I assume by cancer cell, you mean mammalian cells.
This is actually often a straight forward protocol if you protein is well behaved, and usually construct optimization is the key.
You should harvest the mammalian cells in a isotonic buffer.
1) Collect the cell pellet.
2) Lyse the cells in a hypotonic buffer (I usually use 20mM HEPES pH7.4/2mM MgCl2 with 1:100000 benzonase).
3) Spin the lysate at ~ 18000 rpm in ss34 rotor or similar for 15minutes.
4) Recover the pellet (you can discard the supernatant). Take a sample of the supernatant to make sure your protein is in the membrane/insoluble fraction.
5) Solubilize the pellet in the appropriate detergent - usually 1% DDM or 0.5-1%L-MNG to start (dissolved in 20mM HEPES, 150mM NaCl pH7.4). The volume should be equal to 10X the cell paste weight.
Use a dounce homogenizer to really break up the pellet - 10x pestle A and 10x pestle B. (Save 50ml of solubilization buffer for future steps)
6) Solubilize for ~2 hours - try to be consistent - do not leave overnight.
7) Spin @ 20000rpm for 20 min in ss34 rotor or similar.
8) Filter the supernatant over a glass fibre filter to get rid of any particulate or "goop" before binding to appropriate resin.
If using NiNTA - batch bind for 1-2 hours at 4 deg. If using Flag or ProtC resin, bind by gravity.
9) Wash with 0.1X solubilization buffer ~ 50mL.
10) Elute in 1mL fraction using 0.1X solubilization buffer with appropriate elution conditions.
Use a nanodrop or protein dye - to test each fraction from the presence of protein.