How can I scale the predictor variables when performing multiple linear regression by using weighted least squares method?

The constant term of the predictor variables will not be constant after using weighted method, because of that, the original constant term can not be eliminated by scaling. Should I add another constant term ?