I have not heard of this, though guess it might make some sense. When a table is retrieved the data must be loaded into memory on the server. If you have columns that are frequently accessed and ones that are in-frequently accessed in the same table then it is conceivable that without well-made queries that all of this would be loaded into memory. If this does happen the amount of data loaded into memory would be decreased by moving the in-frequently accessed data into another table.
I don't know exactly how MySQL does all of its searches on the very base level, but I assume that if you use the names of the columns when selecting and not *
then this problem would be minimized.
Again, I don't know the details of how all of this is coded, but I don't think it should be a problem as long as your don't select columns you don't need.