A calcium-sensitive promoter construct for gene therapy

Abstract

Targeting diseased cells is a challenging issue in both pharmacological and biological therapeutics. Gene therapy is emerging as a novel approach for treating rare diseases and for illnesses for which there is no other alternative. An important limitation of gene therapy has been the off-target effects and therefore efforts have been focused on increasing the specificity of gene transfer to the targeted organ. Here, we describe a promoter containing six nuclear factor of activated T cells (NFAT) consensus sequences, which is as efficient as the cytomegalovirus (CMV) promoter to drive expression in vascular smooth muscle cells both in vitro and in vivo. In contrast to the CMV promoter it is activated in a Ca²⁺ -dependent manner after endoplasmic reticulum depletion and allows the transgene expression only in proliferative/diseased cells. Overexpression of sarco/ endoplasmic reticulum (SR/ER) Ca^{2 +} ATPase 2a under the control of this NFAT promoter inhibits restenosis after angioplasty in rats. In conclusion, this promoter may be useful for gene therapy in vascular proliferative diseases and other diseases involving upregulation of the NFAT pathway.

INTRODUCTION

The calcium ion, Ca²⁺, controls various cellular functions, including contraction, secretion, cell growth and proliferation.¹ Alteration in Ca²⁺ signaling leads to modifications of the cell fate and is at the origin of many diseases. Defective intracellular Ca²⁺ homeostasis has been reported consistently in heart failure (HF)² and in vascular proliferative diseases.^1,3

Ca²⁺ cycling proteins and their regulators are thus potential therapeutic targets. Among them the sarco/endoplasmic reticulum (SR/ER) Ca²⁺ ATPase 2a (SERCA2a), which takes up Ca²⁺ from the cytosol to the SR, is essential for cardiomyocytes relaxation. In HF, SERCA2a activity is decreased because of reduced expression² and post-translational modifications.⁴ Increasing SERCA2a activity through gene transfer has been used in many preclinical studies and is now at the clinical level.⁵ Similarly, vascular smooth muscle cells (VSMC) proliferation is associated with reduced SERCA2a expression and increased stromal interacting protein 1 (STIM1) expression, a Ca²⁺ sensor in the ER that participates to the store-operated Ca²⁺ entry. In vivo restenosis can be prevented by gene transfer using either a SERCA2a cDNA or a STIM1-specific short hairpin RNA.^6,7

In HF as well as in vascular proliferative disorders, alteration in Ca²⁺ signaling leads to the activation of the transcription factor NFAT (nuclear factor of activated T cells).^1,8–11 The NFAT family (NFAT1 -NFAT5) consists of five ubiquitously expressed transcription factors. Among them NFAT1–4 are regulated by intracellular Ca²⁺ signaling. In resting cells, NFAT is phosphorylated by various kinases and remains in the cytoplasm. When cells are stimulated by different stresses and growth factors, Ca²⁺ influx activates calcineurin, a calcium/calmodulin-dependent phosphatase, which dephosphorylates NFAT allowing its translocation to the nucleus. NFAT then activates different target genes implicated in proliferation and inflammation. NFAT activation also has a role in lymphocytes activation,¹² cardiac growth and smooth muscle proliferation,¹³ tumor angiogenesis, invasive migration and survival of tumor cells.^14,15

Development of a promoter that restricts expression of the transgene to the target tissue/diseased cells will increase the safety of the system by minimizing adverse effects due to expression in the nondiseased cells or nontarget tissue. Short cell-type synthetic composite regulatory elements have been used to drive reporter gene expression in specific types of mammalian cells.^16,17 We hypothesized that a synthetic promoter containing NFAT responsive elements could drive expression of transgenes of interest in proliferative VSMC. We designed a promoter containing six NFAT-binding sequences and used this NFAT-activated promoter to produce an adenovirus overexpressing the firefly luciferase or human SERCA2a and checked their expression in VSMC in vitro and in vivo.

RESULTS

The NFAT promoter drives Ca²⁺ -regulated gene expression We designed a promoter consisting of six repeated consensus NFAT-responsive elements upstream a TATA box (see Materials and Methods). The sequence of this promoter called NF6 is presented in Figure 1a. To determine whether this short promoter could drive expression in a Ca²⁺ -regulated manner, VSMC where infected with AdNF6-Luc construct. Quiescent VSMC were treated with 1 µM thapsigargin, a strong inhibitor of SERCA proteins, to deplete Ca²⁺ internal stores. Luciferase-reporter assay demonstrated that in VSMC, NFAT was activated and luciferase subsequently induced by thapsigargin (Figure 1b). In addition, overexpression of SERCA2a by AdCMV-S2a in proliferating VSMCs co-infected with AdNF6-Luc leads to a decreased luciferase activity compared with AdCMV-bGal;thus, demonstrating the existence of feedback regulation of SERCA2a over NF6 promoter activation (Figure 1c).

Figure 1.

NF6 promoter is a calcium-sensitive promoter. (a) Sequence of NF6 promoter. (b) VSMC were infected with AdNF6-Luc for 24 h, then serum-starved for 24 h and treated with thapsigargin 1µM for 4h. Luciferase activity was normalized to total protein amount. (c) VSMC were coinfected with AdNF6-Luc and AdCMV-βGal or AdCMV-S2a and cultured 72h. Luciferase activity was normalized to total protein amount (*P<0.05).

The NFAT promoter is activated during serum-induced proliferation

The basal activity of the cytomegalovirus (CMV) promoter is 1500 fold higher than the NF6 promoter activity in serum-starved cells (Supplementary Figure 1). Nevertheless, expression of luciferase under the control of the CMV promoter was insensitive to the presence of serum and the proliferative status of the VSMC (Figure 2a), whereas the NFAT promoter markedly induced luciferase expression in presence of 10% fetal bovine serum (FBS) (Figure 2b). This effect was completely abolished using 5 μM of cyclosporine A, which inhibits calcineurin and thus NFAT activation. Expression was induced in a dose- and time-dependent manner (Figures 2c and d).

Figure 2.

NF6 promoter is induced during proliferation. VSMC were transfected with pCMV-Luc (a) or pNF6-Luc (b) and cultured 48h in absence of FBS. Then, cells were stimulated with 10% FBS in absence or presence of 5 µM cyclosporine A for 4h. Luciferase activity was normalized to total protein amount (*P<0.05). To investigate the dose-dependence and time course of NF6 activity, VSMC were infected with AdNF6-Luc and then serum starved for 24 h. Cells were then treated with increasing FBS doses (0–10%) for 4 h (c) or treated with 10% FBS and lysed at different times (0–16h). (d) Luciferase activity was normalized to total protein amount.

SERCA2a protein expression was low in quiescent (0% FBS) passaged VSMC (data not shown) as previously demonstrated.^11,18 Infection of these cells by AdNF6-S2a did not increase significantly SERCA2a expression compared with control cells infected by AdCMV-pGal. However, cells infected by AdCMV-S2a show a strong overexpression of SERCA2a. In proliferative (10% FBS) VSMC, AdCMV-S2a led to the same induction of SERCA2a as in quiescent cells. Interestingly, in this condition AdNF6-S2a allowed an induction of SERCA2a similar to the one observed in cells infected with AdCMV-S2a (Figures 3a and b).

Figure 3.

NF6 promoter is as efficient as CMV promoter in vitro and its expression is limited to proliferative cells. (a) Comparison of CMV-driven and NF6-driven SERCA2a expression in cultured VSMC by western blot analysis. Cells were infected with various adenovirus as indicated and then cultured in the presence or in the absence of serum. Total proteins were harvested 4 days after infection. For western blot, 30 μg of total protein extracts were loaded. (b) Quantification of western blot. SERCA2a expression was normalized to actin expression. (c) VSMC were cultured and infected with adenovirus for 48h and then cultured for 48h with bromo- deoxy uridine (BrdU) in presence or absence of FBS. Proliferation was measured by BrdU incorporation. Data are expressed as a percentage of value in control wells (0% FBS). Bars represent mean ± s.e.m. of four experiments in triplicate (*P<0.05).

The NFAT promoter is as efficient as the CMV promoter Figures 3a and b indicate that the NF6 promoter is as efficient as the CMV promoter in transducing SERCA2a expression in proliferating VSMC. We had previously shown that proliferation of VSMC was inhibited by infection with AdCMV-S2a.⁷ We thus tested the efficacy of AdNF6-S2a on VSMC proliferation. Proliferation was induced by addition of FBS and as expected proliferation was inhibited by AdCMV-S2a. Interestingly, proliferation was inhibited to a similar extend by AdNF6-S2a (Figure 3c). There was no effect of AdNF6-Luc and AdCMV-βGal used as controls.

Then, we tested the efficacy of the NF6 promoter in an in vivo model of VSMC proliferation in which we had previously reported that S2a gene transfer driven by the CMV promoter efficiently prevented post-injury restenosis.⁷ Angioplasty of the rat left carotid artery was followed by infection with AdNF6-S2a or AdNF6-Luc or by saline injection. The animals were killed 14 days after surgery and carotid arteries were removed. Infection was monitored by PCR (Figure 4a). Luciferase or SERCA2a transgenes were observed by immunofluorescence (Figure 4b). Luciferase was mainly in the luminal side of the transduced vessel and never observed in noninfected arteries. SERCA2a was present in noninjured noninfected and in injured AdNF6-S2a-infected arteries, but not in AdNF6-Luc-infected ones in agreement with its disappearance after angioplasty (Figure 4b).

Figure 4.

Evaluation of the infection 14 days after surgery. (a) PCR validating adenovirus infection of rat left carotid arteries at 14 days: positive control corresponds either to NF6-Luc (top panel) or NF6-S2a (middle panel) plasmid DNA or to VSMC genomic DNA (lower panel). (b) Transduction of rat carotid arteries with AdNF6-Luc and AdNF6-S2a. Confocal immunofluorescence with anti-luciferase or anti-SERCA2a of rat carotid arteries cross sections are shown in red. In green is presented the autofluorescence of elastin, which characterize the medial layer of the vessel. The right bottom image shows the Alexa 594-conjugated secondary antibody alone. These images are representative images obtained from five rats for each condition. Ad, adventitia; m, media; ni, neointima.

Morphometric analysis was performed on hematoxylin/eosin stained carotid artery cross sections (Figures 5a and b). Adventitia thickness was increased in all injured arteries as a consequence of injury. Two weeks after injury, abundant neointimal thickening was observed in noninfected and AdNF6-Luc-infected vessels, whereas in AdNF6-S2a-infected animals neointima was poorly to not detectable (Figures 5a and b). No difference was observed for media area between control, AdNF6-S2a-infected and AdNF6- Luc-infected arteries (n = 5 in each group) (0.98 ± 0.11; 1.16 ± 0.07; 1.25 ±0.09 mm² respectively, P = ns). The overall intima area and intima/media ratio in injured AdNF6-S2a-infected arteries were significantly lower than in AdNF6-Luc-infected arteries (neointima: 0.13 ±0.03 vs 0.76 ± 0.16 mm², P<0.01; intima/media ratio: 0.11 ±0.02 vs 0.81 ±0.08, P<0.01). Thus, NF6-driven SERCA2a expression was sufficient to significantly reduce neointimal thickening after balloon injury. For comparison the intima/media ratio for the CMV-driven SERCA2a was 0.29 ± 0.04.⁷

Figure 5.

NF6 promoter is efficient to drive SERCA2a expression and limit restenosis in vivo. Angioplasty followed by infection with AdNF6-S2a (n = 5) or AdNF6-Luc (n = 5) was performed on rats left carotid arteries. Rats were killed 14 days after injury. (a) Morphometric analysis of rat carotid cross sections. Bars represent mean ± s.e.m. of media, intima and intima/media thickness ratio. (b) Representative haematoxylin-eosin staining of rat carotid arteries cross sections (**P<0.01).

DISCUSSION

In this study, we described a promoter that allows transgene induction in a Ca²⁺ -regulated manner. This promoter, which contains NF6, was as efficient as the strong CMV promoter to induce expression in proliferative VSMC both in vitro and in vivo. It is only 224 bp long allowing insertion of longer transgenes in some viral vectors with small insert size such as adeno-associated viruses. Moreover, unlike the CMV promoter, NF6 was not active in quiescent VSMC. So this promoter seemed a useful tool for expressing a gene of interest only in proliferating, and thus diseased, cells. In agreement with a feedback control of SERCA2a on NFAT, we observed that NF6-Luciferase expression was reduced in cells overexpressing SERCA2a. We then used SERCA2a as a control transgene as we had previously shown that SERCA2a (1) controls VSMC proliferation and neointima formation and (2) controls NFAT activity.^7,11 Overexpressing SERCA2a under the control of the NF6 promoter limited VSMC proliferation to the same extent as expression of SERCA2a under the control of the CMV promoter in vitro and in vivo.⁷

Transgenic mice with cardiac expression of a NFAT-luciferase construct demonstrated that NFAT is involved in pathological but not physiological cardiac hypertrophy.¹⁹ NFAT is known to be activated by an influx of Ca²⁺ through the store-operated channels¹² after sarco/endoplasmic reticulum Ca²⁺ depletion. TRPC3 and TRPC6 were first described as positive regulators of NFAT during cardiac hypertrophy.^20,21 Moreover, SERCA2a overexpression^1,22 or inhibition of STIM1^6,23 or Orai²⁴ prevents Ca²⁺ influx and NFAT activation in cardiomyocytes and in VSMC. SERCA2a overexpression has been proposed as a therapeutic approach for HF²⁵ and vascular proliferative disorders.⁷ More recently the CUPID trial (Calcium Upregulation by Percutaneous Administration of Gene Therapy in Cardiac Disease), a phase 2 trial of intracoronary of SERCA2a gene therapy demonstrated safety and suggested benefit in patients with advanced HF. All these studies have been performed by overexpressing SERCA2a under control of the strong nonregulated CMV promoter, which induces transgene expression in all cardiac and vascular cells infected. The same strategy using a NFAT-responsive promoter would allow transgene expression only in cells in which NFAT is activated. Furthermore, by expressing proteins implicated in intracellular Ca²⁺ regulation (such as SERCA2a, STIM1, Orai1 or TRPC3 and 6) using this NFAT-activated promoter, their expression will be limited to the period necessary to inhibit the activity of NFAT and thus the diseased state. When NFAT activity is back to normal the expression of the transgene will be shut down.

NFAT is activated in many pathophysiological situations: NFAT inhibitors like cyclosporine A have been used for their immunosuppressive effects in autoimmune diseases and after organ transplant. NFAT is also implicated in chronic inflammatory disease like rheumatoid arthritis.²⁶ A role of NFAT proteins in the progression of cancer is also beginning to emerge.¹⁴ Inducing a gene under the control of the NFAT promoter is conceivable in all these disease. Nevertheless, to avoid adverse effects that could occur in other tissues and cell types where Ca²⁺ variations could activate NFAT, the development of tissue and cell-specific delivery vectors is a challenging issue. Several techniques have been used in drug therapy to enhance delivery of the drug to particular cells including the use of antibody- and peptide-based nanoparticles.²⁷ Gene therapy is a new therapeutic approach especially useful for genetic diseases and rare diseases for which there is no other alternative. One of the first clinical successes of gene therapy used an adeno-associated virus in an inherited form of blindness.²⁸ One way to restrict the location of transgene expression is to use adeno-associated virus with a natural tropism for particular tissues. For instance systemic injection of rAAV9 transduced efficiently the heart in rodents.²⁹ In addition, modifications of the capsids have been introduced to enhance the specificity of transduction.^30,31

Cell-specific promoters are a promising approach to enhance tissue specificity of transgene expression in addition to appropriate vectors. Promoters from cardiac myosin heavy chain, myosin light chain or the cardiac troponin T genes are used to specifically target cardiomyocytes, transgelin (SM22a) or smooth muscle actin for smooth muscle cells, VEGF, Tie2 or Flt1 for endothelial cells. Promoters have been developed which function as switches that are controlled under specific cell situations. This is the case of the hypoxia-response element associated with either the minimal simian virus 40 promoter or with a cardiac-specific promoter that becomes active in the heart under ischemic conditions.^32,33 In this study we provide a new promoter, which is Ca²⁺ -sensitive and activated only in proliferative-diseased VSMC.

In conclusion, the NF6 promoter described here is activated in specific situations where the ER is depleted in Ca²⁺. This regulated promoter could therefore be used to develop gene therapy for disease and pathologies in which Ca²⁺ regulation is altered like vascular proliferative diseases, HF or tumor invasion.

MATERIALS AND METHODS

Materials

Thapsigargin and cyclosporine A were from Sigma-Aldrich (Saint Quentin Fallavier, France). The following primary antibodies were used: anti-SERCA2a (A010–20 Badrilla for immunohistochemistry experiments, anti-SERCA2a provided by Dr F Wuytack (University of Leuven, Belgium) for western blot analysis),³⁴ anti-luciferase (sc-32896, Santa Cruz Biotechnology, Heidelberg, Germany), anti-actin (A2668, Sigma-Aldrich).

Generation of the constructs using the NFAT promoter

The synthetic promoter was generated by using tandemly repeated NFAT- responsive elements from the human interleukin 2 gene. These sequences were generated from single-stranded synthetic oligonucleotides and linkers containing restriction sites using a ligation-PCR protocol. Oligonucleotides contained also specific overhang 5'-CG-3' to control orientation during ligation. NFATc forward: 5'-TGGAGGAAAAACTGTTTCATACAGAAGG CG-3'; NFATc reverse: 3'- GCACCTCCTTTTTGACAAAGTATGTCTTCC-5'; oligos linker forward: 5'-AAGCTTGAGCTCCGTCGACCG-3' and reverse: 3'-GCTTCG AACTCGAGGCAGCTG −5' containing the restriction sites for HindIII, Sac1 and Sal1. Single-stranded oligonucleotides (2 μg) were phosphorylated separately in 8 μl reactions (37°C, 90 min) by T4 polynucleotide kinase (Promega, Charbonnieres, France). For annealing, each pair of single- stranded oligonucleotides was combined, heated to 90°C and slowly cooled to room temperature and 2 ml linkers were added to the phosphorylated monomers (monomer:linker ratio = 10:1). Ligation was performed directly to the annealed oligonucleotides. Products of ~600 bp were subjected to DNA extraction and used as a PCR template. PCR products were purified, subjected to enzymatic digestion by Sac1 and resolved on a 2% agarose gel. Extracted DNA fragments were subcloned to pGL3 basic plasmid (Promega) into Sac1 site. The efficacy of different NFAT-multimers promoters was tested using relative luciferase activity. The promoter containing six tandemly repeated NFAT-consensus sites (NF6) was chosen for following cloning. TATA box was inserted by using site-directed mutagenesis (Stratagene, Agilent Technologies, Massy, France) and modified synthetic oligonucleotides: NFAT-TATA forward: 5^'-TGGAGGAAAAACTGTTTCATACAGAAGGCGTCGCGGAGACTCTAGAGGGTAT ATA-3'; NFAT-TATA reverse: 3'-GCACCTCCTTTTTGACAAAGTATGTCTTCCGC AGCGCCTCTGAGATCTCCCATATAT-5^'. Finally, a new vector encoding luciferase under the control of NF6 promoter (NF6-Luciferase vector) was obtained.

Human SERCA2a cDNA³⁵ was inserted in place of the luciferase cDNA to generate the NF6-S2a vector.

Generation of recombinant adenovirus

NF6-Luciferase and NF6-S2a constructs were used to produce adenovirus AdNF6-Luc and AdNF6-S2a (Figure 1a) by using Adeno-X promoter less Expression System 2 (Stratagene, Agilent Technologies) according to manufacturer's instructions. AdCMV-S2a expressed SERCA2a and the green fluorescent protein under the control of two different CMV promoters. Similarly, AdCMV-βGal encoded the beta-galactosidase and the green fluorescent protein under the control of CMV promoters.³⁵

Cell culture, proliferation and promoter-reporter assay (NFAT-luc)

VSMC were isolated from the media of thoracic aortas from male Wistar rats and cultured as described.⁷ Cells were used between passages 2–8 and were infected with adenovirus at 10 PFU per cell. Proliferation was measured by bromo-deoxy uridine incorporation during 48 h from day 2 to day 4 after infection. Bromo-deoxy uridine incorporation was measured using a colorimetric method (Roche Diagnostics GmbH, Mannheim, Germany).

For reporter gene assay, pCMV-Luc and pNF6-Luc plasmids were transfected with lipofectamine 2000 (Life Technologies, Courtaboeuf, France) according to the manufacturer's instructions. Cells were maintained in serum-starved medium for 48 h and stimulated as indicated for 4 h. For reporter gene assay with adenovirus, cells were infected with AdNF6-Luc alone or co-infected with AdCMV-S2a or AdCMV-βGal in presence of serum and cultured 72 h in the same medium. The luciferase activity was measured using Promega's luciferase assay kit (Promega), normalized to total protein, and expressed as fold change of control in relative luciferase units.

Rat carotid artery injury, gene delivery and morphometric analysis

Rat carotid injury and gene transfer were performed as described.⁷ Fourteen days after surgery the animals were killed, the left and right carotid arteries were dissected, flushed with saline, included in cryomatrix (Thermo Fisher Scientific, Illkirch, France) and frozen at −80°C. Hematox- ylin/eosin staining was performed on cross sections. Specimens were measured by using Lucia G computer software (Laboratory Imaging sro, Prague, Czech Republic) on Leica microscope (Leica Microsystems, Nanterre, France). The area of the media and intima layers were measured. The following groups were analysed: (1) injured-non infected (saline); (2) injured + AdNF6-Luc; (3) injured + AdNF6-S2a; and (4) the right carotid arteries were used as noninjured controls.

Confocal microscopy

Immunohistochemistry was performed on acetone-fixed and 0.1%. Triton- PBS-permeabilized sections according to a standard protocol. Proteins were visualized by using secondary antibodies conjugated to Alexa 594 (Life Technologies, Saint-Aubin, France). Slides were examined with a Leica TCS4D confocal scanning laser microscope using a Plan Apochromat × 63 objective (NA 1.40, oil immersion). Green fluorescence was observed with a 505–550 nm band-pass emission filter under 488nm laser illumination. Red fluorescence was observed with a 560-nm long-pass emission filter under 594 nm laser illumination. Pinholes were set at 1.0 Airy unit. Stacks of images were collected every 0.4 μm along the z axis. All settings were kept constant to allow comparison.

Genomic DNA preparation and PCR verification of viral infection

Genomic DNA was extracted from frozen and mounted slices of carotid arteries using QiaAmp DNA microkit (Qiagen, Courtaboeuf, France) according to the manufacturer's instructions for laser-microdissected tissues. Starting from about 10 carotid arteries slices incubated with proteinase K and ATL Buffer for 16 h at 56 °C, DNA was finally resuspended in 20 μl water. Five microliter of genomic DNA from carotid arteries or 1 μl of positive controls were used to perform PCR using Biotaq DNA polymerase (Bioline, Paris, France). Thirty-five cycles of PCR with the following steps were performed: 45 s 95 °C-45s 55 °C-45s 72 °C. Oligonucleotides used were: genomic DNA-f: 5'-TGCATCAGAGGTAA GCGTCTC-3^', genomic DNA-r: 3^'-GCACTCCAAGTTGGTGGACT-5^'; NF6-Luc-f: 5^'-TGGTTTAAACTTAAGCTTGGCATT-3^', NF6-Luc-r: 3^'-GTTCCATCTTCCAGCG GATA-5^'; NF6-S2a-f: 5^'-GCGGAGACTCTAGAGGGTATATAATG-3^', NF6-S2a-r: 3^'-AGCAAGTCTTCAAACTGCTCAA-5^'.

Protein preparation and WB analysis

Total cell lysates were prepared using Cell Extraction Kit (PromoCell GmbH, Heidelberg, Germany) according to manufacturer's instructions. Thirty micrograms of proteins were then loaded on 10% SDS-PAGE gels and blotted on Hybond-C membrane (Amersham Biosciences, GE Healthcare, Munich, Germany). Proteins were visualized by using the ECL + (Amersham Biosciences, Munich, Germany) and Ettan DIGE Imager (GE Healthcare, Munich, Germany).

Statistical analysis

All quantitative data are presented as mean of at least three independent experiments ±s.e.m. Data were analyzed by using StatView software (SAS France, Greégy-sur-Yerres, France). One-way analysis of variance test were performed for multiple comparisons of values. Statistical comparison of two groups was done by an unpaired Student's t-test or Mann-Whitney tests. Differences were considered significant when P<0.05.