It is insoluble or associates with part of its sequence with membrane. Membrane proteins in particular can have a good affinity for particular membrane lipids, like alkylphosphocholines, phosphoethanolamines, and this is used in their crystallization, eg. of PDB entry 4TL3 with PEE ligand 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine. 

Hi there,

To go futher in Julie's way : how have you defined the truncation of your membrane protein? Are you actually sure the truncation is devoided of any transmembrane putative sequence? And even if if is the case, predicted soluble domain of a membrane protein may still exhibit a high affinity for membrane mainly through hydrophobic interaction.

Hi Victor, I suggest you may need an N-terminal exporting signal for proper folding. UCo-transformation with rare codon plasmid may be helpful.

Firstly, it's not known what initial steps have been taken.  Have you checked for inclusion bodies by phase contrast microscopy?  What ever method you use for cell breakage, did you check for complete homogenization (are there any whole cells appearing in your pellet).  The type of homogenization will influence the kinds of membrane fragments generated that in turn influence interaction with and pelleting of protein.  If you do have inclusion bodies then the protein is following the traditional behavior of over expressed recombinant proteins.  Why do you expect this protein domain to have native structure in a foreign reducing environment whether or not it is soluble?