A novel system for mitigation of ectopic transgene expression induced by adenoviral vectors


Adenoviral (Ad) vectors are good candidates for gene therapy in view of their high in vivo gene delivery efficiency. However, greater control over the tissue distribution of transgene expression is required to avoid potentially deleterious effects in non-target organs. In this regard, the liver is particularly at risk due to the high natural tropism of Ad for this organ, where dose limiting toxicity has been seen due to toxic transgene expression. We hypothesized that the cre/loxP system could be utilized to reduce unintended transgene expression at this site. This concept was tested using an Ad vector (AdLCLLL) carrying a reporter gene cassette in which the promoter and luciferase gene were flanked by LoxP sequences. Co-administration of this vector with a second vector carrying the cre recombinase gene in vitro and in vivo resulted in specific down-regulation of transgene expression. This novel approach thus has the potential to improve the safety of gene therapy strategies that rely upon the delivery of genes which may be hepatotoxic.


Adenoviral vectors possess a number of attributes which render them useful gene delivery vehicles for systemic gene therapy. In particular, the in vivo transduction efficiencies achievable with these agents are greater than with currently available alternative vector systems.1 However, measures to control specifically the distribution of delivered transgene expression must be superimposed on the basic vector for optimal applicability. Various approaches to this problem have been proposed, including imparting both transductional and transcriptional targeting properties to Ad vectors.2,3,4,5,6,7 While some measure of success has been achieved with these approaches, ectopic transgene expression in non-target organs remains a potential problem, even when vectors that have been modified so as not to recognize the native Ad receptor, CAR, are used. Thus, additional, complementary, strategies may be superimposed upon these systems to afford an even greater measure of control.

To complement strategies which primarily aim to ensure that transgene expression is sufficiently high in target tissues, we have focused on a strategy to deliberately switch transgene expression off at a site where expression might otherwise be disadvantageous. With respect to Ad vector administration, the majority of vector localization and subsequent transgene expression (using nonselective promoters) is seen in the liver, and as a consequence, hepatotoxicity resulting from ectopic transgene expression is a major concern. In the context of cancer gene therapy using suicide genes, for example, this toxicity has resulted in dose limiting morbidity and mortality in animal models.8,9,10 This problem is clearly an important issue, in addition to concerns about expression of residual viral genes. Thus, an approach which switches transgene delivery off in the liver may improve the safety of suicide gene therapy.

Based on these considerations, we here explore a novel ‘liver off’ system. The approach uses the cre recombinase (Cre)/LoxP system, whereby the Cre enzyme catalyzes recombination at specific DNA sequences (LoxP sites). This technique has previously been used in the generation of transgenic knock-out mice and as an ‘on’ switch in an Ad vector context by excising a DNA stuffer sequence inserted between promoters and transgenes.11,12,13,14,15,16,17,18 The utility of this system to effectively switch off the expression of an Ad-delivered transgene has not been reported. Thus, we hypothesized that a system could be developed whereby the transgene expression resulting from Ad vectors containing LoxP flanked expression cassettes could be effectively switched off by administration of a second vector delivering the Cre gene.

As initial proof of principle we developed a strategy that utilizes the Cre/LoxP system to switch off luciferase reporter gene expression, and evaluated efficacy both in vitro and in vivo. We show here that this approach can indeed inactivate transgene expression and importantly, the approach has efficacy in the liver in vivo.

We wished to evaluate the feasibility of using the Cre/LoxP system specifically to inactivate a delivered transgene. We thus initially constructed a plasmid containing the CMV promoter and the luciferase reporter gene interspersed with LoxP sequences (pLCLLL, Figure 1a). The cytomegalovirus (CMV) promoter was inserted into the EcoRV site between the LoxP sequences of plasmid pBS246 (GibcoBRL, Rockville, MD, USA). CMV-LoxP was then excised as a SmaI/MscI fragment and inserted into the SmaI site of the luciferase reporter gene plasmid PGL3 basic (Promega, Madison, WI, USA). CMV-LoxP-Luc was excised using KpnI and SalI, blunted and inserted into the EcoRV site of pBS246, thus forming plasmid pLCLLL, which was used for initial in vitro analysis. The plasmid pBS185, containing the Cre recombinase gene under the control of the CMV promoter was obtained from GibcoBRL. The plasmid pCMVpA (containing the CMV promoter and a poly A signal but no transgene) was used as a control. Initially, we investigated the ability of cre recombinase to inactivate the expression cassette in a human liver cell line. For these studies, HepG2 cells (American Type Culture Collection) were transfected with either pLCLLL alone or in combination with pBS185, expressing cre recombinase, or the control plasmid pCMVpA. We found that the combination of pBS185 with pLCLLL resulted in a mean 87 ± 2% (mean ± s.d. of three experiments) reduction in luciferase expression compared with pLCLLL alone, or 91 ± 0.8% compared with pLCLLL + pCMVpA (Figure 1b, P < 0.01 by t test). Thus the basic premise that an expression cassette could be inactivated in this way was confirmed.

Figure 1

(a) Diagram of LoxP interspersed luciferase expression cassette. (b) In vitro validation of luciferase expression by cre. Luciferase expression in HepG2 cells transfected with pLCLLL plasmid alone or with pCMVpA control plasmid or pBS185 (expressing cre recombinase). Studies using plasmids were conducted by transfecting cells with the relevant plasmid using Superfect (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions. HepG2 cells were propagated in RPMI (GibcoBRL) medium supplemented with 10% fetal calf serum, penicillin and streptomycin. Cells were plated at 1 × 105 cells per well in 12-well plates. Next day, cells were transfected using a total of 2.5 μg of plasmid per triplicate wells. Forty-eight hours after transfection a luciferase assay using a Luciferase Assay System kit (Promega) was performed according to the manufacturer's instructions. Mean ± s.d. of three experiments shown.

Having established the functionality of our expression cassette, we then evaluated its functionality in the context of Ad vectors. An E1/E3-deleted Ad vector (AdLCLLL) containing the LoxP flanked luciferase cassette was constructed using the AdEasy system.19 The LCLLL segment was excised as a NotI fragment, inserted into the NotI site of the adenoviral shuttle plasmid pShuttle. The adenoviral genome was constructed by homologous recombination with pAdEasy I in BJ5183 cells by standard techniques, then the virus was generated by transfection of the linearized genome into 911 cells. After confirmation of construct by analysis of viral DNA, stocks of virus were generated in 293 cells and quantified by standard plaque assay and optical density titers. AdCMVCre, an adenoviral vector carrying the cre recombinase gene has been previously described, as has AdCMVHSV-TK, which was used as a control vector.18,20 To assess the system in Ad vectors, HepG2 cells were infected with AdLCLLL at a dose of 50, 500 or 5000 viral particles per cell (to demonstrate a dose–response relationship) as well as 500 viral particles per cell in combination with various doses of AdCMVCre (Figure 2). An Ad vector containing the CMV promoter driving luciferase expression (with no LoxP sequences) was used as a control. We found that the combination of AdLCLLL with AdCMVCre resulted in a dose-dependent reduction in luciferase expression (Figure 2a), whereas AdCMVCre infection did not significantly reduce the luciferase expression arising from AdCMVLuc (Figure 2b). In this experiment, the combination of equal particle numbers of AdLCLLL and AdCMVCre led to a reduction in transgene expression equivalent to that seen when the dose of AdLCLLL alone was reduced by 10-fold. Maximum reduction, seen with the highest dose of AdCMVCre used, was by over 99%. Thus, the feasibility of using the Cre/LoxP system to inactivate adenovirally delivered genes was shown.

Figure 2

(a) Cre recombinase inactivation of luciferase in adenoviral vector. Luciferase expression in HepG2 cells infected with AdLCLLL alone (open bars) or with AdCMVCre (solid bars). Numbers on x axis refer to dose of viral particles per cell. Cells were plated at 50 000 per well in 24-well plates. Cells were infected next day with viruses as shown, in culture medium containing 2% FCS. After 1 h the medium was removed and cells were washed with PBS then cultured in complete medium for a further 24 h. Luciferase assay was then performed. (b) Control wells – HepG2 cells plated and infected as in a, but using AdCMVLuc (having no LoxP sites) in place of AdLCLLL. Results shown are representative of two separate experiments. Mean ± s.d. of triplicate determinations.

To evaluate the system further, we determined the effect of infecting cells first with AdCMVCre, then 24 h later infecting with AdLCLLL, thus allowing time for cre recombinase to be generated before the arrival of the LoxP flanked cassette. In a direct comparison of concurrent infection versus temporal separation of infection, greater reduction in luciferase expression was seen when the AdCMVCre was given first (Figure 3). Further temporal manipulations may be optimized depending on the application and may thus allow greater reductions in transgene levels, if required.

Figure 3

Improved effectiveness of cre inactivation concept by temporal separation of infections. HepG2 cells were infected with AdLCLLL, either alone, together with various doses of AdCMVCre, or 24 h after AdCMVCre. Luciferase assay was performed 24 h after AdLCLLL infection. Numbers on x axis refer to dose of viral particles per cell. Mean ± s.d. of triplicate determinations.

To characterize the system further we also looked for any evidence of a bystander effect of cre-expressing cells on adjacent AdLCLLL-expressing cells. HepG2 cells were infected separately with either AdCMVCre or AdLCLLL, then, 24 h later, the AdLCLLL infected cells were mixed at various ratios with either uninfected or AdCMVCre infected cells. As expected, the amount of luciferase expression decreased as the proportion of AdLCLLL cells decreased. However, this decrease was the same whether the other cells were uninfected or whether they expressed cre (Figure 4). Thus no bystander effect was observed. This finding may be advantageous in that the control of the ‘off’ switch should be tightly restricted only to those cells that are infected with AdCMVCre.

Figure 4

No evidence of bystander effect from cells expressing cre. HepG2 cells were infected with AdLCLLL, then 24 h later mixed with uninfected or AdCMVCre infected cells in the proportions shown. Luciferase analysis was performed 24 h after mixing.

The entire rationale for proposing an expression cassette inactivation strategy was to investigate its potential in vivo, especially in the critical context of liver transgene expression. Thus, we conducted in vivo experiments combining AdLCLLL and AdCMVCre to determine the effect on transgene expression. Animal studies were approved by the University of Alabama Institutional Animal Use and Care Committee. Mice were injected via the tail vein on day 1 with either PBS as a control, AdCMVCre (2 × 1010 viral particles) or AdCMVHSV-TK (2 × 1010 particles) which served as an irrelevant virus control. On day 2, mice where injected with AdLCLLL (2 × 1010 particles), then on day 4 mice were killed, livers were harvested and transgene expression was determined (Figure 5a). Approximately one log reduction in transgene expression was seen with the AdCMVCre + AdLCLLL combination whereas no significant reduction was seen with the control combination. Furthermore, no reduction in luciferase expression was seen with the AdCMVCre + AdCMVLuc combination compared to AdCMVLuc alone (Figure 5b). In a separate, repeat experiment, we also evaluated the effect of co-administering AdCMVCre with AdLCLLL on the same day as well as sequentially (Figure 5c). Here, again, a reduction in luciferase expression was seen with the AdCMVCre/AdLCLLL combination, but not with the control AdCMVHSV-TK/AdLCLLL combination. In each experiment using C57 Black6 mice, P < 0.04 (by Student's t test) comparing the logarithmically transformed data from the groups AdCMVCre+ AdLCLLL versus AdHSVTK + AdLCLLL. More variability was seen in the Balb C experiment, although the difference between the AdCMVCre + AdLCLLL and the control combination was still apparent. The incorporation of the relevant controls (irrelevant virus in place of AdCMVCre, and the non-LoxP containing AdCMVLuc in place of AdLCLLL) clearly shows that the effects are specific for the Cre/LoxP combination, and not due either to the potential induction of anti-Ad immune responses or nonspecific toxicity related to AdCMVCre. Thus, we have shown that our inactivating vector strategy does indeed have functionality in vivo.

Figure 5

In vivo effectiveness of liver untargeting concept. Adenoviral vectors were administered by tail vein injection (total volume 200 μl) into female C57/Bl6 mice (a,b) or female Balb C mice (c) aged 6–8 weeks (Charles Rivers, Wilmington, MA, USA). AdCMVCre or control (PBS or AdCMVHSV-TK) was administered on day 1, then AdLCLLL on day 2, and animals killed on day three. Livers were harvested, snap-frozen in ethanol/dry ice, ground to a fine powder using a mortar and pestle sitting in an ethanol/dry ice bath, then powders were assayed for luciferase activity using the Promega Luciferase Assay System kit. (a) (n = 3 per group). (b) Repeat experiment as in a, but including additional control of AdCMVluc showing no effect of AdCMVCre pre-treatment (n = 4 per group). (c) Further repeat experiment using Balb C mice, n = 3 per group. Last two bars represent administration of AdLCLLL together with either AdCMVHSV-TK or AdCMVCre together as one injection. In each case bars = mean ± s.d.

Efforts to achieve improved control of gene expression via improvements in basic vector design have been a major focus of gene therapy research in recent years. These approaches have principally focused on modifying either the transduction properties of vectors via tropism modification, or the gene expression profile of delivered transgenes via the use of tissue-specific promoters. For the greatest possible control to be achieved, it is likely that some combination of targeting approaches will be required. The strategy presented in the article represents a further, complementary approach, especially in those situations where ectopic transgene expression may be toxic. In addition, it would be feasible to add additional LoxP sites in other areas of the Ad genome, thereby potentially inactivating not only transgene expression but also ectopic expression of residual native viral genes. Further optimization may be possible with respect to dose and timing of the two vectors, but importantly, we have determined that the approach has potential use in vivo.

In its present form the approach may have utility in situations where potentially toxic transgenes are administered in a loco-regional context for tumor therapy. In this regard, administration of ad vectors carrying the HSVTK transgene into the peritoneum has lead ectopic transduction of hepatocytes, resulting in dose limiting hepatic toxicity upon gancyclovir administration.10 Pre-treatment by intravenous injection of AdCMVCre, leading to hepatic cre expression, could protect the liver from the toxicity of a peritoneally administered Ad vector carrying a LoxP flanked HSVTK gene. The key principle is to have cre expression being dominant at the site to be protected, and the therapeutic gene dominating at the treatment site. As transductional and transcriptional targeting strategies evolve, further rational combinations can be envisaged. For instance, one Ad vector (carrying a LoxP flanked therapeutic gene) could be transductionally targeted to tumor upon intravenous administration, whereas a second vector having native tropism and carrying the cre recombinase gene would preferentially go to the liver, thereby reducing any ectopic transgene expression that may have arisen due to incomplete specificity of the transductional targeting. In its present form, the lack of a bystander effect indicates the need for co-infection of the one cell for the inactivation strategy to work. To reduce potential inefficiencies resulting from this restriction, cre recombinase could be placed under the control of a liver-specific promoter. Such an approach could then allow for the derivation of single vectors possessing two cassettes – one having a tumor-specific promoter driving a LoxP flanked therapeutic gene, the other having a liver-specific promoter driving cre. Such a system would therefore avoid a net increase in the total Ad vector administration needed with the current approach. Furthermore, the use of conditionally replicative Ad agents has received considerable attention recently and has shown promise as a cancer therapy.21 Incorporation of a Cre/LoxP control system might also offer improved safety in this setting.

In summary, the novel application of the Cre/LoxP system described here provides a basis for a multitude of potential applications to improve significantly the utility and safety of adenovirally delivered transgenes.


  1. 1

    Brody SL, Crystal RG . Adenovirus-mediated in vivo gene transfer Ann NY Acad Sci 1994 716: 90–101

  2. 2

    Reynolds PN et al. A targetable injectable adenoviral vector for selective gene delivery to pulmonary endothelium in vivo Mol Ther 2000 2: 562–578

  3. 3

    Douglas JT et al. Targeted gene delivery by tropism-modified adenoviral vectors Nat Biotech 1996 14: 1574–1578

  4. 4

    Dmitriev I et al. An adenovirus vector with genetically modified fibers demonstrates expanded tropism via utilization of a coxsackievirus and adenovirus receptor-independent cell entry mechanism J Virol 1998 72: 9706–9713

  5. 5

    Wickham TJ et al. Increased in vitro and in vivo gene transfer by adenovirus vectors containing chimeric fiber proteins J Virol 1997 71: 8221–8229

  6. 6

    Adachi Y et al. Midkine promoter-based adenoviral vector gene delivery for pediatric solid tumors Cancer Res 2000 60: 4305–4310

  7. 7

    Koeneman KS et al. Osteocalcin-directed gene therapy for prostate-cancer bone metastasis World J Urol 2000 18: 102–110

  8. 8

    Gao GP, Yang Y, Wilson JM . Biology of adenovirus vectors with E1 and E4 deletions for liver-directed gene therapy J Virol 1996 70: 8934–8943

  9. 9

    Brand K et al. Liver-associated toxicity of the HSV-tk/GCV approach and adenoviral vectors Cancer Gene Ther 1997 4: 9–16

  10. 10

    van der Eb MM et al. Severe hepatic dysfunction after adenovirus-mediated transfer of the herpes simplex virus thymidine kinase gene and ganciclovir administration Gene Therapy 1998 5: 451–458

  11. 11

    Lakso M et al. Efficient in vivo manipulation of mouse genomic sequences at the zygote stage Proc Natl Acad Sci USA 1996 93: 5860–5865

  12. 12

    Ramirez-Solis R, Liu P, Bradley A . Chromosome engineering in mice Nature 1995 378: 720–724

  13. 13

    Wang Y, Krushel LA, Edelman GM . Targeted DNA recombination in vivo using an adenovirus carrying the cre recombinase gene Proc Natl Acad Sci USA 1996 93: 3932–3936

  14. 14

    Sakai K, Mitani K, Miyazaki J . Efficient regulation of gene expression by adenovirus vector-mediated delivery of the CRE recombinase Biochem Biophys Res Commun 1995 217: 393–401

  15. 15

    Lewandoski M, Martin GR . Cre-mediated chromosome loss in mice Nat Genet 1997 17: 223–225

  16. 16

    Okuyama T et al. Efficient Fas-ligand gene expression in rodent liver after intravenous injection of a recombinant adenovirus by the use of a Cre-mediated switching system Gene Therapy 1998 5: 1047–1053

  17. 17

    Bilbao G et al. Construction of a recombinant adenovirus vector encoding Fas ligand with a CRE/Loxp inducible system Transplant Proc 1999 31: 792–793

  18. 18

    Arafat WO et al. An adenovirus encoding proapoptotic Bax induces apoptosis and enhances the radiation effect in human ovarian cancer Mol Ther 2000 1: 545–554

  19. 19

    He TC et al. A simplified system for generating recombinant adenoviruses Proc Natl Acad Sci USA 1998 95: 2509–2514

  20. 20

    Rancourt C et al. FGF2-enhancement of adenovirus-mediated delivery of the herpes simplex virus thymidine kinase gene results in augmented therapeutic benefit in a murine model of ovarian cancer Clin Cancer Res 1998 4: 2455–2461

  21. 21

    Alemany R, Balague C, Curiel DT . Replicative adenoviruses for cancer therapy Nat Biotechnol 2000 18: 723–727

Download references


This work was supported by the following grants: NIH R01 CA74242, NIH R01 HL50255, NIH R01 CA 86881–01 and NCI N01C0–97110.

Author information

Correspondence to DT Curiel.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Reynolds, P., Holmes, M., Adachi, Y. et al. A novel system for mitigation of ectopic transgene expression induced by adenoviral vectors. Gene Ther 8, 1271–1275 (2001). https://doi.org/10.1038/sj.gt.3301511

Download citation


  • adenovirus
  • cre recombinase
  • gene therapy
  • liver

Further reading