Localized Co-delivery of Serca2a and Cx43 Genes Combined With Bone Marrow Mesenchymal Stem Cells Restores Rat Ischemic Cardiac Mechano-electrical Function

Aim: Signicantly reduced expression of Ca 2+ -ATPase 2a (SERCA2a) and Connexin 43 (Cx43) which play important roles in regulating mechano-electrical function was found in an ischemic heart. Bone marrow mesenchymal stem cells (BMSCs) transplantation brings with a weak improvement of ischemic heart in cardiac contractility and anti-arrhythmias effect. The aim of this study was to recover mechano-electrical function of ischemic heart through combining stem cell therapy with SERCA2a/Cx43 co-delivery by biotinylated microbubbles (BMBs) via ultrasound targeted microbubble destruction (UTMD). Methods: Dual gene-loaded BMBs were developed through conjugating SERCA2a-adenovirous (S-Ad) and Cx43-adenovious (C-Ad) onto BMBs via biotin-avidin linkage. UTMD was used to mediate the local co-delivery of S-Ad and C-Ad into the infarct zone where BMSCs were transplanted. The expression of SERCA2a and Cx43 gene, neovascularization in the infarct area and mechano-electrical function of rats were detected. Results: UTMD-mediated dual gene delivery could signicantly enhance the expression of SERCA2a/Cx43 genes in the infarct zone receiving BMSCs transplantation. Signicantly improved neovascularization was observed. More importantly, UTMD-mediated dual gene delivery greatly improved cardiac function and reduced arrhythmia in these myocardial infarct (MI) hearts transplanted with BMSCs. Conclusions: BMSCs-based dual gene therapy can effectively improve cardiac mechano-electrical function and reduce arrhythmia in heart with MI. It is necessary to rebuild the gene network in damaged heart rather than supply a single gene. BMSCs transplantation combined with localized dual-gene delivery by UTMD might point out a novel strategy to recover mechano-electrical function of ischemic heart.


Introduction
Ischemic heart disease is a signi cant cause of morbidity and mortality worldwide [1]. Transplantation with bone marrow mesenchymal stem cells (BMSCs) is considered as an ideal approach in myocardial infarction (MI) therapy due to their advantages as follows: 1) MSCs can secrete some growth factors, such as vascular endothelial growth factor (VEGF) and stromal cell derived factor 1 (SDF-1), to elevate angiogenesis or to promote the formation of novel vasoganglion via paracrine effect [2]. 2) BMSCs are capable to inhibit the production of active in ammatory factors and improve the in ammatory microenvironment or directly regulating the functions of immune cells after MI [3,4]. 3) BMSCs can prevent cardiac brosis by releasing hepatocyte growth factor (HGF), and thus reverse LV remodeling and restore the cardiac function [5]. 4) BMSCs can home to the infarction area after transplantation with the help of SDF-1 and its receptor CXC chemokine receptor 4 (CXCR4). 5) BMSCs have the capacity for multidirectional differentiation, contributing to the regeneration of the vasculature after MI [6]. Despite all of the bene ts brought by MSCs, this approach only shows a faint e cacy in restoring cardiac contractility and preventing ischemia-induced arrhythmias after MI. The main reason which causes unsatisfactory e cacy may attribute to the abnormal Ca 2+ handling accompanied by undermined cellcell coupling within infarct and peri-infarct region.
It has been well established that myocardial ischemic stress strongly modi es the pro le of gene expression in heart [7] including sarcoplasmic reticulum Ca 2+ -ATPase 2a (SERCA2a) and Connexin 43 (Cx43), two key proteins that regulate mechano-electrical function [8,9]. SERCA2a governs the intracellular Ca 2+ handling process [10], by which cardiac contraction and relaxation is maintained through mediating Ca 2+ reuptake into sarcoplasmic reticulum (SR) [11]. A reduced SERCA2a level or impaired function will result in the overload of intracellular Ca 2+ and utterly myocardial dysfunction after MI [12]. Cx43 is another signi cantly decreased protein which takes vital role in the gap junctions (GJs) of myocardial cells in ischemic myocardial tissue. GJs mediate electrical coupling between cardiac myocytes, forming the cell-to-cell pathways for orderly spread of the wave of electrical excitation responsible for synchronous contraction, and the impaired electrical coupling could consequence in increased inclination for arrhythmias [13]. GJs channel is consisted of six transmembrane proteins in atrial and ventricular myocytes and Cx43 is the most abundant member among them. Previous studies have demonstrated that the reduced Cx43 expression and its abnormal distribution eventually result in arrhythmias [14]. In this context, it is necessary to enhance the expression levels of SERCA2a and Cx43 genes in the recovery of cardiac function, especially during the transplantation treatment with stem cells after MI.
Recent years, gene therapy has drawn great attention due to its favorable therapeutic bene ts in the treatment of various diseases [15], including tumors [16], cardiac diseases [17] and nervous system disorders [18]. However, it remains a grand challenge to selectively and effectively deliver the genes into targeted cells with minimal side-effect. To date, various virus gene delivery vectors or vehicles, such as adenovirus (Ad), lentivirus and adeno associated virus (AAV), provide an approach with high gene transfection e ciency, but lack of speci c tissue or organ targeting capability for these virus vectors limit their clinical applications to some extent [19]. Thus, there is great demand to develop a strategy to help these viral vectors to locally deliver into target tissues. More recently, ultrasound-targeted microbubble destruction (UTMD) has been proved a promising approach for local delivery of genes or drugs [20][21][22].
Ultrasound beams transmitted by a focused transducer can penetrate soft tissues and be focused into a special organ such as a heart. When there are microbubbles (MBs) in the blood vessels, the ultrasound energy in the focused site will induce the bubble resonance, producing a series of acoustic cavitation effects, including stable cavitation and inertial cavitation. The former induces microstreaming in the ow of liquid around the MBs that brings with shear stress to cell membranes due to the repetitive MB contraction and expansions. The latter causes abrupt MB destruction to produce stronger mechanical stress, such as shock waves, microjets, etc. These acoustic cavitation effects collectively cause cell membrane perforation and locally enhance blood vessel permeabilization [23].
In this study, taking the great advantages of high virus gene transfection e ciency and localized gene delivery by UTMD, we attempt to combine stem cell-based strategy and gene therapy to improve the e cacy in the restoration of mechano-electrical function after MI (Abstract graphic). Brie y, Ad-loaded MBs were rstly constructed through coating the SERCA2a-Ad (S-Ad) and Cx43-Ad (C-Ad) onto the surface of MBs via a biotin-avidin linkage. After being intravenously injected into the MI rats transplanted with BMSCs, ultrasound irradiation was applied to release and deliver the Ads locally into the infarct myocardial tissues. SERCA2a and Cx43 overexpression will accelerate neovascularization in infarct zone and present better restoration in mechano-electrical function than cell-based method alone.

Characterization of BMBs and gene-loaded BMBs
Particle size and size distribution of MBs were measured with Accusizer 780 Optical Particle Sizer (Particle Sizing Systems, Santa Barbara, CA, USA). A drop (about 20 μL) of each kind of BMBs suspension was applied to the microscope slide and observed under an optical microscope (Olympus, Tokyo, Japan). Furthermore, each kind of BMBs suspension was negatively stained with an aqueous solution of uranyl acetate and observed using a transmission electron microscope.

BMBs stability
The stability of BMBs was manifested by the change in the particle size and concentration over time. The particle size and concentration of 1ml of diluted BMBs were measured immediately after the completion of fabrication at room temperature, or after 15, 30, 45 and 60 mins by Accusizer 780 Optical Particle Sizer.

Gene loading capacity of BMBs
To determine the gene loading capacity of BMBs, FITC-labeled SERCA2a/GFP-encoded adenovirus (S-Ad) or rhodamine-labeled Cx43/RFP-encoded adenovirus (C-Ad) range from 1.25 μl to 20 μl (1×10 9 pfu/ml) was mixed with BMBs to fabricate gene loaded BMBs (S-BMBs or C-BMBs), respectively. Then, the gene loading capacity was determined by ow cytometry and qPCR. After determined the lowest amount of Ad to achieve the maximum gene loading e ciency. Furthermore, S-Ad and C-Ad was added into BMBs to making dual gene-loaded BMBs (S/C-BMBs) at the ratio of 1:1, 1:2 or 2:1, respectively. Next the dual gene loading capacity was determined by ow cytometry and qPCR.

Isolation culture and identi cation of BMSCs
Conforming to the Directive 2010/63/EU of the European Parliament, SD (Sprague Dawley) rats at 4 weeks of age were euthanatized with pentobarbital sodium (60 mg/kg body weight, Sigma-Aldrich Inc., USA) via intraperitoneal injection once. Then, bilateral tibial and femoral bones were dissected from body trunk and stored on ice in 75% alcohol. Subsequently, femurs and tibias were separated and both ends of each femur or tibia were cut, and a 22-gauge needle attached to a 10cc syringe containing complete medium was then inserted into the spongy bone exposed by removal of the growth plate. The marrow plug is then ushed from the bone with 5 ml of complete medium and collected in a 50 ml conical tube.
Cells were then cultured in T-75 cell culture ask, with a cell concentration of 1×10 5 /mL, using Mouse Mesenchymal Stem Cell Growth Medium (Cyagen, China). BMSCs were then puri ed and passaged by attachment method. Cells were incubated under standard cell culture conditions with 5% CO 2 , at 37 °C and 95% relative humidity. The medium was changed every three days, and BMSCs were passaged when 80%-90% con uence was reached. BMSC identity was con rmed on the basis of morphological criteria, plastic adherence, and speci c surface antigen expression: CD29(+), CD90(+), CD45(−).

Ultrasound-mediated gene transfection with BMBs in vitro
Ultrasound-mediated gene transfection was performed by using an ultrasound system including an arbitrary waveform generator (model AFG3102, Tektronix, USA), an RF power ampli er (model AR150A100B, AR, USA), and a single-element planar ultrasound (US) transducer (frequency = 1 MHz; Valpey Fisher Corp., MA, USA). Brie y, BMSCs were seeded in 12-well plates (1×10 5 cells per well) and transfection experiments were conducted when the cell con uence reached 70-80%. Then gene-loaded BMBs were added to the well, and to guarantee a close contact between transfection complexes and cells, the 12-well plate was sealed rmly and inverted for 15 min. Then ultrasound exposure was performed. The multiplicity of infection (MOI) was adjusted to 500 throughout the study. Ultrasound conditions were set as follows: frequency 1 MHz, power 2.0 W/cm 2 , duration 60 s, duty cycle 10%.

Detection of the in vitro gene transfection e ciency
To examine the ultrasound-mediated gene transfection e ciency, the following groups were included: (1) Control group; (2) Ad group, only Ad was used to transfect BMSCs; (3) Ad + US group, Ad was used to transfect BMSCs under the aid of ultrasound; (4) Ad + UTMD group, gene-loaded BMBs was use to transfect BMSCs under the aid of ultrasound. These BMSCs were cultured for another 48 h. Fluorescence microscopy was applied to detect GFP and RFP expression of each group, western blot and qPCR analysis to detect SERCA2a and Cx43 protein and mRNA level of each group.

Cell viability assay and in vivo biocompatibility
Cell viability was measured immediately after the gene transfection using a Cell Counting Kit-8 (CCK-8) according to the manufacturer's protocol (Dojindo, Japan) to determine the possible cell damage caused by UTMD. Relative cell viability (RCV) was assessed using CCK-8 assay and then determined in a 96-well plate reader (BioTek Synergy 4) at 450 nm wavelength with the equation RCV (%) = At − Anc/Apc − Anc ×100%. Furthermore, to evaluate the in vivo biocompatibility of the BMBs, healthy SD rats were intravenously injected with gene-loaded BMBs. On the 15 th day post-injection, fresh blood samples (1.0 mL) were obtained by cardiac puncture from the rats for serum biochemistry study. Subsequently, the major organs including lungs liver, spleen and kidneys of were carefully collected for H&E staining histology analysis. Healthy rats without any intervention were used as the control.

Animal model
The experimental protocol was approved by the Animal Ethics Committee of the First A liated Hospital of Xinjiang Medical University (approval number of IACUC-20160218-057). Following the ARRIVE criteria [25], the experiment was carried out in strict accordance with the "3R" principle of substitution, reduction and optimization to minimize damage to animals. Male or female healthy SD rats (8~10 weeks old, weight 200-260 g) were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. The experimental animals were well housed under standard conditions of room temperature and dark-light cycles with su cient water and food.
On the rst day of the experiment, rats were induced for anesthesia with Dexmedetomidine (0.3 mg/kg body weight, Sigma-Aldrich Inc., USA)/midazolam (4.0 mg/kg bw, Sigma-Aldrich Inc., USA)/butorphanol (5.0 mg/kg bw, Sigma-Aldrich Inc., USA) [26] via intraperitoneal injection once. Then, the rats were intubated with a 16G intravenous catheter (Introcan 16G, Braun Medical Co., Ltd., Germany) and ventilated with a rodent ventilator (HX-100E, Taimeng Software Co., Ltd., Chengdu, China). Left anterior descending artery (LAD) ligation was performed as previously described [27]. The rat's heart was fully exposed by thoracotomy. Finding the coronary vein as a landmark, then permanently ligated the LAD approximately 2~3 mm distal from its origin with a depth of 0.5 mm by a 5/0 suture. Evidence of a MI was con rmed by pale and hypokinesia in the left ventricular anterior wall and ST-segment elevation on an electrocardiogram. Penicillin 160,000 u was intraperitoneal injected for 3 days after the surgery.

Experimental animal groups
Forty-eight successfully modeled SD rats were randomly divided into control and experimental groups:

BMSCs transplantation
Four weeks after the successful ligation, the rats underwent surgery again. The rats from SHAM, Control and UTMD groups were injected with PBS (100 μl). Rats from the rest groups were injected with BMSCs (5×10 6 , 100 μl) at the myocardial infarct zone and the peri-infarct zone.

UTMD-mediated localized co-delivery of genes
Two days after BMSCs transplantation [20], UTMD-mediated gene localized co-delivery was performed. The treatment groups were infused with 100 μl of the S-BMBs, C-BMBs or S/C-BMBs respectively via the tail vein at a constant rate (15 ml/h), and then washed with PBS 100 μl. A Mindray Kunlun 7 ultrasound machine with a line array probe (L 11-3 U) was used at a setting of fundamental frequency 5.6 MHz-11.8 MHz, harmonic frequency 7 MHz ~ 9 MHz and mechanical index 0.53. Contrast mode was activated at the beginning of infusion. On seeing the lling of BMBs in the left ventricle cavity, the FLASH (mechanical index: 1 [20]) function was triggered manually at an interval of 3 s ~ 5 s, lasting for 10 mins [28]. The rats of Control group were infused with PBS 100 μl and performed as above.

Left ventricular function analyzing
Echocardiography was performed 4 weeks after AMI and 2 weeks and 4 weeks post-treatment. A Vevo2100® ultrasound imaging system (High-Resolution Micro-Imaging System, VisualSonics, Canada) equipped with an 18 MHz transducer was used by an investigator blinded to group designation. The left ventricular ejection fraction (LVEF) and left ventricular fractional shortening (LVFS) were calculated by Mmode tracing. The dimension data are presented as the average of measurements of three selected beats.

Electrophysiological examination
Steady-state pacing was performed to test the ability of anti-arrhythmia after treatment. On the 28 th day after BMSCs transplantation, Rats were induced for anesthesia by intraperitoneal injection of sodium pentobarbital (30 mg/kg body weight, Sigma-Aldrich Inc., USA) once. ECG monitoring (BL-420F biological function experiment system, Taimeng Software Co., Ltd., Chengdu, China) was used to record the ECG activity in real time. The stimulation protocol was performed as described previously [29]. Brie y, the left chest was opened to fully expose the heart after anesthesia. A bipolar pacing electrode was placed in the left ventricular wall near the apex of the heart, with a diameter of 0.2 mm and a distance of 3 mm between the two electrodes. Continuous progressive stimulation (S1S1) was given to induce ventricular arrhythmias. Stimulus plan is that as below, pulse width of 0.1ms, step length at -2 ms, stimulus frequency starts at 5 Hz and stimulus voltage begins with 5 V, gradually increasing by 1 V. End point of stimulation is induction of ventricular tachycardia (VT) or ventricular brillation (VF) with more than 6 consecutive ventricular premature beats. After the Electrophysiological examination, the rats were euthanized by pentobarbital sodium (60 mg/kg body weight, Sigma-Aldrich Inc., USA) [26] via intraperitoneal injection once, and the hearts were taken for subsequent test. The procedure conformed to the Directive 2010/63/EU of the European Parliament.
Enzyme-linked immunosorbent assay (ELISA) analysis ELISA was applied to determine the levels of SERCA2a and Cx43 in the infarcted myocardial tissue homogenates. commercial ELISA kits (R&D systems) were used per the manufacturer's instructions.
Samples and standards were prepared according to manufacturer's instructions.

Western blot assay
Protein expression levels of SERCA2a, Cx43, VEGF and cardiac-speci c proteins (cTnT and α-actin) in the infarct zone were detected by Western blot according to the previous study. [30] Total proteins were extracted from treated myocardial infarct tissue using a RIPA buffer with protease and phosphatase inhibitors (G2002, Wuhan Servicebio Co., Ltd., China). The protein concentrations were determined by the Bicinchoninic acid (BCA) method as a protein standard. Proteins were separated by electrophoresis, transferred to a cellulose acetate membrane and blocked. The primary antibody (GB23303, Wuhan Servicebio Co., Ltd., China) was added followed by the secondary antibody (GB23302, Wuhan Servicebio Co., Ltd., China). The density of each band was quantitated by a densitometer with AlphaView Software for FluorChem Systems (ProteinSimpleTM).

Analysis of immuno uorescent protein expression
Immuno uorescence was performed to detect SERCA2a and Cx43 protein expression in the infarcted myocardium. Frozen sections of myocardial tissue were prepared as described previously [31]. Then, the prepared frozen sections were incubated with 5% (volume fraction) bovine serum albumin (Solarbio, China) for 30 mins. The primary antibody (at 1:100 /1:200 volume) was added, incubated overnight at 4 ℃, and then washed thrice for 5 mins with PBS. Next, the uorescently labeled secondary antibody (at 1:300/1:400 volume) was added and incubated at 37 ℃ for 50 min. The slides were washed as above. Subsequently, the image was observed and photographed using a uorescence microscope (Nikon Eclipse TI-SR, Nikon Inc., Japan). The uorescence area ratio was quantitated with Image-Pro Plus version 6.0 software (Media Cybernetics, Bethesda, MD) by 2 observers blinded to the conditions.

Histological evaluation of neovascularization in the infarcted zone
To evaluate neovascularization, formaldehyde-xed rat hearts were dehydrated and then embedded in para n. Para n-embedded sections (4 μm) were dewaxed and incubated with the primary antibody against factor VIII (ThermoFisher Scienti c, America). After visualization with diaminobenzidine (DAB), we counted the positively stained micro-vessels under a Nikon eclipse E100 light microscope (Nikon Inc., Japan; 100× magni cation). Microvascular endothelial cells showed a layer of brown ring-like precipitates with a diameter of less than 20 μm, which were calculated as previously described [32]. All histological analyses were independently performed by two experienced pathologists under doubleblinded conditions.

Measurement of the infarct size
Twenty eighty days after treatment, the rats were euthanasized with overdose of pentobarbital sodium (60 mg/kg body weight, Sigma-Aldrich Inc., USA) via intraperitoneal injection once. Then, after cardiac perfusion with saline hearts of the rats were extracted and xed in 4% paraformaldehyde (Wei Bio Technology Co., Ltd., Shanghai, China). Para n-embedded samples were sectioned at 4 μm, and Masson's trichrome staining was performed. The infarcted zone was evaluated based on the percentage of blue staining, indicative of brosis, and quantitated with Image-Pro Plus version 6.0 software by two observers blinded to the conditions.

Statistical analysis
All data are summarized as mean ± standard deviation. The data were analyzed using SPSS 21.0. Parametric comparisons were tested by One-way analyses of variance (ANOVA) with subsequent posthoc multiple comparisons using the least signi cant difference (LSD) test. Probability values P < 0.05 were considered statistically signi cant.

Fabrication and characterization of biotinylated microbubbles (BMBs)
The fabrication process of BMBs was illustrated in Fig. 1a. Brie y, the mixture of DSPC, DSPE-PEG2000, DSPE-PEG-Biotin dissolved in chloroform was applied to form the lipid shell of BMBs and C 3 F 8 was used as the gas core (for sample preparation details, see Methods). The resulting BMBs were rstly coated with streptavidin, followed by incubation with Ad particles, leading to the production of Ad-BMBs. The lipid solution was transparency liquid (Fig. 1b left), BMBs appeared milky after oscillation (Fig. 1b right). Fig. 1F presents the microscopic image of BMBs, showing a bright gas core surrounded by dark circular rings. Then scanning electronic microscopy (SEM) was performed to further determine the morphology of these bubbles (Fig.1c). Fig. 1d presented the particle size distribution of the BMBs, showing average particle size of 1.10 ± 0.65 μm with a polydispersity index (PDI) of 0.26 ± 0.05. The freshly fabricated BMBs had a concentration of 4.63 ± 0.62×10 9 bubbles/ml as shown in Table 1. Considering that the Ad loading process takes approximately 40 mins, the BMBs must keep stable during this period of time. Fig.  1e revealed that the mean diameter did not show any difference during 60 mins and the bubble concentration stayed the same after 40 mins, proving that the self-assembled BMBs possessed a desirable stability. The ultrasound images of BMBs showed su cient contrast signals detected by ultrasound in vitro and in vivo (Fig. S1). Thus, BMBs were successfully fabricated.

In vitro UTMD-mediated gene transfection
To con rm the UTMD-mediated gene transfection e ciency in vitro, BMSCs were cultivated (Fig. S2a) and identi ed (Fig. S2b), then were seeded in 24-well plates and divided into four groups: (1) control group without any intervention; (2) Ad only group; (3) Ad + US group, and (4) Ad + UTMD group. A uorescence microscope was used to detect the expression of GFP or RFP at 48 h after application of S-Ad or C-Ad. As presented in Fig. 2a, GFP expression displayed a burst elevation in Ad group comparing to the control which did not show any uorescence signals. Strikingly, the GFP positive cells can be increased with the aid of UTMD, while US alone could not bring any bene t to Ad transfection e cacy (Fig. 2a, d). Similar patterns were discovered when using C-Ad and co-application of both Ads (Fig. 2b, c and 2e, f). Furthermore, to gain a quantitative insight of the transfection e ciency, levels of SERCA2a and Cx43 protein and mRNA were con rmed by western blotting and q-PCR analysis 48 h after transfection. Fig. 4g, h demonstrated a prominent elevation in SERCA2a or Cx43 protein level both in Ad and Ad + US group compared with the control group. However, we discovered that the protein level of SERCA2a or Cx43 was signi cantly enhanced in Ad + UTMD group compared to Ad group (Fig. 2g, h). When the S/C-BMBs which carried both S-Ad and C-Ad in a ratio of 1:1 were used, the protein levels of both SERCA2a and Cx43 were declined approximately by 30%. It may be attributed to the decreased loading virus amounts for S-Ad and C-Ad in S/C-BMBs. Intriguingly, we found that the levels of SERCA2a and Cx43 were elevated more prominently in Ad + UTMD group compared to both Ad only and Ad + US group (Fig.  2c, f), indicating that the bene ts of UTMD are more pronounced at lower MOI. As expected, the improvement in SERCA2a and Cx43 mRNA levels showed the same pattern with protein in all conditions ( Fig. 4i, l). Subsequently, the cell viability of BMSCs in different groups was detected by the experimental setup showed in Fig. S3a. No evident cell damage was found in the UTMD group, with the cell viability remaining at 95.47±2.96%, indicating the ultrasound irradiation (1 W/cm 2 , a duty cycle of 20% and a duration of 60 s) was safe for BMSCs (Fig. S3b).
UTMD-mediated co-delivery of S-Ad and C-Ad into infarct zone enhanced SERCA2a and Cx43 expression Next, we further detected the feasibility of gene co-delivery into the infarct zone of MI rats receiving BMSCs transplantation. Firstly, MI rat model was made (Fig. S4). Electrocardiograms showed that STsegment typically elevated after LAD ligation. (Fig. S4a, b). Moreover, echocardiography reports a signi cantly thinning in left ventricular wall and left ventricular dilation 28 d after LAD ligation (Fig. S4ce). LVEF and LVFS also displayed a signi cantly reduction after ligation (Fig. S4f). After echocardiography assessment, the heart was harvested, the myocardium of the anterior wall of the left ventricle was pale and collapsed (Fig. S4g). Masson's trichrome staining showed that viable myocardium and scar tissue were identi ed in red and blue, respectively (Fig. S4h). All of these results showed that the MI model was successfully established. Four weeks after MI, allograft BMSCs were injected directly into the infarct zone and UTMD-mediated localized gene delivery was applied 2 days after this process. Levels of SERCA2a and Cx43 proteins were estimated through immuno uorescence staining at 4 weeks after gene delivery. As presented in Fig. 3a, we observed a drastic reduction in both SERCA2a and Cx43 expression in the infarct zone after MI (control group), and this harsh situation can barely be reversed by BMSCs transplantation (BMSCs group). However, abundant red uorescence signal representing the expression of SERCA2a was discovered in the B+S+U group, while green uorescence signal representing the expression of Cx43 was seen in the B+C+U group. Strikingly, both red and green signals, which were more intense compared with the control group, were found in the B+S/C+U group. Western bolt analysis showed that the SERCA2a level in the B+S+U group was approximately 3.83-fold and 4.00-fold higher than that of the control and BMSCs groups (Fig. 3b). Similarly, the Cx43 level in the B+C+U group was approximately 15.50-fold higher than that of the control and BMSCs groups respectively (Fig. 3c). As expected, an obvious enhancement of SERCA2a could be observed in the B+S/C+U group, despite less than that of the B+S+U group. Interestingly, the Cx43 protein level in the B+S/C+U group was even higher than that of the B+C+U group, implicating a connection between SERCA2a and Cx43. Data from ELISA showed the similar pattern with western blotting analysis among groups except for the level of Cx43 in the B+C+U and B+S/C+U groups ( Fig. 3d and e).

BMSCs-based gene therapy improves neovascularization in the infarct zone
Given by the fact that the infarcted heart for the late phase of MI undergoes brosis and remolding to replace necrotic myocardial cells and eventually led to further deterioration of cardiac function. We evaluated the size of brosis via Masson's trichrome staining at 4 weeks after treatment. No difference of the infarct size was detected among groups according to Fig. 4a and b. Notably, we did observe cardiomyocyte-like cells stained in red within the infract zone in the four groups receiving BMSCs transplantation, except the Control group. In order to con rm this, we further evaluated the expression of myocardial markers in the infarct zone. Results from western blot analysis demonstrated that myocardial ischemia resulted in a robust loss in cTnT (Fig. 4c) and α-actin (Fig. 4d) in the control group. BMSCs transplantation produced the favorable impact on the expression of cTnT and α-actin, with 0.39-fold and 0.23-fold higher than that of the control group, respectively. Remarkably, the expression levels of cTnT and α-actin could be further improved by gene therapy, especially for rats overexpressing Cx43. The cTnT and α-actin levels increased by 1.61-fold and 1.56-fold in the B+S+U group respectively. And in the B+C+U group the increasement was 1.84-fold and 1.95-fold for cTnT and α-actin respectively (Fig. 4c-d).
Since neovascularization is a crucial process in post-MI recovery, we also tested the expression of Factor VIII and VEGF. We evaluated the micro-vessels density (MVD) in the infarct zone by staining the microvessels with anti-VIII factor antibody. Immunohistochemical staining assay revealed that MVD enhanced in the BMSCs group compared with the control (Fig. 5a, b). Remarkably, more MVD could be found in the B+C+U and B+S/C+U groups, while overexpression of SERCA2a brought few bene ts to neovascularization.
Similarly, strong expression of VEGF was detected in the B+C+U group compared with the others (Fig. 5c), indicating that Cx43 might be bene cial for accelerating neovascularization. Thus, BMSCs in combination with Cx43 overexpression can effectively improve the microenvironment in the infarct region, which might directly contribute to sustain the remaining cardiomyocytes.

BMSCs-based gene therapy recovers ischemic cardiac mechano-electrical function
Since BMSCs-based gene therapy can bring with the bene cial structural changes, we further evaluated the functional alterations 4 weeks after treatment through detecting their contractile and electrophysiological properties. Firstly, echocardiographic assessment was applied to record the LVEF and LVFS, which were critical for cardiac functional parameters. As Fig. 6a exhibited, hypokinesia in the anterior wall and the left ventricle dilation were observed, accompanied by a decrease in LVEF and LVFS post-MI (Fig. 6b, c). Two weeks after BMSCs-based gene therapy, LVEF and LVFS signi cantly increased both in the B+S+C (34.6±3.13% and 17.3±1.42%) and B+S/C+U groups (35.8±3.31% and 20.5±4.55%). These tendencies maintained until 4 weeks. By contrast, no signi cant increasements in LVEF and LVFS were discovered in the BMSCs and B+C+U groups compared with the control group, indicating that BMSCs transplantation alone or combined with Cx43 overexpression cannot restore myocardial contractility. Next, rats were subjected to steady-state pacing at progressively faster rating up to 8.0 Hz or the induction of ventricular tachycardia (VT) / ventricular brillation (VF) (Fig. 6d). Fig. 6e showed that none of the rats in the SHAM group exhibited pacing-induced VT/VF when the stimulating voltage was set as 7 volts, while 8 (100%) and 7 (87.5%) of the rats in the control and BMSCs groups were susceptible to VT/VF, indicating that BMSCs transplantation brought little pro ts to electrical stability of myocardium. In contrast, VT/VF propensity was suppressed to around 62.5% in the B+S+U group and 0% in the B+C+U and B+S/C+U groups. Interestingly, VT/VF propensity could still be diminished from 100% to around 50% in the B+C+U and B+S/C+U groups even if we switched the stimulating voltage to 8 volts, revealing that the protective effect of SERCA2a disappeared (Fig. 6f). Thus, our data strongly supported the strategy of localized co-delivery of SERCA2a and Cx43 genes combined with BMSCs can restore rat's ischemic cardiac mechano-electrical functions.
Biosafety assay Finally, we tested the biocompatibility of BMSCs-based gene therapy in vivo. All the measured blood parameters of liver and kidneys from rats on the 15 th day post-injection with gene-loaded BMBs uctuated within the normal ranges, showing no distinct differences compared with the control group ( Fig. Sc-f). Furthermore, H&E staining images of the primary organs, such as the lungs, liver, spleen and kidneys, demonstrated no apparent injury or in ammation in the groups with BMSCs-based gene therapy and control group, as Fig. S3g showed.

Discussion
Despite the multifunction properties of BMSCs, their effect in enhancing cardiac function after MI is still controversial [5] To date, few studies clari ed the impact of regenerative method on electrophysiological properties in the failing heart after MI. Above conclusions indicate that BMSCs transplantation alone is not enough to recover the mechano-electrical function of damaged heart. It is necessary to combine the stem cell-based approach with other intervention strategies to improve the e cacy for MI treatment.
According to the previous studies, overexpression of some critical genes can improve the function of abnormal cardiomyocytes. Cutler and his collages successfully ameliorate the cardiac function of failing heart through overexpressing SERCA2a gene with AVV9 [33]. Another study indicates that elevation of SERCA2a protein level increases contractile properties of cultured rat cardiomyocytes and signi cantly improves ventricular function in mice with heart failure [34]. Also, Roell [35] displayed that overexpression of Cx43 markedly reduces post-infarction VT and emphasizes the concept of non-myocyte electrical conduction as a key target in post-infarction cardio protection [35]. In this context, we proposed a strategy that combine stem cell transplantation and site-speci c dual gene therapy, which is mediated by UTMD.
The current study demonstrates that stem cell-based site-speci c dual gene therapy can effectively improve cardiac function and reduce the post-infarct scar-related arrhythmias.
An appropriate vector is the basis to manage a site-speci c gene delivery. Although viral vectors were highlighted by their satisfactory transfection e ciency, lack of tissue targeted ability limit the administration route of viral vectors to local injection, which might cause tissue damage. Previous studies have proved that a tissue targeted effective gene transfection could be achieved through binding viral vectors to the surface of MBs with the aid of ultrasound irradiation [36,37]. In our latest research, we also demonstrated that ultrasound beam could destroy the virus loading MBs, release the viral vectors at ventricular level, and realize the tissue targeted gene delivery [22]. In the current study, the resulting BMBs had a micro-scale size (around 1 μm) and showed a good stability. Moreover, these BMBs could produce a clear and stable contrast signal both in vitro and in vivo, making it possible to monitor the gene delivery progress by ultrasound in a real-time manner. Most importantly, BMBs showed an ideal gene loading capacity that above 90%, just as proved by ow cytometry and qPCR analysis.
In our study, we validated in vitro that UTMD could release the Ad from BMBs and improve the gene transfection e ciency. Images of uorescent microscope clearly presents that Ad itself possess a relatively high transfection e ciency. However, compared with the cells received UTMD, more cells expressing uorescent protein could be observed, strongly indicating that UTMD helps more virus encoding uorescent protein gene enter cytoplasm by making transient pore on cell membrane [38]. Interestingly, during the in vitro experiment we found that the effect of UTMD was more prominent when the MOI was lower, indicating the effect of UTMD was underestimated, and we hypothesis that a reasonable combination of MOI and UTMD might signi cantly decrease the amounts of viral vectors when achieving an ideal transfection e ciency.
As demonstrated by the immuno uorescence detection in MI rat model, the level of SERCA2a and Cx43 was sharply decreased so much in infarct zone of MI rats that we can hardly found uorescence signals, strongly indicating that gene expressions of SERCA2a and Cx43 failed due to the myocardial ischemic stress [7]. A reasonable explanation is that it's not only the loss of cardiomyocytes but also the absence of crucial protein that induced post-MI heart failure and arrhythmias. Our results also illustrated that the level of SERCA2a and Cx43 in the infarct zone cannot be reversed by only BMSCs transplantation. Thus, cell-based intervention was not enough to regain mechano-electrical function of damaged heart.
Fortunately, the level of SERCA2a and Cx43 showed a considerable elevation through UTMD-mediated dual gene co-delivery. By contrast, the expression of SERCA2a presented a robust increase in B+S+U group, while level of Cx43 did not show any improvement. Similarly, B+C+U group only showed a promotion in Cx43 expression but not in SERCA2a, implying that there was no obvious correlation between these two proteins.
It has been well elucidated that rat BMSCs can secrete paracrine factors including VEGF-1 to trigger angiogenic and migratory effects at the site of the infarct to promote myocardial healing and to improve the cardiac function [39]. In the current study, we also found that the level of VEGF and MVD in the infarct zone had a signi cant increasement after BMSCs transplantation, indicating that paracrine effect plays an essential role in BMSCs tissue repair and angiogenesis. Notably, our results revealed that overexpression of Cx43 upgraded the angiogenesis effect by BMSCs, while overexpression of SERCA2a brought no interest to local VEGF level. With these data, we assume that overexpression of Cx43 can improve the microenvironment of infarct area by promote the level of VEGF. Indeed, more cardiomyocytelike cells and higher level of myocardial markers in the infarct zone were detected among the groups with BMSCs transplantation. Although we cannot identify the origin of these cells, we can speculate that BMSCs in combination with Cx43 will signi cantly relive the harsh conditions within the infarct zone and help to rejuvenate cardiac function.
Myocardial contractility and electrical stability of myocardium are two most critical evaluation indexes for cardiac function. In the current study, we applied echocardiography and ECG to record the alteration of cardiac function in each group after treatment. From the results of echocardiography, we can see that ischemia brings a disastrous blow upon systolic function, which was manifested by plummeting LVEF.
Moreover, arrhythmias could be induced very easily in MI rats by steady-state pacing, which indicates a poor electrical stability of myocardium. Furthermore, as revealed by our results, rats received BMSCs transplantation alone do not show any solid signs of recovery both in myocardial contractility and electrical stability. Until now, we can deduce that the effect of BMSCs transplantation in cardiac function recovery is limited. However, gene therapy combined with cell transplantation can bring with encouraging news. Echocardiography clearly showed that ventricular wall motion of those rats received SERCA2a transfection was signi cantly improved, indicating the essential role of SERCA2a in maintaining mechanical function of myocardium [33,40]. Moreover, we also discovered that rats treated with SERCA2a showed a better resistance to pacing induced arrhythmias than MI group, but this virtue of SERCA2a overexpression vanished when we turn up the voltage, indicating a limited effect of SERCA2a in restoring electrophysiological properties. In particular, we revealed that overexpression of Cx43 in MI heart could effectively suppress pacing induced arrhythmias, but did nothing to restore myocardial contractility.
Based on these ndings, we surmised that there were no interactions between SERCA2a and Cx43 [8], and overexpression of neither of them solely could meet our goal. Guided by these ndings above, we applied two genes together after BMSCs transplantation. Remarkably, rats receiving dual-gene therapy showed an intrinsic improvement both in myocardial contractility and electrophysiological properties, highlighting the necessity for rebuild the gene network in damaged heart rather than the supplement of a single particular gene [41].
Conclusions the ndings in the present study, together with the published studies from our own laboratory and from other research group, came to the following conclusions: 1) UTMD is a reliable tool to realize site-speci c gene delivery, and in the current study, we managed to deliver two genes in one administration; 2) BMSCs transplantation can improve the microenvironment of infract zone, however it does few to restore cardiac function; 3) Cell-based dual gene therapy solidly ameliorates myocardial contractility and electrophysiological properties, emphasizing the necessity for rebuild the gene network in a damaged heart rather than the supplement of a single gene. Tables Table 1 The Characteristics of BMBs.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.