|Home | About | Journals | Submit | Contact Us | Français|
The strategy of modulating gene activities in vivo via CRE/loxP recombination would greatly profit from subjecting the recombination event to an independent and stringent temporal control. Here, we describe a transgenic mouse line, LC-1, where the expression of the cre and luciferase gene is tightly controlled by the Tet system. Using the R26R mouse line as indicator for CRE activity, and mouse lines expressing tetracycline controlled transactivators (tTA/rtTA) in various tissues, we show that; (i) in the non-induced state CRE recombinase is tightly controlled throughout the development and adulthood of an animal; (ii) upon induction, efficient recombination occurs in the adult animal in all tissues where tTA/rtTA is present, including hepatocytes, kidney cells, neurons and T lymphocytes; and (iii) no position effect appears to be caused by the LC-1 locus. Moreover, using the novel rTALAP-1 mouse line, we show that in hepatocytes, complete deletion of the loxP-flanked insert in R26R animals is achieved less than 48 h after induction. Thus, the LC-1 mouse appears suitable for exploiting two rapidly increasing collections of mouse lines of which one provides tTA/rtTA in specific cell types/tissues, and the other a variety of loxP-flanked genes.
Site-specific recombinases, such as CRE and FLP derived from microorganisms, catalyze recombination between their cognate recognition sites, loxP and frt, respectively, with high specificity. Both systems function in mammalian cells and, in particular, the CRE/loxP system has developed into a powerful tool for the in vivo manipulation of genomes in transgenic mice (reviewed in 1–3). While the recombinases permit to invert or delete DNA fragments in cis or switch segments of DNA in trans and thus allow a variety of modifications, including chromosome engineering, the most common application is the conditional deletion of a DNA fragment flanked, e.g., by two loxP sites in order to silence or activate a gene. In the most generally applied approach, conditionality of recombination is based on the specificity of the promoter driving the expression of the cre gene and, accordingly, recombination will occur dependent on the spatial and temporal activity of the respective promoter. Not withstanding its great potential, this approach has a number of limitations. Thus, the activity spectrum of a promoter depends on a rigid developmental/differentiation program of an organism and, accordingly, as recombination induces irreversible alterations, the final pattern of CRE action will reflect the ‘history’ of a promoter’s activity throughout development. While such a cummulative recombination pattern may be of interest in some studies, it diminishes the precision of the approach in others. Moreover, as CRE will act whenever a respective promoter becomes active in a particular tissue/cell type, it will not be possible to induce recombination at a later time in the lifespan of an animal, thus preventing a thorough study of the unperturbed state in the same individual. Finally, there is increasing evidence that by recognizing pseudo loxP sites within the mouse genome, CRE can cause undesired recombination events (4,5). Even if such recombinations occur at a low rate, the outcome will depend on the intracellular concentration of CRE and the duration of its expression.
Obviously, controlling CRE activity from outside at will would enhance the scope of the technology significantly as numerous attempts demonstrate. Two approaches have primarily been followed (reviewed in 2): controlling either the transcription of the cre gene or the activity of CRE itself. Both principles have been applied with some success, but problems concerning the general applicability have remained. They were connected with the inducers of the control systems (interferon, synthetic steroids such as tamoxifen and Ru486), the diminished enzymatic activity of the CRE fusion proteins and the tightness of the control systems over long periods of time. The latter parameter is particularly important in the study of long-lived cells as recombination events, caused by leakiness of the control system, accumulate over time.
When properly set up, the Tet regulatory system allows tight control of transcription (6,7) and indeed control of CRE via doxycycline (Dox) has been reported (8–10). The approach of Utomo et al. (9) seems particularly attractive in this context. Here, a single vector encodes two transcription units, of which one contains the gene of a tetracycline controlled transactivator (tTA or rtTA) driven by a cell type specific promoter, whereas the other one encodes CRE controlled by the tTA/rtTA responsive promoter Ptet-1 (7). Mouse lines derived with this vector were shown to control CRE activity satisfactorily in the expected tissues. Nevertheless, like other regulatory systems which function via promoter activation, the tightness and range of regulation of the Tet system depends strictly on the genomic locus where the Ptet controlled transcription unit is integrated. For example, enhancers which may act over considerable distances can activate the minimal promoter within Ptet when located near the integration site. One would, therefore, predict that the Utomo vector will function in some, but not in other settings.
Here, we describe an alternative strategy which is based on a mouse line, designated LC-1, that contains a Ptet-cre transcription unit integrated in an appropriate chromosomal locus and which exhibits remarkable properties: when used to generate triple transgenic animals,which also contain the lacz gene silenced by a loxP flanked Stop sequence (ROSA R26R line) and which produce rtTA or tTA in various tissues, no CRE mediated recombination is detected via X-gal staining in the uninduced state during development and during at least 14 months of adulthood. In contrast, upon induction, efficient recombination occurs in all tissues examined. Moreover, there is apparently no position effect variegation (PEV) connected with the LC-1 locus.
Considering the rapidly increasing number of mouse lines producing tTA or rtTA in a wide variety of tissues/cell types (reviewed in 11) and the large ‘zoo’ of mouse lines with loxP-flanked genes (12), the LC-1 line will allow researchers to exploit both pools to generate animals where CRE activity is subjected to Tet control. Even though the breeding of triple transgenic mice will be somewhat demanding, it permits, however, to obtain the desired animals without generating and characterizing a new transgenic line.
The bidirectional transcripton unit (Fig. (Fig.1)1) encoding the cre and the luciferase (luc) gene under the control of the promoter Ptetbi-1 (7,13) was derived as described previously (14). The sequence of the transcription unit is available upon request.
The PLAP rtTA2s-S2 transcription unit (Fig. (Fig.1)1) was obtained by first replacing the SV40 poly(A) site in pUHrT61-1 (15) by the β-globin intron/poly(A) sequence. Subsequently, the PCMV sequence was exchanged for the sequence encoding the promoter of the liver enriched activator protein (16) derived from plasmid pUHG15-30 (17). Sequence information for the resulting plasmid pUHrT61-30 is available upon request.
LC transgenic mice were obtained by pronuclear injection of a purified 6.8 kb NotI fragment containing the bidirectional transcription unit into fertilized F2 (C57Bl/6X BALB/c) eggs. Similary, rTALAP mice were generated by transferring the 5.1 kb XmnI–AsnI fragment of pUHrT61-30 into fertilized F2 (C57Bl/6X DBA) eggs, following standard techniques (18).
Genotypes were identified by Southern blot analyses and PCR using DNA from tail biopsies. Probes for Southern blot analyses were, for LC animals, the 1.25 kb SalI fragment from PTZ19Rcre NLS-1 (14), and for rTALAP mice, the 0.75 kb XbaI–BamHI fragment from pUHrT 61-1.
For PCR analyses, the following primers were designed. luc: sense luc1, 5′-TTA CAG ATG CAC ATA TCG AGG-3′; antisense luc2, 5′-TAA CCC AGT AGA TCC AGA GG-3′. cre: sense Cre3, 5′-TCG CTG CAT TAC CGG TCG ATG C-3′; antisense Cre4, 5′-CCA TGA GTG AAC GAA CCT GGT CG-3′. TetR (synthetic): sense sTA, 5′-CCA TGT CTA GAC TGG ACA AGA-3′; antisense sTA, 5′-CTC CAG GCC ACA TAT GAT TAG-3′. tetR (native): sense tet, 5′-AAT GAG GTC GGA ATC GAA GG-3′; antisense tet, 5′-TAG CTT GTC GTA ATA ATG GCG G-3′. internal lacZ: sense lacZ, 5′-TTA CGA TGC GCC CAT CTA CAC-3′; antisense lacZ, 5′-TTA CCC GTA GGT AGT CAC GCA-3′.
PCR was routinely carried out for 28 cycles and resulting material was analyzed by electrophoresis in agarose (1%).
Double or triple transgenic animals were supplied for two weeks with doxycycline hydrochloride via drinking water (5% sucrose, 0.2–2 mg Dox/ml) which was exchanged twice a week. For rapid and short-lived induction, 2 mg of Dox in 0.5 ml of 0.9% aqueous NaCl were injected i.p. Depending on the experiment, injections were repeated twice at intervals of 24 h.
Tissue samples of transgenic animals were processed and luciferase activity determined as described previously (17).
In situ β-galactosidase staining. Sacrificed mice were perfused first with PBS followed by 4% paraformaldehyde in PBS. Tissue samples were incubated overnight in PBS containing 30% sucrose and subsequently frozen on dry ice. To detect lacz expression 20 µm cryosections mounted on glass slides (SuperFrost Plus, Menzel-Gläser) were stained with 4-bromo-3-chloro-2-indolyl-β-galactosidase overnight at 37°C (18). A 0.1% Nuclear fast red-solution (Certistain, MERCK) in 5% aqueous aluminium sulfate solution was used for counterstaining.
Immunohistochemistry. Brain immunohistochemistry of CRE recombinase was carried out as described by Kellendonk et al. (19). Liver slices were processed as 20 µm cryosections mounted on glass slides.
For detecting recombination in the R26R allele, the following primers were used: R26F2 (5′-AAAGTCGCTCTGAGT TGTTAT-3′) as described by Soriano (20) and a modification of lacZ3 (21) designed in our lab (lacZ4, 5′-GTGCGG GCCTCTTCGCTAT-3′). With these primers the intact R26R allele gives rise to an ~3.5 kb fragment which is reduced upon CRE mediated recombination to 600 bp. DNA from fixed tissue samples used for in situ β-gal staining was purified by the DNeasy Tissue Kit from Qiagen. PCR conditions were as follows: 5 min at 94°C, 38 cycles of 40 s at 94°C, 40 s at 58°C and 90 s at 68°C, followed by 7 min at 68°C. Taq polymerase was purchased from Roche Diagnostics GmbH (Expand High Fidelity PCR System). According to Rijnkels and Rosen (21), such DNA is only suitable for identifying PCR products of up to 800 bp. The lacz gene was identified analogously using internal primers (see above).
To facilitate the identification of founder lines where the cre gene is regulated tightly and over a suitable range, the bidirectional promoter Ptetbi-1 (13) was used to coregulate the cre and the luc gene (Fig. (Fig.1).1). Upon transfer of the respective DNA construct into mice, seven founder lines were obtained which transmitted the transgene stably. Animals of these lines were crossed with rTACMV-3 mice which produce rtTA in all organs where the activity of the human cytomegalovirus promoter IE (PhCMV) is supported (Fig. (Fig.2)2) including muscle, pancreas, thymus, kidney and spleen (17). Monitoring the luciferase activity in the various organs in presence and absence of Dox revealed the LC-1 line (TCL-1, in 14) as a most promising one: in the OFF state, i.e. in absence of Dox, luciferase activity was barely or not at all detected in any organ. In contrast, when Dox was supplied (2 mg/ml in drinking water), it was induced to levels reflecting the activity of PhCMV in the respective organs (Fig. (Fig.2).2). Moreover, when LC-1 mice were crossed with animals of the TALAP-2 line (17), a hepatocyte specific tTA mouse line, luciferase was induced to high levels exclusively in the liver (Fig. (Fig.2).2). In all other lines, luciferase was also barely detectable in the uninduced state. However, it was inducible to only low levels when compared to LC-1. One of these lines expressed the cre transgene transiently during embryonal development, mediating the complete deletion of loxP-flanked substrates even in the absence of the rtTA transgene (22).
The promoter of the liver enriched activator protein, PLAP (16), can give rise to hepatocyte specific expression when ectopically inserted in transgenic mice, as previously shown with the mouse lines TALAP-1 and TALAP-2 (17) which produce tTA in a highly hepatocyte specific manner. For various reasons, we were interested in generating an analogous mouse line that expresses rtTA instead of tTA in hepatocytes. Thus, the fusion between PLAP and the gene encoding the novel transactivator rtTA2S-S2 (15) was used for transgenesis resulting in seven DNA positive animals of which five transmitted the transgene. When crossed with individuals of the LC-1 line, noninvasive luciferase monitoring (23) in presence and absence of Dox (2 mg/ml in drinking water) revealed four lines where the luciferase gene was not activatable. By contrast, in one line designated rTALAP-1 the antibiotic induced high activity. When luciferase was measured in extracts of isolated organs of double transgenic rTALAP-1/LC-1 mice, high values were, as expected, found in liver, but also in kidney. As seen in Figure Figure2,2, in absence of Dox, luciferase values are at the level of instrumental background (~1 r.l.u./µg protein). Upon induction by Dox, all organ extracts examined show a low luciferase activity which is, however, 2 to 3 orders of magnitude below the values observed in liver and kidney. Our data show that the luc gene is shut off tightly in the non-induced state in all tissues, while, upon induction, it is activated over 3 to 5 orders of magnitude in liver and kidney, respectively (Fig. (Fig.2).2). Examination of rTALAP-1/nZL-2 double transgenic animals where the lacz gene is under control of Ptetbi-1 revealed that every hepatocyte is stained demonstrating that in rTALAP-1 mice PLAP is not subjected to position effect variegation in the liver (data not shown). Moreover, induction of luciferase expression in rTALAP-1/LC-1 animals is rapid as luciferase activity can be monitored within 1 h after i.p. injection of 2 mg of Dox (23).
Even though it was anticipated that cre and luc are co-regulated by Dox (23), the usefulness of the LC-1 mouse line obviously depends on the tightness of CRE expression in the OFF state and the intracellular level of CRE upon induction which would ensure efficient recombination at the target site. By using the R26R (20) indicator mouse line where constitutive β-galactosidase synthesis via the endogenous ROSA26 promoter is prevented at the transcriptional level by a loxP-flanked stop sequence, we have examined the functional CRE background of the LC-1 mouse line. Thus, LC-1/R26R double transgenic animals were histologically analyzed for β-galactosidase activity at day E14 of embryonic development (data not shown) and through adulthood. As exemplified in Figure Figure3,3, no signs of recombination were detected in several up to 14-month-old animals in any tissue examined, confirming the tight control of the bidirectional transcription unit in the LC-1 locus as already indicated by the barely detectable luciferase activities in LC-1 animals.
The same picture emerges when triple transgenic animals, where tTA or rtTA is expressed tissue-specifically, are examined in the OFF state. As exemplified for liver, kidney and lung, there is no sign of recombination in rTALAP-1/LC-1/R26R animals in the OFF state as shown by in situ β-galactosidase monitoring (Fig. (Fig.4)4) and by PCR of respective DNA (Fig. (Fig.5).5). In contrast, two i.p. injections (2 mg of Dox each) spaced 24 h apart convert the inactive lacz into an active gene within less than 48 h via recombination (Figs (Figs44 and and5).5). This experiment revealed that in the rTALAP-1 mouse line the transactivator is produced not only in hepatocytes but also in some additional rather well defined cell populations of the kidney such as the cortical proximal tubules and of TAL profiles of the inner stripe. Moreover, there is also sporadic expression in a minor fraction of bronchial epithelial cells which, however, do not include clara or mucous epithelial cells (Fig. (Fig.4).4). This particular pattern of CRE activity is obviously due to position effects governing the activity of the LAP promoter in the rTALAP-1 mouse line, as comparisons with TALAP-1 and TALAP-2 mouse lines (17) confirm (data not shown). The high levels of luciferase activity (r.l.u./µg) detected in the kidney, where only a fraction of cells produce the transactivator, suggest that a lower expression of the transactivator in hepatocytes of rTALAP-1 animals may limit luciferase expression. This phenomenon has been observed previously (17).
When LC-1 mice were crossed with individuals of the TACamK-1 mouse line (line B in 24) which constitutively produce tTA in neurons of the forebrain, Dox dependent co-regulation of CRE and luciferase expression was demonstrated by assaying luciferase activity in tissue extracts and by immunostaining of CRE in situ (Fig. (Fig.66).
Detailed analysis of liver sections as depicted in Figure Figure44 shows that in rTALAP-1/LC-1/R26R animals, recombination has occurred in every hepatocyte. Obviously, none of the three mouse lines gives rise to position effect variegation in hepatocytes implying that every hepatocyte produces rtTA which, when activated by Dox, can access the respective LC-1 locus to express CRE which in turn is capable of catalyzing a recombination event at the ROSA26 locus in every cell. These results show that at least in hepatocytes the LC-1 mouse line is not subjected to PEV and that CRE levels are induced that lead to a rapid and complete deletion of an appropriately loxP-flanked DNA segment. This reasoning is supported by data obtained with two sets of triple transgenic animals (TALAP-1/LC-1/polβflox and µEG2/LC-1/polβflox) where complete deletion of the loxP-flanked beta polymerase was demonstrated in hepatocytes and T lymphocytes by Southern blot analysis (14).
Here, we describe two transgenic mouse lines of which the first one, the LC-1, allows us to superimpose an effective and reliable temporal control onto CRE mediated recombination via the Tet regulatory system, whereas the second one, the rTALAP-1, confines Tet regulation via rtTA rather exclusively to hepatocytes, some subsets of kidney cells and a minute population of cells in the bronchial epithel.
The LC-1 mouse line will expand the applications of conditional mutagenesis. A prerequisite for regulating a gene of interest such as the cre gene tightly and over a sufficient range is the integration of the Ptet-cre transcription unit into an appropriate chromosomal site, where no outside activation of Ptet occurs while high levels of expression can be achieved upon induction. Indeed, the locus identified in our LC-1 mouse lines appears to fulfill all criteria required.
First, with the exception discussed below, we have not detected CRE recombination in any tissue and at any time during development and adulthood of up to 14-month-old LC-1/R26R double transgenic animals demonstrating that the bidirectional cre–luc transcription unit embedded in the LC-1 locus generates no or only subfunctional and, thus, negligible amounts of CRE which agrees with the barely measurable luciferase activities monitored in the various organs of the animals. Corresponding results were obtained with rTALAP-1/LC-1/R26R triple transgenic mice. Thus, in the OFF state the LC-1 locus appears to be ubiquitously silent.
Second, expression of CRE and luciferase is inducible in all cell types and tissues examined. For example, by crossing LC-1 mice with individuals of the rTACMV-3 line, luciferase activity is detected in all tissues/organs where PhCMV is known to be active (Fig. (Fig.2;2; 17). Moreover, luciferase is induced in hepatocytes, neurons, kidney cells and T lymphocytes as analyses of respective LC-1 double transgenic animals show. In the LC-1 locus, the cre and the luc gene are transcribed at about the same rate (23). It, therefore, can be safely assumed that CRE is produced in all cells where luciferase can be monitored. This is supported by immunostaining of CRE which was positive in all tissues analyzed, namely in hepatocytes, neurons (Fig. (Fig.6),6), kidney cells (data not shown) and T lymphocytes (14). Moreover, in all these cell types, efficient CRE mediated recombination was detected. Therefore, given the present set of data, it appears not unlikely that LC-1 animals are capable of producing CRE and, thus, of mediating recombination in many, possibly all cell types provided that tTA or rtTA is expressed in the respective cell.
Third, our data obtained with rTALAP-1/LC-1/R26R mice show that upon induction CRE recombination occurs rapidly in all hepatocytes. Accordingly, the LC-1 locus is accessible for tTA/rtTA in every cell. Similar results were obtained in pyramidal cells of the hippocampus (H.Krestel, personal communication). Apparently, the LC-1 locus does not cause PEV in these very different cell types, and it will be interesting to learn from future studies whether absence of PEV will be another general feature of this locus.
The second mouse line characterized here (rTALAP-1) demonstrates that CRE and luciferase controlled by Ptetbi-1 can be regulated over several orders of magnitude via Dox in highly specific cell populations while not causing any measurable activity of luciferase or CRE in the uninduced state. The finding that in the uninduced state even 6–8-month-old triple transgenic animals (rTALAP-1/LC-1/R26R) show no sign of recombination in cells producing rtTA2S-S2 demonstrates the high discrimination potential of this transactivator.
One observation should, however, not go unmentioned here. We have occasionally detected a very small number of X-gal positive cells in kidney and heart sections of some, but not all individuals of a litter of LC-1/R26R double transgenic animals. While this could well be an artefact indicating endogenous activities for X-gal production, it could also reflect individual variations in background activity of the LC-1 locus. As described previously (17), up to 5-fold differences in expression levels can be measured among littermates of transgenic animals. We attribute these differences to epigenetic variations among individuals, a point to be kept in mind when changing the genetic background via breeding.
In conclusion, the properties of the genomic site, where the Ptetbi-1 controlled cre gene is integrated and co-regulated with the luc gene, make the LC-1 mouse line an interesting and probably widely applicable experimental tool for controlling CRE recombination in vivo. Moreover, the fact that the expression of CRE cannot only be tightly suppressed, thus preventing undesired background, but also induced to high levels may be of particular value. Various reports suggest that recombination efficiency also depends on the topography of the loxP-flanked DNA within the chromatin context. Short exposure of such sites to high levels of CRE may be a suitable approach for achieving efficient and complete recombination while avoiding side effects of high constitutive levels of recombinase. Finally, considering the two growing ‘zoos’ of transgenic mouse lines containing either tTA/rtTA genes under the control of various tissue specific promoters or a wide variety of loxP-flanked genes, the LC-1 mouse line provides an attractive link for exploiting the two worlds of transgenic mice.
We like to thank F. Zimmermann, S. Dlugosz and the team in the animal facilities of the ZMBH for expert assistance, P. Soriano for generously supplying us with the R26R mouse line, G. Schütz for anti-CRE antibodies and H.-F. Gröne for help in histological analyses. We are grateful to U. Baron, R. Kühn and R. Witzgall for advice and discussions. The patient help of S. Reinig in preparation of the manuscript is gratefully acknowledged. This work was supported by grants of the Deutsche Forschungsgemeinschaft (SFB 243), the Human Frontier Science Program, the Volkswagen-Stiftung, the Bioregio Program Heidelberg, the European Community and the State of Nordrhein-Westfalen. F.S. was supported by a fellowship of the Boehringer Ingelheim Fonds.