(Questions)_ARDL I(0) and I(1) data

[Shin]

Dear all,

I would like to ask about ARDL estimation.
For instance, the model is Y= a+b1X1+b2X2+b3X3+e, and found Y and X1 is I(1) and other variables are I(0) from unit root tests.

Thus, I would like to know do we need to first difference Y and X1 before input it to "ARDL estimate equation"?
That is using model "d(Y)=a+ d(X1)+ X2+ X3" for estimation or we can apply the original model "Y= a+X1+X2+X3" for estimation.

Your response is highly appreciated.

Thanks
Shin