Multimodal and Dynamic Cross-Linking of Modular Thiolated Alginate-Based Bioinks

Engineered extracellular matrix-mimicking hydrogels can facilitate 3D cell culture and fabrication of tissue-like constructs and biologically relevant disease models. Processing of cell-laden


Introduction
Hydrogel-based three-dimensional (3D) cell culture systems offer possibilities to generate biologically relevant tissue and disease models that can facilitate biomedical research and drug development. 1 Hydrogels are highly hydrated polymer networks that can be designed to mimic key aspects of the native extra cellular matrix (ECM). Hydrogels obtained from harvested ECM are widely used for 3D cell culture. However, the complexity, batch-to-batch variability, and varying amount of bioactive factors in these materials can influence experimental reproducibility, which has motivated development of engineered hydrogel systems with more defined compositions. 2,3 Engineered hydrogels can also offer means to tailor cell-hydrogel interactions and possibilities to tune the mechanical and biochemical properties of the materials. [3][4][5] This development is facilitated by hydrogel chemistries that allow for dynamic modulation of hydrogel properties. 6,7 Changes in the hydrogel properties can be cell-mediated 8,9 or triggered by various physicochemical or biochemical ques. 10,11 Cross-linking density can be altered by for example using photo-responsive cross-linkers, 12 proteolytic degradation, 13 or by exploiting physical crosslinking strategies. 10 Physically cross-linked hydrogels are inherently dynamic and often show a time-dependent mechanical response to extrinsic forces, such as stress relaxation and plastic deformation. 14 Stress relaxing and viscoplastic hydrogels can also be produced using reversible covalent cross-links. 15 Time-dependent changes in mechanical properties, including plastic deformation, influence protease independent migration of cancer cells 16 and are central for organoid culture and morphogenesis. 10,17 On the other hand, triggered hydrogel stiffening, 18 softening, 7 activation 19,20 and addition or release of biofunctional motifs can be utilized to control and guide cell behavior with spatiotemporal control. 21 For instance, Hushka et al. 7 used photocleavable bonds in allyl sulfide hydrogels to demonstrate that temporal softening of the materials influenced crypt formation in intestinal organoids. Guvendiren and Burdick 6 showed that the culture period before or after UV-light mediated stiffening of methacrylated hyaluronan-based hydrogels determined the adipogenic and osteogenic differentiation of human mesenchymal stem cells.
Various fabrication techniques have been developed to produce complex functional tissue-mimetic structures using engineered hydrogels. 22 Inspired by additive manufacturing techniques, 3D bioprinting has been shown to be a promising strategy for generating spatially defined multicellular and multi-material constructs. 23 Extrusion bioprinting, where a filament of a bioink comprising cells encapsulated in hydrogels is extruded on a solid substrate or in a colloidal support gel, is the most common technique. However, although a large number of hydrogels can be used for 3D cell culture with satisfactory results, only a limited number of materials are compatible with extrusion-based bioprinting. 24 In addition to requirements on cytocompatibility, the viscoelastic properties before, during, and after printing have a major impact on the performance of the materials. 25 Highly viscous and non-thixotropic hydrogels are difficult to extrude and require high syringe pressures, which can have detrimental effects on cell viability. 26 Printing of very soft materials, on the other hand, pose challenges for creating self-standing 3D structures. 27 Several methods have been proposed to overcome this limitation, 27 including freeform reversible embedding of suspended hydrogels (FRESH). 28 By printing the hydrogel in a support hydrogel bath of e.g., gelatin beads, the shape of the printed structure can be temporarily maintained until the cross-linking of the printed structure has improved the mechanical stability of the structure.
The support can then be removed, usually by increasing the temperature to liquify the gelatin. This strategy can significantly improve fidelity and structural complexity of printed structures but has primarily been used in combination with a limited number of bioink materials, such as alginate 29,30 and collagen 31 , and mostly for generating acellular structures. Essentially, bioprinting using this technique requires either a cross-linking process that is slow enough to provide a reasonable printing window, or that cross-linking can be initiated when the bioink has been embedded in the support bath. However, controlling the degree and timing of cross-linking and generating structures comprising different cross-linkers or bioinks are very challenging using existing materials and printing technologies.
We have recently demonstrated a bioorthogonal hydrogel cross-linking strategy based on enzyme triggered deprotection sulfhydryl groups conjugated to alginate. 32 Phacm-protected cysteine (Cys-Phacm) was conjugated to alginate to generate AlgCP (Scheme 1a). Gelation was initiated after cysteine deprotection using penicillin G acylase (PGA) because of thiol oxidation and formation of intramolecular disulfide bridges (Scheme 1b). Due to the simplicity, biorthogonality, and large number of available chemistries, thiol-based cross-linking strategies are a very common in hydrogels for 3D cell culture but suffer from the short lifetime of thiols under ambient conditions. 33 Using PGA-mediated deprotection of Cys-Phacm functionalized polymers, this limitation can be circumvented, which dramatically facilitates handling of the polymers and processing of the hydrogels, including bioprinting. Here we show that this strategy can be significantly expanded to allow for multimodal and dynamic cross-linking of alginate-based bioinks. Cross-linking by formation of disulfides is a relatively slow process but is reversible and results in self-healing hydrogels. Alginate can also be physically cross-linked using Ca 2+ , which is usually faster and a widely used strategy for generating hydrogels for bioprinting. The possibility to trigger cysteine deprotection and formation of free thiols using PGA, however, also allow for cross-linking using thiol-reactive cross-linkers. PGA-mediated deprotection prevents premature thiol oxidation enabling precise control over both the onset of cross-linking and the stoichiometric ratio between thiols and thiol-reactive cross-linkers, allowing for better tuning of the resulting viscoelastic properties of the hydrogels. Moreover, we show that it is possible to combine these three crosslinking modalities to generate bioinks with very versatile properties (Scheme 1c). Using 4-armmaleimide PEG (p(Mal)4) as a thiol-reactive covalent cross-linker, hydrogels with tunable and dynamic mechanical properties and excellent cytocompatibility were obtained. The materials were compatible with FRESH bioprinting techniques where PGA could be included in the support bath to initiate AlgCP cross-linking post-extrusion, which enabled printing of complex multi-material structures and allowed for long printing times without risking nozzle clogging. By combining materials and cross-linking modalities we further show possibilities to generate structures that could be selectively dissolved while leaving other features intact. Finally, we used these bioinks to print primary human dermal fibroblasts and breast cancer MCF7 cells to demonstrate the cytocompatibility and potential for fabrication of simple breast cancer models. The AlgCP-based bioinks facilitate bioprinting and allows for new means to tune the properties of hydrogels and bioprinted structures. Schematic 1. a) The structure of alginate. Boc-L-Cys(Phacm)-NH2 (BCP) is conjugated to a fraction of the carboxyl groups, generating AlgBCP. b) Chemical structure of AlgCP before and after enzymatic deprotection of the thiol group followed by disulfide formation. c) Illustration of the possible modes of AlgCP cross-linking, including physical cross-linking through ionic interactions between Ca 2+ ions and guluronic acid units, the formation of reversible disulfide bonds, the irreversible Michael addition reaction between sulfhydryl groups and p(Mal)4 cross-linkers, the combination of disulfides and thiol-maleimide cross-linking, and the incorporation of all three cross-linking strategies together.

AlgCP and AlgBCP Synthesis
AlgBCP and AlgCP were synthesized as previously described. 32  Samples were kept at -20 °C for later use. 1 H-NMR was used to confirm the release of the Boc group (Figure S1 c, Supporting information).

Hydrogel formation and characterization
AlgCP and AlgBCP were dissolved in HEPES to prepare a 22.85 mg ml -1 solution. All hydrogels were prepared with this concentration with the appropriate amount of 4-arm PEG-maleimide (p(Mal)4) (Creative PEGWorks, Chapel Hill, USA) at [Mal]:[SH] stoichiometric ratio of 1:10 or as otherwise stated. Before adding the mixture to the PDMS mold for making hydrogel disks or to the rheometer stage for kinetic analysis of cross-linking, 1 unit of penicillin G acylase (PGA) was added to each µmoles of Phacm in the sample (unless otherwise stated) and thoroughly mixed. The cross-linking kinetics was measured using a Discovery HR-2 rheometer (TA instruments, New Castle, USA). Time sweeps were run at 1 Hz and 1% strain using a 20 mm 1° cone geometry at room temperature for 3600 s. Measurements of rheological properties were done using an 8 mm parallel plate geometry. Hydrogel disks (ø 8 mm) were prepared using a PDMS mold and overnight incubation at 37 °C. The hydrogel disks were swelled in HEPES for 2 h at room temperature before analysis. Frequency sweeps between 1 and 10 Hz at 1% strain were conducted at room temperature, followed by amplitude sweeps between 0.1 and 50% at 1 Hz. To investigate gelation of GelCP, samples were prepared in the presence and absence of PGA and with and without p(Mal)4. Hydrogels were prepared by mixing 33.75 µl GelCP (22.85 mg ml -1 , in 10 mM PBS pH 7.2), 9.75 µl p(Mal)4 (35 mg ml -1 in 10 mM PBS pH 7.2) or the same volume of PBS (10 mM, pH 7.2) and 1.5 µl PGA. Cross-linking kinetics was investigated using a 20 mm diameter and 1° coneplate geometry at a frequency of 1 Hz and a strain of 1% at either 23 °C or 37 °C for 20 hours. To avoid evaporation, silicone oil was added around the edges of the geometry and a vapor chamber with wet tissue was used to seal the geometry.

Post-gelation functionalization
Two disks of AlgCP were prepared as described above. Hydrogel disks were either cross-linked using p(Mal)4 or by disulfide bonds. The samples were then treated with 10 µM cyanine 3maleimide (Cy3-Mal, Lumiprobe, GmbH, Hannover, Germany) at room temperature for 1 hour.
A series of photos recorded the staining process over 1 hour. Samples were then thoroughly washed to remove all unreacted Cy3 dye, followed by keeping the samples in HEPES for 24 hours at room temperature.

Post-gelation softening
Three disks of AlgCP were prepared as described above. After swelling the hydrogels in HEPES for 2 h at room temperature. Samples were then treated with 1 ml glutathione (GSH, 5 mM) in HEPES at room temperature for 1 h. The mechanical properties and the size of samples were measured before and after addition of GSH. The kinetics of disulfide bond reduction was obtained by measuring the changes in mechanical properties of AlgCP hydrogels over time using an 8 mm parallel plate geometry. The time sweeps of the samples before GSH addition were performed for 5 minutes at 1 Hz stress and 1% strain. Then 1 ml of GSH (5 mM) in HEPES was added to the sample followed by time sweeps for 1 hour at 1 Hz stress and 1% strain until the storage modulus which had been admixed with PGA, was supplemented with 50 mM of CaSO4. Prior to this addition, the CaSO4 stock solution was subjected to vigorous agitation for 30 seconds. Thereafter, the solution was casted in a PDMS mold. As a control, samples were exposed to only 50 mM CaSO4 in HEPES. All the samples were incubated overnight at 37 °C. All samples were swelled in HEPES for 2 h at room temperature prior rheological analysis.

Dynamic cross-linking of AlgCP hydrogels
Three disks of AlgCP hydrogels were prepared as described above. Samples were swelled in HEPES for 2 h before measuring their sizes and rheological properties. Samples were then treated with 5 mM GSH in HEPES for 1 h at room temperature. The mechanical properties and the size of the hydrogels were measured. Following reduction, samples were washed with HEPES and then incubated with 50 mM CaCl2 in HEPES for 24 h at room temperature. Photos of the samples were taken following washing and swelling in HEPES for 2 h before measuring the rheological properties. The Ca 2+ was sequestered by addition of 50 mM EDTA in HEPES for 12 h at room temperature. EDTA was washed away before rheological analysis and measurement of sizes.
Analysis was repeated after treatment of the samples with 5 mM GSH in HEPES for 1 h.

Multi-layers structures
Multi-layers structures were prepared using AlgCP solutions stained with three different fluorescent dies to aid visualization. All layers were cross-linked by p(Mal)4 after addition of PGA.
The first layer was stained by mixing 0.02 µmoles of Cy3-Mal with the AlgCP and p(Mal)4. After addition of PGA the hydrogels were allowed to cross-link in a PDMS mold (8 mm diameter and 500 µm depth) at room temperature for 1 h. Fluorescein maleimide 6-isomer (0.01 µmol, FAM-Mal, Lumiprobe GmbH. Hannover, Germany) was mixed with AlgCP and p(Mal)4 as the second layer. After addition of PGA, the second layer was added at the top of the first layer and incubated at room temperature for 1 h. The third layer was prepared the same way, and cyanine 5-maleimide (0.02 µmol, Cy5-Mal, Lumiprobe GmbH. Hannover, Germany) was used for staining, and the hydrogel was added on top of the second layer. The multilayer structure was incubated overnight at 37°C. After removing the PDMS mold, the structure was swelled in HEPES for 2 h prior analysis. By slicing the dried sample, the internal structure was exposed and sputter-coated with platinum for 10 s. SEM imaging was conducted using a LEO 1550 Gemini (Zeiss, Germany) at 3 kV operating voltage.

3D Printing in a PGA loaded support bath
A gelatin slurry was prepared as described previously. 28 Briefly, a 4.5% (w/v) solution of gelatin was prepared in MilliQ water, left overnight at 4 °C, and vigorously blended to obtain a gelatin slurry. The slurry was washed several times and kept in a fridge for later use. Before using the slurry for printing, it was centrifuged at 3000 rpm at 4 °C, washed with PBS (10 mM, pH 7.2) twice, and finally centrifuged at 4000 rpm at 4 °C. Three units of PGA were added to each ml of slurry and vortexted, followed by centrifugation at 4000 rpm at 4 °C before use for printing. Two AlgCP bioinks were prepared: "Blue" bioink was labeled with 10 µM Cy5-Mal and did not contain p(Mal)4 , and "pink" bioink which was labelled with 10 µM Cy3-Mal and was cross-linked using p(Mal)4 at a [Mal]:[SH] ratio of 1:10. The hydrogels were then added to two separate syringes (25 G nozzle) and printed using a BioX bioprinter (Cellink AB, Gothenburg, Sweden). The printing parameters are presented in table S1 (Supporting information). For printing of a two-material tube (5 mm diameter and 5 mm height), the "pink" bioink was used for printing the first 6 layers, followed by 8 more layers of the "blue" bioink, and then 6 layers with the "pink" bioink. Printed cubes had dimensions of 5 mm × 5 mm × 3 mm, with 20% grid infill. As a control, one cube was printed into the slurry that lacked PGA using the "blue" bioink.

3D Bioprinting
The support bath was prepared as described above.

Post printing cell viability
The viability of cells was evaluated using a Live/Dead assay 7 days after printing. Samples were washed with PBS and treated with a solution consisting of 2 µM calcein AM (Biotium, Fermont, the incubator.

Cell fixation and staining
After 7 days post-printing, the cell-laden structures were fixated by a 4% formaldehyde solution.

Scanning confocal microscopy
The stained samples were imaged using point scanning confocal microscopy (Zeiss LSM 780, Carl Zeiss, Oberkochen, Germany). All samples were imaged using a 10× objective, and images of figure 5 g and h were digitally magnified two times.

Statistical analysis
The statistical analysis was carried out using Prism (Graphpad Software, Boston, USA). Testing of two experimental groups was done by unpaired t tests analysis with two-tailed P value, and for more than two experimental groups one-way analysis of variance (ANOVA) with Tukey's post hoc testing was used. The number of repeats is reported for each individual experiment.

PGA dependent hydrogel cross-linking
AlgCP was obtained by conjugation of Boc-L-Cys(Phacm)-NH2 to Alg using carbodiimide chemistry, followed by removal of the Boc protection group. 32

Dynamic tuning of hydrogel stiffness
Although the disulfide-mediated cross-linking is slower than cross-linking by p(Mal)4, these crosslinks may continue to form over time, contributing to a gradual increase in stiffness of the hydrogels. However, by keeping the Boc protection group on the cysteine N-terminus, the reactivity of the thiol group will decrease, making it less prone to oxidize and form disulfides.  . g) Shear modulus and h)Tan (δ) of hydrogels cross-linked by p(Mal)4 and disulfides ± additional physical cross-linking by Ca 2+ (50 mM). Error bars are standard deviations. In g) n = 3, unpaired t tests analysis, two-tailed P value, * = 0.0256. In h) n=3, unpaired t tests analysis, two-tailed P value, n.s. = 0.0582.

Hydrogel functionalization
In addition to contribute to cross-linking, the free thiols generated upon addition of PGA can

3D Bioprinting
Printing of 3D structures using very soft hydrogels can be challenging. Despite the relatively slow cross-linking kinetics of AlgCP in the absence of p(Mal)4 and/or Ca 2+ , we showed previously that bioprinting of 3D structures was possible using a thermoreversible hydrogel support bath. 32 Freeform reversible embedding of suspended hydrogels (FRESH) is a powerful technique for printing of soft bioinks with low shape fidelity. 28,35 However, when cross-linking is initiated prior printing, which is the case for most covalent bioorthogonal cross-linking strategies, there will be a limited time window for printing before gelation resulting in clogging of the printing nozzle. [35][36][37] Addition of the cross-linker to the support bath can circumvent this problem. 24,28,30,38,39 However, tuning the properties of the printed structures and printing structures comprising different bioinks or crosslinkers can be challenging using this method. For AlgCP-based bioinks, PGA can instead be added to the support bath while the various cross-linkers can be included in the bioink (Figure 4a,b).
Printing without PGA in the support bath resulted in dissolution of the structures upon removal of the support (Figure 4c). However, with PGA in the support bath, the rapid deprotection of AlgCP after printing resulted in stable structures also in the absence of p(Mal)4, due to formation of

Conclusions
We proposed a novel dynamic, modular and tunable hydrogel system based on enzymatic deprotection of Cys-Phacm conjugated to alginate. The controlled generation of free thiols in AlgCP by penicillin G acylase (PGA) allowed for tuning the cross-linking and properties of the resulting hydrogels with high precision. Different modes of cross-linking of AlgCP were explored, using reversible disulfide bonds, Michael addition reaction between maleimides and thiols, and physical Ca 2+ -mediated cross-linking, as well as different combinations of these cross-linking methods. Cys-Phacm was also conjugated to gelatin for fabrication of thermally stable gelatin hydrogels. We also showed the controlled and dynamic modulation of the mechanical properties of AlgCP hydrogels by selectively influencing the density and combination of the different types of cross-links. Finally, the PGA initiated cross-linking enabled printing of dynamic multi-material and cell laden 3D structures using FRESH with PGA supplemented in the support bath. Because of the possibility to enzymatically release the thiol protection groups at any instance during the processing of the hydrogels, this hydrogel system offers unique possibilities to control hydrogel cross-linking density and kinetics and provides new means to develop advanced biofabrication strategies.

Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.