In this paper, we investigate a recent proposed model - so called the Tsallis holographic dark energy (THDE) model. In this case, we consider the non-gravitational and phenomenological interaction between dark sectors. We fit the free parameters of the model using Pantheon Supernovae Type Ia data, Baryon Acoustic Oscillations, Cosmic Microwave Background, Gamma-Ray burst and the the local value of the Hubble constant. We examine the THDE model to check its compatibility with observational data using objective Information Criterion (IC). We find that the THDE model cannot be supported by observational data. Using the Alcock-Paczynski (AP) test we check the deviation of the model compared to $\Lambda$CDM. Surveying the evolution of squared of sound speed $v^2_s$ as an another test we check the stability of the interacting and non-interacting THDE models. In addition, using the modified version of the CAMB package, we observe the suppressing the CMB spectrum at small K-modes and large scale.

We study a holographic dark energy model in the framework of Brans-Dicke (BD) theory with taking into account the interaction between dark matter and holographic dark energy. We use the recent observational data sets, namely SN Ia compressed Joint Light-Analysis(cJLA) compilation, Baryon Acoustic Oscillations (BAO) from BOSS DR12 and the Cosmic Microwave Background (CMB) of Planck 2015. After calculating the evolution of the equation of state as well as the deceleration parameters, we find that with a logarithmic form for the BD scalar field the phantom crossing can be achieved in the late time of cosmic evolution. Unlike the conventional theory of holographic dark energy in standard cosmology ($\omega_D=0$), our model results a late time accelerated expansion. It is also shown that the cosmic coincidence problem may be resolved in the proposed model. We execute the statefinder and Om diagnostic tools and demonstrate that interaction term does not play a significant role. Based on the observational data sets used in this paper it seems that the best value with $1\sigma$ and $2\sigma$ confidence interval are $\Omega_m=0.268^{+0.008~+0.010}_{-0.007~-0.009}$, $ \alpha=3.361^{+0.332~+0.483}_{-0.401~-0.522}$, $\beta=5.560^{+0.541~+0.780}_{-0.510~-0.729}$, $c=0.777^{+0.023~+0.029}_{-0.017~-0.023}$ and $b^2 =0.045$, according to which we find that the proposed model in the presence of interaction is compatible with the recent observational data.