I think it is better to perform overexpression in low expressed cell lines and knockdown in cell lines with high expression. In your case, RGL1 is undetectable in NIH3T3 cell. So endogenous RGL1 may take no or little function in NIH3T3 cell line because its low expression. That may explain why you can observe increased cell migration when RGL1 is overexpressed but no phenotype when RGL1 is knocked down. I would choose another cell line with high RGL1 expression to perform the knockdown experiment to test whether RGL1 have pro-migration role.

If there are other family members that are functionally redundant then you likely should see if these other proteins are upregulated in RGL1's absence (via western blot).  This happens for talin 1, where siRNA knockdown of it leads to an immediate increased expression of talin 2.  I am not familiar with RGL1's complete function so it is difficult to assess with the results of a single overexpression experiment. What type of migration assay did you perform (single cell 2D tracking, wound healing assay, etc)?  I ask because wound healing can often show a false positive since cell proliferation can effect this.  I would also figure out the answer to why it increases cell migration.  Looking at upstream and downstream proteins may help elucidate the "why" question.  I would also test this effect of RGL1 in other cell types to see if it is cell-type specific.

I agree with Andrew's response. I would like to add that several controls are needed to safely conclude that "overexpression of RGL1 in primary NIH3T3 cells (not immortalized) increases cell migration". Further, one needs to know what signaling pathways regulate levels and the functions of the RGL1 in murine fibroblasts, such as NIH3T3.

@Andrew: I did sing cell 2D tracking. Somehow when I did western blot to detect endogenous RGL1, I could not detect it . However, the over-expressed RGL1 could be detected in by using the same antibody. Do you know the reason why? I think the level of RGL1 is very low so that why the anybody could not detected. If so, I dont think I could use WB to detect the reduce of RGL1 and increase of RGL2 or RGL3 when RGL1 is knocked down by siRNA.

@Divaker: I added several control like VAV2, TIAM1, GEF-H1 when I observed cell migration of RGL1. Moreover, I also used FRET biosensor to observe which small GTPase that interact with RGL1.

I think it is better to perform overexpression in low expressed cell lines and knockdown in cell lines with high expression. In your case, RGL1 is undetectable in NIH3T3 cell. So endogenous RGL1 may take no or little function in NIH3T3 cell line because its low expression. That may explain why you can observe increased cell migration when RGL1 is overexpressed but no phenotype when RGL1 is knocked down. I would choose another cell line with high RGL1 expression to perform the knockdown experiment to test whether RGL1 have pro-migration role.

Nguyen, I agree of He Jianping that if RGL1 is so low that you cannot detect it by normal western blotting that it likely is not required for migration.  I have had luck with using a 5X buffer instead of a 2X when getting cell lysates from a 10cm dish then using a fempto kit from Pierce (it worked for calcium channels). Also if there is one available a lycor could help with detection.  If you think that  this protein is normally sequestered in the cytoskeletal fraction you could also isolate the cytoskeletal/triton-X100 insoluble fraction to see.  Based on the low normal expression I would guess that it is possibly competing with another protein when overexpressed, blocking its function.  This happens with GEFs often (knock RhoA down and the GEFS activate cdc42,,etc.).  A recent paper from our lab showed without beta-pix srGAP1 goes nuts on collagen matrices (see attached). I would doubt that something similar is occurring in your system. The excess RGL1 could bind a GAP which indirectly leads to an increase in migration.

Thank you your comments He Jianping and Andrew.

@Andrew: I looked at your paper before but I didnt read it carefully. It is time to read it carefully and learn from your experiments. I  appreciate about this :)

Besides western blot, there are many other means to assess the expression profiles of RGL1-3 (Overexpression of KD), such q-PCR, IF, ELISA. Try one of this (semi)-quantitative approach to test our your hypothesis. 

I like the above answers.  I would like to add that it is important to confirm that the knockdown is complete.  Additionally, it is possible that at the basal level of this protein it is not functional, but stimulates migration when the concentration is reached to higher level.  You may want to test if the knock down blocks cell migration stimulated by any factor that might increase the level of RGL1.  

Thanks Yilin Zhang, I will do as your suggestion.

@RK Rao: could you tell me more detail about your suggestion? What should I do?

I intend to generate siRNA that can block all the family of RGL (RGL1, RGL2, and RGL3). Is it a good idea? Is it scientific sense?

Or I have to generate seperate siRNA RGL1, RGL2, and RGL3, then transfect these 3 siRNA in one cell line to make sure all of protein in RGL family is completely delepted.