Changes in the Gut Microbiome Associated with Intussusception in Patients with Peutz-Jeghers Syndrome

Recent research has established a link between the gut microbiome and polyps and neoplasia, and antibiotic use influences the microbiome and the development of colorectal polyps. Familial adenomatous polyposis (FAP), which is characterized by the early development of benign precursor lesions (polyps), is associated with enterotoxigenic Bacteroides fragilis and Escherichia coli biofilms. ABSTRACT Peutz-Jeghers syndrome (PJS) is a rare hereditary disorder characterized by intestinal polyposis, and intestinal intussusception is one of the most urgent complications. While it is known that imbalance of the gut microbiota is highly associated with intestinal disorders, the role of the gut microbiome in the pathogenesis of PJS has not been reported. In this study, we performed 16S rRNA sequencing on stools from 168 patients and 68 healthy family members who lived together to determine the gut microbiome composition of PJS patients. Metagenomics sequencing was further performed on the representative samples (61 PJS patients and 27 healthy family members) to analyze the functional changes. We found that the fecal microbiome of patients with PJS showed a greater variation in β-diversity. An enhancement of Escherichia coli and a reduction of Faecalibacterium prausnitzii was identified in PJS patients. Further reduction of Faecalibacterium prausnitzii was the characteristic microbial change observed in patients with intussusception. Functional analysis revealed that the abundance of propanoate metabolism was enriched in PJS patients and further enriched in those with intussusception. Escherichia coli was the major contributor to the enrichment of this metabolism pathway, which was associated with the abnormal expression of methylglyoxal synthase (encoded by mgsA) and phosphate acetyltransferase (encoded by pta). Our findings showed a distinct gut microbiome signature in PJS patients and identified the connection between the gut microbiome and intussusception. Alterations in the gut microbiome might be involved in the pathogenesis of PJS and may serve as biomarkers for gastrointestinal surveillance. IMPORTANCE Recent research has established a link between the gut microbiome and polyps and neoplasia, and antibiotic use influences the microbiome and the development of colorectal polyps. Familial adenomatous polyposis (FAP), which is characterized by the early development of benign precursor lesions (polyps), is associated with enterotoxigenic Bacteroides fragilis and Escherichia coli biofilms. However, the relationship between the gut microbiome and the pathophysiology of PJS has not yet been established. In this study, we found that PJS patients had a distinct microbiome composition, with a greater variation in β-diversity, an increase in Escherichia coli, and a decrease in Faecalibacterium prausnitzii. A further reduction of Faecalibacterium prausnitzii was observed in patients with intussusception. Moreover, PJS involved increased propanoate metabolism as well as abnormal mgsA and pta expression. These findings may contribute to a better understanding of the etiology of PJS and improve disease control strategies.

P eutz-Jeghers syndrome (PJS) is a rare hereditary intestinal polyposis condition with an estimated incidence of approximately 1 in 50,000 to 1 in 100,000 (1). Germ line mutations in the serine/threonine kinase 11 (STK11) tumor suppressor gene was found to be associated with PJS (2). Mucocutaneous melanin pigmentation and gastrointestinal hamartomatous polyps are two major essential clinical features of PJS. In particular, hamartomatous polyps associated with PJS can be found throughout the gastrointestinal tract, predominantly in the small intestine and the colon (3). Even at a young age, the presence of pathological hamartomatous polyps in PJS patients may lead to diverse gastrointestinal complications, including hemorrhage, anemia, abdominal pain, intussusception, obstruction, and infarction (4). Among them, small intestinal intussusception is the most urgent and even lethal gastrointestinal complication and produces a substantial impact on the clinical prognosis of PJS patients (5). Over 90% patients require surgical treatment, mainly laparotomy, when presenting with intussusception (6). Gastrointestinal surveillance by regular gastrointestinal endoscopy is an imperative management strategy to limit the risk of emergency laparotomy in patients affected by PJS (7)(8)(9). However, gastrointestinal endoscopy is an invasive procedure which carries a risk for complications. Therefore, it is necessary to find a noninvasive, safe, and reliable approach to monitor disease progression in PJS patients. STK11, also known as live kinase B1 (LKB1), is detected to be mutated in PJS and a wide variety of tumors (10,11). LKB1 is primarily dependent on the phosphorylation and activation of the AMP-activated protein kinase (AMPK) for its actions. As a vital metabolic regulator, AMPK modulates cellular metabolism by regulating metabolic enzyme activities and activating adaptive transcriptional responses (12,13), which might be involved in the pathogenesis of PJS.
Emerging evidence suggests that the gut microbiome is highly linked with the development of a wide spectrum of gastrointestinal disorders, especially polyposis and neoplasia (14)(15)(16). A case-control study conducted in Sweden recently reported that usage of broadspectrum antibiotics, such as tetracyclines, b-lactam antibacterials, and sulfonamides, could affect the composition of the intestinal microbiome and was related to an increased risk of colorectal polyps (17). The pathogenesis of familial adenomatous polyposis (FAP), a disorder characterized by mutation in the adenomatous polyposis coli (APC) gene, was found to be associated with bacterial biofilms constituted primarily of Escherichia coli and enterotoxigenic Bacteroides fragilis (18). Patients with adenomas have greater abundances of proinflammatory genera, which may produce genotoxic or inflammatory metabolites such as H 2 S and secondary bile acids, thus promoting adenoma development and tumorigenesis (19). Compared to serrated polyp, conventional adenoma displayed more severe gut microbiota dysbiosis characterized by a reduction of commensal bacteria and an enhancement of potentially harmful bacteria (20). In addition, significant alteration in microbiome and metabolome was also observed in multiple polypoid adenomas and intramucosal carcinomas, which were considered the early stages of colorectal cancer (CRC) (21). Therefore, it is probable that the gut microbiome influences the development of PJS and provides information for prognostic evaluation. So far, however, the influence of the intestinal microbiome in the pathophysiology of PJS has received little attention.
To expand upon the understanding of the modification of gut microbiome composition and the link between the gut microbiome and intussusception in PJS, 16S rRNA gene sequencing was performed on a Chinese family cohort containing 168 PJS patients and 68 healthy family members. Metagenomic sequencing of the gut microbiota was further performed for 61 PJS patients and 27 healthy family members to investigate the functional changes associated with intussusception. Aware that AMPK signaling pathway was associated with the pathogenesis of PJS, we analyzed specific species contributions to this pathway. Alterations of gut microbiota profiles in PJS patients compared with healthy family members based on the 16S rRNA data. To obtain an overview of the microbiome profiles in PJS patients, we applied 16S rRNA gene sequencing, and an average of 64,040 optimized reads per individual (range, 45,242 to 129,374) were obtained. An adequate sequencing depth was assessed according to rarefaction curves generated from the observed operational taxonomic units (OTUs) (see Fig. S1A in the supplemental material). A Venn diagram shows 121 unique genera in PJS patients and 87 unique genera in healthy family controls, while 354 genera were shared by the two groups ( Fig. 1A). At the community level, the Simpson index revealed no significant differences in a-diversity between the two groups (P = 0.056) (Fig. 1B). To evaluate the similarity of microbial community (b-diversity) between the two groups, principal-coordinate analysis (PCoA) was performed based on unweighted (Fig. 1C) and weighted (Fig. 1D) UniFrac distances to visualize the differences in bacterial communities. The analysis demonstrated that the two groups clustered separately in terms of fecal microbiota composition and that there was greater person-to-person microbiome variability among PJS patients.

RESULTS
The dominant taxa at the phylum level included Firmicutes, Bacteroidetes, Proteobacteria, and Actinobacteria (Fig. S1B). The abundances of Proteobacteria, Fusobacteria, Verrucomicrobia, Cyanobacteria, and Chlamydia were higher in the PJS group (P , 0.001), whereas the abundance of Firmicutes was higher in the control group (P = 0.01). Furthermore, the dominant taxa at the genus level included Faecalibacterium, Bacteroides, Escherichia-Shigella, Agathobacter, and Prevotella_9 (Fig. S1C). Genus-level comparisons revealed that 29 genera were significantly enriched in the PJS group, while 63 genera were significantly enriched in the control group (Table S1). A Sankey diagram shows the major proportions of dominant taxa (phyla and genera) in PJS patients and healthy family controls (Fig. 1E).
To characterize the interaction patterns of differentiated microbial taxa, we next constructed co-occurrence networks of top 30 genera from genus-level comparisons between the two groups (Table S1) based on significant Spearman correlations. Notably, microbial community in the PJS group displayed a co-occurrence network with a more complicated correlation among genera. Faecalibacterium, Subdoligranulum, Escherichia-Shigella, Veillonella, and Enterobacter played a crucial role in the microbial community of PJS patients (Fig. 1H). Overall, these results demonstrated that microbial composition and correlation network in the PJS group displayed alteration compared with healthy controls, further indicating the presence of gut microbiota dysbiosis in PJS patients.
Association of microbiome composition with intussusception. Intussusception is the severe clinical outcome of PJS patients which is characterized by polyps throughout the gastrointestinal tract. To identify the effect of intussusception on the gut microbiome, we further compared the gut microbiome composition of 106 PJS patients who had undergone at least one intussusception (I-PJS) with that of 57 PJS patients who did not experience intussusception (NI-PJS). The remaining 5 PJS patients did not provide information related to the experience of intussusception and were therefore not included in this subgroup analysis. The microbial communities did not show significant differences for the a-diversity indices among PJS without and with intussusception ( Fig. 2A). Principal-component analysis (PCA) revealed that the microbiome profile of the I-PJS group was distinct from that of the NI-PJS group (P = 0.019) (Fig. 2B).
The composition of the microbiome at the phylum level further showed differences between the NI-PJS group and the I-PJS group (Fig. S2A). The proportions of Proteobacteria, Fusobacteria, and Synergistetes were increased in I-PJS patients compared to the NI-PJS group (Fig. S2B). Compositional change of gut microbiota in I-PJS patients is also presented at the genus level (Fig. S2C). Genus-level comparisons revealed that 11 genera were substantially elevated in the I-PJS group, while 17 genera were substantially elevated in the NI-PJS group, exhibiting a similar pattern of alteration between PJS patients and healthy controls (Fig. S2D). For instance, the increased Escherichia-Shigella and the decreased Faecalibacterium in I-PJS compared to NI-PJS also displayed the same trend in the PJS patients compared to healthy controls. In Fig. 2C, the proportional abundances of dominant taxa in I-PJS and NI-PJS patients at the phylum and genus levels are depicted in a Sankey diagram. LEfSe analysis and a cladogram (shown in Fig. 2D and E) further uncovered additional hallmark microbial profiles and main bacterial biomarkers of I-PJS patients as indicated by LDA (LDA score . 3). At the phylum level, the abundance of Proteobacteria and Fusobacteria was enriched in I-PJS patient samples compared to that in the NI-PJS group. At the family level, I-PJS enriched taxa consisted of Enterobacteriaceae and Fusobacteriaceae, whereas Bifidobacteriaceae, Peptostreptococcaceae, and Burkholderiaceae were prevalent in the NI-PJS group. In addition, genera such as Escherichia-Shigella, Fusobacterium, Blautia, and Klebsiella were substantially more abundant in the I-PJS group than in the NI-PJS group. In contrast, Faecalibacterium, Bifidobacterium, and Romboutsia were significantly less abundant in I-PJS patient samples than in NI-PJS samples ( Fig. 2D and E). To investigate the prediction potential of the intestinal microbiome in distinguishing PJS with and without intussusception, the random forest model was constructed on the genus abundance data set (Fig. S2E). The combination of these 20 genera achieved good diagnostic performance, with an area under the curve (AUC) of 0.79 (Fig. 2F). Surgery is the mainstay of therapy in PJS patients with intussusception. Among the 106 PJS patients who experienced at least one intussusception, 9 (8%) did not undergo surgery, 33 (31%) had a single surgery, 31 (29%) had two surgeries, and 33 (31%) had more than two surgeries. When classified by the number of surgeries (0 [t0] versus 1 versus 2 versus $3), the Shannon index of the t3 subgroup was decreased compared to those of the t1 and t2 subgroups (Fig. 2G). Moreover, different microbial profiles in these subgroups were identified at the genus level. The Eubacterium eligens group and Intestinibacter were markedly decreased in patients that had undergone surgery, while Dorea and Lachnospiraceae_unclassified were increased in patients that had undergone surgery (Fig. 2H).
Based on the above-described analysis, the compositions of gut microbiotas across PJS patients with and without intussusception showed statistical differences, which may serve as significant predictors of intussusception, a critical complication in PJS patients. Metagenomic data exhibited significant differences between PJS patients and healthy controls. To further determine the relationship between gut microbial alterations at the species level and functions of the gut microbiome in PJS patients, shotgun metagenomic sequencing was performed on fecal samples from 61 PJS patients (35 with intussusception and 26 without intussusception) and 27 healthy family members. A total of 22,311,093 genes were evaluated, and 2,514,259 genes were grouped into a nonredundant gene catalogue for further study (Fig. S3A). A Venn diagram shows 3,459 unique species in the PJS group and 817 unique species in the control group, while 11,143 species were shared by the two groups (Fig. 3A). PCoA, based upon Bray-Curtis distance at the species level, showed significant dissimilarity in bacterial communities between the PJS and control groups (P = 0.002) (Fig. 3B). The composition of the microbial community also revealed striking variations at the species level (Fig. 3C). LEfSe analysis was further conducted to explore relative taxon abundances between two groups, which were characterized by significant difference (LDA score . 2). Twenty-one species were enriched in the PJS group, among which Escherichia coli was the most abundant bacterial species, followed by unclassified_f _Enterobacteriaceae and Klebsiella pneumoniae. Moreover, 26 species were enriched in the control group, among which Faecalibacterium prausnitzii was the most abundant bacterial species, followed by Bifidobacterium longum and unclassified_o_Clostridiales (Fig. 3D). Consistent with the 16S rRNA analysis, the majority of the differentially identified species in the PJS group belonged to the genera Klebsiella and Escherichia-Shigella as well as the family  Fig. 3E. Differential abundance analysis further revealed significant differences of 995 species between the two groups (Wilcoxon rank sum test [Table S2]). The top 20 differential species are shown in Fig. 3F.

Enterobacteriaceae. A community heatmap of the top 30 species is shown in
Organisms were classified according to the absence or presence of intussusception in the PJS patients. A Venn diagram shows that there were 1,433 unique species in the NI-PJS group and 2,528 unique species in the I-PJS group, while 10,641 species were shared by the two groups (Fig. 4A). PCoA did not reveal a significantly separate clustering in microbiota structures between groups (Fig. 4B). Species-level comparison of the gut microbiomes showed significant differences in taxonomy (Fig. 4C). LEfSe analysis was used to screen the differential microbes among species (LDA score . 2). Seven species were enriched in the I-PJS group, among which Collinsella sp. strain AF08 23, Fusobacterium, and Aeromonas veronii were the dominant species. Moreover, 10 species were enriched in the NI-PJS group, among which Faecalibacterium prausnitzii, unclassified_g_Faecalibacterium, and Faecalibacterium were the dominant species (Fig. 4D). To better understand the differences between two groups, a heat map was constructed of the top 30 species (Fig. 4E). Furthermore, differential abundance analysis showed differences in a total of 459 species between the two groups (Wilcoxon rank sum test [Table S3]). The top 20 differential species are shown in Fig. 4F. In particular, the read number of Faecalibacterium prausnitzii was significantly reduced in PJS patients (P = 0.0358) and further reduced in patients with intussusception (P = 0.0129 [ Fig. 4G]). Taken together, these analyses indicate that these differential species may play a role in the pathophysiology of PJS.
Functional profiling of metagenomic data identified disrupted bacterial functions in PJS patients. In order to gain more insights into the gut microbiome dysbiosis of patients with PJS, we next performed functional analysis of metagenomic sequencing to determine differential genes and pathways between the two groups in our study cohort. A Venn diagram displayed 101 unique KEGG orthologous (KO) in the control group and 337 unique KO in the PJS group, while 6,484 KO were shared by the two groups (Fig. 5A). PCoA, based on KO, demonstrated clustering differences in bacterial functions between two groups (Fig. 5B). Meanwhile, KEGG pathways based on level 2 were disrupted in PJS patients compared to the healthy controls. For instance, the pathways involved in signal transduction, xenobiotics biodegradation and cellular community (prokaryotes) were enriched in the PJS group, while the pathways involved in global and overview maps, biosynthesis of other secondary metabolites, and cell growth and death were enriched in the control group (Fig. 5C). LEfSe analysis further revealed that the three most abundant KEGG pathways (level 3) in the PJS group were two-component system, microbial metabolism in diverse environments, and lipopolysaccharide biosynthesis, while the three most abundant KEGG pathways (level 3) in the control group were biosynthesis of amino acids, biosynthesis of secondary metabolites, and starch and sucrose metabolism (LDA score . 2 [ Fig. 5D]). Furthermore, differential abundance analysis showed differences in a total of 134 pathways (level 3) between the two groups (Wilcoxon rank sum test [Table S4]). The top 20 differential pathways (level 3) are shown in Fig. 5E.
The association of microbiome gene pathways with intussusception was further analyzed. A Venn diagram shows that there were 247 unique KO in the I-PJS group and 128 unique KO in the NI-PJS group, while 6,446 KO were shared by the two groups (Fig. 6A). LEfSe analysis was used to screen the differential KEGG pathways (level 3; LDA score . 2). Five pathways (level 3) were enriched in I-PJS group, among which propanoate metabolism was the most dominant pathway. Moreover, 3 pathways (level 3) were enriched in the NI-PJS group, among which biosynthesis of secondary metabolites was the most dominant pathway (Fig. 6B). Furthermore, differential abundance analysis showed differences in a total of 25 pathways (level 3) between the two groups (Wilcoxon rank sum test [Table S5]). The top 20 differential pathways (level 3) are shown in Fig. 6C. In particular, read number for propanoate metabolism was significantly enriched in PJS patients (P = 0.0243) and further enriched in patients with intussusception (P = 0.0233 [ Fig. 6D and Fig. S4]). Furthermore, we investigated whether specific bacterial species contribute to the propanoate metabolism pathway in our cohort. Escherichia coli was major contributor to the enrichment of the KEGG pathway involved in propanoate metabolism (PJS versus control and I-PJS versus NI-PJS [ Fig. 6E and F]). We further measured the abundance of enzymes involved in the propanoate Given the important role of AMPK signaling pathway in the pathogenesis of PJS (22), we further determined the differential abundance of AMPK signaling pathways between the PJS and control groups. We found that AMPK signaling pathway increased in the PJS group relative to the control group (Wilcoxon rank sum test [ Fig.  S7A]), and Shigella flexneri contributed predominantly to the AMPK signaling pathway in PJS patients (Fig. S7B). Overall, these findings indicated that PJS patients displayed dysfunction in the KEGG pathway, which may play a part in the development of PJS and the occurrence of intussusception.

DISCUSSION
In the present study, we first discovered that the microbial composition and functional profiling were significantly different between PJS patients and healthy family controls and revealed that alterations in gut microbiome are closely linked with intussusception in PJS patients. Recent research indicates that the gut microbiome contributes to the maintenance of intestinal health and to the etiology of gastrointestinal polyposis and neoplasia. Consequently, there is a need to better identify specific modifications associated with common observations in clinical practice. PJS is a rare autosomal dominant condition distinguished by mucocutaneous melanin pigmentation and gastrointestinal hamartomatous polyps. We first studied whether the intestinal microbiome was involved in the pathogenesis of PJS. Although the decrease in a-diversity was not statistically significant, the community-level change associated with PJS showed significant differences from the case for healthy family members. Greater person-to-person variability in microbial composition (b-diversity) was also observed in PJS patients. This indicates that even though PJS patients share the same lifestyle, the same dietary habits, and a similar genetic code with their healthy family members, gut microbiome dysbiosis exists in patients with PJS.
PJS patients were enriched in potentially harmful bacteria in their gut, such as pathogenic bacteria and mucin-degrading bacteria (Escherichia-Shigella, Klebsiella, Fusobacterium, Veillonella, and Akkermansia), but depleted of commensal bacteria, such as butyrate-producing bacteria and lactate-producing bacteria (Faecalibacterium, Subdoligranulum, and Lactobacillus), at the genus level. These findings were rather in line with those recently reported for CRC patients, indicating that pathogenic bacteria (Escherichia-Shigella, Klebsiella, and Fusobacterium) and mucin-degrading bacteria (Akkermansia) were more abundant in the gut microbiota of CRC patients and that butyrate-producing bacteria (Lachnospiraceae) were less abundant (23)(24)(25). At the species level, LEfSe analysis showed that a total of 21 species were increased in PJS patients, most of them belonging to the family Enterobacteriaceae and the genera Escherichia-Shigella and Klebsiella. In particular, Escherichia coli, belonging to the family Enterobacteriaceae, was the most abundant bacterial species in PJS patients. An overgrowth of Escherichia coli, considered a tumorigenic bacterium, has been reported to be related to the pathogenesis of CRC (26)(27)(28). Additionally, we found that Faecalibacterium prausnitzii showed the most obvious decline in PJS patients, which is consistent with previous findings for adenomatous polyps and CRC (29)(30)(31). These results indicate that the structures of the gut microbiome were considerably different between PJS patients and healthy family controls.
It is worth noting that the microbiome and intestinal disorders have intricate interactions. On the one hand, the progression of intestinal disorders modifies microbial profiles; on the other hand, microbiome shifts in certain regions might directly or indirectly induce intestinal disorders. In CRC, gene mutations, especially loss-of-function mutations, disrupt the composition of the gut microbiome (32). Gene mutations can affect the entire microbial network, not just individuals. Additionally, intestinal barrier dysfunction, a prevalent hallmark in gastrointestinal diseases, has been shown to alter the balance between host and microbe in a Drosophila tumor model (33). Moreover, microbiotas can thrive and promote immunological and molecular alterations in the microenvironment of a specific disease. For instance, the increased abundance of Fusobacterium nucleatum in CRC influences colorectal carcinogenesis through the production of adhesins and lipopolysaccharide (34). Overall, the microbial community may be impacted by genetic abnormalities and intestinal barrier dysfunction in PJS. And the altered microbiota may contribute to the development of PJS through multiple mechanisms.
Intussusception is the most urgent gastrointestinal complication of PJS patients, and regular gastrointestinal surveillance is of great importance. We further analyzed the gut microbiota composition of PJS patients with or without intussusception. Although the differences in a-diversity and b-diversity were not statistically significant, the dominant taxa between two groups showed the same trend in the PJS patients relative to healthy family members. We found that Escherichia-Shigella, Fusobacterium, and Klebsiella were also enriched in PJS patients with intussusception (I-PJS), while Faecalibacterium was less abundant. At the species level, LEfSe analysis showed that 10 species were increased in PJS patients without intussusception (NI-PJS), most of which belonged to the genus Faecalibacterium, and Faecalibacterium prausnitzii was the most dominant species. Characteristic reduction of Faecalibacterium prausnitzii abundance was observed in either PJS patients relative to healthy family controls or I-PJS patients relative to NI-PJS patients.
Faecalibacterium prausnitzii, one of the most prevalent bacteria in the human intestine, promotes the health of the gastrointestinal tract by strengthening the intestinal epithelial barrier function, exerting anti-inflammatory activity, and regulating the mucus pathway (35). Faecalibacterium prausnitzii is extremely responsive to fluctuations in the intestinal microenvironment, and variations in the population richness and quantity of this species have been found in a range of digestive system illnesses (36,37). Low fecal counts of Faecalibacterium prausnitzii were connected with the progression of hepatitis B virus-related acute-on-chronic liver failure (HBV-ACLF), while the recovery of this species was associated with the remission of HBV-ACLF (38). In addition, a phase III randomized controlled trial recently reported that the abundance Faecalibacterium prausnitzii markedly rebounded after treatment with ursodeoxycholic acid, a therapeutic bile acid, in patients with colorectal adenoma (39). Therefore, Faecalibacterium prausnitzii may serve as a biomarker to assist in the diagnosis and prognosis of intestinal diseases. Overall, these findings imply that the decline in the abundance of Faecalibacterium prausnitzii is connected with the pathogenesis and progression of PJS. Monitoring the dynamic changes of the gut microbiome is expected to become a noninvasive and straightforward method for gastrointestinal surveillance in PJS patients, providing a reference for the prediction of intestinal intussusception.
In addition to taxonomic differences, we found a carbohydrate metabolism pathway involving genes encoding propanoate metabolism that was more abundant in PJS patients than in healthy family controls, and further enhanced in I-PJS patients relative to NI-PJS patients. Propanoate metabolism is downstream of lipid metabolism. Propanoate transforms into succinyl coenzyme A (succinyl-CoA), which then enters the tricarboxylic acid (TCA) cycle in order to generate energy or promote lipogenesis (40). The enhancement of propanoate metabolism may help to maintain the complex communication network among the microbial community in PJS patients via providing more energy. Abnormal propanoate metabolism was reported to be associated with wide range of disorders, such as intrahepatic cholestasis of pregnancy (41), lower-grade glioma (42), colorectal cancer (43), and pancreatic carcinoma (44). Propanoate metabolism can be utilized as a carbon source for fatty acid production and can stimulate tumor growth (44). Moreover, phosphatidylinositol 3-kinase (PI3K) selective molecule repressed tumor angiogenesis and CRC growth via the downregulation of propanoate metabolism and lipogenesis (45). Therefore, the further enhancement of propanoate metabolism in I-PJS patients relative to NI-PJS patients may be associated with a poor prognosis.
Multiple microorganisms were discovered to make a significant contribution to carbohydrate metabolism pathways. In particular, the species Escherichia coli contributed considerably more to the propanoate metabolism in PJS patients and in those with intussusception. Escherichia coli harboring the genomic island polyketide synthase (pks 1 ) was an essential component of bacterial biofilms found in the intestinal mucosae of patients with familial adenomatous polyposis (FAP) (18). The genotoxin colibactin, produced by pks 1 Escherichia coli, may enhance colorectal tumorigenesis by inducing possibly mutagenic DNA damage (46). Otherwise, the most prevalent virulence genes of Escherichia coli isolated from colonic adenomatous polyps were fimH and fyuA (47). In order to adapt to the local environmental changes in colonic adenomatous polyps, Escherichia coli expressed specific phenotypic traits, resulting in the improved biofilm formation and poor proteolytic activity (48). The identification of genotypic and phenotypic traits of Escherichia coli may improve the understanding of the molecular mechanisms underlying PJS pathogenesis.
We also found that genes encoding two key enzymes in propanoate metabolism pathway were decreased in PJS patients and further decreased in patients with intussusception. Methylglyoxal synthase (EC 4.2.3.3; encoded by mgsA) is a crucial glycolysis bypass protein which produces methylglyoxal from the glycolytic intermediate dihydroxyacetone phosphate (49). A high concentration of methylglyoxal is toxic to cell growth (50). Phosphate acetyltransferase (EC 2.3.1.8; encoded by pta) catalyzes the reversible transition of the acetyl group from acetyl-P to CoA, producing acetyl-CoA and inorganic phosphate and contributing to the acetate assimilation or dissimilation process (51). The abnormal expression of mgsA and pta might be a crucial link in the pathogenesis and progression of PJS. However, the underlying mechanism requires further exploration.
Several limitations in our study should be noted. First, the difference in the ages of the PJS patients and their family members may have an impact on the results of the microbiotas. Second, geographical differences and dietary differences among the various families may affect the results of the microbiotas. Third, the present study was a cross-sectional study which allowed for association but could not address causality. Therefore, a well-designed prospective study and translational studies in animal models should be conducted to confirm our findings.
In summary, the present study demonstrated that PJS patients revealed a unique microbiome composition, characterized by a greater variation in b-diversity, an enhancement of the species Escherichia coli, and a reduction of the species Faecalibacterium prausnitzii. Further reduction of Faecalibacterium prausnitzii was observed in I-PJS patients relative to NI-PJS patients. Functional analysis of the intestinal microbiome identified a greater abundance of propanoate metabolism and abnormal expression of mgsA and pta in PJS. These findings may provide insight into our understanding of the molecular mechanisms underlying PJS pathogenesis.

MATERIALS AND METHODS
Study cohorts. This was an observational and cross-sectional clinical study of a Chinese family cohort containing 168 PJS patients and 68 healthy family members. Case participants were patients with PJS diagnosed according to the diagnostic criteria as defined by the World Health Organization (2) and/or they were identified as being STK11 gene mutation carriers. Control participants were healthy family members of the selected patients who had lived with the patients for at least 1 year. Demographic and clinical information was obtained from each participant via interviews and questionnaires. The following data were obtained: age, gender, body mass index (BMI), region, gastrointestinal symptoms, mucocutaneous pigmented lesions, gastrointestinal polyps, intussusception or obstruction, history of surgery, family history of PJS, diagnosis, and type of tumor. Individuals who took antibiotics within 3 months prior to recruitment were excluded. Adult participants provided informed consent; minors provided assent and informed consent was obtained from their parents or guardians prior to fecal sample collection. Ethics approval was obtained from the Ethics Committee of Southern Medical University Nanfang Hospital.
Fecal sample collection and DNA extraction. Fecal samples were self-collected at home by participants according to the manufacturer's instructions, delivered immediately to the laboratory at low temperatures, and then stored at -80°C until DNA extraction. DNA extractions from fecal samples were carried out using the E.Z.N.A. stool DNA kit (Omega Bio-tek, Norcross, GA, USA). A NanoDrop 2000 instrument (Thermo Scientific, Wilmington, NC, USA) was used to evaluate the concentration and purity of extracted DNA.
16S rRNA gene amplicon sequencing. PCR amplification was performed using primers (forward, CCTACGGGNGGCWGCAG; reverse, GACTACHVGGGTATCTAATCC) (52) that targeted the V3-V4 region of the bacterial 16S rRNA gene. PCR mixtures (25 mL) contained 12.5 mL of PCR Premix, 5 mL of primer mix (2.5 mL of each primer), 25 ng of template DNA, and RNase-free double-distilled water (ddH 2 O) to adjust the volume. The following cycling was used: initial denaturation at 98°C for 30 s and then 32 cycles of denaturation at 98°C for 10 s, annealing at 54°C for 30 s, and extension at 72°C for 45 s, followed by final extension at 72°C for 10 min. The libraries were sequenced on the Illumina NovaSeq platform according to the manufacturer's recommendations. Raw reads were demultiplexed, quality filtered by Vsearch (v2.3.4), and merged by FLASH (v1.2.7). aand b-diversities were calculated with QIIME2, and graphs were drawn with R package. Other charts were generated using R package (v3.5.2).
Gene prediction, taxonomy, and functional annotation. MetaGene was used to identify the open reading frames (ORFs) in the assembled contigs (57). The predicted ORFs with lengths equal to or more than 100 bp were translated into the corresponding amino acid sequences. CD-HIT software (v4.6.1) was used to remove the redundant sequences (95% identity; 90% overlap) and construct a nonredundant gene catalogue (58) (http://www.bioinformatics.org/cd-hit/). Taxonomic annotations of nonredundant gene catalogue against the entire NCBI NR database were performed using Diamond v0.8.35 (http:// github.com/bbuchfink/diamond) (59). The KEGG annotation was conducted using Diamond v0.8.35 against the Kyoto Encyclopedia of Genes and Genomes database (v94.2).
Bioinformatics and statistical analysis. Co-occurrence network analysis and random forest model was conducted using the OmicStudio tools (https://www.omicstudio.cn/tool). Receiver operating characteristic (ROC) analysis was performed using the OECloud tools available at https://cloud.oebiotech.cn. Linear discriminant analysis (LDA) effect size (LEfSe) was used to discover differentially abundant taxa or Gut Microbiome Signatures in PJS Microbiology Spectrum KEGG pathways represented between any two groups. Wilcoxon rank sum test was performed to determine whether there was a significant difference in the abundance of taxa and KEGG pathways between any two groups. Differences were considered statistically significant when the false-discovery rate (FDR)corrected P value was ,0.05. Data availability. The raw sequence data have been deposited in the SRA database under accession number PRJNA905444.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 1.3 MB.