I would try ultrafiltration with a 300k cut off, so that most serum proteins stay in the serum and all aggregates (including HDL, LDL, VLDL) are filtered away. Sartorius Centrisart filters work best for this purpose but you can only filter 2.5 ml at a time.
The lipoprotein-free layer was dialyzed extensively against PBS-Ca at 4ºC
however In some literature, separation was achieved by a discontinuous density gradient centrifugation between the density range of 1.006 and 1.30 g/ml.
my purpose is only to get serum which deplete lipoprotein(in other words,don't collect any lipoprotein). so would i also take a discontinuous density gradient centrifugation or just homogeneous density(1.21g/ml)?
In my experience, the best way to get serum lipoproteins free (LPDS) is a fixed ultracentrifugation ( not gradient) with KBr concentration corresponding to a density slightly higher than 1.21 g / ml. The infranatant of the UC should be thoroughly dialyzed against buffer that interests you. In a SW41 rotor type, a suitable protocol is to place in the bottom of the tube 3 ml serum supplemented with 1.150 g of KBr (can double the amount of serum and KBr to even double the amount mentioned above) and complete the tube volume with KBr of a similar density (1,21-1,23 g/ml). Centrifuge at 39000 rpm for 22 h is sufficient to float all lipoproteins and can recover about 3.5 ml of the bottom of the tube as LPDS. This procedure is even more effective if we use a vertical o near vertical rotor with a UC time significantly minor (compare factor k of rotors )
Ultracentrifugation and KBr density adjustment is the best option. Alfa Aesar in the United States sells humand and bovine LPDS for extremely competitive prices
To my understanding, the main purpose of a discontinuous density gradient is to separate the lipoproteins into their classes, and then extract them for downstream application. The method requires a very high g force, or then very long run time. However, if your purpose id to prepare the lipoprotein-free serum would, then the more appropriate (and less laborious) way might be what Erik already suggested, which is using a suitably small pore-size filter to get rid of the proteins.
We have had some trouble with lipoproteins during ultrafiltration, as they tend to stick or block the filter. For this overnight treatment with PEG-800, or some inert material of your choice, might be necessary (check passivation protocol from Merck)