If you want to analyse a protein that is present in say, for instance, a skeletal muscle of a cow, it is usually not very practical to put the whole cow onto the gel (no, I have not tried that, I’m just making a qualified guess here…). So, firstly, a sample needs to be prepared from the tissue. I will fast-forward through the part where the cow is slaughtered, because I 1) is not very familiar with the details, 2) feel sorry for the cow and would like to pretend that this step doesn’t really occur, and 3) if any cow should happen to be reading this, it may consider it offensive.  

So anyway, fast-forwarding to where you have a small piece of tissue prepared – it needs to be broken down further. This can be done by a number of different methods, for instance homogenisation or sonication. After that, the sample needs to be centrifuged in order to separate cytosolic and nuclear fractions from each other. 

In the same way, if you have a cell culture, you need to get rid of the cell membrane. This can be done by either of the mechanical methods mentioned above, but is also frequently done by for instance lysis – a method in which you dissolve the membrane by disrupting the osmotic balance by altering the electrolytic conditions, sometimes combined with the use of a detergent. 

I am not going to recommend a certain method over another. Which method that is most preferable depend on the nature of the sample. Proteins can very easily by damaged or broken down, so it is important to choose a method that does not interfere with the protein of interest. Obviously, the purification will not be very successful if the protein of interest have been distroyed in the earlier preparation step.