Control and Augmentation of Long-Term Plasmid Transgene Expression In Vivo in Murine Muscle Tissue and Ex Vivo in Patient Mesenchymal Tissue

Purpose. In vivo gene therapy directed at tissues of mesenchymal origin could potentially augment healing. We aimed to assess the duration and magnitude of transene expression in vivo in mice and ex vivo in human tissues. Methods. Using bioluminescence imaging, plasmid and adenoviral vector-based transgene expression in murine quadriceps in vivo was examined. Temporal control was assessed using a doxycycline-inducible system. An ex vivo model was developed and optimised using murine tissue, and applied in ex vivo human tissue. Results. In vivo plasmid-based transgene expression did not silence in murine muscle, unlike in liver. Although maximum luciferase expression was higher in muscle with adenoviral delivery compared with plasmid, expression reduced over time. The inducible promoter cassette successfully regulated gene expression with maximum levels a factor of 11 greater than baseline. Expression was re-induced to a similar level on a temporal basis. Luciferase expression was readily detected ex vivo in human muscle and tendon. Conclusions. Plasmid constructs resulted in long-term in vivo gene expression in skeletal muscle, in a controllable fashion utilising an inducible promoter in combination with oral agents. Successful plasmid gene transfection in human ex vivo mesenchymal tissue was demonstrated for the first time.


Introduction
Gene therapy is now a realistic prospect for the treatment of a variety of musculoskeletal disorders because of the emerging knowledge concerning proteins that govern the processes of growth and regeneration of tissues [1,2]. Many orthopaedic tissues do not heal well after injury [3]. Research has identified numerous growth factors and other gene products with the ability to promote regeneration [4]. The most extensively studied are members of the transforming growth factor (TGF) superfamily, collectively known as bone morphogenetic proteins (BMPs). Because gene therapy provides the gene rather than just a degradable protein, this technique may result in higher and more reproducible levels of protein production with the potential for long-term protein delivery. This would allow either local protein production for tissue healing and so forth or potentiate long-term systemic delivery. Unlike current methods, protein delivery could also be intracellular. Various methods have been employed to deliver genetic material. The most commonly used in the orthopaedic field have been viral vectors and ex vivo manipulation with considerable preclinical success [5,6]. There are concerns regarding viral recombination/infectivity, immunogenicity, and possible carcinogenicity [7][8][9]. They are expensive to produce and as vectors have biophysical and genetic limitations. Also many gene therapy strategies prove ineffective, suffering from gene silencing [10]. Nonviral therapy has not been extensively studied in orthopaedic research [11,12]. In theory, non-viral therapy lacks some of the drawbacks associated with viral systems. In this study, the long-term viability of murine skeletal muscle plasmid gene expression in vivo was assessed using live whole body imaging of luciferase expression, along with the potential for expression control using an inducible model. Finally an ex vivo model was developed in murine musculoskeletal tissue and conditions replicated to examine plasmid gene expression in human ex vivo mesenchymal tissue.

Animals.
B-alb C mice were obtained from Harlan Laboratories (Oxfordshire, England). They were kept at a constant room temperature (22 • C) with a natural day/night light cycle in a conventional animal colony. Standard laboratory food and water were provided ad libitum. Before experiments, the mice were afforded an adaptation period of at least 14 days. Female mice in good condition, without fungal or other infections, weighing 16-22 g and of 6-8 weeks of age, were included in experiments.

In Vivo and Ex Vivo Gene Delivery.
All in vivo and ex vivo animal experiments were approved by the ethics committee of University College Cork. Mice were randomly divided into experimental groups and subjected to specific experimental protocols. Mice were anaesthetized during all treatments by i.p. administration of 200 µg xylazine and 2 mg ketamine. For liver transfection, a 1 cm subcostal incision was made over the liver and the peritoneum opened. The right lobe of the exposed liver was administered plasmid by electroporation as described below. The wound was closed in two layers-peritoneal and skin-using 4/0 prolene sutures (Promed). All ex vivo human tissue experiments were approved by the Clinical Research Ethics Committee of the Cork Teaching Hospitals. Informed consent was obtained from patients scheduled for lower limb amputation and the surgeon's permission obtained. Upon removal of sample, it was immediately immersed in culture medium and brought to the laboratory. Samples were cut to a size approximating the medial thigh muscle of Balb-C mice (1.5 cm 3 ).
For plasmid delivery by electroporation, a customdesigned applicator with 2 needles 4 mm apart was used, with both needles placed through the skin central to the tissue. Tissue was injected between electrode needles with plasmid DNA in sterile injectable saline in an injection volume of 50 µl. Concentration of plasmid was adjusted to administer 8 × 10 11 gene copy numbers. Plasmid concentration was determined using the Nanodrop spectrophotometer (ND-1000 Spectrophotometer, Labtech Int, East Sussex, UK).
The number of plasmid copies was determined using the following calculation: mass of plasmid DNA g = no. of gene copies bp × plasmid size × 1.096 × 10(−21) . (1) Plasmid sizes for pCMV luc and pGTRTL were 6.9 kbp and 7.5 kbp, corresponding to a mass of 6.05 × 10 −6 g and 6.576 × 10 −6 g, respectively.
After 80 seconds, square-wave pulses (1200 V/cm 100 µsec × 1 and 120 V/cm 20 msec, 8 pulses) were administered in sequence using a custom-designed pulse generator (CythorLab, Aditus, Lund, Sweden). In all cases, a drop in impedance measurement was taken as evidence for successful tissue cell poration.
Replication incompetent recombinant Adenovirus 5 particles encoding the luciferase gene under the transcriptional control of the CMV promoter were a kind gift from Prof. Andrew Baker, University of Glasgow; they were generated and titrated as described previously [13]. Viral vector particles were administered by direct intramuscular injection in a volume of 50 µl. 1 × 10 9 VP of replication incompetent recombinant Adenovirus 5 particles was used per administration.

Ex Vivo Optimisation.
The entire medial thigh muscle tissue was harvested from sacrificed B-alb C mice (1.5 cm 3 ). Samples were stored as per the experimental protocol to assess survival in either RPMI (Roswell Park Memorial Institute medium) or DMEM (Dulbecco/Vogt modified Eagle's minimal essential medium) at 37 • C for 24 hours. The effect of electroporation on cell survival was assessed by administering the plasmid as described previously and assessing viability at 24 hours.
Cell viability was assessed using the propidium-iodidebased NucleoCounter kit as previously described [14]. Samples were washed and processed through a fine mesh filter (BD Biosciences, Oxford, UK), using a pestle and PBS (phosphate buffered saline) until a suspension was obtained. The suspension was made up to a standard volume, and cell viability was assessed. Viability at time zero was taken as 100% relative survival. Subsequent viabilities were expressed as a percentage relative to this.

Doxycycline Administration.
Solutions were prepared by adding 200 µg/mL of doxycycline (Sigma) to distilled water containing 10% sucrose as previously described. All drinking bottles were wrapped in aluminium foil and were renewed every 2-3 days. Noninduced groups were administered solutions of distilled water containing 10% sucrose. A 20 µg/mL solution was added to wells with pGTRTL ex vivo samples as the 200 µg/mL resulted in excessive cell death (data not shown).
2.6. In Vivo Imaging. In vivo luciferase activity from tissues was analysed at set time points after transfection as follows: 80 µl of 30 mg/mL firefly luciferin was injected intraperitoneally. Mice were anaesthetised as before. Ten minutes after luciferin injection, live anaesthetised mice were imaged for 3 min at high sensitivity using an intensified CCD camera (IVIS Imaging System, Xenogen). The exposure conditions were maintained at identical levels so that all measurements would be comparable. All data analysis was carried out on the Living Image 2.5 software package (IVIS Imaging System, Xenogen). Luminescence levels were calculated using standardised regions of interest (ROIs) for all 3 anatomical areas. Actual levels were obtained by subtracting the corresponding ROI of an untransfected mouse to account for background luminescence. For comparison between plasmids, flux was calculated per gene copy number.

Statistical Analysis.
At each time point, either oneway a two-sampled t-test was used to compare mean luminescence within each experimental group. Microsoft Excel 11.3 (Microsoft) was used to manage and analyze data. Statistical significance was defined at the standard 5% level.

Duration of Gene Expression In Vivo in Muscle and
Liver. Luciferase-coding plasmid DNA featuring the CMV (pCMV-luc) promoter ( Figure 1) was delivered to liver or thigh muscle by electroporation. IVIS live whole body imaging was used at time points over 370 days to determine luciferase expression ( Figure 2). Expression driven by the CMV promoter in liver was initially high but reduced rapidly to background levels by day 14. A different temporal pattern of expression was observed in muscle tissue, with no gross reduction in expression over time observed. Total luminescence for the period was 1.33 × 10 −6 p/sec/cm 2 /sr/plasmid copy. Unlike liver, muscle therefore has the potential to act as a long-term conduit for gene expression.

Comparison of the Duration and Magnitude of Gene Expression In Vivo in Muscle following Plasmid and Viral
Delivery. Luciferase-coding plasmid DNA featuring the CMV (pCMV-luc) promoter or Adenovirus 5 particles encoding the luciferase gene under the transcriptional control of the CMV promoter was delivered to murine thigh muscle as described. IVIS live whole body imaging was used at time points over 18 days to compare luciferase expression between the two constructs. Maximum luciferase expression for both constructs is shown in Figure 3(a). Maximum expression was significantly higher using adenoviral delivery when compared to that of plasmid delivery. Expression  Figure 2: Duration of plasmid-based CMV promoter activity in vivo in liver and muscle. pCMV-luc was delivered to liver and thigh muscle (n = 10) by electroporation and luminescence analysed in vivo over time using IVIS live whole body imaging. CMV activity in liver is initially high but reduced to background level by day 14. A gross reduction in expression over time was not apparent in muscle.
following adenoviral delivery, however, decreased over the study period to background levels at 18 days (Figure 3(b)).

Regulation of Gene Expression.
Luciferase expression from an inducible "tet-on" promoter was examined in muscle tissue using the pGTRTL plasmid, both in the presence and absence of induction by doxycycline ( Figure 4). There was a low initial expression from reporter gene preinduction from the time of delivery. However, following induction with doxycycline at day 8, expression was increased by a factor of 11 (P = 0.0001) at day 11. Following withdrawal of doxycycline at day 12, this had returned to baseline by day 16 (P = 0.08). Expression was reinduced at day 105 and had increased by a factor of 9.2 by day 108 (P = 0.001) and by a factor of 10.7 at day 116. Expression remained constant at a maximal level between days 108 and 116 with no statistically significant change in magnitude observed (P = 0.77). Following withdrawal of doxycycline at day 117, this had returned to baseline by day 119 (P = 0.06). There was no difference in maximal expression between the two induction periods (P = 0.28). pGTRTL gene expression is thus rapidly induced with doxycycline and dissipates on withdrawal. It can also be reinduced on a temporal basis to give reproducible levels of expression.

Ex Vivo Delivery to and Transgene Expression in Patient
Mesenchymal Tissues. To assess translation of the murine findings to a clinical setting, samples of muscle, tendon, and bone with intact periosteal tissue were harvested from a human leg amputated for end-stage peripheral vascular disease. Absolute ischaemic time following amputation before immersion in media was less than 40 minutes. 4 samples were taken of each tissue type and transfected with pCMV-luc or pGTRTL as previously described. These were incubated overnight in conditions described above and luminescence assessed at 24 h ( Figure 6). Luciferase

Discussion
Localised gene expression offers certain advantages in the setting of tissue healing. It may provide a more accurate system of protein delivery to the desired tissue, whilst lessening systemic side effects. In addition, it has the potential for sustained protein delivery, both intracellularly and extracellularly without the need for repeated administration. However, inactivation of gene expression in specific cell types has important therapeutic implications, as does the prospect of uncontrolled protein production [10]. Here, we have demonstrated that plasmid-based gene expression in murine muscle does not decrease with time. Long-term plasmid gene expression has been observed in muscle previously, although not using an in vivo luminescence system for a period greater than one year [15,16]. In vivo luminescence has previously been shown to be a reliable indicator of transgene expression [17]. It is also an ideal method of assessing an inducible agent as an "on-off " switch to control protein expression, by allowing repeated measurements in the same subjects over a long time period. Successful bone and tendon healing in animal models have been reported following adenoviral transfer of BMP 2 and BMP 14 [5,6,18]. Here, we have compared luciferase expression between a plasmid construct and viral particles. Although maximum expression was significantly greater with adenovirus, this was rapidly silenced to background levels by day 18. The attraction of the plasmid system would therefore lie in its capacity for sustained levels of predictable expression over time. Electroporation-assisted intramuscular delivery of plasmid DNA has been shown to efficiently transfect cells of myogenic origin with no evidence of transgene expression in infiltrating inflammatory cells [19]. This is broadly in keeping with our observation of long-term expression, with transfection of a nontransient cell population in muscle and the absence of a significant immune response. Silencing has previously been observed with adenovirus, with the mechanism postulated as methylation of the CMV promoter [10]. Why this does not occur when the promoter is delivered using a plasmid construct, however, is unclear.
Uncontrolled long-term gene expression, however, has serious potential side effects. For instance, in the clinical use of recombinant bone morphogenetic protein (rhBMP) for augmentation of fracture healing and joint fusion, heterotopic ossification and other complications have been reported [20]. In this instance, the protein is eventually cleared from the body. Following gene delivery, however, it would be difficult to arrest protein production in the event of unwanted side effects. It is, therefore, desirable to have a plasmid with an "on-off " switch. We have demonstrated that this can be achieved using pGTRTL, a tet-on inducible promoter. After plasmid delivery, gene expression initiated with the commencement of oral doxycycline administration and promptly ceased upon withdrawal. This strategy has also proven to be successful in a preclinical arthritis study [21]. In the clinical setting, should side effects occur, gene expression could be turned off simply by omitting the oral agent.
In addition, growth factors may not be clinically beneficial until days after intervention, while the delivery of rhBMP at the time of surgery has been shown to impair tissue healing by inducing antagonist factors [18,22]. By using pGTRTL, we have demonstrated that the timing and duration of gene expression can be tightly controlled in vivo. This would decrease the chance of antagonist induction, whilst optimising the timescale of protein production. Furthermore, expression was reinduced efficiently and to a reproducible level. This would allow us to repeatedly treat tissue with a known quantity of protein in recalcitrant situations. As the system is orally administered, there would be no need for interference with the physical environment of tissue healing after the initial application. One issue with tetracycline compounds is that they do have an effect on the tissue healing environment and as such may not be the ideal tool for the investigation of therapeutic plasmid use [23]. In terms of principle, however, it demonstrates the potential applicability of an inducible plasmid-based system in a clinical setting.
Finally, we have demonstrated, for the first time, successful plasmid-based gene delivery to human ex vivo muscle tissue. Levels of luciferase expression in ex vivo tendon and muscle were similar when examined at 24 hours. This was surprising, given the increased cellularity of muscle tissue. Although the exact reasons for the similar levels of expression are unclear, it may be related to a threshold effect or differential tissue survival in the ex vivo environment. Expression may differ in the human in vivo setting, particularly with regard to increased metabolic activity in muscle tissue. Nonetheless, we believe this to be an important step in the demonstration of the applicability of this technique in  Figure 6: Human musculoskeletal tissue electrotransfected ex vivo with either pCMV-luc or pGTRTL. Expression was observed in all samples. Levels of expression for the "tet-on" plasmid were significantly increased in the presence of doxycycline.
the in vivo setting, as expression capability cannot be definitely extrapolated from the murine model to the human setting. The expression of plasmid-based delivery in human ex vivo musculoskeletal tissue is encouraging on two levels; firstly, it proves that the system can function in human tissue, and secondly, it raises the possibility of autograft or allograft transfection, which could lead to the acceleration of graft incorporation [24].
The luminescence imaging strategy utilised here provided a rapid, robust and reliable method for ex vivo assessment of transgene expression in patient samples. The method of culturing tissue ex vivo has been demonstrated to provide high viability for up to 1 week post resection, while bioluminescence imaging provides a highly sensitive readout for real-time transgene expression in viable cells [25]. Other reporter gene imaging techniques may also be applicable in this context (fluorescence, PET etc.) [26,27]. In summary, these results indicate, when administered using electroporation, that plasmid constructs result in long-term in vivo gene expression and that this expression can be reliably induced with oral agents. Finally, for the first time, successful plasmid gene transfection with electroporation in human ex vivo mesenchymal tissue has been demonstrated.

Disclosure
The authors have no declared professional or financial affiliations pertinent to this work.