Zebrafish Enhancer Detection (ZED) Vector: ANew Tool to Facilitate Transgenesis and theFunctional Analysis of cis-Regulatory Regionsin ZebrafishJose Bessa,1* Juan J. Tena,1 Elisa de la Calle-Mustienes,1 Ana Fernandez-Minan,1 Silvia Naranjo,1
Almudena Fernandez,2 Lluis Montoliu,2 Altuna Akalin,3 Boris Lenhard,3 Fernando Casares,1 andJose luis Gomez-Skarmeta1*
At present, considerable efforts arebeing devoted to understanding tran-scriptional gene regulation. Althoughseveral methods have been developedto identify cis-regulatory sequences invertebrates (Allende et al., 2006; Pen-nacchio et al., 2006), their in vivo test-ing and validation remains a signifi-
cant challenge. To date, most discovery/validation protocols have been carriedout in mouse (Visel et al., 2007). How-ever, mouse as a model organism is bestsuited for small/medium-scale screens.Some of its limitations are overcome byusing zebrafish, which allows for rapid,large-scale genomic screening, and hasbeen shown to be a good model for effi-cient detection of enhancer activity in
vivo (de la Calle-Mustienes et al., 2005;Allende et al., 2006; Navratilova et al.,2009). However, the genetic tools usedin zebrafish for enhancer detection arefar from optimized.
The commonly used transgenesisvectors consist of a shuttle vector, aminimal promoter and an in vivo re-porter gene (Nobrega et al., 2003; Pen-nacchio et al., 2006; Woolfe et al.,
Additional supporting information may be found in the online version of this article.1Centro Andaluz de Biologıa del Desarrollo (CABD), CSIC-Universidad Pablo de Olavide, Campus UPO, Seville, Spain2Centro Nacional de Biotecnologıa (CNB-CSIC), Campus de Cantoblanco, and Centro de Investigacion Biomedica en Red de EnfermedadesRaras (CIBERER), ISCIII, Madrid, Spain3Computational Biology Unit, Bergen Center for Computational Science, University of Bergen, Bergen, NorwayGrant sponsor: Spanish Ministry of Education and Science; Grant numbers: BFU2007-60042/BMC, Petri PET2007_0158, CSD2007-00008;Grant sponsor: Junta de Andalucıa; Grant numbers: Proyecto de Excelencia CVI260, CVI-3488.*Correspondence to: Jose Luis Gomez-Skarmeta or José Bessa, Centro Andaluz de Biologıa del Desarrollo (CABD), CSIC-Universidad Pablo de Olavide, Campus UPO, Ctra. de Utrera km1, E-41013 Seville, Spain. E-mail: [email protected][email protected]
DOI 10.1002/dvdy.22051Published online 3 August 2009 in Wiley InterScience (www.interscience.wiley.com).
2007; Visel et al., 2008). Yet, this sim-ple design faces two major practicalproblems. First, the random integra-tion of the shuttle vector in the ge-nome often causes it to be exposed tothe enhancer activity present in thesurrounding genomic regions, result-ing in reporter gene expression thatdoes not result from the DNA se-quenced cloned in the vector. Thisphenomenon is commonly known as“position effect” (Chung et al., 1993).Second, the lack of a positive control oftransgenesis makes it difficult to de-termine the efficiency of the integra-tion events in F0 injected embryos andif a nonexpressing F1 results from thefish carrying an inactive region or notransgene at all.
Here, we have developed a new toolfor enhancer detection in zebrafishthat avoids these problems: the Ze-brafish Enhancer Detection (ZED)vector. ZED is based on the Tol2Transposon (Kawakami et al., 2000)and introduces several improvements.First, to drive the expression of thereporter gene encoding the enhancedgreen fluorescent protein (EGFP), wehave used a gata2a minimal promoter(Ellingsen et al., 2005), which hasbeen selected from a test set of severalTATA-containing and TATA-less pro-moters because of its high efficiency ofenhancer activity detection. Second,we have placed a Gateway entry site5� of the gata2a minimal promoter tofacilitate the insertion of candidateenhancers in the reporter construct(Hartley et al., 2000; Cheo et al.,2004). Third, the enhancer reportercassette in ZED is flanked by insula-tor sequences that we show decreaseposition effects, thereby increasingthe specificity of the signal. We haveused two different insulators, the5�HS4 insulator from the chicken be-ta-globin gene (Chung et al., 1993; Re-cillas-Targa et al., 2002) and the GABinsulator element from the mouse ty-rosinase gene (Montoliu et al., 1996),which we show here to maintain theirinsulator function in zebrafish. Also,ZED contains a cardiac actin promoterdriving the expression of a red fluores-cent protein (DsRed) that serves ascontrol for transgenesis efficiency invivo in the F0 and the F1 embryos.
Furthermore, ZED contains two ex-cision cassettes, one targeted by Cre-recombinase (loxP sites) and the other
by Flip-recombinase (FRT sites). Weshow here that these cassettes arefunctional and allow the genetic dele-tion, from stable transgenic lines, ofboth the positive control of transgen-esis and the putative enhancer understudy. The latter deletion allows oneto confirm that the enhancer activityin stable lines depends on the elementbeing assayed. Finally, as a collateralproduct of this work, we have alsogenerated an efficient vector for rapidevaluation (in F0 injected embryos) ofinsulator activity in genomic DNA.
RESULTS AND DISCUSSION
Screening for an OptimalPromoter to Test EnhancerActivity
To identify an optimal minimal pro-moter to detect cis-regulatory activity,we compared the sensitivity to en-hancers of different promoters iso-lated from the zebrafish gata2a, irx3a,and rhodopsin genes, and the human�-globin gene (named herein as:pgata2, pirx3, prhodopsin, and p�-glo-bin, respectively). An ideal promotershould be sensitive (i.e., capable of re-sponding efficiently to different en-hancers) and specific (i.e., not able todrive any pattern of expression on itsown). As test enhancers, we isolatedfour different sequences from the ze-brafish genome that are highly con-served among vertebrates and whosehomologous regions have been shownto promote distinct expression pat-terns in mice (Visel et al., 2007). Thesesequences are E187, E200, E261, andE298 (http://enhancer.lbl.gov/; Visel etal., 2007). Each of these sequenceswas cloned upstream of the four pro-moters. EGFP was used as a reportergene. These constructs were assem-bled in a Tol2 transposon and injectedin 0–1 hours postfertilization (hpf)embryos. More than 100 injected (orF0) embryos for each construct wereobserved at 24 hpf (Supp. Fig. S1A,which is available online) and em-bryos with a reproducible EGFP ex-pression pattern were scored. In thisassay, we found that, with the fourdifferent enhancers, pgata2 and pirx3allowed GFP detection on an averagefrequency of 35% and 47% of the in-jected embryos, respectively (Supp.Fig. S1B). These results are consider-
ably higher than the average fre-quency of expression obtained withthe other two promoters tested (Supp.Fig. S1B). In F1s stable lines, pgata2transgenic embryos showed very spe-cific expression patterns. This con-trasted with the expression patternsobserved in pirx3-derived stable lines,which consistently drove some expres-sion in the somites and eye (Fig. 1).These results were confirmed, for eachconstruct, with two or more stablelines derived from independent inser-tion events. Therefore, the highestsensitivity and the good specificity ofpgata2 prompted us to use it as theminimal promoter in the further as-sembling of the ZED vector. In addi-tion, these experiments also demon-strated that the conserved zebrafishsequences maintain their function asenhancers and they are able to driveexpression, to a large extent, in simi-lar anatomical regions than their hu-man counterparts in mouse (Visel etal., 2007; Supp. Table S1).
Implementation of InsulatorSequences to MinimizePosition Effects
An important source of noise in en-hancer detection assays is the cis-reg-ulatory activity from genomic regionssurrounding the transposon insertionsite, also known as position effect(Chung et al., 1993). One way to re-duce this position effect is to flank theintegration cassette with insulators(Montoliu et al., 1996, 2009; Allen andWeeks, 2005). To do this, we have firstdetermined the insulator potential ofknown sequences in zebrafish. Two in-sulator sequences were used, the5�HS4 insulator from the chicken�-globin gene (Chung et al., 1993) andthe GAB insulator element from themouse tyrosinase gene (Montoliu etal., 1996). Although the 5�HS4 insula-tor has been previously tested in ze-brafish (Zhu et al., 2007), no informa-tion in this model was available forthe GAB insulator element. For thatreason, we have developed a new vec-tor to test insulator activity in ze-brafish. This vector is Tol2 transpo-son-based, and contains the zebrafishirx Z48 enhancer (also known asZ54390), which promotes strong ex-pression in the midbrain (de la Calle-
2410 BESSA ET AL.
Mustienes et al., 2005), and the car-diac actin promoter isolated from Xe-nopus laevis driving EGFP. This pro-moter drives strong expression insomites and heart (Mohun et al., 1986;Ryffel et al., 2003). To facilitate thetransference of putative insulator se-quences to the reporter vector, we in-troduced a Gateway entry site be-tween the promoter and the enhancer(Hartley et al., 2000; Cheo et al., 2004;Fig. 2). With this vector design, activeinsulator sequences should decreasethe ability of the midbrain enhancer toactivate the reporter gene. The GFPexpression in the somites, driven bythe Cardiac Actin promoter, servestwo purposes: (1) as an internal con-trol for transformation efficiency (themore GFP-positive somites, thehigher the integration efficiency) and(2) as an internal control for GFP nor-malization. Thus, if compared withmuscle expression, the GFP activity inmidbrain should be reduced in thepresence of an insulator (Fig. 2). Us-ing this vector, with both 5�HS4 andGAB cases we detect a decrease (45–65% of the embryos with decreasedmidbrain enhancer activity, n � 100F0 injected embryos), although not acomplete loss, in the levels of mid-brain-specific GFP when comparedthe control, which carries no insulatorsequence (Fig. 2). This demonstratesthat both insulators function in ze-brafish. Interestingly, these se-quences are not conserved in the ze-
brafish genome. This suggests that,despite of lack of sequence conserva-tion, the molecular machinery that op-
erates through these insulators ispresent and functional in this organ-ism. To test the ability of these insu-
Fig. 1. Capacity of irx3 and gata2 promoters to respond to different enhancers. Two or more independent F1 stable transgenic lines were generatedwith the four tested enhancers (E187, E200, E261, and E298, from left to right columns) positioned 5� of either the irx3 (upper row) or the gata2 (lowerrow) minimal promoters, both driving green fluorescent protein (GFP) expression. Both irx3 and gata2 minimal promoters respond efficiently to all fourenhancers. However, the irx3 promoter shows some promoter-specific expression in somites and eye (arrows).
Fig. 2. The mouse “GAB” and the chicken “5�HS4” insulators are functional in zebrafish. We havegenerated a new Tol2 transposon-based vector to test the insulator potential of candidate sequences.This vector contains a strong midbrain-specific enhancer (Z48) 5� of the Cardiac Actin promoter. A:These two components drive GPF expression in the midbrain (Pale orange bracket), due to activity ofthe Z48 enhancer, and in the somites (pale blue open bracket), due to an endogenous activity of theCardiac Actin promoter. B: If a strong insulator is placed between the enhancer and the promoter, themidbrain expression should be reduced. C–E: When we placed either the mouse GAB or the chicken5�HS4 insulators in this vector, we observe that the ratio between midbrain and muscle expression isdecreased (D,E) compared with that observed in controls bearing the Gateway entry sequence of 1.7kb, which has no insulator activity (C). For each insulator, the reduction of the midbrain enhancer activitywas observed in 45–65% of the injected embryos (n � 100).
NEW VECTOR FOR ZEBRAFISH TRANSGENESIS 2411
lators to reduce the position effects,we injected embryos with a reporterconstruct consisting in the pgata2 pro-moter driving EGFP alone or flankedby 5�HS4 and GAB in the in the 5� and3� region, respectively. Because thesevectors do not harbor any enhancersequence, we assume that any GFPexpression we might observe in theinjected embryos is only due to posi-tion effects. As expected, when the re-porter vector is flanked by insulators,we detect a dramatic decrease in boththe number of embryos showing GFP-expressing cells and in the number ofcells in each embryo with reportergene expression (Fig. 3). Using an in-ternal control of transgenesis de-scribed below, we demonstrated thatthis loss of expression is not due to adecrease in the efficiency of integra-tion events (not shown).
Implementing the CardiacActin Promoter RFPCassette as an InternalControl for Transgenesis
Next, to estimate transgenesis effi-ciency, we introduced the XenopusCardiac Actin promoter driving theexpression of the gene encoding forDsRed (a red fluorescence protein)“outside” of the insulated enhancer re-ported cassette. The Cardiac Actinpromoter activates the expression ofDsRed in heart and, very conspicu-ously, in muscles (Mohun et al., 1986;Ryffel et al., 2003). There are severalbenefits of this DsRed-expressing in-ternal control. First, it allows select-ing among injected embryos thosewhere the transgenesis has occurredefficiently, which correspond to thosewith broad expression of DsRed inmuscles (Fig. 4). Second, this controlallows us to identify unambiguouslytransgenic F1 individuals by observ-ing red fluorescence in heart and mus-cles, independently of whether thetested sequence is capable of induceEGFP expression or not. Of interest,in both F0 and F1 embryos, strongDsRed expression was only detectedat 72 hpf, facilitating the detection ofEGFP at earlier stages (not shown). Adiagram of the final configuration ofthe ZED vector is shown in Figure 5A.To validate the ZED vector, the E187,E200, E261, and E298 enhancers aswell as Z176 (previously denoted
Z54102 in de la Calle-Mustienes et al.,2005) and Z48 were cloned in the ZEDvector, and embryos were injected(Figs. 5B, 6). The F0 embryos showedEGFP expression (30–50% of the in-jected embryos, n � 100) in the ex-pected domains (Fig. 6) and also mus-cle RFP expression (Fig. 5B). As foundwith the empty vector, the injectedembryos showed reduced backgroundas compared to those injected with thenoninsulated pgata2 constructs (Supp.Fig. S1). We also generate several in-dependent stable lines for the Z176and Z48 constructs. These lines ex-press EGFP in territories that showedreporter gene activity in the F0 in-jected embryos (Fig. 6). In addition, inthese lines, RFP is detected at 48–72hpf (insets in Fig. 6).
Flipase- and Cre-MediatedExcision Cassettes
Foreseeing the need to erase the inter-nal control of transgenesis (CardiacActin Promoter-DsRed) or to revertpossible unpredictable or mutageniceffects generated by the GAB insula-tor from stable transgenic lines, wehave designed a Flipase-mediated re-combination cassette by cloning twoFRT sequences (Harrison and Perri-mon, 1993; Werdien et al., 2001)flanking the Cardiac Actin promoter-DsRed-GAB insulator cassette (Fig.5A). In the presence of Flipase, a re-combination event should result in thedeletion of this cassette (Harrison andPerrimon, 1993; Werdien et al., 2001).To test the functionality of these sites,we injected an F1 transgenic stableline with 200 pg of Flipase mRNA. Inthese injected embryos, the homoge-neous muscle-specific expression istransformed into a mosaic DsRed pat-tern, which confirms that the Flipase/FRT-mediated excision of the cassettecan be induced in vivo (Fig. 7A). At theother end of the vector, we have alsodesigned another excision cassettethat includes the 5�HS4 insulator andthe cloning site for the sequence totest. For that, we flanked this regionby Cre recombinase target loxP sites(Fig. 5A). The purpose of this remov-able cassette is to remove the insula-tor or to test if the pattern of EGFPexpression in stable transgenic linescan be only attributed to the clonedsequence (enhancer), or whether it is
contaminated by position effects. Ifthe GFP pattern were exclusively dueto the cloned sequence, Cre-mediatedexcision should transform a homoge-neous tissue-specific EGFP expressioninto a mosaic pattern. To test this cas-sette, we injected 100 pg of Cre mRNAin F1 stable transgenic embryos con-taining the Z48 enhancer-EGFP inser-tion (Fig. 7B). At 24 hpf, most of theCre-injected embryos showed, insteadof a homogeneous EGPF expression, amosaic distribution of the fluorescentprotein in the domain where this en-hancer promotes expression. These re-sults demonstrate that the Cre-medi-ated excision in stable transgeniclines harboring the ZED vector is veryeffective and that, as in this particularcase, it can be used to determine if thepattern of EGFP expression is indeeddriven by the cloned enhancer (Fig.7B). Because in some cases the simul-taneous removal of the insulator couldmake it difficult to discriminate if theresulting EGFP expression dependson the excision of the enhancer or onthe position effect becoming now evi-dent after elimination of the insulator,Cre-mediated enhancer removal shouldbe done in more than one independentstable line. It is very unlikely that twoindependent genomic landscapes pro-mote reporter expression in the samedomain as the enhancer under evalu-ation.
In this study, we have discussed theneed to improve the transgenic vec-tors for enhancer assays in zebrafish.This need arises from the fact thatcurrent standard methods are not ableto cope with common problems associ-ated with this kind of assay, such asthe noise generated by position effectand the ambiguity in the detection ofthe transgenic events. Here, we havepresented an improved vector forfunctional enhancer assays in trans-genic experiments in zebrafish, theZED vector. Several features makethe ZED vector an optimal tool for as-sessing enhancer activity of candidatesequences: (1) it detects enhancer ac-tivity efficiently; (2) it shows reducedsensitivity to the position effect due tothe presence of insulators flanking theEGFP expression cassette; (3) it alsoallows investigators to easily evaluatethe efficiency of transgenesis by directin vivo visualization of a DsRed-posi-tive control cassette in F0 injected em-
2412 BESSA ET AL.
bryos and F1 stable transgenic lines;(4) the implementation of recombi-nation cassettes mediated by Flipand Cre recombinases allow, if de-sired, the deletion of this positivecontrol of transgenesis and/or thecandidate enhancer under evalua-tion. This latter deletion permits tovalidate its enhancer activity by di-rect injecting Cre mRNA in stabletransgenic lines and evaluating itseffects in a matter of days. Despitethe fact that it has been designed fora very specific application, we pro-pose that its basic architecture, com-prising the use of insulators to re-duce the position effect and apositive control for transgenesis, canbe broadly applied as an improvedmethod of transgenesis in zebrafish.Finally, as an important byproductof this work, we have also generateda new vector for easy detection ofinsulator sequences whose applica-tion will be described in more detailelsewhere. These two tools will facil-itate the identification of positive aswell as negative cis-regulatory ele-ments essential for gene regulationand genome organization.
Fig. 3. GAB and 5�HS4 insulators decrease the position effect. A: F0 embryos injected with a Tol2transposon containing the gata2 minimal promoter driving green fluorescent protein (GFP). The GFPbackground expression is due to position effect in independent random insertions and was observedin a large number of cells in 40% of the injected embryos (n � 100). B: Most of this background is lostin embryos injected with similar construct flanked GAB and 5�HS4 insulators, as we observed GFPexpression in only 16% of the injected embryos (n � 100) and in a much reduced number of cells.
Fig. 4. The Cardiac Actin-RFP cassette is effective as an internal control of transgenesis. TheCardiac Actin-RFP cassette drives expression of RFP in the somites. A: The broad RFP expressionin the muscles of this embryo indicates high efficiency of integration. B: In this embryo, thetransgenesis procedure was less effective. Thus, the possibility of germ line integrations is reduced.
Fig. 5. A: Diagram of the Zebrafish Enhancer Detection (ZED) vector. Orange boxes are the Tol2 transposase recognition sequences. This vector iscomposed of two different cassettes. The transgenesis internal control cassette is composed of the Cardiac Actin promoter (pale blue arrow), and redfluorescent protein (RFP; red block). The enhancer detection cassette contains a Gateway entry site, represented by a yellow box, the gata2 minimalpromoter, shown in pale blue arrow, and the enhanced green fluorescent protein (EGFP) reporter gene, marked with a Green box. This enhancerdetection cassette is flanked by two Insulator sequences represented by Violet circles that protect the enhancer detection cassette (dashed purple box)from position effects. The left insulator corresponds to the GAB insulator (G Ins) from the mouse tyrosinase gene and the right to the 5�HS4 insulator(B Ins) from the chicken �-globin gene. Additionally, two excision cassettes are also present. One is mediated by Flipase (black triangles), and the otherby Cre recombinase (gray triangles). B: F0 injected embryos with the ZED vector containing the Z48 enhancer at 48 hours postfertilization (hpf) showboth GFP expression in the midbrain (Green) and RFP expression in the somites (Red) in 70% of the injected embryos (n � 150).
NEW VECTOR FOR ZEBRAFISH TRANSGENESIS 2413
EXPERIMENTALPROCEDURES
pgata2, pirx3, prhodopsin,and p�-globin Vectors
The vectors containing the differentpromoters were built cloning the spe-cific minimal promoter into the XhoIand BamHI sites 5� of EGFP in theT2KHG Tol2 vector (Kawakami et al.,2004). In each vector, a Gateway cas-sette (Invitrogen, Cassette C1 fromGateway Vector Conversion System)was cloned in blunt in the XhoI site.The pgata2 minimal promoter (Menget al., 1997; Perz-Edwards et al., 2001)was directly excised with XhoI andBamHI from the vector pCLGY (Ell-ingsen et al., 2005). The pirx3 mini-mal promoter (de la Calle-Mustieneset al., 2005) was PCR amplified fromzebrafish genomic DNA using theprimers 5�-CTGCAGAGGACCTCGA-CTGGC-3� and 5�-GGATCCGCCCT-CTTGGTGC-3�, which contain a PstIand a BamHI site, respectively (un-derlined). This PCR fragment wasfirst cloned in pBluescript and thenextracted with XhoI and BamHI andtransferred to the Tol2 vector. prho-dopsin was PCR amplified from ze-brafish genomic DNA using theprimers 5�-CTCGAGAAAGTTCTT-ATTATCTACCTGCTGCTCAG-3� and5�-GGATCCGGTTGGATGTGGCGC-TC-3� that contain a XhoI and a BamHIsite, respectively (underlined). p�-glo-bin was directly excised with XhoI andBamHI from the p1230 vector (Man-zanares et al., 2000). The genomic coor-dinates of these promoters and theirdistance from the first methionine ofthe transcript (A from the first ATG isconsidered position 0) are: pgata2(�1031 to �1 from ATG), chr11:2,551,069-2,552,099 (danRer5); pirx3(�746 to �41 from ATG), chr7:27,485,244-27,485,945 (danRer5); prho-dopsin (�287 to �5 from ATG) chr7:2,535,681-2,535,962 (danRer5); p�-globin (�90 to �38 from ATG), chr11:5,204,866-5,204,917 (hg18).
The enhancers used to test thesevectors were PCR amplified from ze-brafish genomic DNA, cloned in TOPOvector (Invitrogen, pCR®8/GW/TOPO®TA cloning KIT) and then recombinedto each vector by the Gateway in vitrorecombination system (Invitrogen,Gateway LR clonase II Enzyme mix).
Fig. 6. The Zebrafish Enhancer Detection (ZED) vector effectively detects enhancer activity. Foreach construct, the number of injected embryos was higher than 100. A: Green fluorescent protein(GFP) expression in 24 hours postfertilization (hpf) mosaic F0 embryos injected with the ZED vectorharboring the E187, E200, E261, E298, Z176 and Z48 enhancers and F1 embryos derived from theZ176 and Z48 constructs. Insets show GFP and red fluorescent protein (RFP) expression in thesetwo last F1 stable lines at 48 hpf. Note the reduced background in these injected embryos ascompared to those injected with the noninsulated pgata2 constructs shown in Supp. Fig. S1. Forthe F1 lines, we obtained more than two independent insertions that showed the same pattern. B:Graphic representation of the percentage of embryos showing GFP expression in the expectedterritories. The reduced background observed with the ZED constructs did not compromise theenhancer detection potential of this vector as the GFP activity was detected in 30–50% of theinjected embryos (n � 100). These percentages are similar to that observed in embryos injectedwith the noninsulated pgata2 constructs containing the same enhancers (Supp. Fig. S1).
2414 BESSA ET AL.
We used the following primers to am-plify these enhancers: E187Forward:5�-TTTGCAAAACATTTTTGACTG-3�;E187Reverse:5�-GATCAGCCATTGT-CACCTCCCAG-3�; E200Forward: 5�-TCTGTGGCATTCTCTTCGGCC-3�;E200Reverse: 5�-TGGCATCTCGCA-CACAAAAGC-3�; E261Forward: 5�-GCAATTTTGCCAATGGTTTCC-3�;E261Reverse: 5�-CCAAAGCCT-TCAAATGCCTCGC-3�; E298Forward:5�-GCGTGGTTCGCATTAAGCAGT-3�;E298Reverse: 5�-CGCTGTGTACAATC-CGAGCCG-3�. The genomic coordinatesof these enhancers in the zebrafish ge-nome (danRer5) are as follows: E187:chr6:30,560,951-30,562,277; E200: chr8:13,856,596-13,858,411; E261: chr16:15,190,519-15,192,608; E298: chr19:38,244,824-38,246,847; E48: chr7:27,681,739-27,682,324; E176: chr7:27,497,995-27,501,393.
We have selected EGFP as the re-porter gene for enhancer detection inour vectors because, in our hands, itgives stronger and earlier signal thanDsRed.
Insulator Vectors
This vector was generated by firstcloning a fragment containing theCardiac Actin promoter driving GFPfrom the pCARGFP vector (Kroll andAmaya, 1996) into the XbaI/NotI re-striction sites of the pminiTol2/MCSvector (Balciunas et al., 2006) to gen-erate the pminiTol2-CARGFP. TheZ48 enhancer was then PCR amplifiedfrom zebrafish DNA using the primers5�-GGTCTAGAGCTCTCGCAGTTGT-GGGC-3� and 5�-CCTCTAGAGGTACCCCCCCTGCTTAAGACACAG-3�,which contain a XbaI restrictionsites in the forward primer and aXbaI and KpnI restriction sites inthe reverse one (underlined). ThisPCR amplified DNA was clonedin the XbaI restriction site of pmini-Tol2-CARGFP to generate pmini-Tol2-Z48-CARGFP. Clones were se-lected having the KpnI site immediatelynear 5� the Cardiac Actin GFP insert.The gateway cassette (Invitrogen,Cassette B from Gateway Vector Con-version System) was cloned in bluntinto the KpnI restriction site generat-ing the insulator vector pminiTol2-
Z48-Gw-CARGFP. Two copies of the�-globin 5� HS4 250-pb insulator (Re-cillas-Targa et al., 2002), previouslycloned in pBluescript, were amplifiedwith the T3 and T7 primers. The PCRfragment was cloned in TOPO vector(Invitrogen, pCR®8/GW/TOPO® TAcloning KIT) and transferred to theinsulator test vector by in vitro recom-bination (Invitrogen, Gateway LR clo-nase II Enzyme mix). The GAB insu-lator element was excised as a 0.57-kbHindIII/EcoRV DNA fragment fromplasmid pGAB (A. Garcia-Diaz and L.Montoliu, unpublished), carrying a fu-sion of the G and AB boundary ele-ments previously described within thelocus control region (LCR) of themouse tyrosinase gene (Giraldo et al.,2003). The fragment was cloned bluntinto the KpnI site of the pminiTol2-Z48-CARGFP plasmid.
The ZED Vector
The 5�HS4 insulator was PCR ampli-fied from the CMV-gfp reporter plas-mid (Allen and Weeks, 2005) usingthe primers 5�-ggccagatgggccATA-ACTTCGTATAATGTATGCTATAC-GAAGTTATGAGTTGCGCGCCTGG-GAGC-3� and 5�-ggccagatgggccATA-ACTTCGTATAATGTATGCTATAC-GAAGTTATGAGTTGCGCGCCTGG-GAGC-3�. The forward primer con-tains an SfiI site (underlined andsmall letters) and LoxP recognitionsite (underlined and capital letters).The reverse primer contains a NdeIsite (underlined and small letters), aLoxP recognition site (underlined andcapital letters) and a XhoI site (under-lined and small letters). This frag-ment was cloned in the SfiI/XhoI sites5� of the pgata2-EGFP-polyA T2KHGplasmid to generate the LoxP-5�HS-
Fig. 7. In vivo excision of the internal control and enhancer detection cassettes of the ZebrafishEnhancer Detection (ZED) vector using Flipase and Cre recombinases. A: ZED F1 stable transgenicembryos showing strong red fluorescent protein (RFP) expression in the muscles were injected withFlipase mRNA. This injection caused reduction of RFP expression in some cells of the right side ofthe embryo (arrow) but not in the left side. RFP reduction upon Flipase mRNA injection wasobserved in 83% of the injected embryos (n � 200). B: Embryos from a F1 Z48-ZED stabletransgenic line. C: Injection of Cre mRNA in these stable transgenic embryos from this line causedmosaic loss of GFP expression. The transition from the homogeneous expression promoted bystable insertion to a mosaic pattern due to Cre-mediated recombination was observed in 85% ofthe injected embryos harboring the transgene (n � 200).
NEW VECTOR FOR ZEBRAFISH TRANSGENESIS 2415
LoxP-pgata2-EGFP-polyA T2KHG.The Gateway B cassette was clonedblunt into the NdeI site. This gener-ated the LoxP-5�HS-Gw-LoxP-pgata2-EGFP-polyA T2KHG plasmid. Wethen introduced a PacI and an SfiIsites into the pminiTol2/MCS plas-mid. To that end, we delete from theLoxP-5�HS-LoxP-pgata2-EGFP-polyAT2KHG plasmid a DNA fragment be-tween NcoI and SacI sites, which in-cludes the LoxP-5�HS-LoxP-pgata2-EGFP-polyA cassette, and we re-ligated the remaining vector. Fromthe resulting truncated vector, we ex-cised a 417-pb HindIII/BglII fragmentthat we cloned in the pminiTol2/MCSplasmid. This fragment contains aPacI and a SfiI sites. We then trans-ferred to these sites the LoxP-5�HS-LoxP-pgata2-EGFP-polyA cassette ex-cised with PacI and SfiI to generatethe pminiTol2- LoxP-5�HS-LoxP-pgata2-EGFP-polyA plasmid. To in-troduce the positive control to this vec-tor, we first excised the Cardiac Actin-DsRed-polyA cassette from theISCarDsR2/Pax6GFP3 vector (giftfrom H. Ogino and R. Grainger) andcloned into the XbaI/BamHI sites ofpBluescript KS� vector. In this plas-mid, we then introduced the GAB in-sulator from the pGAB vector (Giraldoet al., 2003) into the XhoI/KpnI sites.We then amplified the GAB insulator-Cardiac Actin-DsRed-polyA fragmentwith the following primers: 5�ttaattaaT-GAAGTTCCTATACTTTCTAGAGA-ATAGGAACTTCCTCACTATAGGGC-GAATTGG-3� and 5�ttaattaaGAAGT-TCCTATTCTCTAGAAAGTATAGG-AACTTCACGCAATTAACCCTCAC-TAAAGG-3�. These primers contain aPacI restriction site (underlined andsmall letter) and an FRT sequence (un-derlined and capital letters). This PCRfragment was then transferred to thePacI site of pminiTol2-5�HS4-Gw-pgata2-EGFP-polyA plasmid to gener-ate the ZED vector.
In Vitro mRNA Synthesis,Microinjection, andTransgenesis
cDNAs were linearized and mRNAswere transcribed as described (Bessaet al., 2008), and 100–200 pg of eachmRNA was injected in one-cell stageembryos. Transposase (Kawakami etal., 2004), Cre (Langenau et al., 2005),
and Flipase (Werdien et al., 2001) cD-NAs were linearized using NotI re-striction enzyme, and mRNA was syn-thesized using Sp6 RNA polymerase.
The Tol2 transposon/transposasemethod of transgenesis (Kawakami etal., 2004) was used with minor modi-fications. One nanoliter was injectedin the cell of one-cell stage embryoscontaining 50 ng/�l of transposasemRNA, 40 ng/�l of phenol/chloroformpurified DNA, and 0.05% phenol red.
ACKNOWLEDGMENTSWe thank E. Amaya, T. Becker, S.C.Ekker, G. Felsenfeld, R. Grainger, K.Kawakami, T. Look, R. Ryffel, andD. Weeks for reagents. J.B. is a post-doctoral fellow of the PortugueseFundacao para a Ciencia e a Tecno-logia. This work was supported bygrants from the Spanish Ministry ofEducation and Science (BFU2007-60042/BMC, Petri PET2007_0158,CSD2007-00008) and Junta de An-dalucıa (Proyecto de Excelencia CVI-3488) to J.L.G.-S. C.A.B.D. is insti-tutionally supported by CSIC,Universidad Pablo de Olavide andJunta de Andalucıa. Vectors can befreely obtained by contacting J.Bessa ([email protected]).
REFERENCES
Allen BG, Weeks DL. 2005. Transgenic Xe-nopus laevis embryos can be generatedusing phiC31 integrase. Nat Methods2:975–979.
Allende ML, Manzanares M, Tena JJ, Fei-joo CG, Gomez-Skarmeta JL. 2006.Cracking the genome’s second code: en-hancer detection by combined phyloge-netic footprinting and transgenic fishand frog embryos. Methods 39:212–219.
Balciunas D, Wangensteen KJ, Wilber A,Bell J, Geurts A, Sivasubbu S, Wang X,Hackett PB, Largaespada DA, McIvorRS, Ekker SC. 2006. Harnessing a highcargo-capacity transposon for genetic ap-plications in vertebrates. PLoS Genet2:e169.
Bessa J, Tavares MJ, Santos J, Kikuta H,Laplante M, Becker TS, Gomez-SkarmetaJL, Casares F. 2008. meis1 regulates cy-clin D1 and c-myc expression, and controlsthe proliferation of the multipotent cells inthe early developing zebrafish eye. Devel-opment 135:799–803.
Cheo DL, Titus SA, Byrd DR, Hartley JL,Temple GF, Brasch MA. 2004. Concertedassembly and cloning of multiple DNAsegments using in vitro site-specific re-combination: functional analysis ofmulti-segment expression clones. Ge-nome Res 14:2111–2120.
Chung JH, Whiteley M, Felsenfeld G.1993. A 5� element of the chicken beta-globin domain serves as an insulator inhuman erythroid cells and protectsagainst position effect in Drosophila.Cell 74:505–514.
de la Calle-Mustienes E, Feijoo CG, Man-zanares M, Tena JJ, Rodriguez-SeguelE, Letizia A, Allende ML, Gomez-Skarmeta JL. 2005. A functional surveyof the enhancer activity of conservednon-coding sequences from vertebrateIroquois cluster gene deserts. GenomeRes 15:1061–1072.
Giraldo P, Martinez A, Regales L, Lavado A,Garcia-Diaz A, Alonso A, Busturia A, Mon-toliu L. 2003. Functional dissection of themouse tyrosinase locus control regionidentifies a new putative boundary activ-ity. Nucleic Acids Res 31:6290–6305.
Harrison DA, Perrimon N. 1993. Simpleand efficient generation of marked clonesin Drosophila. Curr Biol 3:424–433.
Hartley JL, Temple GF, Brasch MA. 2000.DNA cloning using in vitro site-specific re-combination. Genome Res 10:1788–1795.
Kawakami K, Shima A, Kawakami N.2000. Identification of a functional trans-posase of the Tol2 element, an Ac-likeelement from the Japanese medaka fish,and its transposition in the zebrafishgerm lineage. Proc Natl Acad Sci U S A97:11403–11408.
Kawakami K, Takeda H, Kawakami N,Kobayashi M, Matsuda N, Mishina M.2004. A transposon-mediated gene trapapproach identifies developmentally reg-ulated genes in zebrafish. Dev Cell7:133–144.
Kroll KL, Amaya E. 1996. Transgenic Xe-nopus embryos from sperm nucleartransplantations reveal FGF signalingrequirements during gastrulation. De-velopment 122:3173–3183.
Langenau DM, Feng H, Berghmans S,Kanki JP, Kutok JL, Look AT. 2005. Cre/lox-regulated transgenic zebrafish modelwith conditional myc-induced T cellacute lymphoblastic leukemia. Proc NatlAcad Sci U S A 102:6068–6073.
Manzanares M, Wada H, Itasaki N, TrainorPA, Krumlauf R, Holland PW. 2000. Con-servation and elaboration of Hox gene reg-ulation during evolution of the vertebratehead. Nature 408:854–857.
Meng A, Tang H, Ong BA, Farrell MJ, LinS. 1997. Promoter analysis in living ze-brafish embryos identifies a cis-actingmotif required for neuronal expression ofGATA-2. Proc Natl Acad Sci U S A 94:6267–6272.
Mohun TJ, Garrett N, Gurdon JB. 1986.Upstream sequences required for tissue-specific activation of the cardiac actingene in Xenopus laevis embryos. EMBOJ 5:3185–3193.
Montoliu L, Umland T, Schutz G. 1996. Alocus control region at -12 kb of the ty-rosinase gene. EMBO J 15:6026–6034.
2416 BESSA ET AL.
Montoliu L, Roy R, Regales L, Garcia-DiazA. 2009. Design of vectors for transgeneexpression: the use of genomic compara-tive approaches. Comp Immunol Micro-biol Infect Dis 32:81–90.
Navratilova P, Fredman D, Hawkins TA,Turner K, Lenhard B, Becker TS. 2009.Systematic human/zebrafish compara-tive identification of cis-regulatory activ-ity around vertebrate developmentaltranscription factor genes. Dev Biol 327:526–540.
Nobrega MA, Ovcharenko I, Afzal V, RubinEM. 2003. Scanning human gene desertsfor long-range enhancers. Science 302:413.
Pennacchio LA, Ahituv N, Moses AM, Prab-hakar S, Nobrega MA, Shoukry M, Mino-vitsky S, Dubchak I, Holt A, Lewis KD,Plajzer-Frick I, Akiyama J, De Val S, AfzalV, Black BL, Couronne O, Eisen MB, ViselA, Rubin EM. 2006. In vivo enhancer anal-ysis of human conserved non-coding se-quences. Nature 444:499–502.
Perz-Edwards A, Hardison NL, Linney E.2001. Retinoic acid-mediated gene ex-pression in transgenic reporter ze-brafish. Dev Biol 229:89–101.
Recillas-Targa F, Pikaart MJ, Burgess-Beusse B, Bell AC, Litt MD, West AG,Gaszner M, Felsenfeld G. 2002. Position-effect protection and enhancer blockingby the chicken beta-globin insulator areseparable activities. Proc Natl Acad SciU S A 99:6883–6888.
Ryffel GU, Werdien D, Turan G, GerhardsA, Goosses S, Senkel S. 2003. Taggingmuscle cell lineages in development andtail regeneration using Cre recombinasein transgenic Xenopus. Nucleic AcidsRes 31:e44.
Visel A, Minovitsky S, Dubchak I, Pennac-chio LA. 2007. VISTA Enhancer Brows-er—a database of tissue-specific humanenhancers. Nucleic Acids Res 35:D88–D92.
Visel A, Prabhakar S, Akiyama JA,Shoukry M, Lewis KD, Holt A, Plajzer-Frick I, Afzal V, Rubin EM, PennacchioLA. 2008. Ultraconservation identifies asmall subset of extremely constraineddevelopmental enhancers. Nat Genet 40:158–160.
Werdien D, Peiler G, Ryffel GU. 2001. FLPand Cre recombinase function in Xeno-pus embryos. Nucleic Acids Res 29:E53–E53.
Woolfe A, Goode DK, Cooke J, Callaway H,Smith S, Snell P, McEwen GK, Elgar G.2007. CONDOR: a database resource ofdevelopmentally associated conservednon-coding elements. BMC Dev Biol7:100.
Zhu X, Ling J, Zhang L, Pi W, Wu M, TuanD. 2007. A facilitated tracking and tran-scription mechanism of long-range en-hancer function. Nucleic Acids Res 35:5532–5544.
