Xenopus laevis is an attractive model to analyze gene function in development and organogenesis of vertebrates, as embryonic development occurs outside the female and the developing larvae are highly transparent, allowing an easy observation and manipulation of the organism during development (Sive et al.,2000). Furthermore, embryos are available in large amounts and need a relatively low infrastructure for breeding. The attractiveness of Xenopus has been further increased by the method to generate transgenic frogs (Amaya and Kroll,1999), and several groups succeeded in establishing stable transgenic Xenopus lines transmitting transgenes to the next generations (Bronchain et al.,1999; Marsh-Armstrong et al.,1999; Offield et al.,2000; Hartley et al.,2002). Transgenic lines allow a most reproducible analysis of gene function, because hundreds of highly comparable F1 larvae can be investigated in one experiment, whereas in a transient assay, there is a large variation due to variable integration sites. However, investigating transgenes interfering with normal development, one can hardly obtain sexually mature founder animals; therefore, the use of an inducible system is essential. So far, different inducible systems have been applied in Xenopus laevis, including heat shock controlled transgenes (Wheeler et al.,2000; Buchholz et al.,2004), as well as RU486 or doxycycline inducible systems (Das and Brown,2004). Also, binary systems such as the GAL4/UAS system (Chae et al.,2002; Hartley et al.,2002) and the Cre/loxP system (Ryffel et al.,2003) have been adapted successfully to Xenopus for the conditional expression of transgenes. These binary systems are most versatile, because the independent activator and effector strains can be shared by different labs.
In mice, the DNA recombinases Cre and FLP have been used most successfully for conditional manipulation of gene expression, allowing the dissection of regulatory networks in an entire organism and to label cell lineages (reviewed in Branda and Dymecki,2004; Glaser et al.,2005). These recombinases can be used most efficiently in a binary system with a transgenic reporter strain and an activator strain expressing recombinase. Crossing the strains, the reporter gene is modified by specific recombination of the recognition sites. These modifications by the recombinase may lead to activation, silencing, or integration of a transgene. However, all these manipulations are irreversible, and the establishment of activator and effector strains is time consuming. The bacteriophage P1 recombinase Cre has been used predominantly in mammals, because it works optimally at 37°C, whereas FLP from S. cerevisiae has its optimum at 25°C. In contrast, in cold-blooded vertebrates both recombinases may be equivalent. In addition, Cre and FLP recombinases can be used independently, because they act on different specific recognition sites, loxP and FRT sites, respectively. We have shown previously that both Cre and FLP recombinases are functional in Xenopus laevis (Werdien et al.,2001) and used the Cre recombinase for labeling muscle cell lineages in stable transgenics (Ryffel et al.,2003). We have established two transgenic reporter strains but failed to generate an activator strain (Ryffel et al.,2003).
Here, we describe the establishment of a transgenic Xenopus laevis strain expressing the Cre recombinase in a muscle restricted manner as well as several transgenic reporter strains that can be used for cell lineage labeling. More importantly, these reporter strains will allow the identification and characterization of additional recombinase strains with distinct expression profiles. This first Cre-expressing Xenopus strain constitutes a major breakthrough of the Cre/loxP technology in Xenopus and opens up a new experimental field for this organism.
Establishment of a Cre-Expressing Transgenic Xenopus laevis Strain
To generate activator strains that express the Cre or FLP recombinase in a tissue restricted manner, we used either the cardiac actin promotor (CAR) for muscle-specific activity (Ryffel and Lingott,2000; Satoh et al.,2005) or the HNF1α promotor for expression in kidney, liver, and gut (Ryffel and Lingott,2000). Thus, four different constructs were made: CAR:Cre, HNF1α:Cre, CAR:FLP, and HNF1α:FLP.
Transgenic larvae were selected by polymerase chain reaction (PCR) genotyping (Table 1) and tested for transmission of recombinase activity by crossing the adult frogs with the C5 strain containing the LCMV:ECFP(loxP)(FRT)EYFP reporter construct (Fig. 1A). We could identify two male founder animals A6 and A7 described previously (Ryffel et al.,2003) that showed muscle-specific Cre activity. The remaining 14 strains were inactive, although 9 of these strains inherited the transgene as seen by PCR analysis (Table 1). Because only 2 of 16 (12.5%) founder animals transmitted recombinase activity and all A6 and A7 offspring died during metamorphosis, we suspect negative selection against recombinase expression.
Table 1. Founder Animals for Recombinase Expressiona
Founder tested (N)
Strains with recombinase in F1
By PCR and function
The number of founder animals and the animals transmitting the recombinase transgene are given. The cardiac actin promoter (CAR) was used to drive muscle-specific gene expression. The promoter of HNF1α was used to direct expression to kidney, liver, and gut. PCR, polymerase chain reaction.
A2, A4, A6, A7
D1, D3, D4, D5, D6
As the male founder animals A6 and A7 have been killed for the in vitro fertilization, we tried to recover the lines from a backup of frozen sperm. To establish a transgenic strain from the A7 male, we microinjected thawed A7 sperm into unfertilized Xenopus laevis eggs. From 11,500 injections we obtained 11 living tadpoles reaching metamorphosis. Three of them were positive for the Cre transgene as seen by PCR from tail tip DNA. One tadpole died during metamorphosis, but two frogs could be raised to sexual maturity. We also injected the sperm into eggs of the C5 reporter strain to directly score for Cre activity, which can be identified by activation of the yellow fluorescent reporter protein (Fig. 1A). From 4,900 injections we obtained 13 living tadpoles. Five of them had the reporter gene construct from the heterozygous C5 strain indicated by blue fluorescence and, thus, could be screened for Cre activity directly. In four of these animals, we detected muscle cell-specific activation by the yellow fluorescence (Fig. 2A–C). Yellow fluorescence was most prominent in the tail myotomes of swimming tadpoles (stage 42, Nieuwkoop and Faber,1975) and in myofibrils of the jaw and the abdominal wall of feeding larvae (stage 52). By PCR genotyping of these larvae, we could verify that the activation of the yellow fluorescence was indicative for the presence of the Cre transgene (Fig. 2D). Three such animals could be raised to sexual maturity. From the seven larvae that lacked the C5 reporter gene and, thus, could not be investigated for Cre activity, three contained the Cre transgene as seen by PCR (Fig. 2D) and two sexually mature frogs were obtained. Frozen sperm from the A6 male were also used for in vitro fertilization. From 8,100 injections into wild-type eggs, we obtained one Cre-positive froglet that died shortly after metamorphosis. In summary, we were able to recover from frozen sperm the transgenic A7 strain expressing Cre recombinase in muscle cells.
Characterization of the A7 Cre-Expressing Transgenic Xenopus laevis Strain
To test whether the Cre transgene is stably integrated and Cre activity is inherited in a mendelian manner, we crossed one male of the recovered A7 F1 generation by in vitro fertilization with one female of the C5 reporter strain. At stage 42, 83 of a total of 163 tadpoles derived from approximately 300 fertilized eggs showed blue fluorescence (Fig. 3A,C), indicating the presence of the C5 reporter gene. Of these 83 tadpoles, 41 showed induction of the yellow fluorescent reporter EYFP in muscle cells. Induction of yellow fluorescence was visible in tail myotomes and myofibrils of the jaws in Cre-positive tadpoles (Fig. 3B) compared with Cre-negative tadpoles (Fig. 3D). No activation of EYFP was found in the heart, even when the pictures were taken at long exposure times (Fig. 3E), but very fine myofibrils near the eye were marked by EYFP (Fig. 3F). We also observed weak and diffuse yellow fluorescence in the developing limbs of metamorphic Cre-positive tadpoles compared with control animals (data not shown). Thus, Cre activity is heritable in this heterozygous A7 strain and by crossing with a transgenic reporter strain activates an inducible gene in approximately 50% of the offspring as expected from mendelian inheritance.
Identification and Characterization of Cre-Inducible Transgenic Reporter Strains With DsRed2 or LacZ
With an active muscle-specific Cre strain at hand, we tested potential reporter strains for induction of the reporter gene by crossing with the A7 CAR:Cre strain. First, we tested the C1 strain containing the same LCMV:ECFP(loxP)(FRT)EYFP reporter construct (Fig. 1A) as present in the C5 strain. By crossing a C1 F1 female to the killed A7 male, we obtained 125 stage 42 larvae, with 65 showing blue fluorescence of the ECFP marker gene, but we did not observe any activation of yellow fluorescence in these larvae. Thus, the reporter gene construct LCMV:ECFP(loxP)(FRT)EYFP is not sensitive to Cre activity in the C1 strain, although it is in the C5 strain. In previous experiments, we had shown that A7 activates also the reporter in the Y2 strain that contains the LCMV:ECFP(loxP)EYFP construct (Fig. 1B) with loxP sites only (Ryffel et al.,2003). Altogether, we have obtained two functional reporter strains, C5 and Y2, that upon Cre activity express EYFP.
In a further test, a female of the A7 F1 generation was crossed by in vitro fertilization with an F1 male of the G/R reporter strain containing the construct CAG:EGFP(loxP)DsRed2 (Fig. 1C) that is under the control of the CMV enhancer/chicken β-actin promoter (CAG, Sakamaki et al.,2005). This heterozygous G/R strain shows ubiquitous expression of the green fluorescent marker gene EGFP, which is flanked by loxP sites allowing activation of the red fluorescent protein DsRed2 upon Cre activity (Fig. 1C). We obtained 28 larvae, because the 12-month-young A7 F1 female gave only approximately 50 eggs: 15 larvae showed green fluorescence and 10 of these animals revealed red fluorescence in tail muscle cells at the tadpole stage (Fig. 4A,B). In a few nonmuscle cells, we observed also some red fluorescence (Fig. 4A, lower panel). Surprisingly, there was also induction of green fluorescence in parallel to the induction of red fluorescence in the muscle cells. This was more evident at later stages of development (Fig. 4C,D). In conclusion, in this transgenic G/R reporter strain, the DsRed2 reporter gene could be activated in muscle cells by crossing with the A7 recombinase strain, but the effect was less specific than in the C5 strain.
We also tested the two independent reporter strains GPL6 and GPL8 that contain the construct CMV:GFP(loxP)LacZ (Fig. 1D). These heterozygous strains show ubiquitous expression of the GFP marker gene that is flanked by loxP sites, allowing activation of LacZ upon Cre activity. F1 females of these strains were crossed with an A7 male by in vitro fertilization. From the crossing A7 × GPL6 we obtained 86 free-swimming larvae (stage 42) with 40 showing green fluorescence of the GFP marker gene. Of these 40 larvae, 20 were stained for β-galactosidase activity, but no activation of the reporter was found, although PCR analysis showed the presence of Cre transgene in 5 of 12 animals (data not shown). From the GPL8 female, we could fertilize only approximately 50 eggs, leading to 13 larvae, with 8 showing green fluorescence of the GFP marker gene (Fig. 5A,D). When stained for β-galactosidase activity, five of these eight tadpoles showed activation of the reporter gene in some muscle cells (Fig. 5B,C). This partial activation was found in myotomes of the tail (Fig. 5G), myofibrils of the jaws (Fig. 5H), and myofibrils near the eye (Fig. 5I). In contrast to the C5 reporter strain, we also observed diffuse activation of the reporter by CAR:Cre in the heart (Fig. 5C) as compared with Cre-negative controls (Fig. 5F). Some unspecific LacZ staining of nonmuscle cells was also observed in these animals (Fig. 5C,G–I). PCR analysis (Fig. 5K) revealed that the Cre transgene was present in the five LacZ-positive larvae but absent in the three LacZ-negative animals (Fig. 5E,F). Thus, the LacZ reporter could be activated in this transgenic GPL8 strain, but the effect in muscle cells was only partial and less specific than in the C5 strain.
Taken together, 7 of 11 (64%) founder animals for recombinase-inducible reporter gene constructs transmitted the active transgene. Four of the resulting strains (67%) showed recombinase-mediated induction of the reporter gene upon crossing to the Cre-expressing strain A7 (Table 2).
Table 2. Reporter Strains for Cell Lineage Labelinga
Crossing with CAR:Cre (A7)
Marker gene expression (%)
The reporter constructs are given in Figure 1. Active strains are designated, when the transmission of the active transgene has been found and healthy frogs (F1) have been raised. Marker gene expression gives the percentage of larvae expressing ECFP, EGFP, or GFP, whereas recombination refers to Cre-induced activation of the EYFP, DsRed2, or LacZ gene in the reporter construct by crossing with the recombinase strain A7. The number (N) of total larvae of each crossing is given. Asterisks indicate published data using sperm of the A6 male (Ryffel et al.,2003).
Generating Cre- or FLP-expressing strains, the recombinase activity could not be measured in the larvae; the animals could only be scored for the presence of the recombinase gene by PCR and had to grow to sexual maturity. When crossing these founder animals to a recombinase-sensitive reporter strain, we observed that only 2 of 11 animals transmitted an active recombinase transgene. We suspect a negative selection against high recombinase expression levels, because Cre activity might target some pseudo loxP sites, a phenomenon well established in mammalian cells (Schmidt et al.,2000; Thyagarajan et al.,2000; Loonstra et al.,2001), and cleavage of DNA destroys genes essential for normal development. This assumption is supported by our observation that all the F1 animals derived from the in vitro fertilization of the C5 or Y2 reporter strains with sperm of the CAR:Cre males A6 and A7 died during metamorphosis (Ryffel et al.,2003). Animals from the same breeding lacking the CAR:Cre transgene, but retaining the reporter transgene, developed normally throughout metamorphosis, linking death to the presence of the active Cre transgene. In the case of the reestablished A7 Cre-expressing strain, Cre activity is obviously high enough to recombine artificial loxP sites but low enough not to be harmful by acting on pseudo loxP sites. We cannot definitely explain why the first offspring of A7 all died (Ryffel et al.,2003), and we now have succeeded in establishing the strain from the frozen sperm. However, we assume that a genetic mosaicsm in the germ cells of the A7 founder animal existed that allowed the recovery of a less active Cre transgene.
We were not able to identify any FLP activity in transgenic Xenopus, although we have proven that FLP recombinase is active in Xenopus in a transient assay (Werdien et al.,2001). However, the number of three investigated founder animals is too low to make any firm conclusions, as from our experiments with transgenic Cre animals, we deduce that at least 10 founder animals are needed to establish one transgenic recombinase strain. A way to improve the efficiency might be to circumvent the adverse effect of recombinases by making inducible constructs. Fusion proteins of the Cre or FLP recombinase with the ligand binding domain of a mutated estrogen or progesterone receptor, allowing a tamoxifen- or RU486-dependent activation, respectively, of the recombinase have been used successfully (Branda and Dymecki,2004). Another approach involves the heat-shock promoter (Wheeler et al.,2000) to drive expression of the recombinase and has already been applied in transgenic zebrafish (Thummel et al.,2005). As the heat-shock promoter has been used most successfully for conditional transgenesis in Xenopus without showing any basal activity (Buchholz et al.,2004; Fu et al.,2005), a heat-shock promoter-controlled Cre strain would be the most promising approach to ubiquitously activate silent reporter genes in transgenic reporter strains. Importantly, both conditional systems will allow a defined activation of the recombinase at any time of Xenopus development. To further improve the efficiency generating transgenic recombinase animals, we can now introduce the recombinase transgene into the background of the established reporter strains to directly evaluate recombinase activity in the larvae of the founder animal.
In mice, a multitude of transgenic strains have been generated and the efficiency of obtaining a functional recombinase strain is similar to the efficiency we observed in Xenopus, because for instance with a tamoxifen-inducible Cre recombinase, 3 of 54 founder animals turned out to be functional (Imai et al.,2001). In zebrafish, so far only one Cre strain (Thummel et al.,2005) and one Cre-sensitive reporter strain (Langenau et al.,2005) have been described. Considering the relatively low infrastructure needed to establish transgenic Xenopus and zebrafish compared with the mouse model, the efficiency in generating recombinase strains in these lower vertebrates is most favorable.
Characterizing the recovered A7 strain, we have seen that in 50% of the offspring Cre recombinase is present and can activate an inducible reporter. From this mendelian behavior, we deduce that the Cre transgene is integrated into a single locus in the A7 strain.
Comparing the independent reporter strains C1 and C5, which both have the same gene construct LCMV:ECFP(loxP)(FRT)EYFP, we found that Cre can activate the EYFP reporter in the C5 strain but not in C1. This finding was unexpected, because the transgene was active in both strains as seen by ubiquitous expression of the ECFP marker gene. However, as it is well established that transgenes in Xenopus are frequently integrated in multiple copies, Cre recombinase may just reduce the copy number without efficiently recombining the EYFP cassette into the active position. It is also possible that there is a reduced accessibility of the loxP sites in this strain due to an altered chromatin structure or that the loxP sites have been damaged during transgenesis.
The G/R reporter strain containing CAG:EGFP(loxP)DsRed2 showed perfect induction of red fluorescence in muscle cells when crossed to a CAR:Cre female of the A7 strain. The unspecific activation of red fluorescence in some nonmuscle cells is likely not due to a general unspecific CAR:Cre activity, because in all crossings to the C5 strain, we did not observe any unspecific activation of the EYFP reporter gene. We have no firm explanation for this unspecific activity in this G/R reporter strain and can only speculate that the CAG:EGFP(loxP)-DsRed2 construct is integrated into a chromosomal locus where spontaneous recombination can occur easily. The observation that the red fluorescence is unevenly distributed within one muscle cell after activation by Cre may reflect that DsRed tends to form oligomers (Baird et al.,2000). Using the C5 reporter strain, the induced yellow fluorescence by EYFP is distributed most evenly. However, differences in the expression pattern of the reporter genes between the G/R and C5 strains may also be explained by the presence of the CMV promoter in the C5 strain compared with the CAG (CMV enhancer/chicken β-actin) promoter in the G/R strain. From these experiments, we deduce that different combinations of activator and reporter strains may lead to a distinct expression profile and, thus, any transgene activated by a recombinase has to be checked for proper expression pattern.
Investigating two independent transgenic strains for the CMV:GFP(loxP)LacZ reporter expressing the GFP marker gene, only the GPL8 strain was Cre sensitive while GPL6 was not. Similar to the inactive C1 strain, there may be too many gene copies in GPL6 for sufficient recombination or the loxP sites may have been inactivated. Unexpectedly, the activation of the LacZ reporter in the Cre-sensitive GPL8 strain was only partial. We speculate that inactivation of the loxP sites, which we suspect in the C1 and GPL6 strains, can also be restricted to single cells. Like the G/R reporter strain, GPL8 also shows unspecific activation of the reporter gene. Furthermore, GPL8 revealed CAR:Cre-induced reporter gene activity in the heart, whereas there was no activation of yellow fluorescence from the EYFP reporter gene seen in the C5 strain. In a previous study analyzing CAR:GFP transgenic animals, we also did not see any fluorescence in the heart (Ryffel and Lingott,2000). This finding suggests that the LacZ staining is more sensitive.
We have established and characterized the first Cre-expressing transgenic Xenopus strain, which constitutes an important step into a novel experimental area in Xenopus. This A7 CAR:Cre strain allows diverse applications in Xenopus, because it can be crossed with independent reporter strains to activate a silent reporter gene. This binary recombinase system can also be used for the conditional expression of genes interfering with normal development. In this study, the A7 strain activated a reporter gene in three different transgenic strains, of which two (C5 and G/R) can be used for investigating living tadpoles and one (GPL8) for the more sensitive LacZ staining of fixed tadpoles. The availability of these three reporter strains will allow the efficient identification and characterization of additional recombinase strains with distinct expression profiles to establish defined conditional systems in Xenopus.
The reporter constructs LCMV:ECFP(loxP)(FRT)EYFP, LCMV:ECFP(loxP) EYFP (Ryffel et al.,2003) and CMV:GFP(loxP)LacZ (Werdien et al.,2001) have been described. To construct the reporter plasmid carrying the CAG:EGFP(loxP)DsRed2 transgene, we inserted the DNA fragment encoding an enhanced green fluorescent protein (EGFP, BD-Clontech) joined by the SV40 polyadenylation (polyA) signal sequence, which was sandwiched between the loxP sites, and followed by the DNA fragment encoding a DsRed2 fluorescent protein and the rabbit β-globin polyA signal sequence into the pCAGGS expression vector (Niwa et al.,1991). The CAR:Cre expression vector has been described (Ryffel et al.,2003). The expression vector CAR:FLP was constructed by exchanging the GFP gene in CAR:GFP (kindly provided by Kristen Kroll) as a HindIII NotI fragment with the corresponding FLP gene from pCSFLP. pCSFLP was constructed by replacing the BamHI XbaI GFP sequence in the pCSGFP2 vector with the open reading frame of FLP derived as a BamI XbaI PCR fragment from pOGFlpe6 (Buchholz et al.,1998) using the forward primer 5′-CGCGGATCCACAGC-CACCATGCCACAATTTG-3′ and the reverse primer 5′-TGCTCTAGAATGCGTCTATTTATGTAGGA-3′. The expression vectors HNF1α:Cre and HNF1α:FLP were made by replacing the CMV promoter in pCSCre2 and pCSFLP, respectively, with the Xenopus promoter extending from −5960 to −58 in front of the open reading frame of HNF1α (Ryffel and Lingott,2000). The full-sequence data of all the constructs can be obtained on request.
The original protocol (Amaya and Kroll,1999) was modified by using frozen sperm nuclei and omitting the egg extract to get more normal developing larvae (Ryffel et al.,2003). To generate the CAG:EGFP(loxP)DsRed2 transgenic frogs, the protocol was modified as detailed (Sakamaki et al.,2005). The reporter constructs LCMV:ECFP(loxP)(FRT)EYFP and LCMV:ECFP(loxP)EYFP were digested by Asp700 and NotI, and the NotI fragment lacking the bacterial DNA was purified for transgenesis. All the other plasmids were linearized by NotI digestion without removing the bacterial sequence in the transgene. Fluorescence microscopy was done with a Leica MZ/FLIII stereomicroscope with the appropriate filters as described (Werdien et al.,2001). Xenopus stages are as defined (Nieuwkoop and Faber,1975).
Cryopreservation of Sperm and In Vitro Fertilization With Frozen Sperm
Testes from the A7 male were isolated and minced 1:10 in Holfreter's solution with 10% dimethyl sulfoxide. Aliquots were kept overnight at −80°C and then stored in liquid nitrogen for 3 years. Sperm was thawed on ice and spermatozoa were counted and diluted with sperm dilution buffer (250 mM sucrose, 75 mM KCl, 0.5 mM spermidine trihydrochloride, 0.2 mM spermine tetrahydrochloride, pH 7.5). The dilution was adjusted for the following microinjection into unfertilized eggs to one spermatozoon per injected egg. Tadpoles were grown up by feeding with sera micron (www.sera.de), and after metamorphosis, the froglets were fed a rich diet including fresh tubifex, frozen red larvae, living worms (Dendrobena, www.superwurm.de) and chopped pig heart.
Tail-tip DNA was isolated using the DNeasy tissue extraction kit (Qiagen). PCR was performed using the specific primers CreP1 5′-AAACATGCTTCATCGTCGGTCCG-3′ and CreP2 5′-ATGCCAGATTACGTATATCCTGGC-3′ to detect the Cre transgene or FLPP1 5′-TTGGAAGACATTTGATGACCTCA-3′ and FLPP2 5′-GGCCACGGCAGAAGCAC-3′ for the FLP transgene. As controls, DNA from wild-type larvae was extracted and analyzed in parallel.
Tadpoles were fixed in 1% glutaraldehyde in 50 mM sodium cacodylate buffer pH 7.3 for 15 min and washed for 5 min with Fe/NaP solution (7.2 mM Na2HPO4, 2.8 mM NaH2PO4, 150 mM NaCl, 1 mM MgCl2, 3 mM K3[Fe(CN)6], 3 mM K4[Fe(CN)6] at pH 7.2). Subsequently, the embryos were incubated with fresh Na/P solution containing 0.1% Triton X-100 and 0.027% X-Gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside) for 24 hr and then refixed for 1 hr in MEMFA buffer (0.1 MOPS, 2 mM ethyleneglycoltetraacetic acid, 1 mM MgSO4, 3.7% formaldehyde, pH 7.4), washed, and stored in methanol.
We thank Kirsten Kroll and A. Francis Stewart for DNA constructs used in our work. K.S. and N.U. received a Grant in-Aid for Scientific Research to from the Japan Society for the Promotion of Science.