Method for analyzing characteristics of intestinal flora of atopic dermatitis based on metagenomics and screening biomarkers

Abstract

The application is suitable for the field of biomedical and bioinformatics technology, and provides an atopic dermatitis intestinal flora feature analysis and biomarker screening method based on metagenomics, comprising: obtaining fecal sample metagenomic sequencing data; performing sequence assembly and gene prediction, constructing a non-redundant gene set; performing species annotation and functional annotation, obtaining relative abundance information of microbial species and functional pathways; constructing a microbial symbiotic network based on the species relative abundance information and performing topological property analysis; screening differential marker microorganisms and differential functional pathways between the atopic dermatitis group and the healthy control group; and integrating the two to construct a combined biomarker. The application breaks through the resolution limitation of traditional technology, and systematically reveals the conversion mechanism of the intestinal flora of atopic dermatitis patients from a "steady-state fermentation type" to an "inflammation adaptation type" from three dimensions of species composition, symbiotic network and functional potential, thereby providing technical support for the mechanism research of atopic dermatitis.

Description

Technical Field

[0001] This invention relates to the fields of biomedicine and bioinformatics, specifically to a method for analyzing the characteristics of gut microbiota in atopic dermatitis and screening biomarkers based on metagenomics. Background Technology

[0002] Atopic dermatitis (AD) is a chronic, relapsing, inflammatory skin disease characterized by dry skin, intense itching, and eczematous lesions. In recent years, its incidence has been rising globally, becoming a significant public health issue. Although the diagnostic criteria for AD are relatively well-established, its etiology is complex, involving the interaction of genetic susceptibility, immune dysregulation, skin barrier dysfunction, and environmental factors.

[0003] In recent years, the "gut-skin axis" theory has provided a new perspective for understanding the systemic pathogenesis of Alzheimer's disease (AD). As the largest micro-ecosystem in the human body, the gut microbiota can remotely regulate skin inflammation and barrier function through its metabolites (such as short-chain fatty acids) and microbiome-associated molecular patterns. Numerous studies have confirmed significant gut microbiota dysbiosis in AD patients. However, current research on the gut microbiota in AD suffers from the following limitations: Most studies rely on 16S rRNA gene amplicon sequencing technology; this technology only provides information on bacterial community composition at the genus or phylum level, which is difficult to be accurate to the species level, and cannot directly and comprehensively reveal the functional gene potential of the bacterial community, resulting in insufficient ability to identify key functional strains related to AD. Existing technologies for analyzing gut microbiota mostly focus on species abundance comparisons, neglecting the complex co-occurrence, mutual exclusion, and niche competition relationships among microbial species. They are unable to reveal the overall stability and network topology changes of the gut microbiota ecosystem under AD conditions. Traditional analyses have failed to effectively combine changes in species composition with shifts in community functional potential, and thus cannot provide a deep understanding of the mechanisms by which dysbiosis affects host immune metabolism at the functional level.

[0004] Therefore, in view of the above situation, there is an urgent need to provide a metagenomics-based method for analyzing the gut microbiota characteristics and screening biomarkers in atopic dermatitis, in order to overcome the shortcomings in current practical applications. Summary of the Invention

[0005] The purpose of this invention is to provide a method for analyzing the gut microbiota characteristics and screening biomarkers in atopic dermatitis based on metagenomics, effectively solving the problems mentioned in the background art.

[0006] This invention is achieved through a metagenomics-based method for analyzing the gut microbiota characteristics and screening biomarkers in atopic dermatitis, comprising the following steps: Step S1: Obtain metagenomic sequencing data from fecal samples of the subjects; Step S2: Assemble and predict genes from the metagenomic sequencing data to construct non-redundant initial Unigenes and screen out the final Unigenes; Step S3: Perform species annotation and functional annotation based on the final Unigenes to obtain the abundance information of microbial species and the abundance of species and functional pathways; Step S4: Based on the abundance of the microbial species, construct a microbial symbiotic network and analyze the topological properties of the microbial symbiotic network; Step S5: Based on the abundance of the species and the abundance of the functional pathways, screen for differentially expressed microorganisms and differentially expressed functional pathways that are significantly different between the atopic dermatitis group and the healthy control group. Step S6: Integrate the differentially marked microorganisms with differentially functional pathways to construct a combined biomarker for characterizing atopic dermatitis.

[0007] As a further aspect of the present invention: the method for constructing the microbial symbiotic network in step S4 is as follows: Calculate the Spearman correlation coefficient between any two microbial species, and construct an undirected network based on the correlation between Spearman correlation coefficients with an absolute value greater than a preset threshold and a p-value less than a preset threshold. The topological properties include at least one selected from the number of nodes, number of edges, average degree, average path length, clustering coefficient, and network diameter.

[0008] As a further aspect of the present invention: the preset threshold is an absolute value of the Spearman correlation coefficient > 0.6 and a p-value < 0.01.

[0009] As a further aspect of the present invention, step S4 further includes a step of performing Zi-Pi analysis on the microbial symbiotic network. Calculate the intra-module connectivity Zi and inter-module connectivity Pi of each node in the network, and identify the nodes as peripheral nodes, connection nodes, module hub nodes or network hub nodes with thresholds of Zi=2.5 and Pi=0.62.

[0010] As a further aspect of the present invention: in step S5, linear discriminant analysis effect size analysis is used to screen the differentially identified marker microorganisms, and reporter factor method is used to screen the differentially functional pathways.

[0011] As a further aspect of the present invention: the screening criteria for the linear discriminant analysis effect size analysis are LDA>3 and p<0.05; The screening criteria for the reporting factor method are |Z|>1.96 and p<0.05 after correction.

[0012] As a further aspect of the present invention: the differential marker microorganisms mentioned in step S5 include: Compared to the healthy control group, the abundance of Clostridium praosporum and Ruminococcus brucelli in the intestine of the atopic dermatitis group was reduced, while the abundance of Bacteroides monomorphosum and Bifidobacterium longum was increased.

[0013] As a further aspect of the present invention, the differential marker microorganisms also include the relative abundance ratio of Bifidobacterium longum and Segardella coccidioides, which is significantly higher in the atopic dermatitis group than in the healthy control group.

[0014] As a further aspect of the present invention: the differential functional pathways mentioned in step S5 include: Compared to the healthy control group, the atopic dermatitis group showed significant upregulation of the two-component system pathway, ABC transporter pathway, non-phosphorylated ED pathway module, and pyrimidine deoxyribonucleic acid synthesis module in the gut microbiota. Significant downregulation was observed in fatty acid biosynthesis pathways, fructose and mannose metabolism pathways, and the phosphotransferase system pathway.

[0015] As a further aspect of the present invention: the non-phosphorylated ED pathway module is a KEGG Module M00309.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention employs metagenomic sequencing technology, achieving species-level resolution and overcoming the limitations of traditional 16S rRNA technology. Furthermore, this invention integrates analyses from three dimensions: "species composition, symbiotic network, and functional potential," and for the first time systematically reveals the global transformation of the gut microbiota in atopic dermatitis patients from a "steady-state fermentation type" to an "inflammatory adaptive type."

[0017] This invention, by constructing a microbial symbiotic network and performing Zi-Pi topology analysis, can identify whether the gut microbiota network of patients with atopic dermatitis exhibits characteristics such as "closer connections, predominant positive correlations, and the emergence of specific module hubs," providing a new method for assessing changes in the stability of the microecology associated with AD. Attached Figure Description

[0018] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is a flowchart of the method of the present invention.

[0020] Figure 2 This is a schematic diagram of the composition of the microbial community in the sample of this invention.

[0021] Figure 3 This is a schematic diagram of the neutral microbial community model analysis in this invention.

[0022] Figure 4 This is a schematic diagram illustrating the differences in microbial community composition in this invention.

[0023] Figure 5 This is a schematic diagram illustrating the differences in α-diversity and β-diversity of microbial communities in this invention.

[0024] Figure 6 This is a schematic diagram of differential microbial LEfSe analysis in this invention.

[0025] Figure 7 This is a schematic diagram of the topological structure of the microbial community symbiotic network in this invention.

[0026] Figure 8 This is a schematic diagram of Zi-Pi analysis of the microbial community symbiotic network in this invention.

[0027] Figure 9 This is a visualization of the core differences in the KEGG pathway and modules in this invention. Detailed Implementation

[0028] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] The present invention will be further explained below with reference to specific embodiments.

[0030] Please see Figure 1 The method for analyzing the gut microbiota characteristics and screening biomarkers for atopic dermatitis based on metagenomics provided in this invention includes the following steps: Step S1: Obtain metagenomic sequencing data from fecal samples of the subjects; Step S2: Assemble and predict genes from the metagenomic sequencing data to construct non-redundant initial Unigenes and screen out the final Unigenes; Step S3: Perform species annotation and functional annotation based on the final Unigenes to obtain the abundance information of microbial species and the abundance of species and functional pathways; Step S4: Based on the abundance of the microbial species, construct a microbial symbiotic network and analyze the topological properties of the microbial symbiotic network; Step S5: Based on the abundance of the species and the abundance of the functional pathways, screen for differentially expressed microorganisms and differentially expressed functional pathways that are significantly different between the atopic dermatitis group and the healthy control group. Step S6: Integrate the differentially marked microorganisms with differentially functional pathways to construct a combined biomarker for characterizing atopic dermatitis.

[0031] In a more specific example, the method for constructing the microbial symbiotic network described in step S4 is as follows: Calculate the Spearman correlation coefficient between any two microbial species, and construct an undirected network based on the correlation between Spearman correlation coefficients with an absolute value greater than a preset threshold and a p-value less than a preset threshold. The topological properties include at least one selected from the number of nodes, number of edges, average degree, average path length, clustering coefficient, and network diameter.

[0032] Preferably, the preset threshold is a Spearman correlation coefficient absolute value > 0.6 and a p value < 0.01.

[0033] Preferably, step S4 further includes a Zi-Pi analysis of the microbial symbiotic network: Calculate the intra-module connectivity Zi and inter-module connectivity Pi of each node in the network, and identify the nodes as peripheral nodes, connection nodes, module hub nodes or network hub nodes with thresholds of Zi=2.5 and Pi=0.62.

[0034] In a more specific example, in step S5, linear discriminant analysis effect size analysis is used to screen the differentially identified marker microorganisms, and reporter factor method is used to screen the differentially functional pathways.

[0035] Preferably, the screening criteria for the linear discriminant analysis effect size analysis are LDA > 3 and p < 0.05; The screening criteria for the reporting factor method are |Z|>1.96 and p<0.05 after correction.

[0036] In a more specific example, the differential marker microorganisms mentioned in step S5 include: Compared to the healthy control group, the abundance of Clostridium praosporum and Ruminococcus brucelli in the intestine of the atopic dermatitis group was reduced, while the abundance of Bacteroides monomorphosum and Bifidobacterium longum was increased.

[0037] Preferably, the differential marker microorganisms also include a significantly higher relative abundance ratio of Bifidobacterium longum and Segardella coccidioides in the atopic dermatitis group compared to the healthy control group.

[0038] In a more specific example, the differentially functioning pathways described in step S5 include: Compared to the healthy control group, the atopic dermatitis group showed significant upregulation of the two-component system pathway, ABC transporter pathway, non-phosphorylated ED pathway module, and pyrimidine deoxyribonucleic acid synthesis module in the gut microbiota. Significant downregulation was observed in fatty acid biosynthesis pathways, fructose and mannose metabolism pathways, and the phosphotransferase system pathway.

[0039] Preferably, the non-phosphorylated ED pathway module is a KEGG Module M00309.

[0040] In the embodiments of this invention, this invention breaks through the resolution limitations of traditional 16S rRNA technology and systematically reveals the transformation mechanism of the gut microbiota of atopic dermatitis patients from "steady-state fermentation type" to "inflammatory adaptation type" from three dimensions: species composition, symbiotic network, and functional potential, providing efficient technical support for the study of the mechanism of atopic dermatitis.

[0041] Please refer to Figures 1-9 To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below in conjunction with the practical application of this method, as follows: 1. Research Content and Methods 1.1 Research Subjects This study recruited 20 patients who met the Chinese diagnostic criteria for atopic dermatitis as the AD group, and 20 age-, sex-, and BMI-matched healthy individuals as the control group (CON). All participants signed informed consent forms and had not used antibiotics, probiotics, or enterotropic drugs within 4 weeks prior to sampling.

[0042] 1.1.1 Diagnostic and Severity Classification Criteria for AD Diagnostic criteria: The Zhang diagnostic criteria were adopted. Professor Zhang Jianzhong, by summarizing the characteristics of atopic dermatitis (eczema) in Chinese adolescents and adults, proposed a "Chinese standard" suitable for the diagnosis of atopic dermatitis in my country: ① Patients with symmetrical eczema with a disease course exceeding 6 months; ② Personal and / or family history of specific diseases (including eczema, allergic rhinitis, asthma, allergic conjunctivitis, etc.); ③ Elevated serum IgE and / or elevated peripheral blood IgE and / or positive allergen test (allergen-specific IgE test grade 2 or above positive); Atopic dermatitis can be diagnosed if the first criterion above is met, plus either the second or third criterion.

[0043] Severity classification criteria: The severity of skin lesions, itching, and sleep disturbance in AD patients are classified according to the Scoring Atopic Dermatitis (SCORAD) score. A score <25 indicates mild, 25-50 indicates moderate, and >50 indicates severe.

[0044] 1.1.2 Inclusion and exclusion criteria for experimental cases A. Case inclusion criteria: Those who meet the diagnostic criteria for AD; Age ≥ 18 years old; No antibiotics, probiotics, prebiotics, or drugs that regulate the intestinal nervous system were used in the four weeks prior to stool collection; Comorbid chronic diarrhea, inflammatory bowel disease, or other diseases known to cause intestinal flora imbalance; Patients with severe dysfunction of organs such as the heart, brain, lungs, and kidneys; Patients with concurrent malignant tumors; Pregnant women and breastfeeding women; Those who have signed the informed consent form; Good compliance; agreed to participate in questionnaires and subsequent follow-up treatment.

[0045] B. Criteria for exclusion and dropout of cases: Those who were found to be ineligible for inclusion or eligible for exclusion after inclusion; Those with incomplete information.

[0046] 1.2 Reagents and Equipment The main reagents and consumables are shown in Table 1 and Table 2.

[0047] Table 1. Reagents and instruments used for DNA extraction from fecal samples

[0048] Table 2. Reagents and instruments used for metagenomic library construction

[0049] 1.3 Materials and Methods 1.3.1 Fecal Sample Collection All samples must be collected on an empty stomach. Considering the circadian rhythm of gut microbiota changes, collection time is 8-9 AM. Urinate before collecting stool samples, taking care to avoid urine contamination. Use a sterile scoop to collect 1-3g of midstream stool and place it in a sterile cryovial or centrifuge tube. Immediately flash freeze in liquid nitrogen and then transfer to... Store in the refrigerator.

[0050] 1.3.2 Clinical Data Collection ① Demographic data: age, sex, height, weight; ② Disease characteristics: age of onset, season of onset, personal and family history of allergies, clinical characteristics, severity of the disease (range of skin lesions, severity of skin lesions, degree of itching and sleep impact) and degree of impact on quality of life (severity of the disease and quality of life are assessed using SCORAD / IGA / EASI / scores, DLQI / CDLQI scores); ③ Laboratory tests: eosinophil count / percentage, serum total IgE level, allergen-specific IgE.

[0051] 1.3.3 DNA extraction, detection and library construction Total DNA was extracted from fecal samples using the QIAamp DNA Stool Mini Kit (QIAGEN, Germany) according to the manufacturer's instructions. DNA purity and concentration were assessed using a NanoDrop 2000 spectrophotometer and a Qubit 4.0 quantitative PCR instrument, and DNA integrity was checked using 1% agarose gel electrophoresis. Qualified DNA samples were randomly fragmented to construct metagenomic sequencing libraries, and library quality was tested using an Agilent 2100 Bioanalyzer. Paired-end sequencing (PE150) was performed using the Illumina NovaSeq 6000 platform to obtain raw metagenomic sequence data. The specific operational procedure is as follows: (1) DNA extraction process from fecal samples Preparation before the experiment: Take out the required reagents in advance, observe whether the amount of reagents is sufficient for the day's use, and bring the reagents to room temperature.

[0052] Prepare the required number of sterile pipette tips (1000ul tips need to be cut, 1000ul tips do not need to be cut, 200ul tips, and 10ul tips).

[0053] Thaw the fecal collection tube overnight in a 4-degree refrigerator.

[0054] Pre-wash buffer and Lysis buffer F1 were heated in a 55°C water bath.

[0055] The sterile operating table is sterilized with ultraviolet light for 30 minutes.

[0056] Add 42 ml of isopropanol to Binding buffer F, and mark the time on the cap (newly opened kit).

[0057] Add 45 ml of anhydrous ethanol to Wash buffer F1, and mark the time on the cap (newly opened kit).

[0058] Add 90 ml of anhydrous ethanol to Wash buffer F2, and mark the time on the cap (newly opened kit).

[0059] The magnetic beads are mixed by vortex oscillation before use.

[0060] Pre-wash: Add 30-300 mg of feces to 1000 μL of pre-wash buffer in the lysis tube and vortex for 20-40 seconds to mix the sample thoroughly.

[0061] Centrifuge at 14,000 rpm for 3-5 mins and discard the supernatant (be careful not to aspirate the precipitate at the bottom of the tube).

[0062] Cell lysis: Add 980 µL Lysis Buffer F1, 120 µL Lysis Buffer F2 and 10 µL RNase ASolution, and vortex to mix.

[0063] Grind with a tissue homogenizer at 5 m / s for 40 s, or use a vortex mixer to vortex at the maximum speed for 10 min.

[0064] Centrifuge at 14,000 rpm for 5-10 mins, retain the supernatant and discard the precipitate.

[0065] Remove impurities: Transfer 900 µL of supernatant to a new 2 mL centrifuge tube, add 250 μL of PPS and invert 20 times, being careful not to shake violently.

[0066] Centrifuge at 14,000 rpm for 5 mins, retain the supernatant and discard the precipitate.

[0067] Combination: Transfer 1000 μL of supernatant to a new 2 mL centrifuge tube, add 1000 μL of Binding Buffer F and 10 μL of Magnetic Beads, and vortex to mix. Magnetic beads need to be vortexed before use.

[0068] Place it on an oscillator and oscillate for 5-10 minutes.

[0069] Place on the magnetic rack for 3-5 minutes until the magnetic beads are completely attracted, then discard the supernatant (be careful not to touch the magnetic beads).

[0070] rinsing: Add 1000 μL Wash Buffer F1 and shake on a shaker for 3 mins.

[0071] Place the beads on a magnetic rack and let them attract magnetically for 3-5 minutes until they are completely attracted. Discard the supernatant.

[0072] Add 1000 μL Wash Buffer F2 and shake on a shaker for 3 mins.

[0073] Place the beads on a magnetic rack and let them attract magnetically for 3-5 minutes until they are completely attracted. Discard the supernatant.

[0074] Repeat steps 3 and 4 once.

[0075] Air dry at 55℃ for 5-10 minutes.

[0076] Washout: Add 103 μL TE Buffer, resuspend the magnetic beads, and place the centrifuge tube in a metal bath at 55°C for 5 mins to elute.

[0077] Place the tube on a magnetic rack and magnetically attract the beads for 3-5 minutes until the magnetic beads are completely adsorbed (the supernatant is completely clear). Transfer 101 μL of supernatant (eluted DNA) to a new 1.5 mL centrifuge tube. For long-term storage, please store at -20°C and avoid repeated freeze-thaw cycles.

[0078] Qubit library concentration determination: The concentration of the purified product was determined using a Qubit assay, as follows: Standard products: 190ul buffer + 10ul standard; 190ul buffer + 10ul standard (2 samples); Library: 199ul buffer + 1ul sample.

[0079] (2) Metagenomic library construction procedure Fragmentation and end-of-pipe repair: Prepare the reaction system as shown in Table 3 in a sterile PCR tube placed on ice (FS Pro Enzymes I is added last and must be prepared on ice): Table 3. Preparation of Fragmentation and Terminal Repair Reagents

[0080] Use a pipette to gently blow or shake the mixture to ensure it is thoroughly mixed, then briefly centrifuge.

[0081] Run the reaction program in Table 4. When the PCR instrument reaches 32°C, immediately place the PCR tube in the instrument. Set the PCR instrument's heating cap to 75°C.

[0082] Table 4 Operating Temperature and Duration

[0083] Connector connection: Take the required magnetic beads and sterilized deionized water out of the 4°C refrigerator in advance and bring them to room temperature. Mix the magnetic beads well and allow them to equilibrate at room temperature for at least 30 minutes.

[0084] Configure the reaction system on ice according to Table 5. First, add the Adapter separately to the previous reaction, and then add FSPro Ligation Buffer I and Ligase Enzymes.

[0085] Table 5. Preparation of reagents for connector connection

[0086] Place the PCR tubes on the PCR instrument and execute the procedure in Table 6. The PCR instrument does not have a heat cap.

[0087] Table 6 Operating Temperature and Duration

[0088] Add 64 μL (0.8X) AFTMag NGS DNA Clean Beads to the sample, vortex to mix, incubate at room temperature for 5 mins, and then briefly centrifuge.

[0089] Place the solution on a magnetic rack and let it stand for 2-5 minutes until the solution becomes clear. Then remove the supernatant (be careful not to touch the magnetic beads).

[0090] Keep the centrifuge tubes or 96-well plates on a magnetic rack, add 200 μL of 80% ethanol to rinse the magnetic beads, incubate for 30 s, and then remove the supernatant.

[0091] Repeat the previous step.

[0092] Keep the centrifuge tube or 96-well plate on the magnetic rack, use a 10 μL pipette to remove any residual ethanol from the bottom of the tube, and dry until no ethanol remains.

[0093] Remove the centrifuge tubes or 96-well plates from the magnetic rack, add 103 μL of Nuclease-Free Water to resuspend the magnetic beads, and let stand at room temperature for 2 minutes to allow the DNA on the magnetic beads to be fully released.

[0094] Place the centrifuge tube or 96-well plate on a magnetic rack for 2 minutes, then transfer 100 μL of the supernatant into a new centrifuge tube or 96-well plate.

[0095] This product can be temporarily stored at 4℃ / -20℃ for about 1-2 weeks.

[0096] Segment Filtering: Add 60 μL (0.6X) AFTMag NGS DNA Clean Beads to 100 μL of supernatant, vortex to mix, incubate at room temperature for 5 mins, and then incubate briefly.

[0097] Place the centrifuge tubes or 96-well plates on a magnetic rack for 2 minutes until the solution becomes clear. Then transfer the supernatant to a new centrifuge tube or 96-well plate (remember not to discard the supernatant).

[0098] Add 20 μL (0.2X) AFTMag NGS DNA Clean Beads to the supernatant, vortex to mix, incubate at room temperature for 5 mins, and then briefly centrifuge.

[0099] Place the centrifuge tubes or 96-well plates on a magnetic rack and let them stand for 2 minutes until the solution becomes clear. Then remove the supernatant (be careful not to touch the magnetic beads).

[0100] Keep the centrifuge tubes and 96-well plates on a magnetic rack, add 200 μL of 80% ethanol to rinse the magnetic beads, incubate for 30 seconds, and then remove the supernatant.

[0101] Repeat the previous step.

[0102] Keep the centrifuge tube or 96-well plate on the magnetic rack, use a 10 μL pipette to remove any residual ethanol from the bottom of the tube, and dry until no ethanol remains.

[0103] Remove the centrifuge tubes or 96-well plates from the magnetic rack, add 22 μL of Nuclease-Free Water to resuspend the magnetic beads, and let stand at room temperature for 2 minutes to allow the DNA on the magnetic beads to be fully released.

[0104] Place the centrifuge tube on a magnetic rack for 2 minutes, then transfer 20 μL of supernatant to a new PCR tube or 96-well plate.

[0105] Library PCR amplification and purification: The configuration is shown in Table 7 (2×PCR Mix and MGI UDI primers should be thawed in advance).

[0106] Table 7. Preparation of Reagents for Library PCR Amplification and Purification

[0107] Gently vortex to mix and centrifuge. Run the program in Table 8 on a PCR instrument and heat at 105°C.

[0108] Table 8 Operating Temperature and Duration

[0109] Add 50 μL (1X) AFTMag NGS DNA Clean Beads to the amplification product, vortex to mix, incubate at room temperature for 5 mins, and then incubate briefly.

[0110] Place the centrifuge tubes or 96-well plates on a magnetic rack and let them stand for 2 minutes until the solution becomes clear. Then remove the supernatant (be careful not to touch the magnetic beads).

[0111] Keep the centrifuge tubes or 96-well plates on a magnetic rack, add 200 μL of 80% ethanol to rinse the magnetic beads, incubate for 30 s, and then remove the supernatant.

[0112] Repeat the previous step.

[0113] Keep the centrifuge tube or 96-well plate on the magnetic rack, use a 10 μL pipette to remove any residual ethanol from the bottom of the tube, and dry until no ethanol remains.

[0114] Remove the centrifuge tubes or 96-well plates from the magnetic rack, add 33 μL of Nuclease-Free Water to resuspend the magnetic beads, and let stand at room temperature for 2 minutes to allow the DNA on the magnetic beads to be fully released.

[0115] Place the centrifuge tube or 96-well plate on a magnetic rack for 2 minutes, then transfer 31 μL of the supernatant into a new centrifuge tube.

[0116] Qubit concentration determination: The specific steps for determining library concentration using Qubit are as follows: Standard 1: 190ul buffer + 10ul Standard 1; Standard 2: 190ul buffer + 10ul Standard 2; Sample: 199ul buffer + 1ul sample; Insert size detection for document files; This step uses the Agilent 5400 system to accurately determine the size of the inserted fragments in the library. The specific operation strictly follows the instructions of the instrument and the accompanying reagent kit.

[0117] Sample pretreatment: Take 5 μL of the constructed library sample, add 10 μL of enzyme-free sterile water for dilution, vortex for 10 s to mix, and then place it in a high-speed refrigerated centrifuge and centrifuge at 12000 r / min for 30 s to avoid aerosol contamination.

[0118] Preparation of the detection system: In a clean bench, prepare the detection system according to the proportions specified in the kit instructions. Add 20 μL of the dedicated buffer, 1 μL of the labeling dye, and 2 μL of the molecular weight standard to a new 1.5 mL sterile centrifuge tube. Then add 2 μL of the pretreated library sample and gently pipette five times to mix (avoid vigorous shaking to prevent fragment breakage). The total volume of the system is 25 μL. A blank control (using 2 μL of enzyme-free sterile water instead of the library sample, with all other components the same) is also provided to eliminate interference from the reagents themselves.

[0119] Instrument testing: Slowly add the prepared detection system into the capillary sample inlet of the Agilent 5400 system, and set the instrument parameters according to the instruction manual. After the parameters are set, start the instrument testing program and record the electrophoretic pattern and fragment size data in real time.

[0120] 1.3.4 Data Quality Control and Preprocessing The raw metagenomic sequencing data (FASTQ) was first subjected to quality control. FastP was used to remove adapters and filter low-quality reads. The filtering rules were: (1) remove reads containing adapter contamination; (2) remove low-quality reads: reads with N content > 10%; reads with Q ≤ 10 bases accounting for ≥ 50% of the entire read. After quality control, Clean Data (FASTQ) was obtained, and FastP output a report to statistically analyze the number of reads, base number (BaseNum), Q20, Q30, and GC content for each sample, used to evaluate sequencing quality and consistency between samples. Q20 (%): percentage of bases with a quality value ≥ 20; Q30 (%): percentage of bases with a quality value ≥ 30; GC (%): percentage of G and C bases.

[0121] 1.3.5 Assembly The optimized sequences were assembled using Megahit v1.2.9 assembly software (https: / / github.com / voutcn / megahit) to obtain contigs for each sample. Assembly was performed using MEGAHIT, with a minimum contig length threshold set to ≥500 bp. ORF prediction was performed on the contigs, and protein-coding genes (CDS) were obtained using Prodigal.

[0122] 1.3.6 Gene prediction and abundance analysis Using bwa 0.7.1 7, the Clean Data of each sample was sequence-aligned with the non-redundant initial Unigenes. The number of reads that were aligned to the gene in each sample was calculated, and the final Unigenes (reads > 2) were selected. The abundance information of each gene in each sample was calculated and statistical analysis was performed.

[0123] 1.3.7 Species Annotation Using Kraken software (version 2.1.2, http: / / ccb.jhu.edu / software / kraken), Unigenes were classified into species based on precise k-mers alignment. Unigenes were compared with sequences in databases (Bacteria, Fungi, Archaea) and also with each individual database. Through analysis of bacterial, fungal, and archaea annotations and gene abundance, the abundance and number of genes at the kingdom, phylum, class, order, family, genus, and species levels were statistically analyzed for different treatments. The abundance of a species across treatments was calculated by summing the abundance of all genes annotated for that species; the number of genes in each sample was determined by counting the number of genes annotated to that species with a non-zero abundance.

[0124] The non-parametric factorial Kruskal-Wallis (KW) sum-rank test was used to detect features of significant abundance differences and to identify taxa with significant abundance differences. Metastats analysis and linear discriminant analysis (LEfSe) effect size analysis were then performed. In the metastats analysis, t-tests were conducted at each taxonomic level to obtain p-values, which were then corrected for by FDR to obtain q-values. The LEfSe analysis used LEfSe 1.0 software with linear discriminant analysis (LDA) to estimate the magnitude of the effect of abundance of each component (species) on the differential effect size (screening criteria were as follows). p <0.05, LDA score>3).

[0125] 1.3.8 Microbial Alpha Diversity Analysis Microbial alpha diversity was obtained from the abundance tables at each taxonomic level. The Shannon index (H) and Simpson index (D) were calculated to estimate the microbial community diversity in the sample.

[0126] (1) (2) in P i Indicates species i proportion, S This represents the total number of species. The above analysis was performed using the "vegan" package in R software version 4.0.3.

[0127] 1.3.9 Microbial Beta Diversity, PCoA, and PERMANOVA Analysis To compare the differences in microbial composition between AD and CON, β-habitat diversity was used as a measure. To examine differences between microbial communities, principal component analysis (PCoA) was performed using Bray-Curtis distance, which reflects differences in community species composition. Nonparametric tests, specifically permutational multivariate analysis of variance (PERMANOVA), were used to test for significance. A Bray-Curtis distance matrix was constructed based on relative abundance at the phylum / species level to compare dissimilarity between communities, with a permutation test statistic of 999. PCoA was implemented in R4.0.3 using `vegan::wcmdscale`; PERMANOVA was implemented using `vegan::adonis`; and pairwise comparisons were performed using the `pairwiseAdonis` package.

[0128] (3) In the formula: k It is the number of species (the number of species in the sample-species matrix). y Indicates species abundance. g and h This indicates the numbers of the two sample plots being compared.

[0129] 1.3.10 Neutral Community Model Analysis To clarify the potential importance of stochastic processes in microbial community assembly, a neutral community model (NCM) was used to analyze the contribution of stochastic processes to the microbial community. The relationship between the occurrence frequency and relative abundance of operational taxonomic units (OTUs) in the community was analyzed. The calculation formula is as follows: Freqi =1- I (1 / ) N | N × m × pi , N × m ×(1- pi )) (4) in, Freq i Classification unit i Frequency of occurrence in the community; N It represents the number of individuals in each community; m It is the estimated diffusion rate. m A smaller value indicates that species dispersal is more restricted throughout the community, and vice versa. m A higher value indicates a lower degree of restriction on species dispersal; p i It is a taxonomic unit in the community. i The average relative abundance; I yes β The probability density function of the distribution. The 95% confidence interval is calculated based on 1000 bootstrap tests. This analysis is based on nonlinear least squares fitting, R0. 2 R represents the fitting parameters, i.e., the overall goodness of fit of the neutral community model. 2 A higher value indicates a closer approximation to the neutral model. The MicEco package in R software version 4.0.3 was used.

[0130] 1.3.11 Symbiotic Network Topology Analysis To analyze the interrelationships and species coexistence mechanisms among microorganisms, a meta-network of community symbiosis consisting of bacterial and fungal communities was constructed. The symbiotic network includes all operational taxonomic units (OTUs) of bacteria and fungi. Spearman correlations between pairwise OTUs were calculated, and the false discovery rate (FDR) method was used to analyze the relationships between them. p The value is corrected to retain |r|>0.6. p For correlations <0.01, calculate the adjacency matrix of the correlation coefficient matrix, and construct an undirected network from the adjacency matrix.

[0131] To describe the network topology, at the network level, we calculated the following metrics: average path length, which is the average distance between all pairs of nodes; average degree, which is the average connectivity between each node and every other unique node in the network; and clustering coefficient, which indicates the degree to which nodes tend to cluster together. Higher average degree and clustering coefficient, and lower average path length, indicate a more tightly connected network.

[0132] 1.3.12 Intra-module and Inter-module Connectivity Analysis To identify the response patterns of the microbial communities in groups AD and CON, network analysis was conducted using within-module connectivity (Zi) and among-module connectivity (Pi), calculated as follows: (5) (6) in, Zi , Ki It is a node i In the module Si The number of edges between a node and other nodes; K Si It is a module Si All nodes K The average value K The value is at the node Si The number of connections between the node and other nodes; σKSi It is a module Si All nodes K Standard deviation of the value; Pi , K is It is a node i With modules S The number of edges in a node. k i It is a node i The degree; M Representative module; N M This represents all the modules.

[0133] Nodes are divided into four types: network hubs (Module hubs, Zi >2.5 and Pi >0.62); Module central nodes (Network hubs, Zi >2.5 and Pi≤ 0.62); Connectors, Zi ≤ 2.5 and Pi >0.62); Peripherals, Zi ≤ 2.5 and Pi (≤ 0.62). Typically, the three types of nodes other than peripheral nodes are classified as critical nodes.

[0134] 1.3.13 Functional Path and Module Analysis Sequencing data were aligned and annotated against the KEGG database, with sequences mapped to KEGG Orthology (KO) numbers and then to KEGG pathway and module levels. Differentially functional pathways were identified using the reporter score method, with a Z-score |Z| > 1.96 after multiple corrections. p <0.05 indicates a significantly different pathway.

[0135] 2. Results 2.1 Basic Information of the Research Subjects This study enrolled 20 patients with Alzheimer's disease (AD) as the case group (AD) and 20 healthy individuals as the control group (CON). There were no statistically significant differences between the two groups in terms of age, sex, and BMI.

[0136] 2.2 Metagenomic Sequencing and Assembly Information Illumina sequencing results showed that 2,556,918,228 high-quality clean reads were obtained from 40 samples after filtering. The Q20 percentage for each sample ranged from 97.02% to 98.73% (average 97.90%), and the Q30 percentage ranged from 89.88% to 95.83% (average 92.83%). Except for AD11 (Q30 = 89.88%), the Q30 percentages for all other samples were ≥90%. The GC content of the clean data ranged from 33.73% to 48.50% (average 44.66%). Overall, both Q20 and Q30 remained at high levels, indicating good sequencing data quality and a comprehensive reflection of the microbial community structure of the samples (Table 9).

[0137] Table 9. Analysis of Sequencing Results for Different Samples

[0138] 2.3 Metagenomic Sequencing and Assembly Information According to the metagenomic sequencing results and comparison with the NR database, the microorganisms in the sample were mainly classified as bacteria, eukaryotes, and archaea, accounting for 99.619%, 0.002%, and 0.379% respectively in the AD group. Figure 2 (A in the text); among them, samples AD04, AD05, AD06, AD09, AD10LW, AD12, AD13, AD14, AD15, AD16, AD17, AD18, AD19, AD20, AD22, AD23, and AD3 were all bacteria ( Figure 2 In the AD02 sample, eukaryotes accounted for the highest proportion (0.024%), while archaea accounted for the highest proportion (7.577%) in the AD10GYG sample. Bacteria and eukaryotes accounted for 99.999% and 0.001% of the CON group, respectively; no archaea were found. Specifically, samples CON1, CON2, CON3, CON4, CON5, CON6, CON7, CON9, CON10, CON11, CON12, CON13, CON14, CON16, CON18, CON19, and CON20 were all bacterial; eukaryotes accounted for the highest proportion in samples CON15 and CON17, both at 0.008%.

[0139] Based on species annotation results from the NR database, taxonomic classification statistics of metagenomic sequences from 40 samples were performed. Figure 2 (C) At the six taxonomic levels of phylum, class, order, family, genus, and species, 14 phyla, 26 classes, 40 orders, 74 families, 261 genera, and 699 species were identified, respectively. Overall, the number of identifiable taxa increases with the taxonomic hierarchy, indicating that the microbial community in this study has high taxonomic complexity and species richness, providing a reliable taxonomic basis for subsequent community composition comparisons (such as intergroup differences and dominant group screening) and functional potential analysis.

[0140] Based on community building mechanisms, niche theory and neutrality theory determine the community assembly process. To assess the relative contributions of random drift and diffusion processes during the microbial community assembly in groups AD and CON, this study fitted the model using the Sloan Neutral Community Model (NCM) within both groups. Figure 3 The results showed that the neutral community model had a fitting parameter of Nm = 643 in the AD group, and the goodness of fit was R. 2=0.372. Within the AD group, the frequency of taxa occurrence can be explained to some extent by their average abundance and neutral diffusion processes. The scatter distribution exhibits a typical nonlinear threshold pattern: taxa in the low average abundance range generally correspond to lower occurrence frequencies, while as the average abundance increases, the occurrence frequency rises rapidly and approaches 1. At the same time, a certain number of taxa still deviate significantly from the predicted range: taxa above the upper confidence limit occur more frequently than the neutral expectation, suggesting that they may have stronger colonization capabilities or a wider ecological adaptation range; taxa below the lower confidence limit occur less frequently than the neutral expectation, suggesting that they may be constrained by deterministic processes such as diffusion restriction or environmental filtering.

[0141] The results of the neutral model fitting in the CON group show that the model parameter Nm=729 and the goodness of fit R0 is 729. 2 =0.323. Similar to the AD group, the CON group also exhibits a nonlinear relationship of "increasing average abundance and rapidly increasing frequency of occurrence," indicating that neutral processes also have some explanatory power in the formation of the CON group community structure. However, in terms of goodness of fit, the R-value of the CON group is... 2 The lower value of the CON group compared to the AD group suggests that the neutral model's explanation of the frequency distribution of the CON group is relatively weak. Meanwhile, the higher Nm value of the CON group compared to the AD group can be interpreted within the neutral model framework as a larger combined factor of meta-community size and immigration intensity (N×m), implying that the CON group may have a higher overall strength in random diffusion / immigration processes. Overall, both groups achieve a moderate level of neutral model fit, indicating that random drift and diffusion processes contribute to community assembly.

[0142] 2.4 Analysis of Gut Microbiota Composition in AD Patients Figure 4Figure A illustrates the differences in microbial composition between AD and CON at the phylum level. The AD group contains 12 phyla. Among the top 5 phyla in terms of relative abundance, Firmicutes is absolutely dominant, with a relative abundance of 67.04% ± 4.05%. This is followed by Bacteroidota (17.73% ± 4.05%) and Actinobacteria (11.28% ± 2.48%); these three phyla together contribute 96.05% of the average relative abundance. Furthermore, Proteobacteria (2.13% ± 0.71%) and Verrucomicrobia (1.37% ± 0.79%) both have relative abundances greater than 1%. The remaining phyla in the CON group are all at low abundance levels (<1%). The CON group contains 10 phyla. The top three dominant phyla are consistent with the AD group, with Firmicutes being the most dominant phylum, at a relative abundance of 64.98% ± 4.92%. This is followed by Bacteroidota (27.93% ± 5.32%) and Actinobacteria (4.65% ± 0.86%). These three phyla together contribute 97.57% of the average relative abundance. Furthermore, Proteobacteria's relative abundance in the CON group also exceeds 1% (2.29% ± 1.01%). The remaining phyla are all at low abundance levels (<1%) in the CON group. In summary, both groups are dominated by Firmicutes and Bacteroidetes, but the dominant patterns differ: compared to CON, the AD group shows a structural change of "firmbacteria maintaining a high proportion, Bacteroidetes relatively decreasing, Actinobacteria relatively increasing, and Verrucous Microbes reaching ≥1% on average."

[0143] Figure 4 In the diagram, B represents the top 10 species by relative abundance selected for comparison of species-level microbial composition between the AD and CON groups. A total of 598 species with relative abundance greater than 0% were detected in the AD group, while 541 species were detected in the CON group. The average relative abundance of the top 10 species was 35.38% in the AD group and 41.32% in the CON group, both exhibiting a structure dominated by a few dominant species and low abundance distribution of the remaining species. However, compared to CON, the concentration of dominant species in the AD group was relatively lower.

[0144] Among the Top 10 species, the species with the highest relative abundance in the AD and CON groups were Faecalibacterium prausnitzii (AD: 6.03%±1.13%; CON: 5.70%±1.01%, Table 10) and Ruminococcus bromii (AD: 4.54%±1.08%; CON: 5.60%±1.76%). The remaining Top 10 species showed a clear abundance bias between the two groups: the average abundance of *Segatella copri* in the CON group was 8.60%±4.41%, significantly higher than that in the AD group (0.67%±0.56%), with an average abundance increase of 1175.10% (approximately 12.75 times) compared to the AD group; meanwhile, the average abundance of *Anaerostipes hadrus* (CON: 4.62%±1.01%; AD: 2.91%±0.64%), *Phocaeicola vulgatus* (CON: 4.08%±1.40%; AD: 2.36%±1.01%), and *Eubacterium rectale* (CON: 3.91%±0.97%; AD: 2.35%±0.69%) was also higher in the CON group. Conversely, the average abundance of Bacteroides uniformis (AD: 4.16%±1.62%; CON: 1.38%±0.38%) and Bifidobacterium longum (AD: 4.08%±1.05%; CON: 1.15%±0.34%) in the AD group was significantly higher than that in the CON group, increasing by 200.43% (approximately 3.00-fold) and 255.13% (approximately 3.55-fold) respectively. In addition, Blautia wexlerae was also higher in the AD group (AD: 5.06%±1.10%; CON: 3.90%±0.75%).

[0145] In summary, at the Top 10 species level, both groups show that Faecalibacterium prausnitzii and Ruminococcus bromii are the common dominant species. However, the CON group is more biased towards the enrichment of Segatella copri and some butyrate-producing dominant species (such as Anaerostipes hadrus), while the AD group is more biased towards the relative enrichment of species such as Bacteroides uniformis and Bifidobacterium longum.

[0146] Table 10. Analysis of relative abundance of dominant species

[0147] Based on unigene sequencing data, the α-diversity of the two groups of samples was analyzed. Figure 5 (A) The results showed that the AD group tended to have higher scores than the CON group on both the Shannon and Simpson indices, but the differences did not reach a statistically significant level. p >0.05). Among them, the mean Shannon index of AD group was 3.28±0.50, and the mean of CON group was 3.14±0.61. As for Simpson index, the mean of AD group was 0.927±0.039, and the mean of CON group was 0.890±0.147.

[0148] This study further combined β-diversity ordination (PCoA) with differential abundance analysis (PERMANOVA, 999 permutations) to characterize changes in community structure. Figure 5 (B) Principal coordinate analysis (PCoA) of the Bray–Curtis distance matrix constructed based on phylum-level relative abundance showed that PCoA1 and PCoA2 explained 71.73% and 15.62% of the community variation, respectively, with a cumulative explained rate of 87.34%, indicating that the community variation at the phylum level can be adequately characterized by the first two axes. Visualization results showed that the AD and CON groups exhibited a certain separation trend in the two-dimensional ordination space, but overall still showed significant overlap. Further PERMANOVA based on the Bray–Curtis distance showed that the groups explained R0 of the phylum-level community structure. 2 =0.0469, the difference was not statistically significant (F=1.87, p =0.158,999 permutations). Overall, at the phylum level, the β-diversity difference between AD and CON is relatively weak, and the intergroup separability is limited. At the species level, PCoA1 and PCoA2 explain 13.22% and 10.50% of the community differences, respectively, with a cumulative explained rate of 23.72%. Compared with the phylum level, the variation at the species level is dispersed across more dimensions, suggesting that community differences at the species level are more complex and that low-abundance species account for a higher proportion. In the two-dimensional ordination plot, the AD and CON groups show a clearer distribution difference in the principal coordinate space. The PERMANOVA results further indicate that the group explains R0 of the species-level community structure. 2 =0.0441, the difference between groups reached statistical significance (F=1.75, p =0.013,999 permutations). Therefore, although the overall proportion of variation explained by groups is not high, the differences in community structure between AD and CON are more sensitive and detectable at species resolution. That is, AD-related microbial changes are more likely to be reflected in the compositional rearrangement at the species level rather than a large-scale overall replacement at the phylum level.

[0149] Through LEfSe analysis, this study further screened out differential markers at different taxonomic levels in the two groups of microbial communities. Figure 6 LEfSe combines nonparametric tests with linear discriminant analysis (LDA) for effect size assessment: first, it uses between-group tests to screen for differential groups (in this study, the group with the most significant difference is LEfSe). P The significance of .unadj is determined, and the BH-corrected value is given. P .adj), using LDA values ​​to measure the contribution of differences (the larger the LDA, the greater the contribution of the group to explaining the differences between groups), and a bar chart of LDA value distribution shows the groups with significant abundance differences and the magnitude of their influence (LDA>3). After significance screening ( P (Undj < 0.05), a total of 114 differential marker lineages were obtained, of which 72 were enriched in the AD group and 42 in the CON group. Specifically, the differential markers in the CON group were mainly concentrated in the Bacteroidota lineage, with Prevotellaceae being one of the most significant differentially expressed groups (LDA = 4.65). P .unadj=0.003), under which is the genus Segatella (LDA=4.62, P .unadj=0.002) and the species Segatella copri (LDA=4.60, P Both LDA (0.002) and unadj (0.002) were significantly enriched in the CON group, suggesting that this lineage made a strong contribution to the separation of the two groups. In addition, the genus *Phocaeicola* (LDA=4.61) within the Bacteroidaceae family was significantly enriched in the CON group. P .unadj=0.016) and its representative species Phocaeicolaplebeius (LDA=4.33, P .unadj=0.007) and Phocaeicola coprocola (LDA=3.95, P The *unadj* marker (LDA = 0.003) was also significantly enriched in the CON group. Besides Bacteroides-related lineages, the CON group also enriched several Firmicutes lineage markers. For example, *Clostridium* was significantly enriched in the CON group (LDA = 4.01). P.unadj=0.006). At the order level, Veillonellales and Selenomonadales also show a differential lineage within the CON group, with Selenomonadales-related entries and the genus Megamonas (mean LDA=3.51) showing the most significant differences. P The .unadj value (ranging from 0.002 to 0.001) suggests that this group has a strong discriminative contribution to the CON group. At the genus and species level, the CON group also detected differential markers such as the genus *Lachnospira* (LDA=3.42) and the genus *Lacrimispora* and its species *Lacrimispora amygdalina* (LDA=3.09).

[0150] Compared to the CON group, the differential markers in the AD group were concentrated in Actinobacteria and its downstream lineages, and also included some Firmicutes and Proteobacteria-related taxa. At the phylum level, Actinobacteria was significantly enriched in the AD group (LDA = 4.53). P .unadj=0.048). Meanwhile, a continuous stratified differential lineage was formed within the phylum Actinobacteria—Eggerthellales—Eggerthellaceae—genus *Eggerthella*—*Eggerthella lenta*, all significantly enriched in the AD group (LDA between 3.02 and 2.80). P .unadj is between 2.89 × 10 -4 Up to 4.81×10 -4 This suggests that this lineage may be an important taxonomic "marker chain" for the AD group. At the species level, *Bifidobacterium longum* is one of the most representative differential markers for the AD group (LDA=4.18). P .unadj=0.004). Among other taxa, group AD was also enriched in Klebsiella pneumoniae (LDA=3.32, PThe AD group showed differential markers such as *Hungatella* (LDA=3.21) and *Blautia faecis* (LDA=3.21) within the Firmicutes genus. Furthermore, the AD group still exhibited a variety of differentially expressed species (such as *Lactobacillus gasseri*, *Enterococcus avium*, and *Peptoniphilus lacrimalis*) in the lower LDA range, reflecting a broader range of differential phylogenetic distributions at the species level. The differential markers in the CON group were mainly dominated by Bacteroidota (Prevotellaceae–Segatella / Segatella copri; Bacteroidaceae–Phocaeicola and related species), accompanied by enrichment in some Firmicutes (such as Clostridium, Lachnospira, Lacrimispora) and Negativicutes related orders (Veillonellales, Selenomonadales / Megamonas); while the AD group was more prominently characterized by enrichment in Actinobacteria and its Eggerthellales–Eggerthella (including E. lenta) lineage and Bifidobacterium longum, accompanied by species differences such as Klebsiella pneumoniae.

[0151] 2.5 Changes in the gut microbiota symbiotic network in AD patients The topological structure of the microbial community symbiotic network visually illustrates the changes in complexity of the AD and CON microbial communities. Figure 7 Topological properties encompass the following metrics: Number of nodes: the total number of species or operational taxonomic units (OTUs) in the microbial community, reflecting diversity. Number of edges: describes the total number of interactions (connections) between species in the microbial community; Average degree (D): the average number of connections per node, reflecting the overall connectivity of the network; a higher average degree indicates denser connections between nodes in the network; the number of edges and average degree are used to describe the complexity of the symbiotic network. Positive connection percentage (%): the proportion of mutually beneficial interactions (such as symbiosis, mutualistic coexistence); Negative connection percentage (%): the proportion of competitive interactions (such as resource competition, antagonistic effects).

[0152] Compared to the control group CON, the AD group's symbiotic network exhibits structural changes characterized by "more edges, higher connectivity, shorter paths, and stronger local clustering." In terms of network size, the AD group contains 304 nodes and 867 edges, both higher than the CON group's 292 nodes and 657 edges. With a similar number of nodes, the AD group shows a significant increase in the number of edges, accompanied by an increase in average degree from 4.50 in the CON group to 5.70, indicating that each node in the AD group has an average of more connected neighbors, resulting in a denser overall network and higher complexity of potential interactions. Although both groups have a density of 0.02, the higher number of edges and average degree in the AD group, when node sizes are similar, still suggest a stronger overall connectivity level.

[0153] In terms of interaction direction, both networks are dominated by positively correlated edges, but the AD group has a higher proportion of positively correlated edges: the AD group has 839 positively correlated edges and 28 negatively correlated edges (positive correlation accounts for 96.77%), while the CON group has 594 positively correlated edges and 63 negatively correlated edges (positive correlation accounts for 90.41%). From a statistical perspective, the co-occurrence relationships in the AD group are more biased towards "positively correlated connections," while the CON group contains a relatively higher proportion of negatively correlated connections.

[0154] Regarding global network reachability, the average path length of the AD group is 3.60, shorter than the 4.02 of the CON group; the network diameter is 7.67, significantly smaller than the 12.03 of the CON group. These results indicate that, compared to the CON group, nodes in the AD group network can be connected through fewer "shortest connections," resulting in a more compact network and stronger reachability between nodes; while the larger diameter of the CON group suggests that its network has longer connection links and a relatively dispersed overall structure.

[0155] Regarding local clustering characteristics, the clustering coefficient of the AD group is 0.69, which is higher than that of the CON group (0.62). This indicates that adjacent nodes in the AD group network tend to form closed triangular structures (more obvious local clusters), and have stronger local clustering and potential modularity characteristics.

[0156] In summary, the AD group's symbiotic network exhibits a greater number of edges, higher average degree, shorter paths, smaller diameters, and stronger clustering, suggesting that its microbial community's potential interaction network is more complex and has a more compact structure. The CON group's network, on the other hand, shows a higher proportion of negative correlations and weaker global reachability, suggesting that its community interaction structure may be more dispersed and accompanied by stronger potential mutual exclusion relationships.

[0157] To identify key microbial species influencing the topology of symbiotic networks, this study classified two groups of network nodes based on the Zi-Pi method. Figure 8Zi (within-module degree) characterizes the connection strength of a node within its own module, while Pi (participation coefficient) characterizes the distribution of node connections across different modules. Nodes are classified into four categories based on classic thresholds (Zi = 2.5, Pi = 0.62): Peripherals, Connectors, Module hubs, and Network hubs.

[0158] Overall, both networks were dominated by peripheral nodes, suggesting that most microorganisms occupied peripheral positions and contributed relatively little to connectivity. However, compared to the CON group, the composition of "critical nodes" in the AD group network changed significantly: module hubs appeared in the AD group, while no module hubs were detected in the CON group. Peripheral nodes accounted for 98.36% of the AD group, while module hubs were detected at 1.32%, including Peptostreptococcaceae, Roseburia, Roseburia faecis, and its SGB4925 variant. These module hubs have significantly higher connectivity within their respective modules and are core nodes maintaining the internal structure and local stability of their modules. In contrast, the CON group had a higher proportion of peripheral nodes (99.66%) and no module hubs were detected, suggesting stronger intra-module connectivity in the AD group network. Only a very small number of connectors were detected in both groups, but the bacterial communities differed. In the AD group, only Firmicutes (p_Firmicutes) was detected as a connector (accounting for 0.33%), suggesting that in the AD state, the potential for cross-module connectivity is more reflected in the broad association of multiple modules by higher-level (phylum-level) lineages. In contrast, the connector in the CON group was a specific species, Corynebacterium hadrus (s_Anaerostipes_hadrus t_SGB4540) (accounting for 0.34%). Furthermore, no network hubs were detected in either group, indicating that both the AD and CON networks lack a "global super-hub" with both high intra-module connectivity and broad cross-module connections, and are more likely to be organized by multiple modules with intra-module connections as the primary mode of organization.

[0159] In summary, compared with the CON group, the AD group's symbiotic network exhibited two key changes at the Zi–Pi threshold: First, the AD group showed module hub nodes composed of Peptostreptococcaceae–Roseburia–R. faecis (and its SGB4925 type), suggesting enhanced centralization within network modules under AD conditions; second, cross-module connectors in the AD group were represented by phylum-level Firmicutes, while in the CON group they were represented by species-level Anaerostipes hadrus (SGB4540), suggesting that disease states may be accompanied by a shift in cross-module connectivity patterns from "species-level bridging" to "high-level phylogenetic broad associations." These results indicate that the AD-related gut microbiota interaction network not only changes in overall topological complexity but also exhibits structural reshaping in the role allocation of key nodes and modular organization.

[0160] 2.6 Pathway Analysis This study used reporter factor path analysis based on KO annotation. Figure 9 Table A (Table 11) compares the potential metabolic / regulatory capacity of gut microbiota between the atopic dermatitis group (AD) and the control group (CON) from a functional prediction perspective. The Z-scores at the KO level are aggregated and standardized within each pathway to obtain the Reporter Score; a Z-score < 0 indicates that the pathway is more enriched in AD than in CON, and a Z-score > 0 indicates that the pathway is more enriched in CON than in AD. After multiple regression analysis (BH-FDR), a batch of differentially expressed pathways (p.adjust=0.0107) were identified, all with absolute factor scores exceeding the threshold line (3.5) in the figure. Among 481 KEGG pathways, 17 significantly differentially expressed pathways were selected, representing functional modules that are significantly remodeled in AD patients.

[0161] Compared with the control CON, AD was significantly reduced in multiple basal metabolic and substrate uptake / utilization pathways. Figure 8The scores for A (Table 11) included fatty acid biosynthesis (Z-score 3.27, factor score 5.12), fatty acid metabolism (Z-score 2.04, factor score 4.24), fructose and mannose metabolism (Z-score 1.44, factor score 4.90), pentose and glucuronate interconversions (Z-score 1.33, factor score 4.01), and the phosphotransferase system (PTS, Z-score 1.76, factor score 4.76). Bacterial chemotaxis (Z-score 1.83, factor score 3.71) was also higher in the CON group. This indicates that AD-associated bacteria, at the predicted functional level, exhibited an overall weakening of the carbohydrate / lipid metabolism and transport and uptake systems, suggesting a systemic decline in their role in the "substrate acquisition-energy metabolism-biosynthesis" process.

[0162] Meanwhile, the abundance of stress regulation, transport and host interaction-related pathways was significantly increased in the AD group. (1) The potential related to environmental information processing and transmembrane transport increased, including: two-component system (Z value -2.26, factor score 3.87) and ABC transporters (Z value -3.14, factor score 3.61) were significantly increased in AD. (2) The potential related to genetic information processing and protein processing increased, with spliceosome (Z value -5.61, factor score -3.65) and ubiquitin-mediated proteolysis (Z value -5.30, factor score -3.54). (3) Increased host-interaction and disease-annotation pathways indicate that the AD gut microbiota is shifting towards an inflammation-related state, including: Shigellosis (Z-score -5.56, factor score -3.67), COVID-19 (Z-score -8.60, factor score -5.76), ALS (Z-score -6.58, factor score -3.68), MAPK signaling pathway (Yeast, Z-score -6.00, factor score -3.74), and cell cycle-related pathway (Cell cycle-yeast, Z-score -6.36, factor score -3.70). In summary, compared with CON, AD shows a systemic shift in environmental adaptation, substance transport, and core metabolism, accompanied by enhanced functions related to disease-related / host interactions, suggesting a significant functional imbalance in the AD-related gut microbiota.

[0163] At the KEGG Module level (30 significant differences in total), 3 modules remained significant (Qvalue < 0.05) after multiple correction. Figure 8 (B in the text). Among them, the module with significantly reduced AD is M00169 (CAM, light, ...). p<0.001, Q=0.05), this module suggests that the control group's microbiota is more enriched in these homologous functional components related to carboxylation-organic acid metabolism, while the AD group may show a reduction in the corresponding functional groups or a weakening of the "non-stress" part of the metabolic network. In terms of microecological homeostasis, this reduction can be regarded as a side signal of the decreased functional diversity and metabolic buffering capacity of AD-related microbiota. The module that was significantly elevated in AD was M00309 (Non-phosphorylative Entner-Doudoroff pathway, gluconate / galactonate). glycerate, p <0.001, Q=0.02), M00938 (Pyrimidine deoxyribonucleotide biosynthesis, UDP dTTP, p <0.001, Q=0.02), M00309 is the non-phosphorylated ED pathway, used to break down "sugar acid substrates" such as gluconic acid / galactonate into intermediates such as glycerol, and is an alternative carbon metabolism pathway. M00938 is the ability to synthesize deoxypyrimidine nucleotides (especially dTTP), which is primarily used for DNA replication and repair. The enhancement of the above modules in the AD group indicates that the gut microbiota plays a role in nucleotide supply and DNA synthesis / repair (such as UDP). The predicted metabolic potential of dTTP (dihydrotestosterone-to-protein ratio) and glucose-acid substrate utilization (such as the conversion of gluconic acid / galactonate to glycerol via the non-phosphorylated ED pathway) was simultaneously upregulated: the former points to a gut microbiota more inclined towards a high-turnover / high-replication or stronger repair demand, often associated with oxidative stress, nutrient niche remodeling, and opportunistic taxa expansion under inflammatory backgrounds; the latter suggests that the available substrate profile in the gut may be altered (e.g., more inclined towards oxidative sugars or mucus-associated glucose-acid substrates), thereby promoting the remodeling of microbiota carbon metabolism pathways. Overall, this combined feature supports a functional response pattern in which the gut microbiota under AD background undergoes a systematic shift from a relatively "homeostatic metabolism" to a "proliferative-stress-adaptive metabolism".

[0164] Table 11 Core Differentiation of KEGG Pathway

[0165] 3. Discussion 3.1 Species-level reorganization of AD-associated gut microbiota Recent studies on the gut microbiota of Alzheimer's disease (AD) have increasingly shown a consensus that AD does not necessarily present as a stable decline in alpha diversity. Even with the same clinical phenotype, different regions, diets, or medications can lead to inconsistent diversity directions. However, more commonly, community composition (especially at the species / strain level) undergoes reorganization, which, together with host immune bias, barrier function, and the inflammatory environment, shapes the transorganismal effects of the gut-skin axis. [97, 98] The results of this study clearly support the "microbial recombination rather than overall collapse" model caused by AD: First, at the community composition level, the two groups were generally similar in terms of dominant phyla, both being dominated by Firmicutes and Bacteroidetes, but there were systematic differences in their relative abundance patterns: compared with CON, the AD group showed a relative difference in the dominance of Actinobacteria up-regulated and Bacteroidetes down-regulated. Figure 5 In the A), and at species resolution, clearer community structure separation appears (PCoA + PERMANOVA: species). p =0.013, and the gate p =0.158). This suggests that the differences associated with AD are more concentrated in the relative shifts of several key taxa rather than complete community replacement. This difference was further amplified in species composition analysis: the AD group had more detectable species than the CON group, but the average relative abundance of the top 10 species was lower in the AD group, indicating a relatively lower concentration and more dispersed composition of dominant species in the AD group. Furthermore, although both groups shared common dominant species such as *Faecalibacterium prausnitzii* and *Ruminococcus bromii*, some species showed clear group biases. The CON group was more biased towards enrichment in *Segatella copri* and some dominant bacteria associated with short-chain fatty acids (SCFA) (such as *Anaerostipeshadrus* and *Eubacterium rectale*), while the AD group was more biased towards relative enrichment in species such as *Bacteroides uniformis* and *Bifidobacterium longum*. Figure 6 In Table B of Table 10), LEfSe further focused the differential markers in the CON group on the Prevotellaceae–Segatella lineage and classic butyrate-producing bacteria (such as Anaerostipeshadrus and Eubacterium rectale), while the AD group focused on Actinobacteria-related lineages (Eggerthellales–Eggerthella containing E. lenta) and conditionally pathogenic bacteria (such as Klebsiella pneumoniae) and other "inflammatory niche-adapted bacteria" ( Figure 6This reflects a species-level structural reorganization in Alzheimer's disease (AD). This pattern is highly consistent with the core mechanism of the "gut-skin axis" internationally: short-chain fatty acids (SCFAs) such as butyrate not only reduce systemic inflammatory load by promoting intestinal barrier function and immune tolerance (Treg / IL-10), but also directly improve epidermal barrier maturation and antigen sensitization risk; therefore, when butyrate-producing bacterial lineages are weakened in AD, the barrier-immune imbalance at the skin end is more easily amplified. At the same time, Eggerthella-associated actinomycetes can drive Th17 axis activation under various inflammatory backgrounds, providing an explainable microbial alteration to support the migration of AD from "Th2-dominated" to "Th2-Th17 parallel chronic inflammation". It is worth emphasizing that domestic and international studies often exhibit geographical and population heterogeneity in determining which specific species experienced an increase or decrease: adults versus infants, presence of gastrointestinal symptoms, dietary structure, and history of antibiotic / topical or systemic treatment can significantly alter the outcome (e.g., changes in Prevotella or Faecalibacterium are inconsistent), a point repeatedly highlighted in systematic reviews and studies. Domestic studies, such as the Guangzhou infant cohort and the "skin-gut axis" review, also emphasize the impact of geographical and exposure differences on the AD-gut microbiota association. Therefore, the value of this study lies in its ability to explain "steady-state symbiotic communities represented by butyrate / fermentation" and "inflammatory adaptive communities represented by actinomycetes-opportunistic pathogens" within the same theoretical framework, providing a clear path for subsequent causal verification using metabolites (SCFAs, indole derivatives, etc.) as a bridge.

[0166] 3.2 Functional potential of AD-related gut microbiota alterations At the functional level, this study found that the AD group's microbial community did not exhibit enhanced basal anabolism, but rather displayed a survival mode of "high stress-high turnover-strong adaptation," namely, upregulation of the two-component system and environmental sensing / transmembrane transport modules such as ABC transporters, while the pyrimidine deoxynucleotide synthesis module related to nucleotide supply / repair was enhanced. In contrast, CON was more inclined towards fatty acid metabolism, fructose / mannose metabolism, and typical fermentation and sugar uptake-utilization pathways such as PTS.

[108] This functional shift "from homeostatic fermentation to stress adaptation" is synergistically related to AD-related changes in the gut microenvironment (barrier damage, oxidative stress, and substrate profile changes): when microorganisms rely more on transport and signaling systems to compete for substrates and maintain rapid replication / repair through sugar and acid metabolism and nucleotide synthesis, the community tends to produce or release pro-inflammatory molecules (such as LPS-related signals), leading to "gut permeability."

[0167] Symbiotic network analysis results show that the AD group has a higher number of edges and average degree, a shorter average path length, a smaller network diameter, and a higher clustering coefficient. Figure 7Zi-Pi primarily positions key nodes in the AD group as module hubs (such as nodes related to Peptostreptococcaceae and Roseburia faecis) and a small number of connectors, with peripheral nodes still making up the vast majority overall. Figure 8 These network metrics collectively indicate that, under the same network parameter system, the AD group network exhibits a more "cohesive / tight" co-occurrence structure, with more active core nodes within modules. However, the positive correlation ratio of the AD group's symbiotic network is higher than that of the CON group. This suggests that in networks where competitive constraints are weakened and positive correlations dominate, system resilience may actually decrease, making them more susceptible to state transitions under perturbations. This is logically consistent with the "remission-relapse" volatility observed in AD clinical practice.

[0168] 4. Summary This study, through metagenomic analysis of the gut microbiota of patients with atopic dermatitis (AD), found that their microbial ecosystem undergoes specific changes in multiple dimensions.

[0169] 4.1 Species-level community structure reorganization: The gut microbiota of AD patients undergoes reorganization at the species level, characterized by a relative decrease in butyrate-producing symbiotic bacteria (such as Segatella copri), while the actinomycete lineage (such as Bifidobacterium longum, Eggerthella lenta) and some opportunistic pathogens are relatively enriched.

[0170] 4.2 Enrichment characteristics of key differential species: Compared with healthy controls, the AD group was specifically enriched with species related to inflammatory adaptation, such as Bifidobacterium longum and Eggerthella lenta, while the healthy control group was enriched with symbiotic bacteria related to the production of short-chain fatty acids (such as butyric acid), represented by Segatella copri.

[0171] 4.3 Increased Complexity and Vulnerability of the Symbiotic Network: The gut microbiota symbiotic network in the AD group was more complex and tightly connected than that in the control group. However, the connections in the network were predominantly positively correlated, and specific "module hub" nodes appeared. This structural change may indicate reduced network stability and is potentially associated with the variability of AD clinical manifestations.

[0172] 4.4 Shift in functional potential: The functional potential of gut microbiota in AD group showed a systematic shift. The potential in basic nutritional metabolic pathways such as fatty acid and fructose / mannose metabolism was weakened, while the potential in stress response, substance transport and proliferation-related pathways such as two-component signaling system, ABC transporter, and pyrimidine deoxynucleotide synthesis was significantly enhanced. That is, the model shifted from basic nutritional metabolism to one characterized by environmental stress adaptation and transmembrane transport.

[0173] 4.5 Synergistic effects of functional and species changes: The remodeling of the microbial symbiotic network structure exhibits topological characteristics of tighter connections but potentially lower ecological resilience. These synergistic changes in species and function collectively point to a shift in the AD gut microbiota from a "steady-state fermentation type" to an "inflammatory adaptive type".

[0174] In summary, this study provides new integrative evidence for elucidating the mechanism of gut-epidermal axis dysregulation in Alzheimer's disease (AD) from the perspectives of gut microbiota composition, function, and interaction networks.

[0175] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for characterizing gut microbiota and screening biomarkers in atopic dermatitis based on metagenomics, characterized in that, Includes the following steps: Step S1: Obtain metagenomic sequencing data from fecal samples of the subjects; Step S2: Assemble and predict genes from the metagenomic sequencing data to construct non-redundant initial Unigenes and screen out the final Unigenes; Step S3: Perform species annotation and functional annotation based on the final Unigenes to obtain the abundance information of microbial species and the abundance of species and functional pathways; Step S4: Based on the abundance of the microbial species, construct a microbial symbiotic network and analyze the topological properties of the microbial symbiotic network; Step S5: Based on the abundance of the species and the abundance of the functional pathways, screen for differentially expressed microorganisms and differentially expressed functional pathways that are significantly different between the atopic dermatitis group and the healthy control group. Step S6: Integrate the differentially marked microorganisms with differentially functional pathways to construct a combined biomarker for characterizing atopic dermatitis.

2. The method according to claim 1, characterized in that, The method for constructing a microbial symbiotic network in step S4 is as follows: Calculate the Spearman correlation coefficient between any two microbial species, and construct an undirected network based on the correlation between Spearman correlation coefficients with an absolute value greater than a preset threshold and a p-value less than a preset threshold. The topological properties include at least one selected from the number of nodes, number of edges, average degree, average path length, clustering coefficient, and network diameter.

3. The method according to claim 2, characterized in that, The preset threshold is a Spearman correlation coefficient absolute value > 0.6 and a p value < 0.

01.

4. The method according to claim 2, characterized in that, Step S4 also includes a Zi-Pi analysis of the microbial symbiotic network: Calculate the intra-module connectivity Zi and inter-module connectivity Pi of each node in the network, and identify the nodes as peripheral nodes, connection nodes, module hub nodes or network hub nodes with thresholds of Zi=2.5 and Pi=0.

62.

5. The method according to claim 1, characterized in that, In step S5, linear discriminant analysis and effect size analysis are used to screen the differentially expressed marker microorganisms, and reporter factor methods are used to screen the differentially expressed functional pathways.

6. The method according to claim 5, characterized in that, The screening criteria for the linear discriminant analysis effect size analysis were LDA > 3 and p < 0.05; The screening criteria for the reporting factor method are |Z|>1.96 and p<0.05 after correction.

7. The method according to claim 1, characterized in that, The differential marker microorganisms mentioned in step S5 include: Compared to the healthy control group, the abundance of Clostridium praosporum and Ruminococcus brucelli in the intestine of the atopic dermatitis group was reduced, while the abundance of Bacteroides monomorphosum and Bifidobacterium longum was increased.

8. The method according to claim 7, characterized in that, The differential marker microorganisms also included Bifidobacterium longum and Segardella coccidioides, whose relative abundance ratios were significantly higher in the atopic dermatitis group than in the healthy control group.

9. The method according to claim 1, characterized in that, The differentially functioning pathways mentioned in step S5 include: Compared to the healthy control group, the atopic dermatitis group showed significant upregulation of the two-component system pathway, ABC transporter pathway, non-phosphorylated ED pathway module, and pyrimidine deoxyribonucleic acid synthesis module in the gut microbiota. Significant downregulation was observed in fatty acid biosynthesis pathways, fructose and mannose metabolism pathways, and the phosphotransferase system pathway.

10. The method according to claim 9, characterized in that, The non-phosphorylated ED pathway module is KEGG Module M00309.

Images