|Home | About | Journals | Submit | Contact Us | Français|
Bone marrow mononuclear cell (MNC) therapy is a promising treatment for peripheral artery disease (PAD). This study aims to provide insight into cellular kinetics using molecular imaging following different transplantation methods.
MNCs were isolated from F6 transgenic mice (FVB background) that express firefly luciferase (Fluc) and green fluorescence protein (GFP). Male FVB and C57Bl6 mice (n=50) underwent femoral artery ligation and were randomized into 4 groups receiving: (1) single intramuscular (i.m.) injection of 2×106 MNC; (2) four weekly i.m. injections of 5×105 MNC; (3) 2×106 MNCs intravenously (i.v.); and (4) PBS. Cellular kinetics, measured by in vivo bioluminescence imaging (BLI), revealed near-complete donor cell death 4 weeks after i.m. transplantation. Following i.v. transplantation, BLI monitored cells homed in on the injured area in the limb, as well as to the liver, spleen, and bone marrow. Ex vivo BLI showed presence of MNCs in the scar tissue and adductor muscle. However, no significant effects on neovascularisation were observed as monitored by Laser-Doppler-Perfusion-Imaging and histology.
This is one of the first studies to assess kinetics of transplanted MNCs in PAD using in vivo molecular imaging. MNC survival is short lived and MNCs do not significantly stimulate perfusion in this model.
Peripheral artery disease (PAD) currently affects over 27 million North Americans and Europeans and is associated with impaired leg function and decreased quality of life, leading to significant morbidity and mortality (1,2). Despite a variety of treatment options, including percutaneous transluminal angioplasty, stenting, and bypass surgery, a cluster of patients remain unresponsive to therapy, having no other option than amputation in one third of patients within this group (3).
Recently, stem cell therapy has emerged from bench to bedside as a treatment for end-stage PAD, potentially offering a last option for revascularization of the ischemic limb. While results from pre-clinical experiments using bone marrow-derived mononuclear cells (MNC) appear hopeful, outcomes from clinical studies are divergent (4), raising questions about transplanted stem cell behavior and mechanisms of action involved in the benefits of stem cell transplantation. As to cell behavior, two major issues might be the lack of donor cell survival after introduction into ischemic target tissue and the absence of cell homing to the injured area following systemic administration (5). Three mechanisms are believed to be of importance for the beneficial effects seen after cell transplantation: donor cell death might hamper a lasting scaffolding effect, impair transplanted cell-derived neovascularization, and limit the secretion of protective paracrine factors by the transplanted cells.
To study stem cell behavior, one must be able to monitor cell location, migration, proliferation, and death. Recent proof-of-principle studies have demonstrated the ability to track cell fate following cardiac injections (5,6). In the present study, we monitor by molecular imaging the presence of MNC after transplantation in mice in which hindlimb ischemia is induced. These experiments are designed to answer critical questions regarding cell survival and homing patterns to the affected leg, as well as functional consequences of different transplantation strategies.
Animal study protocols were approved by the Animal Research Committees from both institutions (Stanford University and Leiden University Medical Center). The donor group for imaging experiments consisted of 8-week old male F6 mice, which were bred on FVB background and ubiquitously express green fluorescent protein (GFP) and firefly luciferase (Fluc) reporter genes driven by a β-actin promoter as previously described (7). Recipient animals for these experiments consisted of syngeneic, male FVB mice (10–12 weeks old, Jackson Laboratories). Additionally, C57Bl6 mice were used (10–12 weeks old, Jackson Laboratories).
The long bones were explanted, washed, and flushed with PBS using a 25-gauge needle to collect bone marrow. After passing through a 70 μm strainer, the isolate was centrifuged at 1200 rpm for 5 minutes, washed, and resuspended into PBS. To acquire the MNC fraction, the bone marrow isolate was centrifuged for 40 minutes at 1600 rpm using a 14 mL tube with 3 mL Ficoll-Paque Premium (GE Healthcare, Piscataway, NJ, USA) gradient and 4 mL cell/saline suspension, as described (5). MNCs were prepared freshly before application.
Cells were incubated in 2% FBS/PBS at 4°C for 30 min with 1 μL of APC-conjugated anti-CD31 (eBioscience), anti-CD45 (BD Biosciences), and anti-Gr-1 (BD Biosciences), or PE-conjugated anti-CD34 (eBioscience), anti-CD11b (BD Biosciences), anti-Flk-1, anti-Sca-1 (both eBioscience), and anti-NK1.1 (BD Biosciences), and processed through a FACSCalibur system (BD, San Jose, CA, USA) according to the manufacturer’s protocol.
BLI was performed on the IVIS 200 (Xenogen, Alameda, CA, USA) system. For in vitro characterization of luciferase expression, cells were suspended in different quantities in 1 mL PBS. Following administration of 10 μL (43.5 μg/mL) D-Luciferin, peak signals (photons/s/cm2/sr) from a fixed region of interest (ROI) were evaluated and plotted versus cell number. For in vivo experiments, recipient mice were anesthetized with isoflurane, shaved, and placed in the imaging chamber. After acquisition of a baseline image, mice were intraperitoneally injected with D-Luciferin (400 mg/kg body weight). Mice were imaged on post-injection days 1, 3, 6, 9, 13, 20, and 27. Peak signals (photons/s/cm2/sr) from a fixed region of interest (ROI) were evaluated as described (7). For ex vivo experiments, animals were euthanized immediately following the moment when peak signals were achieved. The organs were rapidly explanted and imaged according to the protocol described above.
Before surgery, mice were anesthetized with either isoflurane (2%) or an intraperitoneal injection of a combination of midazolam (5mg/kg, Roche), medetomidine (0.5mg/kg, Orion) and fentanyl (0.05 mg/kg, Janssen). The effect of MNC-injections was tested in 2 models of hindlimb ischemia: a single electro-coagulation model and a double electro-coagulation model (8). For unilateral single electro-coagulation of the femoral artery, ischemia was created by an electro-coagulation of the femoral artery just proximally to the superficial epigastric artery. Moreover, a double electro-coagulation was performed to create a larger therapeutic window for assessment of possible arteriogenesis (8). For this model, first an electro-coagulation of the common iliac artery was performed, followed by an electro-coagulation of the femoral artery. Subsequently, the skin was closed using 6-0 silk sutures. One day post operation, 40 μL cell/PBS injections were given into the adductor muscle, or 100 μL cell/PBS solution into the tail vein. To compare the efficacy of a single versus repeated injection with cells, FVB mice were randomized into 3 groups: (1) single i.m. injection of 2×106 MNCs, (2) four repeated i.m. injections of 5×105 MNCs, (3) i.m. injection of phosphate buffered saline (PBS) injection as control (n=10 per group). The reason for using FVB mice in the first experiment was to establish a clinically resembling model of autologous cell transplantation as our F6 transgenic donor mice, used for in vivo BLI, were bred on FVB background. In addition, to investigate the functional effects of intravenous MNC injection in another animal model, C57Bl6 mice were randomized into 2 groups: (1) single i.v. injection of 2×106 MNCs and (2) i.v. injection of PBS (n=10 per group).
Neovascularization was monitored by measurements of perfusion of the mouse hindlimbs at the level of the paws and was performed before, directly after, and weekly over a period of 4 weeks after the surgical procedure with Laser Doppler Perfusion Imaging (LDPI) (Moor Instruments) (8). To control for temperature variability during measurements, all animals were kept in a double-glassed jar filled with 37°C water, keeping environment temperature at a constant level during the LDPI-measurements. Each animal served as its own control. Eventually, perfusion was expressed as a ratio of the flow in the left (ischemic) to right (non-ischemic) paw. Before each LDPI-measurement, mice were anesthetized with an intraperitoneal injection of Midazolam (5mg/kg, Roche) and Medetomidine (0.5mg/kg, Orion).
To further explore short-term effect of cell therapy on paw perfusion, we performed an apoptosis specific ELISA on the affected gastrocnemius muscles. The selected muscle was explanted, digested using a stator-rotator homogenizer, and lysed. ELISA was performed directly on the supernatant to quantify histone-associated DNA fragments (mono- and oligonucleosomes), marking early apoptotic cells (Cell Death Detection ELISA, Roche Applied Science, Indianapolis, IN).
To validate in vivo BLI findings, the bone marrow was collected as described above and assayed for GFP expression by flow cytometry as described above.
Immunohistochemistry was performed to visualize smooth muscle cell layers of collateral arteries with an antibody against smooth muscle actin. Furthermore, with an antibody against GFP, GFP+ MNCs were traced in the ischemic skeletal muscle. Five μm-thick paraffin-embedded sections of skeletal muscle fixed with 4% formaldehyde were used. These were re-hydrated and endogenous peroxidase activity was blocked for 20 minutes in methanol containing 0.3% hydrogen peroxide. Skeletal muscle slides were stained with monoclonal anti-α smooth muscle actin (mouse anti-human, DAKO, dilution 1:800). Antigen retrieval was not necessary and sections were incubated overnight with primary antibody. Rabbit anti-mouse HRP (DAKO, dilution 1:300) was used as a secondary antibody. For the negative control, an isotype control instead of the primary antibody was used. The signal was detected using NovaRED substrate kit (Vector laboratories) and sections were counterstained with hematoxylin. Stainings were quantified from randomly photographed sections using image analysis (ImageJ). For tracing of GFP+ MNCs, adductor muscle slides were incubated with anti-GFP (rabbit anti-mouse, Invitrogen, dilution 1:4000) without antigen retrieval. After overnight incubation, labelling was followed by a biotin-conjugated secondary antibody (donkey anti-rabbit, dilution 1:300). As a positive control, a slide of GFP+ cardiac muscle tissue was used.
Statistics were calculated using SPSS 16.0 (SPSS Inc., Chicago, IL, USA). Descriptive statistics included mean and standard error. Comparisons between groups were performed using an independent t-test. P-values were considered statistically significant if P<0.05.
Following isolation and Ficoll selection, the MNC population showed subpopulations of CD31+, CD34+, CD45+, and Sca-1+, but Flk-1− cells, representing hematopoietic but not early endothelial progenitors cells. Moreover, strong expression of CD11b, Gr-1, and NK 1.1, representative of macrophages, granulocytes, and natural killer cells, indicated the largely inflammatory character of this donor cell population (Figure 1a).
For tracing the cells in an in vivo fashion by bioluminescence imaging (BLI), we first set out to characterize the expression of the reporter gene Fluc in vitro. As suggested in Figure 1b, luciferase expression intensity increased concordant to increasing cell number. When maximum expression per well was plotted versus the amount of cells, a robust correlation was observed with an r2 value equaling 0.97 (Figure 1c). Thus, BLI signal intensity is closely representative of the amount of living cells carrying the luciferase reporter gene. Moreover, the activity of GFP in the donor-specific Fluc-GFP double-fusion reporter gene construct was confirmed by in vitro fluorescence microscopy (Figure 1d).
Following single transplantation of 2×106 MNCs, a short-term post-transplant increase in BLI signal from 6.6±1.5×104 at day 1 to 8.9±2.5×104 p/s/cm2/sr at day 3 (P=NS) indicated an increase in cells in the adductor muscle region during the first days. Thereafter, however, cell death resulted in a rapid decrease in signal intensity to background level after 4 weeks (Figure 2). A similar cumulative dose of MNCs, divided in 4 weekly doses of 5×105 MNCs, led to a relatively stable presence of donor cells. No statistically significant difference after 4 weeks (5.1±0.8×103 in single vs 5.7±0.3×103 in multiple dose group; P=NS) was detected.
Skeletal muscles of mice treated with a single injection of 2×106 MNCs and weekly injections of 5×105 MNCs were harvested 28 days after hindlimb ischemia induction. Dismal numbers of GFP+ MNCs were only observed in the adductor muscle of mice that received weekly injections of MNCs (Figure 3). These GFP+ MNCs surrounded vessels within the muscle tissue, suggesting a role for these cells in neovascularization. On the contrary, GFP+ MNCs were not observed in the adductor muscles of mice receiving a single injection of MNCs.
Single femoral artery electro-coagulation resulted in a significant decrease in paw perfusion when compared to the healthy right hindlimb (P<0.001 for all groups, Figure 4). Three days following MNC transplantation, a trend towards better flow recovery with increased MNC number was observed, as the ischemic/non-ischemic paw perfusion ratios in the single 2×106 MNC and four weekly 5×105 MNC injection groups were 0.75±0.07 and 0.67±0.07, respectively, as compared to 0.62±0.07 in the PBS group (P=NS). However, no significant differences were observed as robust recovery of paw perfusion was seen in all groups by week 4.
Figure 5 shows no differences in collateral density and collateral size in the post-ischemic adductor muscle after a single 2×106 MNC injection, four repeated 5×105 MNC injections and PBS control, further confirming the lack of functional improvement in LDPI results. As shown in Supplemental Figure 1, treatment with both single 2×106 MNCs and four weekly 5×105 MNCs led to significantly (P=0.03 and P=0.02, respectively) decreased amount of fragmented DNA (mirroring apoptosis) as compared to the PBS group, which had an almost 3-fold higher expression than its healthy contralateral counterparts.
The initial BLI signals on day 0 (1 hour after intravenous transplantation of MNCs) equaled background, confirming that the cells were spread throughout the circulatory system (Figure 6). No signs of retention in the pulmonary capillaries were observed, in contrast with previous studies using larger size cell types such as mesenchymal stem cells (9). Over time, however, signal intensity increased due to homing and migration to the injured area. In addition, signals arose from the bone marrow, spleen, and liver suggest homing patterns that mimick endogenous myelomonocytic pathways (10).
To validate and further specify the observed in vivo findings after systemic MNC-injection, organs were procured immediately following euthanization. As shown in Supplemental Figure 2a, BLI following dissection of the skin showed in situ signals from liver, spleen, and the long bones similar to in vivo results. However, the signals that were previously observed from the injured area in vivo were now largely concentrated in the subcutaneous fat pad as well as in the femoral bone. Indeed, when the different tissues were explanted, low signal was seen from the adductor muscle, while equally strong signals were observed from the scarred skin, the subcutaneous fat pad, and the bone marrow in the femoral bone. Thus, the ex vivo imaging results confirmed the robust in vivo signals from liver and spleen. Moreover, the presence of GFP+ donor MNCs in the bone marrow was validated with flow cytometry (Supplemental Figure 2b). Taken together, these experiments showed that BLI is a reliable method to monitor MNC trafficking in an in vivo fashion, while homing to the injured area was not limited to the adductor muscle, but also more natural biological niches such as marrow, liver, and spleen.
For the previous experiments, FVB mice were used in order to establish a clinically resembling autologous model of cell transplantation in which the donor group consisted of FVB mice expressing GFP-Fluc. However, in FVB mice a robust endogenous restoration of paw perfusion was observed. Therefore, to investigate the functional effects of intravenous MNC-injection, C57Bl6 mice were used in combination with a double electro-coagulation to ensure a larger therapeutic window (8). After double electro-coagulation of both the femoral and iliac artery, the ischemic/non-ischemic paw perfusion ratio decreased dramatically from an overall mean of 1.04 pre-operatively to 0.04 postoperatively (P<0.0001). Intravenous injection of MNCs was incapable of restoring paw perfusion in a significant improved matter, with a ratio of 0.60 in the MNC group compared to 0.57 in the PBS group (P=NS) at 4 weeks postoperatively (Figure 7).
This is one of the first studies to evaluate post-transplant MNC behavior in a murine model of PAD using in vivo molecular imaging techniques. The major findings can be summarized as follows: (1) MNC survival following a single intramuscular injection is short-lived; (2) repeated MNC injections do not provide a significantly prolonged cell survival; (3) homing of MNCs following intravenous injection is not limited to the area of injury; and (4) neither intramuscular nor intravenous injection of MNCs results in an improved blood flow recovery after hindlimb ischemia induction.
The clinical relevance of these findings is significant. Since the pioneering work of Tateishi-Yuyama and colleagues (11), over 25 clinical trials have been registered on www.clinicaltrials.gov, using either intramuscular or systemic injections into the ischemic leg. Although the findings from this first study raised early hopes, so far these results have not been confirmed by large randomized clinical trials. The initial thought behind the use of progenitor cells in regenerative medicine was that it could truly regenerate the damaged tissue by forming new blood vessels (12), skeletal muscle (13), or even myocardium (14). However, as the true regenerative capacity has been questioned (15), and considering the dismal survival capacity of MNCs and other progenitor cells in this and other studies (5), a more plausible explanation for a possible beneficial effect would be the excretion of protective cytokines as suggested before (16). Indeed, it has recently been shown that a more profound angiogenic response can be achieved in ischemic muscle by transplanting progenitor cells overexpressing both vascular endothelial growth factor (VEGF) and stromal-derived factor-1 (SDF-1) (17). Alternatively, and in order to achieve true regeneration, one could switch to more specialized cell types rather than whole MNCs. In this respect, it has recently been shown that embryonic stem cell-derived endothelial cells can improve perfusion due to the favorable effect of engraftment and biological activity (18). Thus, in the future, it might be a feasible approach to use a set of growth factors by gene therapy, to increase survival of specialized cells (beyond just not at the least embryonic stem cells or induced pluripotent stem cells), or to use a combination of these two.
Previous studies have assessed MNC function and mechanism following transplantation into the ischemic leg largely by using post-mortem histological techniques (19). However, this requires euthanizing the animal, thereby increasing inter-animal variance and hampering longitudinal studies of the same subject. Moreover, the search for scant donor cells on histological slides from all organs is extremely difficult and time consuming. As a consequence, these techniques are less suitable for studying the kinetics of cells through the body over time. In contrast, in this study we were able to take advantage of an advanced molecular imaging platform, based on the double-fusion reporter construct carrying Fluc and GFP, to yield valuable insight into longitudinal cell fate. By doing so, we were able to track the spatiotemporal kinetics of MNC homing, retention, and survival in a murine model of PAD.
Interestingly, we observed a relatively limited cell survival after intramuscular injection in the adductor muscle. After a short-term increase in BLI signal up to day 3, a rapid decrease in BLI signal intensity compared to background after four weeks was observed. The limited cell survival was confirmed by the immunohistochemical staining, against GFP+ cells. No MNCs were detected 28 days after single 2×106 MNC injection with an anti-GFP immunohistochemical staining whereas one week after the fourth transplantation of 5×105 MNCs, a dismal proportion of these cells could be found. These GFP+ MNCs were present near blood vessels, suggesting a role in neovascularization, or indicating these cells prefer the adjacency of oxygenated blood. The dismal survival in the adductor muscle is interesting because femoral artery ligation results in less profound ischemia in the adductor muscle as compared to the gastrocnemic muscle. This suggests that even in a normoxic niche, MNCs require more biologically attractive environments to be capable of robust survival. This once again stresses the need for development of cell survival augmenting approaches such as scaffolds or transduction of cells with pro-survival factors.
Results from this study show that, following systemic injection, MNCs migrated extensively to the bone marrow, spleen, and liver. This pattern indicates MNCs traveled to their natural biological niches as all of these organs play a role in intra- and extramedullary hematopoiesis. Confirming this observation, our BLI findings are concordant with previous leukocyte scans showing retention in the liver and spleen (20). For future experiments, it will be important to improve homing to the ischemic muscles which may increase arteriogenic response as measured by LDPI. This could be realized in two ways: 1) improving the attractiveness of the target environment with, for example, the MNC mobilizer SDF-1 (21); or 2) manipulating the cells to be more specifically guided. In this respect, it might be a better approach to isolate a subset of the mononuclear fraction such as the CD14+ expressing cells that are expected to play a more active role in the restorative process after ischemia (22).
Several limitations of this study should be mentioned. First, GFP+ MNCs were used in the hindlimb ischemia mouse model in order to study post-ischemic neovascularization. It has been suggested that GFP can elicit an immune response (23), which may have influenced collateral artery formation since this is an inflammatory driven process. However, the lack of any effect on post-ischemic perfusion recovery or collateral artery formation at the tissue level makes the interference of immunogenic GFP+ cells on arteriogenesis unlikely. Secondly, the present report studied MNC behavior in an acute model of hindlimb ischemia. Clearly, this model is not truly reflective of PAD, which is a chronic disease. Unfortunately, a superior model with more chronically occluded arteries is not available yet in mice.
Taken together, to our knowledge this is one of the first studies to monitor the kinetics of MNCs in PAD in an in vivo fashion using molecular imaging techniques. Results from this study highlight the caution that should be taken when interpreting results from experimental as well as clinical studies. The poor survival and homing patterns warrant further research that should aim for better retention and increased biological activity of the cells in the injured area. By doing so, cell therapy might develop into a valuable option for treating end-stage PAD. In the meantime, molecular imaging should be fully exploited to provide more insight into the mechanisms of action for cell therapy.
This study was supported by BWF CAMS, NIH RC1HL099117, and R01EB009689 (JCW). Koen van der Bogt was supported by the Michaël van Vloten fund. The authors gratefully acknowledge the support of the TeRM Smart Mix Program of the Netherlands Ministry of Economic Affairs and the Netherlands Ministry of Education, Culture and Science for Alwine Hellingman.