Contents lists available at ScienceDirect

Journal of the Mechanical Behavior ofBiomedical Materials

journal homepage: www.elsevier.com/locate/jmbbm

Characterisation of hyaluronic acid methylcellulose hydrogels for 3Dbioprinting

Nicholas Lawa,b,1, Brandon Doneya,b,1, Hayley Glovera,b, Yahua Qinb, Zachary M. Amanb,Timothy B. Sercombeb, Lawrence J. Liewc, Rodney J. Dilleyc,d, Barry J. Doylea,b,e,⁎

a Vascular Engineering Laboratory, Harry Perkins Institute of Medical Research, Nedlands, and Centre for Medical Research, The University of Western Australia, Crawley,Perth, Australiab School of Mechanical and Chemical Engineering, The University of Western Australia, Perth, Australiac Ear Science Institute Australia, Perth, Australiad Ear Sciences Centre and Centre for Cell Therapy and Regenerative Medicine, School of Medicine, The University of Western Australia, Perth, Australiae BHF Centre for Cardiovascular Science, The University of Edinburgh, UK

A R T I C L E I N F O

Keywords:BioprintingStem cellsHydrogelHyaluronic acidMethylcelluloseCharacterisation

A B S T R A C T

Hydrogels containing hyaluronic acid (HA) and methylcellulose (MC) have shown promising results for threedimensional (3D) bioprinting applications. However, several parameters influence the applicability bioprintingand there is scarce data in the literature characterising HAMC. We assessed eight concentrations of HAMC forprintability, swelling and stability over time, rheological and structural behaviour, and viability of mesenchymalstem cells.

We show that HAMC blends behave as viscous solutions at 4 °C and have faster gelation times at highertemperatures, typically gelling upon reaching 37 °C. We found the storage, loss and compressive moduli to bedependent on HAMC concentration and incubation time at 37 °C, and show the compressive modulus to bestrain-rate dependent. Swelling and stability was influenced by time, more so than pH environment. We de-monstrated that mesenchymal stem cell viability was above 75% in bioprinted structures and cells remain viablefor at least one week after 3D bioprinting.

The mechanical properties of HAMC are highly tuneable and we show that higher concentrations of HAMCare particularly suited to cell-encapsulated 3D bioprinting applications that require scaffold structure and de-livery of cells.

1. Introduction

Hydrogels have been used across a wide range of biomedical ap-plications and are showing great potential in the field of tissue en-gineering. Hydrogels are three-dimensional (3D) cross-linked scaffoldsof water-soluble polymers, which form a macromolecular networkcapable of retaining high water content. Due to their often poor me-chanical integrity, hydrogels are classified as soft gels with structuralsimilarity to some human soft tissues (Bajaj et al., 2014). However,their hydrophilic polymer networks enable the diffusion of glucose andother nutrients, thus supporting the growth of cells. Additionally, byaltering the concentration of hydrogel components, mechanical prop-erties can often be tailored.

Recently, hydrogels have been used with 3D bioprinting. To be ef-fective as a 3D bioink, hydrogels must exhibit desirable characteristics,such as good printability at low air pressure (< 200 kPa) (Murphy

et al., 2013), minimal swelling and contraction (Murphy et al., 2013;Sun et al., 2011), and excellent biocompatibility (Bajaj et al., 2014).Depending on the application, good structural integrity (Irvine et al.,2015), the ability to maintain physiological pH values and fast gelationtimes (Mayol et al., 2014) are also highly desirable.

Although there are several hydrogels showing promise, blends ofhyaluronic acid (HA) and methylcellulose (MC) have been used forapplications including dermal wound repair (Murphy et al., 2013;Mayol et al., 2014), retinal repair (Ballios et al., 2010, 2015), stroke(Cooke et al., 2011; Caicco et al., 2013a; Tuladhar et al., 2015) andspinal cord repair (Gupta et al., 2006; Baumann et al., 2010, 2009;Wang et al., 2009; Caicco et al., 2013b). Most studies have focused onlow concentration HAMC blends (Ballios et al., 2010, 2015; Cookeet al., 2011; Caicco et al., 2013a; Tuladhar et al., 2015) for delivery oftherapies to specific sites. However, these lower concentrations ofHAMC do not have the mechanical properties required to print 3D

http://dx.doi.org/10.1016/j.jmbbm.2017.09.031Received 15 June 2017; Received in revised form 23 September 2017; Accepted 25 September 2017

⁎ Corresponding author at: Harry Perkins Institute of Medical Research, 6 Verdun Street, Nedlands, Perth, WA 6009, Australia.1 Both authors contributed equally.

Journal of the Mechanical Behavior of Biomedical Materials 77 (2018) 389–399

Available online 28 September 20171751-6161/ © 2017 Elsevier Ltd. All rights reserved.

MARK

scaffolds for applications requiring structural shape and controlled re-lease of therapies.

Therefore, here we aim to characterise HAMC parameters that en-able its use as a viable bioink for tissue engineering applications re-quiring cell delivery via either a soluble delivery vehicle or a 3D scaf-fold. We have investigated a wide range of HAMC blends from viscousgels to structurally stable blends capable of bioprinting 3D scaffolds. Wehave characterised these biomaterials using a suite of experimental teststo determine printability, swelling and stability, rheological properties,compressive modulus, and performance of mesenchymal stem cellswithin the biomaterial. Overall, we have shown HAMC to have highlytuneable properties capable of supporting cell viability and that, de-pending on the HAMC blend, could be useful across a range of bio-medical and 3D bioprinting applications.

2. Methods

2.1. Biomaterial preparation

Hyaluronic acid methylcellulose (HAMC) blends were prepared si-milar to previous reports (Gupta et al., 2006). Methylcellulose (MC;viscosity 15 cP, Sigma-Aldrich) was dissolved in deionized water at90 °C with a stir bar for 4 h to wet the polymer and to produce solutionscontaining 0.5, 1.0, 2.0, 3.0, 5.0, 6.0, 7.0, 9.0 MC wt%. We used a heatplate and oil bath to uniformly heat the solution. Phosphate buffersaline (PBS) was added to the solution in equal quantity to water, andsubsequently cooled to 0 °C using an ice bath for an additional 30 min,after which it was allowed to equilibrate for 12 h at 4 °C. Hyaluronicacid (HA; 1000–1500 kDa, Lotioncrafter) was then added to the MCsolution at a range of concentrations (0.25, 0.5, 1.0, 1.0, 2.0, 2.0, 2.0,2.0 HA wt%) and allowed to dissolve for 12 h at 4 °C. Throughout theprocess we used a magnetic stirrer to maintain uniformity within thematerial. We produced the following HAMC wt% blends; 0.25/0.5, 0.5/1.0, 1.0/2.0, 1.0/3.0, 2.0/5.0, 2.0/6.0, 2.0/7.0 and 2.0/9.0.

2.2. Printability

Solidworks (v23, Dassault Systèmes, France) was used to design a 5× 5 mm grid pattern for printability experiments. Each HAMC blendwas loaded into a syringe at 4 °C for use with the BioBot bioprinter(Beta model, BioBots, USA) and a heat plate at 37 °C was positionedbelow the print head. The printing process was controlled using thefreeware software platform Repetier-Host (Hot-World GmbH&Co. KG,Germany) and the deposition needle speed was held constant at 3 mm/sfor all tests. The initial layer height was 0.1 mm with each subsequentlayer height 0.35 mm and the deposition needle size ranged from 0.1 to0.51 mm inside diameter, with the needle diameter increasing withincreasing viscosity to ensure that printing air pressure did not exceed200 kPa (Fredriksson et al., 2008; Harkin et al., 2006). The grid patternwas printed three times for each HAMC blend and the print accuracy ofthe matrix was calculated using the following formula (Eq. (1)):

=−

×Printing Accuracy A AA

100%i(1)

Where Ai is the area of print (mm2) measured using a digital calipersand A is the design area (mm2) (Duan et al., 2013). Although we used agrid pattern, only the outer dimensions of the grids were used to cal-culate print area. An example printed grid is shown in Fig. S1 in theSupplementary data section.

2.3. Swelling and stability

A potential application of HAMC is the treatment of burns, whereburns often exhibit different pH environments during healing.Therefore the swelling and stability of the HAMC blends was examinedby placing the samples into different saline buffers with varying pH

levels from 5.5 to 8.5, in 1.0 increments.

2.3.1. Buffer preparationFour 50 mmol buffers were prepared with 150 mmol sodium

chloride (NaCl) in deionized water. Saline was added to more closelyresemble the extracellular conditions where the gel will be deposited.Tris Buffered Saline (TBS) was used for the pH 7.5 and .85 solutions,bis-tis buffered saline (BTBS) for pH 6.5 and citrate buffered saline(CBS) for pH 5.5. The pH was adjusted to these values at 37 °C and theresulting buffers were filtered at 0.45 µm to remove particulates.

2.3.2. SwellingFour 0.5 mL aliquots of each HAMC blend were gelled at the bottom

of separate pre-weighed replicate glass vials heated to 37 °C and re-weighed. We then added 3 mL of each of the buffers to the samples, andplaced the vials in an incubator. Buffers were changed and vials re-weighed at 1 h, 3 h, 6 h, 1 day, and 3 days. The swelling ratio wascalculated as defined in Eq. (2):

=Swelling ratioweight of hydrogel tweight of hydrogel

( )(0) (2)

Where the weight of the hydrogel was determined by subtracting theinitial weight of the vial from the total weight with sample after thesurface buffer was carefully removed.

2.3.3. StabilityFour 0.5 mL aliquots of each HAMC blend were gelled at the bottom

of 24-well plates at 37 °C. We then added 3 mL of each buffer andplaced in an incubator. Inspections and supplementation of fresh buffertook place every day for a 15 day period to determine if samples re-mained intact or dissolved.

2.4. Rheological characterisation

We performed all rheological experiments using a TA InstrumentsDiscovery Hybrid DHR-3 controlled stress rheometer, equipped with60 mm 2° acrylic cone-and-plate geometry placed atop an integratedPeltier stage for temperature control. In each experiment, we performedthree repeat trials per sample, and performed each experiment in du-plicate with separately prepared batches of HAMC.

2.4.1. Frequency sweep testsEach rheological characterisation experiment consisted of frequency

sweeps at 1% strain, repeated at 4 and 37 °C. Storage (G′) and lossmoduli (G″) were recorded from 0.1 to 100 Hz, with 10 points taken perdecade. Samples were stored at 4 °C prior to testing. For frequencysweep tests carried out at 37 °C, samples were allowed to equilibrate for20 min prior to testing, with the Peltier stage set to 37 °C (Caicco et al.,2013b).

2.4.2. Time sweep testsTime sweep tests were performed to quantify gelation time with

storage and loss moduli for each blend measured as a function of time.The rheometer temperature was increased from 4 to 37 °C at 5 °C/min,and the moduli were recorded at an angular frequency of 1 Hz and 1%strain. The gelation time was taken as the time from when the materialreached 37 °C to the crossover point of the storage and loss moduli.

2.5. Compression testing

2.5.1. Sample preparationThe compressive behaviour of the four higher HAMC blends (2.0/

5.0, 2.0/6.0, 2.0/7.0, 2.0/9.0) were tested as the lower blends could notmaintain a cylindrical structure. We injected the HAMC blends at 4 °Cinto cylindrical moulds 6 mm high and 11 mm diameter which had

N. Law et al. Journal of the Mechanical Behavior of Biomedical Materials 77 (2018) 389–399

390

been coated with a hydrophobic surface to prevent the HAMC fromsticking to the mould. Samples were maintained at 37 °C within anincubator for 3, 6, 12, 24, 48 and 72 h. Three samples were tested foreach time point.

2.5.2. Compression testsCompression testing was performed on a custom built uniaxial test

machine with a 20 N load cell. Samples were tested at a strain rate of0.1 mm/s. For the 2.0/9.0 HAMC blend, strain rates of 0.05 mm/s and0.2 mm/s were also tested. All samples underwent ten preconditioningcycles to a strain of 10% to account for any potential stress softeningand to ensure data was repeatable (Mullins, 1969). Engineering stressand strain were calculated from force-displacement measurements. Foreach test (n = 3), engineering stress and strain data were combinedinto an average response. Compressive modulus was calculated fromthe slope of the stress-strain curve in the low strain (between 0 and 0.1)and medium strain (0.2–0.35) region. Plotted data represents themedian, with error bars displaying the range (minimum and maximum)of values.

2.6. Cell isolation and culture

Mesenchymal stem cells (MSCs) from adult sheep adipose tissuewere cultured in Dulbecco's Modified Eagle's Medium (DMEM; lowglucose, Gibco™) supplemented with 10% fetal calf serum and 1%Penicillin/Streptomycin. MSCs were fed with fresh medium twiceweekly and cultured until they were 80% confluent. Cells were used atpassage number 4–6.

2.7. Bioprinting cell-encapsulated constructs

We transferred 500 µl of each of the four hydrogel blends (2.0/5.0,2.0/6.0, 2.0/7.0 and 2.0/9.0) stored at 4 °C into 2 mL cartridges. MSCswere enzymatically dispersed from confluent cultures and re-suspendedin 20 µl of culture medium (DMEM). The cell suspension was manuallytransferred by pipette into the middle of the hydrogel blend, andmanually stirred into the bioink to create an evenly distributed singlecell suspension. Media to hydrogel suspension ratios are rarely re-ported, so here we determined an optimum ratio by investigating dif-ferent ratios for mixing ability and resulting printability. We tested thefollowing ratios of media to hydrogel; 0:1, 1:1, 25:1, 50:1 and 100:1.Even cell suspensions were observed in the higher ratios (50:1, 100:1)however viscosity was significantly reduced due to the large amounts ofmedia present. We deemed the 25:1 ratio to be optimum as the cellsuspension was uniform and the viscosity was not affected, as demon-strated by printing with the same air pressure. Samples were printedusing 23-gauge needles (inner diameter = 0.508 mm) at room tem-perature into tissue culture wells using the Inkredible+ bioprinter(CellInk, Sweden). A rectangular strip (0.5 mm thick, 12.5 mm long and3 mm wide) of each hydrogel was printed into each well at a pressure of160–175 kPa, with images of the printed morphology presented asSupplementary Data (Fig. S1). Well plates were then incubated at 37 °C,5% CO2 for 1 h in a humidified incubator.

2.8. Cell viability after bioprinting

To determine if MSCs survive the bioprinting process we used theability of viable cells to exclude the nuclear dye Propidium Iodide(Invitrogen), whereby nuclear staining indicates loss of plasma mem-brane integrity at necrotic cell death. Printed samples were incubated indye solution (1 µg/mL in PBS) for 30 min at 4 °C then observed on aninverted fluorescence microscope (Olympus IX50). We captured phase-contrast and fluorescence images from six different fields on threebioprinted constructs for each HAMC concentration and counted thenumber of total cells (phase image) and of dead cells (fluorescenceimage) in each field using a cell counting tool (ImageJ NIH, USA). At

least 500 cells per HAMC blend were counted and percent cell viabilitycalculated as [(total cells minus dead cells)/total cells]*100. We de-termined the variability of our cell counting methods through in-dependent analysis by two observers and found average differences incell counts of less than 3%.

2.9. Long term characterisation after bioprinting

To determine the long-term viability of MSCs in HAMC, we bio-printed strips (0.5 mm thick, 12.5 mm long and 3 mm wide) into cul-ture wells and inspected cell viability, distribution and density overtime with a phase contrast microscope. We performed this inspectionimmediately post-print and then transferred the well plate to an in-cubator at 37 °C, 5% CO2 for 15 min to gel the solution before adding2 mL of culture medium to the well, and again visually inspecting underthe microscope to ensure the gel structure and cell suspension wasmaintained. We then returned the sample to the incubator and ex-amined the sample every 24 h for 4 days. The media was not replacedduring this time.

After 3 days incubation, the media which still had a cell suspensionwas extracted and was re-plated into a culture flask, with fresh mediaadded to the original plate. Both well plates were left to culture for 2weeks in an incubator and were inspected using phase contrast mi-croscopy for cell adherence and spreading, which were consideredevidence of normal cell morphology and physiology, indicating viabi-lity.

2.10. Statistical analyses

Where applicable, we used t-tests to compare data between differentHAMC concentrations and ANOVA for the live/dead cell counting assayfollowed by t-test for pairwise comparisons. We used P<0.05 to deemstatistical significance throughout.

3. Results

3.1. Printability

For HAMC blends capable of being extruded at airpressure< 200 kPa, the mean printing accuracy was 85% (range 70 −95%) (Fig. 1, n = 3 per blend). The two lowest concentrations (0.25/0.50 and 0.50/1.0) were unable to maintain a printed shape andtherefore were discarded for this analysis. The 1.0/3.0 HAMC blendshowed the best accuracy, however the higher HAMC blends are moresuited to multi-layer 3D bioprinting due to their greater viscosity.

3.2. Swelling and stability

Fig. 2 shows the swelling ratios (mean± SEM) of each HAMC blendwith respect to pH. Within the first hour, the three lowest blends (0.25/

Fig. 1. Printing accuracy for HAMC blends. Larger diameter print needles were re-quired for the higher four blends (2.0/5.0, 2.0/6.0, 2.0/7.0, 2.0/9.0) due to viscosity andair pressure.

N. Law et al. Journal of the Mechanical Behavior of Biomedical Materials 77 (2018) 389–399

391

5.0, 0.5/1.0 and 1.0/2.0) had already become viscous solutions andcould no longer be classified as gels. Since the surface buffer was in-distinguishable from the gel before the first time assessment (1 h), theswelling ratios for these blends were not determined. The fourth lowestblend (1.0/3.0) maintained its cross-linked gel structure for only 3 h.The maximum swelling value for the higher blends (1.0/3.0, 2.0/5.0,2.0/6.0, 2.0/7.0 and 2.0/9.0) occurred at 3 h, where all samples nearlydoubled in weight before gradually dropping to about initial weight orless after 3 days. With the exception of those indicated (Fig. 2), wefound no significant difference in swelling ratios between each pH foreach blend. Although most blends lost their gel-like characteristics after3 days, we continued visually monitoring the samples and found that allblends had completely dissolved after 15 days. These data are alsopresented in the Supplementary data (Fig. S2).

3.3. Rheological characterisation

Fig. 3 shows the frequency sweep measurements at 4 °C. The vis-coelastic moduli of all blends were frequency dependent, with bothstorage (G′) and loss (G″) moduli increasing with frequency. For mostblends tested at 4 °C, G″ exceeded G′ over the initial frequency range,indicating that these blends behave as viscous solutions at lower fre-quencies before elastic properties begin to dominate; this behaviour istypical of an entangled network (Mayol et al., 2014). For all tests car-ried out at 37 °C (Fig. S3, Supplementary data), G′ exceeded G″ over theentire frequency range, indicating gel-like behaviour at a physiologicaltemperature.

We used the crossover point of moduli curves to determine the ge-lation point of each blend. This is seen on the plots (Fig. 4) with viscousfluid behaviour (G′<G″) observed at low temperatures, followed byelastic gel-like characteristics (G′>G″) as the temperature increased to37 °C. The lower HAMC blends displayed gelation either during thetemperature increase to 37 °C or approximately upon reaching this set-point, whereas the higher blends gelated shortly after reaching 37 °C.

It was observed that the gelation time in higher concentrationblends (2.0/5.0, 2.0/6.0, 2.0/7.0) was slower than those in lowerblends (1.0/2.0, 1.0/3.0), thus increasing the HA and MC content re-sults in a longer cross-linking time. As temperature is increased, thepolymers lose their water content, leading to the formation of physicalcrosslinks and gelling of the material via polymer-polymer associations.Due to its high-water retention, the addition of HA in solution lowersthe transition temperature of MC, and can alter the gelation process.This is consistent with our data which indicate increased constituentconcentration increases the rigidity of the formed gel. At 37 °C, themoduli of 2.0/5.0 samples however were relatively high compared toother blends. This may indicate that at 2% HA content, increasing therelative concentration of MC increases the transient temperature andslows the gelation process.

3.4. Compressive behaviour of HAMC

The mechanical response of HAMC was non-linear, with the stress-strain behaviour (i.e. stiffness) increasing with MC content (Fig. 5). TheHAMC blends exhibited high initial stiffness (0 – 0.1 strain region),before somewhat plateauing in the mid-strain region (0.2 – 0.35 strainregion) and then exhibiting another rapid increase in stiffness at higherstrains (0.35 – 0.6). The initial compressive modulus increased linearlywith increasing MC content, indicating the hydrogel's mechanicalproperties are tuneable by altering the MC content. However, MCcontent did not influence the stiffness of the mid strain region (0.2 –0.35) to as great an extent. Furthermore, for the first 24 h, HAMCstiffness in the low strain region increases with incubation time at37 °C. However stiffness stabilises after 24 h, fluctuating between 11.4and 14.5 kPa (Fig. 6A). As before, this increase was not as noticeable inthe mid strain region. The stiffness of HAMC was also strain-rate de-pendent with higher test speeds resulting in higher stiffness (Fig. 6B andC).

Fig. 2. Swelling ratios of all eight HAMC blends. Data shows swelling ratios after (A) 1 h, (B) 3 h, (C) 6 h, (D) 1 day and (E) 3 days for four pH values of 5.5, 6.5, 7.5, and 8.5 (n = 4 foreach test). The swelling ratio (1.0 represent initial volume) peaked at 3 h then gradually dropped over 3 days. There was significant difference (*, P<0.05) between pH 5.5 and 8.5 in the2.0/6.0 blend at 1 h, and between pH 6.5 and 8.5 in the 2.0/7.0 blend at 1 h and 6 h. All blends had completely dissolved after two weeks. These data are also presented in theSupplementary data section (Fig. S2).

N. Law et al. Journal of the Mechanical Behavior of Biomedical Materials 77 (2018) 389–399

392

3.5. Bioprinting MSC-encapsulated constructs

As shown in Fig. 7A, the four HAMC blends with MSC suspensionsdemonstrated suitable cell viability 2 h after bioprinting. Similar cellviability was observed for the 2.0/5.0 and 2.0/6.0 blends (77% and75%, respectively), whereas the 2.0/7.0 and 2.0/9.0 blends had sig-nificantly greater cell survival rates (85% and 91%, respectively)(Fig. 7B). Viable cells were confirmed with CFDA (5(6)carboxy-fluorescein diacetate) fluorescence (Fig. 7C).

Fig. 7D shows phase-contrast microscopy for the 2.0/9.0 HAMCblend over time. Initial inspection (15 min after printing) showed themajority of cells were distributed as single cells with spherical mor-phology. In each print for each of the four HAMC blends (n = 6/blend)we observed good cell density and distribution throughout the HAMC,but not in the culture media or adjacent culture well surface. After 20 h,some cells had released into the media and some had fallen to the ad-jacent well plate, adhered as spherical cells or spread to a polygonalshape (actin skeleton reorganising). The majority of cells were stillsuspended at various planes in the HAMC and appeared viable, with no

clear evidence of shape change (cell spreading) within the hydrogel. By44 h, in all samples, some HAMC-suspended MSCs had been releasedfrom the gel and adhered onto the plastic surface where they had beenprinted, with most flattened to form a patterned surface reflecting theprinted structure. Adjacent well surface was relatively free of cells.Rounded cells floating in the medium were collected after 3 days andre-plated in a fresh dish to test their viability and engraftment potential.They also were viable and mostly adhered after 48 h, with about 50%spreading to take on the typical appearance of MSCs. Over one week ofculturing the adherent cells remained viable and proliferated to form an80% confluent monolayer on the culture well surface. Similar cell be-haviour was observed for all blends, however after 10 days, cells sus-pended in the 2.0/9.0 HAMC blend were still visible in its originalgeometry (rectangular strip), with all other blends partially dissolved.At this time point, the cells encapsulated within the 2.0/5.0 and 2.0/6.0blends had dispersed from the gel suspension and covered the well platesurface, whereas the cells within the 2.0/7.0 and 2.0/9.0 blends hadadhered entirely within the contact region of the HAMC and well plate.

Fig. 3. Frequency sweep measurements of alleight HAMC blends at 4 °C. Three tests per samplewere performed and data showed high repeatabilityat this temperature (data points are practicallyoverlaid due to the repeatability). The blend is in-dicated on each graph.

N. Law et al. Journal of the Mechanical Behavior of Biomedical Materials 77 (2018) 389–399

393

4. Discussion

In this study, we have characterised a range of HAMC blends andhave specifically focussed on the use of HAMC for bioprinting appli-cations. Mechanical properties (compressive, storage and loss moduli)can be easily tuned and MSCs remain viable after the bioprinting pro-cess and in culture within the HAMC material itself.

4.1. Printability and gelation time of HAMC blends

The printability of HAMC has only been investigated once before(Murphy et al., 2013) and was found to be unprintable at 2.0/7.0HAMC, however, the needle diameter was not reported. Here we showthat all printable HAMC blends have print accuracies greater than 70%,demonstrating suitability for a range of applications requiring pressure-

driven extrusion. However, besides needle diameter, other factors affectprintability, especially in multi-layered structures. The gelation time ofthe material and the heat transfer from the print bed are limiting factorswhen printing multi-layered objects, as subsequent layers rely on theinitial layers to be capable of holding additional weight. Gelation timeis therefore a key material parameter of bioinks as this will directlyimpact the control of geometry in multi-layered structures. We haveshown that the HAMC blends tested here gelate either approximately at37 °C or shortly thereafter (< 4 min) which is important as gelationtime can be used to control bioprinting speed of multi-layered prints.However, although we report printability, it should be noted that inmost previous applications of HAMC, geometric control is not as im-portant as the ability to deliver cells to a region (Ballios et al., 2015;Tuladhar et al., 2015; Gupta et al., 2006; Baumann et al., 2010, 2009;Wang et al., 2009; Fredriksson et al., 2008; Duan et al., 2013; Caicco

Fig. 4. Representative time sweep mea-surements of all eight HAMC blends.Temperature (T) was steadily increasedfrom 4 °C to 37 °C during the initial phase(6.7 min) of the experiment with storage(G′) and loss (G″) moduli continuouslymeasured during the test. All blends dis-played gelation either during the tempera-ture increase, at approximately 37 °C, orshortly thereafter (< 4 min). The HAMCblend is indicated on each graph. Note thelog scale y-axis required for the 2.0/5.0 and2.0/9.0 data to show the crossover point ofG’ and G″.

N. Law et al. Journal of the Mechanical Behavior of Biomedical Materials 77 (2018) 389–399

394

et al., 2013b).Rapid gelation upon contact with the body is thought to increase

scaffold longevity and is therefore desirable for the material to sustainand control active release of cells once printed. Most previous studiesused the inverted test tube method to determine gelation times(Murphy et al., 2013; Ballios et al., 2010). Although simple and useful,the inverted test tube method has lower repeatability, requires morevisual interpretation, and is therefore prone to inter-observer error.Using this inverted tube method, Ballios et al. (2010) examined lowconcentration HAMC blends and found long gelation times after equi-libration at 37 °C (e.g. 0.5/0.5 HAMC blend resulted in ~30 min gela-tion time) whereas here we found low concentration blends to gel eitherbefore reaching 37 °C (i.e. 0.25/0.50 HAMC) or approximately uponreaching 37 °C. In the study of 0.25/0.25, 0.5/0.5, 0.75/0.75, 1.0/0.75and 1.0/1.0 HAMC blends by Caicco and colleagues (Caicco et al.,2013b), they did not show the rate of temperature increase from 4 °C to37 °C, but do describe a 20 min equilibration time at 37 °C. They found

gelation times to range from 1.7 to 5 min after reaching 37 °C. We alsoallowed a 20 min equilibration time and observed different trends forsimilar HAMC blends.

In higher HAMC blends, Murphy et al. (2013) used the inverted tubemethod to test the gelation of 2.0/7.0 HAMC and found it to gel at roomtemperature. Gupta et al. (2006) studied the same blend using the samemethod and found the gelation time to be 2 min. Here, using the moreaccurate and reproducible time-sweep tests, we found the 2.0/7.0HAMC blend to gel after approximately 4 min at 37 °C. Interestingly,this blend exhibited the longest gelation time of those examined here,with the 2.0/9.0 blend being the fastest gelling of the four higher blends(1 min at 37 °C). Therefore, although the 2.0/7.0 blend appears to gel atroom temperature, it is not until it reaches 37 °C and is maintained atthis temperature for several minutes that the storage and loss modulicrossover, indicating that the material has transitioned from liquid tosolid state.

Upon analysis of the printability tests, which compared the

Fig. 5. Stress-strain behaviour of HAMC blends.Average stress strain data after (A) 3 h (B) 6 h (C)12 h and (D) 24 h for the four HAMC blends (2.0/5.0, 2.0/6.0, 2.0/7.0, 2.0/9.0) tested. (E) Averagecompressive modulus in the initial strain (0 – 0.1)region for blends with increasing MC concentrationover time at 37 °C. Error bars show maximum andminimum data. Data points are offset for clarity andR2> 0.98 for each trendline.

N. Law et al. Journal of the Mechanical Behavior of Biomedical Materials 77 (2018) 389–399

395

resolution of printing with changing HAMC concentration (and thusviscosity), in conjunction with the rheological data, it appears that anoptimal range of viscosity correlates to increased printability. Such that,the viscosity must be high enough (i.e. HAMC 1.0/3.0) to ensurestructural integrity of the material during extrusion of the strand fromthe nozzle, and low enough (i.e.<HAMC 2.0/7.0) to prevent hetero-geneous strands extruded. Future work would benefit from additionalstudies to determine the optimum printing parameters to for HAMCblends (Webb and Doyle, 2017).

In particular, the reduction in accuracy at higher MC concentrationsis thought to be a result of the rapid gelation and high storage modulusof high MC content gels which can prevent uniform extrusionthroughout a print. The observed difference in printability at lowerconcentrations is potentially explained by lower MC concentrationsplaying a lesser role in the gelation of the hydrogel as pure HA behavesas a weakly viscous solution whilst pure MC behaves as a gel, allowing amore uniform gel. However, although the lower MC concentrationsshow better printability data (using our accuracy calculations), theupper MC concentrations are better suited to multi-layer bioprintingdue to their higher moduli and better structural integrity.

4.1.1. Stability of HAMC over timeWe studied the swelling and stability of the HAMC blends over time

and in different pH environments, and found that most blends swell to apeak volume after 3 h, before gradually reducing in volume over time.Murphy et al. (2013) reported that 2.0/7.0 HAMC swelled to a ratioabove 2.5 after 24 h, whereas for the same blend we observed a peakratio of approximately 1.66 after 3 h, reducing to 1.57 after 24 h, beforereducing further to 1.25 after 3 days. Peak swelling at 3 h is likely thelimit of water intake into the matrix, where the large volume of water

taken in is facilitated by carboxyl group repulsion between and withinHA polymer links (Solis et al., 2012). Whereas the lower HAMC blendsdissolve almost immediately (indicating dissolution rather than actualdegradation), the higher blends retain their gel-like characteristics forup to 3 days before all blends completely dissolve after two weeks. Thisknowledge is important for most tissue engineering applications whereHAMC will be implanted to release therapies in a controlled mannerover time. Furthermore, for wound repair applications, swelling andcontraction of a wound site can slow healing rates and increase scar-ring.

The swelling of a hydrogel was previously shown to be dependenton pH as well as ionic strength and temperature of the surroundingaqueous solution (Gupta and Shivakumar, 2012; De et al., 2002). ThepH of skin has a vital role in the effectiveness of treatment and healing(Percival et al., 2014) with the pH of chronic wounds being variable,ranging from 5.45 to 8.65 (Gethin, 2007), and as the wound heals, itspH progressively drops from alkaline to acidic (Gethin, 2007; Nagobaet al., 2015). Wound pH affects fibroblast activity, oxygen release, an-giogenesis, bacterial toxicity, the proliferation of keratinocytes andimmunological responses (Percival et al., 2014; Nagoba et al., 2015).Our data shows that there is little difference in the swelling ratios ofHAMC across the range of pH levels tested, indicating suitability for usein wound healing.

4.1.2. Compressive modulus of HAMCThe compressive modulus of HAMC is highly tuneable through the

modification of MC content. Blends consisting of less than 2.0% HA donot have enough structural integrity to form cylindrical samples andcould not be mechanically tested using our methods. The increase incompressive modulus in the initial strain region was found to be

Fig. 6. Compressive behaviour of 2.0/9.0 HAMC.The compressive behaviour of the material is sensi-tive to both time at 37 °C and also strain-rate. (A)Average compressive modulus in the initial strain (0– 0.1) region. There was a significant difference be-tween all time-points (P = 0.04). Compressivemodulus becomes stable after 24 h with significantdifferences (*) observed between the 3 h and both24 h and 48 h data. The compressive behaviour ofHAMC is strain-rate dependent with increasingstiffness at higher strain-rate and time at 37 °C; (B)3 h and (C) 6 h at 37 °C.

N. Law et al. Journal of the Mechanical Behavior of Biomedical Materials 77 (2018) 389–399

396

approximately linear with increasing MC content. However, there wereslight differences in the mid strain and high strain regions where theincrease was not linear with increasing MC content or incubation time(Fig. 5A-D). For instance, the compressive modulus of the 6 h samples

had lower modulus in the high strain region (> 50% strain) than the3 h samples. This difference could be experimental error introducedthrough microbubbles in the hydrogel, or through the preconditioningprotocol whereby more loading-unloading cycles were required for

Fig. 7. Bioprinted MSC behaviour in HAMCblends. (A) Microscopy images at 2 h. Phase-contrast microscopy images show cells suspendedthroughout the HAMC blends (a range of focalplanes is evident in cell suspensions).Fluorescence microscopy images show dead cellsstained with Propidium Iodide (red). Trapped airis visible in 2.0/7.0 images. (B) Cell viability isexpressed as the total number of dead cells as apercentage of the total number of live cells perimage analysed (P< 0.05 denoted by *). The 2.0/7.0 and 2.0/9.0 blends had greater cell survivalrates than the lower blends of 2.0/5.0 and 2.0/6.0.(C) Live/dead image of MSCs (live cells stainedgreen with CFDA dye) in the 2.0/9.0 HAMC blendat 2 h. (D) Phase-contrast microscopy imagesshow cells suspended throughout the 2.0/9.0HAMC blend. A range of focal planes is evident incell suspensions. Images show cells through thetime course from immediately after bioprinting,up to 1 week later. Over time, the spherical cellsbecome released from the hydrogel and adheredto the dish, then spread and multiplied, indicatinglong-term viability. In this time, the HAMC gelalmost completely dissolved into solution.

N. Law et al. Journal of the Mechanical Behavior of Biomedical Materials 77 (2018) 389–399

397

certain samples. In most applications HAMC will be used at physiolo-gical temperatures and HAMC scaffold stiffness increases over the in-itial 24 h at 37 °C before stabilising. However, the scaffold also beginsto slowly dissolve after about 3 days and is completely dissolved afterabout two weeks. Using the material constants provided in this study(Table S1, Supplementary Data) together with this knowledge of dis-solution kinetics, scaffolds could be computationally designed to notonly conform to subject-specific geometries, but also to control therelease of therapies either uniformly or in specific areas of the scaffold.We demonstrate the use of these data in numerical experiments in theSupplementary Data. Furthermore, we found the stress-strain responseof HAMC to be strain-rate dependent which could be important in dy-namic applications where HAMC will undergo significant deformationunder variable loading rates, such as in spinal cord repair (Gupta et al.,2006; Baumann et al., 2010). It is also known that the dynamic me-chanical environment in which MSCs reside has major influence ontheir differentiation and outcome (Kurpinski et al., 2006).

4.1.3. MSC viability in HAMCIt is known that scaffold stiffness influences MSC differentiation

(Discher et al., 2005; Park et al., 2011) and here we show that HAMCstiffness is tuneable, which could be used to tailor differentiation pro-files. We have shown that MSCs remain viable in all printable blends ofHAMC and that viability increases with scaffold stiffness. The shearstress exhibited on cells during the printing process is a major cause ofcell death so we maintained air pressures below 200 kPa to reduce thelikelihood of cell death while also ensuring successful material extru-sion. At this pressure and needle diameter (0.508 mm) we observedgood printing accuracy (> 70%) and cell viability (> 75%).

The stiffer HAMC blends (2.0/7.0 and 2.0/9.0) have increased sto-rage, loss and compression modulus, which is the reason for increasedcell viability, relative to the weaker HAMC blends (2.0/5.0, 2.0/6.0). Itis well known that matrix stiffness regulates cell behaviour (Wells,2008; Hadden et al., 2017). HA concentration is known to increase cellviability (cell ligand junctions) (Burdick and Prestwich, 2011), how-ever, in this case HA concentration was kept constant.

While previous work investigated the biocompatibility of HAMC(Mayol et al., 2014; Gupta et al., 2006; Caicco et al., 2013b), none haveexamined its biocompatibility after bioprinting. Murphy et al. en-capsulated keratinocytes in a 2.0/7.0 HAMC blend and reported ~95%viability (Murphy et al., 2013), compared to ~85% MSC viability afterbioprinting here. In fact, when we replated MSCs that were lost fromthe bioprinted 2.0/9.0 HAMC blend into the media, we found them toremain viable (visibly adherent, spreading and proliferative). Thereforewe show that cells not only survive the printing process, but also wereundamaged and retained their key functions over the investigated 1-week time frame. This is important as when cells are encapsulated in abiomaterial construct, they should culture over time and the deviceshould mature prior to implantation. Ultimately, the rate of hydrogeldissolution could be optimized (in consideration of rate and maturationof adhered cells), through increasing the volume of hydrogel mixture aswell as increasing the cross-linking time to improve the structural in-tegrity of the mixture.

4.1.4. LimitationsFirstly, despite being the largest study of HAMC to date, we only

investigated eight different concentrations. Our selection of blends wasbased on a review of prior work in the area. However it would be usefulto extend our study to higher concentrations of both HA and MC. Mayolet al. (2014) studied higher concentrations of both low and high mo-lecular weight HA in HAMC blends and showed that high molecularweight HA reduces cell viability. As the lower HAMC blends are notsuitable for 3D bioprinting, we did not investigate their cellular per-formance and could not study their compressive mechanical behaviour.It would also be interesting to measure the stress relaxation time ofthese HAMC blends and its influence on MSC behaviour. Chaudhuri

et al. (2016) demonstrated enhanced spreading, proliferation and dif-ferentiation of MSCs cultured in alginate-based hydrogels with fasterstress relaxation times, suggesting that hydrogel stress relaxationcharacteristics could play a key role in cell-matrix interactions and bean important design consideration for biomaterials. Also, here weaimed to show MSC viability after bioprinting and therefore did notperform viability assays over extended time periods. While we are sa-tisfied that MSCs remain viable after the bioprinting process and alsowhen encapsulated in the HAMC itself, it would be beneficial to per-form extended biocompatibility studies and assess viability and differ-entiation in bioprinted constructs over time. Finally, although we havedemonstrated in vitro performance of these HAMC blends, we have notextended the work to in vivo. We aim to address this in future studies.

5. Conclusions

In this work we have characterised a wide range of HAMC hydrogelsand reported mechanical and rheological properties, gelation time,printability, swelling and stability, and cell viability. Compressive,storage and loss moduli of HAMC can be easily tuned by altering theratio of HA and MC, with changes to storage and loss moduli impactinggelation time. We have demonstrated the versatility of HAMC as abioink through printability studies and also the swelling and stability ofthe material over time. Mesenchymal stem cells encapsulated in HAMCsurvive the 3D bioprinting process and remain viable in HAMC for atleast one week, and can be distributed to planar surfaces for patterning.We conclude that HAMC is fast-gelling, printable, exhibits desirablemechanical properties and cell viability, and with further work, couldbe a promising hydrogel for many 3D bioprinting applications.

Acknowledgments

We would like to acknowledge funding support from the NationalHealth and Medical Research Council (grants APP1063986 andAPP1083572).

Appendix A. Supporting information

Supplementary data associated with this article can be found in theonline version at http://dx.doi.org/10.1016/j.jmbbm.2017.09.031.

References

Bajaj, P., et al., 2014. 3D biofabrication strategies for tissue engineering and regenerativemedicine. Annu. Rev. Biomed. Eng. 16, 247–276.

Ballios, B.G., et al., 2010. A hydrogel-based stem cell delivery system to treat retinaldegenerative diseases. Biomaterials 31 (9), 2555–2564.

Ballios, B.G., et al., 2015. A hyaluronan-based injectable hydrogel improves the survivaland integration of stem cell progeny following transplantation. Stem Cell Rep. 4 (6),1031–1045.

Baumann, M.D., et al., 2009. An injectable drug delivery platform for sustained combi-nation therapy. J. Control Release 138 (3), 205–213.

Baumann, M.D., et al., 2010. Intrathecal delivery of a polymeric nanocomposite hydrogelafter spinal cord injury. Biomaterials 31 (30), 7631–7639.

Burdick, J.A., Prestwich, G.D., 2011. Hyaluronic acid hydrogels for biomedical applica-tions. Adv. Mater. 23 (12), H41–H56.

Caicco, M.J., et al., 2013a. A hydrogel composite system for sustained epi-cortical de-livery of Cyclosporin A to the brain for treatment of stroke. J. Control Release 166(3), 197–202.

Caicco, M.J., et al., 2013b. Characterization of hyaluronan-methylcellulose hydrogels forcell delivery to the injured spinal cord. J. Biomed. Mater. Res. A 101 (5), 1472–1477.

Chaudhuri, O., et al., 2016. Hydrogels with tunable stress relaxation regulate stem cellfate and activity. Nat. Mater. 15 (3), 326–334.

Cooke, M.J., et al., 2011. Controlled epi-cortical delivery of epidermal growth factor forthe stimulation of endogenous neural stem cell proliferation in stroke-injured brain.Biomaterials 32 (24), 5688–5697.

De, S.K., et al., 2002. Equilibrium swelling and kinetics of pH-responsive hydrogels:models, experiments, and simulations. J. Microelectromech. Syst. 11 (5), 544–555.

Discher, D.E., Janmey, P., Wang, Y.L., 2005. Tissue cells feel and respond to the stiffnessof their substrate. Science 310 (5751), 1139–1143.

Duan, B., et al., 2013. 3D bioprinting of heterogeneous aortic valve conduits with algi-nate/gelatin hydrogels. J. Biomed. Mater. Res. A 101 (5), 1255–1264.

N. Law et al. Journal of the Mechanical Behavior of Biomedical Materials 77 (2018) 389–399

398

Fredriksson, C., Kratz, G., Huss, F., 2008. Transplantation of cultured human keratino-cytes in single cell suspension: a comparative in vitro study of different applicationtechniques. Burns 34 (2), 212–219.

Gethin, G., 2007. The significance of surface Ph in chronic wounds. Wounds UK 3, 3.Gupta, D., Tator, C.H., Shoichet, M.S., 2006. Fast-gelling injectable blend of hyaluronan

and methylcellulose for intrathecal, localized delivery to the injured spinal cord.Biomaterials 27 (11), 2370–2379.

Gupta, N.V., Shivakumar, H.G., 2012. Investigation of swelling behavior and mechanicalproperties of a pH-sensitive superporous hydrogel composite. Iran. J. Pharm. Res.:IJPR 11 (2), 481–493.

Hadden, W.J., et al., 2017. Stem cell migration and mechanotransduction on linearstiffness gradient hydrogels. Proc. Natl. Acad. Sci. 114 (22), 5647–5652.

Harkin, D.G., Dawson, R.A., Upton, Z., 2006. Optimized delivery of skin keratinocytes byaerosolization and suspension in fibrin tissue adhesive. Wound Repair Regen. 14 (3),354–363.

Irvine, S.A., et al., 2015. Printing cell-laden gelatin constructs by free-form fabricationand enzymatic protein crosslinking. Biomed. Microdevices 17 (1), 16.

Kurpinski, K., et al., 2006. Anisotropic mechanosensing by mesenchymal stem cells. Proc.Natl. Acad. Sci. USA 103 (44), 16095–16100.

Mayol, L., et al., 2014. Effect of hyaluronic acid on the thermogelation and biocompat-ibility of its blends with methyl cellulose. Carbohydr. Polym. 112, 480–485.

Mullins, L., 1969. Softening of rubber by deformation. Rubber Chem. Technol. 42 (1),339–362.

Murphy, S.V., Skardal, A., Atala, A., 2013. Evaluation of hydrogels for bio-printing ap-plications. J. Biomed. Mater. Res A 101 (1), 272–284.

Nagoba, B., et al., 2015. Acidic environment and wound healing: a review. Wounds 27(1), 5–11.

Park, J.S., et al., 2011. The effect of matrix stiffness on the differentiation of mesenchymalstem cells in response to TGF-β. Biomaterials 32 (16), 3921–3930.

Percival, S.L., et al., 2014. The effects of pH on wound healing, biofilms, and anti-microbial efficacy. Wound Repair Regen. 22 (2), 174–186.

Solis, M.A., et al., 2012. Hyaluronan regulates cell behavior: a potential niche matrix forstem cells. Biochem. Res. Int. 2012, 11.

Sun, G., et al., 2011. Dextran hydrogel scaffolds enhance angiogenic responses and pro-mote complete skin regeneration during burn wound healing. Proc. Natl. Acad. Sci.108 (52), 20976–20981.

Tuladhar, A., Morshead, C.M., Shoichet, M.S., 2015. Circumventing the blood-brainbarrier: local delivery of cyclosporin A stimulates stem cells in stroke-injured ratbrain. J. Control Release 215, 1–11.

Wang, Y., et al., 2009. Accelerated release of a sparingly soluble drug from an injectablehyaluronan-methylcellulose hydrogel. J. Control Release 140 (3), 218–223.

Webb, B., Doyle, B.J., 2017. Parameter optimization for 3D bioprinting of hydrogels.Bioprinting. http://dx.doi.org/10.1016/j.bprint.2017.09.001.

Wells, R.G., 2008. The role of matrix stiffness in regulating cell behavior. Hepatology 47(4), 1394–1400.

N. Law et al. Journal of the Mechanical Behavior of Biomedical Materials 77 (2018) 389–399

399