pseudo-inverse of a large symbolic matrix

Hi guys, I'm looking for to solve a least-square solution for symbolic matrix A*x=b .A is 120x110 ,b 110x1 and the rank of A is 109. Since all matrices are symbolic,lsqr command doesn't work, I can't use qr because its not full rank. the only solution that I have found is find LU decomposition that give me full rank matrix L but U is rank deficiency matrix. x=pinv(U) *inv(L)*b but pinv command takes a lot of time and matlab can not solve that. Do you know how I can fix the problem? Any help would be much appreciated.