Assume that $||U||_F^2=trace(FF^T)$, the Frobenius norm. Here $F(R)=trace((A-RY)(A-RY)^T)$. Its derivative is $DF_R:H\rightarrow trace((-HY)(A-RY)^T+(A-RY)(-HY)^T)=trace(-2HY(A-RY)^T)=trace(-2Y(A-RY)^TH)$.
If you seek a gradient, then we consider the gradient $\
abla_R(F)$ associated to the scalar product $(U,V)=trace(U^TV)$. It is defined as follows: for every $H$, $(\
abla_R(F),H)=DF_R(H)$. Thus $\
abla_R(F)=-2(A-RY)Y^T$.