Transgenic retinoic acid sensor lines in zebrafish indicate regions of available embryonic retinoic acid


  • Amrita Mandal,

    1. The Heart Institute, Molecular Cardiovascular Biology Division and Development Biology Division, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
    2. Molecular Developmental Biology Graduate Program, University of Cincinnati and Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
    Search for more papers by this author
  • Ariel Rydeen,

    1. The Heart Institute, Molecular Cardiovascular Biology Division and Development Biology Division, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
    2. Molecular Developmental Biology Graduate Program, University of Cincinnati and Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
    Search for more papers by this author
  • Jane Anderson,

    1. The Heart Institute, Molecular Cardiovascular Biology Division and Development Biology Division, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
    Search for more papers by this author
  • Mollie R.J. Sorrell,

    1. The Heart Institute, Molecular Cardiovascular Biology Division and Development Biology Division, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
    Search for more papers by this author
  • Tomas Zygmunt,

    1. Department of Cell Biology, Helen L. and Martin S. Kimmel Center for Biology and Medicine at the Skirball Institute, New York University School of Medicine, New York
    Search for more papers by this author
  • Jesús Torres-Vázquez,

    1. Department of Cell Biology, Helen L. and Martin S. Kimmel Center for Biology and Medicine at the Skirball Institute, New York University School of Medicine, New York
    Search for more papers by this author
  • Joshua S. Waxman

    Corresponding author
    1. The Heart Institute, Molecular Cardiovascular Biology Division and Development Biology Division, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
    • Correspondence to: Joshua Waxman, The Heart Institute, Molecular Cardiovascular Biology Division and Development Biology Division, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, 45229. E-mail:

    Search for more papers by this author


Background: Retinoic acid (RA) signaling plays a critical role in vertebrate development. Transcriptional reporters of RA signaling in zebrafish, thus far, have not reflected the broader availability of embryonic RA, necessitating additional tools to enhance our understanding of the spatial and temporal activity of RA signaling in vivo. Results: We have generated novel transgenic RA sensors in which a RA receptor (RAR) ligand-binding domain (RLBD) is fused to the Gal4 DNA-binding domain (GDBD) or a VP16-GDBD (VPBD) construct. Stable transgenic lines expressing these proteins when crossed with UAS reporter lines are responsive to RA. Interestingly, the VPBD RA sensor is significantly more sensitive than the GDBD sensor and demonstrates there may be almost ubiquitous availability of RA within the early embryo. Using confocal microscopy to compare the expression of the GDBD RA sensor to our previously established RA signaling transcriptional reporter line, Tg(12XRARE:EGFP), illustrates these reporters have significant overlap, but that expression from the RA sensor is much broader. We also identify previously unreported domains of expression for the Tg(12XRARE:EGFP) line. Conclusions: Our novel RA sensor lines will be useful and complementary tools for studying RA signaling during development and anatomical structures independent of RA signaling. Developmental Dynamics 242:989–1000, 2013. © 2013 Wiley Periodicals, Inc.


The requirement of RA signaling was first studied in the 1940s and 1950s using Vitamin A–deficient rat embryos (Warkany and Schraffenberger, 1946; Wilson and Warkany, 1949; Wilson et al., 1953). Since these initial pioneering studies, inappropriate RA signaling has been found to cause a similar spectrum of developmental defects in all vertebrates, with increases and decreases in RA signaling affecting many organs and tissues, including the hindbrain, forelimbs, and heart (Heine et al., 1985; Lammer et al., 1985; Rizzo et al., 1991; Li et al., 1993; Lohnes et al., 1993, 1994; Mendelsohn et al., 1994; Niederreither et al., 1999, 2001; Abu-Abed et al., 2001; Sakai et al., 2001; Grandel et al., 2002; Emoto et al., 2005; Sandell et al., 2007). RA signaling's earliest known function is in posteriorization of the embryonic axis (Duester, 2008; Niederreither and Dolle, 2008; Mark et al., 2009), which is an evolutionarily conserved function in chordates (Campo-Paysaa et al., 2008). In vertebrates, later developmental roles of RA signaling include promoting differentiation of the epicardium (Braitsch et al., 2012), proliferation of the myocardium (Chen et al., 2002), outgrowth of the forelimbs (Cooper et al., 2011), and patterning of the cardiac outflow tract (Li et al., 2010).

RA is the most active metabolic derivative of Vitamin A (retinol) and is produced through a series of oxidative reactions by alcohol and aldehyde dehydrogenases within specific locations of the embryo (Duester, 2008). During development, RA acts as a morphogen and is a ligand for members of the nuclear hormone family of receptors, the RA receptors (RARs) (Aranda and Pascual, 2001; Bastien and Rochette-Egly, 2004; White et al., 2007; Duester, 2008). RARs function at the transcriptional level as heterodimers with the retinoid X receptors (RXRs), which together bind RA response elements (RAREs). Most often, RAR/RXRs heterodimers bind what are referred to as direct repeat 5 (DR5) sites, defined by a 5-nucleotide spacer between the RAR and RXR binding elements. However, they do have some flexibility and can also bind DR1 and DR2 sites (Bastien and Rochette-Egly, 2004).

The knowledge of general RAR-binding rules has led to the generation of a number of synthetic reporters that contain concatenated DR5 RARE sites, which have been used both in vitro and in vivo in transgenic mice as transcriptional reporters of RA signaling (Rossant et al., 1991; Balkan et al., 1992; Nagpal et al., 1992). More recently, similar strategies were used to create zebrafish with transgenic RA signaling reporter lines (Perz-Edwards et al., 2001). However, a number of the earlier zebrafish lines, which used the same synthetic promoters as utilized in mice, exhibited significant positional effects due to their insertions, even though they were still responsive to RA. Moreover, these RA signaling reporter lines were not expressed in some of the tissues known to require RA signaling (Perz-Edwards et al., 2001). Despite the transgenic RARE reporter mice having broader expression compared to the transgenic zebrafish (Rossant et al., 1991; Balkan et al., 1992), it is still not clear as to whether or not they too reflect all RA signaling.

In an attempt to make a more sensitive RA signaling reporter, we recently constructed a new synthetic RA signaling reporter composed of 12 concatenated DR5 RARE elements (Fig. 1C; Waxman and Yelon, 2011), instead of the 3 used previously in mice and zebrafish (Rossant et al., 1991; Balkan et al., 1992; Perz-Edwards et al., 2001). Like its predecessors, we also used this construct in vitro in cell culture and in vivo to make stable transgenic lines in zebrafish. While the 12XRARE transgenic reporter lines in zebrafish were less sensitive to positional effects and were highly responsive to RA signaling, it is clear that these reporters still do not reflect expression in all tissues with RA signaling (Waxman and Yelon, 2011). Therefore, we have sought to generate additional in vivo tools that will complement our current RA signaling reporter transgenic lines and aid in our understanding of the locations and mechanisms of RA signaling within the embryo.

Figure 1.

Constructs used and schematics of RA reporters. A: Conserved domains found in RARs (Bastien and Rochette-Egly, 2004). The A domain has some limited transcriptional activation ability. The B domain has an unknown function. The C domain is the DNA-binding domain. The D domain is a hinge domain. The E domain is the ligand-binding domain and contains the major transcriptional activation domain. The F domain has an unknown function. The Gal4-DNA-binding domain (GDBD; yellow) was fused to the D–F domains, which contain the RAR ligand-binding domain, to make the GDBD-RLBD fusion protein. The VP16 activation domain (blue) was fused to the N-terminus of the GDBD-RLBD to make the VPBD-RLBD fusion protein. B: Schematic of how the RLBD fusion proteins work. The GDBD-RLBD fusion protein should bind the UAS elements and activate transcription in the presence of RA (red dot). C: Schematic of the previously reported 12XRARE transgene for comparison to the ligand trap method. RA (red dot) interacts with RARs, which drive transcriptional activation through RAR/RXR heterodimers acting on DR5 RAREs.

Previous in vitro studies have used Gal4 DNA-binding domain (GDBD)/RAR domain fusion proteins to elucidate the functional domains of RARs (Nagpal et al., 1992; Allenby et al., 1993). More recently, groups have adapted this method, in what is referred to as a ligand trap strategy, for use with many other nuclear hormone receptors in stable transgenic zebrafish to identify the ligands for orphan nuclear hormone receptors (Tiefenbach et al., 2010). An advantage of the GDBD-RAR fusion is that it should function primarily as a ligand sensor, rather than as a ligand-dependent transcriptional reporter of RXR/RAR interactions. We have used this strategy to make GDBD and VP16-GDBD (VPBD) fusion proteins with the ligand-binding domain of the zebrafish RARab (RLBD) (Fig. 1A, B; Hale et al., 2006; Waxman and Yelon, 2007). We have characterized these transgenic lines and find that both fusion proteins are responsive to RA, although the VPBD-RLBD fusion is significantly more sensitive and more accurately reflects the locale of RA based on the expression domains of the RA-producing and -degrading enzymes. We also compare the expression of a GDBD-RLBD;UAS:mCherry transgenic reporter line to one of our previously reported Tg(12XRARE-ef1a:EGFP) lines using confocal microscopy. We conclude that these novel RA sensor lines will complement existing tools and aid in the in vivo analysis of RA and RA signaling during development and other contexts.


GDBD-RLBD;UAS Transgenes Produce an RA Sensor

In order for the GDBD-RLBD fusion protein lines to report RA, they need to be crossed with UAS reporter lines (Fig. 1B). To examine the efficacy of the Tg(β-actin:GDBD-RLBD) trangene, we crossed stable transgenic carriers to two distinct UAS reporter lines: a Tg(5XUAS:EGFP) (Asakawa et al., 2008), which we refer to as UAS:EGFP, and a Tg(14XUAS-E1b:NfsB-mCherry) (Davison et al., 2007), which we refer to as UAS:mCherry. Reporter expression in either the Tg(β-actin:GDBD-RLBD);(UAS:EGFP) or Tg(β-actin:GDBD-RLBD);(UAS:mCherry) embryos was detectable using in situ hybridization (ISH) by the tailbud stage (TB) stage (Fig. 2A, B, K, L). UAS reporter expression initially appeared to be predominantly expressed in the spinal cord (Fig. 2A–D, K–N). However, by the 16s stage, expression was observed in a bit broader domain, discernibly expanding to the somites and notochord, and later to the pronephros (Fig. 2E–J, O–T). The region of expression within the embryos was largely similar between the two lines, except that in the Tg(β-actin:GDBD-RLBD);(UAS:EGFP) embryos there was stronger and persistent expression in the anterior somites and expression in the ventral eye (Fig. 2I, J, S, T). Importantly, reporter expression from both lines was restricted to a mid-region of the embryos where we and others have previously established there is high RA signaling during these stages of development (White et al., 2007; Waxman and Yelon, 2009, 2011; Feng et al., 2010).

Figure 2.

RA sensor reporters are expressed in established domains of RA responsive gene expression. A, B, K, L: Reporter expression in both the Tg(β-actin:GDBD-RLBD);(UAS:EGFP) and Tg(β-actin:GDBD-RLBD);(UAS:mCherry) lines initiates in the spinal cord (arrows in A and K) by the tailbud (TB) stage. C, D, M, N: RA sensor reporter expression is strengthened in the anterior spinal chord by the 8-somite (s) stage (arrows in C and M). E, F, O, P: RA sensor reporter expression is maintained in the anterior spinal cord (sc) and noticeable expanded to the adjacent tissues, including the notochord (nc) and ectoderm (ec), by the 16s stage. G, H, Q, R: RA sensor reporter expression was maintained in the tissues of the mid-region (anterior trunk and spinal cord), but was now more easily observed in the somites (sm) and pronephros (pn). I, J, S, T: Both the Tg(β-actin:GDBD-RLBD);(UAS:EGFP) and Tg(β-actin:GDBD-RLBD);(UAS:mCherry) lines initiate some expression in the brain by 36 hpf. However, we only found eye expression (arrow in I) in the Tg(β-actin:GDBD-RLBD);(UAS:EGFP) line. Views in A, C, E, G, I, K, M, O, Q, S are lateral. Views in B, D, F, H, J, L, N, P, R, T are dorsal. In all images, anterior is up.

The fluorescent expression from both transgenic lines was delayed relative to the ISH expression. In the Tg(β-actin:GDBD-RLBD);(UAS:EGFP) transgenic embryos, expression was visible by the 16s stage (Fig. 3A), while expression from the Tg(β-actin:GDBD-RLBD);(UAS:mCherry) transgenic embryos was not typically visible until shortly after the 16s stage (Fig. 3E). Bright fluorescence was observed in a central portion of the embryo by 24 hr post-fertilization (hpf) with both UAS lines, which was maintained past 48 hpf (Fig. 3B–D, F–H). Although the domains of expression from these reporters was largely similar, the Tg(β-actin:GDBD-RLBD);(UAS:mCherry) line was significantly more variegated, which is likely due to silencing caused by the additional repetitive elements used with that UAS construct (Davison et al., 2007; Goll et al., 2009). The other major difference we observed in fluorescent reporter expression, other than the variegation due to silencing, was that the Tg(β-actin:GDBD-RLBD);(UAS:EGFP) line had expression in the eyes and stronger expression in the somites (Fig. 3D), as mentioned with the ISH. We think that the expression in the eye is reflecting availability of RA because our previously created transcriptional RA reporter also has expression in the eye after 24 hpf (Waxman and Yelon, 2011). We postulate that the difference we observe between the UAS reporters could either be due to positional effects on the transgene or the shorter UAS repeats (Goll et al., 2009).

Figure 3.

Fluorescence in the Tg(β-actin:GDBD-RLBD);(UAS:EGFP) and Tg(β-actin:GDBD-RLBD);(UAS:mCherry) lines. A, E: Fluorescence in the Tg(β-actin:GDBD-RLBD);(UAS:EGFP) line is found in the spinal cord (arrow) by the 16s stage, although expression in the Tg(β-actin:GDBD-RLBD);(UAS:mCherry) is typically not visible by this stage. B, F: Expression at 24 hpf is visible in the spinal cord, notochord, and somites. Expression in the adjacent non-spinal cord tissues is more easily visible in the Tg(β-actin:GDBD-RLBD);(UAS:EGFP) embryos. C, D, G, H: Expression in maintained in the spinal cord, notochord, and somites at 30 and 48 hpf. Expression is seen in the ventral eye by 48 hpf (arrow in D). Arrowhead in D indicates the somites. Expression in the eye of the Tg(β-actin:GDBD-RLBD);(UAS:mCherry) embryo is from the α-cry:DsRed used in the transgenic construct. Arrow in H indicates autofluorescent pigment cells. All views are lateral with anterior up.

The expression of the β-actin:GDBD-RLBD;UAS lines in the mid-region of the embryo was strongly suggestive that reporters are RA responsive. To confirm that the reporters are RA responsive, we modulated RA signaling using RAR agonists, RAR antagonists, and an aldehyde dehydrogenase inhibitor. Consistent with the GDBD-RLBD being responsive to RA, treatment with RA or the RARα agonist AM580 both produced essentially ubiquitous expression in the Tg(β-actin:GDBD-RLBD);(UAS:EGFP) and Tg(β-actin:GDBD-RLBD);(UAS:mCherry) embryos (Fig. 4A–C, A′–C′, H–J, H′–J′). Similarly, the pan-Cyp inhibitor Ketoconazole (Keto), which inhibits the function of the RA degrading Cyp26 (P450) enzymes, or the use of morpholinos (MOs) for cyp26a1 and cyp26c1, produced an expansion of the domain of reporter expression (Figs. 4D′, K, K′ and 5A–D). Conversely, the retinaldehyde dehydrogenase inhibitor diethyl-aminobenzoic acid (DEAB), which inhibits RA production (Russo et al., 1988), and the competitive RARα antagonist RO41-5253 (referred to as RO41) (Zhang et al., 1995), which should interact with the RLBD, both inhibited expression of the reporter. Altogether, our analysis suggests that our new transgenic lines are functioning as sensors of embryonic RA.

Figure 4.

RA sensor lines are responsive to RAR agonists and antagonists. A, A′, H, H′: Control untreated embryos. B, B′, C, C′, I, I′, J, J′: RA- or AM580-treated embryos show almost ubiquitous expression of the reporter. Embryos treated with RA or AM580 were posteriorized as a consequence of treatment with these agents. D, D′, K, K′: Keto caused an expansion of reporter expression. E, E′, F, F′, L, L′, M, M′: DEAB or RO41 treatment inhibited expression. None of the DEAB-treated (0 of 17) or R041-treated (0 of 17) Tg(β-actin:GDBD-RLBD);(UAS:EGFP) embryos had expression compared to 28% (17 of 61) sibling embryos, which is close to the 25% of expected fluorescent embryos used from an intercross of Tg(β-actin:GDBD-RLBD) and Tg(UAS:EGFP) hemizygous individuals. None of DEAB-treated (0 of 15) or R041-treated (0 of 14) Tg(β-actin:GDBD-RLBD);(UAS:mCherry) embryos had expression compared to 34% (18 of 53) sibling embryos, which is close to the 37.5% of expected fluorescent embryos used from an intercross of Tg(β-actin:GDBD-RLBD);(UAS:mCherry) hemizygous and (UAS:mCherry) hemizygous individuals. A–F and H–M are fluorescent images of live embryos. A′–F′ and H′–M′ are ISH. Arrows in A,A′,D,D′,H,H′,K,K′ indicate the length of the domain of reporter expression. All views are lateral with anterior up.

Figure 5.

RA sensor expression is expanded in Cyp26-deficient embryos. A, C: Control uninjected sibling embryos. B, D: The domain of reporter expression is expanded in embryos injected with MOs targeting cyp26a1 and cyp26c1. Arrows indicate the length of expression. Views are lateral with anterior up.

A Hypersensitive Transgenic RA Sensor

Although the GDBD-RLBD:UAS reporters are sensitive to RA and expressed in regions of the embryo where there is RA signaling, we found it interesting that the reporter expression did not initiate until the TB, likely after RA signaling initiates based on aldh1a2 expression (Grandel et al., 2002), and were expressed in a relatively restricted domain within the embryos, similar to RA responsive genes and the Tg(12XRARE-ef1a:EGFP) lines, here referred to as Tg(12XRARE:EGFP) (Waxman et al., 2008; Feng et al., 2010; Waxman and Yelon, 2011). Therefore, we wanted to explore ways of increasing the sensitivity of the RA sensor. Our previous study found that fusion of the VP16 transcriptional activation domain to either the N- or C-termini of RARs creates a hyperactive RAR, not a constitutively active RAR (Waxman and Yelon, 2011). Therefore, we wanted to determine if a similar fusion would make a hypersensitive RA sensor. We fused the VP16 domain to the N-terminus of the GDBD-RLBD protein (VPBD-RLBD), made stable transgenic lines, and crossed them to the UAS reporter lines.

We found that in Tg(β-actin:VPBD-RLBD);(UAS:EGFP) embryos expression was pretty broad by the TB stage (Fig. 6A, B) and maintained through the 8s stage when fluorescence was easily oberserved (Fig. 6C–E). In contrast to the more limited expression of the β-actin:GDBD-RLBD;UAS transgenic lines, the broad expression of the VP16-RA sensor was reminiscent of the broad, posterior expression of aldh1a2 at the same stages within the early embryo (Grandel et al., 2002; Maves and Kimmel, 2005). Furthermore, given the role of RA signaling in patterning the hindbrain, the sharp anterior border of reporter expression at the TB stage was reminiscent of the expression patterns of RA responsive genes in the hindbrain, including hoxb1b, vhnf1, and hoxb1a (Hernandez et al., 2004; Maves and Kimmel, 2005). Interestingly, we also found expression in the very anterior of the embryo and the polster (Fig. 6A, C, E), suggesting there may be available RA in the anterior of the embryo in addition to more posteriorly. A candidate source of this expression is Aldh1a3, which is expressed in the anterior of the embryo (Liang et al., 2008; Pittlik et al., 2008). However, its stronger anterior expression does not initiate until mid-somitogenesis (Liang et al., 2008; Pittlik et al., 2008), suggesting that Aldh1a2 cannot also be ruled out as a potential source. Importantly, at these stages, Tg(β-actin:VPBD-RLBD);(UAS:EGFP) reporter expression was largely lacking in a large anterior region of the embryo (Fig. 6A–E), which correlates with the early anterior expression domain of cyp26a1 (Emoto et al., 2005; White et al., 2007). By the 16s stage, Tg(β-actin:VPBD-RLBD);(UAS:EGFP) expression was still absent in the majority of the anterior brain and anterior hindbrain (Fig. 6F–H). However, expression expanded to the eyes, where aldh1a3 is expressed (Liang et al., 2008; Pittlik et al., 2008), and was in the hatching gland, the derivative of the polster. At the 8s and 16s stages, lower levels of expression were found in the posterior hindbrain and extended anteriorly up to the level of the otic vesicle, which forms at rhombomeres 5 and 6 (Fig. 6C, E, F, H). By the 16s stage, strong expression was in the posterior spinal cord and extended to the most posterior somites, but was absent in the tip of the tail (Fig. 6F, I, L). The presence and absence of reporter expression again parallels the opposing domains of aldh1a2 expression and cyp26a1 in the trunk and tail (Grandel et al., 2002; Emoto et al., 2005). Expression was maintained in the anterior brain and eyes, and the somites and spinal cord of the embryos past 48 hpf (Fig. 6I–N). Similar domains of expression were also observed in the Tg(β-actin:VPBD-RLBD);(UAS:mCherry) embryos. However, expression was significantly variegated in these lines, so it is not presented here.

Figure 6.

Expression in Tg(β-actin:VPBD-RLBD);(UAS:EGFP) embryos is consistent with the domains of RA-producing and -degrading enzymes. A, B: VP16-RA sensor expression at the TB stage. Arrowhead indicates strong expression in the polster (po). Arrows in A and B indicate the border between the posterior reporter expression and the anterior of the embryo, which lacks reporter expression. C–E: VP16-RA sensor expression via ISH and fluorescence at the 8s stage. Expression is maintained in the polster (po) and is largely absent in the anterior brain (arrows in C and E). There is strong expression throughout the posterior. Lower levels of expression extend into the posterior hindbrain (phb). F–H: VP16-RA sensor expression via ISH and fluorescence at the 16s stage. There is expression in the eyes (ey) and anterior brain, hatching gland (hg), and posterior hindbrain. Expression extends into the posterior hindbrain up to the otic vesicle (dashed outlines in F and H), which forms at the rhombomeres 5 and 6. Expression is absent in the brain and tip of the tail (arrows in F and H). I–K: RA reporter expression via ISH and fluorescence at 30 hpf. L, M: VP16-RA sensor expression via ISH at 36 hpf. N: VP16-RA sensor expression via fluorescence at 48 hpf. A, D, F, G, I, J, L, M, O are lateral views. B, E, H, K, N are dorsal views. In all images anterior is up.

To determine if the Tg(β-actin:VPBD-RLBD);(UAS:EGFP) embryos were sensitive to modulation of RA, we used the same pharmacological agents that were used to analyze the β-actin:GDBD-RLBD;UAS transgenic lines. Treatment with RA or AM580 produces relatively ubiquitous expression (Fig. 7A–C, G–I), while Keto expanded and enhanced the domains of expression (Fig. 7D, J). Interestingly, treatment with DEAB or RO41 was both able to reduce expression, but never able to completely eliminate expression (Fig. 7K, L). Of these two inhibitors, it is interesting that RO41, the competitive RARα antagonist, was typically much more effective at reducing expression (Fig. 7F, L). DEAB treatment could reduce expression as indicated by ISH (Fig. 7K). However, we often could not detect significantly lower expression via fluorescence in live embryos (Fig. 7L). Therefore, our characterization of the Tg(β-actin:VPBD-RLBD);(UAS:EGFP) transgenic line suggests that the VPBD-RLBD fusion protein creates a more sensitive RA sensor than the GDBD-RLBD and that the reporter's expression and absence of expression correlate better with the expression of RA-producing and degrading enzymes, respectively, than the more restricted β-actin:VPBD-RLBD;UAS or the 12XRARE reporters (Waxman and Yelon, 2011).

Figure 7.

Tg(β-actin:VPBD-RLBD);(UAS:EGFP) embryos are sensitive to RAR agonists and antagonists. A, G: Control sibling embryos at 30 hpf. B, C, H, I: RA- and AM580-treated VP16-RA sensor embryos have almost ubiquitous expression. Embryos are posteriorized due to RAR agonist treatment. D, J: VP16-RA sensor embryos treated with Keto have expanded and enhanced expression, particularly in the anterior brain and eye (arrow in D). E, F, K, L: VP16-RA sensor embryos treated with DEAB or RO41. By ISH, DEAB and RO41 reduced expression (arrows in K and L). However, RO41 was more effective. By fluorescence, it was often hard to distinguish a loss in reporter expression in DEAB-treated embryos. While we observed reductions in reporter expression, particularly with RO41 treatment, we never completely lost expression in embryos treated with RAR or Aldh1a antagonists. All views are lateral with anterior up.

Comparison of the Transgenic GDBD-RLBD RA Sensor and RARE Reporter Lines

The more restricted expression of the β-actin:GDBD-RLBD;UAS RA sensor reporters was conspicuously similar to the 12XRARE:EGFP transgenic reporter lines (Waxman and Yelon, 2011). Both are expressed predominantly in the mid-region of the embryo where we postulate that there is higher sustained RA signaling (Figs. 2 and 3) (Waxman et al., 2008; Waxman and Yelon, 2011). Therefore, we wanted to directly compare the temporal and spatial expression from the different reporters. Hemizygous Tg(β-actin:GDBD-RLBD);(UAS:mCherry) adults were crossed with homozygous Tg(12XRARE:EGFP) adults and the triple transgenic embryos were imaged using confocal microscopy at stages from 20s through 78 hpf. As suggested from our ISH and live fluorescent analysis of the current and previously published reporters (Waxman and Yelon, 2011), the fluorescent expression from the β-actin:GDBD-RLBD;UAS:mCherry transgenes was initiated earlier than the 12XRARE:EGFP transgene (Fig. 8A–C). By the 20s stage, β-actin:GDBD-RLBD;UAS:mCherry reporter expression was easily visible in skin epithelial cells, the spinal cord, a few pronephros cells and notochord (Fig. 8A, C). By comparison, EGFP was more weakly expressed and just initiating in a couple pronephros cells and the spinal cord of the Tg(12XRARE:EGFP) embryos (Fig. 8B). By 24hpf, the β-actin:GDBD-RLBD;UAS:mCherry reporter expression had expanded in the skin epithelial cells, the spinal cord, the pronephros and notochord (Fig. 8D, F). By comparison, the 12XRARE:EGFP reporter was still more weakly expressed, but more visible in the pronephros of embryos (Fig. 8E). By 30 hpf, the expression of both reporters was stronger and significantly expanded (Fig. 8G–I). Expression of the β-actin:GDBD-RLBD;UAS:mCherry reporter was still expressed in the skin epithelial cells, the spinal cord, pronephros and notochord (Fig. 8G, I), but now axons projecting anteriorly from the spinal cord were also visible (Fig. 8G, I, insets). At 30 hpf, in the Tg(12XRARE:EGFP) embryos the spinal cord expression was significantly stronger and expanded in the A–P axis in the spinal cord and pronephros (Fig. 8G, I). Additionally, there were also extremely low levels of expression in the somites and notochord, which we have not reported previously (Fig. 8H). Despite the overlap in most tissues, we never observed expression in the skin epithelial cells in the Tg(12XRARE:EGFP) embryos. By 30 hpf, we also observed expression in the dorsal eye (Fig. 9A), which we had not reported previously and correlates with the localization aldh1a2 expression (Liang et al., 2008; Pittlik et al., 2008; French et al., 2009). Expression in the ventral domain of the eye was reported previously and correlates with aldh1a3 expression (Fig. 9A; Liang et al., 2008; Pittlik et al., 2008; French et al., 2009). As mentioned above, we did not find eye expression using the UAS:mCherry line, but did observe it in the UAS:EGFP embryos (Figs. 2I and 3D).

Figure 8.

Direct comparison of the 12XRARE:EGFP and β-actin:GDBD-RLBD;UAS:mCherry transgenes. A–F: At the 20s stage and 24 hpf, the RA sensor is expressed in the skin epithelial cells (ec), the spinal cord (sc), the pronephros (pn) and the notochord (nc), while the 12XRARE:EGFP transgene is less strongly visible in the spinal cord and pronephros. G–I: By 30 hpf, mCherry expression from the β-actin:GDBD-RLBD;UAS:mCherry reporter has expanded in the anterior-posterior axis, but the tissue types expressing the reporter have not changed. MCherry can be seen in axons projecting anteriorly from the spinal cord (G). Insets in G–I are parallel confocal slices to the images in G–I, but focus on the anterior border of expression from the reporters. 12XRARE:EGFP reporter expression has also expanded in the A–P axis and the intensity of the reporter has increased. Low levels of expression are also in the somites (sm), though this is more obvious from mCherry. J–L: By 56 hpf, expression of both reporters has expanded more posteriorly, while the anterior border of expression at the hindbrain–spinal cord boundary has become more distinct. Expression is also seen in the axonal projects from the ventral motor neurons (vmn). M–O: Expression in these same tissues is maintained past 72 hpf. The ventral motor neurons can be seen elaborating further. All images are single confocal slices. Embryos are lateral views with anterior up.

Figure 9.

The 12XRARE:EGFP transgene is expressed in the eye, pectoral fin motor neurons, and brain. A: The 12XRARE:EGFP reporter is expressed in the dorsal retina (dr), by 30 hpf. The dorsal retina expression was visible due to PTU treatment. Expression in the ventral retina (vr) was reported previously (Waxman and Yelon, 2011). B: Expression of the 12XRARE:EGFP reporter is maintained and expanded in the dorsal retina and ventral retina by 56 hpf. C, D: At 78 hr, EGFP expression from the 12XRARE:EGFP reporter is visible in the pectoral motor neuron axons and their termini at the base of the pectoral fin (arrowheads in C and D). Both transgenes are expressed in neurons in the brain (arrow in D). The red lenses in B-D are due to α-cry:DsRed expression, which was used in the Tol2 construct. The β-actin:GDBD-RLBD;UAS:mCherry reporter was not expressed in the eye. Red at the periphery of the eyes in B–D is due to auto-fluorescence from residual pigment cells. Auto-fluorescent cells are also found in the trunk at later stages, despite PTU treatment (arrow in C). Views in C and D are dorso-lateral. Magnification of images in A, B, and D is 200×. Magnification of images in C is 100×.

Through 72 hpf, the expression of both reporters became sharply defined at the hindbrain–spinal cord boundary (Fig. 8J–O), compared to earlier when the boundary between these neural regions was not quite as distinct (Fig. 8G–I). Expression was maintained at these later stages in all the same tissues as at 30 hpf for both reporters (Figs. 8–O, 9B). Interestingly, both these reporters labeled the ventral motor neurons (Fig. 8J–O), which we did not report previously for the Tg(12XRARE:EGFP) lines using standard fluorescent microscopy (Waxman and Yelon, 2011). The extension and expansion of the ventral motor neurons projecting from the spinal cord was easily observable through these stages and individual axons could be followed due to the variegated nature of expression from the UAS:mCherry transgene. Furthermore, by 78 hpf, the pectoral fin motor neuron projections could be seen terminating at the base of the fins (Fig. 9C,D; Ma et al., 2010), while there was additional expression of individual neurons seen medial to the retina in the brain (Fig. 9D).


Our characterization of the β-actin:GDBD-RLBD;UAS transgenic reporters suggests that they are responsive to and report the availability of embryonic RA. Interestingly, the regionalized expression from the GDBD-RLBD;UAS reporter lines in the mid-region of the embryo posterior to the hindbrain overlaps considerably with our previously reported 12XRARE:EGFP transgenic line (Waxman and Yelon, 2011). The expression of the RA sensor reporter initiates earlier and is expressed more broadly than the 12XRARE:EGFP reporter line, as one might predict because it should report the availability of RA and precede target gene transcription, while the RARE reporter is dependent upon RA plus RAR/RXR interactions. The broader expression of the RA sensor reporter also reinforces the notion that there are tissues that are likely responsive to RA, but are not strongly represented by expression in the 12XRARE:EGFP transgenic line (Waxman and Yelon, 2011). Since it is likely that the RLBD interacts with transcriptional activators and repressors (Aranda and Pascual, 2001; Bastien and Rochette-Egly, 2004), even though theoretically it should not be dependent on RXR (Tiefenbach et al., 2010), we cannot rule out that the transcriptional regulatory proteins that interact with the GDBD-RLBD fusion protein through the RLBD influence the temporal and spatial expression of the reporter (Bastien and Rochette-Egly, 2004). Importantly, the regionalized expression of the RA sensor and RARE transgenic lines does correlate with the expression of RA target genes and sensitive tissues during early and mid-somitogenesis stages (Grandel et al., 2002; Hernandez et al., 2004, 2007; Maves and Kimmel, 2005; Waxman et al., 2008; Feng et al., 2010). However, when compared to the β-actin:VPBD-RLBD;UAS reporter, our characterization also suggests that they both may be reporting higher levels of sustained RA.

When comparing the expression of the β-actin:GDBD-RLBD;UAS and 12XRARE:EGFP reporters to the β-actin:VPBD-RLBD;UAS reporter, it is clear that the β-actin:VPBD-RLBD;UAS reporter is expressed much more broadly throughout the transgenic embryos and strongly correlates with aldh1a2, aldh1a3, and cyp26a1 expression (Grandel et al., 2002; Emoto et al., 2005; Maves and Kimmel, 2005; Liang et al., 2008; French et al., 2009). Because we cannot eliminate reporter expression in β-actin:VPBD-RLBD;UAS transgenic embryos, this suggests a couple possibilities, which are not mutually exclusive: there is some transcriptional activation of the UAS reporter from the VPBD domain that is independent of RA or there are low levels of RA that can still be detected by the VPDB-RLBD fusion protein that we cannot eliminate. Although we have not yet distinguished between these possibilities, an argument in favor of the latter possibility is that if there were RA-independent reporter expression, we would not expect to the have localized reporter expression that strongly correlates with RA producing and degrading enzymes. Moreover, previous studies have also demonstrated that there is still RA-responsive gene expression in zebrafish embryos even with high concentrations (5 μM) of DEAB treatment (Maves and Kimmel, 2005), which confirms that DEAB treatments likely do not completely eliminate embryonic RA or RA signaling. Therefore, assuming that the β-actin:VPBD-RLBD;UAS transgenic reporter is reporting available embryonic RA, our analysis suggests the thought-provoking hypothesis that RA may be available almost entirely throughout the embryo, including the most anterior tissues of the embryos, which has not been previously suggested to receive embryonic RA at earlier stages.

While we propose that VPBD-RLBD; UAS transgenic lines will also be useful, informative, and complementary tools, the enhanced sensitivity to RA at this point may actually limit its general utility because we have found that it does not demonstrate significant fluctuations in reporter expression when RA signaling is modulated. Nevertheless, this analysis promotes the idea that in the future similar strategies using alternative transcriptional activation domains combined with the GDBD-RLBD fusion protein may be able to create highly sensitive reporters that accurately report the availability of low concentrations of RA while eliminating the possibility of ligand-independent transcriptional activation.

In addition to being indicators of RA sensitivity and availability, our confocal analysis of the RA sensor and 12XRARE:EGFP transgenic lines suggests that these transgenes will likely be useful tools for studying the development of RA-independent processes. In particular, we envision that these lines will be useful for studying the development of specific populations of neurons in the brain and spinal cord, due to the expression of the fluorescent proteins in the axons, as well as the nephrons, where expression persists as long as we have tracked expression in the embryos. Together, our current and previous studies using transgenic zebrafish with the Gal4 fusion method set the precedence for the creation of similar lines for studying RA signaling and the function of other nuclear hormone receptors in other model organisms. Overall, the novel RA sensors complement the utility of the existing 12XRARE:EGFP transgenic lines and enhance the available toolset to study the mechanisms of RA signaling dependent developmental processes in vivo.


Zebrafish Maintenance and Lines Used

Adult zebrafish were maintained and embryos were raised using standard zebrafish aquaculture conditions. UAS transgenic lines used in crosses were Tg(5XUAS:EGFP) (Asakawa et al., 2008) and Tg(14XUAS:NfsB-mCherry) (Davison et al., 2007). For the stable β-actin:GDBD-RLBD and β-actin:VPBD-RLBD transgenic lines, multiple transgenic lines were recovered from individual founders designated Tg(β-actin:GDBD-RLBD)cch1, Tg(β-actin:GDBD-RLBD)cch4, Tg(β-actin:VPBD-RLBD)cch2, Tg(β-actin:VPBD-RLBD)cch5, and Tg(β-actin:VPBD-RLBD)cch6. Each transgene in the stable lines segregated according to a standard Mendalian ratio, suggesting the integrations were at single loci. Expression from each of the Tg(β-actin:GDBD-RLBD) and Tg(β-actin:VPBD-RLBD), respectively, was overtly found to be indistinguishable. Therefore, the lines Tg(β-actin:GDBD-RLBD)cch1 and Tg(β-actin:VPBD-RLBD)cch2 are presented in the current study, but all lines are available. For experiments, hemizygous Tg(β-actin:GDBD-RLBD) carriers were crossed with hemizygous Tg(UAS:EGFP) carriers and Tg(β-actin:GDBD-RLBD);(UAS:mCherry), which were hemizygous for each transgene, were crossed to hemizygous Tg(UAS:mCherry) carriers. The Tg(12XRARE-ef1a:EGFP)sk72 was reported previously (Waxman and Yelon, 2011).

Construction of the Transgenes and Fusion Proteins

For constructing the Tol2 Gateway destination vector with the α-crystallin:DsRED (α-cry:DsRED), PCR amplification was used to add ClaI restriction sites to the 5′ and 3′ ends of the previously reported α-cry:DsRED plasmid (Lee et al., 2009). This fragment and the vector pDestTol2p2A were then digested with ClaI and the fragment was ligated into the pDestTol2p2A using standard cloning methods. The GDBD-RLBD and VPBD-RLBD constructs were then cloned upstream of the β-actin promoter using the Tol2/Gateway system (Kwan et al., 2007) and transgenic animals were made using standard Tol2 methods (Kawakami, 2005).

The GDBD-RLBD was made using PCR to fuse the amino acids 156–457 of zebrafish RARαb, which comprise the D-F domains, to amino acids 1–147 of the Gal4 protein, which contains the DNA-binding domain. Amino acids 413–490 of VP16 were fused to the N-terminus of the GDBD-RLBD to make the VPBD-RLBD construct. For this fusion, the start methione was deleted from the GDBD and added to the VP16 domain. Both constructs were initially cloned into pCS2p+ as an intermediate vector prior to cloning into the pDestTol2p2A-αcry:Dsred destination vector along with the β-actin promoter for making transgenic animals.

Drug Treatments and Morpholinos Used

Embryos were treated with 1 μM RA (Sigma, St. Louis, MO), 0.5 μM AM580 (Biomol, Plymouth Meeting, PA), 25 μM Keto (Sigma), 1 μM DEAB (Sigma), and 0.75 μM RO41 (Biomol) beginning at the shield stage; 0.003% w/v 1-phenyl 2-thiourea (PTU; Sigma) was used beginning prior to 24 hpf to prevent pigmentation.

Sequences for cyp26a1 MOs used were:


Sequence for the cyp26c1 MO was reported previously (Hernandez et al., 2007). A mixture of 2 ng cyp26a1 MO1 and 1 ng cyp26a1 MO2 and 6 ng cyp26c1 MO were injected into one cell Tg(β-actin:GDBD-RLBD);(UAS:EGFP) embryos. Deficiency of both Cyp26a1 and Cyp26c1 has been shown previously to more effectively cause increases in embryonic RA (Hernandez et al., 2007). One ng of p53 MO was used to prevent not-specific p53 mediated cell death (Robu et al., 2007).


Images were taken using a Zeiss (Thornwood, NY) M2BIO fluorescent stereomicroscope and Axiocam2 camera. Confocal images were taken using a Nikon_A1 inverted confocal microscope. Images were prepared using Adobe Photoshop.


ISH was performed essentially as reported previously (Oxtoby and Jowett, 1993).


We thank Steve Farber for informative discussions. T.Z. and J.T.Z. were supported by NIH grants R01 HL092263 and R01 HL092263-01A1. A.M., A.R., J.A., M.R.J.S., and J.S.W. were supported by NIH grants R00 HL901126 and R01 HL112893-A1 and a CCHMC Trustee Award.