the human body that relate to both brain and gut (outer circle). Its production, availability and activity is influenced in various ways, with gender, genetics (e.g. enzymatic activity, receptor distribution) and medication (e.g. SSRI) as main external or physiological (internal) determinants. Firstly, dietary or supplemental tryptophan (Trp) can be transformed in the gut by the enterochromaffin cells (ECC) to 5-hydroxytryptophan (5-HTP) by the action of tryptophan hydroxylase (TPH)1, and, subsequently, to serotonin by aromatic L-amino acid decarboxylase (AADC). Following its release, 5-HT interacts with receptors on the enteric nervous system to modulate gut motility among others, and, to induce further signaling along the vagus nerve. Vagal afferents further propagate the signal to the dorsal raphe nuclei and the nucleus of the solitary tract. Both nuclei connect with emotion-regulating brain networks that control mood, which in effect may further determine eating behavior. Secondly, 5-HT production via the ECC can also be effectuated by the intake and digestion of dietary fiber or related prebiotics, following which the microbiota produce short-chain fatty acids (SCFA; e.g. propionate, butyrate, acetate). These SCFA stimulate the ECC for additional 5-HT synthesis. Particular strains of gut microbiota can also synthesize neurotransmitters themselves. Importantly, the intermediate of 5-HT synthesis, 5-HTP, can pass the blood-brain barrier from the systemic circulation, whereas 5-HT cannot. In neurons, Trp is transformed into 5-HTP by the action of TPH2, and, further to 5-HT via AADC. The vagus nerve can be considered as the highway along which 5-HT modulates the gut-brain connection, having a reciprocal interaction. With regard to Alzheimer's disease (AD) development, fecal microbiota transplantations, diet and pre-/probiotics could enhance the abovementioned pathways, and boost brain serotonergic neurotransmission in the end (e.g. hippocampus; limbic cortex). This could result in altered behavioral and cognitive outcomes, or, depending on the disease stage, prevent, attenuate or delay neuroinflammation and thus subsequent plaque or tangle formation (upper left corner). Further involvement of the alternative fate of dietary Trp, i.e. the kynurenine pathway, related to neuroinflammatory processes in AD progression has not been included in this figure. Abbreviations: 5-HIAA; 5-hydroxyindoleacetic acid; 5-HT: 5-hydroxytryptamine (serotonin); 5-HTP: 5-hydroxytryptophan; AADC: aromatic Lamino acid decarboxylase; BBB: blood-brain barrier; DRN: dorsal raphe nuclei; ECC: enterochromaffin cell; ENS: enteric nervous system; FMT: fecal microbiota transplantation; IBS: irritable bowel syndrome; MAO: monoamine oxidase; MCI: mild cognitive impairment; NTS: nucleus tractus solitarius (nucleus of the solitary tract); SCD: subjective cognitive decline; SCFA: short-chain fatty acids; SERT: serotonin transporter; SSRI: selective serotonin reuptake inhibitors; TPH: tryptophan hydroxylase; Trp: tryptophan. Brain images (12% formalin-fixated) are at the courtesy of the picture archive of the Neurobiobank of the Institute Born-Bunge (Antwerp, Belgium;FAGG registration no. 190113). Created with BioRender.com. stressful living conditions (Alboni et al., 2017;Severino et al., 2018). In this context, one proposed mechanism might be the enhanced neuronal plasticity following increased serotonergic neurotransmission, rendering the individual more susceptible to the quality of the living environment.

Limitations and reflections
Some methodological limitations and reflections need to be considered firstly. For instance, the narrow focus of the review, which is mostly on serotonin only. Metabolites and precursors related to its synthesis and metabolization pathways, such as melatonin, tryptophan, and, the neuroinflammatory kynurenine pathway (e.g. quinolinic and kynurenic acid), are beyond the scope of this review. Their importance should certainly not be underestimated and can be placed in a general conceptual framework of neuroinflammation in AD (for review: Gheorghe et al., 2019;Maitre et al., 2020). Another important aspect to take into account, is the fact that the observed correlations between altered brain serotonin content and improved clinical outcome and/or attenuated AD pathology, for instance, following SSRI treatment, do not necessarily imply causality. The same goes for the observed serotonergic effects in brain of the enumerated preclinical studies researching pre-and probiotics, FMT, and, whole diet approaches/dietary restrictions. Given that the serotonergic neurotransmitter system both in gut and brain may serve as an intermediate nexus for neighboring and alike neurotransmitter systems, such effects may be rather indirect. It remains to be evidenced still whether serotonin degeneration may be a downstream effect of AD pathology or may have a causative role after all. SSRI treatment does not unequivocally interfere in the progression of human AD, perhaps because of complex effects of chronic SSRI treatment on multiple serotonin receptor subtypes (Gründer and Cumming, 2021). The discrepancy between animal studies with a successful outcome and the lack of replication in clinical trials is often witnessed in that regard. It is, therefore, a difficult enterprise to attribute a causal link for serotonin systems, however, a handful of studies so far have emerged, revealing modifying effects via direct structural and molecular interactions between serotonin and Aβ. A final limitation might be the exclusion of studies that measured serotonin levels, receptors, enzymes or transporters solely in gut and/or blood. These endpoints are often used in human trials due to more expensive, and, perhaps, somewhat more invasive in vivo brain measurements (e.g. PET scans). Although these do not necessarily provide relevant information on brain serotonin content and alterations, such studies certainly could contribute to the overall understanding of serotonin across the microbiota-gut-brain axis. As for imaging studies, these are very much wanted in view of our proposed hypothesis, however, these should be executed with suitable radioligands, and, preferentially, in combination with peripheral analyses of serotonin synthesis or metabolism.

Conclusions and general considerations
All in all, current reviewed evidence suggests that the brain's serotonergic neurotransmitter system is intrinsically involved in the development of AD. Additionally, this system could be modulated through the microbiota-gut-brain axis, using pre-and probiotics, FMT and nutrition, at least as evidenced in various preclinical studies. A next step would be executing randomized placebo-controlled trials focused on pre-and probiotics, FMT and diet, in actual AD mouse models, at different ages of the disease pathology. In this regard, transgenic mouse models that cover at least both the tau and Aβ abnormalities should be preferred (such as APP/PS1/TauP301L transgenic mice). Study endpoints should ideally cover cognitive aspects, neuropsychiatric symptoms (such as depression and aggression), and, central (brain) as well as peripheral (CSF; blood; gut (biopt or fecal materials)) measurements of serotonin levels, receptors, enzymes (IDO, MAO, TPH2, TPH1) and/or transporter expressions. A distinction could be made between neurochemically and behaviorally important brain regions, such as the hippocampus, brainstem, amygdala and frontal cortex. Functional metagenomics approaches using fecal materials to further identify how bacterial metabolites might (in)directly affect serotonergic signaling remain a very powerful tool in this effort (Jameson et al., 2020). Next, largescale human randomized placebo-controlled intervention trials are required to determine in which stage of the Alzheimer's continuum these modulators (e.g. pre-/probiotics; FMT; diet) of the serotonergic system might have the most promising effect, preferably spanning from the prodromal stages, such as subjective cognitive decline or MCI due to AD, up to the milder AD stages, where both high adherence to such therapies, as well as sufficient room for noticeable enhancement are feasible still. In the end, such trials might facilitate the development of a comprehensive approach to tackle this complex multifactorial disease, since serotonin and its derivatives across the microbiota-gut-brain axis might serve as potential biomarkers of disease progression (Tajeddinn et al., 2016), next to forming a valuable target in AD prevention strategy and drug development.

Declaration of Competing Interest
None.