as to the best number I can't really comment. However, there are lots of reports on silenced UAS repeats in zebrafish resulting in patchy reporting or complete silencing of the transgene. Therefore I would recommend looking into the publication from Akitake et al., here they present a 4xUAS element that carries non-repetitive UAS sequences to reduce the silencing effects observed with increasing generation numbers.
Akitake et al. 2011 "Transgenerational analysis of transcriptional silencing in zebrafish" Dev Biol. 2011 April 15; 352(2): 191–201. doi:10.1016/j.ydbio.2011.01.002
I don't have expertise on generatoin of UAS transgenic lines, but I'm currently using a Gal4 line injecting it with a tol2 UAS vector. The repeats in my vector should be(you can check in the Tol2 page, and look for p5E-UAS) 10 repeats, and it works very very good. Hope this could help you.
I got both the 10X and the 4Xnr UAS constructs, I'll try both side by side. Also there is a new 'Q' transcriptional activator system that looks like it may solve some of the silencing issues from Gal4/UAS. http://www.sciencedirect.com/science/article/pii/S1046202313002132. I'll give this a try too.
Matthew, do you have any update on this for us by any chance?
I am also trying to establish a Gal4/UAS system in our lab, we are interested in responder lines driving the expression of reporters useful for functional imaging. We have had problems establishing a certain line. The direct approach of driving expression under the control of our promoter of interest didn't work out.
Using the Gal4/UAS system and increasing transcriptional levels made a difference, we got nice expression and responses. Ideally I would like to increase the expression as much as possible but I would like to avoid silencing effects. A colleague has generated some stable 10 x UAS lines, I was told that some of them showed silencing effects after 4-6 generations.
Also, I wonder if there is any more data on the usage of KalTA4 constructs as described in:
Proc Natl Acad Sci U S A. 2009 Aug 11;106(32):13365-70. doi: 10.1073/pnas.0903060106. Epub 2009 Jul 23.
Optimized Gal4 genetics for permanent gene expression mapping in zebrafish.