The ability to control the timing of transgene expression in specific tissues is a powerful tool in the study of developmental biology, cell biology, and physiology. Ever since the potential of the bacterial Tet-controlled system was first harnessed for use in other cells and organisms there have been many modifications to make control tighter while maintaining the potential for high levels of expression 
. In order to achieve control of the induction of transgene expression in zebrafish rod photoreceptors, we generated stable Tet-On transgenic driver lines using the previously established Tol2-mediated transgenesis method 
. In our constructs, we used the commercially available, second generation modified components of the Tet-On system to create a self-detecting construct that directed Dox-dependent expression exclusively in rods of Tg(Xla.rho:rtTA, TRE:GFP)
larvae and adult zebrafish and in Tg(Xla.rho:rtTAflag)
zebrafish. The amount of transgene expression in the absence of Dox in Tg(Xla.rho:rtTA, TRE:GFP)
was negligible since less than 0.5% of rods expressed GFP in the Tg(Xla.rho:rtTA, TRE:GFP)
larvae. We analyzed the length of Dox treatment required to induce transgene expression in larvae and found that by 24 h greater than 80% of rods expressed GFP in Tg(Xla.rho:rtTA, TRE:GFP)
larvae and Dox-induced expression had reached a plateau by about 24 h of treatment. We also showed that rtTA in Tg(Xla.rho:rtTA, TRE:GFP)
effectively directed Dox-dependent transgene expression in rods from an injected TRE
-containing plasmid as well as from an independent stable transgenic line.
We created the Tg(Xla.rho:rtTA, TRE:GFP) transgenic line as a self-reporter of Dox-induced expression to allow us to examine the tightness of Dox-dependent transgene expression and the kinetics of GFP expression in the entire population of rods. This line is also useful for inducing an epitope-tagged transgene of interest by either injecting a plasmid with the transgene driven by the TRE or by crossing to a stable transgenic line with the transgene driven by a TRE. In this system, the GFP expression can be used to examine the morphology of the tagged transgene-expressing rods and an antibody against the epitope tag can be used to confirm individual rod expression of the transgenic protein. Alternatively, the transgenic protein of interest could be fused to a fluorescent protein to confirm individual rod expression in mosaics or transgenics.
Confocal microscopy is typically used to image, at most, a triple-labeled fluorescent image and thereby produce a three-color digital image with the red, green, and blue colors used to represent the labeling of the fluorescent probes. While the Tg(Xla.rho:rtTA, TRE:GFP) line is very useful, the ability to examine other proteins or markers is limited when the green color is utilized to display the GFP reporter of rtTA function. This limitation led us to design a second construct, Xla.rho:rtTAflag, and create a second transgenic line, Tg(Xla.rho:rtTAflag), thereby making the green color available for FITC-labeled antibodies or GFP fusion proteins. We showed that tagging rtTA with the FLAG epitope did not affect rtTA activity. In addition, we made an HA-tagged rtTA driver construct that was tested and found to be functional (results not shown). By tagging the rtTA protein with FLAG, we were able to use an anti-FLAG antibody to identify germline founders that expressed rtTAFLAG in nearly all rod photoreceptors. The Tg(Xla.rho:rtTAflag) line is especially useful if a transgene of interest cannot be tagged or fused to a fluorescent protein without losing its activity. The GFP in the biTRE response vector can be used as a surrogate for expression of the transgene in individual rods. Further, GFP from the biTRE vector, which also drives the transgene of interest, can be used to examine the morphology of the transgene-expressing rods. Finally, this system allows for the two remaining colors – red and blue – to be used to assess endogenous proteins of interest, which may reveal important cellular processes affected by expression of the transgenic protein.
We detected no detrimental effects on rod survival or changes in rod cell morphology in Tg(Xla.rho:rtTA, TRE:GFP) or Tg(Xla.rho:rtTAflag) or in rods expressing the response constructs (TRE:nls-mCherry or biTRE: EGFP, nls-mCherry) in the two driver transgenic lines. This observation suggests that although these rods express several transgenes (rtTA+GFP+nlsCherry or rtTAFLAG+ EGFP+nls-mCherry), this extra protein expression is tolerable. We examined whether we could manipulate transgene expression levels by altering the concentration of Dox, however we saw no indication that this would be possible. As we decreased the concentration of Dox in Tg(Xla.rho:rtTA, TRE:GFP) larvae treated from 3–6 dpf, the number of GFP-expressing rods decreased but those that expressed GFP appeared to have similar levels of fluorescence to those treated with 10 µg/ml Dox as observed by confocal microscopy (unpublished observation). Furthermore, the experiments presented here represent Tg(Xla.rho:rtTA, TRE:GFP) individuals from the fifth and sixth generations with no variegation or decreased expression, indicating that gene silencing in later generations has not proven problematic.
We have not fully examined the reversibility of Dox-induced transgene expression after removal of Dox because we could not foresee a need in our studies to turn off transgene expression in zebrafish rods. As the larval eye grows (and to a lesser extent, the adult eye) new rod photoreceptors are added to the expanding retina 
. Therefore, examination of rods after induction of transgene expression and removal of Dox would be complicated by the inability to distinguish between rods that had expressed the transgene and turned it off from rods that had been generated during the post-Dox period and had never expressed the transgene. Nonetheless, we have observed GFP expression in rod photoreceptors 11 days after removal of Dox from a 48 h Dox treatment, suggesting that the interaction of rtTA with Dox at the TRE
site was very tight and not quickly reversible (data not shown).
We created both Tg(Xla.rho:rtTA, TRE:GFP)
lines using the pTol transgenesis method and also cloned the response constructs (TRE:nlsCherry
and biTRE: EGFP, nlsCherry
) in pTol vectors 
. When these pTol response vectors are injected along with transposase
mRNA into transgenic lines that were created by pTol transgenesis, there is a possibility that the transposase could remobilize the Xla.rho:rtTA, TRE:GFP
transgenes in individual cells, thus effecting the Dox-inducible expression of the response construct. This possibility seems small, as we have seen many rods in our mosaics that retained the Xla.rho:rtTA, TRE:GFP
transgenes and expressed from the injected response construct ( and ). Certainly, enough rods expressed the transgene from the injected response construct to determine whether or not to proceed towards making stable germline transgenics with the response construct. However, if remobilization of the driver construct (Xla.rho:rtTA, TRE:GFP
) by injection of Tol2 transposase
mRNA was problematic, the response plasmids could be re-engineered to use either the non-cross-reacting Tol1
or the sleeping beauty
transposable element 
. Gateway vectors with the sleeping beauty
transposable element will be available soon, allowing easy construction (Kristen Kwan, University of Utah, personal communication).
All of the experiments we presented show Dox-induced gene expression leading to the overexpression of target proteins. We are currently examining whether the Tet-On systems described here can be used to express shRNAs to temporally disrupt target gene function in rods. If successful, this component of the system will add a very powerful tool for studying the genes involved in rod photoreceptor physiology and maintenance. While the particular Tet-On transgenics described in this study are useful for studying rod photoreceptors and vision, we created several Gateway compatible vectors that should also be useful to the broader community of researchers for creating additional Tet-On transgenic driver lines using other cell/tissue-specific promoters. Components of the response constructs are also Gateway compatible for the rapid cloning of transgenes. The vectors will be made available to the community as a set - the Tet-On Toolkit (Fig. S2
). Given that Gateway vectors customized for zebrafish research are already generally available from the Chien and Lawson labs, a wide variety of Tet-On response conformations can now easily be constructed to make inducible transgenics.