Metabolomics analysis of serum

Twenty chicks from each group were euthanized 24 h post-CpG-ODN treatment in the morning between 9 am to 10 a.m. Blood was immediately collected into serum tubes by severing the necks of the chicks with a sharp pair of scissors. After about 30 min of blood collection, the clotted blood samples were then centrifuged at 1000g force for 15 min, and serum was separated into 1.5 mL microcentrifuge tubes. Serum samples were stored at − 80 °C, transported on dry ice to The Metabolomics Innovation Centre (TMIC) facility at the University of Alberta in Edmonton, and stored at − 80 °C until further analysis.

Serum samples were thawed on ice and prepared in two batches according to a randomization template, with the addition of pooled samples for quality control. Plasma and serum samples contain a significant concentration of large molecular weight proteins and lipoproteins, which can seriously compromise the quality of ¹H-NMR spectra though the generation of intense, broad lines that interfere with the identification and quantification of lower abundance metabolites. Deproteinization can eliminate these peaks. Deproteinization of the serum samples was done by centrifugation and ultrafiltration using 3-kDa cut-off centrifuge filter units (Microcon YM-3; Sigma-Aldrich, St. Louis, MO), following a previously reported deproteinization procedure⁷². The deproteinized serum samples (280 μL) were then transferred to a 1.5 mL micro centrifuge tube followed by the addition of 70 μL standard NMR buffer solution (1 mM DSS (disodium-2, 2-dimethyl-2-silapentane-5-sulphonate), in 10% D₂O). These samples (a total volume of 350 μL) were then transferred to a 3 mm NMR tube for spectral analysis. All ¹H-NMR spectra were collected on a Bruker Avance III Ascend 700 MHz spectrometer with a 5 mm cryo-probe (Bruker Biospin, Rheinstetten, Germany). ¹H-NMR spectra were acquired at 25 °C using the first transient of the noesy-presaturation pulse sequence, which was chosen for its high degree of quantitative accuracy⁷³. Spectra were collected with 128 transients using a 4 s acquisition time and a 1 s recycle delay.

Before spectral analysis, all free induction decays (FIDs) were zero-filled to 240 k data points, and a line broadening of 0.5 Hz was applied. The methyl singlet of the added DSS served as an internal standard for chemical shift referencing (set to 0.00 ppm) and for quantification. All ¹H-NMR spectra were processed and imported into the Chenomx NMR Suite 8.1 software (Edmonton, Canada). The Chenomx NMR Suite software allows for a quantitative analysis of an NMR spectrum by manually fitting spectral signatures from an internal database to the spectrum. Specifically, the spectral fitting for metabolite was done using the standard Chenomx 700 MHz metabolite library. Most of the visible peaks are annotated with a compound name. Each spectrum was processed and analyzed by at least two experienced NMR spectroscopists to minimize compound misidentification and misquantification. Forty metabolites passed the NMR quality measures and underwent further statistical analysis.

The MetaboAnalyst 4.0 package⁷⁴ was used for statistical analysis of the metabolomics data. Data were log-transformed prior to univariate analysis, and also autoscaled prior to multivariate analysis. Principal Component Analysis (PCA) was performed for quality-control assessment. Partial Least Squares-Discriminant Analysis (PLS/DA) was used to classify samples and suggest potential biomarkers for treatment effect. Several univariate analysis tests were also employed. In particular, analysis of variance (ANOVA) was conducted to compare metabolite levels between all three groups, with Tukey’s HSD post-hoc analysis to indicate significant pairs. Student’s t test was used to compare between two experimental groups in terms of fold-change analysis and assessment of significance with regard to metabolite differences. In all tests, P values were further corrected for multiple comparisons by applying the Benjamini–Hochberg method of false discovery rate (FDR), and considering an FDR-adjusted P value of 0.1 as a threshold for inclusion in tables and figures. Pathway enrichment analysis was performed in the MetaboAnalyst 4.0 software on log-transformed and auto scaled data, against the Gallus Gallus pathway database. For each two-group comparison, it consisted of an ANCOVA test with the use of a relative-betweenness centrality algorithm for pathway topology analysis.

The significance of the observed differences in chick survival and the cumulative clinical score (CCS) were analyzed using GraphPad Prism 8.0 (GraphPad Software Inc., San Diego, CA) with a significance level of P < 0.05. The survival data and bacterial scores of both 1 × 10⁵ CFU and 1 × 10⁶ CFU of E. coli challenge were combined for clarity of analysis and presentation. The level of significance with regard to differences among groups in survival patterns and median survival times were analyzed using the log-rank test and chi-square statistic. Clinical scores assigned at each time point were summed up to 7 days post-challenge to generate CCS and thereby daily mean CCSs were calculated for each group. Two-way ANOVA was performed with Dunnett’s multiple comparison tests to compare the significant differences in mean CCS. For the statistical analysis on bacterial loads (quadrant streaking method), birds in each group were divided into two categories (low or high bacterial load). A chi-square test of independence was performed to examine the relationship between the CpG treatment method and the ability to recover viable bacteria based on this categorization. The results were interpreted with a statistical significance of P < 0.05.