Its ok to add 10 % FBS. FBS must be added as it contains chemoattractants and growth factors.

Either 0.4% FBS in the top, or none. Use 5% in the lower, if 5% FBS is what you usually culture your cells in. If you have highly invasive cells, sometimes you don't need a serum gradient. Also, if you are having problems with cells invading, try setting up an optimisation experiment to determine the optimal invasion time.

5% of FBS should be sufficient as a chemoattractant (10% is recommended). However, i recommend to use serum free-media in the upper chamber.

Best

The issue of gradient is discussed briefly in this 'hands on' publication. I suggest same concetnration in both wells, as the gradients are not stable over the duration of the assay. If chemoattractaion is study it preferrably should be done in a specifically designated system.

In Vitro Assessment of Human Natural Killer Cell Migration and Invasion

Chapter · January 2016

DOI: 10.1007/978-1-4939-3684-7_6

In book: Natural Killer Cells, pp.65-74

While FBS is a very common chemoattractant used in the literature, it is actually a poor and non-physiological chemoattractant. In our hands, migration to FBS takes a great deal longer than with other more physiological chemoattractants used at appropriate concentrations. For instance, many use a 24 hr chemoinvasion assay for MDA-MB-231 cells when using FBS, but we need only 3-4 hours using purified growth factors such as LPA, EGF or HGF. One of the reasons for this phenomenon is that FBS has very high levels of LPA, a bioactive lipid that contributes most of the migration and proliferative properties of FBS. When cells are treated with 1uM LPA or more, the main LPA migratory receptor, LPA1, is down regulated leaving predominantly LPA2 (which is poorly migratory in most cases) to facilitate migration. This leads to a slower migration rate seen with FBS. We published these results several years ago:

https://www.researchgate.net/publication/6336195_LPA2_EDG4_mediates_Rho-dependent_chemotaxis_with_lower_efficacy_than_LPA1_EDG2_in_breast_carcinoma_cells 

One of the things we must consider as a field is that FBS may be a reason too few of our in vitro experiments translate to the clinic. We have found that different growth factors use different combinations of signaling pathways to stimulate migration and invasion, as we've shown previously:

https://www.researchgate.net/publication/258101093_LPA_HGF_and_EGF_utilize_distinct_combinations_of_signaling_pathways_to_promote_migration_and_invasion_of_MDA-MB-231_breast_carcinoma_cells 

My recommendation is to find at physiological chemoattractant for your system and use that. If you do, you will need to coat the bottom well with an extracellular matrix such as laminin, collagen or fibronectin so that they cells will have traction and stick. FBS not only provides a chemoattractant with LPA but also provides extracellular matrices such as plasma fibronectin and vitronectin. The protocols are found in our publications above. We often use laminin or collagen as these are physiologically relevant to our cancers and we study a laminin binding integrin. I am also happy to send our protocol if needed.

If you feel you must use FBS, use a lower concentration and don't add it to the top well. The gradient of chemoattractant will be stable for around 4-5 hours. 

Article LPA2 (EDG4) mediates Rho-dependent chemotaxis with lower eff...

Article LPA, HGF, and EGF utilize distinct combinations of signaling...