Alzheimer's disease (AD) is an age-related neurodegenerative disease characterized by slowly progressive memory decline and cognitive dysfunction, and no preventative or disease-modifying treatments are available for it at present (Gaugler et al., 2019). AD is a leading cause of dementia, and its prevalence is increasing drastically among aging populations worldwide. According to the World Alzheimer Report 2015, almost 47 million people worldwide were expected to be affected by dementia in 2015, with 9.9 million new diagnoses each year. This number is estimated to exceed 130 million by 2050, with the greatest increase expected in low- and middle-income countries. In 2017, 121,404 deaths caused by AD were recorded in USA, making it the sixth leading cause of death among Americans of all ages and the fifth leading cause among elderly Americans (age ≥65 years) (Gaugler et al., 2019). In addition, the economic costs of the disease are formidable due to the care needed by the growing number of patients with AD and other dementias. Thus, addressing the rapidly growing incidence of AD should be regarded as a global public health priority.

Numerous studies in recent decades have focused on elucidating the etiopathology of AD, but its pathogeneses remain unclear, and no therapeutic strategy is available to cure this disease. The most important risk-related factor for AD is advancing age, and as lifespans increase and demographic aging occurs worldwide, the number of AD patients is expected to increase drastically. The depositions of amyloid-beta peptide, a product of the cleavage of the amyloid-beta protein precursor, and the abnormal tau protein can be used as diagnostic markers for AD (Holtzman et al., 2011). However, whether they are the causes of AD or its consequences remains unknown. Many recent studies have noted that several infectious agents, such as Chlamydia pneumoniae, herpes simplex virus type 1, and several types of spirochaete and fungi, are involved in the pathogenesis of late-onset AD (De Chiara et al., 2012; Alonso et al., 2014; Balin and Hudson, 2014; Itzhaki, 2014; Miklossy, 2015; Pisa et al., 2015). This has prompted the suggestion that long-term, largely subclinical pathogenic infection might contribute to the characteristic neurodegeneration that occurs due to AD (Balin and Hudson, 2014). Based on these findings, Reis et al. (2010) considered AD itself as an infectious disease. Recent advances have revealed that microbiota of the human gut have numerous beneficial functions, such as immune development and resistance to pathogens, and serve as an important reservoir of pathogenic bacteria, viruses, and fungi. Previous work has clarified the role of gut microbiota in regulating multiple neuro-chemical pathways through the gut–brain axis (Bonfili et al., 2017). The dysbiosis of intestinal microbiota might impair intestinal mucosal integrity, increase intestinal permeability, and then disturb the intestinal homeostasis, where this can contribute to spreading potential pathogens to the target organs, such as the brain. Several studies have found altered gut microbiota in AD patients, which suggests that gut microbiota may be involved in AD pathogenesis (Vogt et al., 2017; Zhuang et al., 2018; Liu P. et al., 2019). Our group has observed that the transplantation of probiotics, prebiotics, and even fecal microbiota can ameliorate cognitive deficits and neurodegeneration in a model of mice with AD through the modulation of the gut microbiota (Sun et al., 2019a,b; Sun et al., 2020). Thus, the eubiosis of gut microbiota might have beneficial roles in preventing the occurrence and development of AD and other gut–brain disorders. However, the changing patterns of the composition and diversity of the gut microbiota are not always uniform in AD patients, but vary with population, geographical location, diet, and habits. The difference in the genetic background of hosts and dietary constitutions between Western and Chinese populations might contribute to the baseline disparity in the composition of the microbiota between them, which might in turn influence the roles of specific bacteria in the etiopathology of AD.

Lishui is a city that features dense mountains with a vegetation coverage of 80.79%, and has ranked second on China's ecological index behind Zhejiang for each of the past 13 years. It is also called the longevity town of China, with nearly 200 centenarians. The average life expectancy of the residents of Lishui is 80.06, 2.76 years higher than the national average. Its heredity, dietary patterns, and natural geographical environment influence the health of and incidence of diseases among the population, and may also influence the overall structure and function of the people's gut microbiota. The higher depth of sequencing and coverage with the advent of advanced sequencing techniques have made it possible to decipher key unknown functional taxa in Chinese AD patients. In this study, the fecal microbiota associated with AD are analyzed in a large AD cohort and matched healthy controls from Lishui by using the 16S rRNA high-throughput gene MiSeq platform, and are correlated with clinical indicators to provide novel targets for the early, non-invasive diagnosis and personalized treatment of AD as well as the development of tailor-made probiotics designed for Chinese AD patients.

In recent years, multi-omics techniques have revealed that the gut microbiota play a crucial role in promoting human health, as a result of which they are often referred to as the “forgotten organ.” Accumulating evidence indicates that the gut microbiota constitute a key factor in maintaining gut homeostasis by various complex mechanisms. Not only have the gut microbiota been invoked as a contributor to every gastrointestinal ailment, but the analyses of their influence have also been extended to other organs, such as the central nervous system (CNS). Exploring the roles and mechanisms of the gut microbiota in neurodegenerative diseases is an emerging field of research. Recently, several lines of research have suggested that changes in the composition and function of gut microbiota significantly affect neuronal function and, consequently, the host's behavior (Wang T. et al., 2015). The gut–brain axis of the microbiota has a proven role in regulating multiple neuro-chemical pathways. Microbiota–gut–brain axis signaling has uncovered a new era in psychiatry that is expected to provide novel targets for the diagnosis and treatment of psychiatric disorders and decipher their pathogeneses.

Aging is associated with an overstimulation of both innate and adaptive immune systems, resulting in a low-grade, chronic state of inflammation defined as inflammaging (Franceschi et al., 2000; Franceschi, 2007). This can increase gut permeability (“leaky gut”) and bacterial translocation (Ulluwishewa et al., 2011; Tran and Greenwood-Van Meerveld, 2013). As a major age-related neurodegenerative disorder, the onset of AD has been closely correlated with alterations in the gut microbiota (Vogt et al., 2017; Zhuang et al., 2018; Li et al., 2019; Liu P. et al., 2019; Wang et al., 2019). Most of these previous “AD microbiome” studies have been mainly conducted on a small scale of patients. In this study, 100 well-controlled Chinese AD patients and 71 age- and gender-matched normal controls were enrolled for AD microbiome analysis. In total, the deeper sequencing and higher coverage allowed us to identify low-abundance taxa in AD-associated fecal microbiota. In the results, structural changes in the fecal microbiota were evident in Chinese AD patients, with decreased alpha-diversity indices and altered beta-diversity ones. Inconsistent with previous clinical studies on AD patients, our study also indicated that Shannon significantly decreased in the AD patients (Liu P. et al., 2019), and such richness indices as the observed OTUs, ACE, and Chao1 were also significantly reduced (Vogt et al., 2017). Liu P. et al. (2019) have also shown significant compositional differences between AD patients and controls in PCoA plots based on Bray–Curtis dissimilarity, which were consistent with our study. Taken together, both alpha-diversity and beta-diversity indices provide powerful evidence of structurally dysbiotic AD microbiota in the AD patients.

In parallel with other similar studies, the structural dysbiosis observed in ours shows that the compositions of the AD-associated fecal microbiota also changed significantly. The distributions of bacterial taxa in the AD patients, at the levels of the phylum, family, and genus, were significantly different from those in healthy controls. An enrichment of the phyla Actinobacteria and Verrucomicrobia, and a decrease in the phylum Firmicutes were observed in AD-associated microbiota, while the amounts of such abundant phyla as Proteobacteria and Bacteroidetes did not change significantly, which is not consistent with previous studies (Zhuang et al., 2018; Liu P. et al., 2019). One aging indicator, the decreased ratios of Firmicutes/Bacteroidetes (Mariat et al., 2009), also did not change significantly in AD-associated microbiota. However, the ratios of Firmicutes/Actinobacteria were found to have decreased significantly (p < 0.05), which could reveal the bacterial dysbiosis in AD-associated microbiota. Several bacterial families—such abundant families as the Ruminococcaceae, Lachnospiraceae, and Clostridiaceae 1—decreased significantly in AD patients, while Bifidobacteriaceae, Verrucomicrobiaceae, Coriobacteriaceae, Erysipelotrichaceae, and Enterococcaceae increased significantly. Ruminococcaceae and Lachnospiraceae can produce different types of short-chain fatty acids (SCFAs). Among the SCFAs, butyrate has received particular attention in research owing to its beneficial effects on maintaining health. Butyrate can influence gastrointestinal physiology, the peripheral immunity of the liver metabolism, and the integrity of the blood–brain barrier, which can indirectly contribute to the functions of the brain (Fung et al., 2017). In addition, it can drive the maturation of the microglia, and is needed for the maintenance of mature microglia (Erny et al., 2015). However, the amounts of Bifidobacteriaceae, Verrucomicrobiaceae, Coriobacteriaceae, and Enterococcaceae, mainly lactate producers, increased in AD patients. Inconsistent with this study, Liu P. et al. (2019) found that the family Enterobacteriaceae is correlated with the presence and progression of AD, which can help distinguish between AD patients and healthy controls (AUC: 0.698). The changed fecal bacteria at the levels of phylum and family represent the dysbiosis of AD-associated fecal microbiota, but this is not suitable for a non-invasive diagnosis of AD using fecal bacteria at higher taxonomic levels.

In agreement with the altered bacteria at the family level, many such genera as the Bifidobacterium, Akkermansia, Faecalibacterium, Collinsella, and Roseburia and changed significantly in content the AD-associated fecal microbiota. Interestingly, traditionally beneficial bacteria, such as the Bifidobacterium and Akkermansia, increase in these AD patients while Faecalibacterium and Roseburia decrease significantly. Our ROC curves also show that these differential genera can be used as biomarkers to discriminate the controls from the AD patients, alone or together, which provides novel targets for a non-invasive diagnosis of AD. Bifidobacterium, mainly a lactate producer, is highly beneficial to humans, and has been used as a food supplement in dairy products (Camfield et al., 2011). An open-label, single-arm, preliminary clinical study conducted by Kobayashi et al. (2017, 2019) found that oral supplementation using Bifidobacterium breve A1 can improve the cognitive function and maintain the quality of life of the elderly by suppressing the gene expression of inflammation and immune-reactive genes. Different from previous studies on animals and clinical studies (Vogt et al., 2017), this traditionally beneficial genus was among the most abundant genera in the AD-associated fecal microbiota, which suggests that the Bifidobacterium genus may play a crucial role in the pathogenesis and development of AD. It is challenging to link species to 16S metagenomic data, but different species of Bifidobacterium may have different effects that can explain why Bifidobacterium spp. are commonly associated with healthy and diverse microbiota but sometimes also isolated in other conditions (Pineiro and Stanton, 2007). Thus, we needed to re-examine the therapeutic potential of Bifidobacterium in terms of maintaining cognitive function and treating dementia. Our data also showed that Bifidobacterium was significantly negatively correlated with the MMSE, WAIS, Barthel, and IL-8, which also shows that Bifidobacterium was not a beneficial genus in our clinical study. Akkermansia, a specialized mucin-degrading genus, can utilize mucin-derived sugars like fucose to produce propionate through the propanediol pathway (Ottman et al., 2017). Previous work has shown that Akkermansia muciniphila (typical strain) is associated with protection against obesity, enhancement of wound healing, augmented antitumor responses, and induced intestinal adaptive immune responses during homeostasis (Everard et al., 2013; Greer et al., 2016; Routy et al., 2018; Ansaldo et al., 2019). Combinations of Akkermansia, two strains of Clostridium, one strain of Eubacterium, one strain of Bifidobacterium, and inulin have recently been used as special synbiotics to treat type-2 diabetes mellitus. The discovery of Akkermansia muciniphila has opened new avenues for the use of this abundant intestinal symbiont in next-generation therapeutic products, and they can be used to target the dynamics of microbiota. Surprisingly, our data indicate that Akkermansia was among the most abundant genera in the AD-associated fecal microbiota. Similarly to Bifidobacterium, Akkermansia was negatively correlated with clinical indicators of AD, such as MMSE, WAIS, and Barthel, and anti-inflammatory cytokine such as IFN-γ. Based on our present observations, Akkermansia cannot always be considered a potentially beneficial bacterium, it might be harmful for the gut–brain axis in the context of the AD development in the elderly. Of these AD-enriched genera, Clostridium IV, Desulfovibrio, and Corynebacterium have been reported to be involved in the pathologic development of AD and other CNS diseases (Zhou et al., 2019), which is consistent with our findings. Clostridium IV was closely associated with type-2 diabetes and obesity in mice as well as the risk factors in AD development (Haan, 2006; Yamaguchi et al., 2016), which in turn is correlated negatively with the MMSE. Sawin et al. (2015) have also shown that Desulfovibrio can induce decreased levels of SCFAs that can influence pathologic conditions of CNS disease (Sampson et al., 2016). Previous studies have also found a decreased amount of Corynebacterium in patients suffering from depression and autism spectrum disorder (Strati et al., 2017; Yu et al., 2017), while Corynebacterium has been positively correlated with pro-inflammatory TNF-α. Different from Vogt's and Liu's studies, we found that such non-abundant genera as Collinsella, Enterococcus, Olsenella, Eubacterium, Christensenella, Anaerotruncus, and Halomonas were also enriched in AD-associated fecal microbiota (Vogt et al., 2017; Liu P. et al., 2019). Collinsella, one of the most abundant genera in the phylum Actinobacteria, was found to have increased prominently in the AD patients, but decreased in cases of relapsing–remitting multiple sclerosis (Chen et al., 2016), which was negatively correlated with MMSE, WAIS, and Barthel. Enterococcus (typical strain E. faecalis) can generate early Alzheimer-like neurofibrillary epitopes in primary rat cortical neurons (Underly et al., 2016), which can serve as harmful bacteria in AD etiopathology. Similarly to our findings, Christensenella was found to be increased in amounts in patients with Parkinson's disease in comparison with healthy controls (Petrov et al., 2017). These enriched AD-associated fecal genera, mainly lactate and propionate producers, may play a crucial role in the pathogenesis and development of AD.

However, reductions in the levels of Faecalibacterium, Roseburia, Clostridium sensu stricto, Gemmiger, Dialister, Romboutsia, Coprococcus, and Butyricicoccus were observed in AD-associated fecal microbiota. In particular, Biagi et al. (2016) and Wang F. et al. (2015) have shown that such butyrate producers as Faecalibacterium, Roseburia, and Coprococcus are negatively correlated with age. Our study found that these bacteria were positively correlated with AD clinical indicators, such as the MMSE, WAIS, and Barthel, and were negatively correlated with inflammatory cytokines, such as TNF-α and chemokines, such as IP-10. Faecalibacterium (typical strain F. prausnitzii), a major member of the Firmicutes phylum, is considered to be among the most important bacterial indicators of a healthy gut, and can modulate the inflammation of the level of the gut epithelium (Sokol et al., 2008). Beneficial Faecalibacterium has been found to be reduced in case of many intestinal disorders. Van Tongeren et al. (2005) observed a decreased relative abundance of Faecalibacterium in frail and elderly patients. In line with these findings, the decreased proportion of Faecalibacterium and increased Bifidobacterium have been found in elderly patients with Parkinson's disease (Scheperjans et al., 2015; Unger et al., 2016). All these changes may lead to a pro-inflammatory gut environment that may altogether lead to the chronic low-grade inflammation found in elderly persons with declining health. Previous studies have found that Faecalibacterium has anti-inflammatory properties due to its capability to produce butyrate and induce a tolerogenic cytokine profile (Sokol et al., 2008; Qiu et al., 2013), which can help extenuate these alterations in elderly AD patients. Liu J. et al. (2020) found that high-altitude Tibetan fermented milk can increase microbial diversity, and can elevate the levels of Bacteroides and Faecalibacterium in AD mice model, which are associated with cognitive improvements in mice afflicted with AD. The clinical comparative analyses and studies on animal mechanics confirm the beneficial roles of Faecalibacterium on mental health, which has prompted interest in considering this bacterium as a new-generation probiotic or psychobiotic. Gut Roseburia is part of commensal bacteria-producing SCFAs, especially butyrate, that affect immunity maintenance, colonic motility, and anti-inflammatory properties. The concomitant decreases in the well-known butyrate-producing bacterial genus Roseburia in many intestinal disorders (including type-2 diabetes, obesity, irritable bowel syndrome, nervous system conditions, and allergies), which suggests the potential of these bacteria as indicators of intestinal health (Tamanai-Shacoori et al., 2017). Consistently with our data, Keshavarzian et al. (2015) demonstrated the anti-inflammatory properties of Roseburia, and found that its levels are more abundant in feces of controls than in those of patients with Parkinson's disease. Neyrinck et al. (2011) found that the amount of Roseburia was inversely correlated with important markers of the metabolism and obesity of the host lipid. The role of Roseburia in protecting the nervous system from diseases has lately been highlighted, and has been shown to reduce neuroinflammation by regulating the gut–brain axis through its metabolite butyrate. Coprococcus, a less abundant bacterium in the large intestine, produces butyrate from fructose and propionate from lactate (via the acrylate pathway) (Reichardt et al., 2018). Together with Faecalibacterium, the butyrate-producing Coprococcus has been consistently associated with higher quality-of-life (QoL) indicators, which have been positively associated with several QoL scores (Valles-Colomer et al., 2019). Our previous study showed that Coprococcus is depleted in patients of depression (Jiang et al., 2015), even after correcting for the confounding effects of antidepressants (Valles-Colomer et al., 2019). Parashar and Udayabanu (2017) also found a reduction in fecal bacteria in the genus Coprococcus in patients with Parkinson's disease. Inconsistent with our microbiome study on AD patients here, Nagpal et al. (2019) found increased levels of Coprococcus in mild cognitive impairment participants in comparison with the controls. On the contrary, our study demonstrated that Coprococcus is positively correlated with clinical indicators of AD. Butyricicoccus, a butyrate-producing Clostridium cluster IV genus, was reduced in the feces of the AD patients. Butyricicoccus was found to be positively associated with the clinical indicators MMSE, WAIS, and Barthel, and anti-inflammatory cytokine IFN-γ, and negatively correlated with the pro-inflammatory cytokine TNF-α. Devriese et al. (2017) also found that reduced mucosa-associated Butyricicoccus activity in patients with ulcerative colitis was correlated with aberrant expressions of claudin-1, supporting its use as a pharmabiotic that preserves epithelial tight junction integrity. Zhang et al. (2017) also showed that the abundance of Butyricicoccus clearly decreases in a mouse model of AD in comparison with age-matched controls. Shen et al. (2019) observed that the regulation of gut microbiota by using silibinin and silymarin, especially with an increase in Butyricicoccus, might prohibit AD. Another Clostridium cluster I genus, Clostridium sensu stricto, was found to have decreased in AD patients (Vogt et al., 2017). Clostridium sensu stricto was positively associated with HDL and negatively associated with VLDL particles (Vojinovic et al., 2019), which are associated with a decreased risk of cardiovascular disease and stroke (Holmes et al., 2018). Gemmiger, also an SCFAs-producing genus, was positively related with the Montreal Cognitive Assessment scale score in patients with post-stroke cognitive impairment (Liu Y. et al., 2020). Our study also found that decreased levels of Gemmiger were positively associated with the MMSE, WAIS, and Barthel, and negatively correlated with inflammatory cytokines such as TNF-α and chemokines such as IP-10. Consistent with our study, Vogt et al. (2017) found that the genus Dialister (belonging to Veillonellaceae) was less abundant in AD participants, and exhibited the strongest correlations with such AD biomarkers of cerebrospinal fluid as Aβ42/Aβ40, p-tau, and p-tau/Aβ42. Therefore, all these decreased levels of AD-associated fecal genera, interacting with the AD-enriched genera, where this contributed to shifts in the SCFAs, might have participated in the pathogenesis and development of AD.

The level of endogenous SCFAs is influenced by many factors, of which gut bacterial metabolism is the most important. The dysbiosis of microbiota in patients of AD can change the balanced levels of SCFAs in the human body, while abnormal levels of SCFAs may negatively affect human health. Zhao et al. (2018) have claimed that the SCFAs serve as the bridge within this associations among diet, intestinal microbiota, and health. We observed that several metabolic pathways, such as those for carbohydrate metabolism, xenobiotics biodegradation and metabolism, and the immune system, changed significantly in AD-associated fecal microbiota. The characteristics of the AD microbial profiles changed from butyrate producers, such as Faecalibacterium into lactate producers, such as Bifidobacterium. These alterations contributed to shifts in metabolic pathways from butyrate to lactate, which might have participated in the pathogenesis of AD. However, the specific roles of the AD-associated signatures and their functions should be explored in further studies.

Our study is limited in some ways. First, it used the 16S rRNA amplicon rather than metagenomic sequencing, which limited us to the finding of specific bacteria related to AD at the species level. Second, our cross-sectional study investigated only healthy controls and confirmed AD participants. To decipher the dynamic interplay between microbiota and AD, a longitudinal follow-up case-control study should include different stages of AD, such as the preclinical stage and the mild cognitive impairment stage, that signify the transition from health to AD. Third, the fecal microbial signatures and the corresponding metabolites as well as the non-invasive diagnostic model associated with AD still need a larger sample size of clinical studies to be validated. Fourth, culturomics should be used to obtain the AD-associated bacteria, and animal experiments can help determine the cause–effect relationship between these bacteria and the pathogenesis of AD.

In summary, we found altered bacterial composition and decreased bacterial diversity of the fecal microbiota in AD patients compared with healthy elderly subjects. The structural dysbiosis of the fecal microbiota of the AD patients was characterized by reductions in butyrate-producing bacteria, such as Faecalibacterium, and increases in lactate-producing ones, such as Bifidobacterium, which were both significantly correlated with host pro- and anti-inflammatory cytokines as well as clinical indicators of AD in the host. These changes in key functional bacteria, such as the Faecalibacterium and Bifidobacterium, can be used as non-invasive biomarkers to discriminate between healthy elderly subjects and AD patients. Transformations of the gut microbiota from lactate producers into butyrate producers through personalized diet or intervention from beneficial microbiota may be useful for patient-tailored early intervention in cases of AD. In addition, the functional dysbiosis in AD-associated fecal microbiota also suggests that the changed fecal microbiota is associated with changed functions and metabolic activities of the patients, which might play vital roles in the pathogenesis and development of AD. Therefore, our investigation of fecal microbiota using a large and confirmed AD cohort provides novel insights into disease pathogenesis, which can provide new avenues for the scientific trajectory of managing neurodevelopmental disorders by modulating the gut microbiome.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

The studies involving human participants were reviewed and approved by the Ethics Committee of Lishui Second People's Hospital (Zhejiang, China). The patients/participants provided their written informed consent to participate in this study.

ZL, MZ, XY, and SW conceived and designed the experiments. ZL, MZ, XY, YC, LS, XL, and RJ performed the experiments. ZL, MZ, XY, and LS analyzed the data. ZL, MZ, LS, and SW wrote the paper and edited the manuscript. The final manuscript was read and approved by all authors.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.