Because the line passes through $(\bar{x}, \bar{y})$

Because the least-squares method minimizes the sum of squared errors

Because the slope $\beta_1$ is calculated using the covariance of $x$ and $y$

Because the model assumes errors are normally distributed