Robust Adenoviral and Adeno-Associated Viral Gene Transfer to the In Vivo Murine Heart
Application to Study of Phospholamban Physiology
Background— Viral gene transfer to the whole heart in vivo has been achieved in several mammalian species but remained difficult to accomplish in murine hearts. We postulated that a key impediment derives from the use of proximal aortic occlusion during virus injection, because this eliminates coronary perfusion gradients in mice as aortic root and left ventricle pressures equalize.
Methods and Results— Pressure-volume analysis confirmed these mechanics. In contrast, descending aortic occlusion with whole-body cooling (20°C) preserved transmyocardial perfusion gradients and allowed for sustained (>10-minute) dwell times in an upper-body perfusion circuit. This approach yielded robust cardiac transfection with adenovirus (AdV) and adeno-associated virus (AAV) injected into the left ventricle cavity or more simply via a central vein. Cardio-specific expression was achieved with a myocyte-specific promotor. Optimal AdV transfection required 9-minute aortic occlusion, versus 5-minute occlusion for AAV. Using this method, we examined the in vivo function of phospholamban (PLB) by stably transfecting PLB-null mice with AAV encoding PLB (AAVPLB). AAVPLB restored PLB protein to near control levels that colocalized with SERCA2A in cardiomyocytes. At baseline, PLB-null hearts exhibited enhanced systolic and diastolic function, but frequency-dependent reserve was blunted versus wild-type controls. These properties, particularly the frequency response, returned toward control 3 months after AAVPLB transfection.
Conclusions— The new simplified approach for murine whole-heart viral transfection should assist molecular physiology studies.
Received February 18, 2003; de novo received May 6, 2003; revision received July 9, 2003; accepted August 4, 2003.
The use of viral-based vectors for myocardial gene transfer is an attractive approach for studying fundamental mechanisms of cardiac function and developing novel disease treatments.1–3 Although cardiomyocytes are good vector targets, efforts have been impeded by low transfection efficiencies and difficulties in culturing and studying mature cells.1–5 In vivo studies have generally used left ventricle (LV) cavity or aortic root virus injection during proximal (ascending) aortic constriction, with successes reported in rat, rabbit, and Syrian hamster.6–11 The addition of total body cooling has facilitated longer cross-clamp times, enhancing gene transfer in smaller animals.11,12
Despite these advances, successful translation of gene transfer methods to the murine heart has remained problematic. One problem may lie with use of proximal aortic occlusion, because this could greatly reduce aortic compliance in the mouse to effectively eliminate a transmyocardial perfusion gradient. Coronary perfusion would depend solely on manual pressures generated during injection or require sustained ventricular asystole that could heighten procedural risks. By contrast, mid-descending aortic occlusion might preserve a perfusion gradient by generating an isolated upper-body circuit and, combined with whole-body cooling, facilitate sustained clamp times to enhance gene transfer. More distal occlusion would also allow intravenous virus injection, simplifying the approach. The present study tested this methodology for both adenovirus (AdV)-mediated and adeno-associated virus (AAV)-mediated gene transfer to the in vivo murine heart. To provide proof of principal for stably altering in vivo cardiac function, we restored murine phospholamban (PLB) to mice lacking the PLB gene and provide novel insights regarding its role to the in vivo modulation of cardiac function by beat frequency.
Control mice (C57/Bl6, 2 to 4 months) and littermate SVJ129/CF-1 mice with and without a null mutation for the PLB gene (PLB−/−, 6 to 8 months) were used.
Replication-deficient recombinant adenovirus (ST5) encoding nuclear-targeted β-galactosidase (AdV β-gal) driven by a cytomegalovirus (CMV) promoter13–16 or cardiomyocyte-specific promotor (mouse α-myosin heavy chain [MHC]) was used. Recombinant AdV was plaque purified, and titer was determined by plaque assay on HEK293 cells in culture. Viral titers for AdV β-gal and AdMHC β-gal were 1.7×1012 and 1.3×1012 particles (pt)/mL, respectively. After purification, virus was suspended in PBS (pH 7.4) with 3% sucrose and kept at −80°C until use. Amplification and purification were done by the University of Iowa Gene Transfer Vector Core. Replication-deficient recombinant adeno-associated virus (ST2) encoding nuclear-targeted β-galactosidase (AAV β-gal; titer, 2.1×1012 pt/mL) or phospholamban (PLB; titer, 3.7×1012 pt/mL) driven by CMV promoter was also prepared.13–16
In Vivo Myocardial Gene Transfer
Mice were anesthetized with isoflurane induction and endotracheally intubated. Animals were placed supine on a thermoregulated table (37°C) and ventilated with isoflurane (2%) vaporized in 100% O2 using a custom constant-flow ventilator (6.7 μL/g−1 tidal volume at 120 minutes−1). ECG and core temperature were monitored. The left external jugular vein was cannulated, and modest volume expansion was provided (100 to 150 μL of 12.5% human albumin) at 30 μL min−1.
Topical 2% lidocaine gel was applied to the chest, and the thorax was entered by 1 of 2 techniques. For the first method, virus was injected into the LV chamber. Mice were placed on their right sides with the left forepaw affixed to the surgical table. The point of maximal intensity was visualized (approximately 0.5 to 0.75 cm below the axilla), and an incision was made to the left side of the point of maximal intensity. The thorax was carefully entered, avoiding cardiac injury, and the thoracotomy was widened by blunt dissection between the ribs to a total length of 0.5 to 0.75 cm. This provided direct visualization of the descending aorta immediately after the takeoff of the right subclavian artery and the course of the aorta to the diaphragm. The pericardium was opened, and a 7-0 suture was placed at the apex of the left ventricle. The aorta and pulmonary artery were identified. The anesthetized mouse was then cooled with a water jacket to a core temperature of 18 to 21°C (heart rate ≈80 to 100 bpm). The aorta was clamped distal to the takeoff of the left subclavian artery using a microserrefine (18055-03, Fine Science Tools). Because the IVC was not easily accessed through the lateral incision, it was not routinely occluded. IVC clamping was used, however, if the right ventricle (RV) became clearly dilated. Modified St Thomas cardioplegia (20 μL, 20°C) was injected into the LV cavity via an apically inserted 27G needle, followed by 20 μL of lipofectamine 2000 (Invitrogen), 1 μg/kg histamine (Sigma), and 20 μL of AdV β-gal, AAV β-gal, or AAVPLB. The second transfection technique introduced virus intravenously. The thorax was accessed by right lateral incision approximately 0.5 cm above the diaphragm. Both descending aorta and inferior vena cava were cross-clamped, and virus was injected into the right internal jugular vein. For both methods, the aortic clamp was maintained for up to 15-minute durations.
After releasing the aortic clamp, isoproterenol (3 to 10 ng/kg per min IV) or transesophageal pacing (NuMed, Hopkinton)17 was used for cardiac support, if needed, and mice were warmed to 37°C over 30 to 40 minutes. The chest was closed with 6-0 proline, negative thoracic pressure was restored, and the animals were extubated. Subsequent analysis was made 1 to 180 days after transfection. Using this method and a typical aortic occlusion time of 9 minutes, surgical survival was 71%. The most common identifiable causes of perioperative mortality were excessive bleeding, complete heart block, and ventricular arrhythmias. Postoperative mortality was 6%, being principally attributable to infection.
Expression of β-Galactosidase
β-galactosidase expression was assessed by enzyme activity in tissue samples and histochemical staining. Tissues were minced, lysed (75 mL per sample; 0.2% Triton X-100 and 100 nmol potassium phosphate; pH 7.8), and centrifuged (12 000 for 10 minutes), and supernatant was removed. Lysate was assayed for β-galactosidase activity by commercial test (Galacto-Light Plus, Tropix) with light emission measured by luminometer (Luminoscan RS, Labsystems) calibrated to a standard curve based on purified E. coli β-galactosidase. Activity was normalized per milligram of protein.
β-galactosidase histochemistry was performed on hearts fixed by 2% paraformaldehyde and 0.2% glutaraldehyde in PBS perfused through the RV for 10 minutes and on isolated myocytes. Tissue was incubated in X-gal stain (in PBS, 20 mmol/L K4Fe[CN]6 3H2O, 20 mmol/L K3Fe[CN]6, 2 mmol/L MgCl2, and 1 mg/mL X-gal [Promega] in DMSO) for 2 hours at 24°C, rinsed in PBS, and postfixed in 7% buffered formalin for 6 hours.
Isolation of Cardiac Endothelial Cells and Myocytes
To assess cell targeting of viral vectors, murine coronary endothelial cells were isolated using anti-endoglin (CD-106) antibodies (Transduction Laboratories) in a mini-MACS separation unit (Miltenyi Biotec [MB], Bisley, No. 421-01) with confirmation of cell isolation performed by flow cytometry.18 Labeled cells were incubated with MACS magnetic goat anti-rat IgG (H1L) (MB-485-01) MicroBeads and streptavidin (MB-481-01). MicroBeads were then separated using a high-gradient magnetic separation column 1 (MS columns, MB-422-01).18 Myocytes were isolated as described.19
Expression of Phospholamban and Colocalization With SERCA
Western blot analysis, immunoprecipitation of PLB, and colocalization of PLB with SERCA2a were performed as described.20–24 Immunohistochemistry was performed on isolated myocytes fixed in 4% paraformaldehyde and 0.5% Triton X-100. Primary overnight incubation with mouse monoclonal PLB and or SERCA2a antibody was followed by 1-hour secondary incubation with anti-rabbit Alexa 488 and anti-mouse Alexa 536 (Molecular Probes). Imaging was performed on a Nikon Diaphot 300 inverted epifluorescence microscope with a PCM-2000 laser confocal scanning microscope (Nikon).
Age-matched littermates (PLB+/+, PLB−/−, and PLB−/−+AAVPLB, 3 months after transfection) were anesthetized, intubated and ventilated, and studied by pressure-volume catheterization.17 A 1.8F catheter (SPR-719) was advanced via the LV apex to lie along the longitudinal axis and connected to a stimulation/analysis system to yield pressure-volume loops. To assess heart-rate modulation of cardiac function, a 2F pacing catheter (NuMed) was placed in the esophagus to achieve atrial pacing.17 Spontaneous sinus rate was first slowed using the If inhibitor ULFS-49 (Boehringer Ingelheim; 15 to 20 mg/kg−1 IP).17
Data are expressed as mean±SEM and analyzed by ANOVA followed by Tukey post hoc test to determine statistical significance between groups or by Student’s t test.
Influence of Ascending Aortic Versus Descending Aortic Occlusion
Figure 1A displays aortic and LV pressure and their difference (transmyocardial gradient) during descending aortic occlusion (DAO) versus ascending aortic occlusion (AAO). Diastolic myocardial perfusion gradients were maintained by DAO, whereas with AAO, aortic root and LV pressures equalized to eliminate a perfusion gradient. Figure 1B displays the data as pressure-volume loops. Application of AAO to the cooled heart resulted in near isovolumic contraction, with high systolic pressures but also marked elevation of diastolic pressure (near 45 mm Hg). In contrast, DAO yielded an ejecting beat with augmented systolic yet normal diastolic pressures. Thus, the hemodynamics of more distal occlusion in the beating heart were far more favorable for sustained occlusion and for generating adequate coronary perfusion to facilitate gene transfer.
Influence of Aortic Occlusion Time on Transfection Efficacy
We next determined the impact of aortic occlusion time (AOT) on gene transfer to determine the optimal duration or if this time varied between AdV and AAV. Animals were cooled similarly, and AOT varied between 1 and 12 minutes (Figure 2). Whole heart, isolated myocytes, and isolated coronary endothelial cells were examined for β-galactosidase activity 3 to 4 or 7 to 10 days after transfection with AdV and AAV, respectively (time points of near-maximal expression of reporter gene determined in pilot studies). There was little to no β-gal activity in AdV-transfected hearts until AOT was ≥9 minutes, at which point a sudden marked rise in activity was observed. For AAV, AOT could be shortened, although optimal activity still required 9 minutes of occlusion. Figure 2B shows results of a similar analysis performed on isolated coronary endothelial cells. For AdV, there was virtually no β-gal activity detected until 15 minutes of aortic clamping. In contrast, AAV resulted in endothelial expression at much earlier time points, with values approximately 60% of those in myocytes at the longest cross-clamp durations. Figure 2C shows results for isolated myocytes in which significant transfection was achieved after 5 minutes of aortic occlusion time with AAV, but 9 minutes of occlusion time was required for AdV. Thus, for a 9- to 10-minute clamp time, transfection with AdV vectors was principally myocyte-targeted, whereas AAV transfected both cell types.
Distribution and Chronicity of β-Galactosidase Expression
Figure 3 shows whole-heart and isolated myocyte histochemistry for β-gal expression. For AdV transfection (using either LV or intravenous injection routes), hearts were stained 3 days after transfection, whereas those transfected with AAV were stained at 7 days. Staining was diffuse but patchy in localized regions and likely reflects some overstaining (ie, leakage of β-gal/X-gal product into the cytosol), because it was difficult to fully rinse X-gal from the intact heart to stop the reaction in a whole-heart preparation. Figure 3B shows myocytes isolated from transfected hearts, revealing nuclear staining in approximately 50% to 65% of cells transfected with AdV, with similar amounts with AAV. Mouse myocytes frequently display polyploidy, so cells rather than nuclei were counted for this analysis. Both healthy and no longer viable (ie, rounded) cells were counted, because the latter were presumably alive when transfected in vivo.
AdV β-gal transfection yielded significant marker gene activity within 24 hours, peaking at 3 days and declining by 2 weeks (Figure 4A). This time course was independent of the route of viral introduction (LV cavity or intravenous, data not shown). With AAV β-gal, enzyme activity rose gradually, first appearing after 1 week, plateauing by 2 weeks, and then stabilizing thereafter for at least 180 days. Figure 4B shows the tissue distribution of β-gal activity. With both AdV and AAV coupled to the CMV promoter, activity was greatest and similar in heart and lung but also observed at lower levels in liver and other organs. AdV transfection with the α-MHC promoter selectively targeted the heart and yet produced similar enzyme activity to that with the CMV promotor. Thus, although use of a descending aortic clamp increased the tissue distribution of transfection, this could be obviated by use of a tissue-specific promotor.
In Vivo Gene Transfer of PLB to the Myocardium of the PLB−/− Mouse
To test for chronic functionality of gene transfer, we examined expression, localization, and functional influence of AAVPLB transfection into myocardium of PLB−/− mice. Analysis was performed 3 months after transfection and contrasted to age-matched wild-type (WT) and PLB−/− littermate controls. PLB−/− hearts had no PLB expression, whereas PLB−/−+AAVPLB hearts displayed PLB protein levels similar to those of WT controls (Figure 5A). Immunoprecipitation of PLB and subsequent probing for SERCA2a by Western blot revealed normal colocalization of both proteins in WT and PLB−/−+AAVPLB mice (Figure 5B). PLB−/− mice had no detectable expression of PLB by immunolocalization, whereas hearts transfected with AAVPLB showed immunostaining for PLB that colocalized to SERCA2a (Figure 5C).
Baseline systolic function (as indexed by dP/dtmax and preload-normalized maximal LV power) was higher in PLB−/− hearts, and systolic kinetics indexed by the time to end systole were faster. The peak rate of pressure decline (dP/dtmin) was greater, although relaxation time constant was not significantly different (Table). Transfection of AAVPLB diminished systolic function modestly, with borderline significant change in dP/dtmax but significant prolongation of systole.
More striking effects of AAVPLB were observed on the frequency dependence of both systolic and diastolic function. Figure 6A shows pooled regression plots for representative functional parameters at different steady-state heart rates. There was significant enhancement of systolic function and accelerated relaxation with higher heart rate in WT, but this was not significant in PLB−/− hearts, with the relations in these hearts being generally flat. Transfection with AAVPLB in the null-mutant heart restored the frequency-following response (FFR). Summary data for the slopes of various heart function–heart rate relations are provided in Figure 6B. Pooled regression analysis was performed using dummy variables for each heart to derive the average slope of each relation. Both WT and PLB−/− transfected with AAVPLB displayed significant relationships, whereas there was little to no heart rate dependence in the PLB−/− hearts.
We developed a novel method for in vivo gene transfer to the mouse using animal cooling, mid-distal aortic occlusion during virus injection, and substantially longer clamp times than have been generally used in larger rodents. Ascending aortic occlusion, a historically more widely used approach, yielded near isovolumic contraction with elevated systolic pressures but also equalized aortic and LV pressures and markedly raised LV diastolic pressures, compromising myocardial perfusion. Although this approach has been successful in larger rodents,1–7 these animals have nearly an order of magnitude (or greater) larger ascending root that may provide adequate compliance to maintain diastolic perfusion. The mouse aortic root is very small, and with AAO, only a trivial compartment remains. Indeed, our initial efforts using an AAO clamp approach consistently yielded low levels of transfection that favored the RV, and occlusion times were much shorter or risked cardiac demise. In contrast, the more distal clamp preserved a coronary perfusion gradient, maintained LV systolic and diastolic pressures, and permitted long clamp times needed for adequate viral-tissue interaction. Importantly, it allowed virus to be introduced intravenously, avoiding the need for proximal aorta and pulmonary artery dissection and LV manipulation.
Several recent studies11,25 have combined AAO with marked animal cooling (as first used in neonatal mice12) to transfect hamsters with AdV and AAV. These investigators used methods similar to Hajjar et al7 but prolonged aortic occlusion by cooling to achieve impressive results.11,25 Vascular dwell time is a key factor for efficient transfection,16,26,27 and in these hamster studies, 4 to 6 minutes of cross-clamp was adequate at 25 to 26°C cooling. However, in mice, this duration seems too short with AdV but may be adequate for AAV. The hamster is approximately 5 to 6 times the size of a mouse and probably more similar to guinea pig or rat in this regard.
A potential limitation of distal over proximal aortic occlusion is the enlarged perfused territory encompassing the upper body. However, cardiac specificity could be achieved by use of a myocyte-specific promoter. It is intriguing that marker gene activity was similar whether provided by CMV or α-MHC promoters despite greater potency of the former, but this may relate to relative high transfection levels. One potential limitation may rest in differences in mouse strains. The present study showed successful transfection in 2 broadly used strains, C57BL/6 and 129SvJ, but this still may not directly translate to other strains that may display different transfection efficiencies.
We did not perform systematic analysis of transfection efficacy as a function of body temperature, although Donahue et al26 have shown that cooling inhibits AdV transfection. Somewhat higher temperatures may even enhance gene transfer while still permitting prolonged aortic clamp times, but this remains to be tested. It also remains unclear whether myocyte infection occurs during cooling in the present or prior studies11,25 or after normal body temperatures are restored. The colder temperature may assist endothelial permeability to help translocate virus into the interstitium, but then the particles get trapped during rewarming as they become more infectious.
PLB Transfection: Functional Studies
An important component of whole-heart transfection is the capacity to alter net chamber performance. We chose to replace PLB in the PLB−/− model because protein expression and localization could only result from transfected gene. Furthermore, although many studies have examined the role of PLB to cardiac function, the present analysis is the first to test the impact of PLB restoration in an intact nonfailing adult PLB-null mouse, providing more direct analysis of its in vivo function. PLB protein levels were similar to WT and colocalized with SERCA2a (shown by immunoprecipitation and confocal microscopy). Our analysis of basal function supports prior data showing enhanced systolic performance in the PLB−/− mouse, and this was modestly blunted by restoration of PLB.7,20,22 One difference was the lack of enhanced relaxation in PLB−/− hearts that has been reported.29 This may be attributable to the much slower heart rates (≈300 minutes−1) and reduced contractility present in the earlier study that could favor PLB-dependent relaxation disparities.
More striking hemodynamic results were related to the influence of PLB restoration on the frequency dependence of systolic and diastolic function. This is thought to be principally attributable to enhanced calcium entry and uptake by the sarcoplasmic reticulum, although the exact role of PLB to the process has been somewhat controversial. In some studies,20,28 absence of PLB shifted the FFR from positive to negative, whereas other investigations found the opposite.29 Others found frequency-dependent acceleration of relaxation was preserved in PLB−/− but depended on calcium/calmodulin-dependent protein kinase II.30 Some of the discrepancy may relate to the preparation conditions and thus level of PLB phosphorylation, because negative influences of PLB have been found in preparations where optimal systolic function is achieved at subphysiological heart rates, with a negative FFR at more physiological rates.
Our in vivo preparation yielded more physiological basal function with typically positive FFRs that did not decline up to 900 minutes−1 in normal mice.17 In this setting, lack of PLB elevated dP/dtmax at lower rates but displayed little rate-dependent rise.17 Relaxation and effects on early peak diastolic filling were similarly largely rate independent in PLB−/− animals. The observed restoration of rate responsiveness in both sets of parameters several months after AAVPLB transfection argues in favor of in vivo FFR modulation by PLB. The exact mechanism remains to be clarified but may be elucidated by transfection with PLB mutants harboring altered phosphorylation sites. Lastly, it is possible that some other compensatory mechanisms may have played a role, although the duration of expression was only 3 months.
The present method for cardiac gene transfer should help the exploration of molecular signaling and development of therapeutic strategies using murine models. Importantly, the systematic identification of key elements for successful transfection, ie, perfusion/clamp time, temperature, and provision of adequate arterial compliance and transmyocardial perfusion gradients, may be relevant to other mammalian species and prove helpful for developing additional approaches for whole-heart gene transfer.
This work was supported by NIH NHLBI grants P01 HL-59408, P50-52307, and T32-HL07227 (to Dr Kass) and NIH NCRR grant P40RR12358 (from Evangelia Kranias, University of Cincinnati, Ohio). Dr Champion is a recipient of the Shin Chun-Wang Young Investigator Award and the Giles F. Filley Memorial Award from the American Physiological Society. The authors thank Evangelia Kranias for her support and Kevin Donahue for his helpful suggestions.
Kim TH, Skelding KA, Nabel EG, et al. What can cardiovascular gene transfer learn from genomics: and vice versa? Physiol Genomics. 2002; 11: 179–182.
Hajjar RJ, Kang JX, Gwathmey JK, et al. Physiological effects of adenoviral gene transfer of sarcoplasmic reticulum calcium ATPase in isolated rat myocytes. Circulation. 1997; 95: 423–429.
Hajjar RJ, Schmidt U, Matsui T, et al. Modulation of ventricular function through gene transfer in vivo. Proc Natl Acad Sci U S A. 1998; 95: 5251–5256.
Shah AS, Lilly RE, Kypson AP, et al. Intracoronary adenovirus-mediated delivery and overexpression of the β2-adrenergic receptor in the heart: prospects for molecular ventricular assistance. Circulation. 2000; 101: 408–414.
Ikeda Y, Gu Y, Iwanaga Y, et al. Restoration of deficient membrane proteins in the cardiomyopathic hamster by in vivo cardiac gene transfer. Circulation. 2002; 105: 502–508.
Christensen G, Minamisawa S, Gruber PJ, et al. High-efficiency, long-term cardiac expression of foreign genes in living mouse embryos and neonates. Circulation. 2000; 101: 178–184.
Lund DD, Faraci FM, Miller FJ Jr, et al. Gene transfer of endothelial nitric oxide synthase improves relaxation of carotid arteries from diabetic rabbits. Circulation. 2000; 101: 1027–1033.
Nakane H, Miller FJ Jr, Faraci FM, et al. Gene transfer of endothelial nitric oxide synthase reduces angiotensin II-induced endothelial dysfunction. Hypertension. 2000; 35: 595–601.
Champion HC, Bivalacqua TJ, Hyman AL, et al. Gene transfer of endothelial nitric oxide synthase to the penis augments erectile responses in the aged rat. Proc Natl Acad Sci U S A. 1999; 96: 11648–11652.
Champion HC, Bivalacqua TJ, D’Souza FM, et al. Gene transfer of endothelial nitric oxide synthase to the lung of the mouse in vivo: effect on agonist-induced and flow-mediated vascular responses. Circ Res. 1999; 84: 1422–1432.
Bluhm WF, Kranias EG, Dillmann WH, et al. Phospholamban: a major determinant of the cardiac force-frequency relationship. Am J Physiol Heart Circ Physiol. 2000; 278: H249–H255.
Feron O, Belhassen L, Kobzik L, et al. Endothelial nitric oxide synthase targeting to caveolae: specific interactions with caveolin isoforms in cardiac myocytes and endothelial cells. J Biol Chem. 1996; 271: 22810–22814.
Senzaki H, Smith CJ, Juang GJ, et al. Cardiac phosphodiesterase 5 (cGMP-specific) modulates beta-adrenergic signaling in vivo and is down-regulated in heart failure. FASEB J. 2001; 15: 1718–1726.
Donahue JK, Kikkawa K, Johns DC, et al. Ultrarapid, highly efficient viral gene transfer to the heart. Proc Natl Acad Sci U S A. 1997; 94: 4664–4668.