I would suggest that the preferable term depends on whether you are talking about the cells or talking about the modifications.
In other words, I would refer to the NCAM2 expression package as the "heterologous expression system", which is then integrated into the CHO platform in order to create a "heterologous cell model."
My reason for this suggestion is that the expression package is the core construct without which there is no expression, and that this package might potentially be transferred into other cell lines to construct other cellular models.