What about good old size exclusion?
Sorry I did not actually address the question. My bad .
If you don't want to reinvent the wheel, you could potentially use a MonoQ or a MonoS column. Depending on the pI of your protein and the fusion you maybe able to separate them. MBP is quite acidic. If your protein is also acidic you may/may not be able to separate them on a MonoQ. If you protein is basic I would use the MonoS as MBP will have problems sticking at high pH. The fusion should stick to the MonoS through your protein. If neutral, you may have problems with ionic exchange.
If your protein of interest has an antibody or known binding partner you could make an affinity column using CNBr activated beads.
Dan
We have had success by moving the MBP (or MBP-fusion) to a C-terminal His vector (pET21 in our case, though others will probably work). We typically use the His-tag as purification and not the MBP. This will result in full length protein. Occasionally we
will observe some degradation. If so we dilute the Ni-Elution fraction 2-fold with MBP binding buffer and then flow over Amylose beads. This will separate only full length protein.
If the MBP is C-terminal, use an N-terminal His-tagged vector.
All the best!
Dan