To further investigate the mechanism of turbulent heat transfer in channel flow under the constant wall-temperature difference (CTD) condition, this work conducts a DNS study with various Prandtl numbers (Pr = 1.0, 0.71, 0.6, 0.3, 0.1, 0.05, and 0.025) at Re = 180 using OpenFOAM. Numerical methods, mesh resolution, and computational domain size are validated by the turbulent length scales, two-point autocorrelation coefficients, and one-dimensional wavenumber spectra. The mean temperature profile, temperature variance and its budgets, and turbulent heat flux are obtained to enrich and improve the DNS database. The wall asymptotic properties and values show that the characteristics of temperature variance and turbulent heat flux under the CTD condition still satisfy the classic power law prediction. The practical physical quantities, such as the eddy diffusivity, time scale ratio, turbulent structure parameter, and turbulent Prandtl number are also gathered to verify and calibrate turbulence models and empirical relationships. Furthermore, it can be found that when Pr <= 0.1, the time-averaged results and behavior of turbulent heat transfer show a remarkable difference compared to medium-to-high Prandtl number fluids (0.3 <= Pr <= 1.0), indicating that the boundary for recognizing low-Prandtl number fluids is Pr = 0.1. This study would expect to gain insight into mechanisms and develop turbulent heat transfer models for equipment in fourth-generation advanced nuclear reactor systems.