Aspects of a Suspended Bioprinting System Affect Cell Viability and Support Bath Properties

Abstract

Suspended hydrogel printing is a growing method for fabricating bioprinted hydrogel constructs, largely due to how it enables nonviscous hydrogel inks to be used in extrusion printing. In this work, a previously developed poly(N-isopropylacrylamide)-based thermogelling suspended bioprinting system was examined in the context of chondrocyte-laden printing. Material factors such as ink concentration and cell concentration were found to have a significant effect on printed chondrocyte viability. In addition, the heated poloxamer support bath was able to maintain chondrocyte viability for up to 6 h of residence within the bath. The relationship between the ink and support bath was also assessed by measuring the rheological properties of the bath before and after printing. Bath storage modulus and yield stress decreased during printing as nozzle size was reduced, indicating the likelihood that dilution occurs over time through osmotic exchange with the ink. Altogether this work demonstrates the promise for printing high-resolution cell-encapsulating tissue engineering constructs, while also elucidating complex relationships between the ink and bath, which must be taken into consideration when designing suspended printing systems.

Impact statement

This work demonstrates the ability to print viable chondrocyte-laden fibers at a resolution on the 50 μm scale, while maintaining viability during printing for up to 6 h using a thermogelling suspended bioprinting system. Factors that ultimately contributed to chondrocyte viability and support bath rheological properties were identified. Additionally, osmotic relationships between the ink and bath materials during the course of printing were observed and characterized, contributing to our understanding of the underlying principles of successful suspended bioprinting systems.

Introduction

Bioprinting, the method of fabricating constructs out of biomaterials and/or cells by three-dimensional (3D) printing, holds significant promise in recreating the complex structure and function of native biological tissue. This promise has led to expansive investigation into bioprinting as a tool for fabrication of constructs within the field of tissue engineering.^1,2 Extrusion bioprinting specifically has been popular for cell-laden construct fabrication, as many common hydrogel systems already popular in tissue engineering can more easily meet the rheological and chemical demands required to successfully print, using an extrusion printer compared to other printing modalities.^3–5

However, extrusion bioprinting has also typically had lower resolution and cell viability when compared to other 3D printing methods.^1,4 The standard extrusion printing requirements of (1) having a material that is both viscous enough to support its own weight after being printed, while also (2) extruding said material through as small a nozzle as possible, all while (3) keeping living cells that have been encapsulated in that material alive, despite putting significant amounts of stress on the material, have traditionally been at odds with another.^6,7

Recent innovations have addressed this conflict in requirements, most notably through the application of hydrogel support baths that can support the structure of less viscous materials until some secondary reaction takes place, which sets the material.^7–15 By enabling the printing of less viscous materials, a library of bioinks that can be extruded through smaller nozzles and place less stress on encapsulated cells is now available for use, increasing both resolution and viability in extrusion printing.

While the use of these support baths has opened new possibilities in extrusion hydrogel printing, it has also complicated the printing system and introduced new underlying relationships that are not fully understood. Most notably, how the inks and support baths, two permeable hydrogel networks, interact with one another when placed in direct contact for extended lengths of time is an important component to understanding the overall function of the system, and is often overlooked in literature regarding these systems.

Specifically, osmotic pressure gradients can drive solvent exchange between the bath and ink, which may affect concentration-dependent properties, such as bath stiffness and yield stress.^15–18 This osmotic pressure has also been shown to drive the expansion of printed hydrogel filaments over extended encapsulation within the support bath, as the ink materials diffuse into the surrounding matrix.¹⁶ These fluid dynamic considerations significantly increase the complexity of accurately modeling these bioprinting systems for predictive printability optimization. Machine learning and computational modeling have begun to be applied to suspended bioprinting systems,¹⁹ but continuing to expand our understanding of how the inks and baths interact with one another through further experimental data can enable us to fully unlock the potential this technique has.

Our laboratory has previously produced one such bioink/support bath system, centered around the use of a four-part poly(N-isopropylacrylamide) (PNIPAAm)-based thermogelling macromer (TGM). Through a unique dual-gelling mechanism, we are able to take advantage of the benefits of nonviscous printing, including using nozzles down to 108 μm in diameter, while relying on the rapid thermal gelation of the TGM to resolve fibers that can then be layered on top of one another.¹⁵

This thermal gelation is followed by a spontaneous epoxy-amine covalent crosslinking reaction that occurs slowly between the TGM and a poly(amidoamine) (PAMAM) crosslinker as the material is printed. This allows the ink to chemically gel and secure a permanent structure before the scaffold is removed from the support bath. The support bath is a poloxamer heated to 37°C, which enables complete thermal gelation of the TGM throughout the entirety of the printed construct. It also allows for printed cells to be incubated within a hydrated, heated environment during the duration of the print.

Using this system, we were successful in fabricating acellular uniform scaffolds consisting of up to 33 layers, with tunable fiber diameters from 80 to 200 μm.¹⁵ In this work, appropriate TGM ink and Synperonic^® bath concentrations were identified and optimized for printability outcomes, and methods for ensuring interlayer adhesion were developed. It was found that the dual-gelling properties of the TGM inks facilitated extrusion through fine nozzles, followed by rapid, initiator-free gelation and crosslinking, which enabled the production of precise and robust scaffolds.¹⁵ Furthermore, the characteristic PNIPAAm behavior of syneresis following thermal gelation may prevent excess diffusion of TGM inks during extended incubation in the support bath, which has been observed in other inks,¹⁶ allowing retained fiber resolution.

The use of TGM as a scaffold material in osteochondral tissue engineering has been previously investigated in our laboratory.^20,21 In particular, TGM possesses regions of relatively high hydrophobicity for a hydrogel, which has been shown to support mineralization during osteogenesis.²¹ As mentioned, we recently developed a method for fabricating TGM scaffolds through suspended bioprinting, which yielded robust and highly uniform scaffolds with tunable fiber diameters below 100 μm.¹⁵

These scaffolds were acellular, so whether cells encapsulated within the TGM inks would survive the temperature change, extrusion through small nozzles, and subsequent incubation within and removal of the Synperonic support material remained to be seen. Our goals in this work, therefore, were threefold: (1) assess what material and printing factors affect encapsulated chondrocyte viability, (2) evaluate the dynamic relationship of the ink and bath materials when printing, and (3) evaluate any change TGM inks and Synperonic baths undergo when used in cell culture conditions compared to prior acellular results.

Materials and Methods

Materials

N-isopropylacrylamide (NIPAAm), glycidyl methacrylate (GMA), (R)-α-acryloyloxy-β,β-dimethyl-γ-butyrolactone (DBA), acrylic acid (AA), acrylamide (AAm), 2,2′-Azobis(2-methylpropionitrile) (AIBN), 1,4-dioxane, piperazine (PiP), N,N′-methylenebis(acrylamide) (MBA), Synperonic F-108, sodium pyruvate, sodium bicarbonate, L-proline, ascorbic acid, HEPES buffer, phosphate-buffered saline (PBS), and dimethyl sulfoxide (DMSO) were purchased from Sigma-Aldrich (St. Louis, MO). Dulbecco's modified Eagle medium (DMEM), antibiotic-antimycotic, nonessential amino acids (NEAA), and mammalian Live/Dead Assay were purchased from ThermoFisher Scientific (Waltham, MA). Fetal bovine serum (FBS) was purchased from Gemini Bio (West Sacramento, CA).

Thermogelling macromer synthesis

The standard PNIPAAm-based TGM was synthesized according to previously established protocols.^15,22 NIPAAm, GMA, DBA, and AA were added to 1,4-dioxane at a molar feed ratio of 86.5:7.5:3.5:2.5, respectively, along with 0.1% w/v AIBN and allowed to react at 65°C under N₂ for 16 h. The dioxane was then removed from the reaction mixture through rotary evaporation and the remaining product was twice purified by solubilization in acetone followed by dropwise precipitation in cold diethyl ether.

The actual mean molar ratio was measured with a combination of proton nuclear magnetic resonance (¹H NMR) (600 MHz Bruker NEO NMR Spectrometer; Bruker, Billerica, MA) and acid-base titration as 87.6:6.3:2.3:3.8. TGM without AA was synthesized in an identical manner at a molar feed ratio of 89:7.5:3.5 of NIPAAm, GMA, and DBA respectively, with a measured mean molar ratio of 90.2:6.4:3.2. Finally, TGM with AAm instead of AA (TGM-AAm) was synthesized at a molar feed ratio of 86.5:7.5:3.5:2.5 of NIPAAm, GMA, DBA, and AAm, respectively, and AAm was confirmed to have been successfully incorporated through ¹H NMR.

The number-average molecular weight (M_n), weight-average molecular weight (M_w), and polydispersity index (PDI) of purified TGM formulations were measured using an ACQUITY advanced polymer chromatography system (Waters Corporation, Milford, MA) with reference to polystyrene standards. Standard TGM was measured to have a M_n, M_w, and PDI of 20,263 ± 1789 Da, 55,663 ± 356 Da, and 2.73 ± 0.27, respectively. TGM without AA had a M_n, M_w, and PDI of 19,548 ± 1487 Da, 53,688 ± 246 Da, and 2.75 ± 0.23, respectively. Finally, TGM with AAm had a M_n, M_w, and PDI of 20,089 ± 1278 Da, 54,845 ± 276 Da, and 2.73 ± 0.21, respectively.

Poly(amidoamine) crosslinker synthesis

The PAMAM was synthesized according to prior protocols.^15,22,23 MBA and PiP were added at a molar feed ratio of 0.75:1 to 30 mL of Milli-Q water at 30°C and allowing to react under N₂ for 48 h. The reaction mixture was then purified by dropwise precipitation in acetone following synthesis. The M_n of the PAMAM was estimated through matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) mass spectrometry²³ to be 1734 ± 345 Da.

Chondrocyte isolation and culture

Primary porcine articular chondrocytes were isolated from the articular joint of female Hampshire pigs donated by the University of Texas McGovern Medical School. Cartilage pieces were excised from the tibial joint and washed twice with 2% antibiotic/antimycotic before being incubated in a mixture of serum-free DMEM and 0.6% collagenase for 24 h. The collagenase solution was then pipetted through a cell strainer before centrifugation. Isolated cells were either frozen in a 90:10 mixture of FBS and DMSO at a concentration of 10 × 10⁶ cells/mL or were cultured in chondrogenic cell culture medium consisting of high-glucose DMEM, 10% FBS, 1% antibiotic/antimycotic, 1% NEAA, 20 mM HEPES, 110 mg/L sodium pyruvate, 50 mg/L ascorbic acid, and 46 mg/L L-proline. All chondrocytes used in the studies were passage number 3 or less and passaged at a maximum confluence of 70%.

TGM ink and Synperonic F-108 bath preparation

TGM, PAMAM, and Synperonic F-108 were sterilized using ethylene oxide gas before use in cellular studies. TGM and PAMAM were solubilized in either PBS or chondrogenic DMEM one day before use, while Synperonic F-108 was solubilized in either PBS or serum-free, phenol red-free chondrogenic DMEM and stored until use. The absence of serum and phenol red in the Synperonic F-108 solutions allowed the support baths to be both transparent and provided contrast for the phenol red-containing ink to aid in observation during printing. Final TGM ink compositions consisted of mixing solutions of TGM and PAMAM together at a constant concentration ratio of 4:3, while increasing the overall concentration of both polymers to increase overall ink concentration.

Here on, “TGM ink(s)” will refer to solutions of TGM and PAMAM combined, unless otherwise specified. Due to the acidic and basic nature of TGM and PAMAM respectively, the two polymers were mixed first before adding cells to the solution, to prevent a drop in cell viability due to pH dysregulation. As the crosslinking reaction between TGM and PAMAM is spontaneous and there was no printing head with mixing capabilities available, the two solutions were mixed immediately before printing and only used for short amounts of time before being replaced with a newly mixed batch. This procedure proved effective in ensuring uniform printing in previous studies.¹⁵

Rheological analyses

The rheological properties of TGM inks and Synperonic F-108 baths were assessed using a Discovery HR-1 hybrid rheometer (TA Instruments, New Castle, DE) in a similar manner to previous studies.¹⁵ TGM ink viscosity was assessed at the printing temperature of 12.5°C by a shear rate ramp from 0.01 to 1000 s⁻¹ using a 20 mm stainless steel parallel plate geometry. Synperonic bath storage and loss modulus and yield stress were measured at 37°C using a 40 mm stainless steel parallel plate geometry containing a solvent trap in a two-step process consisting of a nondestructive oscillatory time sweep at 0.1% strain for 3 min followed by an oscillatory stress ramp between 1 and 1000 Pa.

The storage and loss modulus of a sample were determined by the mean of measured values over the 3-min time sweep, while the yield stress was determined by finding the intersection of a line fitted to the initial plateau of G′ with respect to stress and a line fitted to the linear portion of the downward slope of G′ as stress increases (Supplementary Fig. S1).

Differential scanning calorimetry

The lower critical solution temperature (LCST) of TGM solutions was measured without PAMAM by differential scanning calorimetry using a DSC250 (TA Instruments). All solutions underwent a protocol consisting of an initial ramp down to 5°C, an isothermal period at 5°C for 5 min, followed by a ramp up to 60°C at a rate of 10°C/min. The LCST is visible through an exothermic peak on an otherwise flat isotherm curve. The LCST is presented with both the onset temperature and peak temperature of this exothermic peak.

Printing process

TGM inks and Synperonic bath solution were prepared according to Section “TGM ink and Synperonic F-108 bath preparation.” Around 35% w/v Synperonic was used for all printing, as it was deemed during preliminary studies to generate scaffolds with higher structural stability and uniformity under cell culture conditions than the 30% w/v used in prior studies.¹⁵ Approximately 6 mL of 35% Synperonic solution was added to each well of a 12-well plate and heated to 37°C. TGM and PAMAM solutions were mixed immediately before printing and then replaced every third well to prevent over-crosslinking (Fig. 1A).

FIG. 1.

Experimental procedures. (A) Preparation of chondrocyte-containing TGM inks. TGM and PAMAM were mixed before mixing chondrocytes to ensure a neutral solution. (B) The printing process of two- and six-layered scaffolds for viability, printability, and bath rheology studies. (C) The postprinting process for cell viability studies. (D) The postprinting process for bath rheological studies. PAMAM, poly(amidoamine); TGM, thermogelling macromer.

Two-layered 10 mm × 10 mm scaffolds with 25% infill were printed for the printability and viability assessments, to enable full visualization of fibers and encapsulated cells, while six-layered 10 mm × 10 mm scaffolds were printed for the preprint and postprint bath assessments, to enhance any potential effect, so they can be observed (Fig. 1B). The printing nozzle was calibrated to begin the print approximately at ∼60% of the height of the total bath volume. The nozzle speed was a constant 5 mm/s and the pressure used was dependent on both the ink concentration and the nozzle size (Table 1).

Table 1.

Printing Pressures for Various Combinations of Ink Concentration and Nozzle Size

In addition, the first layer was coded at only 75% of its actual height. This offset the nozzle height downwards for each layer by 75% of the nozzle diameter, ensuring that the nozzle would come into contact with each previous layer, even if the material was underextruding.

Cell viability studies

The scaffolds for the cell viability studies were all prepared in a similar manner. TGM, PAMAM, and Synperonic F-108, along with relevant printing cartridges and nozzles, were sterilized using ethylene oxide before use. TGM and PAMAM were dissolved in chondrogenic media the previous day, while the Synperonic solution could be prepared ahead of time and stored at 4°C. The BIOX printer was relocated within a biosafety cabinet and sterilized before a 12-well plate containing sterile 35% Synperonic F-108 in serum-free, phenol-red free chondrogenic media was placed on the print plate and heated to 37°C.

As the plate was heating, chondrocytes of passage 3 or less were passaged with TrypleE and counted with a hemocytometer, before being diluted in chondrogenic media to an appropriate concentration. Sterile printing cartridges and nozzles were kept on ice as the TGM/medium solution was measured out in aliquots. Immediately before printing, the TGM would be mixed with the PAMAM, neutralizing the two and beginning the spontaneous crosslinking reaction, before the chondrocytes were added and mixed mechanically using a pipette (Fig. 1A). This ink was then pipetted into a printing cartridge and printed, replacing the ink mixture every third well to ensure constant viscosity of the ink, despite the ongoing crosslinking reaction.

Once the cell-encapsulating scaffolds were printed, they were incubated in a cell incubator at 37°C and 5% CO₂ for at least 2 h before being removed with a sterile spatula and placed into excess serum-free chondrogenic media (Fig. 1C). This removal marks time 0 h on the viability timeline. After ∼3 h, the remaining bath was diluted away by the excess media, leaving only the scaffold.

The remaining media were exchanged for fresh, serum-containing media and the scaffolds were either incubated until their appropriate time point or immediately analyzed. This was done by washing the scaffolds thrice with sterile PBS before incubating in a Live/Dead staining solution prepared using the manufacturer's protocol at room temperature for 45 min. Excess staining solution was then removed and the scaffolds were imaged with an ECLIPSE Ti2 inverted fluorescent microscope (Nikon, Tokyo, Japan). These images were robotically taken with a 4 × perspective across the entire scaffold and then stitched together to form fluorescent images of the total scaffold.

The images were thresholded and the live and dead cells counted through particle analysis in the green and red channels, respectively, using ImageJ software to determine cell viability within the scaffold.

Preprinting and postprinting assessment of bath properties

TGM bioinks and Synperonic bath solutions were prepared according to protocols in Section “TGM ink and Synperonic F-108 bath preparation” and then printed according to protocols in Section “Cell viability studies.” As previously mentioned, six-layered 10 mm × 10 mm constructs were printed to enhance any change the bath may undergo when exposed to the ink. Three factors were assessed: ink concentration (17.5%, 26.25%, and 35%), infill density (15%, 25%, and 35%), and nozzle size (27G, 30G, and 32G). When not the factor under analysis, the standard values would be 17.5% ink concentration, a 25% infill density, and a 27G nozzle. Each group also encapsulated 1 × 10⁶ chondrocytes to replicate the conditions of cell-laden printing as closely as possible.

Following printing, the scaffolds and the bath immediately surrounding them would be removed before incubation, to best capture the conditions of the bath at the time of printing (Fig. 1D). A spatula marked for a specific depth would be inserted and used to remove only the scaffold and its local surrounding bath, ensuring the same volume and location of bath were removed for each group. The samples were then sealed to prevent evaporation and incubated at 37°C to allow the scaffolds to crosslink completely. In the meantime, an additional group of baths was heated for the exact same length of time as the experimental baths as a control for changes due to the printing procedure itself and sampled with the marked spatula.

After 2 h of crosslinking, the scaffold-containing baths were refrigerated at 4°C to liquify them, and the crosslinked scaffold was removed with forceps, leaving only a solution of used Synperonic bath. This sample was then assessed using the rheological protocols outlined in Section “Rheological analyses.”

Statistics

Synperonic bath yield stresses were compared using one-way analysis of variance (ANOVA) followed by Tukey's honest significant difference (HSD) (p < 0.05). All other study groups were compared with two-way ANOVA followed by Tukey's HSD (p < 0.05). Values in text and figures are presented as mean ± standard deviation, unless otherwise stated.

Results and Discussion

Material and printing factors and cell viability

While prior studies have indicated that increasing TGM ink concentration can increase print resolution,¹⁵ other studies have shown that increasing TGM concentration can lead to decreased cell viability.²²

To elucidate this relationship, we encapsulated 1 × 10⁶ chondrocytes/mL in 17.5%, 26.25%, and 35% w/v TGM inks and printed them with a 27G nozzle. While the percent viabilities of the three groups were not distinct at a 3-h time point, after 24 h, the viability of the higher concentration inks had declined substantially, to 33.4 ± 7.1% and 35.4 ± 13.5% for the 26.25% and 35% ink groups, respectively, compared to 65.2 ± 7.2% in the 17.5% ink (Fig. 2A). This aligns with previous data regarding TGM concentration leading to cell death, potentially from increased solution osmolality.²² It may also be due to increased printing pressure between the groups (Table 1), which has been shown to be of particular importance in postextrusion cell viability.²⁴ Based on these data, 17.5% TGM inks were exclusively used for the remaining studies.

FIG. 2.

(A–C) Chondrocyte viability at 3 and 24 h within TGM scaffolds printed with varying (A) ink concentrations, (B) nozzle sizes, and (C) cell concentrations. (D) Percent viable chondrocytes at 24 h following incubation in the support bath for different lengths of time. n = 4 for all graphs. (n.s., p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001).

Next, the effect of varying nozzle sizes on encapsulated chondrocyte viability was studied. Among scaffolds printed with a 27G, 30G, and 32G nozzle, no significant difference in viability was observed (Fig. 2B). Prior studies have shown that printing pressure, rather than nozzle size, may contribute more significantly to extrusion-related cell death.²⁴ Since all TGM inks were printed at a relatively low pressure, regardless of nozzle size, this may explain why no significant difference in cell viability was observed, contrary to the well-documented trend of decreasing nozzle size leading to decreases in cell viability.^25–27 This ability to maintain cell viability across nozzle sizes holds promise for continuing to print chondrocyte-laden scaffolds with tunable diameters and high resolution.

To determine the maximum cell concentration achievable within the TGM scaffolds, inks with 1 × 10⁶, 2.5 × 10⁶, or 5 × 10⁶ cells/mL were printed with a 27G nozzle. While cell viability remained constant among all groups at the 3-h time point, a significant decrease was observed within the 5 × 10⁶ cells/mL group, decreasing from 71.6 ± 7.1% at 3 h to 37.2 ± 5.6% (Fig. 2C).

Since the effects of shear-induced cell death can be observed within the same day of printing,²⁶ and the printing pressure was not changed between the three groups, it is unlikely this decrease is due to increased shear forces. The change occurs during postprinting culture conditions, which may indicate the likeliest cause to be inefficient diffusion in and out of the TGM fibers in culture to maintain this number of cells. The current state-of-the-art for printed chondrocyte concentration is 10 × 10⁶ cells/mL and higher,^28,29 so further work will be done to increase the chondrocyte carrying capacity of printed TGM scaffolds.

Finally, to identify the amount of time one could spend printing a single chondrocyte-laden construct using the Synperonic F-108 support bath, scaffolds containing the maximum 2.5 × 10⁶ cells/mL were printed and incubated within the bath for up to 12 h. The minimum required time was 2 h, due to the ongoing crosslinking reaction requiring 2 h to complete. At 2 h, the cell viability was measured to be 73.8 ± 7.7% and did not significantly decrease from this value until the 12-h time point, when the viability was 31.5 ± 17.3%, significantly different than the first three time points of 2, 4, and 6 h (Fig. 2D).

From these data, we can conclude that a concentration of 2.5 × 10⁶ cells can be sustained for up to 6 h within the Synperonic F-108 support bath, indicating a maximum TGM print time of up to 4 h on a single construct, if taking the 2-h crosslinking of the top layers into consideration. At the highest resolution (32G nozzle with ∼50 μm fibers), the TGM is printed at a volumetric flow rate of ∼6 mm³/min (6 × 10⁻³ mL/min). Therefore, over the course of 4 h of printing, a total volume of 1440 mm³ or 1.44 mL can be extruded at that resolution.

The final design of the scaffold, including X-Y infill density, will ultimately affect the total maximum dimensions of a scaffold. Using the dimensions of the scaffold in these studies (100 mm × 100 mm with 25% infill), a maximum construct size of 100 mm × 100 mm × 7.2 mm (144 layers) would be attainable. When the base area is shrunk to 50 mm × 50 mm (2500 mm² or 25 cm²), the maximum height is increased to 28.8 mm (756 layers), to a final scaffold size of 50 mm × 50 mm × 28.8 mm. These dimensions are more than capable of treating most critical-sized cartilage injuries, regardless of location in the body, which on average have an area of 3 cm² and rarely exceed 10 cm² in area,³⁰ with the thickness of the chondral layer not exceeding 2.5 mm.³¹

Furthermore, the ability for Synperonic support baths to maintain viable printed cells for 6 h can be useful in additional applications outside chondrocyte printing with TGM, including printing full osteochondral scaffolds with TGM or printing larger constructs with different ink materials that do not require 2 h of crosslinking and can make use of the full 6-h printing window. This being said, chondrocytes are relatively robust in environments that may be hypoxic or possess minimal transport due to their lower metabolic activity, so further studying how other cell types fare during incubation in Synperonic baths is necessary.

Taken together, these data demonstrate the ability for the TGM bioprinting system to maintain relatively high concentrations of cells at viability between 60% and 75% for substantial print times. This viability can be maintained across a variety of nozzle sizes, enabling the possibility to tune fiber diameters from 175 ± 11 μm (Fig. 3A) to 105 ± 10 μm (Fig. 3B) and 54 ± 15 μm (Fig. 3C). Generally it is accepted that extrusion bioprinting systems can achieve resolutions of ∼150 μm,^32,33 which is larger than other bioprinting systems, but includes benefits related to faster print times and additional multimaterial functionality. The TGM-Synperonic bioprinting system's strengths lie in substantially increasing resolution for an extrusion-based system, down to ∼50 μm features, while still maintaining reasonable print times and multimaterial capabilities.

FIG. 3.

Fluorescent images of TGM scaffolds containing chondrocytes stained with Live/Dead. Live cells are depicted in green, while dead cells are depicted in red. Fiber outlines are depicted in white. (A) Scaffold printed with 27G nozzle. Group fiber diameters were 175 ± 11 μm and chondrocyte viability was 70.3 ± 25.4% (n = 4). (B) Scaffold printed with 30G nozzle. Group fiber diameters were 105 ± 10 μm and chondrocyte viability was 60.8 ± 3.3% (n = 4). (C) Scaffold printed with 32G nozzle. Group fiber diameters were 54 ± 15 μm and chondrocyte viability was 65.9 ± 12.0% (n = 4). Scale bars of large image = 1000 μm, scale bars of insets = 500 μm.

With a 50 μm resolution and a print rate of 360 mm³/h, the system approaches printing capabilities comparable to that of some photolithography systems.³³ While some work is necessary to increase cell carrying capacity and further enhance cell viability, the achievable resolution, print rate, and multimaterial potential indicate that the TGM-Synperonic system holds promise as a method for extrusion bioprinting.

Material and printing factors and bath properties

To assess the relationship between the TGM inks and Synperonic baths, bath storage and loss modulus and yield stress were evaluated before and after printing. Three factors representing different material and printing factors were selected due to their possible relationship with the surrounding support bath. Ink concentration affects the overall osmotic gradient existing between the ink and bath. Infill density affects the total amount of ink being extruded into the bath and the overall ratio of ink:bath in the surrounding area. Finally, nozzle size represents the size of the interface between the ink fibers and surrounding support bath, as the surface area to volume ratio increases as nozzle size decreases.

Changing the ink concentration and infill density had little effect on the bath rheological properties (Fig. 4A, B). The lack of difference between the stock solution and the no ink sample is indicative of minimal evaporation occurring during the printing process within these groups, except for the increase in storage modulus present between the no ink print and stock print during the infill density study. While the regression of the experimental groups, which underwent the same printing process as the no ink print, to a lack of significant difference with the stock solution indicates some level of dilution of the bath, they were not significantly lower than the no ink storage modulus.

FIG. 4.

Storage and loss modulus (left) and yield stress (right) of baths consisting of the preprint stock solution, a control bath that underwent the heating and exposure of the printing process, but was not printed in, and baths that had supported TGM prints of varying (A) ink concentrations, (B) infill densities, and (C) nozzle sizes. n = 4. (n.s., p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001).

On the contrary, there were significant changes in both storage modulus and yield stress in the various nozzle size groups (Fig. 4C). A significant increase in storage modulus and yield stress relevant to the stock solution is likely due to the much longer print time of this study (∼90 min compared to ∼60 min for the ink concentration and infill density studies). However, the significantly decreased values of the 32G storage modulus and yield stress relative to both the no ink print and the 27G groups are indicative of active weakening of these properties during the course of printing.

One possibility for the source of this weakening is the increase in passes that the nozzle will have over the printing space within the support bath when printing at a smaller resolution. This will increase the number of times the bath is deformed and potentially lead to a gradual decrease of bath stiffness and yielding. However, increasing the infill density would also result in an increase in nozzle passes, yet similar results were not observed when changing the infill density (Fig. 4B). In addition, Synperonic gels were previously characterized for changes in storage modulus following cyclic deformation, and while an initial decrease was observed following the first period of high strain, subsequent deformation did not lead to further decrease.¹⁵

Therefore, we believe a more likely cause is the increased size of the interface between the two materials when smaller nozzles are used, enhancing the rate of osmotic exchange between the two and leading to higher quantities of solvent in the surrounding bath. The osmotic exchange between the ink and the bath in suspended bioprinting system has been identified in prior studies, and has led to the observation of phenomena from ink diffusion,¹⁶ scaffold shrinkage,¹⁸ and loss of printability.¹⁵ The significance with which this osmotic exchange and possible dilution of the surrounding support matrix affects the printability will vary and likely need to be assessed for each specific system, and should potentially be included as a regular form of characterization when developing new suspended bioprinting systems.

Effects of cell culture conditions on TGM printability

As discussed, this TGM system has been previously characterized in acellular conditions and was capable of producing highly uniform scaffolds with tunable fiber diameters between 80 and 200 μm.¹⁵ In these conditions, the TGM inks and Synperonic baths were prepared in PBS and analyzed at room temperature. When adapted to a cell culture environment, where the inks and baths were solubilized in cell culture medium (DMEM) and incubated at 37°C for up to 24 h, significant changes in scaffold size and uniformity were observed. Specifically, TGM scaffolds would appear swollen and transparent at room temperature, 25°C (Fig. 5A), and then shrink considerably and appear opaque when incubated at 37°C (Fig. 5B). This is consistent with traditional thermal gelation in PNIPAAm, including syneresis of the material, but indicates an elevated LCST above room temperature.

FIG. 5.

Assessments of the effect of solvent and cell concentration on ink and bath characteristics. (A) TGM scaffold solubilized in DMEM at 25°C and (B) 37°C. The change in opacity of the material is indicative of a change in thermal gelation. (C) Onset and peak LCSTs of TGM inks in various solvents and in the presence or absence of chondrocytes. Groups with different letters have both a significantly different onset and peak LCST (p < 0.01 for both). n = 3. (D) Onset and peak LCSTs of different formulations of TGM both in PBS and DMEM. n = 3. (****p < 0.0001. n.s., p > 0.05 in both onset and peak LCST comparisons). (E) Fluorescent microscopy images of TGM with acrylic acid and (F) TGM with acrylamide scaffolds stained with Live/Dead stain. Green is indicative of live cells, while red indicates dead cells. Scale bar = 250 μm. (G, H) Rheological curves of (G) ink viscosity with respect to shear rate in PBS and DMEM, (H) bath storage and loss modulus in PBS and DMEM, and (I) ink viscosity in DMEM with different concentrations of chondrocytes. n = 5. DMEM, Dulbecco's modified Eagle medium; LCST, lower critical solution temperature; PBS, phosphate-buffered saline.

TGM LCST in varying solvent and cell conditions

Subsequent assessment of the LCST of acellular TGM in both serum-containing DMEM and serum-free DMEM, and TGM containing cells in PBS revealed significant increases in both the onset and peak LCST compared to the acellular PBS control (Fig. 5C). While acellular TGM in PBS has an onset LCST of 16.9 ± 0.7°C and a peak LCST of 24.2 ± 0.5°C, acellular TGM in DMEM has an onset LCST of 21.7 ± 0.2°C and a peak LCST of 29.9 ± 0.3°C, while acellular TGM in serum-free DMEM has a similarly elevated onset LCST of 22.1 ± 0.2°C and peak LCST of 30.3 ± 0.1°C. Finally, TGM in PBS that contained 1 × 10⁶ cells/mL possessed an elevated onset LCST of 19.2 ± 0.3°C and peak LCST of 26.0 ± 0.3°C.

Solvent-dependent behavior of the LCST of differing TGM formulations

Since the LCST of the PNIPAAm-based TGM is a product of the overall hydrophobicity of the macromer,³⁴ varying TGM formulations were studied to determine if one particular component of the macromer was undergoing an external reaction with the DMEM. Due to the epoxy-amine reaction at the center of the crosslinking mechanism for this system, the epoxy-containing GMA group was identified as one possible candidate, potentially reacting with amino acids or other amine-bearing molecules within the media and transforming a relatively hydrophobic GMA group into a hydrophilic side group.

Along a similar line of thought, the carboxylic acid pendant to the AA component is capable of forming peptide bonds with amines, and thus was included in the study as well. Two formulations of TGM, one without GMA and one without AA, were synthesized, with the relative molar amount being converted to additional NIPAAm groups. LCST assessment by DSC revealed that, while TGM without GMA still underwent a similar solvent-dependent LCST elevation as standard TGM (+2.9°C in mean onset LCST and +6.4°C in mean peak LCST), TGM without AA demonstrated no significant difference in LCST between the two solvents (Fig. 5D).

Furthermore, replacing the AA component with another hydrophilic compound, AAm, at the same molar ratio as the original TGM formulation was successful in creating a TGM with a solvent-independent LCST and both an onset and peak LCST comparable to the original TGM formulation (Fig. 5D). Unfortunately, follow-up viability experiments using this AAm-bearing TGM as an ink revealed extensive cell death compared to TGM containing AA (Fig. 5E, F). AAm can have cytotoxic effects at doses above ∼6 mM, which may be present through unreacted monomers,³⁵ particularly at the relatively high polymer concentrations used in the TGM inks. Due to this observed cytotoxicity, it is unlikely that TGM containing AAm is a viable alternative in future studies.

Ink and bath rheological properties in varying solvent and cell conditions

To ensure that observed changes in uniformity were not due to changes in ink or bath rheological properties, ink viscosity was assessed at both low (17.5% w/v) and high (35% w/v) concentrations for TGM dissolved in PBS and DMEM (Fig. 5G) and for TGM containing 0, 1 × 10⁶, or 2.5 × 10⁶ cells/mL (Fig. 5I). In addition, Synperonic bath storage and loss modulus were measured in both PBS and DMEM (Fig. 5H). Minimal changes were seen in each of these studies.

A slight increase in 35% TGM in DMEM compared to 35% TGM in PBS was observed at low shear rates, but at the high shear rates undergone during extrusion, the two inks were comparable. Also, incremental increases in ink viscosity were observed as cell concentration increased, but not to a degree that would affect overall extrudability. These results confirm that the observed changes in printability occur due to chemical changes rather than rheological changes.

Printability and viability at different crosslinker concentrations

While the shrinking and swelling of TGM scaffolds are due to the raised LCST and the thermogelling nature of the macromer, the inclusion of the chemical gelation through covalent PAMAM crosslinking should prevent such dramatic changes in shape.²³ That those changes are occurring is potentially indicative of a lower degree of crosslinking than expected. PAMAM has been observed to potentially interact with proteins in culture medium.^36,37 By increasing the amount of total PAMAM, it may be possible to increase the concentration of PAMAM available for crosslinking. Therefore, TGM scaffolds with increasing concentrations of PAMAM were fabricated and assessed (Fig. 6).

FIG. 6.

(A) Percent area change in TGM scaffolds with differing PAMAM concentrations after incubation at 37°C for 10 min and (B) 24 h. (C) Fiber diameter of TGM scaffolds with differing PAMAM concentrations at 25°C. (D) Percent cell viability of chondrocytes within TGM scaffolds with differing PAMAM concentrations. n = 6 for all groups. (n.s., p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001).

Percent area change for the scaffolds was measured by measuring the side length of the printed scaffold at room temperature, and then incubating at 37°C for either 10 min (Fig. 6A) or 24 h (Fig. 6B) and measuring the side length again. No significant effect was observed on limiting area change using PAMAM concentration, as all groups shrank significantly.

Fiber diameter was measured by allowing the scaffolds to equilibrate at room temperature before fluorescently imaging and measuring the fiber size using ImageJ software. The fiber diameter of the scaffolds containing the standard 7.5% w/v PAMAM amount was measured as 237.4 ± 15.6 μm, while the scaffolds containing 10% w/v PAMAM had smaller fibers of 198.2 ± 8.3 μm, and the 15% w/v PAMAM scaffolds had significantly smaller 168.3 ± 12.8 μm fibers (Fig. 6C). This is notable as also approaching the mean measured diameter of ∼150 μm that TGM scaffolds printed in acellular conditions possessed with this 27G nozzle.¹⁵

However, this smaller fiber diameter was also accompanied by a lower cell viability, with the 15% PAMAM group possessing a cell viability of 75.1 ± 4.0% compared to 87.3 ± 1.8% and 85.9 ± 3.1% for the 7.5% and 10% PAMAM groups, respectively (Fig. 6D). This decrease is likely because PAMAM possesses 2° amines, which are known to have cytotoxic properties.^36,37

The reduced fiber swelling by increased PAMAM concentration, while still undergoing substantial macroscopic changes, may indicate that intrafiber crosslinking was increased, while the interfiber crosslinking that maintains overall scaffold shape was insufficient to resist deformation. While this reduced fiber swelling is ideal, our goal was to optimize cell viability within this system, so we proceeded with the standard 7.5% PAMAM within our TGM ink formulations.

One additional alternative may be identifying an alternative crosslinker. One such alternative, poly(glycolic acid)-poly(ethylene glycol)-poly(glycolic acid)-di(but-2-yne-1,4-dithiol) (PdBT), has been demonstrated by our laboratory as effective systems for creating functionalized TGM scaffolds using a thiol-epoxide reaction rather than the amine-epoxide reaction present in the PAMAM system.^20,38,39 However, the kinetics of the PdBT reaction are on the order of minutes, rather than hours,⁴⁰ and so we were unable to investigate this route due to insufficient hardware, specifically a dual-mixing head that has cooling abilities.

Conclusions

In this work, we have assessed factors affecting the printability and viability of chondrocytes in a PNIPAAm-based thermogelling ink and poloxamer support bath bioprinting system. Material factors, such as ink concentration and cell concentration, were found to have significant effects on encapsulated chondrocyte viability, with maximum values of 17.5% w/v and 2.5 × 10⁶ cells/mL, respectively. Printing process factors, such as nozzle size, were not observed to have an effect on cell viability, indicating the potential for continued printing at high resolution using this system. However, nozzle size was identified as significantly affecting the rheological properties of the support bath over the course of printing. This observation emphasizes the complex and dynamic nature of gel-in-gel printing systems and identifies another potentially key measurement when characterizing new systems.

In addition, Synperonic baths were found to support encapsulated cells for up to 6 h. This outlines a possibility for extended print times that may become necessary as print resolution and construct size continue to increase. Furthermore, while the Synperonic bath is required for the TGM inks to thermally gel properly, the bath design is not perceptibly limited to only this application. The success of this material as a support bath and its off-the-shelf availability and preparation emphasize the interesting possibility of investigating its use as a support bath for other ink materials.

Finally, significant secondary reactions, both with the TGM and potentially with the PAMAM crosslinker, significantly reduced the uniformity of the system observed in acellular conditions. The AA component of the TGM was identified as the source of side reactions on the TGM, ultimately changing the LCST of the ink. It was demonstrated that this could be replaced with another hydrophilic group, creating a TGM with a similar LCST, but absent of any solvent-dependent behavior. However, a replacement that was cytocompatible was not yet identified. Potential candidate copolymers in future studies include poly(vinyl alcohol) and poly(2-hydroxyethyl methacrylate), which are both hydrophilic polymers that have been used extensively in cytocompatible applications. In addition, alternative crosslinkers have been demonstrated by our laboratory as effective systems for creating functionalized TGM scaffolds, but would require additional hardware components to effectively investigate.

As a whole, this work demonstrates the feasibility of fabricating bioprinted TGM constructs laden with viable chondrocytes and composed of fibers between 50 and 200 μm in diameter. It establishes the significant effects that ink and cell concentration have on chondrocyte viability, as well as how printing parameters such as nozzle size and bath residence time affect the maintenance of bath rheological properties and chondrocyte viability, respectively. The analysis of bath storage and loss modulus and yield stress before and after printing demonstrates the effect an extruded ink can have on the surrounding support bath during the printing process, and suggests the need for characterization of these relationships as new suspended hydrogel systems are developed.

Authors' Contributions

A.M.N.: Conceptualization; data curation; formal analysis; investigation; methodology; validation; visualization; writing – original draft; and writing – review and editing. Y.X.: Formal analysis; investigation; and methodology. M.R.P.: Formal analysis; investigation; and methodology. A.G.M.: Conceptualization; funding acquisition; methodology; project administration; resources; supervision; validation; and writing – review and editing.

Disclosure Statement

The authors declare that there is no conflict of interest.

Funding Information

This work was funded by a grant from the National Institutes of Health (P41 EB023833) and A.M.N. and M.R.P. received funding from the National Science Foundation Graduate Research Fellowship Program.