Granulocyte-macrophage colony-stimulating factor gene based ...

THE JOURNAL OF GENE MEDICINE RESEARCH ARTICLE J Gene Med 2009; 11: 345–353. Published online 4 February 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/jgm.1298

Granulocyte-macrophage colony-stimulating factor gene based therapy for acute limb ischemia in a mouse model Chester Bittencourt Sacramento1† Vanessa Dionisio Cantagalli1† Mariana Grings1† Leonardo Pinto Carvalho1 Jos´e Carlos Costa Baptista-Silva2 Abram Beutel3 Cassia Toledo Bergamaschi3 Ruy Ribeiro de Campos Junior3 Jane Zveiter de Moraes4 Christina Maeda Takiya5 V´ıvian Yochiko Samoto5 Radovan Borojevic5 Flavia Helena da Silva1,6 Nance Beyer Nardi6 Hans Fernando Dohmann7 Hamilton Silva Junior7 Valderez Bastos Valero1 Sang Won Han1,4 * 1

Interdisciplinary Center for Gene Therapy, Federal University of S˜ ao Paulo, S˜ ao Paulo, Brazil 2 Department of Surgery, Federal University of S˜ ao Paulo, S˜ ao Paulo, Brazil 3 Department of Physiology, Federal University of S˜ ao Paulo, S˜ ao Paulo, Brazil 4 Department of Biophysics, Federal University of S˜ ao Paulo, S˜ ao Paulo, Brazil 5 Department of Histology and Embryology, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil 6 Department of Genetics, Bioscience Institute, Federal University of Rio Grande do Sul, Rio Grande do Sul, Brazil 7 Excellion Biomedical Services, Petropolis, RJ, Brazil *Correspondence to: Sang Won Han, CINTERGEN-UNIFESP, Rua Mirassol, 207 S˜ ao Paulo-SP, 04044-010, Brazil. E-mail: [email protected] † These investigators contributed contributed equally and should be considered as senior authors. Received: 17 August 2008 Revised: 17 December 2008 Accepted: 22 December 2008

Copyright  2009 John Wiley & Sons, Ltd.

Abstract Background Granulocyte-colony-stimulating factor (GM-CSF) is a pleiotropic factor for hematopoiesis that stimulates myeloblasts, monoblasts and mobilization of bone marrow stem cells. Therefore, the GM-CSF gene is a potential candidate for vessel formation and tissue remodeling in the treatment of ischemic diseases. Methods A new mouse limb ischemia was established by surgery and gene transfer was performed by injection of 100 µg of a plasmid carrying GM-CSF. Muscle force and weight, histology, capillary density, circulating stem cells and monocytes were determined after 3–4 weeks. Results More than 60% of nontreated ischemic animals showed gangrene below the heel after 4 weeks, whereas the GM-CSF gene-treated animals showed only darkening of nails or toes. These animals demonstrated a full recovery of the affected muscles in terms of weight, force and muscle fiber structure, but the muscles of nontreated ischemic animals lost approximately 50% weight, 86% force and their regular structure. When the GM-CSF gene was injected into the contralateral limb, only partial loss was observed, demonstrating a distant effect of GM-CSF. The capillary density in the GM-CSF-treated group was 52% higher in relation to the nontreated group. Blood analysis by flow cytometry showed that the GMCSF-treated group had 10–20% higher levels of circulating monocytes and Sca-1+ . Conclusions We conclude that the direct administration of GM-CSF gene in limb ischemia had a strong therapeutic effect because it promoted the recovery of muscle mass, force and structure by mobilizing therapeutic cells and augmenting the number of vessels. Copyright  2009 John Wiley & Sons, Ltd. Keywords angiogenesis; arteriogenesis; gene therapy; GM-CSF; ischemia; vasculogenesis

Introduction Peripheral arterial diseases (PAD) are mainly caused by atherosclerosis, stenosis or thrombus formation. When these conditions appear, there is an increase in vessel resistance that can lead to a reduction in the distal perfusion pressure and blood flow. In the case of critical limb ischemia (CLI), approximately 200 lower limb amputations per million occur each year in the nondiabetic population and 3900 per million in diabetic patients [1,2]. Approximately half of CLI patients can be treated by vascular surgery, but the remainder depend mainly on the velocity of adaptation of the existing collateral vessels, a process known as arteriogenesis, so as to avoid necrosis and, consequently, amputation or death [3].

346

Vessel formation in adult phase can occur in three different ways: arteriogenesis, vasculogenesis and angiogenesis [4]. Several growth factors, cytokines and proteases are required for arteriogenesis, which is the process of remodeling pre-existent vessels during the ischemia. The monocytes, attracted by the presence of tumor necrosis factor-α and monocyte chemoattractant protein-1 at the inflammatory sites, are the main producers of factors for arteriogenesis [5–7]. However, during transendothelial migration, the endothelial fas-ligand binds to the fas-receptor of the monocytes and induces apoptosis to control the excessive production of the mediators of arteriogenesis [8–11]. Because the time and concentration of the factors required to complete arteriogenesis varies in each case of PAD, the action of the activated monocytes may not be enough to resolve the ischemic problem. One way to prolong the life of monocytes is by inhibiting apoptosis by granulocyte-colony-stimulating factor (GMCSF), which is a well-known hematopoietic-stimulator that enhances the survival, proliferation and rate of differentiation of hematopoietic cells [12–14]. The administration of GM-CSF also mobilizes bone marrow stem cells, especially the endothelial progenitor cells, which can promote vasculogenesis even in the adult phase [4]. Therefore, arteriogenesis and vasculogenesis, which are two important processes to formation of vessels during ischemia, can be stimulated simultaneously by GM-CSF. In the present study, we report on the therapeutic effects of the GM-CSF gene transferred into mice with surgery-induced limb ischemia. Physical and physiological parameters, in addition to histology, capillary density and recruitment of bone marrow derived cells, were evaluated.

Materials and methods Vector construction The plasmid vector expressing mGM-CSF (murine GMCSF) was constructed in our laboratory by inserting a sequence of DNA containing cytomegalovirus (CMV) intron 1 with splicing signals between the CMV promoter and polycloning site of pVAX (Invitrogen, Carlsbad, CA, USA) [15]. Therefore, all features of the pVAX vector were maintained including the cloning sites. The mGMCSF gene was inserted between the EcoRI and EcoRV sites, and was named uP-mGM.

Induction of hind limb ischemia and gene transfer in mice All mice used for experiments were male BALB/c mice, aged 10–12 weeks old, and were obtained from the animal house of the Federal University of Sao Paulo (UNIFESP). All experiments were carried out in accordance with the recommendations for the proper care and use of laboratory animals, as recommended by Copyright  2009 John Wiley & Sons, Ltd.

C. B. Sacramento et al.

the Ethic Committee of the UNIFESP (CEP 298/04 and CEP 0531/04). To induce limb ischemia, under anesthesia with an intraperitoneal injection of ketamine (40 mg/kg of animal) and xylazine (10 mg/kg of animal), the deep and superficial femoral arteries were excised from their origin as a branch of the external iliac artery, without damaging the femoralis vein or nervus, and ligated to the point of distal bifurcation into the saphenous and popliteal arteries. These last arteries and the circumflex artery were only ligated. Perfusion of the distal limb was consequently limited to collateral arteries from the internal iliac artery, rendering the distal hind limb severely ischemic. Gene therapy was performed by injecting 100 µg of plasmid vector in 100 µl of phosphate-buffered saline (PBS) at the middle of the thigh soon after the surgery of ischemia with an insulin needle. After the procedure, the animals were maintained under analgesia with daily peritoneal injections of 5 mg/kg Carprofen. Histological analysis and determination of GM-CSF were performed at the end of the experiments, when all animals were subjected to euthanasia by cervical dislocation.

Determination of GM-CSF by enzyme-linked immunosorbent assay (ELISA) For the extraction of protein GM-CSF from the muscle, mice were subjected to euthanasia and the thighs were surgically removed. A piece of muscle from the centre of the thigh measuring approximately 1 cm was isolated for protein extraction [16]. Muscles were frozen at −80 ◦ C overnight, immersed in Tris-HCl buffer (25 mM, pH 7.4) containing 50 mM NaCl, 0.5% Na-deoxycholate, 2% NP-40, 0.2% sodium dodecyl sulphate, 1 mM phenylmethylsulphonyl fluoride, and homogenized with the motorized homogenizer (Ultra-Turrax T8; Ika-Werke, Baden-Wurttemberg, Germany). The homogenized tissues were centrifuged at 4500 g for 10 min at 4 ◦ C (Eppendorf, Westbury, NY, USA) and the supernatants were used for protein determination [17] and ELISA. The samples were kept in an ice bath for all protein extraction procedures. To determine GM-CSF by ELISA, 96-well plates were coated with the rabbit polyclonal anti-murine GM-CSF antibody (1 µg/ml) (Peprotech Inc., Rocky Hill, NJ, USA). After blocking with 1% [w/v bovine serum albumin (BSA)-PBS], samples diluted 1 : 5 (v/v in 0.1% BSAPBS) were added and incubated overnight at 4 ◦ C. After three washings with 0.05% (v/v) Tween 20 in PBS, plates were incubated for 1 h at 37 ◦ C with 1 : 250fold diluted biotin-conjugated rabbit anti-murine GM-CSF antibody (Peprotech Inc.). After washing, the plates were incubated for 30 min at room temperature with 1 : 1000 diluted streptavidin conjugated with horseradish peroxidase (Pierce Biotechnology, Rockford, IL, USA). After thorough washing, the reactions were developed by adding o-phenylen-diamin diluted in 0.05 M citrate buffer (3 mg/ml) and interrupted with 4 N H2 SO4 . J Gene Med 2009; 11: 345–353. DOI: 10.1002/jgm

347

GM-CSF gene therapy for limb ischemia

The absorbance at 492 nm was determined with an ELISA reader (EL 808 Ultra Microplate Reader, Bio-Tek Instruments, Inc., Winooski, VT, USA). Each sample was analysed in triplicate to obtain the mean.

software. At least 50,000 events were collected for frequency determination.

Visual assessment of ischemic limbs

Muscle samples were fixed in 10% formaldehyde, dehydrated and embedded in paraffin. Sections (6 µm) were obtained and stained with hematoxylin and eosin. A descriptive and a semi-quantitative analysis were carried out considering the parameters: degree of muscle regeneration, adipose infiltration, cellular infiltrates, muscle degeneration and atrophy. Parameters were assessed according to a histological score (HS): 0, absence; 1, rare to few foci of lesion; 2, moderate amount; 3, intense amount; 4, abundant. To count the amount of capillaries in muscle tissue other sections were collected in poly L-lysine coated slides and submitted to histochemistry with biotinylated Griffonia (bandeira) (Vector Laboratories, Peterborough, UK). Antibodies were revealed with streptavidin conjugated with horseradish peroxidase (Sigma, St Louis, MO, USA) using diaminobenzydine as the chromogen. Blood vessel density was monitored by using a computerized image analysis of 20 randomly selected fields/animal in sections stained with the antibodies, obtained using a ×40 objective lens. Results of the different groups were expressed as the mean number of capillaries per mm2 .

Three grades were used to measure the degree of limb gangrene for the evaluation of gene therapy procedure: I, nail darkening or no gangrene; II, gangrene in toes; III, gangrene below heel.

Determination of muscle force Mice were anaesthetized with urethane at concentrations of 1.3 g/kg animal weight. The anaesthetized animals were placed on a table and the limb was fixed rigidly to avoid movements. The right gastrocnemius muscle was totally isolated, leaving the vascular connections and muscle origin intact. The distal end of the muscle’s tendon was isolated, cut and attached with a strong suture to a force transducer (iWorx/CB Science, Inc., Dover, NH, USA). The sciatic nerve was cut and the distal portion placed over a bipolar platinum electrode connected to a stimulator (Grass S88 Grass Instruments, Quince, MA, USA). Muscle was covered with paraffin oil and gauze continually moistened with warm saline solution. Muscle function was assessed by measuring the isometric contractile response of the right gastrocnemius muscle. Muscle resting tension was adjusted to obtain the maximal tension response using a length-to-peak force of contraction curve. Supramaximal tetanic tension was achieved by 200-ms trains at 100 pulses/s with square wave pulse duration of 0.2 ms at 60 Hz. The voltage was supramaximal (i.e. 20% higher than that required to elicit a maximal twitch). Each tetanic contraction was separated by a 2-min recovery period to prevent fatigue between contractions. Stimuli were generated by a stimulator connected to a stimulus isolation unit (Grass SIU5; Grass Instruments) with platinum electrodes; the muscle force was recorded and analysed with the PowerLab/800 software (ADInstruments, Bella Vista, NSW, Australia).

Flow cytometry of peripheral blood Blood samples were incubated for 30 min at 4 ◦ C with phycoerythrin- or fluorescein isothiocyanate-conjugated antibodies against murine CD45, as pan-leukocyte marker (clone 30-F11), Sca-1/Ly6A-E (clone D7) and CD117 (clone 2B8) as a pan-progenitor markers (Pharmingen BD, San Diego, CA, USA). After staining, erythrocytes were lysed and samples fixed with BD lysing Solution. The frequency of Sca-1/Ly6A-E/CD117 positive cells among CD45-positive leukocytes was determined using a FACSCANTO cytometer equipped with 488 nm argon laser (Becton Dickinson, Brazil) with the FACSDiva Copyright  2009 John Wiley & Sons, Ltd.

Histology

Statistical analysis All values are expressed as the mean ± SD. The GraphPad Prism, version 5.01 (GraphPad Software Inc., San Diego, CA, USA) was used to perform variance analysis using Bonferroni method for multiple comparisons or the Mann–Whitney test for comparisons of two groups. p < 0.05 was considered statistically significant.

Results Vector uP-mGM and its expression in vivo A sufficient level of therapeutic gene expression in vivo is essential for gene therapy. To achieve an increased transgene expression, the vector uP was constructed by inserting intron 1 and splicing sequences from CMV into the commercially available vector pVAX. Even though the vector uP became larger than pVAX, in vitro expression of GM-CSF by uP was approximately 50% higher than pVAX [15]. Thus, the vector uP was used in all experiments. To establish the best conditions for in vivo transfection, different plasmid concentrations were tested. The injection volume had to be always maintained at 100 µl because it is the largest volume that can be injected in the thigh of a mouse weighing 25 g without leaking. With this condition, there was no significant variation of J Gene Med 2009; 11: 345–353. DOI: 10.1002/jgm

348

Figure 1. Score of the degree of limb gangrene for the evaluation of the gene therapy procedure. Three grades were used to measure the degree of limb gangrene: I, nail darkening or no gangrene; II, gangrene in toes; III, gangrene below heel. Is1w and Is4w represent 1 and 4 weeks after ischemia, respectively; GM1w and GM4w represent 1 and 4 weeks after gene therapy with GM-CSF, respectively, for ischemic animals. Four groups of ten mice were used in this experiment. Non-ischemic and sham-operated animals did not present any sign of necrosis

gene expression among animals injected with 50, 100 and 200 µg of uP-mGM, which produced 1, 1.4 and 1.6 ng of mGM-CSF/mg of protein, respectively, after 72 h of injection. For transfection based on the hydrodynamic force, it is likely that the amount of plasmid has less influence than the volume of liquid [18]. Based on our results of in vivo transfection, for all experiments carried out in the present study, 100 µg of plasmid in 100 µl of PBS was used.


Figure 2. Determination of gastrocnemius muscle force. At the indicated time, the mice were anaesthetized to determine muscle force. Control animals were non-ischemic and sham-operated mice and did not receive gene therapy. The group designated uP-mGM left hindlimb had the uP-mGM vector administrated at the right thigh, whereas ischemia was induced on the left thigh. Five groups of eight animals were used in the experiment. Is, Ischemic animals; Non-Is, non-ischemic animals; w, week; Sham-op, Sham-operated; GM, ischemic animals treated with mGM-CSF. ∗ p < 0.05

Evaluation of the ischemic limb model Visual evaluation of ischemic limbs after 4 weeks showed that most of the animals presented a gangrene sign up to the ankle (Figure 1). At this time, the gastrocnemius muscle force was drastically reduced to 10% in comparison to non-ischemic animals (Figure 2). This result may be explained by the loss of muscle mass (approximately 45%) verified after 4 weeks of observation (Figure 3). Therefore, this model of ischemic limbs provided physical and physiological parameters clearly indicating post-ischemic damage, which were in contrast to that observed in control non-ischemic animals.

Determination of circulating monocytes and Sca-1+ cells by flow cytometry Three days after ischemic surgery, the frequency of circulating mononuclear cells increased to greater than 20% and a similar level was maintained for 1 week, but this value returned to the baseline at day 14 (Figure 4A). The GM-CSF-treated group had a profile similar to that of the ischemic group, but the concentration of monocytes was maintained at a level 15–20% higher than in the ischemic group during 21 days of follow-up. The profile of Sca-1+ cell production was very different Copyright  2009 John Wiley & Sons, Ltd.

Figure 3. Determination of gastrocnemius muscle mass. At the indicated time, the mice were euthanized and gastrocnemius muscles were removed surgically and weighed. Control animals were non-ischemic and sham-operated mice that did not receive gene therapy. The uP-mGM left hindlimb group received gene therapy at the right thigh and ischemia was induced on the left thigh. The same animals as those shown in Figure 2 were used. Is, Ischemic animals; Non-Is, non-ischemic animals; w, week; Sham-op, Sham-operated; GM, ischemic animals treated with mGM-CSF. ∗ p < 0.05

from that of monocytes. During the first week, there was no response against ischemia, even with the GMCSF treatment (Figure 4B). A considerable increase in the J Gene Med 2009; 11: 345–353. DOI: 10.1002/jgm

GM-CSF gene therapy for limb ischemia

Figure 4. Determination of Sca-1+ cells in mouse peripheral blood. The frequency of monocytes among mononuclear cells (A) and of Sca-1+ cells within the population of CD45+ /CD117+ leukocytes (B) was determined by flow cytometry of blood samples withdrawn at the indicated time. When treated and nontreated groups were compared, p = 0.02 for Sca-1+ and p = 0.0002 for monocytes (∗ ). p < 0.05 was observed for all comparisons of ischemic groups (treated and nontreated) with the sham-operated group (!). , animals with limb ischemia; ♦, animals with limb ischemia and treated with uP-mGM; , Sham-operated animals. Three independent experiments were carried out for cell counting. In total, 60 animals were used and distributed in three groups of five animals. ∗ p < 0.05

frequency of Sca-1+ cells was detected on day 14 in the ischemic group, but the GM-CSF-treated group presented at a level higher than 10% of Sca-1+ cells (Figure 4B). One week later, the values decreased significantly.

Evaluation of the therapeutic use of uP-mGM for ischemic limbs To evaluate the therapeutic effects in our study, the following parameters were used: muscle mass, muscle force, muscle fiber regeneration and capillary density, in addition to visual evaluation. In the first week, the visual evaluation of gene therapytreated ischemic animals did not show any significant improvement compared to the group of ischemic animals that were not treated (Figure 1). However, the ischemic animals that had not been treated were in a worse condition compared to the gene therapy-treated ischemic animals after 4 weeks. In the GM-CSF treated group, there was no animal with degree III (gangrene below heel), but animals with degree III were predominant in the nontreated group. One of the main consequences of limb ischemia is muscle atrophy, which can be observed in Figure 3, where ischemic animals lost 18% of muscle mass in Copyright  2009 John Wiley & Sons, Ltd.

349

Figure 5. Determination of capillary density. At the indicated time, the mice were euthanized and gastrocnemius muscles were removed for immunohistochemistry, as decribed in the Materials and methods. Capillary density was determined by counting numbers of vessels per field at ×40 magnification using a light microscope. More than 20 fields were counted for each group. The same animals as those shown in Figure 4 were used, but only three representative animals of each group are considered here. Is-7 and Is-21 represent ischemic animals after 7 and 21 days, respectively. GM-7 and GM-21 represent ischemic animals treated with GM-CSF after 7 and 21 days, respectively. Non-Is, non-ischemic normal animals; Sham-Op-21, Sham-operated group after 21 days. ∗ p < 0.05

the first week and 45% of muscle mass in the fourth week. The progressive loss of muscle mass was totally reversed in the fourth week with GM-CSF gene therapy and, consequently, the muscle force was fully recovered (Figure 2). To investigate whether GM-CSF gene therapy could stimulate a physiological effect only locally or, alternatively, whether the effect could be extended to a distant site, one group of animals with right hind limb ischemia received the uP-mGM vector injected in the contralateral non-ischemic limb. Four weeks after therapy, the recovery of muscle force reached 47% and the muscle mass recuperation was 75% compared to non-ischemic animals. To verify whether GM-CSF production in these animals was sufficient to reach the ischemic limb, the production of GM-CSF was monitored in the sera of treated animals by ELISA on days 0, 3, 5, 7 and 9. GM-CSF was not detected in the GM-CSF-treated animal sera (data not shown), despite the sensitivity (0.05 ng/ml) of the test. However, the local production of GM-CSF after 48 h of plasmid injection was 1.6 ng/mg protein from the thigh. Another parameter analysed was the capillary density, which was determined by labeling endothelial cells with biotinylated Griffonia. One week after surgery, nontreated ischemic animals (Is-7) spontaneously presented an increase of 46% in their local vessel density compared to non-ischemic animals (non-Is), and a slightly higher augmentation (Is-7 versus Is-21) was observed thereafter (Figure 5). On the other hand, ischemic animals treated with uP-mGM (GM-7) showed an augmentation that was approximately 10% higher than that observed in nontreated ischemic animals (Is-7) in the first week and became 52% better than the counterpart group (GM-21 versus Is-21) after 3 weeks of therapy. J Gene Med 2009; 11: 345–353. DOI: 10.1002/jgm

350


Figure 6. Histological analysis. Mice were euthanized at the indicated time and their gastrocnemii muscles were removed for staining with hematoxylin and eosin. (A) Ischemic group after 1 week; (B) ischemic group treated with uP-mGM after 1 week; (C) ischemic group after 4 weeks; (D) ischemic group treated with uP-mGM after 4 weeks. A semi-quantitative histological assessment is presented in Table 1. The same animals as those shown in Figure 1 were used, but only three representative animals per group are considered here

Table 1. Histological score of muscle tissues

Muscle fiber regeneration Inflammation Adipose tissue infiltration Muscle fiber degeneration

1w Is

1w + GM

4w Is

4w + GM

3 3 2 2

2 2 1 1

2 2 2 1

1 2 1 1

0, absence; 1, rare to few foci of lesion; 2, moderate amount; 3, intense amount; 4, abundant. Is, Ischemic animals; w, week; GM, GM-CSF-treated.

We next examined the histopathology of muscle tissues semi-quantitatively using the HS (see Materials and methods). In the 1-week ischemic limbs, some degenerated cells still could be observed occupying approximately 10% of the regenerated zones (Figure 6 and Table 1). Replacing necrotic areas of the tissue were numerous small, basophilic multinucleated myocytes (approximately 35% of the surface; HS3) interspersed with mononuclear inflammatory cells (approximately 30% of the regenerating zone; HS3) and with a moderate adipocyte infiltration (approximately 20% of the regenerating zone; HS2), all indicative of the Copyright  2009 John Wiley & Sons, Ltd.

physiological regeneration process [19]. After 4 weeks, all these histological inflammatory parameters diminished. Rare muscle fibers were calcified, as a consequence of ischemic injury (1% of the regenerated zone; HS1). Regenerating fibers were scarce (1% of the regenerated zone; HS1), but inflammatory cells still persisted at a moderate scale (10% of the regenerated zone; HS2) as well as adipose cells (