Intestinal microbiota-metabolite-gene network early warning method for breast cancer bone metastasis

Abstract

The application discloses a breast cancer bone metastasis intestinal flora-metabolite-gene network early warning method, and belongs to the technical field of biomedical and tumor risk prediction. The method collects intestinal contents and tumor tissue samples, respectively carries out microbial sequencing, metabolomics detection and gene expression analysis, carries out cross-omics correlation analysis, constructs a ternary correlation network with intestinal microorganisms as one end, metabolites as intermediate connection and host genes as the other end, and screens key early warning markers. The results show that Subdoligranulum, LysoPC22: 6 and Cd300lb and / or Nrros gene can form a breast cancer spinal bone metastasis early warning marker combination as core correlation nodes. The application integrates multi-omics information of intestinal flora, metabolites and host genes, realizes early identification and risk stratification of breast cancer bone metastasis, and provides a basis for clinical screening and intervention.

Description

Technical Field

[0001] This application relates to the fields of biomedical and tumor risk prediction technology, specifically a gut microbiota-metabolite-gene network early warning method for breast cancer bone metastasis. Background Technology

[0002] Breast cancer is a prevalent malignant tumor among women worldwide, and the spine is one of the most common sites of bone metastasis. Bone metastasis not only significantly reduces patient survival but also causes clinical complications such as bone pain, pathological fractures, spinal compression, and other serious skeletal-related events (SREs). However, the regulatory mechanisms behind spinal metastasis remain poorly understood, especially the role of the microbiota in influencing systemic tumor progression. Currently, there is a lack of highly specific, early, and non-invasive diagnostic and predictive methods for breast cancer spinal metastasis, relying heavily on imaging examinations for early warning. Studies have shown that the gut microbiota plays a crucial role in tumor development, progression, and metastasis. The microbiota can influence tumor progression by regulating host metabolism, immune responses, and inflammatory responses. However, traditional single-omics studies cannot fully elucidate the multi-level regulatory mechanisms of spinal metastasis. Therefore, there is an urgent need to establish an early warning model for breast cancer spinal metastasis based on a gut microbiota-metabolite-gene association network to fill this clinical gap.

[0003] Furthermore, this invention compares orthotopic breast cancer models, spinal metastasis breast cancer models, and sham-operated control models within the same experimental system, simultaneously analyzing gut microbiota, metabolites, and host gene expression. This allows common changes associated with spinal metastasis to be isolated from model differences, providing a unified comparative basis for subsequent construction of a ternary association network. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a gut microbiota-metabolite-gene network-based early warning method for breast cancer bone metastasis, which integrates multi-omics detection results, performs risk stratification, and supports subsequent target screening, given that existing imaging examinations are insufficient for early warning and single-omics studies are inadequate to reflect multi-level regulatory relationships.

[0005] By gaining a deep understanding of the current state of research on breast cancer spinal metastasis and integrating it with clinical problems, this study aims to construct a cross-omics framework that links gut microbiota, metabolites, and host gene expression to achieve early, non-invasive, and precise assessment of breast cancer spinal metastasis. This invention utilizes nude mouse orthotopic breast cancer (BC) and breast cancer spinal metastasis (BCSM) models. Gut contents and tumor tissue were collected and subjected to: 16S rDNA gene sequencing (microbiota analysis), non-targeted metabolomics detection, and transcriptomics sequencing. Through differential analysis, pathway enrichment analysis, correlation analysis, and Lasso regression analysis, metabolites, genes, and microorganisms that showed significant changes in the metastatic group were screened, and a gut microbiota-metabolite-gene association network was constructed. To achieve the above objectives, the specific technical solution of this invention is as follows: In this context, group BC represents the in situ breast cancer model, group BCSM represents the spinal metastasis breast cancer model, and group Sham represents the sham-operated control model where only surgical procedures were performed without implantation of tumor microparticles. Unless otherwise specified, the group names in the following examples are used in accordance with the above meanings.

[0006] In a first aspect, this invention provides a gut microbiota-metabolite-gene network early warning method for breast cancer bone metastasis, comprising the following steps: constructing a breast cancer orthotopic transplantation model, a breast cancer spinal metastasis model, and a sham surgery control group model, and verifying successful spinal metastasis through bioluminescence imaging or X-ray imaging. Further, 16S rDNA gene sequencing is performed on the gut contents to obtain microbial community composition data; non-targeted metabolomics detection is performed on the gut contents and tumor tissue respectively to obtain metabolite expression profiles; transcriptome sequencing is performed on the tumor tissue to obtain host gene expression profiles. Further, intersection analysis is performed on differentially expressed metabolites identified in tumor tissue and gut contents to screen for metabolites that are co-upregulated or co-downregulated in all comparison groups, including LysoPC22:6, SQDG (16:0 / 22:6), etc. Further, a ternary association network centered on key metabolites and connecting key microbial genera and key host genes is constructed to identify potential biomarkers or intervention targets for breast cancer spinal metastasis.

[0007] The intersection analysis used common changes across three comparison groups as the screening criteria: comparison of intestinal contents between Sham and BCSM, comparison of intestinal contents between BC and BCSM, and comparison of BC and BCSM in tumor tissue. The resulting common metabolites served as intermediate nodes connecting the microbiota and host genes.

[0008] Secondly, the present invention provides detection reagents and analysis modules for constructing the above-mentioned correlation network, including sample extraction reagents, sequencing primers and data analysis programs.

[0009] Thirdly, this invention provides the application of the gut microbiota-metabolite-gene association network in early warning of spinal metastasis in breast cancer, drug target screening, and intervention strategy development.

[0010] In practical use, the abundance of bacterial genus, expression levels of metabolites, and expression levels of host genes obtained from the sample to be tested are input into the analysis module according to the format required by the model. The total score is calculated by the Logistic regression model, and the total score is mapped to the risk probability through Nomogram. The web tool is an interactive implementation of this model.

[0011] Beneficial effects 1. For the first time, a cross-omics association network related to spinal metastasis in breast cancer was constructed, integrating information from three levels: gut microbiota, metabolites, and host genes, revealing the key role of the glycerophospholipid metabolic pathway.

[0012] 2. A combination of biomarkers with high specificity and sensitivity was provided. Experimental data analysis revealed differentially expressed genes highly correlated with 9930111J21Rik2, Cd300lb, Enho, Fgf10, Nrros, Zic1, and the key bacterial genus g__Subdoligranulum, which can be used for early identification and prognosis of spinal metastases.

[0013] 3. It enables non-invasive testing. Fecal samples can be used to simultaneously detect microbial and metabolite indicators, avoiding the invasiveness of tissue biopsies and facilitating clinical application. Attached Figure Description

[0014] Figure 1 This is an experimental flowchart for constructing a nude mouse model of breast cancer and spinal metastasis; Figure 2 It is a heatmap of the expression of metabolites co-regulated by tumor tissue and intestinal feces; Figure 3 These are the Lasso regression cross-validation error plot and the Lasso regression gene coefficient path plot; Figure 4 It is a nomogram model nodal plot; Figure 5 It is a multi-omics integration framework that connects the microbiome, metabolites, and host genes in spinal metastases of breast cancer. Detailed Implementation

[0015] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0016] Example 1: Construction of a spinal metastasis model for breast cancer 1.1 Laboratory Animals and Cells Five- to six-week-old female BALB / c-nu nude mice were housed under SPF conditions. MDA-MB-231 breast cancer cells stably expressing green fluorescent EGFP (5 × 10⁻⁶) were used. 7 (cells / mL).

[0017] In this embodiment, the animals, cells, and inoculation conditions were kept consistent. The difference between the BC group and the BCSM group was the implantation site of the tumor microparticles and the corresponding model construction method. The Sham group only underwent the corresponding surgical procedure to provide a surgical factor control.

[0018] 1.2 Orthotopic model of breast cancer (BC group) The breast cancer cells were injected into the right mammary fat pad of 5-week-old supine nude mice in the breast cancer group using a syringe. The injected cell volume was 1×10⁻⁶. 7 A breast cancer nude mouse model was successfully constructed after 6 weeks using cells / 200μL, characterized by a localized mass in the right breast. Tumor tissue was then aseptically removed and minced into a paste-like state in a culture dish using a scalpel. Tumor particles with a diameter less than 1mm were immersed in a tumor cell basal culture medium and used to construct breast cancer orthotopic implantation models and breast cancer spinal metastasis models. 1.3 Breast Cancer Spinal Metastasis Model (BCSM Group) The backs of 5-week-old nude mice with spinal metastasis of breast cancer and nude mice in the Sham group were cut open with a sterile scalpel to expose the lumbar spine. The cortical bone of the vertebral lamina was scraped off to expose the medulla. A small hole was prepared by drilling a needle, and 1 mm breast cancer tumor microparticles were implanted into the small hole prepared on the vertebral lamina of the nude mice with spinal metastasis of breast cancer.

[0019] After implantation, the wound was sutured and the animals were observed under the same feeding conditions. The establishment of the spinal transfer model was confirmed by bioluminescence imaging or X-ray imaging, and the confirmed animals were used for subsequent sample collection.

[0020] 1.4 Sham surgery group (Sham group) The procedure was performed without implanting tumor particles; the wound was then sutured. Therefore, the Sham group corresponded to the breast cancer spinal metastasis group in terms of surgical exposure and wound management, without introducing tumor microparticles.

[0021] Example 2: Sample Collection and Preservation Euthanasia was performed on mice with complete hind limb paralysis. The intestinal contents of the nude mice were rapidly and aseptically collected and preserved in glycerol to preserve tumor tissue (including carcinoma in situ and spinal cord metastases).

[0022] The source of each sample was recorded during collection. Intestinal contents were used for 16S rDNA sequencing and metabolomics detection, while in situ carcinoma tissue and spinal metastasis tissue were used for metabolomics detection and transcriptomics sequencing, thus ensuring that different omics data could be analyzed according to the sample source.

[0023] Example 3: 16S rDNA gene sequencing and microbiome analysis 3.1 DNA Extraction and PCR Amplification Total DNA was extracted from intestinal contents using the CTAB method. The V3-V4 region was amplified using primers 341F (5'-CCTACGGGNGGCWGCAG-3') and 805R (5'-GACTACHVGGGTATCTAATCC-3'). After purification, the PCR products were used to construct libraries and sequenced at 2×250 bp using a NovaSeq 6000 sequencer.

[0024] 3.2 Data Processing Primer sequences and balancing base sequences were removed from the original data using cutadaptv1.9, tag sequences were flash-merged, quality filtering was performed using fqtrim (100bp window, average Q < 20 truncation), and chimeras were removed using Vsearch. ASVs were obtained by DADA2 noise reduction using QIIME2, and species were annotated using the silvav138 database (confidence level 0.7).

[0025] After the above quality control, splicing, dechimeric removal and noise reduction, ASV was used as the subsequent abundance analysis unit, and species annotation was performed based on the silvav138 database; the annotation confidence level was set to 0.7.

[0026] Example 4: Non-targeted metabolomics analysis 4.1 Sample Pretreatment Weigh 50mg of intestinal contents or tumor tissue, add 500μL of pre-cooled 80% methanol, grind with steel balls, incubate at -20℃ for 30min, centrifuge at 20000g for 10min at 4℃, centrifuge again for 5min and collect the supernatant.

[0027] 4.2 Chromatographic-Mass Spectrometry Conditions A Waters ACQUITY UPLCH SST3 column (100 mm × 2.1 mm, 1.8 μm) was used. Mobile phase A consisted of 5 mM ammonium acetate + 5 mM acetic acid + water, and mobile phase B consisted of acetonitrile. Gradient elution was performed as follows: 0–0.8 min B 2%; 0.8–2.8 min B 2%–70%; 2.8–5.6 min B 70%–90%; 5.6–6.4 min B 90%–100%; 6.4–8.0 min B 100%; 8.0–8.1 min B 100%–2%; 8.1–10.0 min B 2%. Flow rate… 0.35 mL / min, 4 μL injection, column temperature 40℃.

[0028] TripleTOF6600 mass spectrometer, ESI source, positive and negative ion modes. Positive ion voltage +5000V, negative ion voltage -4500V; curtain gas 30psi, Gas1 / Gas260psi; TOFMS scan range 60-1200Da (150ms), IDA scan range 25-1200Da (30ms), dynamic exclusion 4s.

[0029] 4.3 Data Processing and Differential Metabolite Screening XCMS was used for peak detection and alignment, while MetaX was used for missing value filtering (removing peaks missing in >50% QC or >80% of samples), KNN filling, PQN normalization, and QC spline correction. Metabolite identification was performed using KEGG and a self-built secondary spectral library.

[0030] Differential metabolite screening criteria: |log2FC|>0.263, P<0.05, VIP>1 (PLS-DA model). 420 differentially expressed metabolites were identified in the intestinal contents of Sham vs BCSM (173 up, 247 down); 87 in the intestinal contents of BC vs BCSM (36 up, 51 down); and 149 in the BC vs BCSM tumor tissue (77 up, 72 down). Three metabolites (LysoPC22:6, BMP44:12, and SQDG38:6) were upregulated in all three comparison groups.

[0031] The above screening criteria were used for both orientation determination and statistical significance determination; the commonly upregulated LysoPC22:6, BMP44:12 and SQDG38:6 were retained as candidate metabolites after cross-sample source comparison, among which LysoPC22:6 was further used to associate microbial genera and host genes.

[0032] Example 5: Transcriptomics Analysis 5.1 RNA Extraction and Library Construction Total RNA was extracted from tumor tissue using the TRIzol method, with a RIN > 7.0. mRNA was isolated using Dynabeads Oligo(dT), fragmented, and then reverse transcribed into cDNA. A second-stranded RNA containing dUTP was synthesized, terminated with A-terminal repair, ligated with adapters, and degraded using UNG enzyme. PCR amplification was performed for 8 cycles, and fragments of 300 ± 50 bp were selected using AMPureXP. Sequencing was performed using an Illumina NovaSeq 6000 PE150.

[0033] 5.2 Data Processing and Differential Gene Analysis FastP quality control, HISAT2 alignment to ftp: / / ftp.ensembl.org / pub / release-107 / fasta / homo_sapiens / dna / , StringTie assembly and FPKM calculation. DESeq2 analysis of differentially expressed genes, standard: |log2FC|>1 and corrected P<0.05. A total of 619 differentially expressed genes were identified (228 upregulated and 391 downregulated). Functional enrichment analysis was performed using clusterProfiler, with GO and KEGG pathways considered significant at P<0.05.

[0034] After obtaining the gene expression matrix, candidate genes are first identified based on the differential expression threshold, and then the functional background of the candidate genes is interpreted in conjunction with the GO and KEGG enrichment results. The above functional analysis is used for auxiliary explanation and does not replace the subsequent Lasso regression screening.

[0035] 5.3 Lasso Regression Screening of Key Genes The correlation between LysoPC22:6 and all differentially expressed genes in tumor tissue was calculated (Spearman, |cor|>0.7, P<0.05), yielding 289 related genes. Similarly, 74 related genes were obtained from fecal intestinal samples. The intersection of these two analyses yielded 33 co-regulated genes. Lasso regression analysis was performed on these 33 genes, and six key genes were selected based on the minimum cross-validation error (Lambda.min): 9930111J21Rik2; Cd300lb; Enho; Fgf10; Nrros; and Zic1.

[0036] This step first obtains differentially expressed genes associated with LysoPC22:6 from two sample sources: tumor tissue and intestinal feces. Then, the bias caused by the source difference is reduced by taking the intersection. Finally, the key gene to be included in the model is determined by Lambda.min, which has the smallest cross-validation error.

[0037] 5.416S genus screening The correlation between LysoPC22:6 and the abundance of genus-level flora in fecal gut was calculated (Spearman, |cor|>0.5, P<0.05), revealing 10 significantly correlated genera. Further calculations of the correlation between the aforementioned 6 key genes and each genera showed that g__Subdoligranulum was significantly correlated with all 6 genes (|cor|>0.5, P<0.05), and therefore included it in the model.

[0038] Therefore, the genus Subdoligranulum simultaneously meets the screening criteria for correlation with core metabolites and six key genes, and is regarded as a representative key microbial genus in the ternary association network.

[0039] Example 6: Construction of Multi-omics Connection Networks 6.1 Construction of the Prediction Model A logistic regression model was constructed based on the metabolite LysoPC22:6; genes 9930111J21Rik2, Cd300lb, Enho, Fgf10, Nrros, Zic1; and bacteria Subdoligranulum to predict the risk of spinal metastasis in breast cancer. A nomogram was constructed using R packages, where each variable in the model corresponds to a score. The scores of all variables are added together to obtain the total score (TotalPoints). The total score is mapped to the risk axis to obtain the predicted probability of spinal metastasis in breast cancer in this sample.

[0040] In the model calculation, metabolites, genes, and bacterial genera are used as input variables; each variable corresponds to an independent score in the nomogram. The scores are read according to the variable test values ​​and summed, and the resulting TotalPoints are then mapped to the risk axis. This calculation process is consistent with the Logistic regression model.

[0041] 6.2 Model Application and Web Tools Deploy the Nomogram model described above as an interactive web application (based on Shiny) at: https: / / opmodel16smetrna.shinyapps.io / MD_mdel / . After the user inputs the test values ​​for each variable and clicks the "Predict" button, the system automatically calculates the total score and outputs the probability of developing spinal metastasis in breast cancer, while also displaying the ranking of the contribution of each variable to the risk.

[0042] The web application serves only as a visualization and interactive execution interface for the model. The model variables, calculation logic, and output meanings are consistent with those in Example 6.1. After the user inputs the detection value, they can obtain the total score, prediction probability, and variable contribution ranking.

[0043] Example 7: Statistical Analysis All data were analyzed using R software. Student's t test or Wilcoxon's rank-sum test was used for comparisons between two groups. Tukey's HSD was used for comparisons between multiple groups after ANOVA. Spearman's analysis was used to assess correlation. The significance level was set at α=0.05.

[0044] In statistical analysis, the appropriate test method is selected for continuous variables based on the data distribution and the number of comparison groups. The Spearman method is used uniformly for correlation analysis, and the significance level is set at α=0.05.

[0045] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for early warning of gut microbiota-metabolite-gene network in breast cancer bone metastasis, characterized in that, Includes the following steps: S1. Construct animal models of breast cancer spinal metastasis, orthotopic breast cancer animal models, and sham-operated control animal models, and verify the breast cancer spinal metastasis animal models through imaging. S2. Collect intestinal contents and tumor tissue from each model animal; S3. Perform 16S rDNA gene sequencing on the intestinal contents to obtain intestinal microbial community composition data; perform non-targeted metabolomics detection on the intestinal contents and tumor tissue to obtain metabolite expression profiles; perform transcriptome sequencing on the tumor tissue to obtain host gene expression profiles. S4. Perform differential analysis on the microbial community composition data, metabolite expression profiles and host gene expression profiles to screen for microorganisms, metabolites and host genes that are commonly different between the breast cancer spinal metastasis model and the breast cancer in situ model and the sham surgery control model. S5. Using the differentially identified metabolites as the core, conduct correlation analysis on differentially identified microorganisms, differentially identified metabolites and differentially identified host genes, and construct a ternary association network of gut microbiota-metabolites-host genes. S6. Based on the ternary association network, key microorganisms, key metabolites and key host genes are screened to construct an early warning model for breast cancer bone metastases and output the predicted probability of breast cancer bone metastases.

2. The method for early warning of gut microbiota-metabolite-gene network in breast cancer bone metastasis according to claim 1, characterized in that: The common differential metabolites obtained in S4 include LysoPC22:6, BMP44:12 and SQDG38:

6.

3. The method for early warning of gut microbiota-metabolite-gene network in breast cancer bone metastasis according to claim 1, characterized in that: In S5, the key microorganisms include the genus Subdoligranulum, and the key host genes include 9930111J21Rik2, Cd300lb, Enho, Fgf10, Nrros, and Zic1.

4. The method for early warning of gut microbiota-metabolite-gene network in breast cancer bone metastasis according to claim 1, characterized in that: The correlation analysis in S5 uses Spearman correlation analysis, with the absolute value of the correlation coefficient being greater than 0.5 and the p-value being less than 0.05 as the screening criteria.

5. The method for early warning of gut microbiota-metabolite-gene network in breast cancer bone metastasis according to claim 1, characterized in that: The core association axes of the ternary association network include the associations between the genus Subdoligranulum, LysoPC22:6, and the Cd300lb and / or Nrros genes.

6. The method for early warning of gut microbiota-metabolite-gene network in breast cancer bone metastasis according to claim 1, characterized in that: The key host genes were screened using Lasso regression, and host gene feature combinations were established based on the screening results.

7. The method for early warning of gut microbiota-metabolite-gene network in breast cancer bone metastasis according to claim 1, characterized in that: The 16S rDNA gene sequencing used 341F and 805R primers to amplify the V3-V4 variable region, and paired-end sequencing was used to obtain microbial community composition data; the non-targeted metabolomics detection was performed using liquid chromatography-mass spectrometry; the transcriptome sequencing was performed using the Illumina sequencing platform, and differential expression analysis was performed using DESeq2.

8. The method for early warning of gut microbiota-metabolite-gene network in breast cancer bone metastasis according to claim 1, characterized in that: In S1, spinal metastases of breast cancer are verified by bioluminescence imaging and / or X-ray imaging; In S2, euthanasia is performed when the hind limbs of breast cancer spinal metastases become completely paralyzed, and intestinal contents, in situ carcinoma tissue, and spinal metastases are aseptically collected.

Images