Based on the radiation hydrodynamic model for black hole (BH) growth and incorporating the chemical evolution of the early-type host galaxy, we construct the coevolution model of the quasar (QSO) BH and host galaxy. As a result, it is found that after a galactic wind epoch, the luminosity is shifted from the host- dominant phase to the active galactic nucleus-dominant phase (QSO phase) in the timescale of a few 10$^8$ yr. The former phase corresponds to the early stage of a growing BH and can be regarded as a ``proto-QSO'' phase. It has observable characteristic properties as follows: (1) The broad emission lines are narrower than those of ordinary QSOs and are typically less than 1500 km s$^-1$. (2) The BH-to-bulge mass ratio, M$_BH$/M$_bulge$, is in the range of 10$^-5.3$ to 10$^-3.9$. (3) Host galaxies are bluer than QSO hosts, by about 0.5 mag in the colors B-V at the rest bands and V-K at the observed bands, with assumed galaxy formation redshifts of z$_f$=3-5. (4) The metallicity of gas in the galactic nuclei is åisebox-0.5ex~8 Z$_solar$ and that of stars weighted by the host luminosity is \i̊sebox-0.5ex~3 Z$_solar$. (5) The central massive BH (\rs̊ebox-0.5ex~=10$^7$ M$_solar$) is surrounded by a massive dusty disk (>10$^8$ M$_solar$), which may obscure the nucleus in the edge-on view and make a type 2 nucleus. By comparing these predictions with recent observations, radio galaxies are a possible candidate for proto-QSOs. Also, it is anticipated that the proto-QSO phase is preceded by an optically thick phase, which may correspond to ultraluminous infrared galaxies (ULIRGs). In this phase, M$_BH$/M$_bulge$ is predicted to be much less than 10$^-3$ and to grow with metallicity. Moreover, as precursors of ULIRGs, optically thin star-forming galaxies are predicted. These may be in the assembly phase of Lyman break galaxies (LBGs) or Ly\ensuremathα emitters.