Materials and Methods

Results

We chose drug metabolism genes and the multiple drug resistance gene MDR1 (also termed ABCB1) for this study because they are activated by well-described mechanisms involving nuclear hormone receptors, which are the sensors of pharmacologically important xenobiotic ligands. Through these receptors, physiological ligands (such as hormones and bile acids) regulate the expression of target genes involved in bile acid and sterol homeostasis (Chawla et al., 2001). We first determined the minimum dose of plasmid required to allow measurement of expression of the reporter gene in untreated mouse liver. Mice were injected with various concentrations of plasmids in which the CYP3A4 or the MDR1 promoter was immediately upstream of the luciferase gene (CYP3A4-LUC or MDR1-LUC, respectively). Seven hours later, we administered luciferin and optically imaged the bioluminescence through the skin of each mouse. We observed dose-dependent expression of luciferase, and basal transcriptional activity was measurable with as little as 5 μg of plasmid (Fig. 1, A and B). (Luciferase expression was measurable with only 1 μg of CMV-LUC plasmid; data not shown.) Optical imaging of the excised livers of these mice and the anhepatic mouse demonstrated that all measurable bioluminescence was localized to the liver (data not shown). When luciferase enzyme activity was measured in homogenates of the excised livers (Liu et al., 1999), the results corresponded closely with those of the in vivo optical bioluminescence assays in the same mice (Fig. 1, A and B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

External imaging of luciferase reporter activity in mouse liver. Plasmids bearing the indicated reporter constructs were combined with SV40-LacZ plasmid and hydrodynamically infused. After 7 h, mice were anesthetized and injected with firefly d-luciferin and optically imaged. Emitted light is shown as a pseudocolor graphic over the black and white image of each animal. A, mice that received 1, 5, or 10 μg of −7836-CYP3A4-LUC. B, mice that received 1, 5, or 10 μg of −10-kb-MDR1-Luc. C, mice that received 2 μg of −362-CYP3A4-LUC, −7836-CYP3A4-LUC, or −13,000-CYP3A4. Graphs at right show in vivo and in vitro bioluminescence measurements obtained from the livers of the same mice, according to the quantity of DNA injected. Each graphed data point represents measurements from two to three animals.

Because transcription assays are routinely used to define regulatory regions important in gene transcription, we designed experiments to determine whether the hydrodynamic infusion method can be used to analyze the promoter activity of various deletion constructs. We found that the basal transcription of three CYP3A4-LUC constructs containing various deletions in the CYP3A4 5′-flanking region was readily detectable by optical imaging after infusion of 2 μg of plasmid. When the distal CYP3A4 5′-flanking nucleotides (−7836- to −7208-bp) were fused to the 5′ end of the promoter in the CYP3A4-LUC construct, transcription of the reporter was markedly enhanced in vivo (Fig. 1C), as it had been when the same plasmids were used in vitro (Goodwin et al., 1999).

We next examined the transcription of P450s induced by unique receptor pathways. Environmental toxins, such as dioxin (2,3,7,8-tetrachlorodibenzo-p-dioxin), planar polychlorinated biphenyls, and 3-methylcholanthrene, activate the aryl hydrocarbon receptor (AHR)-AHR nuclear translocator heterodimer and induce transcription of CYP1A1. Anticonvulsant drugs, such as phenobarbital, activate the constitutive androstane receptor (CAR) and enhance transcription of CYP2B, causing some drug interactions. Induction of CYP2B expression by phenobarbital and other xenobiotics, such as the organochlorine pesticide transnonachlor, is a biomarker of hepatic tumor promotion in rodents (Lubet et al., 1989). Expression ofCYP3A and MDR1 is induced by endobiotic and xenobiotic molecules that ligate the nuclear PXR/steroid and xenobiotic receptor (SXR) (also termed NR1I2) (Synold et al., 2001). Because CYP3A and MDR1 participate in the detoxification and elimination of many drugs, their PXR/SXR-induced expression gives rise to many drug interactions. Infusion of the various CYP3A4-LUC and MDR1-LUC plasmids and treatment with well characterized ligands of the various receptors activated reporter transcription in each case (Fig.2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

In vivo monitoring of induction of the transcription of xenobiotic detoxification genes. Mice were hydrodynamically infused with plasmids encoding CYP1A1-LUC (10 μg), −2413-CYP2B1-LUC (2 μg), −2145-CYP2B1-LUC (2 μg), −7836-CYP3A4-LUC (2 μg), or −10-kb-MDR1-LUC (5 μg). Plasmids were administered with SV40-LacZ. Each mouse received 25 μg of DNA. Diagrams at left show XRE binding sites for the AHR-AHR nuclear translocator heterodimer in the CYP1A1 promoter, nuclear receptor 1 and 2 (NR1 and NR2) binding sites for the CAR(C)-retinoid X receptor (R) heterodimer in the rat CYP2B1 promoter, and the direct or everted repeat (DR3, DR4, and ER6) binding elements for the PXR(P)-RXR(R) heterodimer in the CYP3A4 and MDR1 promoters. Photographs show mice after intraperitoneal administration of dexamethasone (DEX) or PCN, spironolactone (SLN), rifampin (RIF), sodium phenobarbital (PB), transnonachlor (TN), 3-methylcholanthrene (3-MC), or vehicle control (CT). Graphs show hepatic bioluminescence measured in vivo and in vitro as described in Fig. 1.

Chen et al. (2001) reported that the extent of expression of DNA that is linearized before hydrodynamic infusion is 10 to 100 times that of equivalent circular DNA and that the transgene is expressed for a period of at least 9 months because the linear plasmids efficiently concatenate in vivo and remain extrachromosomal in hepatic cells. In contrast, the expression of genes in circular plasmids 1 week after injection is reportedly less than 1% of the level of expression 24 h after injection (Liu et al., 1999). We therefore injected 40 μg of linear or closed circular CYP3A4-LUC plasmid into mice and investigated luciferase transcription at various time points. An inducer of CYP3A transcription (PCN) was administered to mice 7 h before imaging. Expression of luciferase from the circular plasmid declined steadily; it had declined 98% 3 weeks after injection, and after 4 weeks, little fluorescence was detected. Three and four weeks after injection, bioluminescence in the mice that had received CYP3A4-LUC–linearized plasmids was 10 times that in mice that had received closed circular plasmids (Fig.3). Twelve weeks after DNA injection, bioluminescence was still detectable in mice with linearized but not circular CYP3A4-LUC plasmid (data not shown). Therefore, this technique allows serial real-time measurement of hepatic gene transcription in a single live mouse. Extracorporal monitoring of bioluminescence from a transcription activation assay provided results equal to or better than those obtained by traditional in vitro assays conducted with the same constructs (Figs. 1 and 2) and allowed these measurements to be made in vivo in a physiologically normal environment and in real time. Because the animal need not be sacrificed, the time course of transcription can be documented.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Bioluminescence is more persistent after administration of linearized plasmids than after administration of circular plasmids. Mice were hydrodynamically infused with circular or linearized −7836-CYP3A4-LUC. Mice were treated with PCN 7 h before optical imaging.

We next showed a practical application of this technique. We determined whether we could use the in vivo transcription assay to identify new PXR ligands. Hydrodynamic infusion of CYP3A4-LUC and treatment with a number of anticancer drugs demonstrated that etoposide, tamoxifen, and topotecan increased CYP3A4-LUC transcription in vivo (Fig.4A). To determine whether this was a PXR-dependent mechanism, we cotransfected NIH 3T3 cells, which are deficient in PXR nuclear hormone receptor, with CYP3A4-LUC and PXR. Addition of 10 μM rifampin or 10 and 50 μM etoposide or topotecan produced a dramatic increase in CYP3A4-LUC reporter activity but only in those cells cotransfected with the human PXR plasmid (Fig.4C). Tamoxifen (at 10 and 50 μM) was extremely toxic to both NIH 3T3 and other cells tried and, thus, could not be definitively established in this cellular model as a PXR ligand. To determine whether etoposide and topotecan bind directly to PXR as ligands, we cotransfected NIH 3T3 cells with a GAL-PXR/SXR construct containing only the ligand binding domain of PXR/SXR fused to the DNA binding domain of the yeast transcription factor GAL4, along with a reporter containing the GAL4 response element fused to a luciferase reporter. Etoposide and topotecan at concentrations ranging from 10 to 50 μM readily increased PXR-GAL4 transcriptional activity in a PXR-dependent fashion (Fig. 4B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Anticancer drugs induce CYP3A4 transcription. Mice were hydrodynamically infused with −7836-CYP3A4-LUC (A). Mice were treated with the indicated concentrations of drugs and optically imaged 7 h later. Red, in vivo; blue, in vitro; two separated experiments are represented. NIH 3T3 cells were transfected with TK(MH100)₄-LUC reporter and either vector (PM2) or GAL-hPXR/SXR plasmid (B). NIH 3T3 cells were transfected with −7836-CYP3A4-LUC with hPXR or empty vector (pSG5) (C). Cells were treated with rifampin (RIF), etoposide (VP-16), or topotecan (Topo). Luciferase activities were normalized to β-galactosidase activity, and the fold increases in treated groups over untreated controls were determined and expressed as a percentage of rifampin induction. Values represent the mean ± S.D. from triplicate determinations of a single experiment that was performed at least three times.

Discussion

Although a variety of methods allow the rapid introduction of DNA into hepatocytes in vivo (e.g., direct injection of DNA into the liver and transduction with adenoviral vectors), such approaches are limited by expression of the DNA only at the site of injection and by the development of an immune response to adenovirus that specifically inhibits the expression of many drug detoxification genes (Muntane-Relat et al., 1995). Our new in vivo real-time transcription assay is free of such limitations. We used six different gene reporter constructs identical to those widely used to measure transcription activity in cell lines. Our results demonstrate that this technique is broadly applicable and requires only the technology already at hand in most laboratories. Because animals need not be sacrificed, transcription can be measured as it occurs, under the most informative biological conditions (i.e., in a living, intact organ). This technique is not limited to investigation of the transcription of P450 or other drug-metabolizing genes. If promoter constructs for genes encoding growth factors or cytokines and their receptors were introduced into animals, their transcription could be monitored in real time during physiological processes, such as organ regeneration or inflammation, or during chronic pathological processes, such as the development of cirrhosis or neoplasia. Only a small set of mice would be needed. With existing technology, such studies would require that several mice be sacrificed at each time point.

The power of this technique was demonstrated in the rapid discovery in vivo of several anticancer drugs (etoposide, topotecan, and tamoxifen) as transcriptional activators of CYP3A4. Traditional in vitro PXR screening techniques further identified etoposide and topotecan as PXR ligands. This result demonstrates a further advantage of the in vivo system for testing compounds that are not toxic to quiescent hepatocytes in vivo (e.g., tamoxifen) but cannot be screened well in vitro in replicating cell models due to extensive toxicity. Moreover, this in vivo approach capitalizes on the additional pharmacokinetic parameters that can influence transcriptional activation. For example, because replicating cells lose expression of drug-metabolizing enzymes, these in vitro screening models would not detect those chemicals where the metabolite, but not parent drug or steroid, is the ligand for a nuclear hormone receptor and transcriptional activator (Gant et al., 1991). Thus, the in vivo model would complement existing in vitro screening methods. Our finding that tamoxifen, etoposide, and topotecan induce CYP3A4-LUC and are PXR ligands is important because the efficacy of anticancer drug therapy is related to systemic exposure. Indeed, it has previously been demonstrated that concurrent treatment with CYP3A4 inducers and chemotherapies metabolized by CYP3A leads to decreased therapeutic effect and long-term survival (Relling et al., 2000) and that CYP3A expression correlates with sensitivity to anticancer drugs in human cancer xenografts (Zembutsu et al., 2002).

In conjunction with the increasing development of genetically modified mice, our real-time transcription assay will permit immediate testing of the in vivo roles of cis-acting regulatory elements and their cognate binding factors in the transcription of reporter constructs. Many additional pharmaceutical and toxicological applications can be envisioned. Reporter constructs that carry DNA regulatory sequences of genes implicated in the hepatotoxicity of xenobiotic agents could be used with traditional toxicity screens to aid in the selection of drug candidates for development. Indeed, microarray analysis of changes in gene expression in mouse liver after exposure to various chemicals has revealed a diagnostic set of 12 RNAs that can be used to classify chemicals into five toxicologically informative subsets (Thomas et al., 2001). “Toxicogenomic” profiling of transcripts could be used to assess the potential toxicity of specific chemicals. When DNA sequences that regulate expression of genes implicated in hepatotoxicity are well defined, linearized reporter gene constructs can be employed to monitor the expression of the genes during acute or chronic toxicity testing of drugs or chemicals.

Because many drugs act through receptors, reporter constructs could also be used in vivo to confirm mechanisms of drug action. Reporter constructs bearing DNA-regulatory sequences activated by steroid and nuclear hormone receptors would be powerful sensors of physiological endocrine, lipid, retinoid, fatty acid, and sterol signaling in vivo (Chawla et al., 2001). Those that carried target sequences for xenobiotic sensors, such as PXR/SXR and CAR, would be particularly useful in drug development. For example, mice that express the human SXR transgene could be infused with a plasmid in which the reporter gene is linked to the promoter of a drug-detoxifying gene, such asCYP3A4 or MDR1, to evaluate the ability of human SXR receptor ligands to induce expression of these genes (Xie et al., 2000). Similarly, infusion of CYP3A4-LUC plasmid into the livers of mice with and without the MDR1 efflux pump would simultaneously identify drugs that are able to induce expression of CYP3A4 and those that will undergo MDR1-mediated efflux.

Hydrodynamic infusion of DNA can be used with genes other than reporter genes (Yant et al., 2000; Yang et al., 2001). Sequences that encode P450 gene products that participate in detoxification, growth factors, cytokines, or their receptors could be substituted for the reporter genes, and the resulting constructs could be used to create mice that carry the human transgene of interest. Transgenic “humanized” mice could be of significant value in the preclinical evaluation of new drugs. These in vivo approaches would greatly complement and validate in vitro drug screening assays that detect potential drug interactions. Moreover, the in vivo assay would faithfully recapitulate all of the pharmacokinetic variables in effect in the intact liver, such as absorption, distribution, metabolism, export, and inflammatory processes, that affect responses to xenobiotic agents. Although these variables cannot be recapitulated in cell cultures, they influence ligand availability, receptor activation, and the initiation of transcription in the entire animal.

These same imaging techniques could be used to examine protein-protein interactions in mouse liver in vivo. For example, two-hybrid assays using the ligand binding domain of nuclear receptors fused to the yeast transcription factor GAL4-DNA binding domain hydrodynamically introduced along with a reporter plasmid containing the GAL4 response element fused to a luciferase reporter could be used as ligand screening assays. Furthermore, because many cell processes require protein-protein interactions, construction of any combination of plasmids that ultimately monitor the protein-protein interaction with a luciferase reporter readout would be amenable to hydrodynamic infusion into mouse liver and noninvasive screening in vivo. An additional likely benefit of this technique, a reduction in the number of animals needed for investigation of gene expression, will benefit the entire scientific community.