Sequential modes of crosslinking tune viscoelasticity of cell-instructive hydrogels

Abstract

Materials that can mimic the fibrillar architecture of native extracellular matrix (ECM) while allowing for independent regulation of viscoelastic properties may serve as ideal, artificial ECM (aECM) to regulate cell functions. Here we describe an interpenetrating network of click-functionalized alginate, crosslinked with a combination of ionic and covalent crosslinking, and fibrillar collagen type I. Varying the mode and magnitude of crosslinking enables tunable stiffness and viscoelasticity, while altering neither the hydrogel's microscale architecture nor diffusional transport of molecules with molecular weight relevant to typical nutrients. Further, appropriately timing sequential ionic and covalent crosslinking permits self-assembly of collagen into fibrillar structures within the network. Culture of human mesenchymal stem cells (MSCs) in this mechanically-tunable ECM system revealed that MSC expression of immunomodulatory markers is differentially impacted by the viscoelasticity and stiffness of the matrix. Together, these results describe and validate a novel material system for investigating how viscoelastic mechanical properties of ECM regulate cellular behavior.

Figure 1.

Artificial extracellular matrix (aECM) hydrogel system features hierarchical and tunable covalent and ionic crosslinking. a) Collagen type I protein (green) is mixed with calcium carbonate nanoparticles (red) to obtain a homogenous suspension at 4° C. b) Next, an alginate polymer solution a nd cells are added to the mixture while maintaining a neutral pH and 4° C. c) After the addition of weak acid glucono-delta-lactone, the gel mixtures are rapidly cast and form hydrogels at 37° C. Alginate (blue) and self-assembled fibers of collagen type I form an interpenetrating network (IPN) hydrogel. Crosslinking domains of alginate (dotted circle) contain both covalent and ionic bonds (inset). d–e) Sequential ionic and covalent crosslinking of alginate permits collagen type I self-assembly into a network. d) Collagen self-assembles with a representative time scale (t_Collagen). e) In parallel, alginate reversibly crosslinks within blocks of glucuronic acid residues (inset) following addition of divalent cations, such as calcium, (red circles), with a representative reaction time (t_Ionic). Following initial alginate gelation and collagen self-assembly, click chemistry groups norbornene (Nb, purple) and tetrazine (Tz, light blue) provide secondary covalent crosslinks within the G-blocks over an addition time (Δt_NbTz) and release N₂ gas.

Figure 2.

Sequential ionic and covalent crosslinking tune the viscoelasticity of aECM. a–d) Oscillatory shear rheology (1 Hz, 1% strain, 25° C) with a cone -plate geometry was used to determine the timescales of transition from a more fluid phase (loss angle > 45°) to solid-like gel phase (loss angle < 45°) of alginate hydrogels with either ionic or covalent crosslinking. Storage modulus, G’ (solid line), loss modulus, G” (dotted line), loss angle (open circle). a–b) Unmodified alginate with only ionic crosslinking, Alg (blue, 1.5% w/v total alginate, 4 mg/mL collagen type I) at low and high crosslinking densities (0.15% and 0.3% w/v calcium carbonate). c–d) A combination of alginate polymers functionalized with either norbornene (Nb) or tetrazine (Tz) form hydrogels upon mixing via Nb-Tz covalent crosslinking, NbTz (red, 1.5% w/v total alginate, 4 mg/mL collagen type I) at low and high crosslinking density (low - 1.0% w/v NbTz, 0.25 N:T ratio; high – 1.5% w/v NbTz, 1 N:T ratio). e–f) Frequency sweeps after equilibrium of aECM hydrogels formed with either only ionic (Alg, blue) or sequential ionic and covalent crosslinking (NbTz, red) at low and high crosslinking densities (low – 0.15% Ca, 1.0% NbTz, 0.25 N:T ratio; high – 0.3% Ca, 1.5% NbTz, 1 N:T ratio). Storage modulus, G’ (solid line), loss modulus, G” (dotted line), loss angle (open circle). Graphs show representative data from rheological measurements.

Figure 3.

Macromolecular diffusion of bovine serum albumin (BSA) and interpenetration of aECM hydrogels are uniform across Alg and NbTz hydrogels. a) Cumulative release of fluorescently-labeled BSA from Alg and NbTz hydrogels at low and high crosslinking densities. Low – 0.15% Ca, 1% NbTz 0.25 N:T ratio; High – 0.35% Ca, 1.5% NbTz 1 N:T ratio. b) The diffusion coefficient (D) of BSA, calculated from release curves, for various hydrogels. Error bars represent the 95% confidence interval (CI) of D for each condition. c) Confocal fluorescent imaging of rhodamine (TAMRA)-modified Alg (left) and anti-collagen I antibody staining of Alg (right). d) Mean distributions of fluorescence intensities for TAMRA-Alg hydrogels, n = 3. e) Confocal fluorescent imaging and f) mean distribution of FastGreen staining of collagen protein in NbTz hydrogels. Images shown are representative of across all conditions. Scale bar 100 μm.

Figure 4.

Self-assembly of collagen fibrillar architecture in click-crosslinked aECM depends on rate of collagen and norbornene (Nb) and tetrazine (Tz) crosslinking. a) Gelation kinetics of collagen type I (4 mg/mL) crosslinking at 37° C with telopeptides (Tel o-Col I, black) or without telopeptides (Atelo-Col I, gray) with 1.5% w/v alginate and 0% w/v calcium carbonate. Storage modulus, G’ (solid line), loss modulus, G” (dotted line), loss angle (open circle). b) Gelation kinetics of NbTz hydrogels can be controlled by their composition. Shear moduli (left) and loss angle (right) with only click alginate crosslinking (1.5% w/v total alginate, 4 mg/mL collagen type I, 0% w/v calcium carbonate) of 1.5% w/v NbTz (red, 1 N:T) and 1.0% w/v NbTz (purple, 1 N:T). Storage modulus, G’ (solid line), loss modulus, G” (dotted line), loss angle (open circle). c) Frequency sweeps after equilibrium of 1.0% NbTz aECM hydrogels with high calcium (0.3%) and high covalent crosslinking (1 N:T ratio) at 1% strain (1.5% total alginate, 4 mg/mL telo-collagen type I). Storage modulus, G’ (solid line), loss modulus, G” (dotted line), loss angle (open circle). Graphs show representative data from rheological measurements. d) Second harmonic generation (SHG) confocal imaging of fibrillar collagen in Alg (blue), 1.5% NbTz (red), and 1.0% NbTz (purple) with high crosslinking (0.3% Ca, 1 N:T ratio), 4 mg/mL telo-collagen type I, and 1.5% w/v total alginate. 40x objective, maximum intensity projection of 15 µm z-stack, scale bar 25 µm.

Figure 5.

Viscoelastic properties of aECM regulate the immunomodulatory phenotype of bone marrow-derived MSCs. a–b) Shear moduli (a) and loss angle (b) of aECM with various compositions. a) G’ storage modulus (black), G” loss modulus (gray), no statistically significant difference between Alg and NbTz groups, two-way ANOVA with Sidak’s multiple comparison test, p>0.05. b) * statistically significant difference between Alg and NbTz, student’s T test, unpaired, p<0.05, n > 3. c–e) Relative gene expression of cyclooxygenase-2 (PTGS2), TNFα stimulated gene 6 (TNFAIP6), and interleukin-1 receptor antagonist (IL1RN) by human primary MSCs after 72 hours, normalized to housekeeping gene GAPDH and cells adhered on tissue culture plastic (TCP) control. aECM – MSCs were encapsulated in Alg (blue) or NbTz (red) of various moduli and loss angles. * statistically significant difference, two-way ANOVA, Sidak’s post hoc test, p<0.05, n ≥ 3. Controls – MSCs adhered to TCP were cultured with empty ionically crosslinked Alg (Ca) to determine the effects of freely soluble calcium in these gels on gene expression. 8h pulse – MSCs adhered to TCP were pulse stimulated with TNFα and IFNγ (+, 10 ng/mL) or vehicle control (−) for 8 hours. ND, not detected. * statistically significant difference, unpaired T-test, two-tailed, p<0.05, n = 4. f) Prostaglandin E2 (PGE2) in conditioned media measured by ELISA after 72 hours in Alg (blue) and NbTz (red) hydrogels of various moduli and loss angles. * statistically significant difference, two-way ANOVA, Sidak’s post hoc test, p<0.05, n ≥ 3.