Figure 2.

FT-IR-based spectral analysis revealed distinct chemical patterns. (a) Cecal microbial pellets from mice fed the same diet clustered together (n=12 mice per diet; trial 1 and 2). Clusters were calculated using the Ward's algorithm and vector normalized first derivatives of the spectra in the range from 3000 to 2800 cm⁻¹ and 1800 to 700 cm⁻¹. Mouse ID numbers are shown. Capital letters indicate diets (CARB, carbohydrate; HF, high-fat). (b) For each diet, one representative original spectrum is shown. For the sake of clarity, spectra were shifted vertically and the region between 1800 and 700 cm⁻¹ was expanded. Regions 1 to 4 depict water-, lipid-, protein- and carbohydrate-specific absorbance wave number ranges, respectively. New peaks arising upon changes in diet are labeled individually. (c) Principal component analysis of cecal FT-IR spectra. Input data were spectral intensities from 3000 to 2800 and 1800 to 850 cm⁻¹ (598 data points per spectrum). Data were normalized according to integrated spectra (area under the curve) and Pareto scaling. Factor 1 is a projection of mainly lipid (2850 and 2920 cm⁻¹) and protein (1539 and 1576 cm⁻¹) wave numbers. Factor 2 is a projection of mainly carbohydrate (920 to 1080 cm⁻¹) wave numbers. These factors explain 58.5% and 21.0% of the quantitative variations within the wave number span of the spectra, respectively. The mean body weight of mice monocolonized with B. thetaiotaomicron was 22.4±1.3 g (CARB) and 26.2±1.9 g (HF) after 3 weeks of experimental feeding (P=0.017; t-test). CARB, conventional mice on control diet (black dots); HF, conventional mice on high-fat diet (gray triangles); BtCARB, mice monocolonized with B. thetaiotaomicron DSM 2079^T on control diet (circles); BtHF, monocolonized mice on high-fat diet (empty triangles).