Author + information
- Received December 4, 2006
- Revision received March 22, 2007
- Accepted March 28, 2007
- Published online July 17, 2007.
- David M. Kaye, MD, PhD⁎,1,2,⁎ (, )
- Arthur Preovolos, BS⁎,1,3,
- Tanneale Marshall, BS⁎,
- Melissa Byrne, PhD⁎,1,3,
- Masahiko Hoshijima, PhD†,
- Roger Hajjar, MD‡,
- Justin A. Mariani, MD⁎,
- Salvatore Pepe, PhD⁎,
- Kenneth R. Chien, MD, PhD‡ and
- John M. Power, PhD⁎,1,2
- ↵⁎Reprint requests and correspondence:
Prof. David M. Kaye, Wynn Department of Metabolic Cardiology, Baker Heart Research Institute, P.O. Box 6492, St. Kilda Road Central, Melbourne, VIC 8008, Australia.
Objectives The purpose of this study was to develop a clinically applicable high-efficiency percutaneous means of therapeutic gene delivery to the failing heart.
Background Substantial advances in the understanding of the cellular and molecular basis of heart failure (HF) have recently fostered interest in the potential utility of gene and cell therapy as novel therapeutic approaches. However, successful clinical translation is currently limited by the lack of safe, efficient, and selective delivery systems.
Methods We developed a novel percutaneous closed-loop recirculatory system that provides homogeneous myocardial delivery for gene transfer in the failing large animal heart. After 4 weeks’ rapid pacing in adult sheep to induce HF, the animals were randomly allocated to receive either adenovirus expressing a pseudophosphorylated mutant (AdS16E) of phospholamban (PLN) or Ad–β-galactosidase (AdLacZ).
Results Two weeks after gene delivery, in the presence of continued pacing, left ventricular (LV) ejection fraction had significantly improved in the AdS16E-treated animals (27 ± 3% to 50 ± 4%; p < 0.001), whereas a further decline occurred in the AdLacZ group (34 ± 4% to 27 ± 3%; p < 0.05). In conjunction, AdS16E delivery resulted in significant reductions in LV filling pressures and end-diastolic diameter (both p < 0.05). In conjunction, AdS16E-treated animals showed significant improvement in the expression of PLN and Ca2+-adenosine triphosphatase activity. In separate animals, recirculating AdLacZ delivery was shown to achieve superior myocardial gene expression in contrast to intracoronary delivery and was associated with lower systemic expression.
Conclusions We report the development of a novel closed-loop system for cardiac gene therapy. Using this approach delivery of AdS16E reversed HF progression in a large animal HF model.
The development of new technology for the local organ-restricted delivery of molecular-based therapeutic agents would represent a major advance in the design of biologically targeted therapy by achieving satisfactory local concentrations without the potential for systemic toxicity. Toward this end, the convergence of device technology and biologic therapy has recently become a major focus for research and development in the translational area. For example, the development of drug-eluting stents for the prevention of vascular restenosis represents a prime example of the combination of a drug with a device, in this case to selectively inhibit the proliferation of intimal cells within the device itself.
Although some disease paradigms may require very localized delivery, other disease processes may require more homogeneous high-efficiency delivery to specific organs. Heart failure (HF) is a common clinical cardiovascular disorder, characterized by complex pathophysiology and by substantial morbidity and mortality. While the key underlying mechanism in HF is contractile failure of the myocardium, the only currently available pharmacotherapeutic strategies that prolong life are somewhat indirectly acting agents, in particular directed at the sympathetic nervous system and renin-angiotensin system (1). Thus, although the cellular and molecular causes of myocardial failure are well understood, current therapies do not directly target these disorders. Accordingly, in HF there is a substantial interest in the potential role of gene therapy or in therapies that control gene expression as a therapeutic tool in the treatment of HF, given that the extensive identification of the molecular deficiencies of the failing heart provides a logical basis for the application of specific gene therapy (2–4).
In particular, in HF an extensive body of data indicate that key defects in the regulation of intracellular Ca2+play a pivotal role in the development and progression of contractile failure. In both experimental models and clinical heart failure, it has been reported that expression and activity of the sarcoplasmic reticulum Ca2+-adenosine triphosphatase (ATPase) type 2 (SERCA2) is markedly reduced, typically in conjunction with an increase in the expression and the activity of the regulatory protein phospholamban (PLN) (5–7). Together, these defects provide a clear molecular basis for the observations of reduced sarcoplasmic reticulum (SR) Ca2+release and diastolic Ca2+overload in cardiomyocytes from failing myocardium (8,9). In support of these observations, studies performed in transgenic animals or by using gene-transfer strategies in small animals confirm the role of these molecular defects in causing myocardial failure (10,11).
The major current limitation for the clinical translation of gene therapy in HF is the development of suitable delivery tools (12,13). Preferably, an ideal delivery mechanism would provide safe, homogeneous myocardial delivery with limited systemic “spillover” or expression and would require as little delivery vector as possible. In the present study, we report the therapeutic delivery of an adenoviral vector encompassing a pseudophosphorylated mutant of PLN, using a novel closed-loop percutaneous catheter-based recirculating cardiac perfusion system in large animals with HF.
Recirculating delivery of myocardial perfusate and detection of viral distribution through fluorescent imaging
Recirculating perfusate delivery was achieved with the use of a novel cardiac perfusion circuit. Under fluoroscopic guidance, coronary venous blood was recaptured from the coronary sinus with the use of a percutaneously positioned occlusive balloon recovery catheter. The draining catheter was placed in such a position to exclude the azygous vein, either by occluding it with the balloon or beyond the entry point of the azygous vein. Further, the recovery catheter incorporated a structural element to prevent dynamic collapse during the application of suction (VFocus, VKardia Inc., Minneapolis, Minnesota). Venous return was facilitated by the use of a roller pump, followed by reoxygenation using an oxygenator membrane. Oxygenated perfusate was then directed to the left coronary territory via a nonocclusive catheter placed percutaneously in the left main coronary artery. Right coronary artery cannulation was not performed, given that in the sheep this vessel is generally small, with a limited myocardial distribution. Optimization of pump flow is determined by progressively increasing the roller pump speed to achieve a pump head pressure of −80 to −100 mm Hg on the venous side, which we demonstrated as corresponding to a coronary sinus pressure of 0 to 5 mm Hg. At the conclusion of the recirculation period, blood continued to be removed from the coronary sinus for an additional 2 min and this, together with the remaining circuit blood, was discarded to avoid systemic delivery of residual perfusate.
To determine the distribution of myocardial perfusate delivered in this manner we characterized the pattern of delivery of the fluorophore indocyanine green (ICG). Ten milliliters of ICG (2 mg/ml; Akorn, Decatur, Illinois) was injected into the circuit and recirculated for 10 min. Hearts were then explanted and examined using near-infrared spectroscopy (NIRS). Spectroscopic images were acquired as described in detail previously (14).
Animal procedures and gene delivery
Heart failure was induced in sheep by rapid ventricular pacing (180 beats/min) for a period of 4 weeks, as previously described (15). On the day of gene transfer and on the final study day, echocardiographic (Cypress, Acuson, Malvern, Pennsylvania) and left ventricular hemodynamic assessments (Millar catheter) were performed 1 h after rapid pacing was stopped. For gene delivery, animals were randomized to receive either adenoviral serine-to-glutamate “pseudophosphorylated” PLN mutant (AdS16EPLN; 1 × 1012vp; n = 9) or adenoviral β-galactosidase (AdLacZ; 1 × 1012vp; n = 6), delivered in the antegrade recirculating mode as described in the preceding section. Both vectors were driven by a cytomegalovirus promoter and the AdLacZ was non-nuclear localizing. Repeat hemodynamic and echocardiographic assessment was performed 2 weeks after gene transfer. At the conclusion of the studies, left ventricular samples were collected, rapidly frozen in liquid nitrogen and stored at −80°C for subsequent molecular, biologic, and biochemical analysis. In a separate cohort of normal animals, we compared the myocardial, pulmonary, and hepatic expression of β-galactosidase after intracoronary (n = 3) and recirculating (n = 3) delivery of AdLacZ (1 × 1012vp). All animal studies were performed with the approval of the Institutional Ethics Review Committee.
Histology and immunohistochemistry
For histology, paraffin-embedded sections were stained with hematoxylin and eosin and examined under light microscopy. For β-galactosidase immunohistochemistry, sections were deparaffinized in histolene and rehydrated with ethanol. Sections were then washed in 1% H2O2to block endogenous peroxidase activity. After brief rinsing, nonspecific antibody binding was suppressed by 1% normal goat serum diluted in Tris-buffered saline (TBS; 10 mmol/l Tris-HCl, pH 7.6, 150 mmol/l NaCl) with 1% bovine serum albumin (BSA). Sections were then incubated with rabbit anti–β-galactosidase antibody (Abcam, Cambridge, United Kingdom) diluted 1:1,500 in TBS with 1% BSA and 0.1% Tween overnight at 4°C. After rinsing in TBS the sections were incubated with horseradish peroxidase-conjugated goat antirabbit antibody (BioRad) diluted 1:250 in TBS with 0.1% Tween for 1 h at room temperature. After several rinsings, the bound antibody was visualized using 3,3′-diaminobenzidine tetrahydrochloride as chromogen. Control sections incubated without the primary antibody, as well as sections from nontreated animals, showed very low background.
Molecular and biochemical analyses
Expression of PLN and phospho (S16) PLN was analyzed by Western blot analysis using commercially available antibodies (Affinity Bioreagents, Golden, Colorado; and Abcam). For the determination of myocardial SERCA activity, samples were pulverized in liquid nitrogen, and Ca2+-dependent ATPase activity was determined using a pyruvate-nicotinamide adenine dinucleotide 3-step coupled reaction (16). Free Ca2+was determined as previously described (17,18). Assays were run in quadruplicate, and conditions were validated in independent reactions using the mycotoxin cyclopiazonic acid (10 μmol/l) to inhibit SR Ca2+-ATPase (19), butanedione monoxime (30 mmol/l) to inhibit potential residual levels of Ca2+-dependent actomyosin ATPase, and EGTA (4 mmol/l) to determine basal ATPase activity.
Data are presented as mean ± SEM. Between-group comparisons were performed using an unpaired ttest for normally distributed data or a Mann-Whitney test for data that were not normally distributed (as assessed by the Kolmogorov-Smirnov test). Within-group comparisons were performed using a paired ttest as appropriate. A p value <0.05 was considered to be statistically significant.
Establishment of recirculating perfusate delivery to the heart
We developed a novel recirculating technique for the selective percutaneous delivery of gene therapy to the failing heart. As shown in Figures 1Aand 1B, the coronary sinus is selectively cannulated with a balloon drainage catheter system that incorporates a retractable nitinol device which maintains venous patency during coronary sinus drainage. This venous recovery system is connected to an extracorporeal pump-oxygenator circuit, with return to the myocardium via a percutaneously placed left coronary arterial catheter for antegrade perfusate delivery. To demonstrate the homogeneity of perfusion, the pattern of distribution of indocyanine green throughout the myocardium was examined using near-infrared spectroscopy. After 10 min of recirculation, a homogeneous transmural pattern of perfusate delivery was evident in a regional pattern consistent with the perfusion territory of the left coronary arteries (Fig. 2).In conjunction, there was no significant accumulation of lactate within the circuit blood during this period (baseline vs. 10 min: 2.9 ± 0.8 mmol/l vs. 3.1 ± 0.6 mmol/l; p = NS), and circuit blood oxygen saturation was maintained at 100% throughout the procedure. The combined application of a recirculating approach to myocardial perfusion with adenoviral delivery showed no histologic evidence of myocardial inflammation or infarction (Fig. 3).
To evaluate the relative efficacy of this novel approach over intracoronary gene delivery, we examined the intensity of myocardial β-galactosidase expression in the myocardium by immunohistochemistry. In preliminary studies we compared myocardial expression of β-galactosidase after delivery of 1010versus 1012vp AdLacZ. Delivery of the lower amount resulted in minimal or undetectable levels (data not shown). Accordingly, delivery of adenovirus in the range of 1012vp was subsequently used. Subsequently, we compared intracoronary (n = 3) with recirculating delivery (n = 3) of 1 × 1012vp AdLacZ in normal sheep. These studies visually indicated greater LacZ expression in animals treated with recirculating gene delivery. In conjunction, we attempted to quantify β-galactosidase enzymatic activity in myocardial tissue samples. However, we were not able to distinguish activity distinct from that in control samples, possibly owing to the influence of endogenous β-galactosidase as has been previously reported (20). Next, we also determined the capacity of the recirculation system to limit the systemic leakage of adenovirus by investigating the expression of β-galactosidase in the liver and lungs. As shown in Figure 4,the use of the recirculating approach yielded a higher intensity of β-galactosidase expression in the myocardium and lower expression in the liver and lungs. Notably, during direct intracoronary infusion of AdLacZ there was some evidence of alveolar infiltration consistent with previous reports (21).
Echocardiographic, hemodynamic, and molecular effects of recirculating S16EPLN delivery
To evaluate the therapeutic potential of percutaneous recirculating gene delivery in the failing myocardium, we compared the effects of delivery of AdLacZ to those of AdS16EPLN in sheep with pacing-induced HF. Four weeks after the commencement of rapid ventricular pacing, echocardiographic and hemodynamic evaluation demonstrated the presence of significant ventricular dysfunction. Before gene delivery, the group mean left ventricular ejection fraction (LVEF) was 29 ±2% and the group mean left ventricular end-diastolic pressure was 24 ±3 mm Hg (n = 15). Despite a further 2 weeks’ rapid ventricular pacing, animals treated with AdS16EPLN showed significant hemodynamic improvement in conjunction with reverse ventricular remodeling, whereas animals treated with AdLacZ demonstrated progressive ventricular failure, as indicated in Figure 5.Specifically, in the AdLacZ animals there was a continued reduction in LVEF from 35 ±6% to 27 ±3% (p < 0.05), whereas in the AdS16EPLN animals the LVEF increased from 27 ±3% to 50 ±9% (p < 0.001). Consistent with the induction of a process of reverse remodeling, the left ventricular end-diastolic area fell significantly in AdS16EPLN-treated sheep (37.5 ±1.7 cm2to 32.4 ±1.7 cm2; p = 0.01), whereas a continued increase in the left ventricular end-diastolic area was observed in the AdLacZ-treated animals (36.2 ±2.6 cm2to 41.3 ±2.0 cm2; p < 0.01). Measures of diastolic function also appeared to be favorably affected by AdS16EPLN delivery. In treated animals, the left ventricular end-diastolic pressure fell from 26.3 ±2.6 mm Hg to 19.9 ±1.8 mm Hg (p < 0.05), whereas it remained unchanged in the AdLacZ group (Fig. 5). Directionally similar changes also occurred in the peak −dP/dt (AdLacZ: −1,957 ±219 mm Hg/s to −1,562 ±206 mm Hg/s; p = 0.09; AdS16EPLN: −1,572 ±166 mm Hg/s to 1,671 ±116 mm Hg/s; p = NS) and in the time constant of relaxation, tau (AdLacZ: 39.2 ±2.4 ms to 41.8 ±3.5 ms; p = NS; AdS16EPLN: 47.6 ±1.4 ms to 41.8 ±3.3 ms; p = NS). In animals receiving AdS16EPLN, the heart rate did not significantly change with gene delivery (64 ±2 beats/min to 69 ±3 beats/min). The AdLacZ-treated animals had a significantly higher baseline heart rate (p < 0.05), driven by 2 animals with a significant resting tachycardia; however, AdLacZ delivery did not significantly alter heart rate (95 ±13 beats/min to 78 ±9 beats/min).
To determine the molecular basis for the functional improvement in animals that underwent AdS16EPLN gene transfer, we measured the expression of total and phosphorylated (S16) PLN by immunoblotting and SERCA activity. As illustrated in Figure 6,immunoblot analysis demonstrated an increase in the abundance of phospho-PLN in S16E-treated animals, and, consistent with this finding, SERCA activity was significantly increased in S16EPLN HF animals compared with those treated with AdLacZ (Fig. 7).
Over the past decade, exponential growth in the molecular characterization of HF has led to the identification of a growing number of targets for disease modification. However, many of these are proteins that play critical roles in signaling in other organ systems and would require restricting their expression to the heart in vivo. In the case of HF, the rapid identification of a myriad of molecular targets also presents the same underlying issues in the quest to translate research into the development of a therapeutic tool. For HF, many of the most appropriate targets are intracellular proteins involved in the regulation of intracellular calcium and/or involvement in signaling (4). Accordingly, manipulation of the expression of these proteins is likely best addressed by a gene therapy-based approach, which for safe and effective translation into clinical practice requires the conjunction of several key elements. These include the use of a high-efficiency viral vector that has the capacity to incorporate an appropriate therapeutic gene.
Defective cellular Ca2+handling, particularly that mediated by the sarcoplasmic reticulum, is a well recognized feature of HF. This phenomenon has been reported in both clinical and experimental HF and has been variously attributed to a reduction in the expression of SERCA protein or its activity (6,22). Alterations in the amount or activity of SERCA have been shown to significantly influence the rate and amplitude of myocardial contraction as well as the rate and extent of relaxation (23). In further support of this notion, SERCA2a gene transfer has been shown to improve both systolic and diastolic function and survival in experimental HF models (24,25). In conjunction with the central role of SERCA in the pathophysiology of HF, the regulatory protein PLN is also pivotal, given its role as a major regulator of SERCA activity. In the nonphosphorylated state, PLN inhibits SERCA activity, whereas when it undergoes phosphorylation this negative regulation is relieved. As such, many studies have shown that PLN is a major determinant of cardiac contractility and of the myocardial response to β-adrenoceptor agonists (26). Given the important role of PLN as a regulator of SERCA, particular emphasis has also been placed upon the influence of the ratio of PLN to SERCA myocardial contractility (27).
In conjunction with altered SERCA expression, disordered PLN expression per se has also been suggested to possibly play a role in HF, although this remains controversial. In particular, it has been shown that PLN is relatively hypophosphorylated in HF, which contributes to the reduction in SERCA activity (28). Evidence also exists to indicate that PLN expression varies across the left ventricular wall in HF, although the precise impact of this heterogeneous pattern of expression on ventricular function is not clear (29). Consequently, several strategies to reduce the expression of PLN have been shown to improve myocardial contractility (30–32), although these findings are not consistent (33). Furthermore, several mutations of PLN, resulting in either complete ablation or alterations in the protein structure, have been accompanied by cardiomyopathy in man (34,35).
In the context of HF, myocardial gene delivery requires a safe well tolerated delivery tool that provides homogeneous tissue distribution. Preferably, such an approach would also avoid the potential adverse effects of systemic leakage with its attendant risk of systemic gene expression and an immune response to the viral capsid (36,37). Although many attempts have been made to develop techniques for somatic gene transfer in HF, many are inefficient. Current techniques with possible relevance to HF in particular range from simple intracoronary injection, to coronary sinus retroperfusion and myocardial perfusion during cardiopulmonary bypass. Although direct single bolus intracoronary injection may be the simplest approach, its limitations include the need for coadministration of potentially toxic adjuvant agents and an inherent inability to prevent systemic leakage of the delivery vector (13,38,39). Coronary sinus retroperfusion (40) has been shown to provide some myocardial gene delivery; however, it has not been evaluated in the context of HF. Of note, we have previously observed that balloon coronary sinus occlusion alone causes a rapid rise in the coronary venous pressure (unpublished observation), which would likely be poorly tolerated by the failing heart. Gene delivery to the heart during full cardiopulmonary bypass has also been reported (38), although this approach would have limited application to a broader HF population.
With these issues in mind, in the present study we developed a percutaneous closed-loop system for myocardial gene delivery which could be applied to the failing heart in large animals and ultimately in man. We hypothesized that a closed-loop recirculation system for myocardial gene delivery might achieve several desirable features. First, by avoiding peripheral systemic delivery or even single-pass intracoronary infusion the concentration of vector reaching the myocardium would be higher. In the present study we showed that the expression of a reported gene was higher during recirculation compared with intracoronary delivery. Second, we also aimed to reduce systemic delivery of vector to avoid potential deleterious effects of transgene expression in other organs. Indeed, in the present study we demonstrated a reduction in the pulmonary and hepatic expression of β-galactosidase. In conjunction, we observed some possible evidence of lung inflammation following intracoronary infusion of AdLacZ, which was not evident with closed-loop cardiac recirculation. From a practical clinical perspective, the application of this approach uses commonly performed cardiac catheterization techniques combined with standard blood perfusion methods commonly used in the cardiothoracic and intensive care settings. In the present study the recirculating approach was well tolerated in animals with evidence of moderate to severe HF.
On the basis of data suggesting that the restoration of defective control of intracellular calcium homeostasis may be beneficial in HF (5,12,25), we elected to deliver a pseudophosphorylated mutant of PLN, which has been previously shown to rescue the HF phenotype in small animals with HF (10). In the present study, we showed that the recirculating delivery of S16EPLN was able to significantly improve indices of both systolic and diastolic myocardial function in concert with a reversal of the process of ventricular remodeling that typifies progressive HF. These observations are consistent with earlier studies (10).
Although in the present study we documented that delivery of S16EPLN resulted in an increase in the ratio of phospho-PLN to total PLN, we were not able to specifically quantitate the relative abundance of endogenous phospho-PLN to S16EPLN, because of the lack of an antibody specific to S16EPLN. In the present study we elected to measure SERCA activity rather than SERCA protein expression, given that activity per se rather than protein levels contributes to myocardial function (27). In earlier studies, SERCA delivery has been shown to reduce the frequency of ventricular arrhythmias (41); however, in the present study we did not perform detailed electrophysiologic testing or telemetric monitoring.
Going forward, the likely ultimate utility of adenoviral vectors in the clinical setting is limited by their short duration of expression, inflammatory potential, and low transduction efficiency. Additionally, the immunogenic properties of the encoded protein also play a significant role in the ultimate level of expression. Accordingly, current attention is focused on the utility of recombinant adeno-associated viruses (rAAV), which demonstrate higher efficiency and greater tissue specificity (36,42–44). Despite the attraction of rAAV as vectors for human gene therapy, translation into the clinic also potentially remains limited. Although tissue specificity may be greater for specific rAAV serotypes (36), recent studies indicate that inflammatory responses may significantly limit transgene expression and elicit significant target organ damage (45). Adeno-associated virus gene expression is also increasingly understood to be limited by the prevailing level of AAV antibodies (46). In addition, rAAV are substantially more complex to generate in quantities likely required for clinical use if delivered systemically, thereby limiting possible clinical translation. Together, these potential limitations of AAV application to somatic gene transfer will also require the development of an interventional approach for targeted therapeutic delivery.
In conclusion, we report the development of a novel enabling technology for the percutaneous delivery of molecular or cellular therapy to the failing human heart.
The authors greatly appreciated the excellent technical assistance of Mr. Adam Bilney, Dr. Fabrice Prunier, Dr. Yoshi Kawase, Ms. Freya Sheehan, and Ms. Shaolini Arunogiri.
↵1 Drs. Kaye, Power, and Byrne and Mr. Preovolos are inventors of the VFocus system.
↵2 Drs. Kaye and Power are founders and stockholders in VKardia, which holds the intellectual property.
↵3 Dr. Byrne and Mr. Preovolos also are VKardia stockholders.
Supported by grants from the Atherosclerosis Research Trust (U.K.) and the National Health and Medical Research Council of Australia.
- Abbreviations and Acronyms
- heart failure
- left ventricular ejection fraction
- serine-to-glutamate “pseudo-phosphorylated” phospholamban mutant
- sarcoplasmic reticulum Ca2+-ATPase
- sarcoplasmic reticulum
- Received December 4, 2006.
- Revision received March 22, 2007.
- Accepted March 28, 2007.
- American College of Cardiology Foundation
- Hunt S.A.,
- Abraham W.T.,
- Chin M.H.,
- et al.
- Sesti C.,
- Kloner R.A.
- Kaye D.M.,
- Byrne M.,
- Alferness C.,
- Power J.
- Seidler N.,
- Jona I.,
- Vegh M.,
- Martonosi A.
- Zsengeller Z.,
- Otake K.,
- Hossain S.,
- Berclaz P.,
- Trapnell B.
- Tomaselli G.F.,
- Marban E.
- Schwinger R.H.,
- Brixius K.,
- Bavendiek U.,
- et al.
- Miyamoto M.I.,
- del Monte F.,
- Schmidt U.,
- et al.
- del Monte F.,
- Williams E.,
- Lebeche D.,
- et al.
- Luo W.,
- Wolska B.M.,
- Grupp I.L.,
- et al.
- Prestle J.,
- Dieterich S.,
- Preuss M.,
- Bieligk U.,
- Hasenfuss G.
- He H.,
- Meyer M.,
- Martin J.L.,
- et al.
- Eizema K.,
- Fechner H.,
- Bezstarosti K.,
- et al.
- Dieterle T.,
- Meyer M.,
- Gu Y.,
- et al.
- Schmitt J.P.,
- Kamisago M.,
- Asahi M.,
- et al.
- Leiden J.M.
- Davidson M.J.,
- Jones J.M.,
- Emani S.M.,
- et al.
- Raake P.,
- von Degenfeld G.,
- Hinkel R.,
- et al.
- del Monte F.,
- Lebeche D.,
- Guerrero J.L.,
- et al.
- Zhu T.,
- Zhou L.,
- Mori S.,
- et al.