Mouse Reporter Strain for Noninvasive Bioluminescent Imaging of Cells that have Undergone Cre-Mediated Recombination

Abstract

Conditional alleles containing LoxP recombination sites, in conjunction with Cre recombinase delivered by a variety of means, allows for spatial and temporal control of gene expression in mouse models. Here we describe a mouse strain in which a luciferase (Luc) cDNA, preceded by a LoxP-stop-LoxP (L-S-L) cassette, was introduced into the ubiquitously expressed ROSA26 locus. Mouse embryo fibroblasts derived from this strain expressed luciferase after Cre-mediated recombination in vitro. ROSA26 L-S-L-Luc/+ mice expressed luciferase in a diffuse or liver-restricted pattern, as determined by noninvasive, bioluminescent imaging, when crossed to transgenic mice in which Cre was under the control of a zygotically expressed (EIIA-Cre), or a liver-restricted (albumin-Cre), promoter, respectively. Organ-specific luciferase expression was also seen after intraparenchymal administration of an adenovirus encoding Cre. The ROSA26 L-S-L-Luc/+ strain should be useful for characterizing Cre mouse strains and for following the fate of cells that have undergone Cre-mediated recombination in vivo.

Introduction

The use of conditional alleles containing LoxP sites is increasingly common in genetically engineered mouse models because they allow for spatial and temporal control of gene expression [1,2]. Such alleles are recombined by Cre recombinase, which is typically introduced using a viral vector or by mating to one of an expanding number of transgenic Cre mouse strains. Bioluminescent imaging based on luciferase expression is a powerful technique for studying diverse processes in living mice [3–6]. We therefore set out to develop a mouse strain in which Cre activity could be noninvasively monitored based on luciferase activity.

Materials and Methods

ROSA26 Targeting Construct

The Rosa-26 PA and the pBig-T vectors were generous gifts of Drs. Soriano and Constantini. The pBigT plasmid contains two LoxP sites that flank a neomycin resistance cassette and a strong transcriptional stop sequence [7]. A multiple cloning site (MCS) is located after the second LoxP. All these elements are located between unique Pad and Ascl restriction sites. The Rosa-26 PA plasmid contains ROSA26 genomic DNA into which has been inserted a linker containing Pad and Ascl sites [7]. This plasmid also contains a diphtheria toxin cassette 3′ of the ROSA26 genomic DNA to select against nonhomologous recombinants.

The Luciferase cDNA from pGL3-control (Promega, Madison, WI) was first excised as a HindIII-XbaI fragment and ligated into pcDNA3 (Invitrogen, Carlsbad, CA) cut with these two enzymes to make pcDNA3-Luciferase. To make pBigT-luciferase, pcDNA3-Luciferase was cut with HindIII (followed by Klenow) and ApaI. The liberated luciferase cDNA was then ligated into pBigT that had been cut with XhoI (followed by Klenow) and ApaI. The Pad-Asd fragment from pBigT-luciferase containing the floxed neomycin cassette and luciferase cDNA was ligated into pRosa-26PA cut with the same enzymes to make the L-S-L Luciferase ROSA26 targeting construct.

Homologous Recombination

The ROSA26 Luciferase targeting construct was linearized with KpnI and electroporated into TCI embryonic stem (ES) cells (derived from 129SVEV strain) using standard techniques. Two out of 100 G418-resistant ES clones underwent successful homologous recombination, as determined by Southern blot with a 5′ ROSA26 probe (external probe in Figure 1A), and were micro-injected into C57/BL6 blastocysts. High-percentage chimeric mice were obtained and bred to FVB mice.

Figure 1.

Conditional luciferase allele. (A) Structure of the targeted ROSA26 locus before and after Cre-mediated excision. LoxP sites, which are in the same orientation, are indicated by the arrowheads. PCR primers are indicated by the arrows. External and internal probes used for Southern blot analysis are indicated by black rectangles. (B) Southern blot analysis of EcoRV-digested genomic DNA (tail) from mice with the indicated genotypes using the external probe. Molecular weights of bands were determined based on migration of a 1-kb DNA ladder and agreed with their predicted sizes.

Polymerase Chain Reaction

Polymerase chain reaction (PCR) of genomic DNA was performed with AmpliTaq Gold DNA (Applied Biosystems, Foster City, CA) under standard conditions. Primers used were: Forward = 5′-CGGTATCGTAGAGTCGAGGCC-3′, L-LUCB = 5′-CAGGGCGTATCTCTTCATAGCC-3′, L-S-L-LUCB = 5′-GGCCAGAGGCCA-CTTGTGTAGC-3′, Cre Forward = 5′-CCTGGAAA-ATGCTTCTGTCCG-3′, Cre Rev = 5′-CAGGGTGTTATAAGCAATCCC-3′; GABRA Forward = 5′-CAATGGTAGGCTCACTCTGGGAGATGATA-3′; and GABRA Rev = 5′-AACACACACTGGCAGGACTGGC-TAGG-3′.

Mouse strains. E2A-Cre mice were a gift of Dr. R. DePinho. Alb-Cre mice and FVB mice were purchased from Jackson Laboratories (Bar Harbor, ME).

MEF luciferase assays. MEFs were generated from Day 13–5 embryos as previously described [8] and grown in six-well plates (1.5 × 10⁵ cells/well) in the presence of 10% fetal calf serum. After retroviral infection with pBabe-Cre or pBabe-Cre-ER cells were selected in puromycin (1 μg/mL). Activation of the Cre-ER was achieved by addition of Tamoxifen (0.1 μM) for 48 hr. Luciferase activity of cell extracts was measured with the Dual Luciferase Assay System (Promega) according to the manufacturer's instructions and a Berthold Technologies luminometer.

Cre Adenovirus

Mice were administered 150 μL of Ad5CMVcre (10⁹ pfu/mL in PBS) (Gene Transfer Vector Core, University of Iowa) by tail-vein injection and imaged 5 days later. For brain injection, the skull was exposed with a 1-cm scalp incision and the bregma and lambda sutures were identified as landmarks. A 0.5-mm hole was drilled through the skull, 2 mm caudal and 2 mm lateral (right) of the bregma suture. Ten microliters (10¹⁰ pfu/mL) of Ad5CMVcre was injected into the brain using a 27-gauge needle and the skin was approximated by staples. Images were obtained 6 days later. For kidney injection, the kidney was exposed with a 1-cm incision through the skin and peritoneum. One hundred microliters (10⁹ pfu/ml) of Ad5CMVcre was injected using a 27-gauge needle that had been passed from the lower pole of the kidney to the upper pole. PBS was injected in the same way into the contralateral kidney. Skin was approximated by Vetbond (3M Animal Care Products, St. Paul, MN) and images were obtained 12 days later.

In vivo imaging of luciferase activity. Mice were given a single intraperitoneal injection of a mixture of luciferin (50 mg/kg), Ketamine (150 mg/kg), and Xylazine (12 mg/kg) in sterile water. Fifteen minutes later, mice were placed in a light-tight chamber equipped with a charge-coupled device (CCD) IVIS imaging camera (Xenogen, Alameda, CA). Photons were collected for a period of 2 min and images were obtained using LIVING IMAGE software (Xenogen) and IGOR image analysis software (WaveMatrics, Lake Oswego, OR).

Results

Placement of a LoxP-stop-LoxP (L-S-L) cassette between a promoter and an open reading frame has been used successful to create alleles that are active only after Cre-mediated DNA excision [7,9–14]. Heterologous genes placed under the control of the ROSA26 locus are ubiquitously expressed [7,10–12,15]. Based on these two considerations, we placed an L-S-L cassette upstream of a firefly luciferase cDNA, which was then introduced into the ROSA26 locus by homologous recombination (Figure 1). Embryonic stem cells that had undergone successful recombination, as determined by Southern blot analysis and PCR, were micro-injected into C57/BL6 blastocysts to generate chimeric mice, which were then bred to FVB mice to confirm germline transmission.

In pilot experiments, mouse embryo fibroblasts were isolated from ROSA26 L-S-L-Luc/+ embryos or wild-type littermate controls and infected with retroviruses encoding Cre recombinase or a Cre-ER fusion protein, which is a latent form of Cre until activated by Tamoxifen (Figure 2). As expected, the ROSA26 L-S-L-Luc/+ MEFs, but not the control cells, became luciferase positive when infected to produce Cre recombinase. The induction of luciferase activity was dependent upon Cre recombinase enzymatic activity because luciferase induction required Tamoxifen in the cells producing the Cre-ER fusion protein.

Figure 2.

Cre-dependent luciferase activity in vitro. (A) PCR-based genotyping using genomic DNA obtained from MEFs or mice Presence of L-S-L-Luc allele is indicated by PCR product with “Forward” and “L-S-L-LUCB” primers (see Figure 1). Amplification of Gabra using internal primers served as a control. (B) MEFs with the indicated genotypes were uninfected (mock) or infected with a retrovirus encoding Cre or Cre-ER. After a 24-hr recovery period, the infected cells were grown in the presence of puromycin for 72 hr, lyzed, and raw luciferase values were determined. Where indicated Tamoxifen was added 48 hr before lysis. LU = light units.

Next, the ROSA26 L-S-L-Luc/+ mice were crossed to transgenic mice in which Cre is under the control of the EIIA promoter. Since the EIIA promoter is active during very early embryogenesis [16,17], many somatic cells in ROSA26 L-S-L-Luc/+; EIIA-Cre/+ mice should contain the recombined, active, L-Luc allele. After genotyping by PCR, selected Fl progeny (Figure 3A) were anesthetized, given luciferin by intraperitoneal injection, and imaged using a photon counting CCD camera. As expected, ROSA26 L-S-L-Luc/+; EIIA-Cre +– mice emitted light diffusely, in contrast to mice that harbored either the ROSA26 L-S-L-Luc/+ allele or the EIIA-Cre allele alone (Figure 3B and data not shown). In back-crosses to FVB mice, the recombined L-Luc allele was transmitted to F2 progeny as determined by PCR and by similarly diffuse light emission following luciferin administration, indicative of successful germline transmission (data not shown).

Figure 3.

In vivo imaging of luciferase activity after Cre-mediated excision of Lox-stop-Lox cassette. (A) PCR-based genotyping of mouse genomic DNA The presence of L-S-L-Luc allele detected as in Figure 2 and the presence of L-Luc allele are indicated by PCR product with “Forward” and “L-LUCB” primers (see Figure 1). Note that L-S-L-Luc is not recombined in tail DNA obtained from Rosa26 L-S-L-Luc/+; Alb-Cre/+ mice. Cre and Gabra genes were amplified using internal primers specific for these two genes. (B) Photon emissions captured over 2 min from mice with the indicated genotypes after intraperitoneal administration of luciferin. Color bar indicates photons/cm²/sec/steradian with min and max threshold values.

The light signal observed in the ROSA26 L-S-L-Luc/+; EIIA-Cre mice is consistent with ubiquitous luciferase expression, but formal conclusions cannot be drawn because currently available detectors generate 2-D, rather than tomographic, images and because of special resolution issues. These problems are expected to diminish as newer camera systems are developed. Unfortunately, none of the commercially available anti-luciferase antibodies performs well in immunohistochemical studies. For example, we have been unable to detect a specific luciferase immunohistochemical signal in mouse xenograft tumors formed by HeLa cervical carcinoma cells that have been engineered to produce high levels of luciferase (data not shown).

To ask whether the ROSA26 L-S-L-Luc/+ strain could detect organ-confined Cre activity, the experiments were repeated with a transgenic mouse in which Cre is under the control of the albumin promoter, which is active only in the liver [18]. In contrast to ROSA26 L-S-L-Luc/+; EIIA-Cre+/– mice, ROSA26 L-S-L-Luc/+; Alb-Cre+/– mice emitted light exclusively from the liver (Figure 3B). A similar signal was obtained 5 days after tail-vein injection of ROSA26 L-S-L-Luc/+ mice with an adenovirus encoding Cre, which is consistent with the profound liver tropism of adenoviral vectors when administered systemically [19,20] (Figure 4A). Direct injection of the Cre adenovirus into the brains or kidneys of ROSA26 L-S-L-Luc/+ mice also led to organ-confined light signals (Figure 4B and C, respectively). No signal was obtained in control mice or in animals receiving PBS instead of the virus (Figure 4 and data not shown).

Figure 4.

In vivo imaging of mice treated with Cre adenovirus. Mice with the indicated genotypes were administered an adenovirus encoding Cre recombinase by tail-vein injection (A) or by stereotactic brain injection (B). (C) The left kidney of a Rosa26 L-S-L-Luc/+ mouse was exposed surgically and injected with Cre adenovirus. Bioluminescent imaging was performed 5 days (A), 6 days (B), or 12 days (C) later. Color bar indicates photons/cm²/sec/steradian with min and max threshold values. Note signal over spine in one mouse in (B), possibly due to AdCre entry into the spinal fluid. In (C), a slightly higher min threshold was set due to low-level background signal in the area of the PBS-injected kidney. The origin of this signal, which was not observed in wild-type mice studied in parallel, is currently being investigated.

Discussion

These results suggest that the ROSA26 L-S-L-Luc/+ mouse will be useful for characterizing new transgenic Cre strains, for testing new methods (viral or nonviral) of introducing Cre into animal cells, and for monitoring the fate of cells, over time, that have undergone Cre-mediated recombination events in vivo. In this regard, Vooijs and coworkers [21] recently showed that pituitary tumors arising in Rb-1+/– animals could be noninvasively monitored based on luciferase expression driven by a pituitary-specific promoter. ROSA26 L-S-L-Luc/+ mice might be generally useful in settings in which tumor development is intimately linked to a Cre-mediated recombination event in a genetically engineered mouse [9]. For example, the ROSA26 L-S-L-Luc/+ allele could be introduced, by mating, into mice that harbor a conditional oncogene, which is activated after Cre recombination, or into mice that are homozygous or hemizygous for a conditional tumor suppressor gene, which is inactivated after Cre recombination. Cre recombinase could then be delivered somatically (such as by using a viral vector) or after an additional cross to a mouse Cre strain. In many cases, it should be possible to image tumor development in situ (e.g., by using limiting amounts of viral vector or a promoter whose activity is normally limited to a small number of cells) or metastasis (e.g., by using promoters or somatic delivery systems that lead to spatially restricted Cre activity). Likewise, the ROSA26 L-S-L-Luc/+ mouse could be used to monitor the mobilization and deployment of specific migratory cell populations (e.g., immune effector cells or bone marrow-derived stem cells) that have been marked by virtue of Cre activation using, for example, a lineage-specific promoter.

Mice that produce β-galactosidase or green fluorescent protein (GFP) following Cre-mediated excision have been reported [7,10–12,22]. Colorimetric and fluorescent imaging of these two reporters is superior to bioluminescent imaging of luciferase with respect to spatial resolution [4]. Moreover, GFP can be imaged in living mice [23–25]. On the other hand, imaging GFP (or similar fluorescent proteins) requires an excitation light and gives rise to background due to autofluorescence, in contrast to bioluminescent imaging [4]. Thus, it might be impossible to distinguish between background fluorescence and fluorescence resulting from low-level, stochastic, Cre-mediated recombination in a given tissue. Moreover, luciferase is clearly superior to GFP in vivo with respect to sensitivity, especially for deep-seated organs [4]. Indeed, as few as 1000 luciferase-positive tumor cells can be detected in vivo with currently available cameras and signal can still be detected if luciferase-positive cells are surrounded by bone (such as skull or long bones) [21,26] (Andrew Kung, personal communication). The ROSA26L-S-L-Luc/+ mouse is therefore complementary to Cre reporter strains reported so far and will likely be superior in certain settings.