Author + information
- Received November 13, 2008
- Revision received March 23, 2009
- Accepted April 2, 2009
- Published online July 1, 2009.
- Hiroko Fujii, MD, PhD⁎,
- Zhuo Sun, MD⁎,
- Shu-Hong Li, MD, MSc⁎,
- Jun Wu, MD⁎,
- Shafie Fazel, MD, PhD⁎,
- Richard D. Weisel, MD⁎,
- Harry Rakowski, MD†,
- Jonathan Lindner, MD‡ and
- Ren-Ke Li, MD, PhD⁎,⁎ ()
Reprint requests and correspondence:
Dr. Ren-Ke Li, Toronto General Research Institute, MaRS Centre, Toronto Medical Discovery Tower, Room 3-702, 101 College Street, Toronto, Ontario, Canada M5G 1L7
Objectives This study evaluated the capacity of ultrasound-targeted microbubble destruction (UTMD) to deliver angiogenic genes, improve perfusion, and recruit progenitor cells after a myocardial infarction (MI) in mice.
Background Angiogenic gene therapy after an MI may become a clinically relevant approach to improve the engraftment of implanted cells if targeted delivery can be accomplished noninvasively. The UTMD technique uses myocardial contrast echocardiography to target plasmid gene delivery to the myocardium and features low toxicity, limited immunogenicity, and the potential for repeated application.
Methods Empty plasmids (control group) or those containing genes for vascular endothelial growth factor (VEGF), stem cell factor (SCF), or green fluorescent protein (to visualize gene delivery) were incubated with perflutren lipid microbubbles. The microbubble-deoxyribonucleic acid mixture was injected intravenously into C57BL/6 mice at 7 days after coronary artery ligation (MI). The UTMD technique facilitated transgene release into the myocardium. Twenty-one days after MI, myocardial perfusion and function were assessed by contrast echocardiography. Protein expression was quantified by Western blot and enzyme-linked immunosorbent assay. Flow cytometry quantified progenitor cell recruitment to the heart. Blood vessel density was evaluated immunohistochemically.
Results Green fluorescent protein expression in the infarcted myocardium demonstrated gene delivery. Myocardial VEGF and SCF levels increased significantly in the respective groups (p < 0.05). The physiologic impact of VEGF and SCF gene delivery was confirmed by increased myocardial recruitment of VEGF receptor 2– and SCF receptor (c-kit)–expressing cells, respectively (p < 0.05). Consequently, capillary and arteriolar density (Factor VIII and alpha-smooth muscle actin staining), myocardial perfusion, and cardiac function were all enhanced (p < 0.01 relative to control group) in recipients of VEGF or SCF.
Conclusions Noninvasive UTMD successfully delivered VEGF and SCF genes into the infarcted heart, increased vascular density, and improved myocardial perfusion and ventricular function. The UTMD technique may be an ideal method for noninvasive, repeated gene delivery after an MI.
Cell transplantation offers the potential to restore ventricular function after a myocardial infarction (MI). Unfortunately, this treatment was less effective in initial clinical trials (1,2) than in pre-clinical animal studies (3), due in part to limited engraftment of the implanted cells. Angiogenic gene therapy (4–7) might enhance the benefits of cell therapy by preparing the myocardium for cell delivery, but will require a clinically effective, noninvasive technique that allows repeated, targeted delivery. Viral vectors for gene therapy are associated with safety concerns (8); for example, viral proteins elicit an immune response within the target tissue (8,9) and can cause an intense inflammatory activation of endothelial cells (10). Nonviral delivery vehicles, including plasmids, avoid these concerns but exhibit lower transfection efficiency and induce only transient expression of the gene product (11). However, we recently demonstrated that transient up-regulation of vascular endothelial growth factor (VEGF) was associated with a prolonged (6 months) increase in vessel density and perfusion (12).
Ultrasound-targeted microbubble destruction (UTMD) provides a noninvasive method to effectively deliver drugs, proteins, and plasmids to targeted organs (13,14). UTMD plasmid gene delivery uses myocardial contrast echocardiography (MCE) and features low toxicity, limited immunogenicity, and the potential for repeated, targeted application. The current study evaluated UTMD delivery of plasmid DNA-containing genes for VEGF and stem cell factor (SCF) after an MI in mice.
Preparation of plasmid DNA solution
Three plasmids (all with cytomegalovirus promoter) were used: pCEP4, containing the VEGF165 gene (human); pcDNA3, containing the SCF gene (murine); and pEGFP-N2, containing the green fluorescent protein (GFP) gene (to determine in vivo transfection efficiency). Empty plasmid pcDNA3.1 was used as a control group. The plasmid deoxyribonucleic acid (DNA) solution was prepared as described in the Online Appendix.
Preparation of microbubble-DNA solution
The microbubble-DNA solution (25 μl of activated perflutren lipid microbubbles [DEFINITY, Lantheus Medical Imaging, Billerica, Massachusetts], 25 μl of phosphate-buffered saline [PBS], and 20 μl of DNA solution) was prepared as detailed in the Online Appendix. Briefly, the microbubbles were first activated using a Vialmix (Lantheus Medical Imaging) apparatus. Next, the activated microbubbles were diluted with PBS (1:1 ratio). Finally, the plasmid DNA solution was added to the mixture. The microbubble-DNA solution was incubated in an Eppendorf tube for 20 min with gentle shaking.
In vitro determination of % plasmid DNA associated with the microbubbles
The microbubble-DNA solution (prepared using 12 μg of plasmid DNA) was allowed to stand in the Eppendorf tube in a vertical position for 30 min after the 20-min incubation in order to facilitate separation of the microbubble and liquid phases. After standing, the solution separated into 2 phases: an upper, milky-white layer containing microbubbles and a lower, clear layer containing almost no microbubbles. The quantity of plasmid DNA was measured separately for each layer as described in the Online Appendix, and then the % of plasmid DNA isolated from the microbubbles in the milky layer was calculated using the following formula: [(DNA in milky layer)/(DNA in milky layer + DNA in clear layer)] × 100.
We used female C57BL/6 mice (age 8 to 10 weeks; average mouse body weight 19 to 22 g). All animal experimental procedures were approved by the Animal Care Committee of the Toronto General Research Institute, according to the Guide for the Care and Use of Laboratory Animals.
MI and UTMD
MI was induced in female mice as we previously described (15). Seven days later, mice with an infarct wall length of between 25% and 35% of the left ventricular (LV) circumference were randomly separated into 4 groups: control group (empty plasmid DNA), VEGF group, SCF group, and GFP group. Gene therapy was administered at 7 days after injury in order to separate the effect of the gene therapy from the acute inflammatory response of the host. In each mouse, the microbubble-DNA solution (prepared using 0.6 mg of plasmid DNA/kg of mouse body weight) was infused into the tail vein over a period of 1 min, immediately after the 20-min incubation. After the infusion, UTMD was carried out with an Acuson sequoia C256 system (Siemens Medical Solutions Inc., New York, New York). An ultrasound beam delivered with a 15L8 transducer (8 MHz, mechanical index = 1.6, depth = 2 cm) was directed to the heart for 20 min in an intermittent mode (1 burst of ultrasound every 500 ms), and the heart was scanned repeatedly from base to apex (16). Contrast Pulse Sequencing technology (Cadence, Siemens) was used to detect microbubble destruction.
Evaluation of myocardial GFP expression
We used a plasmid containing the gene for GFP to establish the proof of concept that UTMD can facilitate transfection of the plasmid into the heart. To confirm gene transfection, cells expressing GFP were immunohistochemically identified in hearts collected at 5 days after UTMD, using an antibody against GFP. The heart tissue was fixed in 4% paraformaldehyde for 2 days, and then incubated at 4°C in an increasing gradient of sucrose (10% and 20% for 1 h each; 30% overnight). The tissue was then frozen in optimal cutting temperature Tissue-Tek (Sakura Finetek USA Inc., Torrance, California) and cut into 10 μm-thick sections. The samples were fixed in acetone at –20°C for 10 min, washed, incubated in 0.05% Triton X-100 in PBS for 5 min, and then blocked with 1% to 2% bovine serum albumin and 1% to 2% goat serum in PBS. Each sample was incubated with anti-GFP antibody (1:200, Alexa Fluor, Molecular Probes, Invitrogen Canada Inc., Burlington, Ontario, Canada) for 1 h at room temperature. After washing, nuclei were stained with Hoechst 33342 (1:100). The slides were washed twice with PBS (10 min each), and representative sections from the infarct, border zone, and remote (normal) myocardium were examined under a confocal microscope. The border between infarct and remote regions is clearly identifiable in fixed myocardial tissue (the infarcted myocardium appears visibly thinner and paler in color than the normal tissue). In the tissue sections, the infarct and border zone were identified using hematoxylin and eosin staining for basic histological structures on the preceding serial section.
Evaluation of perfusion using MCE
Myocardial perfusion was assessed 7 days after MI (before UTMD) and 21 days after MI (14 days after UTMD). A microbubble solution diluted with an equal volume of PBS (50 μl) was infused into the tail vein, and imaging was performed at the midpapillary muscle level with the 15L8 transducer set at 8 MHz (mechanical index = 1.6). The microbubble signal intensities were determined by recording images in an intermittent mode (every 50 to 150 ms pulsing interval until the rise in intensity, and then every 550 to 650 ms pulsing interval for the plateau of intensity). Rise intensity was calculated (difference between intensities at 50 and 150 ms, divided by 150 – 50 ms) as a measure of myocardial blood flow velocity. In this study, a 550-ms pulsing interval corresponded approximately to the plateau intensity (17); the anterior wall represented the infarct region, and the posterior wall represented the remote region. The intensity ratio was calculated (signal intensity in the anterior wall divided by that in the posterior wall) to estimate myocardial blood flow volume. The data were analyzed offline using the digital images.
Evaluation of cardiac function by echocardiography
Cardiac function was evaluated 7 days after MI (before UTMD) and 21 days after MI (14 days after UTMD). Mice were sedated with ketamine (100 mg/kg) and xylazine (10 mg/kg). In each mouse, echocardiographic images were recorded as previously described (18) with the 15L8 transducer set at 13 MHz. Left ventricular end-diastolic diameter, end-systolic diameter, end-diastolic area (LVEAd), and end-systolic area (LVEAs) were measured. Left ventricular end-diastolic volume (LVEDV) and left ventricular end-systolic volume (LVESV) were calculated using the following formulas: LVEDV = 1.047 × LVEDd3 and LVESV = 1.047 × LVEDs3. Percent ejection fraction (%EF) and fractional area change (%FAC) of the LV were calculated as follows: %EF = [(LVEDV − LVESV)/LVEDV] × 100; %FAC = [(LVEAd − LVEAs)/LVEAd] × 100 (15,19).
Measurement of infarct wall length and thickness
Infarct wall length and thickness were measured 14 days after UTMD using echocardiography and compared among VEGF, SCF, and control groups. Briefly, images from the B-mode were collected at the LV short axis view of the papillary muscle level at the end-diastolic phase. The infarct wall was identified as the area of reduced wall motion (akinetic region). Infarct wall length was measured as the endocardial length of the akinetic region and expressed as a percentage of the LV circumference. Infarct wall thickness (in mm) was measured across the center of the infarct. Because all animals were pre-selected for similar infarct sizes, the relative group differences measured were not likely due to pre-existing differences in the dimensions of individual infarcts.
Quantification of blood vessel density
Mean vascular density per 0.2 mm2 was quantified immunohistochemically (Factor VIII- or alpha-smooth muscle actin [SMA]–positive structures) in the infarct region (identified using hematoxylin and eosin staining for basic histological structures on the preceding serial section) of heart sections obtained at 14 days after UTMD, as detailed in the Online Appendix (15).
Quantification of myocardial VEGF and SCF protein levels
VEGF and SCF protein levels were quantified (using Western blot analysis or a Mouse SCF Immunoassay kit, respectively) separately in the infarct and remote regions of LV samples collected at 14 days after UTMD, as detailed in the Online Appendix. The VEGF levels were normalized to the levels of housekeeping protein glyceraldehyde 3-phosphate dehydrogenase and expressed as VEGF/glyceraldehyde 3-phosphate dehydrogenase band intensity ratios.
Quantification of progenitor cells in the myocardium
Heart samples collected at 5 days after UTMD were washed, minced into fine pieces, and digested in 0.2% collagenase solution. Flow cytometry was used to quantify the number of VEGF receptor 2 (VEGF-R2)– and c-kit–expressing cells using anti–VEGF-R2 and anti–c-kit antibodies (BD Biosciences, Mississauga, Ontario, Canada). After immunolabeling, the cells were fixed with 2% paraformaldehyde and analyzed using a Cytomics FC500 MPL flow cytometer (Beckman Coulter Inc., Fullerton, California) with CXP acquisition FC500 software.
Data are expressed as mean ± standard error. Analyses were performed using SPSS software version 12.0, with the critical alpha level set at p < 0.05. Statistical tests were as follows: comparisons between 2 groups (VEGF or SCF levels, VEGF-R2– or c-kit–positive cell counts) were made using Student t tests. Comparisons among 3 groups (vascular density, cardiac function, infarct size) were made using 1-way analyses of variance. Repeated measures analyses of variance tested the main effects and interactions of gene treatment and time after UTMD on myocardial perfusion (rise intensity, intensity ratio) and cardiac function (%FAC, %EF). When F values were significant, differences between the groups were specified with Tukey's multiple comparison post-tests.
Efficiency of plasmid DNA delivery
An in vitro analysis performed by separating the microbubble and liquid phases of the mixture before injection indicated that 72.7 ± 10.8% (n = 5) of total plasmid DNA was contained in the microbubble layer. Five days after UTMD, GFP expression identified in the hearts of mice that received microbubble/plasmid solution containing the GFP gene confirmed successful in vivo transfection using the microbubble technique (Fig. 1) Using this approach, we determined that the plasmid was able to enter cells (with the assistance of ultrasound energy) and induce functional gene expression in the target tissue.
Myocardial VEGF and SCF protein levels
Myocardial expression levels of VEGF and SCF were quantified 14 days after UTMD. VEGF protein levels in the whole heart and the infarct region (n = 7 for control group, n = 7 for VEGF) were significantly greater (p < 0.05) in the VEGF than in the control group (Fig. 2), and whole heart and remote region SCF levels (n = 7 for control group, n = 9 for SCF) were greater (p < 0.05) in the SCF than in the control group (Fig. 3). Interestingly, VEGF was most abundant in the infarct region, whereas SCF was most abundant in the remote region of the heart.
Progenitor cell recruitment to the treated hearts
Five days after UTMD, VEGF-R2– and c-kit–expressing cells were quantified in the treated hearts by flow cytometry (n = 5 for control group, n = 7 for VEGF, n = 7 for SCF). In VEGF- or SCF-treated hearts, the increased myocardial expression of VEGF or SCF was associated with a greater recruitment to the heart (compared with control group) of VEGF-R2– or c-kit–expressing cells, respectively (p < 0.05 for both groups) (Fig. 4).
Vascular density and myocardial perfusion
Fourteen days after UTMD, vascular structures were identified in the infarct region by immunohistochemical staining using antibodies against α-SMA (to identify the smooth muscle coating of arterioles, n = 5 per group) and Factor VIII (to identify all endothelium-lined structures, n = 5 per group) (Figs. 5A and 5B). More α-SMA– and Factor VIII–positive structures were observed in the infarct region of VEGF- and SCF-treated hearts than in those of control hearts (p < 0.01). Surprisingly, both types of structures were most abundant (p < 0.01) in the SCF group (compared with the VEGF group) (Figs. 5C and 5D).
Left ventricular myocardial perfusion was evaluated by MCE before and 14 days after UTMD (n = 9 for control group, n = 10 per group for VEGF, SCF) (Fig. 6). We observed a significant interaction effect (p < 0.01) between gene treatment and time after UTMD. In agreement with the immunohistochemical data, myocardial rise intensity, which represents blood flow velocity, increased remarkably in the SCF group (p < 0.01 relative to VEGF and control groups). Intensity ratios, which represent blood flow volume, increased significantly (p < 0.01) in both VEGF and SCF groups compared with the control group, but again, the improvement was significantly greatest in the SCF group (p < 0.01 relative to the VEGF group).
Cardiac function and infarct morphometry
Cardiac function (n = 10 per group) was evaluated by echocardiography before and 14 days after UTMD (7 and 21 days after MI). We monitored heart rate in all animals at both time points and recorded no significant differences among groups or between time points (mean heart rates in beats/min: control group = 199 ± 17 at day 0, 245 ± 11 at day 14; VEGF = 251 ± 29 at day 0, 230 ± 20 at day 14; SCF = 225 ± 16 at day 0, 254 ± 43 at day 14). Before UTMD therapy, measures of systolic function (%FAC, %EF) were similar among the 3 groups. However, we observed a significant interaction effect (p < 0.01) between gene treatment and time after UTMD. Whereas %FAC and %EF decreased over the course of 14 days in the control group, significant improvements were registered over the same time period in both VEGF (p < 0.01 for both measures) and SCF (p < 0.05 for %FAC, p < 0.01 for %EF) groups relative to the control group (Figs. 7A to 7D).
At the end of the study, the control group exhibited characteristic infarct thinning and expansion (lengthening). We found that infarct wall lengths were preserved in the VEGF and SCF groups compared with the control group (p < 0.01 for both groups), and were further preserved (p < 0.01) in the SCF group compared with the VEGF group. Both VEGF and SCF groups also exhibited greater preservation of infarct wall thickness than the control group (p < 0.01 for both groups) (Figs. 7E and 7F).
The UTMD technique provides a noninvasive means to deliver naked DNA to the heart. In the present study, both VEGF and SCF were transferred into the infarcted myocardium, the products of gene translation were identified, and the elevated protein levels were associated with improved perfusion, greater vascular density, and enhanced cardiac function after an MI.
After encouraging pre-clinical studies of angiogenic gene therapy, clinical trials produced disappointing results (20,21), possibly due to incomplete or inaccurate gene delivery. Early clinical trials of cell transplantation demonstrated improved perfusion (1), but limited cell survival may have diminished the benefits of this approach for cardiac restoration. UTMD is a noninvasive, targeted delivery technique that may induce sufficient angiogenesis to increase cell engraftment after transplantation. Potentially, infarcted myocardium could be prepared to support the engraftment of injected and recruited cells after an MI by pre-treating the damaged tissue using UTMD gene treatments repeated over a 3- to 5-day period.
We used perflutren lipid microbubbles (the contrast agent used for MCE) to deliver plasmid DNA to the myocardium. These microbubbles are electrically neutral in their unmodified state. However, we demonstrated that diluting the activated microbubbles with PBS (1:1 ratio) generated a net positive charge in the solution that facilitated DNA attachment to the lipid microbubbles (H. Fujii et al., unpublished data, 2009). The mechanism by which the plasmids bind to the microbubble shell is unclear and will require a thorough evaluation, but our in vitro study indicated that more than 72% of the plasmid DNA was associated with the microbubbles. Future studies should compare lipid- and albumin-based microbubbles to determine the optimal type for gene delivery; however, perflutren lipid microbubbles have a high membrane affinity that may facilitate gene transfer (14). In addition, UTMD may induce transient endothelial injury, which could facilitate the transit of the plasmid DNA into myocardial tissue (22,23). In our model, treatment with naked plasmid and UTMD did not cause a decline in cardiac function, suggesting that minor injury to the endothelium did not produce a functional consequence. The UTMD technique did, however, induce significant increases in the levels of GFP, VEGF, and SCF in the gene recipients. And although additional regional assessments are required to confirm our results in larger animal models because the small rodent model is limited by the potential for variations in functional measurements, our data suggest that even a single treatment was sufficient to produce an improvement in global function.
Although we did not calculate gene transfection efficiency in vivo, our studies with GFP-carrying plasmids allowed us to visualize the cells that accepted the delivered gene. We did not determine specifically which cell types were transfected by the UTMD technique; however, significant GFP expression was identified in fibroblasts and within the walls of blood vessels in both the infarct and remote regions. These findings suggest that the UTMD facilitates gene delivery not only to blood vessels but also to deeper tissues. It is unclear why SCF expression was principally increased in the remote region, whereas VEGF was principally increased in the infarct region, but comparing effects in the border zone with those in the infarct would allow a more precise localization of gene delivery.
Beyond our qualitative evaluation of cardiac perfusion via contrast echocardiography, we also completed a quantitative immunohistochemical assessment of blood vessel density that identified significant local effects of VEGF and SCF overexpression. These findings are in agreement with our previous report, in which cells transfected with the VEGF gene promoted angiogenesis after they were injected into the myocardium (24). Our histological findings (α-SMA and Factor VIII staining) indicated more vascular structures in the SCF compared with the VEGF group; rise intensity was also highest in the SCF group. The reason for these effects is presently unknown, but suggests a critical role for SCF in mediating flow velocity through arteriogenesis. The effect may be driven by the recruited c-kit–expressing cells, which we have previously shown are critical to post-MI angiogenesis (15). Future studies might include an accurate characterization (using multiple surface markers) of the recruited progenitor cells and their origins and should also determine the benefits of combined gene therapy or the delivery of other genes, such as fibroblast growth factor, which might have clinical utility. In this study, because we administered gene therapy at 7 days after MI, the improvements in global ventricular function may have been related to the prevention of adverse remodeling. However, UTMD might effectively be used earlier after MI to mitigate reperfusion injury or rescue myocardial tissue.
To facilitate the delivery of plasmid DNA into the myocardial tissue, we used a high mechanical index to destroy the microbubbles in an intermittent mode (1 burst ∼ every 4 to 5 heartbeats). This timing was selected to allow the microbubbles to fill the capillaries during each pulsing interval and to minimize endothelial damage during UTMD. Our results suggest that these settings were effective for in vivo gene delivery and produced no obvious myocardial/endothelial injury by histological or functional examination. Additional experiments will confirm the optimal concentrations of microbubbles and plasmid DNA for gene delivery and the optimal ultrasound settings, and may also examine the location of the approach's effects in normal heart tissue.
The ultrasound settings were adjusted not only to maximize microbubble destruction but also to optimize the evaluation of myocardial perfusion, which was accomplished using both early and late phase plateau intensities. We also reduced the risk for endothelial injury (22,23) by maintaining a high intensity with a short pulsing interval. Higher frequencies are required for high-quality perfusion images, but lower frequencies are required to destroy the microbubbles: we therefore used a mechanical index of 1.5 to 1.6, which is the highest setting in clinical use, and set the frequency at 8 MHz, which is the lowest frequency on this transducer. This combination allowed us to deliver the genes to the heart and detect significant differences in perfusion despite variability between groups and low signal intensity. Improved echocardiographic techniques will enhance the quality of the diagnostic images in both small and larger animal models while maintaining adequate gene delivery. Quantitative measures of myocardial perfusion will also be required before the clinical application of UTMD.
In summary, we successfully delivered VEGF and SCF genes into the infarcted heart using UTMD. We cannot exclude the possibility that inflammation resulting from vector delivery contributed to endogenous VEGF/SCF expression and increased angiogenesis, but plasmid delivery was associated with significantly greater vascular density, protein expression, and progenitor cell homing in gene recipients relative to controls, with a net effect of improved myocardial perfusion and function. Our findings further suggest that the SCF gene plays an important role in arteriogenesis. Perhaps by inducing a greater influx of c-kit–expressing cells than VEGF, SCF may have the greater potential for myocardial restoration.
The authors thank Heather McDonald Kinkaid for her assistance with manuscript preparation, editing, and revision.
For supplementary methods, please see the online version of this article.
Ultrasound-Targeted Gene Delivery Induces Angiogenesis After a Myocardial Infarction in Mice
This research was supported by grants from the Canadian Institutes of Health Research (CIHR; RMF82498) to Dr. Li from the CIHR (MOP14795, MOP86661) and to Dr. Weisel from the Heart and Stroke Foundation of Ontario (T6148). Dr. Li is a Career Investigator of the Heart and Stroke Foundation of Canada and holds a Canada Research Chair in Cardiac Regeneration.
- Abbreviations and Acronyms
- ejection fraction
- fractional area change
- green fluorescent protein
- left ventricle
- left ventricular end-diastolic area
- left ventricular end-systolic area
- left ventricular end-diastolic volume
- left ventricular end-systolic volume
- myocardial contrast echocardiography
- myocardial infarction
- phosphate-buffered saline
- stem cell factor
- smooth muscle actin
- ultrasound-targeted microbubble destruction
- vascular endothelial growth factor
- Received November 13, 2008.
- Revision received March 23, 2009.
- Accepted April 2, 2009.
- American College of Cardiology Foundation
- Meyer G.P.,
- Wollert K.C.,
- Lotz J.,
- et al.
- Losordo D.W.,
- Vale P.R.,
- Symes J.F.,
- et al.
- Patterson C.,
- Runge M.S.
- Chen S.,
- Shohet R.V.,
- Bekeredjian R.,
- Frenkel P.,
- Grayburn P.A.
- Bekeredjian R.,
- Chen S.,
- Frenkel P.A.,
- Grayburn P.A.,
- Shohet R.V.
- Bekeredjian R.,
- Grayburn P.A.,
- Shohet R.V.
- Wei K.,
- Jayaweera A.R.,
- Firoozan S.,
- Linka A.,
- Skyba D.M.,
- Kaul S.
- Fujii H.,
- Tomita S.,
- Nakatani T.,
- et al.
- Henry T.D.,
- Annex B.H.,
- McKendall G.R.,
- et al.
- Teupe C.,
- Richter S.,
- Fisslthaler B.,
- et al.