Design and fabrication of MPS
The MPS consists of two main parts: a SA-CNTs-based flexible pacing electrode in direct contact with the surface of the heart, and a CNTs-film-based power assist unit packed in nontoxic PDMS (Fig. 1b and d). The power assist unit is an electrical driven electrochemical actuator consisting of two CNTs-film electrodes and an ions-filled polymer gel electrolyte. CNTs are widely explored for the wearable and implantable devices due to their unique structure, excellent mechanical property, large surface area, excellent charge transport capability and tunable morphology [27,28,29,30]. The SACNTs in the pacing electrode part were prepared by dryspinning process from a spinnable CNT forest, which was synthesized via chemical vapor deposition (CVD) method [31]. SACNTs show a highly aligned morphology, where CNTs tend to form CNT bundles due to the Van der Waals’ force, thus leaving nano-scaled free space among them (Fig. S1, Supporting Information) and can offer an anisotropic interface between SACNTs and contacted cells.
Biocompatibility and interaction between SA-CNTs and Epicardial cells
Biocompatibility is essential to developing implantable device. The MPS was mostly packaged in biocompatible PDMS [32, 33], except for the SA-CNTs based electrode, which act as the interface between MPS and heart and is directly contacting the epicardium of a heart. Therefore, interfacial interactions between SA-CNTs and cardiac epicardial cells were firstly investigated. In our previous study, besides an excellent biocompatibility, SACNTs demonstrated a function of guiding the orientation and extension of cardiomyocytes, and showed an efficient transmitting of electrical signal for the cultured cardiac tissue in vitro [15]. In this work, the effect of SA-CNTs on epicardial cells was investigated by seeding primary epicardial cells onto the SA-CNTs and compared with those cells seeding on the Coverglass under the same condition. Primary epicardial cells were obtained by outgrowth culture from embryonic mouse hearts [26]. By planting an embryonic mouse heart (E 12.5) on the substrate and culturing in vitro, epicardial cells at the heart surface/substrate interface migrated and expanded spontaneously from the heart to the surface of the substrate as a result of epithelial spreading (Fig. S2, Supporting Information) [26]. Epicardial cells were identified with characteristic mesothelial marker (WT1) and organized epithelial tight junctions (ZO1) (Fig. S3, Supporting Information). After a certain amount of epicardial cells had reached to the substrate, the original embryonic mouse heart can be removed, and the morphology and physiologic properties of these epicardial cells on the substrates were further evaluated.
Epicardial outgrowth from the original embryonic mouse heart to both Coverglass and SA-CNTs were observed within 24 h after explant planting (Fig. S4, Supporting Information). The epicardial outgrowth primarily consisted of mesothelial cells with cobblestoned morphology, which is consistent with previous reports [21, 26]. The cell outgrowth was then traced and recorded during 4 days of incubation. At Day2, epicardial outgrowth on Coverglass (control) formed a tightly packed sheet of epithelial cells, while cells on SA-CNTs begun to exhibit a lamellipodia and migratory behavior, especially for those peripheral cells in the epithelial sheets (Fig. S5, Supporting Information). At day3, the outermost cells from the center heart on Coverglass seemed stuck to the substrate and stopped moving forward, resulting in a newly formed “Dam”. The latter outgrowing cells would be trapped in this “Dammed lake”, until the trapped cells accumulated to an extent that exceed the capacity of the “Dam”, they broke through the embankment and pour out (Fig. 2a-c and S6, Supporting Information).
In comparison, the migration behavior of peripheral cells on SA-CNTs was quite different, because the surface provided by SA-CNTs is anisotropic. Firstly, the formation of cellular “Dam” was rarely observed on SA-CNTs (Fig. 2d-f). Secondly, when the epicardial cells on Coverglass migrated around in a non-directional manner (La = Lb, Fig. S7, Supporting Information), the cells on SA-CNTs tended to migrate more along the orientation of the SA-CNTs (La > Lb, Fig. 2g). Therefore, the migration distance of epicardial cells in the SA-CNTs aligned direction is longer than that in the vertical direction (Fig. 2h). The SA-CNTs is ultra-thin (around 20 nm) compared to the micrometer-scaled cells, therefore the CNT bundles showed no obvious hindrance to cell migration at the vertical direction. However, the SA-CNTs promoted cell migration at the orientation direction, which was corresponding to their guiding function of elongated cell morphology for cultured cardiomyocytes in our pervious study [15]. This cell migration guidance phenomenon was confirmed by quantitative measurements (Fig. 2h and S8, Supporting Information). These results suggest that by controlling the physical interface interactions, which is the aligned morphology of CNTs in this work, more pro-regenerative epicardial cells can be directed to the infarcted area to participate in cardiac repair.
On the other hand, the SA-CNTs-based flexible pacing electrode is expected to provide electrical pacing signals to the contacted heart. So, continuous electrical impulse (rectangular, 2 ms, 2 V cm− 1, 1 Hz) was applied to the epicardial cells by connecting the conductive SACNTs to a pacemaker. The electrical stimulation started at 48 h after the heart planting onto the substrates, when the epicardial outgrowth had become relatively stable. Results showed that the cell migration behavior with electrical stimulation was similar to that observed on SA-CNTs without stimulation (Fig. S9, Supporting Information).
Moreover, the biocompatibility of SA-CNTs was also evaluated by detecting the cytotoxicity of SA-CNTs with/without electrical stimulation using TUNEL (TdT-mediated dUTP nick-end labeling) staining of apoptotic cells. Both the SA-CNTs with and without electrical stimulation showed no significant cytotoxicity to the epicardial cells compared to the Coverglass (Fig. S10, Supporting Information).
The morphological and behavioral changes observed in epicardial cells on SA-CNTs suggest that the cells on SA-CNTs may undergo EMT. EMT is a biological process, in which epithelial cells lose cell polarity and epithelial phenotypes, transform into mesenchymal phenotype cells, and obtain higher interstitial phenotypes such as migration and invasion [34, 35]. To verify this assumption, expressions of EMT markers of outgrowing cells on SA-CNTs and Coverglass were assessed by immunocytochemistry and quantitative real-time polymerase chain reaction (qRT-PCR). Immunostaining of α-smooth muscle actin (α-SMA) indicated that its expression was increased in outgrowing epicardial cells on SA-CNTs, and further slightly enhanced after external electrical stimulation (Fig. 3a and b). qRT-PCR was also performed targeting on EMT markers α-SMA, Vimentin and Smooth muscle myosin heavy chain (SM-MHC), and EMT inducer Transforming growth factor (TGF)-β1. For these epicardial cells, SA-CNTs increased significantly the mRNA expression of EMT inducer TGF-β1 (Fig. 3b). The mRNA expression of α-SMA, SM-MHC and Vimentin were also increased in SA-CNTs and E-SA-CNTs group, but the difference was not statistically significant (Fig. 3b). Moreover, corresponding to the above microscopy observation, immunostaining results indicated that the SA-CNTs increase the cells in lamellipodia, and enhance the cell separation (a significant increase in distance between nuclei, Fig. 3c and d), which further confirmed the enhanced cell migration and the accelerated EMT.
Performance and mechanism of power assist unit
The second part of the MPS is a power assist unit, which is directly attached onto the other side of the flexible pacing electrode. The power assist unit is an electrical driven actuator, providing assisted systolic-diastolic force along with the required synchronous electrical impulses for the seriously damaged heart. In this work, a flower-shaped MPS was designed to envelop the apical part of the heart (where frequently develop akinesis or even dyskinesis due to infarction expansion [36, 37]) and provide centripetal contractions-relaxation (Fig. 1a and d). The shape and size of the MPS can be tailored to suit the target heart and envelop the desired area (Fig. S11, Supporting Information). The electrical driven actuator was assembled by sandwiching an ions-filled polymer electrolyte between two flexible CNTs-film electrodes and packed in nontoxic PDMS (Fig. 1b). Each actuator petal was light-weight (0.3 × 2 cm2, 3.6 mg) with a thickness of around 50 μm (Fig. S12, Supporting Information). Different from the SA-CNTs in the pacing electrode, CNTs-film was commercially produced with a thickness of 5–10 μm, where the CNTs was randomly dispersed (Fig. S13, Supporting Information). This film showed excellent conductivity and flexibility (3.0 × 104 S m− 1, 60–120 MPa), which can meet the demands of an actuator at low cost.
The working principle of the electrochemical actuator petal is based on the electric double layer phenomenon [38,39,40,41]. When potential difference was applied at the two electrodes, ions (positive: H+, negative: SO42−) in the gel electrolyte run towards their corresponding electrodes to balance the internal electric field. Due to the different volume of positive and negative ions, their directional movement and accumulation led to asymmetric volume changes near these two electrodes, consequently resulting in a deformation of the actuator petal (Fig. 4a). When the voltage of external circuit reversed, the actuator petal bent to the opposite direction. A petal in deformation can lift an object ≈ 4.7 times of its own weight and push an object ≈ 35.3 times of its own weight, which is similar to that of an adult human heart (250-350 g) outputting a force of ≈ 5.90 times of its weight (Fig. S14, Supporting Information) [42, 43]. Therefore, the heart beating mimicked systolic-diastolic behavior can be realized through cyclic alternating voltage.
Alternating voltage at different frequency was provided by CV method (Electrochemical workstation) with different scan rates. The as fabricated electrochemical actuator is also an electrochemical supercapacitor [40, 41, 44, 45], which was indicated by the CV curves (Fig. 4b). The device showed a typical supercapacitor behavior at increased scan rates from 1 to 25 V s− 1 with the potential range of − 2.5 V to + 2.5 V or − 2.0 V to + 2.0 V (Fig. 4b, and S15, Supporting Information). The specific capacitance (Csp) based on the CV curves was calculated with:
$$ {C}_{sp}=\frac{1}{2\bullet s\bullet m\bullet \Delta V}{\int}_{V0}^{V0+\Delta V} idV $$
Where i is current and V is potential, ∆V is potential range, s is scan rate, and m is mass of the whole device. Csp was 1.3 F g− 1 at low scan rate of 1 V s− 1, and was 0.7 F g− 1 at high scan rate of 25 V s− 1 (− 2.5 V to + 2.5 V). The shape of CV curves was mostly maintained at high scan rates, indicating a fast response capability.
The calculated Csp decreased to 0.6 g− 1 at low scan rate of 1 V s− 1, and 0.4 F g− 1 at high scan rate of 25 V s− 1 when the potential range narrowed to − 2.0 V to + 2.0 V, which was also corresponded to the actuator’s systole-diastole mimicked behavior. Here, contraction ratio (CR) was defined as (end diastolic dimension – end systolic dimension)/end diastolic dimension× 100%. The CR of a flower-shaped power assist unit was 19.4% when under the operation voltage range of − 2.5 V to + 2.5 V, and was 9.4% when under the operation voltage range of − 2.0 V to + 2.0 V (Fig. S16, Supporting Information). Although the CR decreased with the increased potential scan rate, the CR at 25 V s− 1 still remained 58% of that at 5 V s− 1 (Fig. S17, Supporting Information).
On the other hand, the electrolyte prepared here involved a small amount of water (lower than 10% weight percentage), while the industry produced CNTs films contained some metal catalyst (e. g., Fe, Ni, 5–15% weight percentage). Therefore, both the CV test of the assembled device or the three-electrode electrochemical testing system (CNTs film as working electrode v.s. Ag/AgCl electrode, Fig. 4c-d, S18, Supporting Information) revealed an irreversible electrode redox reaction. The irreversible reaction tent to be more noticeable at the wider operation voltage window, but was reduced after 10,000 cyclic scan process (Fig. 4c). Meanwhile, no significant change was observed during 10,000 cyclic operations at − 2.0 V to + 2.0 V (Fig. 4d).
Contraction behavior of the power assist unit was further evaluated by finite element method analysis. In fact, a normal heart contracts in a centripetal manner when bumping blood. However, for a failed heart post-infarction, the infarcted area is replaced by a thin fibrous scar and thus lost the ability to contract, or even exhibits dyskinesis and bulges out during systole; while the non-infarcted area undergoes remodeling, thinning and develops hypokinesis (Fig. 4e). The computer simulated stress distribution map and structure deformation map indicated that the bottom of the flower-shaped device has a larger stress with a smaller deformation, while the top part of each petal reveals a small stress with large position shifting (Fig. 4f and S19, Supporting Information). This result suggests that the flower-shape design may offer a structure restriction to the thin scar area while providing additional inward force to the myocardium with reduced contractility. On the other hand, the Young’s modulus of the actuator petal prepared in this work was 334 ± 11 MPa (Fig. S20, Supporting Information), which can be tuned and further influences the stress distribution of the device. Thus, the inward force for the failed myocardium can be enhanced by increasing the Young’s modulus of the actuator petals (Fig. S21, Supporting Information).
In general, the actuator’s systole-diastole mimicked behavior was controlled by the alternating voltage. 1) The power assist unit always showed a fast responsiveness when the voltage change frequency accelerated from 5, 10, 15, 20 to 25 V s− 1, corresponding to the heart beat of 30, 60, 90, 120, and 150 beat per minute (bpm) (Fig. 4g, Video S1, Supporting Information), which far covered the normal heart rate range (60–100 bpm); 2) The contraction amplitude (expansion dimension minus contraction dimension) decreased 10% when the voltage change frequency accelerated 25 times or the operation voltage window decreased from ±2.5 V to ±2.0 V. This phenomenon is similar to the case in human, when the heart rate goes up, the ejection fraction decreases. For patients with heart diseases, an increase in heart rate from 83 bpm to 154 bpm results in 31% reduction in ejection fraction [46]. It is worth noting that one aim of this work is to explore the feasibility of our design, such as the matchability between the stress and deformation distribution, and the response frequency of the flower-shaped electrochemical driven actuator compared with organs. By learning from the achievement of other works, optimized actuator with stronger contraction force and safer electrolyte can be further explored [38, 47, 48].
Application and biostability of the MPS in vivo
The clinical applications of the MPS were demonstrated with ex vivo organs. Due to the excellent flexibility and light-weight, the MPS was easily attached directly onto the surface of a Tyrode’s solution perfused rabbit heart, covering the apical portion of the heart where includes both the left and right ventricles (Fig. 5a-b). The electrocardiogram (ECG) of the rabbit heart was monitored in real-time to indicate the beating rhythm of the heart (Fig. 5c and S22, supporting information). When the pacemaker connected MPS device was turned on to output controlled pacing signals, a continuous and regular ECG tracing was observed. This result suggests that the device offers a pacing function owe to excellent attachment, and is capable of contracting and dilating along with the beating heart (Fig. 5b-c). Therefore, the MPS is designed for failed hearts with deteriorative contractility, especially for those hearts accompanied with asynchronous contractions of left and right ventricles.
Finally, this MPS was subcutaneously implanted in mice to assess in vivo biodegradation and local interaction between the implant and the tissue. Implanted MPS was excised for evaluation at 2 weeks and 4 weeks. The shape and size of the implants remain the same after 4 weeks, indicate that the MPS is stable in vivo without structure destruction or biodegradation (Fig. 5d and e, and S23, supporting information). Hematoxylin and eosin staining showed that inflammatory cells increased near the interface of MPS and tissue at 2 weeks, which indicates a postoperative inflammatory response. However, this inflammatory response disappeared at 4 weeks after implantation (Fig. 5e and S23, supporting information). The boundary between the MPS and the tissue is regular and clear, suggesting that the interaction between the MPS and tissue is very mild without noticeable fibrous tissue proliferation or adhesion (Fig. 5e and S23, supporting information). These results indicate that the MPS has excellent biocompatibility and structural stability.
The applications of the electrical driven power assist unit can be further expanded to other dynamic organs with disordered or failed function, such as urinary bladder and stomach (Fig. 5f), to provide rational choice for their future clinical therapeutic scheme. In the future work, optimized actuator devices with stronger contraction strength and bigger contraction amplitude can be further explored. It should be noted, in practical applications, the stronger or bigger contraction does not necessarily lead to better outcomes. The functional characteristics of the device, such as contraction intensity and amplitude, should be carefully customized to match the target organs before clinical transformation.