Multi-omics study reveals gut microbiota dysbiosis and tryptophan metabolism alterations in GH-PitNET progression

The study, including all experimental procedures, was approved by the Ethics Committee of Peking Union Medical College Hospital, adhering to the guidelines of the Council for International Organizations of Medical Sciences (Ethics approval number: I-24PJ0158). All participants provided written informed consent prior to inclusion in the study, ensuring they were fully aware of the study’s aims, procedures, potential risks, and benefits. Ethical guidelines were strictly followed, including confidentiality of participant information and the option to withdraw from the study at any time without penalty.

In this study, a sample of 20 GH-PitNET patients and 30 controls was selected based on preliminary power calculations to ensure sufficient statistical power. Using G*Power software, we performed a two-tailed independent t-test with an assumed effect size of 0.9, reflecting a large effect based on similar studies. We set the significance level at 0.05 for a two-tailed test and aimed for a power level of 0.8. The analysis indicated that a minimum of 20 patients per group would be necessary to detect meaningful differences. To further account for variability and enhance the robustness of our findings, we included 20 GH-PitNET patients and 30 healthy controls. The formulas used are as follows:where Zα/2 is the Z-value corresponding to the significance level (α). For a two-tailed test at α = 0.05, Zα/2 = 1.96; Zβ is the Z-value corresponding to the power level (1−β). For a power of 0.8, Zβ = 0.84; d = 0.9, the anticipated effect size (Cohen’s d). \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$n=\frac{2\times {({Z}_{\frac{\alpha }{2}}+{Z}_{\beta })}^{2}}{{d}^{2}}$$\end{document}

Between January 2023 and June 2023, a cohort of 20 patients diagnosed with sporadic GH-PitNETs and 30 healthy control (HC) volunteers were recruited from Peking Union Medical College Hospital (Beijing, China). These individuals underwent routine medical investigations, including pituitary plain magnetic resonance imaging (MRI) and dynamic contrast-enhanced MRI. Prior to their inclusion in the study, written informed consent was obtained from all GH-PitNET patients and healthy volunteers. The researchers meticulously recorded the patients’ general clinical data, encompassing disease course (CD), age, sex, body mass index (BMI), Knosp classification, history of obstructive sleep apnea syndrome (OSAS), history of diabetes mellitus (DM), history of hypertension (HTN), growth hormone nadir during the oral glucose tolerance test (OGTT-GH nadir), GH level, and p53 and Ki67 immunohistochemical staining results. Notably, all GH-PitNET patients and volunteers were of Han Chinese ethnicity and shared similar geographic regions and dietary habits. The definitive diagnosis of GH-PitNETs in all patients was established through postoperative pathologic examination. To ensure the integrity of the study, several exclusion criteria were applied: patients aged 80 years or older or younger than 25 years, patients with multiple endocrine neoplasia type 1 (MEN1), patients who had undergone preoperative radiotherapy or octreotide treatment, patients with concurrent tumors, and patients with metastatic PitNETs. Healthy volunteers were selected based on the criteria of normal bowel habits and the absence of a history of tumors or other serious gastrointestinal diseases. Additionally, both patients and volunteers who had used antibiotics, probiotics, prebiotics, or symbiotics within two months prior to sample collection were excluded. To facilitate further experiments, all eligible fecal samples were self-collected by participants and promptly transported to the laboratory. Upon arrival, each sample was divided into three aliquots, packed into separate freezer tubes, and stored at − 80 °C.

The antibodies employed in this study included anti-phospho-ERK1/2 (Thr202/Tyr204) (Proteintech, 28733-1-AP), anti-ERK1/2 (Proteintech, 11257-1-AP), anti-phospho-CREB1 (Ser133) (Proteintech, 28792-1-AP), anti-CREB1 (Proteintech, 12208-1-AP), anti-Pit1 (Abcam, ab313642), anti-AHR (Proteintech, 67785-1-lg) and anti-β-actin (Proteintech, 66009-1-Ig). The reagents utilized in this study were 3-indoleacetic acid (IAA) (10 mM, dissolved in DMSO, MCE, HY-18569) and forskolin (1 mM, dissolved in DMSO, MCE, HY-15371).

The GH3 cell line, a rat pituitary tumor clone that synthesizes and secretes prolactin and growth hormone, was used for in vitro studies. The cell line was purchased from the National Infrastructure of Cell Line Resources (Beijing, China) and was certified by the institute. GH3 cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS) and 1% penicillin (50 U/ml)/streptomycin (50 mg/ml) in an incubator at 37 °C in a humidified environment with 5% CO₂.

Male C57 mice, 5 weeks old, were used for this experiment, housed in a controlled environment at 22 ± 2 °C with a 12-h light/dark cycle, with ad libitum access to food and water. Mice were randomly divided into two groups: a treatment group receiving indole-3-acetic acid (IAA) at a dose of 100 mg/kg (p.o.) and a control group receiving an equal volume of saline, both administered by oral gavage once daily for seven consecutive days. On the eighth day, blood was collected from the retro-orbital sinus under anesthesia and centrifuged at 3000 rpm for 10 min to separate the serum, which was then stored at − 80 °C until ELISA analysis. This experiment was approved by the Animal Welfare Ethics Committee of Beijing MDKN Biotechnology Co., LTD., and was conducted in strict accordance with the experimental animal care and use guidelines of Beijing Animal Control Committee.

All patient samples were collected during the active disease phase, specifically at the time of diagnosis, prior to any treatment interventions. First, fecal samples are homogenized in a lysis buffer containing SDS, proteinase K, and bead-beating to mechanically disrupt the cells. After incubation, the sample undergoes phenol–chloroform extraction, followed by a purification step using a silica column to isolate DNA. To assess DNA quality, we measure the A260/A280 ratio to check purity and run an agarose gel to verify integrity. We also include a final wash step to remove contaminants that might affect subsequent PCR or sequencing processes. Subsequently, the DNA was diluted to a concentration of 1 ng/µL using double-distilled water.

A specific, barcoded primer pair was used to amplify the 16S rRNA gene from distinct regions. PCR reactions were performed using 15 µL of Phusion® High-Fidelity PCR Master Mix (New England Biolabs), 2 µM of each forward and reverse primer, and 10 ng of template DNA. PCR products were mixed with an equal volume of loading dye containing SYBR Green and visualized on a 2% agarose gel. Subsequently, the PCR products were pooled and purified using a Qiagen Gel Extraction Kit (Qiagen, Germany). A TruSeq® DNA PCR-Free Sample Preparation Kit (Illumina, USA) was employed to generate sequencing libraries following the manufacturer’s protocol. Library quality was assessed using a Qubit® 2.0 Fluorometer (Thermo Fisher Scientific) and an Agilent Bioanalyzer 2100 System. Finally, the library was sequenced on an Illumina NovaSeq platform to generate 250 bp paired-end reads.

Paired-end reads were demultiplexed according to their unique barcodes, assigning each read to its originating sample. Subsequently, the paired-end reads were merged using FLASH (version 1.2.7). Quality filtering was performed using QIIME (version 1.9.1) to generate high-quality, clean sequence tags. These tags were compared against the Silva reference database (https://www.arb-silva.de/↗). Vsearch (https://github.com/torognes/vsearch/↗) was used to identify and remove chimeric sequences, resulting in a set of effective clean tags. Sequence analysis was performed using UPARSE (version 7.0.1001). Sequences with a similarity greater than 97% were clustered into operational taxonomic units (ASVs). A representative sequence from each ASV was used for annotation with the Silva Database. Alpha diversity, including the Shannon and Simpson indices, was calculated for the fecal samples using QIIME (version 1.9.0). Beta diversity analysis was also performed using QIIME (version 1.9.0). The Mann–Whitney U test was used to assess differences in alpha diversity between groups, with a significance level set at p < 0.05. The Bray–Curtis distance matrix was calculated using the ade4 R package for Principal Coordinates Analysis (PCoA), and the ggplot2 R package was used to generate the sorted plots. Permutational Multivariate Analysis of Variance (PERMANOVA) was employed to compare beta diversity. For comparing the differential Operational Taxonomic Units (OTUs) between groups, the Student’s t-test was utilized to evaluate the differences in OTUs, with a significance level set at p < 0.05. The Benjamini–Hochberg method was applied for p-value correction to identify significant OTUs. OTUs with significantly increased or decreased abundance were displayed in a volcano plot based on fold-change and p-value. Differences in the abundance of significant OTUs across different groups were shown in boxplots, all visualized using the ggplot2 R package. Using SPSS software, we calculated the Spearman correlation between OTUs and clinical indicators along with the corresponding p-values, and a heatmap was created for visualization using the heatmapR package.

One hundred microliters of blood were added to Eppendorf tubes and mixed with pre-chilled 80% methanol using a vortex mixer. The mixture was then incubated on ice for 5 min, followed by centrifugation at 15,000 × g at 4 °C for 20 min. A portion of the supernatant was diluted with liquid chromatography–mass spectrometry (LC‒MS)-grade water to achieve a final methanol concentration of 53%. The diluted samples were transferred to new Eppendorf tubes and centrifuged under the same conditions.

Analytical procedures were conducted using a Vanquish UHPLC system coupled with an Orbitrap Q Exactive HF mass spectrometer, both from Thermo Fisher Scientific (Germany), at Novogene Co., Ltd. in Beijing. Samples were introduced onto a Hypesil Gold column (100 × 2.1 mm, 1.9 μm) and separated using a 17-min linear gradient at a flow rate of 0.2 mL/min. For positive polarity, a mobile phase consisting of 0.1% formic acid in water (A) and methanol (B) was employed, while for negative polarity, 5 mM ammonium acetate (pH 9.0) in water (A) and methanol (B) were used. The gradient program involved an initial 2% B hold for 1.5 min, a linear increase to 85% B over 3 min, a ramp to 100% B over 10 min, a decrease to 2% B in 10.1 min, and a final 12-min hold at 2% B. The mass spectrometer was operated in both positive and negative ion modes with the following settings: spray voltage of 3.5 kV, capillary temperature of 320 °C, sheath gas flow rate of 35 psi, auxiliary gas flow rate of 10 L/min, S-lens RF level of 60, and auxiliary gas heater temperature of 350 °C.

UHPLC-MS/MS raw data underwent processing using Compound Discoverer 3.1 (Thermo Fisher) for peak alignment, selection, and metabolite quantitation. Parameters included a retention time tolerance of 0.2 min, a mass tolerance of 5 ppm, a 30% tolerance for signal intensity, a minimum signal-to-noise ratio of 3, and a specified minimum peak intensity. Following peak intensity normalization to total spectral intensity, molecular formula prediction was facilitated by considering additive ions, molecular ion peaks, and fragment ions. mzCloud (https://www.mzcloud.org/↗), mzVault, and the MassList database were then employed for metabolite identification and relative quantification. Statistical analysis, excluding compounds with a relative peak area coefficient of variation (CV) exceeding 30% in quality control (QC) samples, was performed using R (version 3.4.3), Python (version 2.7.6), and CentOS (version 6.6). Finally, metabolite identification and relative quantification were achieved, with annotations drawn from KEGG (https://www.genome.jp/kegg/pathway.html↗), HMDB (https://hmdb.ca/metabolites↗), and LIPIDMaps (http://www.lipidmaps.org/↗) databases. ROC curves for relevant serum metabolites were calculated using metaboanalyst (https://www.metaboanalyst.ca/↗) and visualized with pROC [1.18.0] and ggplot2 [3.3.6]. Additionally, Spearman correlation analysis was employed to assess correlations between serum metabolites and clinical indicators, with results visualized as heatmaps generated by ggplot2 [3.3.6].

CTG proliferation assays were conducted in accordance with the manufacturer’s protocol. Briefly, 5000 cells per well were seeded in 96-well plates. On the following day, cells were treated with varying concentrations of IAA. After a three-day incubation period, CTG reagent (Promega) was added to lyse the cells. Luminescence was subsequently measured using a BioTek Synergy H1 Plate Reader (BioTek, Winooski, VT).

GH3 cells were seeded in six-well plates and treated with varying concentrations of IAA for 24 h to assess cell proliferation via 5-ethynyl-2′-deoxyuridine (EdU) incorporation assay. Subsequently, cells were incubated with EdU for 2 h. Following fixation with 4% paraformaldehyde and permeabilization with 0.3% Triton X-100, cells were incubated with a pre-prepared Click-iT reaction buffer for 30 min. Finally, nuclei were counterstained with DAPI and visualized using a Cytation 7 imaging system (BioTek Instruments, Winooski, VT, USA).

Growth hormone (GH) secretion was quantified using a Rat GH ELISA Kit and a Mice GH ELISA Kit (Elabscience Biotechnology Co., Ltd.) according to the manufacturer’s protocol. Cells were resuspended in PBS containing 1% protease inhibitor, subjected to ultrasonic disruption, and centrifuged at 1000 × g for 20 min. The supernatant was collected for GH measurement. Absorbance was detected at 450 nm using a BioTek Synergy H1 Plate Reader. Protein concentration was determined using a BCA protein quantification kit (Beyotime Biotechnology) for normalization. GH in GH3 cell supernatant was assayed using a Rat GH ELISA Kit (Elabscience). Supernatants were centrifuged (1000 × g, 20 min), aliquoted, and stored at − 80 °C. Samples and standards were loaded in triplicate onto pre-coated plates. After incubation, detection antibody, HRP-streptavidin, and TMB substrate were added sequentially. Absorbance (450 nm) was measured (BioTek Synergy H1).

The collected cells were lysed in RIPA buffer supplemented with 1% phosphatase and protease inhibitors and incubated on ice for 30 min. Following centrifugation at 4 °C and 12,000 × g for 15 min, protein concentration in the supernatant was determined using a BCA protein quantification kit (Beyotime Biotechnology, Shanghai, China). Equal amounts of protein was separated by SDS-PAGE, transferred to PVDF membranes, and probed with a specific antibody. Target protein bands were visualized using a Bio-Rad chemiluminescence imager (Hercules, CA, USA) and quantified by densitometry analysis with ImageJ.

In the analysis of experimental data, all data were presented as mean ± standard deviation (SD). Student’s t-test was used to determine whether there were statistically significant differences between groups, with a significance level set at p < 0.05. Statistical results were visualized using GraphPad software. For the analysis of omics data, several R packages were utilized, including ggraph, ggClusterNet, Hmisc, and igraph, to calculate the correlations between bacteria and metabolites and to construct their respective co-abundance networks. Additionally, the igraph package was used to compute the correlations between key bacteria, metabolites, and clinical indicators. The node and edge files were imported into Cytoscape software to visualize the correlation network among the three entities. The remaining visualizations were generated using RStudio.

Furthermore, we investigated the interrelationships between the differential gut microbiota, metabolites, and clinical indicators. Based on the observed correlations between microbiota and clinical indicators (Fig. 3C), we identified five differential bacterial species significantly associated with GH levels. Subsequently, we constructed a correlation network involving these bacterial species, targeted metabolites, and clinical indicators (Fig. 6E). Notably, Intestinibacter bartlettii, Fusicatenibacter faecihominis, and Massilioclostridium coli were correlated with IAA, a metabolite also associated with GH-related clinical indicators.

Next, we investigated the effect of IAA on growth hormone secretion in vivo and vitro. As shown in Fig. 7D,E, IAA significantly stimulated growth hormone production in GH3 cells and serum of mice. Treatment with different concentrations of IAA also significantly increased the concentration of growth hormone in the supernatant of GH3 cells (Figure S1). To further explore the underlying mechanisms, we performed RNA-sequencing analysis on IAA-stimulated GH3 cells. PCA confirmed distinct gene expression patterns in IAA-treated cells (Fig. 7F). Volcano plots were used to identify differentially expressed genes (DEGs) (Fig. 7G). Enrichment analysis revealed significant upregulation of the KEGG pathways involved in growth hormone synthesis, secretion, and action (map04935), as well as the cAMP (map04024) and MAPK (map04010) signaling pathways. These findings suggest that IAA promotes GH3 cell growth hormone secretion through activation of the cAMP signaling pathway³¹ (Fig. 7H).

We further validated these findings at the protein level. Western blot analysis demonstrated that IAA upregulated the expression of AHR, p-CREB, p-ERK1/2, and Pit1. These results indicate that IAA could activate both the cAMP and MAPK signaling pathways, thereby promoting GH3 cell proliferation and growth hormone secretion (Fig. 7I,J). To further confirm the role of the cAMP signaling pathway in growth hormone secretion, we employed the cAMP agonist forskolin, which stimulated growth hormone secretion (Fig. 7K). As shown in Fig. 6L,M. BAY-218 and SQ22536 are inhibitors of AHR and cAMP respectively. The results showed that blocking AHR and cAMP could inhibit the increase of GH secretion caused by IAA, that is, IAA up-regulated AHR protein expression. Then activation of cAMP pathway promotes GH secretion.