I like AHiggins answer as it solves the problem in SQL as it is probably expected. But what initially stuck me when I read your question is: Why solve it in SQL at all?
It seems like the data is not originally created inside the database but somehow imported.
If I am correct - or if exporting the table, editing and re-importing - is an option then you could solve this with a regular expression. All notable text editors can do this for you (I tested this with NotePad++).
If the table is in a text file using a tabulator \t
as delimiter you could simply replace all
NULL\tNULL\t(.*)\n(.*)\t(.*)\tNULL
with \2\t\3\t\1
.