Digital Light 4D Printing of Bioresorbable Shape Memory Elastomers for Personalized Biomedical Implantation

Abstract

Four-dimensional (4D) printing unlocks new potentials for personalized biomedical implantation, but still with hurdles of lacking suitable materials. Herein, we demonstrate a bioresorbable shape memory elastomer (SME) with high elasticity at both below and above its phase transition temperature (T_trans). This SME can be digital light 3D printed by co-polymerizing glycerol dodecanoate acrylate prepolymer (pre-PGDA) with acrylic acid monomer to form crosslinked Poly(glycerol dodecanoate acrylate) (PGDA)-Polyacrylic acid (PAA), or PGDA-PAA network. The printed complex, free-standing 3D structures with high-resolution features exhibit shape programming properties at a physiological temperature. By tuning the pre-PGDA weight ratios between 55 wt% and 70 wt%, T_trans varies between 39.2 and 47.2 °C while Young's moduli (E) range 40-170 MPa below T_trans with fractural strain (ε_f) of 170%-200%. Above T_trans, E drops to 1-1.82 MPa which is close to those of soft tissue. Strikingly, ε_f of 130-180% is still maintained. In vitro biocompatibility test on the material shows > 90% cell proliferation and great cell attachment. In vivo vascular grafting trials underline the geometrical and mechanical adaptability of these 4D printed constructs in regenerating the aorta tissue. Biodegradation of the implants shows the possibility of their full replacement by natural tissue over time. To highlight its potential for personalized medicine, a patient-specific left atrial appendage (LAA) occluder was printed and implanted endovascularly into an in vitro heart model.

Graphical Abstract

1. Introduction

Personalized care and minimally invasive surgery (MIS) hold paramount importance in clinical biomedical settings[1]. Among them, the devices that cater to these clinical needs are called patient-specific biomedical implants. Three-dimensional (3D) printing technologies in combination with advanced imaging techniques have offered enormous opportunities for high-rate and cost-effective fabrication of these personalized devices for various biomedical applications, e.g., patient-specific airway stents[2], orthopedic instruments and implants[3], and spine surgery implants[4]. Despite such significant advancements in the past decades, their practical in vivo functionalities remain hindered by challenges concerning biocompatibility, mechanical properties of the materials as well as limitation on geometrical complexity of the printed implants[3, 4]. More critically, the shapes of the as-printed implants are static, limiting their capability in adaptation to the tomography and topology of the targeted organs or tissues, thus restricting their practical implantation via the MIS techniques[5].

Recently, 4D printing, known as 3D printing of responsive materials, has captured much attention of the scientific community[6, 7]. The printed complex 3D structures have mutable, transformative configurations, features, and functions in response to external stimuli over “time”— the 4^th dimension. Among various types of responsive materials[8] such as anisotropic hydrogels[9] and liquid crystal elastomers[10] used as the raw materials for 4D printing, shape memory polymers (SMPs)[11], with their inherent phase-changing properties, can respond to a multitude of stimuli like heat[12], electric fields[13], and light[14]. Thermo-responsive SMPs, in particular, have the capability to recover their shapes when heated above their phase transition temperatures (T_trans), rendering their applications in 4D printing of biomedical implants for enophthalmic invagination[15], endovascular embolization[16], heart occlusion[17], respiratory tract patency[18], and vascular grafting[19]. Nevertheless, a major limit inherited in these 4D printing studies includes a lack of biocompatible SMPs with suitable thermomechanical properties. Most of them are not biocompatible and have no suitable T_trans (either much lower than 20 °C or much higher than 37 °C), thus not easy for deployment at body temperature[20, 21]. Further compounding these challenges, the literature reveals that while there is abundant research on biodegradable and elastic SMPs [22-24], their potential in 4D printing has not been fully realized [25]. Pioneering efforts have established a foundation for 4D printing bioresorbable SMPs by DLP to fabricate a tracheal stent from a methacrylated poly(ε-caprolactone) (PCL) based resin [26]. Despite these advances, the elasticity and toughness of these SMPs at elevated temperatures (above T_trans) remain suboptimal. For instance, Zhang et. al. reported 4D printing of a biocompatible SMP which shows greatly reduced fractural strain (ε_f), or called elongation at fracture, from 209 to 35% above its T_trans[19]. In contrast, Zhang et. al. introduced a shape memory elastomer (SME) with high elasticity with ε_f of ~500% and ~250% at below and above T_trans[12], respectively, but it appears to lack biocompatibility based on its monomer composition and has too high T_trans of > 45 °C, which would restrict its potential in biomedical implantation. A recent study demonstrates a possibility of maintaining good elasticity and stretchability of 4D printed biodegradable polymer to address some of these challenges. But the biocompatibility of the prints has not yet been shown in vivo [27]. Moreover, although many 3D printing technologies such as inkjet printing (polyjet)[28], fused deposition modeling[29], and selective laser sintering (SLS)[30] are available for printing SMP constructs, their printing resolutions are much lower than those of light-based ones, e.g., digital light processing (DLP). But there are limited options on the photo-curable inks that can be used for DLP[31]. These technical obstacles greatly restrict the widespread applications of DLP based 4D printing in biomedical implant fabrication[20].

Herein, we present a DLP based 4D printing of a biocompatible, biodegradable shape memory elastomer (SME) with tunable elasticity and toughness. Our SME, with a phase transition temperature range suitable for biomedical applications, waswas synthesized from a UV-active poly(glycerol dodecanoate) acrylate (PGDA) prepolymer and acrylic acid (AA), forming a crosslinked PGDA-polyacrylic acid (PAA) copolymer. This formulation ensures high-resolution printing, biocompatibility/biodegradability of resulting materials [32], and favorable shape memory properties for in vitro and in vivo applications. Leveraging a patient-specific occluder designed from CT scans, we demonstrate an adaptable left atrial appendage occlude (LAAO) in vitro implantation. The successful demonstration of the patient-specific LAAO serves as a proof-of-concept to illustrate a promise of our material in personalized medical device fabrication. This early-stage study can be a promising step forward in vascular occlusion/embolization in patients with arterial aneurysm, bleeding or need for therapeutic embolization of complex structures such as portal veins or vascular malformations. This innovation propels 4D printing into new realms of biomedicine and tissue engineering, promising to bridge gaps between synthetic and biological structures.

2. Materials and Methods

2.1. Materials.

4-Dimethylaminopyridine (DMAP, 99%), Glycerol (>99.5%), dodecanedioic acid (DDA, 99%), triethylamine (> 99%), Dichloromethane (>99.8%), and acryloyl Chloride (> 97%) were purchased from Sigma-Aldrich (St. Louis, MO, USA). ethyl acetate (99.5%), diphenyl(2,4,6-trimethylbenzoyl)phosphine oxide (TPO, > 98%), and Acrylic acid (AA) (98%) were purchased from Fisher Scientific (Pittsburgh, PA, USA). 4-methoxyphenol (99%) was purchased from Acros Organics. These materials were used without further purification.

2.2. PGDA pre-polymer synthesis.

PGD prepolymer (pre-PGD) synthesis was conducted using a methodology previously described in our work [19]. In a three-necked flask, glycerol and dodecanedioic acid (DDA) were mixed in a 1:1 molar ratio. This mixture was then heated to 120 °C by an oil bath for 24 hours under nitrogen flow with magnetic stirring using the synthesis protocol outlined in previous studies [33, 34]. The resulting pre-PGD had a number molecular weight (Mn) of ~1300, a weight molecular weight (Mw) of ~ 2200, and a polydispersity index of ~1.6 [34]. To acrylate the pre-PGD to get pre-PGDA, a base solution was prepared by mixing 0.1 g of 4-methoxyphenol, 0.2 g of DMAP, 4.9 mL of triethylamine, and 200 mL of dichloromethane. This base solution was then used to dissolve 20 g of pre-PGD. After cooling the pre-PGD solution to 0 °C under nitrogen for 10 min, 3 mL of acryloyl chloride (0.18 mol/mol hydroxyl groups on the pre-PGD), which was pre-diluted in 30 mL dichloromethane, was added dropwise. The acrylation process was performed in accordance with the procedure reported before, aiming to achieve an acrylation percentage of ~18% [35].

After that, aluminum foil was used to seal the reaction vessel, which was then stirred at room temperature (T_R). After reaction for 12 hours, additional 0.1 g of 4-methoxyphenol was added. The solution was dried in a rotary evaporator by removing the dichloromethane, after which it was then dissolved in 100 mL of ethyl acetate. The supernatant was first dried in a rotary evaporator and then further dried for three days in a vacuum chamber. To separate the solubilized pre-PGDA from the triethylamine salt by-product, the mixture was centrifuged at 10,000 rpm for 10 minutes.

2.3. Ink preparation and printing process.

pre-PGDA and AA were mixed at various pre-PGDA weight percentages of 70%, 65%, 60%, 55% and 50%. The photoinitiator diphenyl(2,4,6-trimethylbenzoyl) phosphine oxide (TPO) was added to pre-PGDA-AA resin with a concentration of 3 wt%. TPO was first dissolved in AA, and the melted pre-PGDA at 65 °C was slowly added to the AA/TPO solution while being stirred with a magnetic stir rod. After they were homogenously mixed, the resin was poured into an in-built vat of a B9Creations DLP printer (Core 550). B9Create software was used for G-code generation from the 3D models. The constructs were printed by an irradiation wavelength of 405 nm under a power density of ~ 5 mW/cm² and the thickness of each layer was set to 50 μm. The printed objects were detached from the collector surface, washed with ethanol to remove the unreacted resin, and finally post-cured by 405 nm UV light for 600 s.

2.4. Material characterizations.

A Thermo Nicolet 380 FTIR spectrometer with DIAMOND ATR was used to collect FTIR spectra. Viscosity of resion was evaluated by a modular rotation and interface rheometer MCR302 equipped with a C60/2°. The test was performed at T_R with shear rates changing from 0.1 to 100 1/s. Differential scanning calorimetry (DSC) measurements were done with TA Instruments, Q-600 DSC where the temperature was decreased to −30 °C followed by ramping from −30 to 100 °C at a constant rate of 10 °C/min.

2.5. Thermomechanical properties characterization.

Tensile tests were performed on a Mark-10ESM303 universal testing apparatus. For these tests, ASTM-D638 Type IV dog bone-shaped specimens were oriented flat against the DLP print plate, building up the sample layer by layer in the thickness direction. The initial length of the samples was 25 mm and a strain rate of 50 mm/min was applied. The cross-sectional dimensions of each sample were measured using a digital caliper for calculating the cross-sectional areas. For each set of experiments, four samples were tested both at T_R and above T_trans. to get statistical results. For the tensile tests performed above the transition temperatureT_trans), a heat gun was utilized to uniformly raise the temperature of each sample above its T_trans, which was maintained throughout the testing process. The temperature was continuously monitored using an adjacent infrared thermometer to ensure accuracy and consistency. The resulting stress-strain curves were used to calculate the mechanical properties of the printed samples. Storage modulus, loss modulus, and tangent delta were measured using Hitachi Dynamic Mechanical Analyzer (DMA7100) under the tensile mode, with printed rectangular 50 mm x 10 mm x 1.5 mm samples.

2.6. Shape memory behavior characterization.

We investigated the shape memory behavior of the PGDA-PAA SME samples by following a typical shape memory cycling method. The sample was first stretched by 100% at a programming temperature (i.e., T_trans + 30 °C). Then, the temperature was decreased to 25 °C. After reaching the targeted programming temperature, the sample was held isothermally for 2 min. The strain of the temporary shape was measured after removing the external load. In the free recovery step, the temperature was gradually increased to the recovery temperature (i.e., T_trans + 30 °C). The sample was held isothermally for another 1 min to observe the free recovery behavior.

2.7. In vitro biodegradability evaluation.

Printed materials were submerged in 50 mL pure phosphate-buffered saline (PBS) and 50 mL PBS with 0.1 mM NaOH, respectively, at 37 °C. Samples were removed at designated time points and then dried overnight at T_R. They were weighed after being dried to determine mass loss.

2.8. Cytotoxicity test.

The biocompatibility of the printed materials was tested by co-culturing them with pluripotent mesenchymal progenitor C3H10T1/2 (10T1/2) cells in a 6-well plate for the cell counting or on coverslips in a 24-well plate for fluorescent imaging in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2mM L-glutamine, 100U/mL penicillin and 100 μg/mL streptomycin at 37 °C in a humidified atmosphere with 5% CO2. Before the co-culturing assay, the printed materials were first sterilized in ethanol and then incubated in DMEM for 48 hours to diminish the effect of the materials on a medium PH value. Cell viability was tested on Day 1, 2, 3 and 4 after the co-culture, respectively. Images were captured using a Nikon microscope after each day of culturing. For cell quantification, the cells were dissociated with trypsin and then counted using a Bio-Rad TC10 automated cell counter. To take the fluorescent images, cells on coverslips were rinsed with PBS and then incubated in propidium iodide (PI) staining solution at 4 °C for 15 minutes. The samples were then protected from light, rinsed twice with PBS, and then fixed in 10% formalin at T_R for 10 minutes. Coverslips were then rinsed with PBS twice and mounted on glass slides with prolong gold antifade mounting medium containing DAPI. Fluorescent images were captured using a Keyence microscope and processed using ImageJ. To calculate survival rates, the number of cells positive for PI was divided by the total number of cells. Analysis on the surface of printed, degraded, and dried materials was done using scanning electronic microscopy (SEM, FEI Quanta 600F Environmental SEM).

For cell attachment assay, 13 cm disc scaffolds were printed. The discs were sterilized in 100% ethanol and then soaked in DMEM for 48 hours. 10T1/2 cells were plated with the scaffold and allowed to adhere and proliferate for 1 week. After 1 week of culture, the scaffolds were rinsed twice with PBS then fixed in 10% formalin for 1 hour, rinsed twice with PBS, and allowed to dry completely. The following day, scaffolds were prepared for SEM analysis.

2.9. Anastomosis of printed tubes implanted into mouse aorta.

The anastomosis of the 3D-printed PGDA-PAA tube implanted to the mouse aorta was conducted using a well-established aortic transplantation method in a mouse model [36]. In preparation for the procedure, the recipient mice were anesthetized using a carefully measured blend of 1.5% (by volume) isoflurane and pure oxygen, delivered via a face mask. This was followed by a meticulous process of hair removal and disinfection in the abdominal area to minimize the risk of infection. A mid-line incision was made, stretching from the xiphoid process down to the pelvis. Special attention was given to dissect the infrarenal aorta located between the renal arteries. Any small branches off this segment were skillfully ligated using an ultra-fine 11-0 monofilament suture to prevent bleeding. Proximal and distal portions of the aorta were secured using clamps, and the intervening aortic tissue was carefully excised. The aortic lumen was then flushed with sterile saline to remove any residual blood or debris. The printed tubes were positioned in the orthotopic location, effectively replacing the removed aortic segment. Anastomosis was performed on the proximal and distal ends of the abdominal aorta, employing an end-to-end pattern using an 11–0 polyamide monofilament suture. This suturing technique ensured the secure attachment of the graft while minimizing potential damage to the vessel walls. Following the completion of the anastomosis, the clamps were gently removed to restore the aortic blood flow. We closely monitored the graft, noting the presence of visible pulses as an indication of successful blood flow restoration. The abdominal cavity was then carefully closed, ensuring that all contents were correctly repositioned to avoid any postoperative complications. The wound was closed using a 4-0 polyglycolic acid suture, which was chosen because of its strength and biocompatibility. Postoperative care included euthanizing the animals 21 days after transplantation as per the experimental design. The 3D printed tube, along with any tissues that had adhered to it, was then meticulously excised. The harvested tube material was preserved in a 4% paraformaldehyde (PFA) solution, embedded in 2% paraffin for structural stability, sectioned using a precise microtome, and subsequently analyzed using various histological stains to evaluate tissue integration and graft performance. This animal surgery procedure was conducted under the approval of the Institutional Animal Care and Use Committee of the University of Missouri (ACQA # 40285), ensuring that all operations were performed under stringent ethical guidelines.

2.10. Histology.

The paraffin-embedded sections of the 3D printed tubes underwent a detailed Hematoxylin and Eosin (H & E) staining process to reveal key histological features of the adhered tissue. This staining was performed using high-quality reagents sourced from StatLab, based in McKinney, TX, USA. Briefly, the paraffin sections were deparaffinized and rehydrated through a series of graded alcohol solutions. Once the samples were properly prepared, they were subjected to Hematoxylin staining. Hematoxylin, which colors cell nuclei a deep blue/purple, was applied to sections, allowing enough time for adequate staining. The samples were then rinsed to remove excess Hematoxylin and differentiate the staining with a bluing reagent to accentuate the color. The second component of the staining process involved the application of Eosin, which provides a pink color to the cytoplasm of the cells, creating a contrast against the Hematoxylin-stained nuclei. The sections were thoroughly rinsed again to remove any unbound Eosin. Finally, the stained sections were dehydrated through an ascending alcohol series, cleared in xylene, and mounted with a coverslip using a mounting medium. This process preserved the stained sections and prepared them for microscopic analysis. Images of these H&E-stained sections were captured using a high-resolution Nikon microscope. The microscope was meticulously calibrated to ensure accurate representation of the staining. The images were taken with a focus on areas of interest and adhered tissues to analyze the interactions between the printed tube and cells. The results of this staining and subsequent imaging provide vital information on the biocompatibility and tissue integration of the 3D printed PGDA-PAA tube. Thrombosis formation rate and material degeneration rate was calculated based on previous publication[37].

2.11. Immunofluorescence (IF) staining.

The immunofluorescent staining was conducted based on a previously established protocol[38], designed to assess the presence and location of specific cell types within the sections of the transplanted PGDA-PAA tubes. The fixed sections of the tubes underwent a deparaffinization process to remove paraffin wax used during embedding. The sections were then rehydrated using a graded series of alcohol solutions, transitioning from a high concentration down to water. This process restored the aqueous environment needed for further staining procedures. After rehydration, the sections were rinsed with Phosphate-Buffered Saline (PBS), a standard solution used to maintain the pH and osmolarity of the samples. This was followed by permeabilization with a solution of 0.5% Triton X-100 in PBS. Permeabilization is a crucial step, as it increases the permeability of the cell membrane, allowing for antibodies to penetrate the cells and bind to their targets. To prevent non-specific antibody binding, sections were blocked with a solution of 2% Bovine Serum Albumin (BSA) for 1 hour at T_R. BSA is a common blocking agent used to occupy potential binding sites, reducing background staining. The sections were then incubated overnight with primary antibodies against CD31 (ab222783, Abcam) or FSP-1 (S100A4) (MA5-31332, Invitrogen) at 4 °C. These antibodies bind specifically to their respective antigens, allowing for detection of endothelial cells and fibroblasts, respectively. Following overnight incubation, the sections were incubated with fluorescent dye-conjugated secondary antibodies for 1 hour at T_R. These antibodies bind to the primary antibodies, providing a fluorescent signal that can be visualized under a microscope.

After three meticulous washes with PBS to remove any unbound secondary antibodies, the sections were mounted using an antifade reagent, which also contained DAPI (D21490, Invitrogen). DAPI is a fluorescent stain that binds strongly to DNA and is used to highlight the cell nucleus. Finally, the sections were observed under a Nikon fluorescent microscope, and images were captured. The resulting images provided a visualization of the distribution and localization of the endothelial cells and fibroblasts within the tissue adhered to the printed tube, offering insights into the biological integration of the graft.

2.12. In vitro demonstration of transcatheter LAA closure.

The clinical applicability of the Left Atrial Appendage Occluder (LAAO) was evaluated using a computed tomography (CT) scan from a 23-year-old patient. Using the scanned data, a heart model with detailed LAA was recreated and then 3D printed. This served as the model for testing the feasibility of the uniquely designed LAAO, which was developed using Fusion 360 based on the isolated LAA from the patient's heart scan. The LAAO was 4D printed using our newly developed PGDA-PAA SME system. To visually accentuate the shape recovery of the LAAO, the 3D-printed heart model was fabricated from a transparent material. The printed LAAO, programmed into a tubular form to fit a 4 mm inner diameter tube, was then delivered to the LAA. To emulate physiological conditions and facilitate the observation of occluder recovery, the heart model was immersed in saline maintained at body temperature.

2.13. In vivo implantation of locking devices.

The in vivo procedure necessitated the attentive preparation of a 6-month-old mouse under a rigorously controlled anesthesia administration. Using an expertly proportioned blend of 1.5% isoflurane (calculated by volume) and pure oxygen, the anesthesia was precisely delivered via a purpose-built face mask. During the procedure, the mouse's snout was carefully placed within a nose cone linked to the anesthesia system. This positioning was key in maintaining a steady state of sedation throughout the procedure, achieved by the delivery of a meticulously proportioned mixture of 1.0% to 1.5% isoflurane and 0.5 L/min 100% O2. To confirm that the sedation was effective, a gentle toe or tail pinch was performed. The anesthesia levels were attentively adjusted, aiming to reach a target heart rate of 450 ± 50 beats per minute (bpm), ensuring both the safety and efficacy of the surgical procedure. Preparation of the animal begins with meticulous shaving and disinfection of the ventral surface of the neck, performed via three alternating applications of 70% ethanol and betadine scrubs. The mouse was then positioned on a warming pad, and Bupivacaine was administered as a local anesthetic via subcutaneous injection at the forthcoming incision site minutes prior to making the incision. A midline incision of approximately 1.0 cm was then created on the neck's ventral surface. The locking device, which had been previously subjected to a heating process and shape-programmed to assume a flattened state at ambient T_R, was delicately maneuvered into place beneath the artery of the mouse, which had been surgically exposed in readiness. With an attentive and gentle approach, the heat-responsive device was positioned strategically to ensure the best results upon activation.

Following the precise positioning, a regulated amount of heat was applied, triggering the device's unique properties. It began its transformative process, gradually coiling around the aorta in a controlled and secure manner. This process was attentively monitored to guarantee the device wrapped completely and uniformly around the aorta, minimizing any potential for complications. Upon achieving its fully wrapped configuration, the device self-locked, securing itself firmly in place. Its placement was characterized by a high level of stability, firmly adhering to the contours of the aorta without causing undue pressure or damage. Post-placement, an intensive evaluation was conducted to ascertain the device's stability and fixation at the site of implantation. This critical assessment ensured the device was perfectly anchored, providing the optimal conditions for the successful continuation and conclusion of the procedure. This level of detailed scrutiny ensured that the device met all the necessary criteria for a successful and safe implantation.

2.14. Statistical analysis.

Experiments were repeated at least 3 times for statistical analysis. Values are expressed as the mean with standard deviation. Data values were first analyzed by comparing experimental values to control values by analyzing for Gaussian distribution using D’ Agostino & Pearson and Shapiro-Wilk normality tests (alpha=0.05, p<0.05). After passing normality, parametric statistical test, unpaired t-test with Welch’s correction was performed. Statistical analysis was conducted using GraphPad Prism 9 software, statistically significant differences were considered when nominal p < 0.05. All p-values and the corresponding statistical tests are provided in the figure legend.

3. Results and Discussion

3.1. Description on the patient-specific printing process of PGDA-PAA by DLP

Producing patient-specific, stimuli-responsive biomaterials for cardiovascular applications requires precision and thoughtful planning. While DLP presents a potential for creating high-toughness SMP at micron-scale resolution[12], its application to produce materials with confirmed in vitro cell studies and in vivo biocompatibility remains largely underexplored[5]. Poly(glycerol dodecanoate) acrylate (PGDA) is a biocompatible SMP with adjustable T_trans, presenting a promising candidate for biomedical applications[19, 39]. While 4D printing of PGDA vascular grafts was reported by us[19], it contends with issues concerning relatively low printing resolution and low material toughness above T_trans[19, 39]. Because they were printed direct ink writing (DIW), which also suffers from a comparatively slow printing pace. In response to these limitations, we demonstrate digital light 4D printing of personalized biomedical implants made from PGDA-PAA SME (Fig. 1a).

Fig. 1 ∣. DLP-based 4D printing of SME PGDA-PAA.

(a) Schematic showing the workflow of 4D printing a personalized LAA occluder, shape programming, implantation, and in situ deployment. Models and 3D printed structures: a standing man (b), a hand (c), a diamond structure (d), and a locking device with a wall thickness of 200 μm and an inner diameter of 1 mm (e). Scale bars: 2 mm.

It is a meticulous six-step process. First, CT scanning of a patient's heart is performed to map the intricate cardiac structures, providing a blueprint for the following steps (Fig. 1a-i). Armed with this precise anatomical data, a digital 3D design of an occluder is created, specifically tailored to well fit the geometry of the patient's LAA to ensure efficient occlusion (Fig. 1a-ii). Subsequently, the designed occluder comes to life through DLP of the developed PGDA-PAA SME (Fig. 1a-iii). After the occluder is printed, it is then subjected to shape programming, transforming it into a compacted, tubular structure that is suitable for transcatheter delivery (Fig. 1a-iv). Once prepared, the occluder is delicately transferred via a catheter to the LAA of a printed heart model, which is a pivotal step in the MIS (Fig. 1a-v). The process culminates in deployment of the occluder within the LAA, accomplishing occlusion to minimize the risk of thromboembolic events (Fig. 1a-vi). To demonstrate efficacy of this process, we first tested the resolution and accuracy by printing free-standing, complex 3D structures: a standing man, a hand, diamond, and a locking device (Fig. 1b-e). They show sub-200 um features with well-matched geometries with the designed models. To demonstrate the shape programming capability, the printed locking device was tested on a tube. The devices were first elevated above T_trans and then programmed to a temporary flat shape. When it was placed beneath the tube and subjected to hot air, it was securely wrap around the tube (Fig. S1a and Movie S1). To test its in vivo capability, it was implanted into a 6-month-old mouse (Fig. S1b). It shows that the device can securely lock around the carotid artery. These results prove the device's implantability and stability within a live model, representing a significant step towards potential clinical applications.

3.2. Printability assessment and material characterization

To print the 3D structures, we customized a photoactive ink consisting of pre-PGDA and AA monomer. The pre-PGDA was synthesized based on our previously reported recipe[19]. The structures of pre-PGDA and AA are illustrated in Fig. 2a. In the formulation of the photoresin, we used a photoinitiator Diphenyl(2,4,6-Trimethylbenzoyl) Phosphine Oxide (TPO) due to its effective polymerization initiation and comparatively low toxicity, as supported by its use in biomedical applications [40] Aware of the potential toxicity of photoinitiator derivatives, we carefully controlled the concentration to 3 wt% and employed thorough post-processing washing steps to remove any uninitiated photoinitiator. These steps are critical for minimizing the potential long-term toxicity and ensuring the biocompatibility of the printed structures, as evidenced by our in vitro and in vivo biocompatibility results. Upon UV irradiation on the resin, TPO is activated to generate free radicals that permeate through AA and pre-PGDA. These activated precursors then undergo copolymerization to create a crosslinked PGDA-PAA network (Fig. 2b). Fig. 2c shows a detailed structure of a unit of the resulting network. Compatibility of the pre-PGDA/AA resin with DLP was determined by rheological and UV sensitivity characterizations. Notably, the weight ratio of pre-PGDA significantly influences the rheological behavior of the resin (Fig. 2d). It shows that all resins with varied pre-PGDA weight ratios behave as a Newtonian fluid at T_R, independent of the testing frequency. Decrease of the pre-PGDA weight ratio from 70 to 55 wt% reduces the viscosity from 90.25 mPa·s to 23.95 mPa·s at a shear rate of 19 Hz. Nevertheless, they are still lower than the threshold value of 1.3 Pa·s for DLP[41]. Adding AA into the precursor resin reduces the viscosity of pre-PGDA, thus improving printability. Meanwhile, presence of the resulting PAA in the PGDA-PAA network improves the elasticity and toughness, which will be discussed later. The UV-initiated crosslinking of the pre-PGDA/AA resin was monitored by ATR-FTIR (Fig. 2e, Fig. S2, Fig. S3a), from which the conversion rates of the C═C bonds were calculated (Fig. 2f, Fig. S3b). Fig. 2e and Fig. S2 show that intensity of the acrylate C═C peaks in AA and pre-PGDA at 1635 cm⁻¹ and 1618 cm⁻¹, respectively, was rapidly reduced within 1s, indicating fast polymerization of the pre-PGDA/AA resins with 50 wt% − 80 wt% pre-PGDA. Fig. 2f shows increase in the corresponding C═C conversion rate as the AA weight ratio increases because of the higher photo-reactivity of AA than that of pre-PGDA. We also investigated the evolvement of the C═C peaks as function of the UV exposure durations (Fig. S3a). As the exposure time increases, the C═C peak intensity in the pre-PGDA/AA resin with 70 wt% pre-PGDA is reduced. As shown in Fig. S3b, the resin was rapidly polymerized within the first second, with ~80% consumption of the C═C bonds.

Fig. 2 ∣. PGDA-PAA precursor details, printability assessment, and material characterization.

(a) Chemical structures of pre-PGDA, AA, and TPO that are used to print PGDA-PAA. (b) Scheme showing photo-polymerization of PGDA-PAA. (c) Scheme of the crosslinked PGDA-PAA chemical structure. (d) Viscosity–shear rate curves of pre-PGDA/AA resin with different pre-PGDA weight ratios. (e) ATR-FTIR spectra of pre-PGDA/AA with 70 wt% pre-PGDA and crosslinked PGDA-PAA after 1 s UV exposure. (f) Conversion of C═C bonds evaluated from ATR-FTIR curves. (g, h) Stress-strain curves of PGDA-PAA with various PGDA weight ratios below and above T_trans. (i) Cyclic tensile testing of PGDA-PAA with 70 wt% PGDA above T_trans. (j) Toughness, modulus plots of PGDA-PAA with different PGDA weight ratios below and above T_trans. (k) Toughness-modulus plot of some reported implantable materials, orthopedic hard and soft tissue, and cardiovascular tissue in comparison to PGDA-PAA above and below T_trans. Note: data about other materials is adopted from reference[42].

T_trans of PGDA-PAA was investigated using Differential Scanning Calorimetry (DSC). Fig. S4 shows DSC curves of the PGDA-PAA samples with different PGDA weight ratios of 50 wt%, 60 wt%, and 70 wt%. All three curves exhibit endothermic peaks. The increase in the weight percentage of PGDA from 50 wt% to 70 wt% results in a decrease in T_trans from 47.2 to 39.2 °C. This can be attributed to the decreased crosslinking density of PGDA-AA caused by a lowered reactivity of the resin, which results from the reduced AA content. Simultaneously, the increased PGDA content may lead to a higher degree of chain mobility, which further contributes to the reduction of T_trans. To investigate the effect of PGDA weight ratios on the mechanical behavior of PGDA-PAA, we performed uniaxial tensile testing below and above the T_trans. As shown in Fig. 2g, the tensile strength (σ_T) at T_R (below T_trans) decreases from 15 MPa to 7 MPa as the PGDA weight percentage increases from 55% to 70%. That can be because σ_T (~5 MPa) of pure PGDA is much lower than that (~150 MPa) of pure PAA at T_R. ε_f slightly decreases from 200% to 170% when the PGDA weight ratios increases from 55% to 70%. Above T_trans, similarly, both σ_T and ε_f of PGDA-PAA are reduced as the increase of the PGDA weight ratios (Fig. 2h). For instance, σ_T of PGDA-PAA with 55 wt% PGDA is reduced to 2.57 MPa from 15.41 MPa. But ε_f is well maintained 180% and 130%, slightly decreased from 200% and 170% for PGDA-PAA with PGDA weight ratios of 55 wt% and 70 wt%, respectively. This result is quite striking because normally the phase transition in SMPs significantly lowers the ductility of the material, which was also observed in pure PGDA[19]. We hypothesize that introduction of the linear PAA chains can augment the stretchability of the crosslinked network (Fig. 2c). Because they can be coiled and entrapped within the PGDA network. The hydrogen bonds introduced by PAA further improves the stretchability. Breaking the hydrogen bonds together with stretching the coiled linear PAA chains require much energy, thus allowing the material to withstand large deformations.

When the sample is unloaded at room temperature, some of the deformation can recover due to entropic change. However, the restoration of broken hydrogen bonds takes an extended period, leading to an observable residual strain. The residual strain can be eliminated through thermal treatment, which expedites the recovery of hydrogen bonds and enhances the mobility of the linear PAA chains. These assertions are corroborated by cyclic tensile tests above T_trans using PGDA-PAA samples with 70 wt% PGDA (Fig. 2i). After five continuous cycles to a strain of 100%, the sample exhibits a residual strain of 14.7%, indicating a well-maintained elasticity. This high elasticity is vital as it facilitates the shape reprogramming. A hysteresis loop is seen in the first cycle, indicating energy dissipation. The hysteresis in the second cycle is slightly lower than that of the first one. It is gradually becoming smaller as the number of cycles increases.

Correspondingly, at both below and above T_trans, the Young's modulus (E) and toughness (U_T) of the PGDA-PAA with different PGDA weight ratios were calculated from the tensile testing data. Below T_trans, E and U_T of PGDA-PAA are in the range of 40-170 MPa and 10-20 MJ/m³, respectively (Fig. 2j). Increase of the PGDA weight ratios reduces both E and U_T. Above T_trans, E is significantly reduced. For instance, E of PGDA-PAA with 70 wt% PGDA decreases from 41.76 MPa to 1.02 MPa. U_T of PGDA-PAA with 70 wt% PGDA is also reduced from 10.18 MJ/m³ to 0.43 MJ/m³. Nevertheless, the E values above T_trans match well with those of the articular cartilage and cardiovascular tissue, while the toughness is 10-100 times greater than these tissues. This considerable self-driven change in the mechanical properties enables the printed 3D structures to emulate characteristics of the natural tissue while maintaining robust properties under cyclic deformations, making them apt for cardiovascular applications.

Fig. 2k includes U_T versus E for some representative synthetic materials in clinical use[42]. The materials are distinctly categorized into three major groups: high-E and high-U_T metals and their alloys[43] such as 316L stainless steel (SS), NiTi, Ti-4Al-6V, commercially pure titanium (cpTi), and Co-Cr, which exhibit; high-E but low-U_T ceramics[44] like zirconia, alumina, and Pyrolytic carbon; and polymers[45] including silicone, Segmented Polyurethane (SPU), Ultra-High Molecular Weight Polyethylene (UHMWPE), Polytetrafluoroethylene (PTFE), Polyamide 6-6 (nylon 6-6), Polyethylene Terephthalate (PET), Polyether Ether Ketone (PEEK), Polylactic Acid (PLA), and Poly(Methyl Methacrylate) (PMMA) which demonstrate a wide range of E and U_T values. This categorization underpins the rationale behind the material choice for specific biomedical applications, such as the preference for metals in joint replacements due to their high-E and high-U_T. An interesting trend observed in Fig. 2k is that clinically successful materials generally possess high toughness. In fact, U_T (10-100 MJ/m³) of most synthetic materials surpasses those (0.01-10 MJ/m³) of most living tissue. Notable exceptions include PMMA, PLLA, and ceramics, each with unique attributes that justify their use in specific circumstances. For instance, PMMA is often used as bone cement[46], while PLLA, despite its lower toughness, is valued for its biocompatibility. Additionally, PCL is an FDA-approved high-toughness polymer that shows no shape memory behavior at body temperature[47]. Nevertheless, there remains a need for biomaterials with high U_T, suitable E and additional functionalities such as shape programming response for MIS involving soft tissue. Bridging this gap is a challenge due to the intrinsic tradeoffs between E and U_T. Our PGDA-PAA SME holds a great potential to address this challenge.

3.3. Thermomechanical performance of PGDA-PAA

We also investigated the thermomechanical properties of PGDA-PAA. Fig. 3a depicts their shape-fixing and recovery mechanisms. To program the sample to a temporary shape by strain, it first is heated above T_trans. PGDA-PAA, cross-linked as described above, exhibits rubber-like elasticity above T_trans and can be easily deformed with an external force to form a new secondary shape. Then this programmed shape is cooled below T_trans. When the sample is cooled, the crystalline domains, dominated by PGDA, percolate to fix the strained shape. This temporary shape (though possibly subjected to prolonged warpage) is maintained below T_trans. When heated above T_trans again, the PGDA crystalline domains melt to form an amorphous, homogeneous phase with high mobility, allowing the fixed shape to recover to the original shape. The role of PGDA in PGDA-PAA is to impart the shape memory effect into the 3D printed structures to realize 4D printing. The PAA chains improve the ductility of PGDA-PAA while maintaining the high tensile strength at both below and above T_trans. As presented in Fig. 3b (Movie S2), a programmed 3D structure recovers to its as-printed shape when heated above T_trans within 25 second.

Fig. 3 ∣. Thermomechanical properties of PGDA-PAA SME.

(a) Shape programming and recovery mechanism. (b) Photographs showing the whole shape memory cycle of a printed object. Scale bar: 2 mm. (c) Storage modulus (E ′) and TanD values of PGDA-PAA with 70 wt% PGDA as a function of temperature. (d) Shape-fixing ratios (R_f) and shape-recovery ratios (R_r of PGDA- PAA SME with different PGDA weight ratios.

Fig. 3c presents DMA results for the PGDA-PAA SMEs with varying weight ratios of PGDA. It displays the temperature-dependent storage modulus, which corresponds to a sample's elastic response[48]. It shows a clear drop in the storage modulus when the temperature is above T_trans. Additionally, Fig. 3c plots the temperature-dependent TanD, which is defined as the ratio of loss modulus (corresponding to energy dissipation) to the storage modulus. The curve shows transition temperature at its peak for a polymer, such as the glass transition temperature for an amorphous one and the melting/crystallization temperature for a semi-crystalline one[49]. The peak in the curve for the PGDA-PAA SME with 70 wt% PGDA appears at transition temperature of 39.38 °C, which closely agrees with the value obtained from the DSC test (Fig. S4).

We further investigated the effect of PGDA weight ratios on the shape memory (SM) behaviors of the PGDA-PAA samples. To quantify the SM behaviors, we calculated the shape fixity ratio (R_f) as R_f = ε_u/ε_m × 100% and the shape recovery ratio (R_r) as R_r = (ε_u − ε_r)/ε_u × 100%, where ε_m is the maximum strain before unloading, ε_u is the strain immediately after unloading, and ε_r is the instantaneous strain after recovery. In the evaluation, they were first stretched by a 100% strain at 45 °C. After being cooled to T_R, their shapes were fixed. Reheating the samples above 45 °C, the shapes are supposed to recover back to the original ones. As shown in Fig. 3d, with the decrease of the PGDA weight ratio from 70 wt% to 55 wt%, R_f is significantly decreased from 92% to 43%. It can be expected because R_f of a semi-crystalline SMP is highly related to the crystalline regions. An increase in the PGDA ratio results in a higher amount of crystalline regions in the SME system, leading to improved R_f. In contrast, all the PGDA-PAA samples with different PGDA weight ratios exhibited R_f of > 95%, and the increased PGDA ratio only slightly decreases R_f (Fig. 3d). After balancing the mechanical and the thermomechanical properties shown in Fig. 2 and Fig. 3, we decided to use the PGDA-PAA SME with 70 wt% PGDA as the material for the following in vitro and in vivo studies.

3.4. Cytocompatibility of PGDA-PAA

Upon implantation, PGDA-PAA is expected to provide temporary support for cell growth until it is replaced by regenerated tissue. Therefore, biodegradability and biocompatibility are crucial for this purpose. As shown in Fig. S5, in six-month degradation experiments, the printed PGDA-PAA with 70 wt% PGDA exhibited 45% weight loss and 15% weight loss in 0.1 mM NaOH solution and PBS, respectively. As the PGDA weight ratio decreases, the degradation rate decreases. It can be expected because PGDA is a polyester, which is more prone to be degraded than PAA is. SEM images of the PGDA-PAA samples show severe surfaces erosion after 6-month degradation in both 0.1 mM NaOH and PBS solutions (Fig. S6). The cytocompatibility of PGDA-PAA was assessed using 10T1/2 cells co-cultured in vitro. Stereo microscope images taken over the four-day culturing period show significant cell proliferation (Fig. 4a). Fig. 4b indicates that the number of the grown cells co-cultured with PGDA-PAA is close to the control with a 2.5% difference, suggesting biocompatibility of PGDA-PAA. An unpaired t-test shows statistical similarity between the cell count of co-cultures with printed materials and the control with a P value equal to 0.8761. Propidium Iodide (PI) and DAPI staining showed no significant difference in the cell survival among the groups (Fig. 4c, Fig. S7). The cell survival rates were quantified to be > 90% after four days of cell culture (Fig. 4d), indicating that PGDA-PAA is cytocompatible.

Fig. 4 ∣. Biocompatibility test of PGDA-PAA.

(a) Stereo microscopy images of 10T1/2 control cells (Control) and cells co-cultured with PGDA-PAA for 1, 2, 3, and 4 days. Scale bar: 100 μm. (b) Cell counts at the end of the 4-day cell culture. (c) Fluorescent microscopy images of 10T1/2 cells (Control) or cells co-cultured with PGDA-PAA for 1, 2, 3 and 4 days, respectively. Propidium Iodide (PI, red) was used to detect dead cells (white arrows), nuclei were stained with DAPI (blue). Scale bar: 10 μm. (d) Quantification of the cell survival rate (%) for control or cells co-cultured with PGDA-PAA. (e-h) SEM images of 10T1/2 cells cultured on PDGA-PAA at different magnifications. Scale bars: 500 μm (e); 100 μm (f); 30 μm (g); 20 μm (h).

We also investigated how the PGDA-PAA surface topography affects cell morphology and the way cells adhere, grow, and differentiate. After an incubation period of 7 days, the used 10T1/2 cells appeared to well adhere to the PDGA-PAA surface (Fig. 4e-h). In addition, they spread widely and formed lamellipodia at the leading edge. The cells also developed short filopodia at their apical poles, spreading from the lamellipodia (Fig. 4g), which is consistent with the properties as fibroblast-like cells[50, 51]. The lamellipodia is a dense network of cross-linked actin filaments that drive cellular distribution and motility. The filopodia are exploratory extensions formed via parallel bundles of actin filaments from the plasma membrane[50]. The thin actin protrusions can probe the extracellular environment to guide cell migration towards specific sites of interest. As filopodia is involved in cellular processes of wound healing, extracellular matrix adhesion, chemoattractant guidance, and neuronal growth-cone pathfinding[51], formation of the filopodia, which senses cell surrounding and acts as sites for signal transduction[50], is important for directing cell migration on the PDGA-PAA surface. These results prove that the surface topography directly affects the extension and adhesion of the filopodia and lamellipodia.

3.5. In vivo Study of a printed PGDA-PAA vascular graft

To test if a printed PGDA-PAA tube can serve as a biocompatible scaffold for vascular grafting, we implanted the tube onto mouse aorta through an end-to-end anastomosis (Fig. 5a). Detailed in Fig. S8, an implanted tube featuring a height of 15 mm, an outer diameter of 2 mm, and a wall thickness of 150 μm was designed to facilitate the grafting procedure. After 21 days, tissues began to adhere to both the inner and outer surfaces of the implanted tube. Gross images and H&E staining of the tubes revealed thrombosis formation in the lumens (Fig. 5b, c) of a few tubes, and material degradation over time (Fig. 5d). Occurrence of the thrombosis and postimplantation material degradation are important considerations when evaluating the value of a vascular graft. Thrombosis is a significant clinical concern in vascular grafts, as it can obstruct blood flow and compromise graft function. It is noteworthy that thrombosis is a common complication following surgical procedures, including aorta transplantation. This is true even for FDA-approved artificial aortas used in aneurysm repair surgeries [52, 53]. In this study, we observed the formation of thrombi after 21 days post-implantation, as indicated in the H&E staining image (Fig. 5c). Thrombi appeared in the lumen of the graft, potentially as a response to the foreign material or due to alterations in blood flow caused by the graft. The observed thrombosis rate of ~ 25% is comparable to, if not lower than the rates reported for other materials used in similar contexts, suggesting that our 4D printed vascular grafts did not excessively promote the thrombus formation [54, 55]. Nevertheless, Continuous monitoring of thrombosis rates is essential for understanding the long-term performance and safety of the vascular graft. Material degradation is another critical factor when assessing biocompatible scaffolds for tissue engineering. Degradation allows for the scaffold to be gradually replaced by natural tissues, leading to a more natural, functional graft. H&E staining revealed signs of material degradation (Fig. 5d), suggesting that the PGDA-PAA tube is not only biocompatible but also biodegradable. This is advantageous as it means that the graft can potentially be replaced by the patient's own tissue over time, improving an overall integrity and function of the repaired vessel. Both thrombosis and material degradation highlight the dynamic interaction between the implanted graft and the surrounding biological environment. Understanding these processes in depth will be crucial for further development and improvement of the PGDA-PAA vascular grafts.

Fig. 5 ∣. In vivo study of a PGDA-PAA tube for vascular grafting.

(a) Photograph of a printed tube after implantation into a mouse aorta. (b) Photographs of the printed tube implanted into a mouse aorta for 21 days. Scale bar: 2 mm. (c) Thrombosis formation rates following vascular graft surgery, with a representative image showing the thrombosis in the lumen (arrow) as depicted by H&E staining. Scale bar: 30 μm. (d) Material degradation derived from the H&E staining images (arrows). Scale bar: 100 μm. (e) Formation of new tissue on tube after 21-day implantation as shown by H&E staining images. Black arrows point to the migration of cells into outer layers, while green arrows indicate the adventitia growth. Scale bar: 100 μm. (f) Immunostaining of endothelial cell marker CD31 (green) and fibroblast marker FSP-1 (red) with DAPI staining of cell nuclei (blue). Scale bar: 100 μm.

On the outer surface of the tube, cell migration was evident (indicated by black arrows in Fig. 5e), as was growth of the adventitia (green arrows in Fig. 5e). Immunostaining revealed CD31-positive endothelial cells (Fig. 5f, green) lining the inner surface of the tube, indicative of a biocompatible environment conducive to endothelial cell growth and endothelium formation. This is crucial for maintaining vessel integrity, preventing coagulation, and controlling blood flow. The outer surface of the tube was populated with FSP-1-positive fibroblasts marked as red in Fig 5f, suggesting the development of a tentative adventitial layer. The combination of CD31-expressing endothelial cells and FSP-1-positive fibroblasts signifies the beginning stages of medial and adventitial layer formation around the graft. Given the complexity and functionality of native arteries and veins, these results highlight the promising potential of the synthetic PGDA-PAA materials for arterial grafting applications. In this study, we employed a mouse model for the biocompatibility testing of the bioresorbable shape memory elastomers. While this model offers advantages such as ease of handling, cost-effectiveness, and established experimental protocols, there are some limitations. There exist significant anatomical and physiological differences between mice and humans, including smaller vessel sizes, distinct blood pressure levels, and different arterial wall properties, which may potentially affect the outcomes of the implantation procedure and the relevance to human conditions. Furthermore, the progression of diseases and healing processes in mice may not accurately mimic the human pathophysiology. Transitioning from a mouse model to human involves more than just scaling up device sizes, but requires careful consideration of biomechanical properties, blood flow dynamics, and tissue response. Therefore, a large animal model such as a pig model is required to make our study more clinically relevant, which will be pursued in the future.

3.6. Design and printing of a personalized PGDA-PAA LAA occluder and in vitro implantation

A CT scan from a 23-year-old female patient, with a medical history inclusive of hypertension, hyperlipidemia, diabetes, and smoking, was utilized for the dual purpose of designing the LAA occluder (LAAO) and 3D printing a model of the heart model. As illustrated in Fig. 6a, this personalized LAAO design was then printed with PGDA-PAA. Fig. 6a also presents the patient's heart, including the LAA, which was included in the heart's 3D model. The model was printed by a transparent elastomer that emulates the mechanical properties of heart tissue, thereby enhancing the realism of in vitro trans-catheterization feasibility studies. The in vitro test of transcatheter LAA closure procedure was executed using a silicone tube to mimic the role of a catheter, delivering the LAAO to the LAA of the printed heart immersed in saline maintained at physiological temperature. Before delivery, the LAAO was programmed to a compact, temporary shape above T_trans and then was fixed under T_trans. As demonstrated in Fig. 6b (Movie S3), the LAAO can be smoothly extruded from the catheter, underscoring the potential for minimally invasive implantation. Upon being released near the LAA orifice, the shape of the LAAO was recovered to the originally printed shape. For a more tangible representation of LAA occlusion effectiveness, red coloring was introduced into the saline as an analog for colored saline that might diffuse into the LAA post-LAAO deployment. It shows that the LAAO can effectively enclose the LAA, indicating a potential clinic application in future. However, it is imperative to clarify that the demonstration is just an early-stage proof-of-concept. The development of a ready-to-use implant entails a series of intricate and rigorous studies that have yet to be performed.

Fig. 6 ∣. In vitro study of a printed PGDA-PAA occluder for LAA occlusion.

a) Design and printing of a personalized LAAO according to a CT scanned patient’s heart model. b) In vitro implantation procedure of the printed LAAO by a transcatheter.

4. Conclusion

From a manufacturing perspective, the developed pre-PGDA/AA resin possesses appropriate rheological properties for printing a variety of complex 3D structures with sub-200 μm features and printing speeds up to 100 mm/h. Although printing tubular structures is challenging due to over-curing of residual resin inside the channel in the vat photopolymerization (VPP)-based 3D printing techniques, we successfully printed tubes with a wall thickness of 200 μm and an inner diameter of 750 μm. The printed 3D strctures exhibit shape memory properties with a shape fixity ratio of 98% and a shape recovery ratio of ~100%.

From an application perspective, addressing the gap between the growing demand for personalized medical devices and the availability of advanced technologies is crucial for improving human health and quality of life. The promising 4D printing techniques can produce such devices whose shapes and functions can dynamically respond to stimulation for MIS. However, traditional SMPs often result in printed structures with non-biocompatibility, low resolution, unsuitable mechanical properties, e.g., low toughness, and undesired T_trans. In this study, we demonstrated 4D printing of the PGDA-PAA SME with suitable T_trans for easy shape programming and deployment. Co-polymerizing PAA within the PGDA network dramatically increases tensile strength and toughness at both below and above T_trans. The printed structures exhibit sufficient stiffness to maintain programmed shapes at T_R, while the reduced E and well-maintained elasticity above T_trans make shape reprogramming and implantation relatively easy. Moreover, the self-driven change in mechanical properties above T_trans allows the printed structures to mimic natural tissue while maintaining robust performance under cyclic deformations, making them suitable for implantation. Furthermore, the material is biocompatible and degradable as demonstrated by in vitro cell studies and in vivo animal studies. Finally, LAAOs that well adapt to a patient’s heart geometry were designed and printed for in vitro test. Successful trans-catheterization procedure demonstrated the geometrical and mechanical adaptability of the printed constructs. In summary, this PGDA-PAA SME-based 4D printing technology would offer a new possible path to personalized biomedical implantation.

Statement of Significance.

4D printed shape-memory elastomer (SME) implants particularly designed and manufactured for a patient are greatly sought-after in minimally invasive surgery (MIS). Traditional shape-memory polymers used in these implants often suffer from issues like unsuitable transition temperatures, poor biocompatibility, limited 3D design complexity, and low toughness, making them unsuitable for MIS. Our new SME, with an adjustable transition temperature and enhanced toughness, is both biocompatible and naturally degradable, particularly in cardiovascular contexts. This allows implants, like biomedical scaffolds, to be programmed at room temperature and then adapt to the body's physiological conditions post-implantation. Our studies, including in vivo vascular grafts and in vitro device implantation, highlight the SME's effectiveness in aortic tissue regeneration and its promising applications in MIS.