Break the cells. Sonication probably best for small volumes, using washed cells resuspended to approx. 0.2-0.3 g/ml. French press works well too. Clarify by centrifugation - you have your cytosolic proteins in the supernatant. Membrane proteins are in the pellet. Resuspend this in about the same volume in ml as you started with g of cells. Detergent is then required to solubilise them for SDS-PAGE. If activity is not an issue, the detergent can be quite harsh. 1-2% Triton X-100, LDAO or one of the shorter-chain glcyosides or maltosides should work. Add this from a stock of e.g. 10% to your membrane fraction. Takes 60-90 min, stirred, to solubilise most protein. Clarify by ultracentrifugation - supernatant will be solubilised membrane protein. There will be a pellet of non-solubilised material, cell debris etc. which can be re-processed if you want to be really thorough, but may not give much more protein. Take care to use buffers compatible with your aims, use protease inhibitors, reducing agents etc as required. To make sure SDS-PAGE runs well you may need to dialyse your membrane protein sample to reduce detergent; you need enough detergent to keep the protein soluble but not so much it distorts the gel. It may take some optimising. (e.g. LDAO or Triton above 1% will likely distort the gel)