How do I apply the bonferroni correction in hierarchical regression?

I am conducting a hierarchical multiple regression with 6 predictor variables (2 dichotomous, 1 continuous, 3 2-way interactions) entered over 4 steps. I know that I need to control for Type I error with so many predictors, and I'm thinking of using a Bonferroni correction. My question is: do I apply the adjusted p-values when determining if the overall regression test of the model is significant, or do I apply it only when determining if the t-test of the individual coefficients for each predictor variable are significant?

More specifically, to hold a family-wise alpha of .05, should I correct using .05/4 (because I have 4 steps to the regression) or using .05/6 (because I have 6 predictor variables); if it's .05/6 - should I use that for every test (i.e. to determine if the Rsquare change is significant AND to determine if the t-test of the coefficients are significant), or could I use test the overall model at .05 and the predictors at .05/6 (i.e. .008)?

Thank you in advance for any clarity and assistance you can provide!