Abstract
As articular cartilage has very limited self-repair capability, the repair and regeneration of damaged cartilage is a major challenge. This review aims to outline the past, present, and future of cell therapies for articular cartilage defect repair. Autologous chondrocyte implantation (ACI) has been used clinically for more than 20 years, and the short, medium, and long-term clinical outcomes of three generation of ACI are extensively overviewed. Also, strategies of clinical outcome evaluation, ACI limitations, and the comparison of ACI clinical outcomes with those of other surgical techniques are discussed. Moreover, mesenchymal stem cells and pluripotent stem cells for cartilage regeneration in vitro, in vivo, and in a few clinical studies are reviewed. This review not only comprehensively analyzes the ACI clinical data but also considers the findings from state-of-the-art stem cell research on cartilage repair from bench and bedside. The conclusion provides clues for the future development of strategies for cartilage regeneration.
Introduction
Cartilage is a poorly vascularized tissue and has a limited capability of self-repair. After trauma or diseases that cause cartilage degradation, it never spontaneously heals, often leading to osteoarthritis. Osteoarthritis affects 21 million people in the US, resulting in over 50% of the total joint replacements, and costing more than US$15 billion per year. The number of incidents is increasing. It is estimated that almost 1,500,000 arthroscopic surgical procedures of the knee are performed each year (155). Hence, the prevention and treatment of osteoarthritis is urgent (49). Currently, therapies such as microfracture (78), abrasion, drilling (37), and osteochondral grafting (116) have been applied to articular cartilage repair. Indeed, these methods help to reduce pain in affected patients to some degree. However, complications and injuries caused by these methods are huge and the results have been largely unsatisfactory. So, autologous chondrocyte implantation (ACI) with or without additional biomaterials has emerged (56,101).
Treatment outcomes of cell-level therapies and tissue engineering treatment for cartilage repair are encouraging. Since the 1980s, ACI has been widely used in clinical practice. More than 15,000 patients have received this treatment (92,121). Besides chondrocytes, cell therapies based on stem cells (136) such as mesenchymal stem cells (MSCs) and embryonic stem cells (ESCs) are emerging and promising. MSCs have been applied for clinical cartilage repair (195). In the future, ESCs will be the candidates for complete cartilage regeneration as they can self-replicate and differentiate into tissues from all three germ layers (79). This review summarizes the most recent advances in ACI and MSC-based clinical cartilage repair as well as the development of MSC- and ESC-based cartilage regeneration research. Also, the limitations of current cell-based cartilage repair techniques and the direction of future research are outlined and discussed.
Autologous Chondrocyte Implantation (ACI)
Development of ACI
ACI began with O'Driscoll et al.'s studies using periosteal grafting to treat chondral defects in rabbits (135). Grande and colleagues then refined that technique to culture autologous chondrocytes (61,62). In the 1980s, ACI was applied in clinical practice for the first time to treat full-thickness chondral defects in the knee (23). Basically, ACI comprises a two-stage procedure: cell harvest and implantation. Chondrocytes are first harvested from the non-load-bearing area of the joint and expand in vitro to acquire enough cells before implantation. In the past 20 years, three generations of ACI have been developed in implantation method. The first generation followed a standard implantation procedure involving preparation of defects, periosteal harvesting, suturing periosteum over defects, application of fibrin glue sealant, and implantation of chondrocytes. The second generation made use of a collagen membrane to suture over prepared cartilage defects, and then a suspension of cells was injected underneath. The third generation was based on biomaterials, and so was named matrix-induced autologous chondrocyte implantation (MACT or MACI). Chondrocytes were seeded on biomaterials, which were trimmed to match the defect size. These “all-in-one” grafts were then implanted without either a periosteal cover or fixing sutures. Collagen membrane, hyaluronan polymer, and collagen gel were used most often (114,147).
Clinical Evaluation of ACI Efficiency
The evaluation methods depend on the clinical purpose. Generally, MRI is the main effective method for ACI evaluation (73,183) because it is noninvasive, and hence causes no pain or secondary injury. The use of sensitive sequences, fat-suppressed proton density-weighted T2, fast spin-echo (PD/T2-FSE) (183), and three-dimensional gradient-echo (3D GRE) (190) are often reported for cartilage detection. Ongoing developments include isotropic 3D sequences (57). To improve morphological analysis and in vivo biochemical imaging, many techniques are used in diagnosis, such as delayed gadolinium-enhanced MRI of cartilage (dGEMRIC), a technique for evaluating the glycosaminoglycan content of articular cartilage (55,180), quantitative T2 mapping for valuating water and collagen content (177,197), combination of T2 and dGEMRIC measurements (97), and combined dGEMRIC, T2 mapping, and diffusion-weighted imaging for analyzing cartilage (48,108,183). Among these, the most promising technique for ACI evaluation is 3D-dGEMRIC (182) with magnetic field ranges from 1 to 7 Tesla (177,199).
For evaluation of knee joint function, the International Cartilage Repair Society (ICRS) system and the Lysholm score system are most often used in case reports. The Lysholm score has an eight-question format on the restoration of function: limping, pain, support, locking, swelling, stair climbing, instability, and squatting. It was designed for knee ligament injuries but has also been validated for other knee injuries such as articular cartilage damage (167). Other clinical scoring systems for cartilage function are also used, such as knee injury and osteoarthritis outcome score (161), HSS (Hospital for Special Surgery) (111), mean Mazur ankle score (200), modified Cincinnati score (95), Hannover ankle rating score, American Orthopaedic Foot and Ankle Society ankle-hindfoot score (16), McDermott score (102), and Western Ontario and McMaster score (189).
For accuracy, maturation of cartilage repair tissue can be assessed by second-look arthroscopy and histology of biopsies. The histological quality of the repair tissue formed after ACI improves with time after implantation (54). Hematoxylin and eosin staining and safranin O staining are often used to show the morphology and matrix of cartilage (22), and immunohistochemical staining for aggrecan, type II collagen, and type X collagen are used to test the composition and location of repaired tissue (177,201). The ICRS system, which includes visual histological scoring, was developed to quantitatively compare the structure of cartilage repair. Peterson et al. studied the histology of cartilage repair and found that samples from 8 of 12 ACI patients showed hyaline characteristics with safranin O staining and a homogeneous appearance under polarized light (141). In another ACI study, 8 of 14 patients regenerated hyaline cartilage that contained type X collagen in the deepest layers and type 2 collagen in the deep layers, and three fibrous and eight hyaline biopsy specimens were stained positive for aggrecan and for cartilage oligomeric matrix protein (22). In a large histological study of 41 cases, the graft consisted of hyaline cartilage in four cases, mixed fibrohyaline cartilage in 10, fibrocartilage in 25, and fibrous tissue in two (177).
Overview of ACI Clinical Outcome
Based on our literature search and data mining, 125 articles related to ACI clinical case reports have been published since 1994 (Table 1).
References for Details of ACI Studies
Some of the therapeutic effects were reviewed by Marlovits et al. (114).
According to the literature review (Fig. 1), 6,520 patients were followed up after ACI treatment. Among them, 2,609 cases had 0–2 years of follow-up, 3,123 had 2–5 years, and 788 had more than 5 years. The overall therapeutic efficacy in the short-to-medium term (0–5 years) is 70–90%, as evidenced by relief of symptoms and improvement of joint function. Based on the 11 reports with more than 5-year follow-up (Table 2), the efficacy of ACI treatment can be maintained for more than 5 years, and even as long as 10–20 years (85,187). If the long-term efficacy can be confirmed with more clinical cases, cell-based joint repair and regeneration will revolutionize joint surgery.
ACI case reports. Number of patients with ACI and number of ACI-related articles in each year. Also shown in the columns is the number of cases reported in different follow-up time periods: white (0–2 years), gray (2–5 years), and black (>5 years).
ACI Case Reports With Over 5 Years of Follow-up
Comparison of Different Generations of ACI
Several studies compare the different generations of ACI. The purpose of such studies is to optimize the treatment so that simpler operations with better therapeutic effects can be developed for patients.
It was reported that the standard procedure used in the first generation gives better results than the type I/III collagen-covered method used in the second generation. After a 2-year follow-up, 33 cases of first generation and 35 cases of second generation ACI showed 74% and 67% good or excellent results (60), indicating comparable therapeutic effects.
Several other studies compared the first and third generations of ACI. Bartlett and colleagues designed a prospective, randomized study in which 44 cases were given periosteum-covered treatment (generation 1) and 47 cases were treated with collagen bilayer seeded with chondrocytes (generation 3). No difference was found between these generations in terms of the clinical, arthroscopic, and histological outcomes (15). Manfredini et al. used MRI, second-look arthroscopy, histology of biopsies, and molecular analysis to identify differences between ACI (n = 17) and MACI (n = 10). According to the HSS and ICRS scores, there were no statistically significant differences between the groups at 6 and 12 months after surgery (111).
In another randomized controlled trial comparing ACI (generation 1) and MACI (generation 3), besides the IKDC (International Knee Documentation Committee) score, several parameters such as health-related quality of life (Short Form-36 Health Survey), knee functionality (Lysholm and Gillquist score), and physical activity (Tegner activity score) were used to evaluate the therapeutic effects. There was no difference in the efficacy between the original and the advanced ACI techniques 12 and 24 months after surgery regarding the IKDC score, Tegner activity score, and Short Form-36; however, with respect to the Lysholm and Gillquist score, better efficacy was found in the periosteal flap technique group (207).
The comparative studies demonstrated that no matter what kind of ACI is used, either periosteum or collagen bilayer, there are no significant differences among the three generations of ACI by current evaluation methods and parameters. However, none of these studies had more than 2 years of follow-up, a relatively short time to detect degeneration or hypotrophy of the new cartilage. Long-term follow-up and more parameters (81,207) need to be considered in comparative studies.
Comparison of ACI with Current Surgical Therapeutics
Although there are no significant differences among different generations of ACI, the improved clinical outcomes demonstrate the potential of ACI for cartilage defect repair. Moreover, studies comparing ACI with other current surgical therapeutics reported that ACI is clearly superior to mosaicplasty, autologous osteochondral graft, and microfracture.
Mosaicplasty or autologous osteochondral grafts are often used in clinical practice, although they have the inherent disadvantages of limited graft source and donor site morbidity. A prospective, randomized comparison trial showed 82% good or excellent regeneration in the ACI group (n = 58), but only 34% in the mosaicplasty group (n = 42) after a 19-month follow-up (20). Another study compared MACT with osteochondral autograft transplantation (OCT). Eighteen patients (9 MACT, 9 OCT) were followed with MRI T2 assessment and clinical evaluation. The results of 41.6 months of follow-up confirmed that MACT is consistently superior to OCT (160).
Microfracture (MFX) is widely used because of its low cost and minimal invasiveness. Reports comparing ACI with MFX have been published since 2007. In terms of short-term therapeutic outcome, ACI leads to cartilage defect filling with better structure and content than after MFX. Brown et al. evaluated the MRI results from 112 patients at 15 months after ACI (n = 35) or MFX (n = 86). The ACI group exhibited consistently better filling of the defect (24). Trattnig et al. (180) used the dGEMRIC technique to evaluate the relative glycosaminoglycan (164) content of repaired tissue after MFX (n = 10) or MACT (n = 10) of the knee joint after an average of 33 months. Higher GAG content in repaired tissue was found in the group treated with MACT (180). Saris et al. compared MACI with MFX (MACI = 57, MFX = 61) using the Knee Injury and Osteoarthritis Outcome Score system, and found that chondrocyte implantation was associated with superior tissue regeneration at 1-year follow-up (161).
However, when reviewing the records of long-term therapeutic effect, there are controversial findings. Knutsen et al. compared the effect of ACI (n = 40) and MFX (n = 40) on cartilage repair and found that both methods gave satisfactory results in 77% of patients after 5 years (85). Kon et al. compared second-generation ACI (n = 40) with MFX (n = 40) for treatment of chondral lesions of the knee after 5 years of follow-up and reported that both methods gave satisfactory clinical outcomes. But better clinical results and resumption of sport activity were reported in another study comparing second-generation ACI with MFX based on a 5-year follow-up (88). Maturation of the repaired cartilage tissue with time may be the reason why ACI provides more satisfactory results in the long term (152). Although reports have not provided consistent conclusions in comparisons of the long-term clinical outcomes of ACI and MFX, generally, the host regeneration induced by MFX might be insufficient, and application of transplanted cells is necessary when the defect area is relatively large (3,19). Thus, the size of the defect should be a major concern when choosing ACI or MFX in clinical practice.
Disadvantages of ACI
Although complications of ACI are rare, they do occur. The United States Food and Drug Administration reported that the most common adverse events of ACI are graft failure, delamination, and tissue hypertrophy. The incidence rate is 3.8% (77,107,201). Besides these, the inferior quality of cartilage repair tissue is also a major concern. LaPrade et al. collected failure cases of 10 patients who underwent autologous chondrocyte implantation (n = 6), MFX (n = 3), or periosteal transplantation (n = 1). Histological and immunohistochemical assays showed that fibrous connective tissue and fibrocartilage were the primary components, and all samples expressed type I instead of type X collagen; 80% of samples were stained positively for type I collagen, higher than for type II (100). Moreover, it was reported that other factors such as the surgical procedure influence the quality of repair and failure rate. Minas and colleagues reviewed 522 ACI cases with 2-year follow-up and reported an increased failure rate of ACI after previous treatment with marrow stimulation techniques (124).
Also, the application of ACI is limited by low cell numbers upon harvest, loss of chondrocyte markers, dedifferentiation in culture, limited life span (74), potential cell leakage upon implantation of chondrocyte suspension (84), and uneven distribution of cells in three-dimensional space (10). Many patients waiting for ACI may not get the treatment because of chondrocyte shortage. In addition, the complexity of the operative procedure remains a major concern for patients.
In conclusion, ACI or MACI efficiently repair articular cartilage defects to some degree and prevent the development of osteoarthritis. However, extensive applications are limited by intrinsic problems. Among these, cell source shortage is a key issue that needs to be resolved by exploring other resources such as stem cells. Unless these problems can be resolved successfully, it will be difficult to meet the needs of all patients with cartilage defects by using cell therapy.
Allogeneic Chondrocyte for Cartilage Repair
As ACI has several complications, allogeneic chondrocyte has been investigated as a candidate seed cells for cartilage repair. It was reported that allogeneic chondrocyte provocates immune response and gradually destroys the resulting cartilage tissue (148). It limits the clinical application of allogeneic chondrovytes. However, very recently there is a hot strategy of using a biomaterial as vehicle for allogenic chondrocyte delivery to repair cartilage (46,115,140). A clinical trial shows that allogenic chondrocytes cultured in alginate beads can be used to treat osteochondral lesions. The 2-year follow-up of 21 patients showed apparent functional improvement (4).
MSCs for Cartilage Repair
MSC-based repair of full-thickness articular cartilage defects has been widely attempted in animal models, together with various growth factors and carrier matrixes (9,34,80,130,196).
However, the development of MSCs for cartilage regeneration and osteoarthritis treatment is still at the infant stage; genetically modified MSCs/ECs for cartilage regeneration have only been tested on small animals. Genetically modified MSCs are usually with some factors such as transforming growth factor-β (TGF-β) and bone morphogenetic proteins (BMPs) (44,99,103,168,208). Preclinical results of cartilage repair from small animals have shown BMP-2-modified MSC injections were successful in achieving repair of both bone and cartilage in vivo (205). As the concern of clinical application, most gene therapy has utilized virus-based transfer of genes such as adenoviruses and retroviruses. Insertional mutagenesis due to integration events after retrovirus-mediated gene transfer or immune response induced by direct injection of adenovirus vectors are the safety concerns that need to be carefully tested in future.
Different MSCs Sources for Cartilage Repair
MSCs can be isolated from bone marrow, synovium, periosteum, skeletal muscle, and adipose tissue. Most of them have similar epitope profiles (139), but significant differences in MSC properties have been noted, depending on the source. For expansion, cells from synovium, periosteum, and bone marrow proliferate faster than those from muscle and adipose tissue. Regarding the capability for differentiation into specific lineage cells, synovium-derived cells have the greatest ability for chondrogenesis (105), synovium and adipose tissue-derived cells show the highest percentage of adipogenesis, while bone marrow, synovium, and periosteum-derived cells exhibit predominantly osteogenic differentiation (159). Several in vivo studies indicate that stem cells isolated from perichondrium, periosteum, and bone marrow are superior to those from adipose tissue in terms of forming hyaline cartilaginous tissue (159,204).
Clinical Research
MSC-based cartilage repair has been widely attempted in animal models, but there are few reports of clinical applications. Early preclinical and clinical data have demonstrated the safety, availability, and effectiveness of MSC therapy. Wakitani et al. reported several cases of clinical bone marrow-derived MSC transplantation. The first study recruited two patients, one 26-year-old female and one 44-year-old male who suffered knee pain and could not walk normally. Autologous bone marrow stromal cells embedded in a collagen gel with an autologous periosteum cover were transplanted into the articular cartilage defects in the patellae. After 6 months, the patients were pain free and their walking improved greatly. These improvements can be maintained for a long time, 5 years and 9 months in one case, and 4 years in the other (193). The second study reported 12 cases of MSC transplantation for repairing articular cartilage defects in the medial femoral condyle. Forty-two weeks after the operation, the defects were covered with white soft tissue and partial hyaline cartilage-like tissue in the treatment group. Compared to the control group (n = 12), the MSC group had better arthroscopic and histological grading scores, although the clinical improvement was not significantly different (192). The third report concerned a 31-year-old male athlete. The size of the full-thickness articular cartilage defect in the right knee was 20–30 mm2, corresponding to a grade IV defect in ICRS. Seven months later, the defect was filled with hyaline-like cartilage tissue. One year later, the clinical symptoms improved and the patient even regained his previous activity level and had no pain or other complications (98). In the fourth clinical report, autologous bone-marrow stromal cell transplantation even showed potential to repair osteochondral damage after septic arthritis of the knee. MSCs delivered in interconnected porous hydroxyapatite ceramic (IP-CHA) were used to treat a large osteochondral defect in the knee of a 21-year-old man who suffered septic arthritis. Regeneration of cartilage-like tissue and bone formation were observed in the repaired tissue (1). In all, these clinical studies demonstrate that autologous bone marrow MSC transplantation is a promising approach for articular cartilage defect repair.
Limitations of MSCs for Cartilage Repair
MSC-based therapy for cartilage defects is effective to some extent, and there is no clinically significant toxicity (98). However, the degree of improvement may be related to age. Young and active patients give excellent results (11,118,144). Besides that, there are still limitations, including morbidity of donor site, lower cell numbers, and limited proliferative potential with increasing donor age (144). Before transplantation, MSCs must be expanded several thousand-fold in vitro to provide sufficient cells for cartilage repair. This long-term in vitro expansion may change the cells phenotype and lead to spontaneous transformation (76).
Another concern is the loss of immunosuppression and gain of immunogenicity during the chondrogenesis of allogeneic MSCs. Li's group reported that chondrogenic differentiating MSCs promote h-dendritic cell maturation, resulting in lymphocyte proliferation and cytotoxicity (29). This implies that the immune tolerance of MSCs might not be effective in their application for cartilage tissue repair. In another study of clinical application of MSCs for osteogenesis imperfecta treatment, an immune response caused by anti-FCS antibodies was observed (13).
In addition, many other questions remain to be answered, such as the mechanisms of engraftment, homing, in vivo differentiation, and therapeutic efficacy of transplanted cells (176). Could MSCs cause tumors in patients receiving MSCs transplantation for cartilage repair? Also, newly grown tissue is biomechanically inferior to normal articular cartilage. Wide clinical application and clinically satisfactory results cannot be achieved before all of the above questions are properly answered.
Embryonic/Pluripotent Stem Cells for Cartilage Repair: The Future
ESCs are considered to be the best potential candidates as seed cells for all tissue engineering and regeneration, including cartilage regeneration (12). However, induction of ESC differentiation into the chondrolineage is a challenge for cartilage tissue engineering.
In vitro chondrogenesis of ESCs has been investigated with the use of growth factors and cytokines, the microenvironment, or culture conditions. BMP-2 (77,174) and TGF-β1 (202) are directly used for induction, and retinoic acid followed by TGF-β1 are also effective for chondrogenic differentiation of ESCs (91). The culture models of micromass (170), and coculture with limb bud progenitor cells helps to promote chondrogenesis of ESCs compared with monolayer culture. EB (ESC-derived embryoid bodies) culture has also proved to be a suitable model to study development in vitro (79). Efforts to improve the chondrogenic differentiation of ESCs via EBs have been made by many research groups. Providing 3D culture environments in scaffolds (43,203) such as poly(ethylene glycol)-based (PEG) hydrogels (172), or encapsulated in alginate together with dexamethasone (172), screening for optimal size of EBs (119), administering different growth factors such as bone morphogenic proteins, TGF-β (79,178,202), and controlling other environmental variables such as oxygen level (87) are some techniques used to improve chondrogenic differentiation.
Chondrogenic differentiation of ESCs is mainly studied at the animal cell level, such as with mouse ESCs. Recently, human ESCs (hESCs) have been under investigation and, according to Vats et al., BMP-2 has been used to induce chondrogenic differentiation of hESCs in the EB direct-plating method, and a high-density 3D microenvironment greatly enhances chondrogenic differentiation (188). Moreover, hESCs cocultured with chondrocytes and implanted into SCID mice on a poly-D,L-lactide scaffold do not form teratomas (12).
Repairing cartilage by implanting ESCs or ESC-derived cells has been under evaluation and positive results have been reported. Repair of osteochondral defects is promoted by implanting ESCs into the cartilage defect site (32,191). Implantation of ESC-derived chondrogenic cells in mice produces cartilage tissue in vivo in 3–4 weeks (43).
Clinical trials using hESC-derived cells have just obtained approval. In 2009, the US FDA approved phase I of clinical trials with hESC-derived neural stem cell implantation for spinal cord regeneration (146), making it possible for applying ESCs to the treatment of other severe diseases in future studies.
However, ESCs will not be extensively applied to cartilage tissue engineering until several key problems are resolved. First, pure chondrocytes have not been obtained from ESCs in vitro or in vivo (70). Second, prevention of ESC-derived chondrocytes from hypertrophy and calcification is a major challenge for researchers (185). Another major problem is the risk of teratoma formation where ES cells are transplanted. Wakitani et al., after implanting ESCs into the joint space and subcutaneous space of mice, found teratomes on both sides (194). Immunological rejection also needs to be resolved in applications of allogeneic ESCs. In addition, the harvest and use of hESCs faces ethical issues (82).
The successful induction and isolation of iPSCs provide a new hope to avoid immune rejection. Transformation from fibroblasts to pluripotent stem cells can expand the cell source and lead to a “personalized therapy” in the future. Several groups reported that iPSCs can differentiate to multiple types, including dendritic cells, macrophages (163), cardiomyocytes (173), retinal cells (75), adipocytes, and osteoblasts (174). However, the differentiation of iPS for chondrogenisis remains to be investigated.
Conclusion and Perspectives
Great progress has been made in cell-based cartilage repair and prevention of osteoarthritis. ACI has shown solid evidence of therapeutic effect and bone marrow-derived MSC implantation has also shown promising clinical outcomes. The large amount of clinical data clearly illustrates that cell transplantation is the future direction of cartilage repair. Although some success has been achieved, cell therapy is still far from being reliable and satisfactory. In the future, several challenges need to be overcomed before cell therapy achieves full cartilage regeneration.
First, how do we find a lineage-specific cell source that can produce hyaline cartilage without fibrocartilage formation or terminal differentiation? Stem cells remain the best option for full cartilage regeneration. MSCs from different tissues are a readily available stem cell source, but the lack of specific markers for the isolation of MSCs or cartilage progenitors impedes efforts towards homogenous cartilage differentiation and clinical application. In future, ESCs or iPS-derived cartilage stem cells will be the major cell sources for cartilage regeneration.
Second, what are the specific differentiation factors for hyaline cartilage tissue regeneration? Currently, MSCs induced in vitro or implanted into defects and induced by local stimuli cannot fully and homogenously differentiate into hyaline cartilage. Also, hypertrophy and calcification often occur. So, investigations of the hyaline chondrocyte–ECM interaction as well as the molecular mechanisms of hyaline cartilage development and differentiation are warranted for better control of stem cell differentiation into stable hyaline cartilage.
Besides the cell source and cell differentiation, mechanical stability of regenerated cartilage, fixation of the transplant to the subchondral bone in the joint, and immunological problems remain to be solved (150). How to improve, standardize, and generalize a noninvasive method in clinical practice, and functional evaluation in vivo are works for the future. When we overcome these challenges, cell-based therapy will bring a revolution to clinical cartilage repair and regeneration from which many patients will benefit and could avoid end-stage artificial joint replacement.
Footnotes
Acknowledgments
The author thanks Ms. Xiu Juan Lv for MACI data sorting and Ms. Do Thi Hai Van and Mr. Faisal Rehman for English polishing. This work was supported by Program for New Century Excellent Talents in University (NCET-08–0487), Zhejiang Provincial Program for the Cultivation of High-level Innovative Health talents, ISTCP grant (2008DFB30090), National Science Foundation of China (NSFC, 81071461,31000436), and Zhejiang Province grants (Y2080141,Z2100086).