$P$-wave coupled channel effects arising from the $Dbar{D}$, $Dbar{D}^*+c.c.$, and $D^*bar{D}^*$ thresholds in $e^+e^-$ annihilations are systematically studied. We provide an exploratory study by solving the Lippmann-Schwinger equation with short-ranged contact potentials obtained in the heavy quark limit. These contact potentials can be extracted from the $P$-wave interactions in the $e^+e^-$ annihilations, and then be employed to investigate possible isosinglet $P$-wave hadronic molecules. In particular, such an investigation may provide information about exotic candidates with quantum numbers $J^{PC}=1^{-+}$. In the mass region of the $Dbar{D}$, $Dbar{D}^*+c.c.$, and $D^*bar{D}^*$ thresholds, there are two quark model bare states, i.e. the $psi(3770)$ and $psi(4040)$, which are assigned as $(1^3D_1)$ and $(3^1S_1)$ states, respectively. By an overall fit of the cross sections of $e^+e^-to Dbar{D}$, $Dbar{D}^*+c.c.$, $D^*bar{D}^*$, we determine the physical coupling constants to each channel and extract the pole positions of the $psi(3770)$ and $psi(4040)$. The deviation of the ratios from that in the heavy quark spin symmetry (HQSS) limit reflects the HQSS breaking effect due to the mass splitting between the $D$ and the $D^*$. Besides the two poles, we also find a pole a few MeV above the $Dbar{D}^*+c.c.$ threshold which can be related to the so-called $G(3900)$ observed earlier by BABAR and Belle. This scenario can be further scrutinized by measuring the angular distribution in the $D^*bar{D}^*$ channel with high luminosity experiments.