Nothing at all.

In ANOVA, The Design Matrix is not full rank always, hence, not every linear combination of the parameters can be estimated by linear estimate functions of the form L'*y, where y is the observation vector and L vector belongs to the column space of design matrix , linear regression is a special case of ANOVA where the design matrix is of full column rank, hence, the all possible linear combinations of model parameters can be estimated by the linear estimate function of the above form.

Dependent variable is a variable needs to be explained/predicted/controlled by other variables whatever the statistical model/method/test is being applied.

thank you so much for replying me