Acceleration of the Healing Process and Myocardial Regeneration May Be Important as a Mechanism of Improvement of Cardiac Function and Remodeling by Postinfarction Granulocyte Colony–Stimulating Factor Treatment
Background— We investigated whether the improvement of cardiac function and remodeling after myocardial infarction (MI) by granulocyte colony–stimulating factor (G-CSF) relates to acceleration of the healing process, in addition to myocardial regeneration.
Methods and Results— In a 30-minute coronary occlusion and reperfusion rabbit model, saline (S) or 10 μg · kg−1 · d−1 of human recombinant G-CSF (G) was injected subcutaneously from 1 to 5 days after MI. Smaller left ventricular (LV) dimension, increased LV ejection fraction, and thicker infarct-LV wall were seen in G at 3 months after MI. At 2, 7, and 14 days and 3 months after MI, necrotic tissue areas were 14.2±1.5/13.4±1.1, 0.4±0.1/1.8±0.5*, 0/0, and 0/0 mm2 · slice−1 · kg−1, granulation areas 0/0, 4.0±0.7/8.5±1.0*, 3.9±0.8/5.7±0.7,* and 0/0 mm2 · slice−1 · kg−1, and scar areas 0/0, 0/0, 0/0, and 4.2±0.5/7.9±0.9* mm2 · slice−1 · kg−1 in G and S, respectively (*P<0.05, G versus S). Clear increases of macrophages and of matrix metalloproteinases (MMP) 1 and 9 were seen in G at 7 days after MI. This suggests that G accelerates absorption of necrotic tissues via increase of macrophages and reduces granulation and scar tissues via expression of MMPs. Meanwhile, surviving myocardial tissue areas within the risk areas were significantly increased in G despite there being no difference in LV weight, LV wall area, or cardiomyocyte size between G and S. Confocal microscopy revealed significant increases of cardiomyocytes with positive 3,3,3′,3′-tetramethylindocarbocyanine perchlorate and positive troponin I in G, suggesting enhanced myocardial regeneration by G.
Conclusions— The acceleration of the healing process and myocardial regeneration may play an important role for the beneficial effect of post-MI G-CSF treatment.
Received September 2, 2003; de novo received December 5, 2003; accepted February 20, 2004.
Granulocyte colony–stimulating factor (G-CSF), which can mobilize multipotential progenitor cells of bone marrow (BM) into peripheral blood, may improve post–myocardial infarction (MI) left ventricular (LV) remodeling and function. At present, the main mechanism is believed to be transdifferentiation of BM progenitor cells into the cell lineages of the heart, including cardiomyocytes, endothelial cells, etc, in the MI tissues.1,2 However, recent studies have suggested that the transdifferentiation of c-Kit–positive BM cells into cardiomyocytes is controversial3 and the number of the transdifferentiated cardiomyocytes from BM stem cells may be too low to explain the improvement of cardiac remodeling and function.4
The healing process after MI begins from absorption of necrotic tissues (acute stage: within several days after MI), moves into granulation with numerous myofibroblasts, rich microvessels, and collagen (subacute stage: 1 to 3 weeks after MI), and then forms scar tissues consisting primarily of collagen, with rare vessels via apoptosis of granulation cells (chronic stage: >1 month).5 However, to the best of our knowledge, the precise quantitative studies on the dynamic changes of the above-described tissue factors are rare. Orlic et al6 reported that scar tissue areas were reduced after G-CSF treatment, although the mechanism and the role were not examined. In the field of dermatology, it has been established that G-CSF enhances the healing process of various types of skin wounds via expression of various cytokines, such as the matrix metalloproteinase (MMP) family.7–9 We hypothesized that G-CSF may also accelerate the healing process of MI wounds and that the acceleration may play an important role in the beneficial effects. Thus, the purpose of the present study was to define whether post-MI G-CSF treatment modifies the healing process via expression of MMPs, in addition to regeneration of myocardial tissues.
In this study, all rabbits received humane care in accordance with the Guide for the Care and Use of Laboratory Animals, published by the US National Institutes of Health (NIH publication 8523, revised 1985). The study protocol was approved by the Ethical Committee of Gifu University School of Medicine, Gifu, Japan.
The standard methods of 30-minute coronary arterial occlusion and reperfusion in male Japanese White rabbits weighing 1.9 to 2.2 kg were performed as previously reported.10 Briefly, under anesthesia and mechanical ventilation with room air, a left thoracotomy was performed, and 4-0 silk string was placed beneath the large coronary arterial branch coursing down the middle of the anterolateral surface of the left ventricle. Then, the rabbits were killed by an overdose of pentobarbital after heparinization (500 U/kg).
The subjects were 120 rabbits with 30-minute ischemia and 2-day, 7-day, 14-day, or 3-month reperfusion as described above. Saline at 0.5 mL in the saline group or 10 μg · kg−1 · d−1 of recombinant human G-CSF (Lenograstin, Chugai Pharmaceutical Co, Ltd, Tokyo, Japan) in the G-CSF group starting 24 hours after infarction once per day for 5 days was administered subcutaneously in the 7-day groups (n=15 each in the saline and G-CSF groups), the 14-day groups (n=15 in each), and the 3-month groups (n=15 in each), but that in the 2-day groups (n=15 in each) was administered only once at 24 hours after infarction, when each rabbit was enrolled in the saline or G-CSF group by lot.
Echocardiography (SSD2000, Aloka Co Ltd) and the measurement of arterial blood pressure were performed before and 14 days after MI in the 14-day groups (n=15 in each of the saline and G-CSF groups) and before and 3 months after MI in the 3-month groups (n=15 in each). The measurements were performed by 2 persons blinded to treatment.
Blood samples (0.3 mL each) were taken from an ear vein before and 7 days after MI in the 7-day groups (n=15 each in the saline and G-CSF groups) and before and 14 days after MI in the 14-day groups (n=15 in each) for peripheral blood cell counts and hemograms.
The excised hearts of a total of 120 rabbits were mounted on a Langendorff apparatus, and Evans blue dye at 4°C was injected for 1 minute from the aorta after reocclusion of the coronary branch by a silk string for the measurement of risk areas. Then, the whole heart was perfused with 10% buffered formalin (4°C) for 2 minutes. The LV was weighed and sectioned into ≈7 transverse slices parallel to the atrioventricular ring. Each slice was fixed with 10% buffered formalin for 4 hours, embedded in paraffin, and sectioned with a microtome (4 μm thick). These sections were stained with hematoxylin-eosin and Sirius red. Under light microscopy, the risk area without blue dye, nonrisk area with blue dye, necrotic areas, granulation areas, and scar areas were clearly demarcated on the above-described stained preparations.
On the transversely sliced preparations with infarction in each heart, LV wall area, risk area, necrotic areas, granulation areas, scar areas, infarct areas, surviving areas in the risk areas, and collagen areas with positive Sirius red were calculated by use of an image analyzer connected to a light microscope (Luzex-F, Nireco) and were expressed as mm2/slice/body weight (kg) corrected by mean risk area of each group for precise comparison. These were performed by 2 persons blinded to treatment.
By use of an indirect immunoperoxidase method, immunohistochemical stainings were performed using monoclonal mouse anti-troponin I antibody (Chemicon International, Inc) at 1:10, monoclonal mouse anti-human CD31, endothelial cell antibody (Dako) at 1:100, monoclonal mouse anti-human α-smooth muscle actin (Dako smooth muscle actin, 1A4) at 1:250, monoclonal mouse anti-macrophage antibody (Dako RAM11) at 1:100 and monoclonal mouse anti-human MMP1 antibody (Daiichi Fine Chemical Co Ltd, F-67) at 1:400, each of which cross-reacts with rabbit tissues. Morphometric analyses were performed by 2 persons blinded to treatment.
In 14 rabbits, BM (≈10 mL) was aspirated from the right and left iliac crests 2 days before ischemia-reperfusion. To evaluate the incorporation of BM cells into the myocardium, BM mononuclear cells were labeled with fluorescent carbocyanine 1,1′-dioctadecyl-1-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI).11 The DiI-labeled autologous BM mononuclear cells (≈1×107 cells) were returned into the BM space in the right and left iliac crests of each rabbit. Two days later, 30-minute ischemia and reperfusion was performed. Then, the saline group (n=7) and the G-CSF group (n=7) were made up by the same method as shown above and killed 14 days after MI. In addition, 7 other rabbits, as a negative control in which non–DiI-labeled BM cells were returned to BM of iliac crests, were killed 14 days after MI.
The hearts excised 14 days after MI were placed in iced PBS at <4°C immersion immediately after the animals were killed. The tissues (≈3×3×2 mm each) obtained from the risk area, including MI, and from the nonrisk area (3 tissues each) of each heart were embedded in OCT compound (Miles Scientific) and snap-frozen in liquid nitrogen. The OCT compound tissues were sectioned at 4-μm thickness with a cryostat for immunohistochemical analysis. In addition, the BMs of iliac crests were examined immunohistochemically.
Immunohistochemical staining was performed with Hoechst 33342 for nuclear staining, in addition to those detailed above. These were observed with confocal microscopy (LSM510 NLO, Zeiss), which can simultaneously analyze the relation among 3 different fluorescences and phase-contrast illustration. Morphometric analyses were performed by 2 persons blinded to treatment.
Other saline (n=7) and G-CSF (n=7) groups with 30-minute ischemia and 7-day reperfusion and a sham group (n=7) were prepared for the measurement of MMP-1 and 9. The excised hearts were placed in iced PBS at <4°C immersion immediately after the animals were killed. Tissues of ≈200 mg each obtained from the center of the risk area and from the nonrisk area of LV wall were snap-frozen in liquid nitrogen.
For the measurement of MMP-1, a collagenase, ≈50 mg of the above-described frozen tissues obtained from each heart of the sham, saline, and G-CSF groups, was homogenized in lysis buffer and centrifuged at 10 000g at 4°C for 10 minutes. MMP-1 was measured by Western blot analysis using anti-human MMP-1 mouse monoclonal antibody (Daiichi Fine Chemical Co, F-67, clone No. 41-1E5). The signals were quantified by densitometry. The measurement of MMP-9, a gelatinase, in the remaining samples of the above-described frozen tissues was performed by zymography using the recommended methods of the Gelatinzymo electrophoresis kit (Yagai Research Center).
All values are presented as mean±SEM. The differences between the saline and G-CSF groups were assessed by 2-way repeated-measures ANOVA. Differences at P<0.05 were considered statistically significant.
All rabbits in the 3-month saline and G-CSF groups enrolled 24 hours after MI survived during 3 months of experimentation.
Echocardiography and Blood Pressure
As shown in Figure 1, echocardiography showed a significant decrease in the LV end-systolic and LV end-diastolic dimensions and significant increases in the LV ejection fraction and fractional shortening in the G-CSF group compared with those of the saline group at 14 days and 3 months after MI. The ratio of the anterior LV wall thickness with infarction to the posterior wall thickness without infarction was greater in the G-CSF group than in the saline group. These suggest the improvements of LV remodeling and function by G-CSF. There was no significant difference in heart rate or blood pressure between the 2 groups (data not shown).
Peripheral Blood Cell Counts
White blood cells, granulocytes, and monocytes increased from 8800±588, 4021±278, and 357±91 before MI to 19 537±2641, 12 076±2663, and 1351±196/μL at 7 days after MI in the G-CSF group and were restored to 10 363±764, 5106±575, and 449±115/μL at 14 days after MI, respectively. There was no significant change in lymphocytes, red blood cells, or thrombocytes throughout the experiment. The saline group showed no significant changes.
At 2 days after MI, large necrotic tissue areas were surrounded by numerous acute inflammatory cell infiltrations consisting of neutrophils, lymphocytes, and macrophages with positive RAM 11 in both the saline and G-CSF groups. However, the extent and density were clearly greater in the G-CSF group than in the saline group (Figure 2).
At 7 days after MI, necrotic tissue areas became smaller and acute infiltrated inflammatory cells were markedly reduced compared with the 2-day groups. The MI areas were to a large extent replaced by granulation in both the saline and G-CSF groups. At 14 days after MI, necrotic tissues were completely absorbed, and granulation tissues alone were observed in the saline and G-CSF groups. The densities of the myofibroblasts with positive α-smooth muscle actin and vessels with positive CD31 were clearly increased in the G-CSF group compared with those of the saline groups (Figure 3). The number of macrophages with positive RAM 11 was increased in the saline groups according to the duration of 2, 7, and 14 days (Figure 3). Meanwhile, it was similar in the G-CSF groups among the 3 durations. Thus, the number of macrophages was significantly greater in the G-CSF group than the saline group at 2 days after MI, similar at 7 days after MI, and lower in the G-CSF group than in the saline group at 14 days after MI (Figure 3).
At 3 months after MI, the granulation tissues were replaced by scar in each of the saline and G-CSF groups. The number of macrophages became small in each of the G-CSF and saline groups (Figure 3). However, it was significantly lower in the G-CSF group than the saline group.
Risk Area, Necrotic Area, Granulation Tissue Area, Scar Tissue Area, Collagen Area, Surviving Area, and Size of Cardiomyocytes
Among the 2-day, 7-day, 14-day, and 3-month groups, there were no significant differences of LV weights, LV areas, and nonrisk areas in any of the G-CSF and saline groups (Figure 4). The risk areas were significantly reduced in each of the G-CSF and saline groups at 14 days and 3 months after MI, compared with those at 2 days after MI, although the risk areas at 2 days after MI were similar between the G-CSF and saline groups (Figure 4).
The MI areas at 2 days after MI (necrotic tissue areas) were similar between the G-CSF and saline groups. However, they were reduced slightly in the saline groups and markedly in the G-CSF groups at 7 days (necrotic areas+granulation areas), 14 days (granulation areas), and 3 months (scar areas) after MI (Figure 4). The differences between the G-CSF and saline groups were significant. The necrotic areas, which were similar between the G-CSF and saline groups at 2 days after MI, were reduced moderately in the saline group and markedly in the G-CSF group at 7 days after MI (Figure 4). The difference between the 2 groups was significant. The granulation areas at 7 and 14 days after MI were significantly smaller in the G-CSF groups than the saline groups (Figure 4). In addition, the scar areas at 3 months after MI were significantly smaller in the G-CSF group than the saline group (Figure 4). Collagen areas with positive Sirius red staining were significantly smaller in the G-CSF groups than the saline groups at 14 days and 3 months after MI (Figure 5).
The surviving areas of the saline groups within the risk areas were similar at 2 days, 7 days, 14 days, and 3 months after MI (Figure 4). However, those of the G-CSF groups were significantly larger at 14 days and 3 months after MI compared with that at 2 days after MI. The difference between the G-CSF and saline groups was significant at 14 days and 3 months after MI.
The transverse size of cardiomyocytes in the border zone between the surviving and infarct areas was significantly greater than that of the nonrisk areas in the G-CSF and saline groups even 2 days after MI (Figure 5). However, the sizes were similar between the saline and G-CSF groups and among 2 days, 7 days, 14 days, and 3 months after MI (Figure 5).
Confocal microscopy revealed the DiI-labeled cells with positive troponin I, a specific marker of cardiomyocytes (Figure 6); CD31, a marker of endothelial cells; or α-smooth muscle actin in the myocardial tissues obtained from the risk area but not in the tissues from the nonrisk area in the saline and G-CSF groups 14 days after MI (Figure 7). The DiI-labeled cells with positive α-smooth muscle actin were seen in the small vessels, indicating smooth muscle cells, and in the extravascular areas, indicating myofibroblasts (Figure 7). The percentages of DiI-labeled cells in troponin I–positive cells were significantly increased in the G-CSF group (0.13±0.03% versus 0.05±0.02% of the saline group). The percentages of DiI-labeled cells in CD31-positive cells and α-smooth muscle actin–positive cells were significantly increased in the G-CSF group (6.9±1.7 and 5.4±1.6% versus 3.3±1.1 and 1.7±0.7% of the saline group, respectively).
In the BM of iliac crests, DiI-positive cells with nuclei showed a scattered distribution (Figure 7), suggesting the reconstruction of the injected BM cells.
MMP 1 was significantly increased in the risk area of the saline and G-CSF groups and in the nonrisk area of the G-CSF group compared with that of the sham group (Figure 8). It was greatest in the risk area of the G-CSF group. MMP 9 was significantly increased in the risk area and nonrisk area of the G-CSF group.
Acceleration of the Postinfarction Healing Process by G-CSF
The present study revealed that necrotic tissue areas as acute MI size were similar in the G-CSF and saline groups 2 days after MI but were smaller in the G-CSF group 7 days after MI. At 2 days after MI, the numbers of infiltrated neutrophils and macrophages in the acute MI areas, which have strong phagocytosis, were definitely increased in the G-CSF group. Therefore, the prompt absorption of necrotic tissues by G-CSF may be related to the enhanced mobilization of neutrophils and macrophages from BM.
Granulation, scar, and collagen areas were smaller in the G-CSF groups than the saline groups. MMP 1, a collagenase, and MMP 9, a gelatinase, were overexpressed in the G-CSF groups. This is similar to previous findings that G-CSF enhances the MMP family in cancer and blood cells.12,13 Therefore, the reducing effect of collagen by G-CSF may be related to overexpression of the MMP family. Many studies showed that G-CSF improves the healing process of skin wounds such as doxorubicin-induced skin necrosis and burn injury of animal models and infected ulcers of the foot in patients with diabetes mellitus.7–9 Improvement of immunocompromised states by G-CSF has been suggested.8 This is supported by the lower level of macrophages at the subacute and chronic stages in the present study. Thus, post-MI G-CSF treatment accelerates the healing process of MI wounds.
Regeneration of Myocardial Tissues by G-CSF
At present, progenitor cells of cardiomyocytes, in addition to those of endothelial cells and smooth muscle cells, may be present in the myocardium itself, as in the BM. To define whether MI alone and G-CSF can mobilize BM-derived progenitor cells from BM into the heart, BM mononuclear cells labeled with DiI were returned into BM of the iliac crests. DiI-positive and troponin I–, CD31-, or α-smooth muscle actin–positive cells, suggesting the transdifferentiated cardiomyocytes, endothelial cells, or smooth muscle cells (or myofibroblasts), respectively, were seen in the risk areas of the saline and G-CSF groups. The percentages were increased by G-CSF treatment. Thus, MI itself may induce regeneration of myocardial cells via mobilization of BM stem cells, and G-CSF may enhance the process.
However, several recent in vivo studies reported the presence of fusion between cardiomyocytes and BM progenitor cells.14 In the present study, we cannot deny its possibility. In addition, the number of regenerated cells mobilized from the BM of the whole body is unknown because of the methodological limitations of the present study. However, surviving myocardial tissue areas within the risk areas in the G-CSF groups were increased 14 days and 3 months after MI. The transverse size of cardiomyocytes within the risk areas was similar between the G-CSF and saline groups. Death of cardiomyocytes is determined within several hours after reperfusion. In the present study, the first G-CSF injection was performed 24 hours after reperfusion, and acute infarct size at 2 days after reperfusion was similar between the G-CSF and saline groups. Therefore, it is considered that cell fusion caused by G-CSF therapy may occur between the BM cells and surviving cardiomyocytes in the risk areas, but it could not rescue the cardiomyocyte death in the present study. This suggests that an increase of surviving myocardial tissue areas by G-CSF is a result of regeneration of myocardial tissues, including cardiomyocytes. A part of the cells in the regenerated myocardial tissues may originate from BM progenitor cells. There is a possibility that hematopoietic stem cells are involved in the progenitor cells that were responsible for myocardial regeneration, because G-CSF generally mobilizes hematopoietic stem cells.
Mechanism of Improvement of LV Remodeling and Function by G-CSF
First, the enhanced myocardial tissue regeneration would contribute to the beneficial effects as detailed previously.6 Second, rapid absorption of necrotic tissues relating to the higher level of neutrophils and macrophages at the acute stage and improvement of chronic inflammation suggested by the lower level of macrophages at the subacute and chronic stages may contribute to the beneficial effects. Third, production of fibrosis after MI prevents structural fragility. Previous studies reported that a MMP family was increased in the postinfarction heart failure models with permanent occlusion and large infarction and that the inhibitors beneficially affected cardiac remodeling and function.15,16 Thus, it is suggested that an increase in MMP has an aggravating effect on heart failure via collagen degradation. However, it is well known that the volume of reactive granulation and/or scar tissues after skin injury after burn, surgery, etc, frequently becomes excessive, and this is called hypertrophic scarring. The excessive extent of fibrosis without contractility or relaxation would accelerate cardiac remodeling and decrease cardiac function, as seen in ischemic or idiopathic dilated cardiomyopathy. In such cases, an increase in the MMP family may be one of the protective mechanisms via proteolysis of excessive collagen. In fact, this concept is supported by findings of several previous studies: an inhibition of MMP caused cardiac failure,17 targeted deletion of MMP 9 attenuated LV remodeling and collagen accumulation via overexpression of MMP-2 and MMP-13,18 and an increase in MMP-1 by hepatocyte growth factor was beneficial on post-MI heart failure via its anti-fibrotic action;19 reduction of scar tissue and improvement of remodeling were observed simultaneously in myocardial regeneration therapy.1,6 An adequate content of fibrosis may be different because of the various conditions of MI models, such as small or large MI, small or large amount of myocardial regeneration, and permanent or transient ischemia. Thus, in the present model with transient ischemia and nonlarge infarction, the beneficial effect of G-CSF relies on a higher number of migrated neutrophils/macrophages and upregulation of MMPs and increased participation of mobilized stem cells, which supports the concept of myocardial repair after infarction by Orlic et al.6 Also, further investigation using MMP gene–defective mice or mice with blocking antibody would be warranted to define the precise role of MMP.
The standard therapy for human MI is reperfusion therapy, and our study showed the beneficial effect of G-CSF using a reperfusion model. The dose of G-CSF and the increased level in peripheral leukocytes were similar to those in a normal human donor for BM transplantation.20 In addition to modification of the healing process and regeneration of myocardial tissues, this would offer the important suggestion of using the clinical application of G-CSF as a noninvasive therapy.
Both the acceleration of the healing process and myocardial regeneration play an important role in the beneficial effects of G-CSF.
This study was supported in part by Research Grants of Frontier Medicine (15209027, 15590732, and 14570700) from the Ministry of Education, Science, and Culture of Japan.
Asahara T, Masuda H, Takahashi T, et al. Bone marrow origin of endothelial progenitor cells responsible for postnatal vasculogenesis in physiological and pathological neovascularization. Circ Res. 1999; 85: 221–228.
Wagers AJ, Sherwood RI, Christensen JL, et al. Little evidence for developmental plasticity of adult hematopoietic stem cells. Science. 2002; 297: 2256–2259.
Taylor DA, Hruban R, Rodriguez ER, et al. Cardiac chimerism as a mechanism for self-repair: does it happen and if so to what degree? Circulation. 2002; 106: 2–4.
Takemura G, Ohno M, Hayakawa Y, et al. Role of apoptosis in the disappearance of infiltrated and proliferated interstitial cells after myocardial infarction. Circ Res. 1998; 82: 1130–1138.
Orlic D, Kajstura J, Chimenti S, et al. Mobilized bone marrow cells repair the infarcted heart, improving function and survival. Proc Natl Acad Sci U S A. 2001; 98: 10344–10349.
Kalka C, Masuda H, Takahashi T, et al. Transplantation of ex vivo expanded endothelial progenitor cells for therapeutic neovascularization. Proc Natl Acad Sci U S A. 2000; 97: 3422–3429.
Spinale FG. Matrix metalloproteinases: regulation and dysregulation on the failing heart. Circ Res. 2002; 90: 520–530.
Creemers EEJM, Cleutjens JPM, Smits JFM, et al. Matrix metalloproteinase inhibition after myocardial infarction: a new approach to prevent heart failure? Circ Res. 2001; 89: 201–210.
Taniyama Y, Morishita R, Nakagami H, et al. Potential contribution of a novel antifibrotic factor, hepatocyte growth factor, to prevention of myocardial fibrosis by angiotensin II blockade in cardiomyopathic hamsters. Circulation. 2000; 102: 246–252.