Stem cell recruitment into the heart is determined by a concentration gradient of stromal-derived factor 1 (SDF-1) from bone marrow to peripheral blood and from blood to injured myocardium. However, this gradient is decreased in chronic myocardial infarction (MI). This study evaluated the effect of cell therapy using bone marrow stromal cells (BMSCs) on an SDF-1 gradient in post-infarction rabbits.
Methods and results
Myocardial infarction was induced in male New Zealand white rabbits (2.5–3 kg) by ligation of the left anterior descending coronary artery. Two months later, the rabbits were randomized to either saline or BMSC (2 × 106 autologous BMSCs injected into the left ventricular cavity) treatment. Four weeks after therapy, the SDF-1 gradients from bone marrow to blood and from blood to myocardium increased in the BMSC group compared with the saline group. This was accompanied by an increase in cells positive for CD34, CD117, and STRO-1 in the myocardium, resulting in more capillary density, better cardiac function, and a decrease in infarct size.
Generation of an SDF-1 gradient towards the heart is a novel effect of BMSC-based cell therapy. This effect facilitates stem cell recruitment into remodelled myocardium and supports improvement in cardiac function.
Continual loss of cardiomyocytes after myocardial infarction (MI) increases the risk of heart failure., Since cardiomyocyte differentiation is one of the characteristics of bone marrow stromal cells (BMSCs), these cells have been used to repair damaged myocardium in pre-clinical studies and recent clinical trials. However, the treatment effect in humans is modest, in contrast to that in animals, where the effect is marked and significant. Thus, it is necessary to know the mechanism underlying successful cell therapy to resolve the efficacy discrepancy between species. The efficacy of cell therapy is known to be associated with the recruitment of stem cells and chemokines to the injured organ., One critical chemokine is stromal-derived factor 1 (SDF-1), the first chemokine found to link stem cell recruitment and post-MI cardiac repair.
Stromal-derived factor 1 is crucial for directing stem cell migration along a low to high SDF-1 gradient. In the bone marrow, SDF-1 is produced by BMSCs to anchor haematopoietic stem cells. When SDF-1 is degraded by proteases, a gradient between bone marrow and peripheral blood is established. This gradient promotes stem cell migration into the systemic circulation. The subsequent fate of circulating stem cells is determined by another SDF-1 gradient from the peripheral blood to the damaged tissue. For instance, immediately after MI, not only does the concentration of SDF-1 in the myocardium increase, but the number of stem cells in the infarcted myocardium also increases. However, this over-expression of SDF-1 in the myocardium is only maintained for 1 week and the concentration of myocardial SDF-1 is decreased in ischaemic cardiomyopathy. Thus, in chronic MI, an SDF-1 gradient that is unfavourable to stem cell recruitment from blood to remodelled myocardium may occur.
Taken together, an SDF-1 gradient from bone marrow to blood and from blood to the myocardium may play a critical role in stem cell recruitment into the heart. Importantly, BMSCs are cells that can secrete high concentrations of SDF-1. Thus, the beneficial effects of BMSC-based cell therapy could be associated with the change in the SDF-1 gradient. In this study, using rabbits as an experimental model of chronic MI, we determined whether BMSC transplantation could recruit more stem cells to the heart in order to improve left ventricular (LV) remodelling, and we explored the changes in SDF-1 levels in the bone marrow, peripheral blood, and myocardium before and after cell therapy.
Animals and protocols
Male New Zealand white rabbits (2.5–3 kg) underwent ligation of the left anterior descending coronary artery to induce MI, as described previously. Two months after induction of MI, the chest wall was reopened and the rabbits were randomized to receive either 2 × 106 autologous BMSCs in 1 ml of phosphate-buffered saline (BMSC group) or saline only (saline group) by direct injection into the LV cavity. Sham operated animals that underwent the same surgical procedure but without actual ligation of the coronary artery were used as controls (sham group). These three groups underwent bone marrow sampling, peripheral blood sampling, and cardiac magnetic resonance imaging (MRI) before ligation, 2 months after ligation, and 1 month after treatment. At the end of the study, the rabbits were euthanized and the left ventricle of each was removed for analysis. Observers blinded to the experimental treatment given to each animal performed the analyses. The procedures used conformed to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996).
Bone marrow stromal cell preparation
Bone marrow stromal cells were isolated by combining gradient density centrifugation with plastic adherence. One month after coronary artery ligation, bone marrow (10 mL) was harvested from rabbit femoral bone and mixed with an equal volume of normal saline. The concentrated leukocyte band, or buffy coat, was separated from red cells by using Ficoll–Paque Plus density gradient centrifugation (density 1.077 g/cm3, Amersham Pharmacia Biotech, Piscataway, NJ, USA) at 700 g for 20 min at 4°C. The mononuclear cells collected at the interface were resuspended in Iscove's modified Eagle's medium (Gibco Invitrogen) containing 20% foetal bovine serum (Biological Industry, Kibbutz Beit Haemek, Israel), 100 U/mL of penicillin, 100 µg/mL of streptomycin, and 0.25 µg/mL of amphotericin B (Biological Industry), plated at a density of 2 × 105 cells per 25 cm2 flask (Corning, NY) and incubated at 37°C in 5% CO2 in air. After 3 days, the medium and non-adherent cells were discarded, after which the medium was changed every 2 days.
Preliminary studies determined that 10 ml of rabbit bone marrow contained 2.2 ± 0.5 × 108 mononuclear cells, in which there were 2.0 ± 0.4 × 103 BMSCs assessed by colony forming unit fibroblast (n = 5). On Day 3 of culture, 70% of BMSCs were positively stained for STRO-1, thus confirming their stromal cell character. On Day 14, BMSCs spontaneously expressed cardiac specific troponin-I, myosin heavy chain, and transcription factor GATA-4. Moreover, BMSCs did not express the surface markers of endothelial cells or haematopoietic stem cells such as CD31, CD34, and CD117 in immunocytochemical analysis.
Four weeks after culture, the BMSCs were removed from the culture disk using 0.25% trypsin, incubated with culture medium to neutralize the trypsin, collected by centrifugation at 900 g for 5 min at 4°C, and then suspended in PBS at a concentration of 2 × 106 cells/mL for autologous transplantation.
Cardiac magnetic resonance imaging
In order to measure the LV volume, mass, function, and infarct size, animals were examined using a 3 T MRI unit (Trio; Siemens, Erlangen, Germany) with an eight-channel cardiac phased array coil for signal reception. An ECG-gated turbo fast low angle shot (TurboFLASH) cine pulse sequence was acquired in two long-axis and five to seven short-axis views. After cine imaging was completed, an intravenous bolus dose of 0.20 mmol/kg of gadodiamide (Nycomed Imaging AS, Oslo, Norway) was administered, and late gadolinium enhancement images acquired 10 min later using an inversion-recovery prepared segmented TurboFLASH sequence. Left ventricular volume, mass, and ejection fraction were assessed using cine MRI and an automated boundary detection algorithm. To assess infarct size, we quantified the late gadolinium enhancement using a signal intensity threshold criterion of >2 SD above the mean signal intensity of the remote myocardium, and expressed it as a percentage of the total LV mass. Imaging analysis was performed using the Mathematica software package (Wolfram Research, Inc., IL, USA) and Matlab (MathWorks, Inc., Natick, MA, USA). The intra-observer variability for the LV volume results was 3%.
Haemodynamic variables were measured in rabbits anaesthetized with 10 mg/kg of xylazine and 50 mg/kg of ketamine given intramuscularly at the end of the study. A polyethylene Millar catheter was inserted into the right carotid artery and connected to a transducer (Model SPR-407, Miller Instruments, Houston, TX) to measure the LV systolic and diastolic pressures as the mean of five consecutive pressure cycles. The maximal rates of the LV pressure rise (+dP/dt) and decrease (−dP/dt) were recorded and measured on a personal computer with data analysis software (PowerLab data acquisition system, AD Instruments, Castle Hill, NSW, Australia).
After the haemodynamic study, the rabbits were sacrificed. The heart was weighed and the left ventricle sectioned into 3–4 transverse slices parallel to the atrioventricular ring along the cardiac long axis. The section at the papillary level was embedded in optimal cutting temperature (OCT) embedding compound and frozen at −70°C for immunological staining. The other sections were divided into remote, border, and infarcted myocardium and frozen at −70°C for protein and nucleic acid analysis.
In order to investigate the spatial distribution of stem cells and the capillary density, we performed immunohistochemical staining for STRO-1 (1:1000; R&D Systems, Minneapolis, MN), CD117, CD34, and CD31 (1:1000; DakoCytomation, Glostrup, Denmark) on the ventricular myocardium using 5 µm thick OCT embedded sections. Immunostaining was performed using an immunoperoxidase technique (3,3′-Diaminobenzidine chromogen with haematoxylin counterstain) and an automated immunostainer (Benchmark™, Ventana Medical Systems, Tucson, AZ, USA). For quantification, five sections from within the remote or infarcted myocardium of each animal were analysed for positive cells by ImageJ software (National Institutes of Health image, http://rsb.info.nih.gov/ij/).
Local stromal-derived factor 1 concentrations
Stromal-derived factor 1 was measured in platelet depleted samples, which had been centrifuged at 11 000 g for 10 min at 4°C, using high sensitivity enzyme linked immunosorbent assay kits (Bender MedSystems, R&D). Blood was obtained by cardiac aspiration, and bone marrow was collected by aspiration of the right femur. The plasma and marrow concentrations of SDF-1 were expressed in picograms per millimetre. Myocardial tissues from the infarcted, border, and remote zones were obtained for measurement of local SDF-1 concentrations. The myocardium was minced and homogenized on ice using a Polytron homogenizer for 60 s in 10 volumes of lysis buffer [25 mmol/L Tris, 1% Triton X-100, 0.5 mmol/L EDTA, 150 mmol/L NaCl, 10 mmol/L NaF, and a protease inhibitor cocktail (Roche Mannheim)]. The lysate was immediately centrifuged at 11 000 g for 10 min at 4°C, and the supernatant was used for SDF-1 analysis. The myocardial concentration of SDF-1 was normalized to tissue weight and expressed in picograms per millimetre, with a tissue density of 1.0 g/mL.
Protocol II: labelling of bone marrow stromal cells and polymerase chain reaction detection of enhanced green fluorescent protein DNA in the heart
In order to investigate the location of the transplanted BMSCs, 2 weeks after cell culture, 106 BMSCs were transfected with 10 µg of pEGFP-N1 plasmid encoding enhanced green fluorescent protein (EGFP) (Clontech, Palo Alto, USA) using Transfast transfection reagent (Promega, Madison, WI, USA) according to the manufacturer's instructions. Transformed cells were selected by culturing with 200 µg/mL of the neomycin analogue G418 (Clontech) for 14 days. The transfected cells were expanded for 7 days to obtain more than 2 million cells, of which 60% expressed GFP. Then they were removed from the culture disk and suspended in PBS at a concentration of 2 × 106 cells/mL for transplantation.
One month after EGFP–BMSCs transplantation, PCR analysis of EGFP gene expression was performed on scar, border zone, and remote zone myocardial tissue. DNA extraction was performed using a PureGene DNA isolation kit (Gentra Systems, Minneapolis, MN, USA) according to the manufacturer's instructions. The total DNA including genomic DNA and plasmid DNA was then used as a template for PCR. The primers for the EGFP gene were 5′-CGA CAA CCA CTA CCT GAG CA-3′ (forward) and 5′-AAA ACC TCC CAC ACC TCC C-3′ (reverse) with a product size 212 bp. The reaction conditions for the 35 cycles of the amplification step were programmed on a computer.
The Shapiro–Wilk test for normal distribution was performed on the data and it was demonstrated that the data were normally distributed. The data could thus be presented as the mean ± SD. Comparisons between the three groups were performed using ANOVA with Bonferroni's correction for multiple comparisons. Within each group, comparisons between baseline and follow-up results were performed using a paired Student's t-test. P < 0.05 was considered significant. Statistical analyses were performed using the software package SPSS, version 12.02 (SPSS, Chicago, IL, USA).
Coronary ligation and mortality
A total of 79 rabbits underwent surgery, of which 59 were subjected to MI and 20 to sham surgery. An additional 30 rabbits were used to study local SDF-1 before ligation. Of the rabbits with MI, six died within 24 h of the first operation and three died during the second open chest operation (19% post-operative mortality). This left 50 rabbits with MI for (i) cardiac MRI evaluation and local SDF-1 change (40 rabbits) and (ii) EGFP–BMSCs tracking (10 rabbits). These infarcted rabbits were randomly assigned 1:1 to the saline- or BMSC-treated groups, with 25 in each group. Between the saline- and BMSC-treated groups, there were no significant differences in terms of body weight, cardiac structure, and function, or infarct size 2 months after coronary ligation (Table1). All rabbits in the saline and BMSC groups survived until sacrificed 2 or 3 months post-MI.
Table 1. Comparison of LV remodelling and function in the sham or ligated and saline or BMSC-treated groups 2 months after coronary artery ligation (2 months) and 1 month after therapy (3 months)
Sham (n = 10)
Infarction treated with
Saline (n = 10)
BMSC (n = 10)
aData are presented as mean ± SD.
*P < 0.05 compared with the sham controls
†P < 0.05 compared with the saline group
#P < 0.05 compared with the corresponding data at 2 months.
Cell therapy favours stem cell recruitment into the heart
To track the injected cells, the cultured BMSCs of five rabbits were labelled with EGFP before delivery into LV cavity (Figure1). After BMSC administration, PCR analysis for EGFP DNA fragment revealed a positive response in heart tissue treated with BMSC for 1 h (1/1), 1 day (2/2), and 4 weeks (2/2), but not in saline-treated heart tissue (0/5) or sham controls (0/3). The injected cells were detected 4 weeks later by fluorescence microscopy in BMSC-treated heart tissue but not in saline-treated heart tissue (data not shown). Immunohistochemical staining showed that these cells were STRO-1 positive, located around blood vessels, and in both remote and infarcted myocardium (data not shown).
Treatment with BMSCs was associated with much greater numbers of cells positive for CD34 and CD117 than in animals treated with saline (Figure2). When quantified, the numbers of Stro-1-positive, CD34-positive, or CD117-positive cells were three to four times greater in animals that received BMSCs than in those given saline. The number of stem cells in the remote myocardium was increased up to the level in sham controls.
One month after treatment, capillary density in the remote myocardium was decreased by 77% in the saline-treated infarcted rabbits compared to sham controls (Figure2). In contrast, along with the significantly greater number of stem cells recruited to the myocardium in the animals that received BMSCs, we found a greater vascular density within the infarct zone in BMSC-treated animals than in those given saline. The capillary density in the remote myocardium was increased up to the level in sham controls.
Cell therapy attenuates left ventricular remodelling
At 2 and 3 months after ligation, saline-treated infarcted rabbits had a significantly increased total heart weight, LV mass index, and LV end-diastolic pressure compared to the sham controls (Table1). Left ventricular ejection fraction and pressure changes at the maximal systolic and diastolic phases were significantly decreased. The wall of the infarcted myocardium was significantly less thick. The infarct size at 2 months after ligation was 16 ± 5%, with no change in the follow-up MRI at 3 months (Figure3).
In contrast, BMSC-treated infarcted rabbits showed a significant decrease in LV mass index compared to the corresponding data at 2 months and the saline-treated group. Left ventricular ejection fraction and LV pressure changes at the maximal systolic and diastolic phases were significantly increased. The infarcted myocardium trended towards increasing wall thickness (P = 0.1) and the infarct size was significantly decreased by 34% after cell therapy.
Before coronary ligation the concentration of SDF-1 in bone marrow was higher than that in the myocardium, and both of them were higher than that in peripheral blood (Figure4). Two months after MI, SDF-1 levels in both bone marrow and peripheral blood tended to be increased, whereas SDF-1 levels in the whole myocardium significantly decreased with the highest level in the remote zone and the lowest in the infarct zone. These SDF-1 changes led to an unchanged SDF-1 gradient from bone marrow to peripheral blood but a decreased gradient from peripheral blood to the myocardium.
One month after therapy, BMSC transplantation significantly decreased the SDF-1 concentration in the bone marrow by 18% but increased the concentration in the peripheral blood by 12% and that in the myocardium by 75%. Thus, there were increases of the SDF-1 gradients from bone marrow to peripheral blood and from peripheral blood to myocardium after cell therapy. In contrast, no interval change in the SDF-1 gradient was seen in either the sham controls or the saline-treated group.
The main findings of our study are that BMSC-based cell therapy increased the SDF-1 gradient from bone marrow to peripheral blood and from peripheral blood to the myocardium, BMSC-based cell therapy increased the number of stem cells in the remodelled myocardium, and BMSC-based cell therapy significantly improved LV remodelling with respect to infarct size, capillary density, and cardiac function. These results suggest that reversal of the SDF-1 gradient towards the myocardium is a novel effect of cell therapy, which acts to recruit more stem cells for cardiac repair.
Interestingly, we observed that SDF-1 levels in both bone marrow and peripheral blood tended to be increased in rabbits with chronic MI. One possible explanation for the increase of marrow SDF-1 is that persistent coronary ligation increases myocardial apoptosis, leading to an increase in DNA damage which in turn causes a significant increase in SDF-1 in the bone marrow. On the other hand, in line with previous reports, we found that myocardial SDF-1 levels decreased in chronic MI. A cellular loss after MI could partly explain the decrease in myocardial SDF-1 in the infarcted heart because the majority of heart cells constitutively expresses SDF-1. Thus, our data illustrate that an unfavourable SDF-1 gradient for stem cell mobilization from bone marrow to the heart exists in chronic MI.
Despite the unfavourable SDF-1 gradient in chronic MI, we found that EGFP-labelled BMSCs were retained and detected in both the remote and infarcted myocardium at 1 day and 4 weeks after LV intracavitary injection. Indeed, SDF-1 is crucial for directing stem cell migration. However, adhesion of circulating BMSCs in the heart appears to be an endothelium dependent process and is enhanced by TNF-alpha and IL-1beta. In chronic MI, the levels of these cytokines are increased in the non-infarcted remote myocardium. Thus, infused BMSCs can end up in the myocardium, especially in the non-infarcted myocardium as shown in Figure1. These findings support the feasibility of incorporating cell therapy in the treatment of chronic MI.
In our present study using a rabbit model of chronic MI, the levels of SDF-1 in all parts of the hearts of rabbits receiving BMSCs were higher than those in rabbits receiving saline treatment. Similarly, in a rat model of acute MI, a study by Tang et al. also demonstrated that SDF-1 increases in the BMSC-treated rat hearts compared to media-treated hearts 3 weeks after MI. This increase in myocardial SDF-1 is not an unexpected finding; in our study, the transplanted BMSCs actually engrafted into the myocardium, and BMSCs can produce high levels of SDF-1. In addition, as shown in the present study, preservation or regeneration of cardiomyocytes and an increase in capillary density due to cell therapy may also contribute to the increase in myocardial SDF-1 levels. On the other hand, after cell therapy, the SDF-1 level in the bone marrow is decreased, whereas the SDF-1 level in peripheral blood is elevated, creating an increasing SDF-1 gradient from bone marrow towards the peripheral blood. A possible explanation for the decrease of SDF-1 levels in bone marrow may be that BMSC transplantation reduces myocardial apoptosis and thus decreases the response of the bone marrow to DNA damage and the over-expression of SDF-1. Moreover, before haematopoietic stem cells mobilize from the bone marrow they are divided and differentiated into neutrophils. When excessive stem cells mobilize, active neutrophils accumulate leading to an accumulation of neutrophil proteases able to directly cleave SDF-1 in the bone marrow.
The goal of cell therapy in chronic MI is to optimize LV remodelling and regenerate myocardial structures. The ability of cardiac MRI to assess the extent of scar tissue in the myocardium with high spatial resolution is one of the major strengths of the present study. We found that along with the three- to four-fold increase in the number of stem cells, the infarct scar size decreased by 34% at 1 month after cell therapy. This observation supports the notion that transplanted BMSCs or recruited stem cells participate in the myocardial regeneration or cardiomyocyte preservation in infarcted rabbits. Furthermore, our results demonstrate that myocardial SDF-1 levels increase after cell therapy. SDF-1 has been shown to promote survival of BMSCs and cardiomyocytes. Thus, the increase in myocardial SDF-1 might also play a role in improving LV remodelling.
The next question is whether these findings in rabbits are applicable to humans. Based on our study, it is conceivable that cell therapy may be more effective in improving LV function if the cell source is BMSCs and if an adequate number of BMSCs is given. Compared to haematopoietic stem cells, BMSCs are a better source for cell therapy because they are more easily expanded in culture and produce more SDF-1. To date, in most clinical studies, investigators were concerned about the risk of contamination occurring during the in vitro expansion step and used unselected bone marrow cells or bone marrow mononuclear cells as the source of cell therapy. A recent meta-analysis of these studies showed an increase in mean LVEF of 4%. In contrast, animal studies have mainly used BMSCs and have shown an increase in mean LVEF of 8% in MI models in various species with cell doses of 0.06–4 × 106 cells/kg body weight., Bone marrow mononuclear cells contain an extremely low number of BMSCs (1/104–1/105). Thus, even though clinical investigators transplanted a total mononuclear cell number of up to 108, they actually delivered less than 104 BMSCs (or 170 cells /kg). In our study with infarcted rabbits, a delivery of 2 × 106 BMSCs (0.7 × 106 cells/kg) into rabbit was enough to change the SDF-1 gradient and improve cardiac remodelling. Considering the difference in BMSC amount between rabbits and humans, we speculate that a dose of 106 BMSCs/kg may be sufficient for humans. Supporting this notion, Chen et al. administered high doses of BMSCs (108 cells/kg) in patients with recent MI. At 6 months, there was an absolute increase in LVEF of 18% in the BMSC group. Further studies measuring cell dose-responses would allow us to quantitatively compare the outcomes and suggest the optimal dose for human studies.
In summary, this study demonstrates that BMSC-based cell therapy generates an SDF-1 gradient towards the heart, concurrently recruit more stem cells to the heart, and improves LV remodelling. Our results suggest that this novel effect of cell therapy could occur in post-infarction patients if adequate numbers of BMSCs were transplanted.
The authors would like to thank Su-Chun Huang for assistance with the MRI data and Che-Hui Chen, Ming-Zhou Wu, Mai-Jun Lai, and Bo-Zu Pan for assistance with the biochemical analysis.
Conflict of interest: none declared.
National Science Council of the Republic of China (95-2314-B002-004).