Regression trees multiple responses

As an extension of this questions in here, assuming I have:

    Xtrain = [a,b,c]; %dimension [n,3]
    Ytrain = [y1,y2]; %dimension [n,2]

I want to determine the random forest parameter using fivefold cross validation with varying max. forest depths of 5,6,7,8,9,10; varying minimum sample per leaf of 5, 10, 15, 20; and the number of forests is 5. How do I set up the trees and predictions?