Search
Close this search box.

A 3D printable tissue adhesive – Nature Communications

Protocol approval

All animal studies were approved (protocol 1119-074-22) under the guidelines of the MIT Committee on Animal Care (CAC) and all surgical procedures and postoperative care were supervised by the MIT Division of Comparative Medicine veterinary staff.

Materials

All chemicals were obtained from Sigma Aldrich unless otherwise stated and used without further purification. For synthesis of the 3D printable tissue adhesive, acrylic acid (AA), hydrophilic polyurethane (PU, HydroMed D3, AdvanSource Biomaterials), benzophenone, α-ketoglutaric acid, N-Hydroxysuccinimide (NHS), and 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) were used. All porcine tissues used for ex vivo experiments were purchased from a research-grade tissue vendor (Sierra Medical Inc.).

Grafting of poly(acrylic acid) to polyurethane

A precursor solution was prepared by combining 32 w/v % AA, 8 w/v % PU, 20 w/v % vacuum-degassed deionized water, and 40 w/v % ethanol and stirring until the PU was fully dissolved. 1.1 w/v % benzophenone and 0.1 w/v % α-ketoglutaric acid were added to the precursor solution and homogeneously mixed, then transferred to a sealed glass vial and cured in a UV crosslinker (364 nm, 15 W power) for 120 min. Benzophenone functions as a Type II free radical photoinitiator that enters an excited triplet state under UV irradiation, generating radical sites in PU (for example, by abstracting hydrogen from carbon-hydrogen containing molecules along the polyether backbone). These radical sites may then react with acrylic acid, initiating the growth of PU-grafted PAA chains (Supplementary Fig. 18). Benzophenone ketyl radicals can eventually combine with each other to form benzopinacol, which is removed during the dialysis process. For dialysis, the cured product was transferred to a cellulose membrane (Sigma Aldrich, typical molecular weight cut-off = 14,000) and purified in a pure ethanol bath for 24 h (replacing the ethanol every 12 h) followed by in a deionized water bath for 24 h (replacing the water every 12 h) with continuous magnetic stirring. The purified PU-PAA was cut into small pieces and dried in a desiccating oven at 70 °C for 48 h.

Preparation of the 3D printable tissue adhesive ink

The dried PU-PAA was redissolved at a concentration of 20 w/w % in 70% ethanol and mixed in a 25:2 (v/v) ratio with a solution comprising 33.3 w/v % EDC and 33.3 w/v % NHS in 70% ethanol to yield around 10% (mol/mol) NHS functionalization of the carboxyl groups. The combined solution was then mixed in a 10:3 (v/v) ratio with 20 w/w % PU in 95% ethanol. To mitigate hydrolysis of NHS, the ink was prepared directly before use.

3D printing procedure

3D printing of the tissue adhesive ink and other polymer solutions was performed using a custom-designed 3D printer with a Cartesian gantry system (Aerotech). Under air pressure, inks were extruded from 5 mL syringe barrels through nozzles ranging in size from 50 to 200 µm (EFD Nordson). Printing paths were designed using Adobe Illustrator and CADFusion (Aerotech), then translated into G-code using a custom Python script. To achieve continuous printing of the tissue adhesive ink, we selected a printing pressure of 250 kPa (Ultimus V, Nordson EFD) and printing speeds ranging from 500−1800 mm/min. The structures were printed onto a glass slide (Corning) treated with hydrophobic coating (Rain-X). After printing, the structures were completely dried and sealed in plastic bags with desiccant (silica gel packets) before use.

FTIR

32 scans were recorded for droplets of polymer on a Bruker Alpha II FT-IR spectrometer with a monolithic diamond crystal at a resolution of 4 cm-1. An equal number of background scans were recorded on air prior to each sample measurement. For analysis, each spectrum was normalized based on the peak around 2900 cm-1. As shown in Supplementary Fig. 4, the stretching bands of the C = O bond and the O-H bond in carboxylic acid can be clearly observed at around 1710 cm-1 and 3300 cm-1 for the samples containing PAA, indicating the retention of PAA via grafting and entanglement. The stretching and bending vibration bands of the C-NH bond on O = C-NH-C at around 3350 cm-1 and 1530 cm-1 and the stretching vibration band of the N-H bond at around 1700 cm-1 can also be observed for the PU-containing spectra, indicating the presence of a urethane group42. The strong stretching vibration band associated with the ether bond at 1100 cm-1 reflects the polyether character of the ether-based PU.

1H NMR

Proton (1H) nuclear magnetic resonance (NMR) spectra were measured at 400 MHz on a Bruker Avance III DPX 400. Approximately 100 mg of each sample were dissolved in 500 μL deuterated dimethylsulfoxide (DMSO-d6) for analysis. To evaluate the ratios of PAA:PU and NHS:PAA, each spectrum was normalized based on the peak around 2.22 ppm attributed to the methylene peak in PAA. The peaks at 7.06 ppm, attributed to the nitrogen-attached hydrogen in PU, were integrated with respect to the normalized peaks to approximate the molar ratios of PAA:PU in the final purified products (Supplementary Fig. 5). The NHS alkyl peaks at 2.76 ppm were integrated with respect to the normalized peaks to estimate the degree of NHS functionalization (Fig. 2d).

Adhesion characterization

Adhesion tests were performed on in the inner surface of porcine skin washed with phosphate buffer solution (PBS). 3D printed tissue adhesive samples were adhered by applying gentle pressure upon the tissue substrate for 10 s. Commercial sealants were applied according to manufacturer instructions. Unless otherwise indicated, adhesion characterizations were performed 20-30 minutes after initial application to allow for swelling of the tissue adhesive material. Adhered samples were covered with gauze soaked in PBS to maintain a wet environment prior to measurement.

To measure interfacial toughness, tissue samples with widths of 2 cm were adhered to the various adhesives and tested via the standard T-peel test (ASTM F2256) using a mechanical testing machine (2.5 kN load cell, Zwick/Roell Z2.5). Data was collected using testXpert Testing Software. All tests were conducted with a constant peeling speed of 50 mm min-1. The measured force reached a plateau as the peeling process entered steady state. Interfacial toughness was determined by dividing two times the plateau force by the width of the tissue sample (Supplementary Fig. 8a). Hydrophilic nylon filters (1 µm pore size, TISCH Scientific) were used as a stiff backing for the 3D printed tissue adhesive.

To measure shear strength, tissue samples with an adhesion area of 2 cm x 2 cm were joined using the various adhesives and tested via the standard lap shear test (ASTM F2255) using a mechanical testing machine (2.5 kN load cell, Zwick/Roell Z2.5). All tests were conducted with a constant peeling speed of 50 mm min-1. Shear strength was determined by dividing the maximum force by the adhesion area (Supplementary Fig. 8b).

To measure burst pressure, 3 mm holes were introduced in 2.5 cm x 2.5 cm pieces of porcine skin using biopsy punches based on the ASTM F2392-04 standard defect size for burst pressure measurement. The holes were then sealed using 1.5 x 1.5 cm samples of the 3D printed tissue adhesive, or an equivalent area of Coseal, TachoSil, or Tegaderm. The size was determined based on previously reported measurements of burst pressure strength of surgical sealants32. The samples were fixed in a testing rig and PBS was injected at a constant rate of 5 mL min-1 to the point of failure (i.e., fluid leakage). Pressure was recorded by a pressure transducer (PX409, Omega) using OMEGA PC software. The burst pressure was determined as the maximum pressure upon which a leakage formed (modified ASTM F2382-04; Supplementary Fig. 8c).

Rheological characterization

Rheological measurements of the tissue adhesive inks were performed using a rotational rheometer (AR-G2, TA instrument) with 40-mm diameter 2° steel cone geometry at 25 °C. Data was collected using TA Rheology Advantage. Apparent viscosity was measured as a function of shear rate using a continuous ramp over a logarithmic sweep from shear rate 1 to 1000 s-1. Shear storage modulus (G’) was measured as a function of shear stress using an oscillatory procedure with 1 Hz frequency over a logarithmic sweep from shear stress 1 to 10000 Pa. For all measurements, an aqueous solvent trap was used to minimize ink drying.

Mechanical characterization

The tensile properties of mesh samples were measured following the standard tensile test (ASTM D412) using a mechanical testing machine (2.5 kN load cell, Zwick/Roell Z2.5). Samples were soaked for 20-30 min in PBS at 37 °C before measurement. Effective Young’s moduli were determined as the initial slope on the stress-strain curve.

Microscope imaging

Microscopic 3D printed structures were imaged using an epifluorescence microscope (Nikon Eclipse LV100ND). Confocal microscope images of the mesh structure were obtained by an upright confocal microscope (SP 8, Leica) with 360 nm excitation wavelength for blue fluorescent beads. ImageJ (version 2.1.0) was used for image processing and analysis.

Histology of porcine tissues

Porcine tissues samples (skin, tendon, aorta, and rectal sheath) were procured from Sierra Medical Inc. and fixed in 10% formalin for 24 h for histological analyses. Fixed tissue samples were placed into 70% ethanol and submitted for histological processing and H&E staining at the Hope Babette Tang (1983) Histology Facility in the Koch Institute for Integrative Cancer Research at the Massachusetts Institute of Technology. One sample for each tissue type was processed for histology.

Fabrication of the collagen fiber-inspired patches

A silicone elastomer ink was prepared by mixing Dragon Skin 30 (Smooth-On) and SE 1700 (Dow Corning) together. Specifically, Dragon Skin 30 part A, Dragon Skin 30 part B, SE 1700 base, and SE 1700 catalyst were added in a 10:10:10:1 weight ratio and mixed thoroughly using a Thinky mixer (AR-100, Thinky). The ink was printed onto a glass slide (Corning) treated with a hydrophobic coating (RainX) into the desired geometry and cured in the oven at 120 °C for 30 min. After curing and cooling, a layer of the tissue adhesive ink (prepared as described above) was printed on top of the silicone layer, following the same geometry. After drying, the patch was removed from the glass slide and evaluated using a standard tensile test.

Fabrication of the liquid-infused blood resistant patch

Tissue adhesive ink was prepared as described above and printed onto a glass slide (Corning) treated with hydrophobic coating (RainX) into a 25 cm by 25 cm repeating lattice pattern with filament width ~150 µm and gap width ~150 µm. After drying, the adhesive structure was integrated with a collagen wound dressing sponge (Puracol) of the same dimensions by exposing the uppermost surface of the printed structure to steam, allowing the surface to become slightly hydrated, then immediately placing the collagen sponge on the hydrated surface to allow crosslinking between the two substrates. The integrated patch was removed from the glass slide and infiltrated with oleic acid. To demonstrate blood-resistant tissue adhesion, samples of the oil-infused patch were applied to porcine skin samples covered with heparinized porcine blood (Lampire Biological Laboratories, Inc.) using gentle pressure for 10–30 s.

In vitro biocompatibility

To evaluate the in vitro biocompatibility and cytotoxicity of the mesh, a LIVE/DEAD assay was used to assess BALB/c 3T3 clone A31 mouse fibroblasts (American Type Culture Collection®; CCL163™). To prepare conditioned media, 500 mg of Coseal, TachoSil, and the 3D printed tissue adhesive were each incubated in 10 ml of Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10 v/v % fetal bovine serum and penicillin-streptomycin (100 U mL−1) at 37 °C for 24 h. The supplemented DMEM (without any material incubation) was used as a control. 3T3 cells were plated in confocal dishes (20-mm diameter) at a density of 0.5 × 105 cells cm−2 (N = 4 for each group). The cells were then treated with either conditioned or control media and incubated at 37 °C for 24 h in a 5% CO2 atmosphere. Cell viability was determined by a LIVE/DEAD viability/cytotoxicity kit for mammalian cells (Thermo Fisher Scientific). A confocal microscope (SP 8, Leica) was used to image live cells with excitation/emission at 495 nm/515 nm and dead cells at 495 nm/635 nm, respectively. Cell viability was calculated by counting live (green fluorescence) and dead (red fluorescence) cells using ImageJ (version 2.1.0).

In vivo biocompatibility

Female Sprague-Dawley rats (225-250 g, Charles River Laboratories) were used for all in vivo studies. Before implantation, the 3D printed patch was exposed for 1 h under UV light. Commercially available tissue adhesives were used as provided in sterile packages following the provided user guide or manual for each product.

For implantation in the dorsal subcutaneous space, rats were anesthetized using isoflurane (2-3% isoflurane in oxygen) in an anesthetizing chamber. Anesthesia was maintained using a nose cone. The back hair was removed, and the animals were placed over a heating pad for the duration of the surgery. The subcutaneous space was accessed by a 1–2 cm skin incision per implant in the center of the animal’s back. To create space for implant placement, blunt dissection was performed from the incision towards the animal shoulder blades. Samples of the 3D printed patch and TachoSil with the size of 20 mm in width and 20 mm in length were placed in the subcutaneous pocket created above the incision without detachment. The incision was closed using interrupted sutures (4-0 Vicryl, Ethicon) and 3–6 mL of saline was injected subcutaneously. Up to four implants were placed per animal ensuring no overlap between each subcutaneous pocket created. 2 or 4 weeks after the implantation, the animals were euthanized by CO2 inhalation. Subcutaneous regions of interest were excised and fixed in 10% formalin for 24 h for histological analyses. Fixed tissue samples were placed into 70% ethanol and submitted for histological processing and H&E or Masson’s trichrome (MT) staining at the Hope Babette Tang (1983) Histology Facility in the Koch Institute for Integrative Cancer Research at the Massachusetts Institute of Technology. Representative histology images of each group were shown in the corresponding figures. The subcutaneous implantation model was performed on 3 independent samples with similar results.

In vivo rat trachea defect repair

For the in vivo trachea defect repair model, rats were anesthetized using isoflurane (2-3% isoflurane in oxygen) in an anesthetizing chamber. Anesthesia was maintained using a nose cone. Hair covering the throat area was removed, and the animals were placed over a heating pad for the duration of the surgery. The trachea was exposed by a 2 cm midline skin incision followed by separation of the sternohyoid and sternothyroid muscles. A longitudinal oval-shaped defect was created by using a 1 mm-diameter biopsy punch to create two adjacent holes in the trachea. A 3D printed patch or TachoSil patch with the size of 5 mm in width and 10 mm in length was applied over the defect by gently pressing with a sterile cotton tip applicator for 10–30 s. After adhesion, leakage from the sealed defect was tested by introducing warm saline solution and checking for bubbles. Following confirmation of an air-tight seal, the muscle and skin layers were closed with sutures (4-0 Vicryl, Ethicon). 2, 4, or 6 weeks after the surgery, the animals were euthanized by CO2 inhalation. Tracheal regions of interest were excised and fixed in 10% formalin for 24 h for histological analyses. Fixed tissue samples were placed into 70% ethanol and submitted for histological processing and H&E or MT staining at the Hope Babette Tang (1983) Histology Facility in the Koch Institute for Integrative Cancer Research at the Massachusetts Institute of Technology. Representative histology images were shown in the corresponding Figs. 1, 3, and 6 weeks after surgery, the animals were imaged using Micro-CT. The trachea defect repair model was performed on 3 independent samples with similar results.

Micro-CT imaging of rat tracheas

Micro-CT was performed at the Preclinical Modeling, Imaging, & Testing Core (PMIT) at the Koch Institute. Micro-CT images were obtained using a SkyScan 1276 (Bruker). To perform Micro-CT imaging, rats were anesthetized using isoflurane (1.5-2.5%) and transferred to an animal holder that was mounted into the scanner. Scans were obtained using the following parameters: pixel size 40.2 µm, source voltage 85 kV, source current 200 µA, exposure 179 ms, rotation step 230.5 mm, no frame averaging, and aluminum filter (1 mm). The images were reconstructed using NRecon software (Bruker). Luminal perimeter fraction was computed using ImageJ to measure the ratio of the injured trachea cross-sectional perimeter to the non-injured perimeter for each animal. Luminal area fraction was computed using ImageJ (version 2.1.0) to measure the ratio of the injured trachea cross-sectional area to the non-injured area for each animal. After the conclusion of the 6 week timepoint, the animals were euthanized by CO2 inhalation.

In vivo rat colon defect repair

For the in vivo colon defect repair model, rats were anesthetized using isoflurane (2-3% isoflurane in oxygen) in an anesthetizing chamber. Anesthesia was maintained using a nose cone. Abdominal hair was removed, and the animals were placed over a heating pad for the duration of the surgery. The colon was exposed by a median laparotomy. The exposed colon was packed with moistened sterile gauzes before creating a defect to prevent contamination of the abdominal cavity. A 10 mm incision was made to the colon using scissors and repaired using a 3D printed patch (10 mm in width and 20 mm in length) or sutures. After repair, warm saline was injected into the colon by a 32-gauge needle syringe to confirm formation of a fluid-tight seal. The muscle and skin layers were closed with sutures (4-0 Vicryl, Ethicon). 2 or 4 weeks after the surgery, the animals were euthanized by CO2 inhalation. Regions of interest were excised and fixed in 10% formalin for 24 h for histological analyses. Fixed tissue samples were placed into 70% ethanol and submitted for histological processing and H&E or MT staining at the Hope Babette Tang (1983) Histology Facility in the Koch Institute for Integrative Cancer Research at the Massachusetts Institute of Technology. Representative histology images were shown in the corresponding figures. The colon defect repair model was performed on 3 independent samples with similar results.

In vivo rat liver defect repair

For the in vivo rat liver defect repair model, rats were anesthetized using isoflurane (2-3% isoflurane in oxygen) in an anesthetizing chamber. Anesthesia was maintained using a nose cone. Abdominal hair was removed, and the animals were placed over a heating pad for the duration of the surgery. The liver was exposed by a laparotomy. An injury 5 mm in length and 2 mm in depth was made to the liver using a surgical scalpel. To seal the injury, a liquid-infused patch measuring around 2 cm by 3.5 cm was placed onto the bleeding defect site and gently pressed for 10 s. To confirm hemostasis, the region was washed with saline and observed for 30 min to check for any signs of further blood loss. At the termination of the study, the animals were euthanized by CO2 inhalation. The liver defect repair model was performed on 2 independent samples with similar results.

In vivo rat femoral artery defect repair

For the in vivo rat femoral artery defect repair model, rats were anesthetized using isoflurane (2-3% isoflurane in oxygen) in an anesthetizing chamber. Anesthesia was maintained using a nose cone. Leg hair was removed, and the animals were placed over a heating pad for the duration of the surgery. The femoral artery was exposed via an incision into the thigh. A snip around 2 mm in length was made to the artery using surgical scissors. To seal the injury, a liquid-infused patch measuring around 2 cm by 2 cm was placed onto the bleeding defect site and gently pressed for 10 s. To confirm hemostasis, the region was washed with saline and observed for 30 min to check for any signs of further blood loss. At the termination of the study, the animals were euthanized by CO2 inhalation. The femoral artery defect repair model was performed on 2 independent samples with similar results.

Fabrication of the adhesive bioelectronic patch

Tissue adhesive ink was prepared as described above and printed in a pattern featuring circular void spaces for electrodes onto a glass slide treated with hydrophobic coating. For the insulator layer, 20 w/v % polyurethane (HydroThane AL93A, AdvanSource biomaterials) dissolved in 1:1 tetrahydrofuran (THF) and dimethylformamide (DMF) was prepared and printed over the adhesive layer (400 µm-diameter printing nozzle, 200 kPa, 500 mm min-1). Next, silver conductive ink was used to print the electrodes and circuitry (100 µm-diameter printing nozzle, 50 kPa, 800 mm min-1). LEDs were attached to the circuit using a small amount of the silver ink. After drying, the bioelectronic patch was removed from the glass slide and adhered to an ex vivo porcine heart and a power source was used to run a current through the tissue to confirm illumination of the LEDs.

Fabrication of the mock drug delivery patches

To prepare the fluorescent mock drug-loaded ink, tissue adhesive ink was prepared as described above and mixed with a small amount of fluorescein. The non-fluorescent and fluorescent inks were printed in corresponding positive and negative patterns onto a glass slide treated with hydrophobic coating. After drying, the patch was removed from the glass slide and adhered to ex vivo porcine skin. At sequential timepoints, a small cross-sectional slice of the patch/tissue interface was cut and imaged using an epifluorescence microscope (Nikon Eclipse LV100ND) to capture the diffusion profile into the skin.

To prepare the drug reservoir patch, tissue adhesive ink was prepared as described above and printed with two different lattice dimensions at the reservoir interfaces. The reservoir walls were printed using three layers of an elastomer ink comprising 20 w/w % polyurethane (HydroThane AL93A, AdvanSource biomaterials) dissolved in 1:1 tetrahydrofuran (THF) and dimethylformamide (DMF). Before the walls were fully dried, spin-coated films of the HydroThane solution were cut to shape and attached to provide enclosures for the reservoirs. After drying, the patch was removed from the glass slide and adhered to a gelatin hydrogel. The mock drug solution (a mix of blue food dye and water) was injected into the reservoirs using a syringe needle, and the diffusion of the dye into the gelatin hydrogel was photographed over 5 h.

Statistics & reproducibility

MATLAB (version R2021b) and Microsoft Excel (version 2310) were used to analyze all data in this work. Data distributions were assumed to be normal for all parametric tests but not formally tested. No statistical method was used to predetermine sample size. For the statistical analyses between two groups, statistical significance and p values were determined using a two-tailed Student’s t-test with unequal variance. The following significance thresholds were used: ns p > 0.05; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.