Total least square cubic splines

Question

I am interested in the non-monotonic relationship between two variables.

For this, I used cubic splines in R (mgcv::gam). However, I am not interested in predicting y using x. Rather, I want to describe their relationship, and both x and y data points are prone to errors.

Is there a simple way, in R, to fit cubic splines using the total least square approach rather than the ordinary least square one ?