Sleep, circadian rhythm and gut microbiota: alterations in Alzheimer’s disease and their potential links in the pathogenesis

ABSTRATC In recent years, emerging studies have observed gut microbiota (GM) alterations in Alzheimer’s disease (AD), even in individuals with mild cognitive impairment (MCI). Further, impaired sleep and circadian patterns are common symptoms of AD, while sleep and circadian rhythm disruption (SCRD) is associated with greater β-amyloid (Aβ) burden and AD risk, sometimes years before the clinical onset of AD. Moreover, reports have demonstrated that GM and its metabolites exhibit diurnal rhythmicity and the role of SCRD in dampening the GM rhythmicity and eubiosis. This review will provide an evaluation of clinical and animal studies describing GM alterations in distinct conditions, including AD, sleep and circadian disruption. It aims to identify the overlapping and distinctive GM alterations in these conditions and their contributions to pathophysiology. Although most studies are observational and use different methodologies, data indicate partial commonalities in GM alterations and unanimity at functional level. Finally, we discuss the possible interactions between SCRD and GM in AD pathogenesis, as well as several methodological improvements that are necessary for future research.


Introduction
Alzheimer's disease (AD) is a degenerative central nervous system (CNS) disorder, characterized by a progressive onset of neurocognitive symptoms, including amnesia, aphasia, disorientation, etc. 1 While the etiology of AD remains largely unknown, AD is generally featured by the deposition of β-amyloid (Aβ) and the formation of neurofibrillary tangles of tau protein in CNS.
The human body harbors a large variety of microorganism communities which intensively interact with host and each other through direct contacts or metabolites. 2 It has long been postulated that human gut microbiota (GM), the collection of all microorganism communities in the human digestive tract, holds great significance to human health and disease. 3,4 However, not until recently have we been able to investigate their composition and function with the advances in DNA sequencing and metagenomic analysis techniques. 5 Moreover, brain-gut-axis (BGA), which studies the interactions between GM and CNS, has gained significant attention in recent years. There is much evidence showing altered GM composition in several neurological diseases, including Parkinson's disease (PD) and autism spectrum disorder (ASD). [6][7][8] Changes in GM composition and richness have also been observed in AD patients and individuals with mild cognitive impairment (MCI), 9,10 suggesting a potential role of GM dysbiosis in AD pathogenesis.
Several neurodegenerative diseases including AD, PD and Huntington disease (HD) have been implicated with sleep disturbance and circadian rhythm dysfunction. 11 While sleep and circadian rhythm disruption (SCRD) are usually recognized as the consequences of these diseases, studies have reported the existence of sleep disorders long before the onset of AD and PD, even by decades. [12][13][14][15] Moreover, growing evidence indicates that sleep disturbance and circadian rhythm misalignment may contribute to neuroinflammation, low Aβ clearance efficacy, increased concentration of reactive oxygen species (ROS), compromised blood-brain-barrier (BBB) and GM dysbiosis. [16][17][18] However, the present work revealed the correlation between SCRD and AD, but not causality, and further work is needed to resolve this issue.
Studies in the last few decades have long examined common determinants of the human GM, including diet, medicine and stress. 19,20 Recent findings suggest a novel role of sleep and circadian rhythm in shaping and modulating the composition of GM. 21 However, to the best of our knowledge, no reviews to date have considered the possible contributions of synergistic interactions between SCRD and GM dysbiosis to the pathogenesis of AD. In this review, we first present recent studies that examined the GM alterations in AD and SCRD. We summarize those findings and compare the GM changes at both compositional and functional levels across studies. We observe commonalities in GM alterations of individual bacteria and unanimous changes at functional level between AD and SCRD conditions. Therefore, we discuss possible interactions between SCRD and GM, which contribute to AD onset by inducing peripheral and central inflammation ( Figure 1). We reason that this is achieved through various pathways including disrupted gut barrier integrity, compromised blood-brain barrier (BBB), decreased short-chain fatty acids (SCFAs) production and increased pro-inflammatory metabolites.

GM and AD
The role of microorganisms in the pathogenesis of AD was initially proposed by Alois Alzheimer, the first describer of this progressive neurodegenerative disorder. 22 After decades of insufficient research, there has been a resurgence of interests in this hypothesis, largely owing to a growing body of evidence from clinical and animal tests. Several kinds of infectious agents such as bacteria, fungi, virus and protozoa that are highly associated with AD have been reviewed elsewhere. 1,[23][24][25] In this part, we focus on GM alterations, probiotic and antibiotic treatments, and fecal microbiota transplantation (FMT) in both AD patients and models. A hypothetical model of linking SCRD, GM and AD pathogenesis. SCRD caused by sleep disorders or working night shift impairs brain functions in many ways, one of which acts through GM. SCRD leads to GM dysbiosis, with increase in pathobionts and decrease in beneficial bacteria. In the bottom of the figure, blue color represents symbionts such as beneficial bacteria, while red color represents pathobionts. Integrated gut barrier and BBB normally block pathogens such as bacteria metabolites from entering the brain. However, GM dysbiosis caused by SCRD disrupt gut barrier and BBB by degrading mucin and releasing proinflammatory agents and neurotoxic metabolites. These pathological changes can cause aberrant neuroinflammation, and subsequently lead to Aβ deposition and AD onset.

GM alterations in AD: from clinical and animal literature
Recent clinical observations have found significant GM alterations in both AD and MCI patients. Here, we summarize the alterations of GM composition in AD patients compared to controls in Table 1 (top). 9,10,[26][27][28] In addition, animal models are also used in other studies, and the relevant findings are summarized in Table 1 (bottom). [29][30][31][32][33][34][35][36] Note that transgenic mice including APP/PS1, SAMP8, 5xFAD and their derivatives were the most frequently used AD models. 37 Substances such as D-galactose, Aβ protein and lipopolysaccharide (LPS) were also used in several studies to induce AD pathology. 38 It has been suggested that α-diversity analysis and Firmicutes/Bacteroidetes (F/B) ratio, two frequently used criteria in microbiome analysis, are not reliable in investigating the association between GM alteration and PD. 6,39 Interestingly, we also found inconsistent results of α-diversity, F/B ratio and GM changes at high phylogenetic rank (e.g., phylum, class and order level) in both AD and   SCRD studies. The findings showed better concordance at higher taxonomic resolution. Therefore, GM alterations at family, genus and species level are presented in the following tables (Tables 1-5). Generally, we have identified higher level of pathobionts and lower level of beneficial bacteria in both AD patients and animals ( Figure 2). The pro-inflammatory taxa Escherichia and Shigella of Enterobacteriaceae have long been proposed to contribute to series of gastrointestinal diseases. 10 Increased level of E. coli LPS has also been detected in the postmortem brain samples of AD patients. 40 The exotoxin of Escherichia and Shigella could disrupt the integrity of epithelial cell further leading to leaky gut and facilitates the translocation of bacteria into the blood. 41 E. coli along with several gram-negative bacteria possess systems for producing bacterial Aβ which is able to penetrate intestinal barrier and BBB and initiate crossseeding in the CNS. 42,43 In addition to Escherichia, bacterial Aβ producing systems have also been found in Staphylococcus, highlighting its potential role in contributing to AD pathogenesis. 44 Although Staphylococcus was not detected in human fecal sample, its higher abundance was found in the blood of AD patients. 9 Studies have reported that strains of Ruminococcus gnavus which belong to the family Lachnospiraceae use terminal mucin glycans to degrade mucus layer of intestinal barrier. 45 Increased level of Ruminococcus gnavus has been associated with inflammatory bowel disease, suggesting the potential role of Ruminococcus gnavus in promoting inflammation. 46 The two families Ruminococcaceae and Clostridiaceae, major SCFA-producing taxa in mammalian GM, have been reported to be decreased in various metabolic and neurodegenerative diseases. 47 The relative abundance of Ruminococcaceae was found to be positively correlated with higher Mini-mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA) scores, which indicates better cognitive functions. 10 Lower level of anti-inflammatory taxa Eubacterium rectale and Bacteroides fragilis along with increased pro-inflammatory cytokines such as IL-1β, NLRP3 and CXCL2 have been also detected in AD patients. 28 Lactobacillus and Bifidobacterium are two common probiotic taxa capable of producing neurotransmitter gamma-amino butyrate (GABA) whose metabolism has been reported to be disrupted in AD patients. 48 Lactobacillus and Bifidobacterium play an important role in protecting intestinal cells and inducing anti-inflammatory responses. 49,50 Studies have shown that probiotic treatment using strains of Lactobacillus and Bifidobacterium was able to ameliorate symptoms associated with AD. 51,52 GM interventions restore the progression of AD As stated above, most studies focusing on GM and AD presented correlations but not causal relationships. While it remains an open question in the field, 53 several studies have begun to demonstrate how GM affect AD pathology by showing the beneficial effects through GM intervention in animal models, including probiotic supplement, 51,52,54-58 antibiotic treatment, [59][60][61][62][63][64] germ-free (GF) animals 36,63,65 and fecal microbiota transplantation (FMT). 31,32,35,36,61,62 These successful trials support the role of GM dysbiosis in contributing to AD pathogenesis and progression and suggest potential benefits of GM modulation for AD treatment (Table 2) ( Figure 3).

Sleep, circadian rhythm and GM
Although human gut ecosystem maintains rather resilient, perturbation by antibiotics, high-fat food and stress could damage intestinal homeostasis. 3,66 These key determinants of GM have been studied extensively over the past decades, but the role of sleep and circadian rhythm in regulating GM was underestimated. 67 Recent studies have shown that human GM display diurnal oscillation at both compositional and functional levels. 68 It has been suggested that SCRD may lead to GM dysbiosis through several indirect ways, including disrupting the rhythmic fluctuation of GM, activating the HPA axis, increasing food and energy intake, decreasing physical activity and damaging gut barrier integrity. 21,69,70 In this part, we summarize recent progress regarding the correlation between SCRD and GM dysbiosis as well as how SCRD impacts GM (Tables 3, 4). Like the findings in AD, increased pathobionts and decreased beneficial bacteria were identified in SCRD conditions in both human and animal models.  (1) vs. (2) and (3) -↑ MAPK signaling pathway in (1) vs. (2) and (3) 63 Male 5xFAD mice Exp: GF AD mice Con: SPF AD mice GF mice were generated through embryo transfer -↑ ceca size and weight  (1) and (2) -Higher level of increased brain Aβ42 in (1)  Prevotellaceae is also an immunogenic bacterial taxon highly coated by IgA. 82 It has also been suggested that species of Prevotellaceae could induce intestinal inflammation, slow the   development of mucus layer and are involved in various intestinal diseases including IBD and colitis. 83 Note that although sleep disturbance increased abundance of Ruminococcaceae and Lachnospiraceae in murine subjects, it is mainly due to increased food-intake as both families are highly fermentative bacteria utilizing the plantderived fiber and polysaccharides in chow food. 75

Decreased bacterial taxa by sleep disturbance
In human studies, a decline in the relative abundance of Ruminococcus is correlated with poor sleep quality ( Eubacteriaceae along with Clostridiaceae, Lachnospiraceae and Ruminococcaceae are important SCFAs producers of mammalian GM. 49 The SCFA butyrate plays an important role in maintaining gut barrier and regulating immune responses toward anti-inflammatory status. 84 The genus Eubacterium makes significant contribution to butyrate production since Eubacterium rectale makes up about 13% of the clostridial cluster XIVa. 49 Therefore, loss of Eubacterium caused by sleep disturbances could lead to a decline in butyrate level and disrupt the integrity of gut barrier. It has been found that the SCFA-producing taxon Akkermansia can successfully mitigate the development of obesity and diabetes, protect gut barrier integrity and stimulate anti-inflammatory responses. 85

Circadian rhythm disruption and GM alterations
In addition to sleep loss, circadian rhythm disruption is also receiving increasing attention, given the increased prevalence of altered sleep-wake cycle and jet lag, which are largely due to working night shift and traveling across time zones. Aberrant light exposure, high fat diet, alcohol consumption and irregular eating behavior have been found to induce circadian misalignment. 86 Numerous studies have indicated a link between circadian rhythm disruption with higher risk of pathological conditions including obesity, cardiovascular diseases and neurodegenerative diseases. The diurnal oscillation of human GM is partially controlled by central clock, 68 indicating the regulatory roles of circadian in GM eubiosis. Thus, we summarized recent studies focusing on the effects of circadian rhythm disruption on GM components in Table 4. 68,[87][88][89][90][91]

Increased bacterial taxa by circadian rhythm disruption
The GM of human after undergoing shift work or jet lag exhibited increased abundance of Erysiopelotrichaceae, Prevotellaceae and Lachnospiraceae at family level, Dorea at genus level, and Ruminococcus torques and Ruminococcus gauvreauii at species level (Table 4, top). In murine models, circadian rhythm disruption (mainly achieved by altering light-dark cycles) resulted in an increase of Erysiopelotrichaceae and Prevotellaceae at family level, Prevotella at genus level and Ruminococcus torques at species level, largely consistent with observations in humans ( Table 4, bottom).
Dorea, Ruminococcus torques and Ruminococcus gauvreauii utilize glycoside hydrolases to breakdown mucus layer and produce propionate. 92 Despite their SFCA-producing capacity, increased abundance of mucolytic bacteria has been associated with disrupted gut barrier and inflammatory bowel diseases. 93 Studies have suggested the role of Dorea spp. in inflammation through the promotion of IFNγ production and mucin degradation. 84,94 Significantly abundant pathobiont Ruminococcus torques has been found in patients with ulcerative colitis (UC) and CD. 93 Ruminococcus gauvreauii has been found to be positively correlated with proinflammatory parameters in rats with fatty liver. 95

Decreased bacterial taxa by circadian rhythm disruption
In human studies, circadian disruption led to decreased levels of genus Faecalibacterium and species Faecalibacterium prausnitzii (Table 4, top). Ruminococcaceae at both family and genus level, Turicibacter at genus level and Eubacterium plexicaudatum at species level were decreased in animal studies after the disruption of light-dark cycles ( Table 4,

bottom).
Faecalibacterium was the only diminished bacterial taxa caused by circadian rhythm disruption at genus level. Faecalibacterium prausnitzi, the sole species of genus Faecalibacterium, is one of the most abundant bacteria in human GM representing more than 5% of bacterial population in intestine. 96 It acts as an important SCFA butyrate producing taxon, similar to other members in Ruminococcaceae family. 97 Moreover, studies have reported a negative association of Faecalibacterium prausnitzi with various inflammatory bowel diseases including UC and CD, suggesting that it could be a health indicator. 96

GM and AD -causal or coincidental?
What is the role of GM dysbiosis in AD? It remains debatable whether GM dysbiosis plays as causal or merely consequential role in AD.
Recently, studies have started to support the idea that GM dysbiosis precedes the onset of AD and even contributes to AD pathogenesis. Li et al. found that AD and MCI groups had distinct GM compositions from healthy controls in both fecal and blood samples, largely consistent with a previous report by another group. 9,10 These findings provide a new perspective that GM dysbiosis starting at early MCI is a developing process with the cumulation and depletion of specific bacterial taxa. Studies of GM intervention in AD including probiotic supplement, antibiotic treatment, germ-free animals and FMT further reinforced the causal role of GM dysbiosis in AD pathogenesis. What causes GM dysbiosis before the onset of AD? Human GM is determined by multiple factors including early life exposure, medical intervention, diet, stress, sleep and circadian rhythm. 21 Many studies have associated these factors with GM eubiosis, and their potential impacts on AD pathogenesis. A recent paper proposed a perspective that dietinduced GM dysbiosis plays a role in the pathogenesis of AD. 44 Multiple reviews summarized GM alterations in AD and SCRD, respectively, but no reviews to date have systematically analyzed the patterns of GM changes in AD and SCRD simultaneously, or made a hypothesis linking SCRD, GM dysbiosis and AD.

Linking SCRD to AD through GM dysbiosis
As shown in the previous parts, GM alterations were observed in AD, sleep and circadian disruption, respectively. Reports have also indicated that GM alterations might contribute to AD pathogenesis. 98,99 Studies which have been reviewed elsewhere have shown that SCRD was associated with greater Aβ burden and AD risk, sometimes decades before the clinical onset of AD. 16 Therefore, we hypothesize that the interactions between SCRD and GM lead to GM dysbiosis indirectly; as a consequence, chronic systematic and neuro-inflammation and Aβ deposition occur, together with a plethora of metabolic and immunogenic responses that may finally contribute to the onset of AD ( Figure 4).
First, we check the uniformity in GM alterations and their potential contributions to health and disease under AD and SCRD conditions. We compared the GM alterations and their potential roles (beneficial bacteria, pathobionts or controversial taxa) in a taxonomic view under distinct conditions: AD, sleep and circadian disruption ( Table 5). We observe higher abundance of highly immunogenic Erysiopelotrichaceae at family level in both human and rodents in each condition, but most other changes in individual bacteria were inconsistent between human and rodent (Table 5), which may be caused by the differences in GM components between these two species. 100 Thus, Figure 4. Time-line for the development of AD via SCRD-induced GM dysbiosis. Long-term SCRD (e.g., insomnia, fragmented sleep, night shift work and frequent traveling between time zones) leads to chronic alteration of GM with overabundant pathobionts and reduced beneficial bacteria. GM dysbiosis disrupts gut barrier integrity and facilitates the invasion of pathogens and their metabolite (e.g., LPS, exotoxins and bacterial Aβ). These pro-inflammatory agents induce inflammation responses and compromise BBB structure, leading to neuroinflammation and the onset of early MCI. As MCI develops, progressive enrichment of pathobionts such as Enterobacteriaceae further exacerbate neuroinflammation, cognitive dysfunction and Aβ burden, which in the end contribute to the pathogenesis of AD.
when analyzing the overlapping of GM alterations in different conditions, we conduct separate evaluations in humans and rodents. In humans, SCFAsproducing Ruminococcaceae at family or genus level is shown to be significantly lower in either condition, whereas highly immunogenic bacteria including Erysiopelotrichaceae and Coriobacteriaceae at family level are shown to be significantly higher in each condition. Most other GM components are inconsistent between different conditions, sometimes due to no relevant data available at present (Table 5). In animal models, similar trends are observed in several bacteria individuals between different conditions. For example, beneficial bacteria including Lactobacillaceae, Bifidobacteriaceae, Turicibacteraceae and Highly immunogenic Pro-inflammatory Lachnospiraceae at family and/or genus level are significantly decreased in AD, sleep disturbance and/or circadian disruption, and other parts of pathobionts are uniformly increased, with the exception of Ruminococcaceae. As stated above, the increase in Ruminococcaceae during sleep disturbance was probably due to aberrant food intake. Next, we elucidate the potential role of GM dysbiosis in the development of AD by providing the evidence of how GM interventions, including probiotics, antibiotics, germ-free treatment and FMT, restore cognitive functions and alleviate AD pathology (Table 2) (Figure 3). Although various factors modulate GM composition, emerging evidence has indicated that SCRD could disturb GM and lead to GM dysbiosis. Most human studies merely investigated the correlation between SCRD and GM dysbiosis, while animal studies provided more insights into GM alterations under different SCRD conditions such as sleep deprivation, sleep fragmentation and circadian rhythm reversal. Studies have also revealed several possible mechanisms underlying how SCRD contributes to GM dysbiosis, including increased food intake, decreased physical activity, activation of HPA axis and compromised gut barrier integrity, and this topic has been reviewed elsewhere. 21,101 Finally, we evaluate the specific roles of each individual bacteria and its potential contributions to health and disease. Intriguingly, dysfunctions mediated by the GM alterations are ideally unanimous in AD and SCRD conditions. Both AD and SCRD are associated with more abundant pathobionts leading to pro-inflammation and lower SCFAs, and less level of anti-inflammatory, SCFAproducing, and gut barrier-protecting bacteria (beneficial bacteria) ( Table 5). These analyses demonstrate that GM dysbiosis caused by SCRD is largely consistent with the ones in AD, supporting our hypothesis that SCRD may contribute to AD partially by impacting on GM ( Figure 5).

Future directions
In this review, we intend to summarize and evaluate the commonalities and distinctiveness of GM alterations in different conditions including AD, sleep disruption and circadian rhythm misalignment. Although data implied commonalities in these conditions, there were also condition- Figure 5. Schematic diagram of how SCRD contributes to AD pathogenesis through GM dysbiosis. SCRD, such as sleep deprivation, sleep fragmentation and jet lag, disrupts gut homeostasis with increased pathobionts (e.g., Enterobacteriaceae, Erysiopelotrichaceae and Prevotellaceae) and decreased beneficial bacteria (e.g., Eubacteriaceae, Ruminococcaceae and other SCFA-producing taxa). On one hand, pathobionts could damage gut barrier and cause leaky gut through the degradation of mucus layer. Pathogens and their metabolites induce pro-inflammatory responses and lead to increased BBB permeability. Bacteria-derived Aβ and LPS invade CNS and are associated with neuroinflammation and Aβ pathology. On the other hand, the compromised functions of beneficial bacteria (e.g., inhibiting infection, promoting mucin expression, producing neuromodulators and anti-inflammation SCFAs) are overwhelmed by overabundant pathobionts. Thus, the elevated neuroinflammation and aggravated Aβ burden facilitate the onset of AD. specific changes in certain species. Significantly, heterogeneity of methodologies applied for genetic material extraction, DNA sequencing, the lifestyle of subjects and methods for data analysis could compromise the results among different studies and lead to inconsistency, which could be expected in human studies. We suggest that further work is needed to specify the alteration of GM at species and even strain level, and incorporate metabolic and functional analysis to reveal possible mechanisms linking GM dysbiosis and diseases using standardized experimental design and data analysis.

Phylogenetic analysis of GM needs to be conducted at a high taxonomic resolution
Studies have implicated that GM can be altered at lower taxonomic level without achieving alteration at high taxonomic level. 39 For example, Firmicutes and Bacteroidetes are the two largest bacterial phyla of the mammalian gastrointestinal tract, and their ratio (F/B) was commonly used in GM analysis. 102 However, reviews have reported inconsistent changes in F/B ratio across a series of neurodegenerative diseases and metabolic disorders, making F/ B ratio a debatable and controversial criterion. 6,99,103,104 In agreement with our findings, one review summarizing the GM alterations in patients with PD found that, at high taxonomic ranks like phylum and class level, the changes in bacterial taxa are neither disease-specific nor consistent among different studies, but a more concordant trend was observed at family and genus level. 39 Additionally, α-diversity was thought to be a good indicator of health and diseases, and has been frequently investigated in GM analysis. 105 However, we found that neither AD studies nor SCRD studies showed concordant variation of GM α-diversity. And α-diversity analysis was not included in several studies. This is supported by another review which examines the association between GM and PD. They found that the confounding results of α-diversity alteration reported by different studies did not substantiate the role of α-diversity analysis as reliable methods for identifying PD and its progression, suggesting that higher α-diversity was not necessarily a predictor of better health. 6

Future studies need to focus more on metabolic and functional analysis
Most studies examining GM alterations in AD or SCRD only evaluated compositional changes of GM, and few conducted function-related analyses such as Kyoto Encyclopedia of Genes and Genomes (KEGG) test or metabolite screening. However, reviews have indicated that two taxonomically distinct bacterial taxa could share similar functions, while two closely related taxa may act antagonistically. 92,106 This suggests that phylogenetic analysis which is based on the hypervariable regions of bacterial 16s RNA gene cannot alone represent GM alterations at both taxonomic and functional level. It is possible that an increase of one genus could be neutralized or even reversed by a decrease of predominant genus in the same family. Thus, it would be confusing and misleading to simply conduct compositional analysis in discussing GM alterations. Moreover, metabolic and functional analysis have provided some important molecular and signaling pathways including possible interaction mechanisms between SCRD and GM and how GM dysbiosis could contribute to AD development. 10,28,30,33

Controversial roles of specific bacterial taxa
Lachnospiraceae and Akkermansia muciniphila, two taxa frequently investigated by the abovementioned studies, still remain controversial in their functions. As a core component of mammalian GM, Lachnospiraceae acts as a double-edged sword in health and disease. 92 On the one hand, several members of Lachnospiraceae like Blautia, Coprococcus and Roseburia are crucial producers of butyrate and acetate, which induce antiinflammatory responses, modulate insulin and lipid metabolism, and serve as the main nutrition source for colonic epithelial cells. [107][108][109] But on the other hand, other members, especially those capable of both producing propionate and degrading mucin, such as Dorea spp, Ruminococcus gnavus and Ruminococcus torques, have been associated with series of inflammation-related disorders and increased gut barrier permeability. 93,94 Unfortunately, the phylogenetic analyses in most studies were limited to the family level, possibly leading to the inconsistent data regarding the role of Lachnospiraceae in health and disease.
Akkermansia muciniphila (A. muciniphila) is another important SCFA-producer that utilizes mucin as carbon source. 110 However, reduced abundance of A. muciniphila has been associated with inflammatory bowel diseases and elevated inflammation. 85 Several reviews have also suggested A. muciniphila as a promising probiotic in treating metabolic disorders and modulating immune responses. 111,112 Different from other mucindegrading taxa, A. muciniphila was also found to promote mucin production, despite its ability to breakdown mucus layer. 113 Nevertheless, increased level of A. muciniphila was found in PD patients and some opposite effects have been reported. 6,85

Controlling variables in human studies
At compositional level, a weak connection of GM changes between human and animal studies can be established since human and murine harbor similar yet distinct microorganisms, although a shared trend of GM alterations was observed at functional level. However, compared to human, animal models exhibited more consistent GM alterations in both AD and SCRD studies. This discrepancy is mainly due to the limited studies available, heterogeneous samples and different methodologies applied in human studies.
In animal studies, mice and rats were born with identical genetic background, housed in constant environment and fed with unified food, and variables that could compromise the study have been carefully controlled as possible. Whereas in human studies, multiple factors including race, nationality, culture background and education may have substantial impacts on the lifestyle, daily diet and eating habit of participants, which directly affect GM composition. 114 For example, participants of the five AD patients studies we have discussed above were from three continents with diverse culture background. It has been reported that diet plays a fundamental role in health and is a key determinant of GM. 115,116 Western-style diet, high in animal protein, sugar and fat and low in vegetables, favors the growth of Bacteroidetes, especially Prevotella, which has been associated with colon cancer and several bowel diseases. 117 Mediterranean diet, featured by fruit, plant fiber and unsaturated fat, shifts GM toward more abundant Akkermansia, Bifidobacterium and Lactobacillus. 117 Also, food rich in dietary fiber and carbohydrates promotes the growth of highly fermentative bacteria such as Lachnospiraceae, Lactobacillaceae and Ruminococcaceae in the phylum Firmicutes. 92 Thus, the diverse dietary could contribute to the discrepant GM alterations in AD patients from different countries. Moreover, the varied experimental designs and heterogeneous methods, including fecal sample acquirement, DNA extraction and sequencing, as well as the criteria in determining cognitive function and sleep quality, make it difficult to conclude a consistent trend of GM alterations from different studies.
Therefore, it seems improper to compare GM alterations in human studies solely based on lowlevel phylogenetic analysis, which can be easily affected by the abovementioned factors. However, we observed a coherent trend by taking the perspective of metabolism and functions (Table 5, Figure 4).

Conclusion
Based on the evaluations from different studies on GM at both compositional and functional levels, this review suggests a possible link between SCRD and AD by GM. We propose that long-term SCRD may indirectly lead to chronic GM dysbiosis by altering eating habit, lifestyle, metabolism, etc. SCRD and GM dysbiosis could work synergistically to contribute to the onset and progression of AD ( Figure 5). However, the contribution of this alternative pathway in the development of AD remains unclear and requires further elucidation, since the etiology of sporadic AD varies from person to person. 118 Also, more studies are needed to further demonstrate the specific mechanisms of how SCRD leads to GM dysbiosis and how probiotic and antibiotic treatment ameliorate AD pathology, as well as the potential implications of pathobionts such as Erysiopelotrichaceae and Coriobacteriaceae in health and disease.

Disclosure statement
No potential conflict of interest was reported by the authors.