Dynamic peptide-folding mediated biofunctionalization and modulation of hydrogels for 4D bioprinting

Hydrogels are used in a wide range of biomedical applications, including three-dimensional (3D) cell culture, cell therapy and bioprinting. To enable processing using advanced additive fabrication techniques and to mimic the dynamic nature of the extracellular matrix (ECM), the properties of the hydrogels must be possible to tailor and change over time with high precision. The design of hydrogels that are both structurally and functionally dynamic, while providing necessary mechanical support is challenging using conventional synthesis techniques. Here, we show a modular and 3D printable hydrogel system that combines a robust but tunable covalent bioorthogonal cross-linking strategy with specific peptide-folding mediated interactions for dynamic modulation of cross-linking and functionalization. The hyaluronan-based hydrogels were covalently cross-linked by strain-promoted alkyne-azide cycloaddition using multi-arm poly(ethylene glycol). In addition, a de novo designed helix-loop-helix peptide was conjugated to the hyaluronan backbone to enable specific peptide-folding modulation of cross-linking density and kinetics, and hydrogel functionality. An array of complementary peptides with different functionalities was developed and used as a toolbox for supramolecular tuning of cell-hydrogel interactions and for controlling enzyme-mediated biomineralization processes. The modular peptide system enabled dynamic modifications of the properties of 3D printed structures, demonstrating a novel route for design of more sophisticated bioinks for four-dimensional bioprinting.


Introduction
Hydrogels are water-swollen polymer networks that can mimic mechanical, structural and chemical properties of the extracellular matrix (ECM) [1]. ECM-mimicking hydrogels are used for both fundamental studies of cell-ECM interactions [2] and for numerous biomedical applications [3], such as tissue engineering [4], three-dimensional (3D) cell culture [5], and cell-based therapies [6,7]. ECMmimicking hydrogels further facilitates the development of human ex vivo tissue and disease models [8], including organs-on-chips [9], for cancer research [10], drug screening [11], and toxicology [12]. This can reduce both the need for animal models as well as time and costs in drug development [13]. Progress is also boosted by the rapid development of 3D bioprinting technologies that enables fabrication of more complex tissue mimetic constructs, comprising of multiple cell types organized in spatially defined structures [14,15]. The improvements in 3D cell culture strategies and 3D bioprinting put new requirements on hydrogels, both with respect to ECM mimicking capabilities and processability [16,17].
A large number of different biomolecules and synthetic polymers, such as collagen [18] and collagen mimetic peptides [19], cellulose [20], alginate [21], hyaluronan [3], and poly(ethylene glycol) (PEG) [22], are commonly used as components in ECMmimicking hydrogels. The polymer networks can be cross-linked by either covalent bonds or by noncovalent supramolecular interactions [23]. The latter strategy results in supramolecular hydrogels that can self-assemble under benign conditions, and show shear thinning and self-healing properties [24], which facilitates both bioprinting and cell encapsulation. In particular, alginate hydrogels, crosslinked by Ca 2+ , have been widely used in this context [25,26]. Other supramolecular hydrogel systems based on host-guest recognition [27], oligonucleotide-hybridization [28], peptide-folding [29], and protein-ligand interactions [30], have also been explored. However, because of the reversibility of the interactions involved in the cross-linking [31], supramolecular hydrogels tend to dissociate over time, which can complicate long-term cell culture experiments and result in poor shape fidelity during bioprinting. Covalently cross-linked hydrogels, on the other hand, are typically more robust and can demonstrate a wider range of viscoelastic properties [32]. Unfortunately, many conventional strategies for covalent cross-linking suffer from problems with cross-reactivity with cell surface receptors, or require precursors, initiators or catalysts that are cytotoxic or result in the generation of harmful by-products, which can have a negative impact on cell viability [23,33]. Development of bioorthogonal chemistries [34], such as strain-promoted alkyneazide 1,3-dipolar cycloaddition (SPAAC) [35], have in recent years emerged as promising techniques for covalent cross-linking of hydrogels that circumvent these issues. SPAAC does not require any additional reagents or catalysts to be initiated and show no or limited cross-reactivity with biomolecules, which facilitate cell encapsulation [36,37]. Irrespectively, covalently cross-linked hydrogels can be difficult to bioprint and often fail to mimic the dynamic nature of the ECM [17]. The native ECM is not static but constantly subject to change as a result of specific cellmatrix interactions and interaction of various endogenous biomolecular cues, including association and dissociation of growth factors and other proteins to ECM components [38]. The integration of time as a parameter in hydrogel design allows for controlling changes in shape or function of the materials [39]. The possibility to dynamically modify the properties of hydrogels can enable fundamental studies of cellmatrix interactions with temporal resolution [40], and is critical for development of stimuli-responsive structures and four-dimensional (4D) bioprinting strategies [41,42].
In the current work, we have developed a novel strategy for 4D bioink development based on a modular peptide-polymer hybrid hydrogel system that combines the structural and mechanical robustness of bioorthogonal covalent cross-linking with specific supramolecular interactions that enable dynamic tuning of both mechanical properties and the biochemical functionality of the hydrogel. The hyaluronan (HA) and PEG-based hydrogel polymer network was covalently cross-linked by SPAAC by modifying HA with a bicyclo[6.1.0]nonyne (BCN) to enable bioorthogonal covalent cross-linking using an eight-armed PEG with terminating azides (p(N 3 ) 8 ) (figure 1(a)) [37]. In addition, a helix-loophelix polypeptide (JR2EK), designed to fold into a helix-loop-helix motif and dimerize into four-helix bundles, was also conjugated to the HA backbone [43]. The peptides facilitate dynamic modulation of multiple properties via specific peptide-folding mediated interactions, including supramolecular cross-linking, tuning of cell-hydrogel interactions, and specific recruitment of active enzymes. In addition, the combined possibility to control the cross-linking kinetics and density by exploiting the temperature-sensitive SPAAC reaction [37] and peptide-dimerization, enabled bioprinting of intricate 3D structures using the freeform reversible embedding of suspended hydrogels (FRESH) technique [44]. The homodimerization and folding of the conjugated peptides further allowed for changes to the cross-linking density of the hydrogels, and thus the size and mechanical properties of the bioprinted structures. Peptide heterodimerization [45], on the other hand, was utilized to dynamically sequester alkaline phosphatase, triggering mineralization of the bioprinted hydrogels.
Combining bioorthogonal covalent cross-linking with dynamic and specific peptide-folding mediated supramolecular functionalization strategies can enable fabrication of modular ECM-mimicking hydrogels that can be tailored and dynamically modified for a wide range of applications, including 4D bioprinting, and facilitate fundamental studies of both cell-and protein-matrix interactions.
Formation of hydrogels: HA-BCN/JR2EK or HA-BCN was mixed with p(N 3 ) 8 at a volume ratio of 5:1HA-BCN/JR2EK:p(N 3 ) 8 (estimatedBCN:N 3~1 :2) at 1.5-2.5% (w/v) total polymer concentration and added on top of a hydrophobic piece of parafilm. Except for the gelation experiments and if nothing else is stated, all hydrogels were allowed to gel at 20 • C for 24 h prior further experiments.
Oscillatory rheology measurements: The rheological properties of the hydrogels were assessed with a Discovery HR-2 rheometer from TA instruments. Gelation kinetics studies were conducted with a 20 mm 1 • cone geometry at 1% strain and 1 Hz oscillatory frequency. Samples were prepared in situ on the rheometer by mixing all components and quickly lowering the geometry. The gelation point was defined as G' = G". A solvent trap was used for all gelation time studies. Fully formed hydrogels were examined with (1) frequency sweeps at fixed strain of 1%, and (2) amplitude sweeps at fixed oscillatory frequency of 1 Hz using an 8 mm plate geometry. To adjust for any variability of thickness between the hydrogels, the axial force was set to be within 0.2-0.4 N for all measurements. All samples were run in at least duplicates.
Swelling ratio: Preformed hydrogels (2.5% (w/v)) were incubated in excess Bis-Tris buffer (30 mM) at pH 7, pH 4 or with Zn 2+ (10 mM) to swell. After 24 h, the swelled hydrogel was weighted (W w ) and sequentially dried to acquire the dry weight (W d ). The swelling ratio was determined by (equation (1)): All samples were run in triplicates.
Post-modification with JR2KK-Cy5: Preformed hydrogels (2.5% (w/v)) were added to 1 ml PBS (10 mM PO 4 3− , 137 mM NaCl, and 2.7 mM KCl, pH 7.4) to swell for an additional 24 h. Subsequently, the swollen hydrogels were transferred to 1 ml PBS with JR2KK-Cy5 (1 µM) and allowed to incubate. After 24 h the solution was collected, and absorption spectra were recorded to estimate the amount of JR2KK-Cy5 that had been taken up by the hydrogels.
BCIP experiment: Hydrogels pre-modified with ALP (see above) were added to BCIP-buffer (5-Bromo-4-chloro-3-indolyl phosphate,~0.7 mM in Bis-Tris pH 7.0). After 2 h the hydrogels were transferred to Bis-Tris-buffer to quench the reaction and left for 24 h prior analysis. The chromogenic response was measured with a microplate photometer (CLARIOstar Plus from BMG Labtech) at 600 nm.
Biomineralization: Hydrogels pre-modified with ALP (see above) were added to calcification-buffer (50 mM Calcium glycerylphosphate and 30 mM Bis-Tris, pH 7.4). The calcification process was monitored over time with an optical microscope (Nikon Eclipse TI) and compared against a sample incubated with JR2K instead of JR2KK-Biotin. Z-stacks (22 images, 50 µm steps) were focus-stacked using Adobe Photoshop CC (20.0.3 Release). At day 5 the calcified hydrogels were prepared for SEM-analysis. The samples were chemically dried in increasing concentration of EtOH (50%, 70%, 80%, 90%, 100%, 10 min each) and finally twice in hexamethyldisilazane (10 min each). The samples were left in a desiccator overnight prior to SEM-analysis. The SEManalysis was conducted with LEO 1550 Gemini from Zeiss operated at a voltage of 5 kV. An elemental analysis of the samples was performed with energy dispersive x-ray spectroscopy (EDS) (X-Max EDS System Detector, Oxford Instruments) using Aztec 3.1 (Oxford Instruments). The operating voltage for EDS analysis was set to 20 kV.
Cell culture: Human hepatocyte carcinoma cells (HepG2) were acquired from ATCC and expanded in cell culture medium consisting of Dulbecco's Modified Eagle Medium (DMEM, VWR) with 10% fetal bovine serum (FBS, Sigma-Aldrich), 1% Penicillin-Streptomycin (Sigma-Aldrich) and 1% non-essential amino acids (Sigma-Aldrich). Human pluripotent stem cell (hPSC) derived cardiac bodies were prepared at the Hannover Medical School, Hannover, Germany (Dr. Robert Zweigerdt's lab) as previously described [46][47][48], and cultured in RPMI1640 (Life-Technologies) supplemented with B27 minus insulin (LifeTechnologies). Dry powders of HA-BCN/JR2EK and p(N 3 ) 8 were UV-sterilized (60 000 J cm −2 ) and subsequently dissolved in DMEM when used for HepG2 cells or RPMI1640 when used for cardiac bodies. Prior to encapsulation of HepG2, the cells were dislodged via trypsinization and pelleted via centrifugation. Cells were resuspended in cell culture medium and counted. 25 000 cells were suspended in a 2.5% (w/v) p(N 3 ) 8 -solution (5 µl) that sequentially were added to 2.5% (w/v) HA-BCN/JR2EK (25 µl). After thorough mixing and 5 min incubation, the hydrogel-mixture with encapsulated cells was added to a custom mold (glass plates with a separation of 4.7 mm) and incubated (37 • C, 5% CO 2 ) for 1 h. Subsequently, the hydrogels were removed from the mold and placed in fresh DMEM in a 96-well plate to allow the hydrogel to swell and for further incubation. Cell viability was assessed using a Live/Dead® Viability/Cytotoxicity Kit (Thermo Fisher). Samples were imaged with a confocal microscope (Zeiss LSM 700). Human neuroblastoma cells (SH-SY5Y) were acquired from ATCC and expanded in cell culture media consisting of Dulbecco's Modified Eagle Medium (DMEM) and Ham's F-12 Nutrient Mixture (F-12, VWR) in a 1:1 ratio, with 10% fetal bovine serum (FBS, Sigma-Aldrich), 1% Penicillin-Streptomycin (Sigma-Aldrich) and 1% non-essential amino acids (Sigma-Aldrich). Dry powders of HA-BCN/JR2EK and p(N 3 ) 8 were UVsterilized (60 000 J cm −2 ) and subsequently dissolved in DMEM/F12 1:1 cell media. Hydrogels of HA-BCN/JR2EK and p(N 3 ) 8 (volume ratio of 5:1) were formed in half-area 96 well plates (Corning) and incubated for 1 h (37 • C, 5% CO 2 ) before allowing to swell overnight in cell media. The cell media was aspirated and peptides of JR2K or JR2KK-cRGD were added to the surfaces. PBS was added as a negative control. These were incubated for four hours (37 • C, 5% CO 2 ) and then washed very gently with PBS. Cells (SH-SY5Y) were added to the surface of the hydrogel, 40 000 in 100 µl of cell media, and the plate placed in the incubator (37 • C, 5% CO 2 ). Cell viability was assessed Live/Dead® Viability/Cytotoxicity Kit (Thermo Fisher) and using Alamar blue (Thermo Fisher) as a measurement of the reducing power of live cells. Cells were imaged with a confocal microscope (Zeiss LSM 780).
Gradient fabrication: Hydrogels consisting of HA-BCN/JR2EK and p(N3)8 (volume ratio of 5:1) were formed within the central chamber of µ-slide Chemotaxis channel slide (ibidi GmbH, Gräfelfing, Germany) by incubating for 1 h at 37 • C. PBS was added to both reservoirs to allow the hydrogel to swell and to equilibrate overnight. The PBS was gently aspirated and then replaced with the corresponding Cy(3/5) fluorophore conjugated JR2KK peptide. Imaging was carried out using a Nikon Ti-Eclipse fluorescence microscope. The excitation wavelength was 550-570 nm and 630-650 nm for Cy3 and Cy5 respectively. The emission was recorded over the interval 580-640 nm and 660-695 nm for Cy3 and Cy5, respectively.
3D bioprinting: 3D printing was conducted using the freeform reversible embedding of suspended hydrogels (FRESH) technique [44]. A slurry of gelatin was made by dissolving 4.5% (w/v) gelatin in 30 mM Bis-Tris buffer pH 7 in the presence or absence of 10 mM ZnCl 2 and incubated at 4 • C for 12 h to solidify. The the jar was topped up with buffer and the gel was vigorously mixed for 120 s using a kitchen blender. The resulting slurry was transferred to 50 ml centrifuge tubes and the gel particles were pelleted by centrifugation at 1500 rpm for 4 min at 4 • C. The pellet was washed with cold (4 • C) buffer (30 mM Bis-Tris buffer pH 7 ± 10 mM ZnCl 2 ). This process was repeated five times to completely remove any dissolved gelatin from the solution. The bioinks were prepared by mixing p(N 3 ) 8 (25 mg ml −1 ) and HA-BCN/JR2EK (25 mg ml −1 ) at a 5:1 volume ratio and incubated on ice for 5 min and then transferred to a 3 ml syringe affixed to a 27-gauge blunt end needle. The syringe was loaded into a Bio X bioprinter from Cellink (Gothenburg, Sweden). The printing head and the bed temperature were set to 8 • C. The structures were printed inside the gelatin slurry using 5 kpa pressure and 5 mm s −1 printing speed. After printing, the structures were incubated at room temperature in the slurry for 1 h. Then structures were harvested by dissolving the bath by heating to 37 • C and washed in buffer. Human dermal fibroblasts (0.25×10 6 ) fibroblast where centrifuged and resuspended in 50 ul of p(N 3 ) 8 (25 mg ml −1 ) in high glucose DMEM with 10% fetal bovine serum and 1% Penicillin-Streptomycin. The cell-p(N 3 ) 8 suspension (8 µl) was mixed with 40 µl of HA-BCN/JR2EK and incubated (37 ºC, 5% CO2) for 5 min to initiate the cross-linking. The fibroblast containing hydrogel was printed on glass coverslips using a 27-gauge blunt end needle with a printer speed of 4 mm s −1 and a pressure ranging from 1 to 5 kPa. The printed structures were incubated at 37 ºC and 5% CO 2 for 20 min prior to addition of 3 ml of high glucose DMEM with 10% fetal bovine serum and 1% Penicillin-Streptomycin and incubated for 18 h. Media was removed and cell viability was investigated using Live/Dead® Viability/Cytotoxicity Kit (Thermo Fisher). The cells were imaged with a widefield microscope (Zeiss Axiovert) with a Zeiss HRm camera and a 20x/0.4 objective.
Statistics: Data presented are shown as mean ± S.D. For statistical analysis paired sample t-test using Origin version 9.0.0 was used. In graphs, * indicates a statically difference of P < 0.05 and * * indicates a statically difference of P < 0.01.

Hydrogel fabrication
To allow for bioorthogonal SPAAC reaction with azide-moieties, HA was first modified with BCN groups (HA-BCN) using carbodiimide chemistry (supporting figure S1) as previously described by us [37,43]. The degree of modification was estimated as~7% based on 1 H-NMR signals originating from the BCN group (supporting figure S2). The azidemodified polypeptide JR2EK(N 3 ) was then conjugated to HA-BCN yielding HA-BCN/JR2EK (figure S1, supporting information). The azide moiety in JR2EK(N 3 ) was included in the loop-region of the polypeptide to minimize the influence of conjugation on dimerization and folding. The design of the polypeptide JR2EK was based on the helix-loop-helix polypeptide JR2EC, and has previously been explored for self-assembly of HA-based hydrogels [43]. The degree of polypeptide conjugation to HA (~4%) was estimated based on the 1 H-NMR signals originating from histidine and phenylalanine residues in the polypeptide (supporting figure S2). Roughly half of the initial BCN groups were thus left unreacted and could be used for covalent cross-linking of HA-BCN/JR2EK using the SPAAC reaction. Covalent cross-linking was achieved by mixing HA-BCN/JR2EK with p(N 3 ) 8 , resulting in the formation of a fully transparent hydrogel HA-BCN/JR2EK-p(N 3 ) 8 ( figure 1(b)).
To determine a suitable polymer ratio between HA-BCN/JR2EK and p(N 3 ) 8 , the two components were mixed at different ratios and the change in viscoelastic properties caused by gelation was monitored with oscillatory rheology. A HA-BCN/JR2EK:p(N 3 ) 8 polymer ratio of 5:1 (with 2.5% w/v polymer stocksolutions) was found to generate hydrogels with the most rapid gelation kinetics and highest storage modulus (G') after 2 h (supporting figure S3). The 5:1 polymer ratio was thus used in all future experiments. Furthermore, using this ratio, different total polymer concentrations (1.5-2.5% (w/v)) were mixed at room temperature (~20 • C) and the rheological response was evaluated. The G' after 2 h ranged from 2 to 150 Pa ( figure 1(g), table 1). The gelation point, i.e. when G' = G", occurred within 20 min after mixing the components at concentrations of 2.0-2.5% (w/v). When lowering the concentration to 1.5% (w/v) the gelation was slower, and the gelation point was reached after approximately 100 min. No further increase in G' was seen after 24 h incubation, with G´values ranging from 200 to 500 Pa ( figure 1(c), table 1, supporting figure S4). The mechanical properties of the hydrogels were thus in a biologically relevant range for mimicking soft tissues [49].

Cell encapsulation
The possibility to control the viscosity and gelation kinetics of hydrogels is critical for both 3D cell encapsulation as well as 3D bioprinting [50]. Although the gelation time for the HA-BCN/JR2EK-p(N 3 ) 8 hydrogels was fairly rapid (20-100 min) for the concentrations tested, it was not as fast as many supramolecular hydrogels, which can reach their maximum G' within seconds [29,51]. Other SPAAC-based hydrogels have also shown a more rapid gelation [34,50], although they typically have a higher degree of BCN and azide modification. However, by increasing the temperature to 37 • C the gelation rate could be greatly enhanced (figures 1(d)-(e), table 1). At 2.5% (w/v) and 37 • C the gelation point was reached within 3 min, and at 2.0% (w/v) it was reached within 10 min. At 1.5% (w/v) the gelation time was reduced to 25 min, as compared to 97 min at 20 • C. Likely, this is a result of the faster diffusion of the polymers and polymer segments at elevated temperatures, thus increasing the collision frequency and the probability that a BCN and an azide moiety will align correctly to react. To confirm that it was possible to encapsulate cells in 3D, and that the hydrogel components were cytocompatible, human hepatocyte carcinoma cells (HepG2) and human pluripotent stem cell (hPSC) derived cardiac spheroids (cardiac bodies, CBs) were encapsulated and cultured in the hydrogels. 3D cell culture of both HepG2 and CBs are of interest for in vitro drug screening and toxicology testing [52]. It has previously been shown that HepG2 express improved liver-like properties when cultured in 3D [12]. In addition, HA is abundant in the ECM of the heart where it is involved in cardiac development during embryogenesis and healing after myocardial infarction, for example [53]. To encapsulate the cells in the hydrogel, the cells were premixed with the hydrogel components (2.5% (w/v)) and allowed to cross-link for 1 h at 37 • C, and then submerged in cell culture medium. The HepG2 cells showed high viability in the hydrogels and an increase in cell density and formation of spheroids occurred over a time course of seven days (supporting figures S5(a) and (b)). We have previously seen that HepG2 cells tend to form spheroids in HA-based hydrogels lacking cell adhesion motifs [37]. The diameter of the spheroids at day 7 was in the range of 80 µm, a size that is still sufficient for effective oxygen diffusion into the center of the spheroid [54]. Furthermore, the viability of the HepG2 remained high after 7 d and very few dead cells could be seen in, or close to, the spheroids (supporting figures S5(c)-(f)). The CBs showed high viability with characteristic beating several days after encapsulation in the hydrogels (figures 2(a)-(c), supporting movie 1). Initially separated CBs were able to selfassemble into larger structures during the time course of days and synchronize their beating. The bioorthogonal SPAAC reaction thus allow for encapsulation and 3D culture of both HepG2 cells and CBs, promoting high cell viability and maintaining characteristic cell functions.

Supramolecular cross-linking
In addition to the bioorthogonal covalent crosslinking of the hydrogels, the JR2EK polypeptide conjugated to the HA can provide supramolecular cross-links that can assist in the formation and contribute to the final viscoelastic properties of the hydrogel ( figure 3(a)). JR2EK is designed to fold into a helix-loop-helix motif and homodimerize into four-helix bundles at acidic pH (pH < 6) or in the presence of Zn 2+ [55]. The isoelectric point (pI) of JR2EK is 4.5 and protonation of Glu residues at the dimer interface readily promote homodimerization in a pH range close to the pI. To assess the impact of polypeptide folding and dimerization on the HA-BCN/JR2EK-p(N 3 ) 8 hydrogel properties, the gelation kinetics was examined at both acidic pH (pH 4.5) and after addition of Zn 2+ (10 mM) to HA-BCN/JR2EK and p(N 3 ) 8 (2.5% (w/v), 20 • C). As expected, the addition of Zn 2+ gave an instant increase in G' from~1 to~100 Pa ( figure 3(b), table 2). This is in sharp contrast to the gradual and slower increase in G' seen when relying on the SPAAC cross-linking alone. Furthermore, the gelation kinetics and initial storage modulus was similar to fully supramolecular hydrogels obtained by the Zn 2+ -triggered and folding-dependent self-assembly of JR2EK when conjugated to HA [43]. After the inital rapid supramolecular peptide-mediated crosslinking of the hydrogels, a slower increase in G' over time indicates the formation of additional covalent cross-links. Interestingly, the combined covalent and supramolecular cross-linking resulted in hydrogels with a significantly higher loss modulus (G") as compared to hydrogels that were only covalently crosslinked (table 2). The higher G" implies that the resulting hydrogels can dissipate more energy as heat, or reform when subjected to stress and strain, making them less brittle as compared to covalently crosslinked hydrogels. A similar response could be seen when triggering peptide dimerization by reducing the pH to 4.5, albeit with a less dramatic initial increase in G' (figure 3c, table 2). Both strategies for triggering peptide homodimerization resulted in a distinct hydrogel deswelling (figure 3(d)) and a decrease in swelling ratio (figure 3(e)) as a result of the increase in cross-linking density.
In control experiments using hydrogels based on HA-BCN, i.e. HA without conjugated JR2EK, no significant change in the viscoelastic properties upon addition of Zn 2+ or in acidic pH were observed (supporting figure S6). Hence, the results imply that it is possible to simultaneously cross-link the hydrogel by both covalent and supramolecular interactions. The supramolecular cross-links, however, appear to form on a much shorter timescale than the covalent bonds, providing efficient means to instantly tune the viscoelastic properties and the kinetics of hydrogel formation. Moreover, by combining robust covalent interactions that can provide elasticity with weaker and more dynamic supramolecular cross-links, it is possible to create hydrogels with a wider repertoire of mechanical properties.

Dynamic hydrogel functionalization
In addition to the possibility to tune the degree and rate of cross-linking, the JR2EK-polypeptide could also be used to dynamically modify the chemical composition of the formed hydrogels ( figure 4(a)). As a proof of concept, a fluorophore labeled complementary helix-loop-helix polypeptide (JR2KK-Cy5, supporting figure S1(d)) was synthesized. JR2KK is designed to fold and heterodimerize with JR2EK to form an antiparallel four-helix bundle [45,56]. Hydrogels were incubated in PBS buffer (10 mM PO 4 3− , 137 mM NaCl, and 2.7 mM KCl, pH 7.4) for 24 h prior to incubation with JR2KK-Cy5 (1 µM, in PBS) for an additional 24 h. Buffer was then collected, and absorption spectra were recorded to estimate the amount of JR2KK-Cy5 adsorbed in the hydrogels. Although a certain amount of JR2KK-Cy5 was found to adsorb nonspecifically to the control hydrogel, the possibility to form heterodimers resulted in a significantly larger uptake of JR2KK-Cy5 ( figure  4(b)). Furthermore, already after only one hour of incubation, a distinct color change could be noticed for the JR2EK-functionalized hydrogels that could not be seen in the JR2EK-lacking control (supporting figure S7). In addition, after seven days of incubation in PBS, the polypeptide-functionalized hydrogel still maintained its distinct teal color caused by the associated Cy5-functionalized peptides, whereas the control was almost transparent ( figure 4(c)).
The high loading and retention capacity of JR2KK-Cy5 in the HA-BCN/JR2EK-p(N 3 ) 8 hydrogels strongly indicate the possibility to use heterodimerization to dynamically change the composition of the hydrogel post-synthesis. Moreover, because of the relatively high local concentration of conjugated peptides (JR2EK) in the hydrogel, the dimerization with the complementary peptide JR2KK was diffusion limited, which facilitated fabrication of well-defined linear 3D gradients of peptide heterodimers. Gradients were fabricated using a microfluidic device (ibidi © µslide Chemotaxis) comprising a linear channel, 1 mm wide, with openings on each side exposing the hydrogel to liquid in two larger reservoirs ( figure 4(d)). HA-BCN/JR2EK-p(N 3 ) 8 hydrogels were injected into the channel and after 1 h incubation at 37 • C to allow the hydrogel to cross-link, complementary peptides labeled with Cy3 (JR2KK-Cy3) and Cy5 (JR2KK-Cy5) in the loop were added to the left and right reservoir, respectively. The diffusion of the peptides into the hydrogel and subsequent dimerization with conjugated JR2EK resulted in formation of almost linear gradients of the peptides that could be visualized using fluorescence imaging ( figure 4(e)). Previous attempts to produce long-term stable gradients of immobilized biomolecules in 3D hydrogels typically rely on photopatterning using UV light [57,58], which requires more complex fabrication techniques. Here, the slope of the gradient and the degree of functionalization could be controlled by simply changing peptide concentration and diffusion time ( figure 4(f)).
The rapid association of the complementary peptide to conjugated JR2EK further enabled dynamic surface functionalization of hydrogels, which can be used for tuning of cell adhesion and spreading of cells cultured on the hydrogels. Seeding of SH-SY5Y neuroblastoma cells on the HA-BCN/JR2EK-p(N 3 ) 8 hydrogels resulted in cell aggregation due to the lack of cell-surface adhesion sites (figure 5). In contrast, surfaces modified with the complementary peptide JR2KK, with and without a cyclic RGD (cRGD) peptide motif in the loop region (JR2KK-cRGD), showed a substantial improvement in cell spreading and cell viability. In addition to the specific integrin binding to the cRGD peptide, the high lysine-content in JR2KK likely contributed to promoting cell adhesion.

Biomineralization
To further explore the potential to dynamically introduce new functionalities in the hydrogels using specific peptide-mediated interactions, we modified the complementary peptide with a biotin moiety in the loop (JR2KK-Biotin, supporting figure S1(d)). When associated to the HA-BCN/JR2EK-p(N 3 ) 8 hydrogels, this peptide could recruit avidin-modified proteins, which opens up for a plethora of functionalization possibilities. The enzyme alkaline phosphatase (ALP) catalyzes the dephosphorylation of several organic orthophosphates and is central for the mineralization of bone [59]. For functionalization of the hydrogels with ALP, the hydrogels were first incubated with JR2KK-Biotin, and subsequently exposed to streptavidin-modified ALP ( figure 6(a)). To verify that ALP was adsorbed in the hydrogel, 5-Bromo-4chloro-3-indolyl phosphate (BCIP) was added to the buffer. BCIP is a chromogenic probe for ALP activity, and dephosphorylation of BCIP result in formation of a visible blue precipitate of 5,5 ′ -dibromo-4,4 ′ -dichloro-indigo white. Already after 2 h incubation, a visible and distinct color difference could be seen between the samples with JR2KK-Biotin and the control containing JR2K (figures 6(b)-(c)). The JR2KK-Biotin hydrogels had turned teal, whereas the control remained colorless. The presence of JR2KK-Biotin was hence crucial for recruiting and sequestering ALP to the hydrogel. The role of ALP in biomineralization relies on the ability of the enzyme  to generate inorganic phosphate in the presence of Ca 2+ , for example by addition of calcium glycerylphosphate (CaGP) [60]. When CaGP was added to the ALP-loaded hydrogels, mineral depositions were clearly seen after 24 h incubation, and continued to accumulate over time (supporting figure S8(a)). In addition, microscope images further indicated that mineral-like structures were formed in the hydrogel (figure 6(d), supporting figure S8(c)). SEM images clearly showed deposition of mineral in the hydrogel with an element composition of Ca (20.6 ± 0.2 Wt%) and P (12.0 ± 0.1 Wt%) confirmed by EDS analysis. The ratio of Ca to P (Ca/P) corresponds to 1.7, which is close to the stoichiometric Ca/P ratio in hydroxyapatite (1.67) (figures 6(f)-(g)). No mineral deposition was seen in control hydrogels modified with JR2K and subsequent incubation with ALPstreptavidin (figure 6(e), supporting figure S8(b)), confirming that the specific peptide-mediated interaction between conjugated JR2EK and JR2KK-Biotin was required to trigger the ALP catalyzed biomineralization of the hydrogels. The stiffness of the hydrogels increased dramatically after biomineralization, resulting in a G´that was about one order of magnitude higher at 1 Hz than the non-mineralized hydrogel (figure 6(h), supporting figure S9). Biomineralization of hydrogels have previously been widely investigated for bone tissue engineering for repair of bone defects [61,62]. The possibilities to promote a high local concentration of ALP in the HA-BCN/JR2EK-p(N 3 ) 8 hydrogels using defined peptide-interactions represent a novel and efficient method for biomimetic mineralization.

3D and 4D bioprinting
Because of the tunable cross-linking kinetics of the HA-BCN/JR2EK-p(N 3 ) 8 hydrogels they can be processed into complex structures using additive biofabrication techniques. 3D printing of the hydrogels was conducted on a Cellink Bio X. using the freeform reversible embedding of suspended hydrogels (FRESH) technique [44]. To demonstrate the possibility to print elaborate architectures requiring high shape fidelity, a buckyball structure comprising 44 layers and with a smallest feature size of 0.6 mm was designed and printed (figures 7(a) and (b)). A thermoreversible gelatin slurry was used as a temporary support for the soft hydrogels during printing to improve fidelity. Printing was conducted at room temperature and the gelatin slurry was removed by heating to 37 • C. The HA-BCN/JR2EK-p(N 3 ) 8 hydrogels could be printed both in the absence and presence of Zn 2+ (supporting movie S2), although the latter resulted in significantly more robust structures due to the additional contribution from the supramolecular interactions to the cross-linking process (supporting movie S3 and movie S4). The subsequent removal of the Zn 2+ by addition of EDTA resulted in an instant swelling of the printed structures by more than 20% (supporting figures S10(a) and (b)), due to the reduced cross-linking density caused by the dissociation of the peptide homodimers. Peptide dimerization and dissociation can thus be used to change the morphology of the printed structures. Longer incubation time with Zn 2+ after printing resulted in reduced swelling after removal of the Zn 2+ , indicating that the supramolecular cross-linking facilitated the SPAAC-mediated covalent cross-linking of the hydrogels (supporting figures S10(c) and (d)). In addition, fibroblast encapsulated in the hydrogels showed high viability after printing (supporting figure S11).
To further investigate the possibility to modify the composition and function of the hydrogels after removal of the support hydrogel, the peptide JR2KK-Cy5 was added to the buffer and allowed to heterodimerize with the complementary and conjugated peptide JR2EK in the hydrogels. The specific association and gradual accumulation of dye-labeled JR2KK-Cy5 gave the printed hydrogel structures a teal color under ambient light and a bright red fluorescent emission when exciting the fluorophore (figures 7(a)-(c)). When instead introducing the biotin-labeled complementary peptide JR2KK-Biotin, the possibility to bind ALP-streptavidin added to the buffer resulted in a distinct biomineralization within the printed structures in the presence of CaGP (figures 7(d)-(f)). This possibility to modify the properties of biomaterials with explicit molecular and temporal control can facilitate fabrication of more complex biofunctional materials and pave the way for development of advanced 4D bioprinting strategies.

Conclusions
In conclusion, a polypeptide-functionalized hyaluronan (HA) and poly(ethylene glycol) (PEG) based hydrogel, HA-BCN/JR2EK-p(N 3 ) 8 , have been developed that can be cross-linked both covalently using a bioorthogonal SPAAC click reaction and by means of specific peptide interactions. The gelation due to covalent cross-linking occurred within minutes at 37 • C and resulted in cytocompatible hydrogels with a final G' within the range of many soft tissues (100 < G' < 500 Pa). The conjugated peptide (JR2EK) was designed to fold into a helix-loophelix motif and dimerize into four-helix bundles at acidic pH (< 6) or at neutral pH in the presence of Zn 2+ . Triggering peptide homodimerization resulted in supramolecular cross-linking and a dramatic increase in the initial gelation kinetics. Changing the degree of homodimerization could also be used for tuning the cross-linking density of the hydrogels postsynthesis. In addition, a complementary helix-loophelix peptide (JR2KK), capable of heterodimerizing with JR2EK, was synthesized with a set of different functionalities in the loop. This small library of peptides was used to investigate the possibilities to dynamically change the properties of the hydrogels with spatiotemporal resolution using the conjugated JR2EK to attract and anchor the complementary peptides. The peptides associated rapidly and remained bound to the hydrogels, which enabled fabrication of 3D gradients using Cy5-and Cy3-labled peptides (JR2KK-Cy5/Cy3), as well as functionalization with cell adhesion motifs using cyclic RGD modified peptides (JR2KK-cRGD). The latter enabled tuning of cell-hydrogel interactions and improved cell spreading and viability of neural cells cultured on the hydrogels. Incorporating a biotin moiety in the loop (JR2KK-Biotin) provided means to bind streptavidin-functionalized molecules to the hydrogels. Streptavidin-labeled alkaline phosphatase (ALP) was bound and enzymatically active after allowing JR2KK-Biotin to first associate in the hydrogel, resulting in biomineralization of the hydrogels in the presence of CaGP. Furthermore, the hydrogels could be 3D printed using the freeform reversible embedding of suspended hydrogels (FRESH) technique. The possibility to use both peptide homo-and heterodimerization to modulate the properties of the printed structures was demonstrated and show that this modular toolbox can be utilized to dynamically engineer the properties of complex soft material architectures. In addition, this strategy is not limited to the peptides used here. A large number of different well-defined de novo designed peptides with tunable self-assembly properties are available [63,64], which can provide an almost infinite number of possibilities to tailor materials properties. The possibility to modify the properties of 3D printed biomaterials with temporal control using highly defined molecular interactions can fascilitate development of versatile bioinks for 4D bioprinting that can open new opportunities to create responsive and dynamic tissue-mimetic constructs for a wide range of biomedical applications. Robert Selegård  https://orcid.org/0000-0002-1781-1489 Daniel Aili  https://orcid.org/0000-0002-7001-9415