Ubiquitous transgene expression and Cre-based recombination driven by the ubiquitin promoter in zebrafish

A ubiquitous promoter/enhancer to drive transgenes is a key component of a complete transgenesis toolkit in any model organism. Enhancer trap screening in the mouse previously identified the Rosa26 locus, which became the current gold standard for ubiquitous transgene expression and for Cre/loxP-based lineage tracing at all stages of development and in all adult tissues (Friedrich and Soriano, 1991; Soriano, 1999; Zambrowicz et al., 1997). The lack of a homologous tool in zebrafish has particularly limited the full potential of Cre/loxP-based lineage-tracing approaches that complement the powerful imaging possibilities and genetic malleability of the model system. As genomic knock-in methods for zebrafish are currently missing, a broadly applicable ubiquitous transgene control element needs to be accessible in simple cloning vectors. Strong candidates for ubiquitous zebrafish promoters/enhancers are thus control regions of genes encoding factors required in fundamental cellular processes.

Several native zebrafish as well as exogenous control elements have been previously deemed ubiquitous in vivo and successfully applied in genetic experiments; these include the zebrafish h2afx, tbp, versions of the β-actin control elements (Burket et al., 2008; Gillette-Ferguson et al., 2003; Higashijima et al., 1997; Kwan et al., 2007), and the Xenopus laevis-derived elongation factor 1a promoter (XlEef1a1) (Johnson and Krieg, 1994; Kawakami et al., 2004). Each of these promoter fragments has shortcomings in different developmental stages and cell types. A common problem is their progressive inactivation during the course of development, particularly in parts of the nervous system, the fin fold, hematopoietic cell populations, or even ubiquitously in all tissues. For example, the commonly applied XlEef1a1 reportedly only expresses strongly during gastrulation and midsomitogenesis before being restricted to specific cell types, although prolonged expression has been reported for isolated transgenes (Collins et al., 2010; Hans et al., 2009). The β-actin promoter fragments do retain expression in adults, yet show no significant activity in erythrocytes or fins, or several other cell types (Burket et al., 2008; Traver et al., 2003). This likely reflects specialization of differentiating cells, potential cell-type-specific requirements for basic molecular mechanisms controlling translation or chromatin maintenance, as well as transgene dependence on the genomic integration context.

Lineage-tracing transgenes using currently available transgene promoters in zebrafish therefore do not allow complete cell-fate tracing to all descendant tissues nor lineage analysis over prolonged time periods. Furthermore, the extensive creation of loxP cassette-based transgenes requires a reproducible and reliable driver element with broad cloning compatibility. Seminal work by Hans et al. (Hans et al., 2009) established the functionality of tamoxifen (TAM)-inducible Cre^ERt2-mediated loxP recombination (Feil et al., 1996; Feil et al., 1997) in zebrafish. The fast uptake and action of the active TAM metabolite 4-hydroxytamoxifen (4-OHT) triggers dose-dependent Cre^ERt2-mediated loxP excision events within 2-4 hours of administration (Hans et al., 2009). As the drug is easily applied through simple addition to the embryo medium, 4-OHT-inducible Cre^ERt2 is a desirable tool for studying zebrafish development. However, to reliably and flexibly trace loxP excision events in cell-lineage experiments at all stages of development, or to trigger Gal4 and other genetic modifiers using tissue-specific Cre^ERt2 sources, a ubiquitously expressed loxP switch construct driver is eagerly needed (Blackburn and Langenau, 2010; Hans et al., 2009).

In Drosophila, the strongest and most faithful ubiquitous control element currently in use is derived from the 5′ region of the ubiquitin (ubi) gene, which encodes the evolutionarily conserved multimeric Ubiquitin (Ubi) peptide precursor (Lee et al., 1988). ubi-driven green fluorescent protein (GFP) transgenes are used widely as reliable markers for various genetic applications at all stages of Drosophila development. The evolutionarily conserved ubi control region has also been cloned from maize and has found widespread application in plant transgenesis (Christensen et al., 1992). The human ubi promoter has been isolated and drives strong transgene expression in a variety of cell culture systems (Schorpp et al., 1996). Notably, UBI-GFP/BL6 transgenic mice express EGFP in virtually all cells of the hematopoietic system, thus greatly enhancing its utility in bone marrow stem cell transplantation experiments (Schaefer et al., 2001).

Based on this precedent, we sought to identify and characterize a ubiquitin locus in zebrafish to generate an endogenously derived ubiquitous transgene driver. Analogous to the reported Drosophila, maize, and human loci, we cloned a 3.5 kb 5′ region upstream of the zebrafish ubiquitin B translational start site spanning the proximal upstream region, the first noncoding exon, and the only intron of the gene. In transient injections into zebrafish embryos, the ubi control region drives strong and ubiquitous visible expression of an EGFP reporter gene within 4 hours post-injection, likely starting at the onset of zygotic transcription. Stable ubi:EGFP lines reveal strong expression in all analyzed external and internal organs, including the retina, fin fold, and across all blood cell types from embryo to adulthood. The blood expression of ubi-driven fluorophores enabled us to perform adult zebrafish transplants with whole kidney marrow (WKM), which harbors the adult hematopoietic stem cells, into irradiated recipients and to monitor true multilineage repopulation from the transplanted cell populations. We further created ubi:cre^ERt2 transgenes and showed they confer rapid and ubiquitous 4-OHT-inducible Cre activity at different stages of development. We further established ubi:loxP-GFP-loxP-mCherry lineage tracer transgenics that provide strong reporter activity upon Cre exposure. The ubi promoter and derived fluorescent reporters, the newly established lineage tracer transgenic lines, and the Cre^ERt2/loxP procedures reported here therefore provide an effective and valuable addition to the current zebrafish transgenesis tool box.

Zebrafish ubi (ZDB-GENE-050411-10) promoter sequence was PCR amplified from bacterial artificial chromosome (BAC) CH211-202A12 using the Expand High Fidelity PCR Kit (Roche, Indianapolis, IN, USA) with primers 5′-AAAACTCGAGACCAGCAAAGTTCTAGAATTTGTCG-3′ (forward) and 5′-AAAGGATCCCTGTAAACAAATTCAAAGTAAGATTAGC-3′ (reverse), and TOPO-cloned into pENTR5′ (Invitrogen, Carlsbad, CA, USA) to create pCM206 (pENTR5′_ubi). pCM206 thus frames the 3.5 kb ubi promoter fragment, including the first non-coding exon and the intron, with L4 and R1 sequences for MultiSite Gateway, as well as 5′ XhoI and 3′ BamHI restriction sites (underlined in primer sequence) to facilitate cloning of ubi-driven constructs.

pENTR/D_cre^ERt2 was created by PCR amplification of the cre^ERt2 open reading frame (ORF) from pCre^ERt2 (Feil et al., 1997) using the Expand High Fidelity PCR kit (Roche) with primers 5′-CACCATGTCCAATTTACTGACCGTACACCAAAATTTGCC-3′ (forward, ATG start codon underlined) and 5′-CAGACATGATAAGATACATTGATGAGTTTGGACAAACC-3′ (reverse) and directional TOPO cloning into pENTR/D (Invitrogen).

pENTR5′_ubi:loxP-GFP-loxP was created by BamHI excision of the loxP-EGFP-loxP cassette from β-actin-loxP-EGFP-loxP-kras^G12D (Le et al., 2007) and subcloning into the BamHI site of pCM206.

All subsequent MultiSite Gateway assemblies were carried out using LR Clonase II Plus (Invitrogen) according to standard protocols and Tol2kit vectors described previously (Kwan et al., 2007), if not stated otherwise.

Tg(–3.5ubi:EGFP) and Tg(–3.5ubi:mCherry) are multisite gateway assemblies of pCM206, Tol2kit #383 (pME-EGFP), #302 (p3E_SV40polyA) and #394 (pDestTol2A2) (ubi:EGFP), or pCM206, Tol2kit #386 (pME-mCherry), #302, and #394 (ubi:mCherry).

Tg(–3.5ubi:cre^ERt2;cmlc2-EGFP) derives from vector pDestTol2pA2_ubi:cre^ERt2;cmlc2-EGFP (pCM283). pCM283 was created by combining pCM206, pENTR/D_cre^ERt2, and Tol2kit vectors #302 and #395 (pDestTol2CG).

Tg(–3.5ubi:loxP-EGFP-loxP-mCherry) (ubi:Switch) bases on pDestTol2A2_ubi:loxP-EGFP-loxP-mCherry, which was assembled with pENTR′5_ubi:loxP-GFP-loxP and Tol2kit vectors #383, #302, and #394. Detailed digital plasmid maps of these vectors are available upon request.

Zebrafish were maintained in accordance with Animal Research Guidelines at Children's Hospital Boston. Zebrafish transgenics were generated using Tol2-mediated transgenesis: 25 ng/μl of each transgenesis vector were combined with 2.5 ng/μl Tol2 mRNA as described (Kwan et al., 2007) and injected into one-cell-stage eggs derived from AB or Tu wild-type crosses, and the injected animals grown up to adulthood. Individual F0 founders were outcrossed to TL, and their F1 progeny were screened for the respective transgenesis marker (ubi:EGFP, ubi:mCherry, or cmlc2-EGFP, respectively). On average we achieved efficiencies of F0 transmission of higher than 50%. Positive individual F1 adults were subsequently outcrossed to wild-type zebrafish, and their offspring were then used to establish lines when 50% transgene transmission was observed in the F2 generation, indicating single transgene insertions. If promising clutches had more than 50% positive F2 embryos, individual F2 adults were outcrossed to wild type and their offspring analyzed for 50% transgene positivity before establishing lines. Genotyping of adult ubi:EGFP fish was facilitated by the use of UV filter goggles (Modular Fluorescence Head set type: FHS, frame FHS/F-01, lightsource FHS/LS-1B, emission filter FHS/EF-2G2; BLS Ltd, Budapest, Hungary), which reveal EGFP fluorescence (or other fluorophores depending on the filter set) in zebrafish under regular light conditions and do not require any manipulation of the animals.

Flow cytometry was performed on a BD LSR II Flow Cytometer (BD Biosciences, San Jose, CA, USA) as described previously (Traver et al., 2003). Retro-orbital adult WKM transplantation of ubi:mCherry cells into casper recipients was performed as described previously (Pugach et al., 2009).

4-Hydroxytamoxifen (4-OHT, H7904; Sigma, St Louis, MO, USA) was dissolved in ethanol at a final stock concentration of 10 mM and kept in single-use aliquots in the dark at –20°C. 4-OHT is light- and temperature-sensitive, and we found it loses activity after prolonged storage. To induce Cre activity in cre^ERt2-expressing embryos, 25-30 stage-matched embryos per single well of a six-well culture plate were washed with E3 embryo medium, all medium removed and replaced with E3 medium freshly mixed with 5 μM 4-OHT. The treated embryos were immediately put into a closed and dark 28.5°C incubator to allow for efficient induction and remained in 4-OHT-containing E3 medium for the rest of the experiment. For the 4-OHT pulse experiment, the 4-OHT-containing E3 medium was removed after 15 minutes, the embryos re-suspended in minimal volume of E3 medium, and twice transferred into a fresh glass beaker with 500 ml of fresh E3 medium to remove all traces of 4-OHT. The embryos were subsequently placed in a fresh well of a six-well culture plate in E3 medium and grown as described previously.

Adult zebrafish were euthanized using approved protocols with tricaine and then fixed in 4% paraformaldehyde (PFA). Paraffin embedding, sectioning, Hematoxylin and Eosin (H&E) staining, and immunohistochemistry (IHC) for enhanced GFP (EGFP) were performed using standard protocols by the Dana Farber/Harvard Cancer Center pathology core facility. EGFP was detected using the JL-8 mouse monoclonal antibody (Clontech, Mountain View, CA, USA).

BLAST searches using the human Ubiquitin peptide ORF as query identified the predicted zebrafish gene locus zgc:172187 on linkage group five, referred to here as zebrafish ubiquitin B (ubi) (Fig. 1A). The zebrafish ubi gene features a 71 bp non-coding first exon, a 2 kb spanning intron, followed by a second exon encoding a multimeric (eight repeats) Ubiquitin peptide precursor (Fig. 1A), a gene structure homologous to ubi loci in other species.

Analogous to the previously reported Drosophila and human ubi control region fragments (Lee et al., 1988; Schaefer et al., 2001; Schorpp et al., 1996), we PCR-cloned a 3.5 kb sequence from BAC CH211-202A12 immediately upstream from the ATG start codon at the 5′ end of exon 2, including the putative intron/exon splice junctions (Fig. 1A). To test if this 5′ ubi fragment contains functional transcriptional regulatory sequences, we assembled the Tol2(–3.5ubi:EGFP) reporter vector and injected it together with Tol2 transposase-encoding mRNA into one-cell-stage zebrafish embryos (Kikuta and Kawakami, 2009). Within 4 hours postinjection, we could detect strong mosaic EGFP expression in the injected embryos by fluorescence microscopy (Fig. 1B) confirming that our isolated 5′ ubi fragment contains sufficient regulatory sequences to drive EGFP transgene expression. This early detection of ubi:EGFP expression suggests an onset of transgene transcription at the mid-blastula transition (2.75 hours post fertilization, hpf), in agreement with active histone marks at the ubi locus at the mid-blastula transition (Vastenhouw et al., 2010). The injected embryos robustly showed mosaic EGFP-positive cells in a variety of cell types and organ systems over the first days of development (Fig. 1C).

To assess the expression pattern of germline transmitted transgene integrations, we grew mosaic EGFP-positive embryos from the F0 generation to adulthood while periodically monitoring EGFP expression. This was facilitated by using UV filter goggles to nonintrusively detect EGFP in vivo (see Materials and methods for details). Throughout development to adulthood, the ubi:EGFP-injected animals displayed strong mosaic EGFP in all visible external organs, such as the skin, eyes, and fins (Fig. 1D).

Individual EGFP-positive F0 males were crossed to wild-type females, and their F1 progeny screened for EGFP expression. In several independently derived F1 clutches from different founder parents, EGFP expression became obvious by 4-5 hpf in all blastula cells. These EGFP-positive embryos showed strong ubiquitous expression throughout development and adult stages (Fig. 1F-G′). To establish stable transgenic lines, we selected three independent male F1 founders that give rise to progeny with 50% EGFP-positive F2 embryos, suggesting a single Tg(–3.5ubi:EGFP) integration (henceforward referred to as ubi:EGFP, results shown for transgenic lines F and V). Outcrossing of adult F2 females to wild-type males revealed strong maternal deposition of ubi-derived EGFP into the egg (Fig. 1E). Incrossing of ubi:EGFP F2 parents generated approximately 25% of EGFP-positive embryos emitting notably more intense fluorescence signal than their other positive siblings. These animals were subsequently found to be homozygous for ubi:EGFP (data not shown). Our observations thus reveal the strong transgene driving activity of the zebrafish ubi promoter and its functional characteristics including maternal contribution, allele dosage effects, and permanent promoter expression throughout zebrafish development.

To determine if ubi:EGFP truly expresses ubiquitously, F2 animals that were heterozygous based on Mendelian transgene inheritance were examined and displayed EGFP in a homogenous pattern in all externally visible body structures including the fins (Fig. 1G). These tissues do not show expression with commonly used ef1α, tbp or β-actin promoter-driven reporters. Anti-EGFP IHC on sagittal sections of heterozygous F2 ubi:EGFP adults further revealed ubi:EGFP expression in a multitude of internal organs (Fig. 2). Importantly, tissues that have so far remained untargeted by other putatively ubiquitous promoters, such as retina and erythrocytes, were EGFP-positive (Fig. 2B,C,F). Notably, EGFP IHC revealed that EGFP expression is not homogenous throughout all cell types. We found small regions in internal organs that, despite positive staining, showed substantially lower EGFP levels than adjacent tissues or cell types, such as cell bodies of subsets of CNS neurons (Fig. 2E), or columnar epithelial cells of the kidney tubules (Fig. 2G).

Overall, we therefore conclude that our newly isolated and MultiSite Gateway-compatible zebrafish –3.5 kb ubi promoter drives transgene expression ubiquitously in the vast majority of cell types, if not completely ubiquitously. ubi expression includes tissues previously untargeted by transgene expression, features maternal deposition, and persists through all stages of zebrafish development, starting at the mid-blastula transition.

The β-actin promoter that is commonly used as a transgenic marker in zebrafish blood cell transplantation experiments is not active in erythroid cells (Traver et al., 2003). Our sagittal sections and IHC indeed captured erythrocytes as ubi-positive (Fig. 2F), yet these are the only unambiguously identifiable blood cell type using this method. We therefore sought to analyze ubi expression in the hematopoietic system and profiled the blood expression of adult ubi:EGFP and ubi:mCherry heterozygous transgenic zebrafish using flow cytometric analysis of dissected WKM (Traver et al., 2003), the major hematopoietic tissue in adult zebrafish. The analyzed ubi:EGFP and ubi:mCherry transgenic lines showed robust fluorophore expression in all resolvable blood cell gates (Fig. 3A and Fig. 4A) and in the vast majority of cells in these respective populations, including erythroid, myeloid, and lymphoid lineages, plus precursor cell populations (Fig. 3B,C and Fig. 4B,C). This is in stark contrast to β-actin promoter-driven EGFP transgenics, which overall show weaker expression in hematopoietic lineages, particularly in erythrocytes, which barely express β-actin:EGFP (Fig. 3D-F).

This broad hematopoietic expression should therefore allow assessment of multilineage engraftment in transplantation experiments using ubi:EGFP or ubi:mCherry. To test this application, we isolated WKM from either ubi:mCherry or ubi:EGFP heterozygous adults, and performed retro-orbital WKM transplantation into irradiated roy^–/–; nacre^–/– (casper) (White et al., 2008) transparent recipients to allow for easy visualization. Three weeks post-transplantation, the transparent casper recipients showed strong localized fluorescence marking the kidney territory (Fig. 4D). Flow cytometry analysis of WKM from these casper recipients revealed chimeric contribution of fluorophore-expressing donor cells to all light-scatter resolvable blood cell populations, as described above (Fig. 4E-G). We have successfully monitored ubi:EGFP and ubi:mCherry WKM-transplanted casper recipients with long-term transplants over 3 months post transplant, as well as performed transplantation with mixed ubi:EGPF and ubi:mCherry populations in competitive transplants, without detecting any diminishing fluorophore intensity (data not shown). ubi-driven transgenic reporters are therefore suitable reagents for hematopoietic transplantation experiments, and their true multilineage expression offers a clear advantage over previously used transgene reporters.

Cre/loxP system-based lineage tracing tools are becoming increasingly effective in zebrafish with single-insertion loxP cassettes generated through Tol2-mediated transgenesis (Kawakami, 2007; Yoshikawa et al., 2008). The Cre^ERt2 recombinase fusion protein is inducible by exposure to TAM or its active metabolite 4-OHT (Feil et al., 1996; Feil et al., 1997) and is thus ideal for spatiotemporal tracing of cell types during zebrafish development (Hans et al., 2009). The features of ubi-driven reporter transgenes prompted us to create Tg(–3.5ubi:cre^ERt2;cmlc2-EGFP) (subsequently referred to as ubi:cre^ERt2) using Tol2-mediated transgenesis to provide a ubiquitous inducible Cre^ERt2 source (Fig. 5A). We established three independent zebrafish lines carrying ubi:cre^ERt2 insertions and subsequently used Tg(–3.5ubi:cre^ERt2;cmlc2-EGFP)V as a representative line.

To characterize ubi-driven Cre^ERt2 function, we crossed ubi:cre^ERt2 males (to circumvent maternal Cre^ERt2 contribution and assure all Cre^ERt2 derives from de novo expression) to lmo2:loxP-dsRed-loxP-EGFP (lmo2:Switch) females (Wang et al., 2008) (Fig. 5A). lmo2 promoter-driven transgenes start to express from 10-12 hpf in hematopoietic and endothelial precursors (Wang et al., 2008; Zhu et al., 2005). When incubated with 5 μM 4-OHT starting from shield stage, ubi:cre^ERt2; lmo2:loxP-dsRed-loxP-EGFP double-positive embryos express strong and mosaic EGFP in the vasculature and hematopoietic system by 24 hpf (Fig. 5B-E), thus revealing 4-OHT dependent Cre^ERt2 activity. These crosses revealed that ubi-derived Cre^ERt2 is absolutely 4-OHT dependent for its activity, as ubi:cre^ERt2; lmo2:loxP-dsRed-loxP-EGFP double-positive embryos never showed any detectable GFP expression by fluorescence when treated with ethanol solvent control (Fig. 5F-I). Strikingly, fluorophore-expressing cells were always single-positive for either EGFP or mCherry (Fig. 5C-E), suggesting that 4-OHT activity, and thus the probability of Cre^ERt2-mediated loxP recombination events, fades during the time window between induction at shield stage and onset of lmo2 switch expression.

To independently test ubi:cre^ERt2 functionality, we crossed ubi:cre^ERt2 males to females carrying the previously reported fusion promoter-driven eab2:[EGFP-T-mCherry] (FlEX) (Fig. 5J). Prior to recombination, FlEX broadly expresses EGFP and has been proposed to be ubiquitous (Boniface et al., 2009). Cre^ERt2 activation with 4-OHT at shield stage triggered highly mosaic and strong mCherry expression in double-transgenic embryos as observed 3 days post fertilisation (dpf) (Fig. 5K,L), whereas ethanol solvent control-treated embryos never produced detectable mCherry expression during the observation period (Fig. 5K,L). Together, these experiments establish that ubi promoter-driven Cre^ERt2 recombines loxP site-flanked transgenes specifically upon 4-OHT addition in zebrafish embryos.

We noted in our ubi:cre^ERt2 experiments (Fig. 5 and see above) different degrees of switch construct expression mosaicism despite early 4-OHT induction at shield stage. This suggests that either: (1) ubi-driven Cre^ERt2 is not efficiently excising the tested loxP switch; or (2) the loxP switch transgenes have integrated in suboptimal genomic locations that decrease the probability of Cre-dependent loxP site recombination. To create new loxP switch insertions devoid of functionally relevant position effects and to generate a lineage tracing reagent with the temporally stable expression benefits of the ubi promoter, we assembled Tg(–3.5ubi:loxP-GFP-loxP-mCherry) (subsequently referred to as ubi:Switch, Fig. 6A). We established three independent transgenic ubi:Switch zebrafish lines and verified their ubiquitous EGFP expression in the F2 generation [the subsequent experiments use Tg(–3.5ubi:loxP-GFP-loxP-mCherry)III].

To assess Cre^ERt2 responsiveness of ubi:Switch, we crossed ubi:cre^ERt2 males to ubi:Switch F2 females, again to ensure active Cre^ERt2 de novo expression through paternal transmission (Fig. 6A). A first group of the resulting embryos was induced with 5 μM 4-OHT at shield stage, while a second group of siblings was mock treated with ethanol solvent. To also assess Cre^ERt2 induction dynamics, we performed a pulse treatment experiment by exposing a third group of embryos for 15 minutes to 5 μM 4-OHT, before thoroughly washing the embryos with E3 medium to attempt to remove all 4-OHT in solution. All the embryos subsequently grew up in the dark for 3 days before analysis to allow for mCherry accumulation for assessment of resulting switch mosaicism.

ubi:cre^ERt2; ubi:Switch double-positive embryos induced with 4-OHT at shield stage showed strong and widespread mCherry expression at 3 dpf, which intensified further over time with steadily decreasing EGFP fluorescence (Fig. 6B). Ethanol-treated control, as well as ubi:cre^ERt2-negative (not shown), embryos remained solely EGFP-positive (Fig. 5B). Strikingly, the 15 minute 4-OHT incubation group revealed strong and widespread mCherry mosaicism that remained less complete than for the first embryo group left in 4-OHT (Fig. 6B). This observation suggests that 4-OHT rapidly enters zebrafish embryos and conveys broad Cre^ERt2 activity.

We next induced ubi:cre^ERt2; ubi:Switch double-positive embryos with varying concentrations of 4-OHT starting at bud stage. These experiments revealed concentration-dependent loxP recombination efficiency when read out at 30 hpf, with 5 μM 4-OHT having the strongest effect (Fig. 6C), with a weakened response in the group exposed to 1 μM 4-OHT (Fig. 5D). However, 0.5 μM 4-OHT only inefficiently triggered mCherry expression (Fig. 6E,G).

To assess ubi:cre^ERt2 activity during later stages, we induced a cohort of ubi:Switch; ubi:cre^ERt2 double-positive embryos with 4-OHT at 24 hpf, and 6 dpf. Compared with their siblings induced at shield stage, the embryos exposed to 4-OHT at 24 hpf showed slightly lower mCherry mosaicism with weaker mCherry levels when observed at 3 dpf (Fig. 6H), which again intensified over developmental time with decreasing EGFP fluorescence. The 6 dpf cohort, when observed at 9 dpf, showed widespread mCherry expression, with faint expression in the head region (Fig. 6I). These effects are expected because: (1) more cells exist in the embryo at 24 hpf and 6 dpf, increasing the number of successful Cre^ERt2-mediated loxP excisions required to give homogeneous mCherry expression; (2) mCherry has less time to accumulate and mature starting from 24 hpf than after loxP excision events at shield stage; and (3) tissue penetration in the growing zebrafish larvae may limit 4-OHT access, particularly in the head region.

In summary, we conclude that: ubi:cre^ERt2 generates potent Cre^ERt2 over a broad developmental window to induce tissue-specific loxP switch lines; different loxP switch transgenes have different sensitivities to Cre^ERt2 activity, potentially depending on their genomic integration site and loxP sequence accessibility; and 4-OHT is rapidly taken up by zebrafish embryos and is able to trigger transient Cre^ERt2 activity within minutes of application, in agreement with previously reported findings (Hans et al., 2009). This last observation suggests the use of temporal pulses of 4-OHT induction during zebrafish development to trace temporally distinct cell populations, an application that is technically challenging in mouse development.

The number of potent zebrafish tools to manipulate genetic events and trace cell lineages through development has steadily increased in recent years; however, a common obstacle has been the lack of a ubiquitous transgene driver (Blackburn and Langenau, 2010; Collins et al., 2010; Hans et al., 2009). Our identification and characterization of the zebrafish ubi promoter described here adds a promising new candidate to fill this gap. ubi expresses from early development throughout the zebrafish life cycle (Figs 1, 2, 3). We have further monitored stable ubi-driven transgene expression through multiple generations (F5 with ubi:EGFP), suggesting no particular susceptibility to transgene silencing. The ubi promoter fragments practical size (3.5 kb) plus compatibility with both MultiSite Gateway and traditional restriction enzyme cloning allows for simple use in applications that depend on persistent ubiquitous transgene expression, as illustrated by our adult WKM transplantation experiments with ubi-driven reporters (Fig. 4). Although our analysis cannot rule out cell types that fail to express ubi-driven transgenes, ubi provides altogether a substantial and significant improvement compared with existing supposedly ubiquitous promoters in zebrafish.

Still, a `ubiquitous' promoter should not be confused with a `homogenous' promoter: ubi does not express at the same level in all cell types. We noted markedly different expression levels by EGFP expression in subsets of cell types within various tissues, such as the brain, retina, kidney (Fig. 2) and blood (Fig. 3). When comparing independent ubi:EGFP or ubi:mCherry transgenics, we also found different fluorophore intensity distributions in our fluorescence-activated cell sorting analysis for WKM (compare Fig. 3B and Fig. 4B; and data not shown), as also described for UBI-GFP/BL6 and ROSA26 mice (Schaefer et al., 2001). Such effects are anticipated for randomly integrated transgene constructs, which are susceptible to environmental influences of the genomic integration locus.

Position effect influences are especially pronounced in loxP:Switch constructs. We found several independent ubi:Switch lines that showed weak and greatly varying Cre-mediated excision potencies (data not shown). Although significantly increasing the required workload, these observations suggest that loxP switch transgenes need to be functionally screened for efficient integration lines, by using ubi:cre^ERt2 for example. Screening with Tg(hsp70:cre), particularly when this transgene is maternally contributed and thus pre-loads the egg cell with mRNA or possibly protein, probably greatly overestimates the loxP recombination efficiencies of switch lines (Hans et al., 2009). Cre^ERt2-mediated recombination induced at different developmental time points exposes loxP transgene dynamics more reliably through de novo transcription events after recombination (see also Fig. 6).

By combining the tissue-specific Tg(pax2.1:cre^ERt2) as a recombinase source with ef1α:loxP-dsRed-loxP-GFP, Hans et al. (Hans et al., 2009) reported Cre^ERt2 sensitivity to 0.5 μM 4-OHT, which at the time points we tested triggered minimal recombination with our ubi-driven transgenes (Fig. 6). This is probably owing to stronger cell-type-specific specific Cre^ERt2 expression in the diencephalon and hindbrain through the pax2.1 driver. These observations combined with our findings suggest that the optimal 4-OHT dose for a given Cre^ERt2 driver paired with a loxP switch will require individual elucidation.

Nonetheless, the use of Cre^ERt2 in zebrafish reveals a striking simplicity and efficiency for drug-inducible loxP cassette recombination, thus crucially extending existing genetic manipulation methods. Furthermore, the possibility of pulsed Cre^ERt2 induction in zebrafish allows for control of the onset of cell lineage labeling and misexpression in restricted developmental time windows. Similar experiments in mouse embryos are technically challenging and ambiguous in their readouts, as TAM needs to be injected or administered in the drinking water of the mother, which does not allow for uniformly controlled temporal induction. Zebrafish molecular genetics have received a major boost with the dawn of transposon-mediated transgenesis (Kawakami, 2005; Kwan et al., 2007; Villefranc et al., 2007); yet much work remains to establish reliable cloning standards and transgenesis markers. The ubi promoter now provides a solid basis to the construction of elaborate recombinase-dependent or drug-inducible transgene systems for all developmental stages in zebrafish.

We thank A. Chen and K. Lawson for technical assistance, C. Lawrence and I. Adatto for fish husbandry, and Drs S. Lacadie and K. Kathrein for insightful discussion and critical comments. C.M. received support through an EMBO long-term fellowship and a HFSP long-term fellowship. C.K.K. is supported by NIH grant 5T32CA09172-34. O.J.T. is funded by fellowships from the Heart and Stroke Foundation of Canada and the Canadian Institutes of Health Research. This study was supported by National Institutes of Health grant 5PO1HL32262-28. L.I.Z. is an HHMI investigator. Deposited in PMC for release after 6 months.