Author + information
- Received January 5, 2007
- Revision received April 3, 2007
- Accepted April 22, 2007
- Published online October 23, 2007.
- Thomas R. Payne, PhD⁎,‡,§,
- Hideki Oshima, MD, PhD⁎,§,
- Masaho Okada, MD⁎,§,
- Nobuo Momoi, MD†,
- Kimimasa Tobita, MD†,
- Bradley B. Keller, MD†,
- Hairong Peng, MD, PhD⁎,§ and
- Johnny Huard, PhD⁎,‡,§∥,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Johnny Huard, 3705 5th Avenue, 4100 Rangos Research Center, Pittsburgh, Pennsylvania 15213.
Objectives We investigated whether vascular endothelial growth factor (VEGF) was associated with the angiogenic and therapeutic effects induced after transplantation of skeletal muscle-derived stem cells (MDSCs) into a myocardial infarction (MI).
Background Because very few MDSCs were found to differentiate into new blood vessels when injected into the heart, the mechanism underlying the occurrence of angiogenesis after MDSC transplantation is currently unknown. In the present study, we used a gain- or loss-of-VEGF function approach with skeletal MDSCs engineered to express VEGF or soluble Flt1, a VEGF-specific antagonist, to identify the involvement of VEGF in MDSC transplantation-induced neoangiogenesis.
Methods Vascular endothelial growth factor- and soluble Flt1-engineered MDSCs were injected into an acute MI. Angiogenesis and cardiac function were evaluated by immunohistochemistry and echocardiography.
Results Both control and VEGF-overexpressing MDSCs induced angiogenesis, prevented adverse cardiac remodeling, and improved function compared with saline-injected hearts. However, these therapeutic effects were diminished in hearts transplanted with MDSCs expressing soluble Flt1 despite successful cell engraftment. In vitro experiments demonstrated that MDSCs increased secretion of VEGF in response to hypoxia and cyclic stretch (likely conditions in ischemic hearts), suggesting that transplanted MDSCs release VEGF in vivo.
Conclusions Our findings suggest that VEGF is essential for the induction of angiogenesis and functional improvements observed after MDSC transplantation for infarct repair.
Progenitor cells of skeletal muscle have been considered advantageous for clinical use, since skeletal muscle is an abundant and accessible tissue source, isolated cells are highly proliferative in culture, and transplantation can be performed in an autologous approach. However, limitations associated with skeletal muscle-derived cells for cardiac repair are based on reports showing inadequate differentiation of transplanted cells into a cardiac phenotype and limited electromechanical coupling and synchronous contraction with one another or with host myocardium via the expression of connexin-43 gap junction protein (1–7). Despite these deficiencies, myogenic cells still have demonstrated therapeutic improvements of cardiac function in various experimental animal models (8). Although the exact mechanism for the beneficial effect of myoblast transplantation is unknown, the inhibition of adverse remodeling and the induction of neoangiogenesis appear to be major factors (9,10).
We have previously shown that mouse skeletal muscle-derived stem cells (MDSCs) injected into hearts after myocardial infarction (MI) improve cardiac function more effectively than do committed skeletal myoblasts (1). Because only a very low percentage of transplanted MDSCs expressed connexin-43 gap junctions and acquired a cardiomyocyte phenotype through either differentiation or fusion with host cardiomyocytes (1,5), we largely attributed this functional advantage to a greater ability of MDSCs to survive, engraft, and induce neoangiogenesis within the infarct (1). Furthermore, very few donor-derived MDSCs differentiate or fuse into blood-vessel-like structures, indicating that the neovasculature resulting after MDSC transplantation is comprised mostly of host-derived cells (1,5). This led us to consider the mechanism for neovascular induction after MDSC transplantation into ischemic myocardium.
Recent evidence suggests that cardiac cell therapy provides therapeutic benefit through the paracrine actions of factors released by transplanted cells (11). Indeed, we have previously demonstrated that one of the most potent angiogenic factors, vascular endothelial growth factor (VEGF), was expressed by many of our engrafted MDSCs between 1 and 12 weeks after transplantation into an MI. On the basis of these observations, we hypothesized that VEGF could be a primary factor in the ischemic milieu that is responsible for the induction of angiogenesis within the infarct (1). We, therefore, designed experiments to investigate the angiogenic and therapeutic role of VEGF in infarcted hearts injected with MDSCs.
To test this, we performed gain- and loss-of-VEGF function experiments using genetically engineered mouse MDSCs that over-express VEGF or the VEGF-specific antagonist soluble Flt1 (sFlt1), to enhance or inhibit the biological effects of VEGF within ischemic myocardium. Our results demonstrate that VEGF stimulates, and is necessary for, angiogenesis in MDSC-transplanted hearts. Additionally, the therapeutic benefit observed, after MDSC transplantation, was diminished when VEGF-induced infarct neovascularization was inhibited with sFlt1. Finally, we demonstrated that MDSCs up-regulate the secretion of VEGF when exposed to hypoxia and cyclic stretch (likely conditions within ischemic myocardium), further supporting the notion that engrafted MDSCs release VEGF to the milieu of infarcted hearts. These results confirm that VEGF is necessary for the therapeutic induction of angiogenesis after MDSC transplantation for MI repair.
Cell culture and transduction to express VEGF and sFlt1
Mouse MDSCs were isolated and cultured, and construction of the retroviral vectors containing genes encoding human VEGF 165, human sFlt1, or bacterial nuclear-localized LacZ(nLacZ) was performed as previously described by members of our laboratory (12,13). Muscle-derived stem cells were retrovirally transduced to express the VEGF transgene (MDSC-VEGF), the sFlt1 transgene (MDSC-FLT), and the nLacZtransgene (MDSC-LacZ) (12). Because the long-term overexpression of VEGF induces deleterious side effects in vivo, including hemangiomas (14–16), we avoided this problem by decreasing the dosage of VEGF released from transplanted MDSCs. We combined MDSC-VEGF cells with control MDSCs expressing nLacZ(MDSC-LacZ) at 2 different dilutions: 1) 50% MDSC-VEGF cells and 50% MDSC-LacZcells (MDSC-VEGF50); and 2) 25% MDSC-VEGF cells and 75% MDSC-LacZcells (MDSC-VEGF25).
MI and cell transplantation
All animal experiments and surgical procedures were approved by the Institutional Animal Care and Use Committee of Children’s Hospital of Pittsburgh (protocol no. 07/03). Acute MI was induced in 56 non-obese diabetic-severe combined immunodeficient (NOD-SCID) mice (male, age 16 weeks, 25 to 30 g, NOD.CB17-PrkdcSCID/J, Jackson Laboratory, Bar Harbor, Maine), as previously described (1). Immediately after ligation, 3 × 105cells were transplanted directly into the ischemic region, as previously described (1). The investigators were blinded to the solution contents injected into each group of mice: phosphate-buffered saline (PBS) only (n = 11 mice), MDSC-LacZ(n = 11), MDSC-FLT (n = 12), MDSC-VEGF25 (n = 11), or MDSC-VEGF50 (n = 11).
Tissue processing, histological and immunohistochemical stainings
Mice were euthanized, and their hearts were harvested and frozen in 2-methylbutane precooled in liquid nitrogen, and serially cryosectioned (from the apex to the base) into 7 μm-thick sections. A mouse antifast skeletal myosin heavy chain (fskMyHC) antibody (1:400, MY-32 clone, Sigma, St. Louis, Missouri) and a rat antimouse CD31 antibody (1:100, BD Pharmingen, San Diego, California) were used to immunostain skeletal myofibers and capillaries as previously described (1,5,12). We used Masson’s Modified IMEB Trichrome Stain Kit (IMEB, San Marcos, California) to stain the infarct scar according to the manufacturer’s instructions.
All fluorescent and brightfield microscopy was performed with a Nikon Eclipse E800 microscope (Nikon Corp., Tokyo, Japan) equipped with a Retiga EXi digital camera (Q Imaging, Burnaby, Canada). Images were acquired with Northern Eclipse software (version 6.0, Empix Imaging, Inc., Cheektowaga, New York). Capillary density, cell engraftment size, left ventricle (LV) scar tissue area ratio, and LV infarct size measurements were performed with ImageJ software (version 1.32j, National Institutes of Health, Bethesda, Maryland). We measured CD31[+] capillary density (CD31[+] capillary structures per high-power field [HPF]) within the infarct and fskMyHC[+] area in the digital images of HPF (200× magnification) obtained from each group (n = 3 hearts/group/time point at 2, 6, and 12 weeks after injection). To measure LV infarct size and scar tissue area ratio, digital images of low-power fields (20×) of the entire LV cross section were captured from the Masson’s trichrome-stained hearts. From these images, scar tissue area ratio was defined as the ratio of LV fibrosis to the area of normal LV myocardium, and LV infarct size was measured as the percentage of LV endocardial surface length infarcted to the total LV endocardial surface.
Echocardiograms assessed LV dimensions and systolic function (blinded investigator; 2, 6, and 12 weeks after cell transplantation), as previously described (1). Two-dimensional images were obtained at the midpapillary muscle level. Left ventricular end-diastolic area (EDA) and end-systolic area (ESA) were measured from short-axis images of the LV, and both LV end-diastolic dimension (EDD) and end-systolic dimension (ESD) were measured from at least 6 consecutive beats via M-mode tracing. To measure LV contractility, fractional shortening (FS) was calculated as FS (%) = [(EDD − ESD) ÷ EDD] × 100, and fractional area change (FAC) was calculated as FAC (%) = [(EDA − ESA) ÷ EDA] × 100.
In vitro stimulation of MDSCs
Hypoxia was induced by culturing MDSCs in an incubator (Heraeus, Newtown, Connecticut) for 24 h in proliferation medium (PM) (Dulbecco’s Modified Eagle’s Medium [DMEM], 10% FBS, 10% horse serum, 1% penicillin/streptomycin, and 0.5% chicken embryo extract) or serum-free medium (DMEM) at 37°C and 2.5% O2(n = 3 samples/group). Cyclic stretch was applied with a Flexercell Strain Unit (Flexcell International, Export, Pennsylvania). Muscle-derived stem cells were cultured to high confluency on collagen type 1-coated Bioflex culture plates (Flexcell International) and subjected to a 10% average surface elongation at 30 cycles/min for 1, 4, 10, and 24 h (n = 3 samples/group/time point). After each assay, the cell culture supernatant was collected and analyzed for VEGF by enzyme-linked immunoadsorbent assay (R&D Systems, Minneapolis, Minnesota). Cells were also collected and counted with a hemacytometer.
All measured data are presented as mean ± standard error. Statistical differences were determined by 2-way analysis of variance. When statistical differences were observed, the Student-Newman-Keuls multiple comparison test was used to perform post-hoc analysis (Sigma Stat, version 2.0, Jandel Scientific, San Rafael, California).
The injection of MDSC-LacZcells induced greater neovascularization of the infarct than did the injection of PBS (p < 0.01) (Fig. 1);however, we observed less angiogenesis in the MDSC-LacZgroup than in both the MDSC-VEGF25 and MDSC-VEGF50 groups at all time points (p < 0.01) (Fig. 1). The MDSC-VEGF25 and MDSC-VEGF50 groups displayed the greatest capillary density at all time points when compared with the other groups (p < 0.01) (Fig. 1). Despite the expected secretion of considerably more VEGF in the MDSC-VEGF50 group than in the MDSC-VEGF25 group, the MDSC-VEGF50-injected hearts displayed similar capillary density to MDSC-VEGF25-injected hearts at all tested time points (Fig. 1). Antagonism of VEGF with sFlt1 in the MDSC-FLT group resulted in levels of infarct vascularization comparable to the PBS group and considerably less than the MDSC-LacZgroup at all tested time points (p < 0.01) (Fig. 1). Taken together, these results suggest that VEGF alone stimulated more angiogenesis when overexpressed, and that antagonism of VEGF with sFlt1 inhibited the angiogenesis associated with MDSC transplantation.
Prior research shows that VEGF overexpression may result in abnormal vasculature at high micro-environmental dosages (14–17). Despite our efforts to control the amount of VEGF delivered by the VEGF-engineered MDSCs, hearts injected with the MDSC-VEGF50 group developed disorganized vascular structures 12 weeks after MI (Fig. 2).However, hearts in the MDSC-VEGF25 and MDSC-LacZgroups exhibited normal morphology and spatial organization of blood vessels within the MI at all tested time points (Fig. 2).
Donor cell engraftment
Previously, we have shown that the vast majority of MDSCs transplanted into the heart differentiate into skeletal myocytes expressing fskMyHC, and that very few donor-derived cells acquire a cardiac or endothelial phenotype (1,5). As expected, regions of the heart showing fskMyHC reactivity colocalized with regions containing nLacZ, a marker for the injected cells (data not shown). Based on measurements of engraftment size in fskMyHC-stained hearts, we observed comparable engraftments of MDSC-LacZcells and MDSC-VEGF25 cells, suggesting that VEGF overexpression in the MDSC-VEGF25 group had no effect on cell engraftment. However, VEGF overexpression adversely affected the engraftment of MDSC-VEGF50 cells at the 6- and 12-week time points (Table 1)suggesting that long-term expression of VEGF at high concentrations in the MDSC-VEGF50 group may have hindered donor cell engraftment at later time points, possibly because of VEGF toxicity or the formation of abnormal vascular structures (Fig. 2). At all tested time points, the MDSC-FLT cells showed fskMyHC areas comparable to the control MDSC-LacZcells, indicating that inhibition of VEGF by sFlt had no effect on cell engraftment.
LV infarct size and scar tissue area
In cross sections of Masson’s trichrome-stained hearts, we measured LV infarct size (Fig. 3A)and LV scar tissue ratio (Fig. 3B). Phosphate-buffered saline–injected hearts displayed very large LV infarcts and scar tissue area ratios at 2, 6, and 12 weeks after MI (Figs. 3A and 3B). In contrast, hearts in the MDSC-LacZand MDSC-VEGF25 groups had smaller infarcts and scar-tissue areas than hearts in the PBS and MDSC-FLT groups (p < 0.05) (Figs. 3A and 3B). The MDSC-VEGF50 group also displayed small infarcts and scar lesions at 2 and 6 weeks, but showed increased LV infarct size and scar tissue by 12 weeks (Figs. 3A and 3B). This increase in LV infarct size and scar tissue may again be related to the formation of abnormal vasculature observed in the MDSC-VEGF50-injected hearts after 12 weeks (Fig. 2). Hearts injected with MDSC-FLT cells displayed large infarcts and scar lesions comparable in size to levels observed in hearts injected with PBS. Thus, VEGF appears to play an integral role in MI wound healing after MDSC transplantation.
Assessment of LV function
Representative images of M-mode echocardiography tracing for each group are shown in Figure 4A.As assessed by EDD, progressive enlargement of the LV cavity occurred over time in hearts in the PBS group, indicating adverse cardiac remodeling after MI (p < 0.05, 2 vs. 12 weeks) (Fig. 4B). In contrast, no significant enlargement of the LV cavity occurred in the MDSC-LacZ– or MDSC-VEGF25–injected hearts over time (Fig. 4B). Progressive enlargement of the LV was also observed in the MDSC-VEGF50 hearts and the MDSC-FLT–injected hearts (p < 0.05, 2 vs. 12 weeks) (Fig. 4B). Of note, these findings appear to be consistent with the histological results in Figure 3.
As assessed by FS and FAC, hearts injected with PBS demonstrated poor cardiac contractility 2 weeks after MI, which continued to decline by 12 weeks (p < 0.05, 2 vs. 12 weeks). Hearts injected with MDSC-LacZcells or MDSC-VEGF25 cells displayed better LV contractility than did control PBS-injected hearts at all time points (p < 0.05) (Figs. 4C and 4D). The MDSC-VEGF50–injected hearts also displayed good LV systolic function at 2 and 6 weeks; however, 12 weeks after MI, hearts in the MDSC-VEGF50 group displayed reduced cardiac function compared with MDSC-LacZand MDSC-VEGF25 groups (p < 0.05) (Figs. 4C and 4D). This could be due to the formation of hemangioma-like structures (Fig. 2). Hearts injected with MDSC-FLT cells displayed poor systolic function compared with hearts injected with MDSC-LacZor MDSC-VEGF25 cells at all time points (p < 0.05) (Figs. 4C and 4D). In comparison with the PBS-injected hearts, MDSC-FLT–injected hearts revealed no difference in FS and FAC at any tested time points (Figs. 4C and 4D). Thus, these results suggest that preservation of LV dimensions and improvements in cardiac function elicited after MDSC transplantation are related, at least partly, to the paracrine effects of VEGF produced in the ischemic milieu.
Secretion of VEGF by MDSCs in response to hypoxia and mechanical stretch
Because hypoxia is a well-known initiator of angiogenesis signaling (18,19), we cultured MDSCs under hypoxic conditions (2.5% O2) for 24 h, and measured the VEGF secreted into the cell culture supernatant. Muscle-derived stem cells cultured in normal PM under hypoxic rather than normal culture conditions (20% O2) displayed a 6-fold increase in VEGF secretion (p < 0.05) (Fig. 5A).Furthermore, MDSCs cultured in serum-free medium under hypoxic conditions secreted 9 times the amount of VEGF expressed by MDSCs cultured in PM under normal conditions (p < 0.05) (Fig. 5A), indicating that a low-nutrient environment enhances the effect of hypoxia on the level of VEGF secretion by MDSCs.
Because the myocardium experiences continuous cyclic load, we subjected MDSCs to cyclic stretch in vitro to determine whether mechanical forces induce a VEGF response similar to that elicited by hypoxia. Ten or 24 h of cyclic stretch resulted in a 2-fold increase in VEGF secretion by MDSCs (p < 0.05 vs. non-stretch control) (Fig. 5B). Although cyclic stretch increased secretion of VEGF from MDSCs, our results indicate that hypoxia is a stronger initiator of VEGF expression by MDSCs. The combination of hypoxia, nutrient deprivation, and mechanical stress (likely conditions within ischemic myocardium) could induce high levels of VEGF secretion by MDSCs, which could explain our previously reported findings that MDSCs expressed VEGF for up to 12 weeks after transplantation into infarcted myocardium (1).
We observed that angiogenesis induced after intramyocardial MDSC transplantation into infarcted hearts is dependent on VEGF within the ischemic milieu. Our results show that VEGF increases angiogenesis and is important for the neovascularization that occurs within an MDSC-treated MI. Inhibition of VEGF bioactivity by sFlt1 resulted in decreased neoangiogenesis, increased infarct size, and decreased cardiac function, thus corroborating the angiogenic and therapeutic value of VEGF for MI repair in MDSC-treated hearts. Finally, we show that MDSCs secrete VEGF when stimulated by hypoxia and cyclic stretch, implying that transplanted MDSCs are a potential source of the VEGF that is released into the ischemic milieu.
Previous studies showed that the delivery of VEGF transgene overexpressing myoblasts into cardiac and skeletal muscle resulted in aberrant angiogenesis at high micro-environmental dosages of VEGF (14–16). Likewise, we observed the formation of abnormal vascular structures in our highest VEGF overexpressing cell population (MDSC-VEGF50) at only the latest time point (12 weeks), suggesting that we surpassed the threshold for VEGF micro-environmental dosage that would lead to normal induction of angiogenesis as demonstrated by Ozawa et al. (16). Additionally, we observed that this was detrimental to the persistence of the donor cell engraftment in the MDSC-VEGF50 group (Table 1). We speculate that this abnormal vasculature and poor engraftment was detrimental to the healing of the infarct scar (Fig. 3) and to sustained functional recovery of the LV (Fig. 4) 12 weeks after an acute MI. Such adverse events associated with abnormal vasculature are consistent with a report by Lee et al. (14), who showed that hemangiomas resulting from long-term VEGF transgene expression by myoblasts injected into nonischemic myocardium caused deleterious effects. Our results further support previous findings that VEGF must be regulated to optimize its therapeutic benefit (16).
The antagonism of VEGF with sFlt1 resulted in decreased neovascularization, adverse remodeling, and diminished cardiac function. This is consistent with a report by Hiasa et al. (20) demonstrating that antagonism of VEGF in hearts injected with bone marrow-derived mononuclear cells resulted in adverse remodeling and larger infarcts. Vascular endothelial growth factor may act therapeutically by reducing cardiomyocyte apoptosis and promoting cell proliferation and migration early after MI (21). The ability of the transplanted MDSCs to release VEGF immediately into the ischemic milieu early after infarction could theoretically help to salvage at-risk myocardium and result in reduced infarct size, a finding that was observed in this study. In addition, the persistent expression of VEGF by transplanted cells for up to 12 weeks could lead to a reduction of ischemia in the infarct through the induction of angiogenesis, and thereby attenuate progressively deleterious remodeling of viable infarct and myocardial tissue (1,22).
Interestingly enough, we did not observe any significant improvement in cardiac remodeling or cardiac function when we overexpressed VEGF in comparison with our control MDSC-injected hearts. This differs from reports by Suzuki et al. (23) wherein they observed greater improvements in cardiac function after transplantation of VEGF overexpressing myoblasts when compared with control myoblasts. A number of differences between these studies could account for this discrepancy in functional outcome, including the cell type (mouse MDSC vs. rat myoblast), method for VEGF transgene delivery to cells (retroviral transduction vs. plasmid), VEGF dosage, and animal MI model (mouse vs. rat; cell injection immediately after ligation vs. cell injection 1 h after ligation) (1,23).
Host-derived and donor cells could both be contributing VEGF to the ischemic milieu. Host-derived cells known to secrete VEGF in the setting of an MI include mobilized c-kit cells from the bone marrow (24), myofibroblasts (25), peripheral blood mononuclear cells (26), endothelial cells lining the blood vessels (1,27), smooth muscle cells (28), infiltrating macrophages (28), and cardiomyocytes (28). In addition to the secretion of VEGF from host cells, we suspect that engrafted donor cells release a significant amount of VEGF to the ischemic milieu in cell-injected hearts. In our previous study, we revealed that engrafted MDSCs express VEGF within infarcted myocardium (1). Here, we further support this by demonstrating that MDSCs secrete VEGF in response to hypoxia and cyclic stretch, conditions likely experienced by the cells after transplantation into an acute MI. A similar response to these stresses has been documented by other cell types (11,19,29). Collectively, these findings suggest that the ischemic milieu of a myocardial infarct is an important trigger for VEGF production by engrafted cells, and that transplanted cells are a significant source of VEGF within the ischemic myocardium. However, because sFlt1 secreted from donor MDSCs in this study could theoretically block endogenous (host cell-expressed) and exogenous (donor cell-expressed) VEGF, we could not precisely determine whether the VEGF secreted directly from the engrafted cells was solely or partially responsible for the angiogenic and therapeutic effects elicited after cell injection into an MI. Future experiments designed to specifically inhibit donor cell expression of VEGF will be performed to address this issue.
This study highlights the angiogenic and therapeutic value of VEGF in the setting of an acute MI treated by an intramyocardial transplantation of MDSCs. The findings of this study signify an important relationship between VEGF, angiogenesis, and functional recovery for cardiac cell therapy.
The authors wish to thank Ryan Sauder and David Humiston for their excellent editorial assistance with this manuscript and Maria Branca for her outstanding technical support.
This work was supported, in part, by grants from the National Institutes of Health (1U54AR050733-01 [to Dr. Huard], R01-HL 069368 [to Dr. Huard], R01HL65219-04), Muscular Dystrophy Association, Pittsburgh Tissue Engineering Initiative, the Donaldson Chair, the Hirtzel Foundation, the Henry J. Mankin Chair, and the American Heart Association (0315349U [to Dr. Payne]). This investigation was conducted in a facility constructed with support from Research Facilities Improvement Program Grant C06 RR-14489 from the National Center for Research Resources, National Institutes of Health.
- Abbreviations and Acronyms
- fractional shortening
- fast skeletal myosin heavy chain
- left ventricle/ventricular
- muscle-derived stem cell
- myocardial infarction
- nuclear-localized LacZ
- phosphate-buffered saline
- soluble Flt1
- vascular endothelial growth factor
- Received January 5, 2007.
- Revision received April 3, 2007.
- Accepted April 22, 2007.
- American College of Cardiology Foundation
- Leobon B.,
- Garcin I.,
- Menasche P.,
- Vilquin J.T.,
- Audinat E.,
- Charpak S.
- Reinecke H.,
- MacDonald G.H.,
- Hauschka S.D.,
- Murry C.E.
- Reinecke H.,
- Minami E.,
- Poppa V.,
- Murry C.E.
- Dowell J.D.,
- Rubart M.,
- Pasumarthi K.B.,
- Soonpaa M.H.,
- Field L.J.
- Gnecchi M.,
- He H.,
- Noiseux N.,
- et al.
- Qu-Petersen Z.,
- Deasy B.,
- Jankowski R.,
- et al.
- Lee R.J.,
- Springer M.L.,
- Blanco-Bose W.E.,
- Shaw R.,
- Ursell P.C.,
- Blau H.M.
- Hojo Y.,
- Ikeda U.,
- Zhu Y.,
- et al.
- Rehman J.,
- Traktuev D.,
- Li J.,
- et al.