Gelatin-Based Colloidal Versus Monolithic Gels to Regulate Macrophage-Mediated Inflammatory Response

Abstract

Unlike conventional monolithic hydrogels with covalent cross-linkage that are typically elastic, colloidal gels assembled by reversibly assembled particles as building blocks have shown fascinating viscoelastic properties. They follow a gel-sol transition upon yielding and recover to the initial state upon the release of the shear force (so-called shear-thinning and self-healing behavior); this makes them an ideal candidate as injectable and moldable biomaterials for tissue regeneration. The immune response provoked by the implantation of the colloidal gels with special viscoelastic and structural features is critical for the successful integration of the implants with the host tissues, which, however, remains little explored. Since macrophages are known as the primary immune cells in determining the inflammatory response against the implants, we herein investigated in vitro macrophage polarization and in vivo inflammatory response induced by gelatin-based colloidal gels as compared to monolithic gels. Specifically, self-healing colloidal gels composed of pure gelatin nanoparticles, or methacrylate gelatin (GelMA) nanoparticles to allow secondary covalent cross-linkage were compared with GelMA bulk hydrogels. We demonstrated that hydrogel's elasticity plays a more dominant role rather than the structural feature in determining in vitro macrophage polarization evidenced by the stiffer gels inducing pro-inflammation M2 macrophage phenotype as compared to soft gels. However, subcutaneous implantation revealed a significantly alleviated immune response characterized by less fibrous capsule formation for the colloidal gels as compared to bulk gels of similar matrix elasticity. We speculated this can be related to the improved permeability of the colloidal gels for cell penetration, thereby leading to less fibrosis. In general, this study provided in-depth insight into the biophysical regulator of hydrogel materials on macrophage behavior and related inflammatory response, which can further direct future implant design and predict biomaterial–host interactions for immunotherapy and regenerative medicine.

Impact statement

Macrophages response to implanted biomaterials is a highly regulated process that influences device functionality and clinical outcome. Nowadays, the viscoelastic properties of colloidal versus monolithic hydrogels on macrophage phenotype in vitro and the host inflammatory response are not known. Our study found that colloidal hydrogels composed of nanoparticles of gelatin and methacrylate gelatin (GelMA) led to more anti-inflammatory polarization especially on soft colloidal gel (5.9 KPa) compared to bulk GelMA hydrogels. It suggested that macrophage response can be mechanically regulated by the viscoelastic signals of the hydrogels, which could be a promising strategy for the future design and application of novel biomaterials.

Introduction

Inflammation plays an important role in the successful tissue integration of the implanted biomaterials and medical devices with the host tissues.^1,2 After implantation of a biomaterial, immune cells, including dendritic cells, mast cells, and neutrophils, immediately migrate to the implantation sites and subsequently trigger an inflammatory cascade.³ Macrophages play a critical role in determining the acceptance or rejection of the implants by the host through pro- or anti-inflammatory polarization of macrophages.⁴

Generally speaking, macrophages can adopt either a classically pro-inflammatory M1 or alternatively activated anti-inflammatory M2 phenotype depending on the complicated environmental cues, which can be divided into biochemical or biophysical signaling factors.^5,6 For instance, M1 and M2 polarization are induced by biochemical factors or interferon-γ or lipopolysaccharides (LPS), and interleukin (IL)-4 (IL-4) or IL-13 respectively.^7–9

Since macrophage polarization is related to many physiological functions and diseases, it is important to control this differentiation process.¹⁰ Specifically, during the wound healing process M1 macrophages begin to dominate the inflammatory response in the early stage (2–3 days) after injury, as characterized by a high level of secretion of tumor necrosis factor-α (TNF-α), IL-1β, and IL-6. Later, macrophages follow M2 polarization and start to dominate the second stage of wound healing, characterized by the expression of IL-10 and arginase. The interruption or disorder of these processes can lead to persistent inflammation and a compromised regenerative process and is usually accompanied by inefficient phenotypic switching or failure to spatiotemporally correct M1 polarization.¹¹ Promoting macrophage transition from M1 to M2 phenotype, thereby biasing toward an anti-inflammatory environment, could be a valid therapeutic approach to facilitate tissue regeneration.^12,13

Biophysical cues, including surface topography, physical confinement, and mechanical properties have shown a profound effect on cellular behavior.^14,15 Since Engler et al. reported the substantial effect of hydrogel elasticity as a vital regulator to steer stem cell differentiation mechanosensitivity of cells has attracted increasing attention.¹⁶ More recently, matrix viscoelasticity has been found to play a more profound role in influencing cell behavior. Specifically, Chaudhuri et al. found that a more viscoelastic hydrogel matrix characterized by faster stress-relaxation can induce an enhanced osteogenic differentiation of mesenchymal stem cells.¹⁷ Despite the recent progress in the mechanosensitivity of stem cells, studies on the effect of biochemical cues on immune cells like macrophages are less explored. Our recent study has shown that the stiffness of the hydrogel matrix can significantly affect macrophage polarization and in vivo inflammatory response.¹⁸ However, how the viscoelasticity of gel matrix can influence macrophage behavior remains to be unclear.

Colloidal gels are a novel class of hydrogel materials that enable a “bottom-up” strategy for the establishment of macroscopic scaffolding materials for regenerative medicine.^19–21 By employing micro- or nanometer-scale colloidal particles as building blocks to assemble into the integrated particulate network, colloidal gels are inherently porous and allow more efficient mass exchange than bulk hydrogels, which is important for biomedical applications.²² Upon introducing attractive interparticle interactions such as electrostatic, magnetic, or hydrophobic interactions, colloidal gels can be rendered with shear-thinning and self-healing behavior, controllable mechanical strength, inherent interconnected and porous structure.²³ These appealing properties make colloidal gels a promising candidate as injectable/moldable hydrogels for tissue engineering scaffolds or delivery carriers.^24,25

The previous studies have demonstrated that colloidal gels composed of gelatin nanoparticles or GelMA nanoparticles can be used as an extracellular matrix (ECM) for tissue regeneration, in which their mechanical properties can be finetuned to adapt to different applications.^26–28 Specifically, a recent study by Bertsch et al. showed that colloidal hydrogels can offer the possibility to modulate viscoelasticity characterized by tailorable stress relaxation rates, which serve as an artificial ECM to indicate cell migration, proliferation, and differentiation.²⁹ However, how this viscoelastic feature of colloidal gels in contrast to the bulk gels mechanically affects macrophage behavior and in vivo inflammatory response has been rarely explored.

Herein, we explored the effect of different hydrogel matrices with distinguished mechanical signals on macrophage behavior and inflammatory response. Specifically, colloidal gels of gelatin or GelMA nanoparticles (denoted as cG-gel or cG-gelMA) were compared with bulk GelMA (bG) hydrogels. All three groups of hydrogels were modulated to possess low or high matrix elasticity, but with different stress-relaxation behavior. Thereafter, macrophage cell line RAW264.7 was seeded on hydrogel substrates of different groups and in vitro morphology polarization was assessed. Further in vivo inflammatory response induced by hydrogel implants was investigated by a subcutaneous implantation model.

Materials and Methods

Preparation of gelatin-based hydrogels

Lyophilized GelMA, GelMA nanoparticles and pure gelatin nanoparticles were offered by Huanova Biotech (Shenzhen, China). Bulk GelMA hydrogels were prepared by dissolving 7% or 15% (w/v) GelMA in Dulbecco's phosphatic buffer solution (Gibco, USA), followed by ultraviolet (UV) irradiation (360 nm; Run LED, China) for 16 s; 0.5 w/v% 2-hydroxy-1-[4-(2-hydroxyethoxy)phenyl]-2-methyl-1-propanone (Irgacure 2959; Sigma-Aldrich, United Kingdom) was used as the photoinitiator. Colloidal gels were prepared by dispersing 7% or 15% (w/v) gelatin or GelMA nanoparticles in phosphate-buffered saline (PBS), followed by vigorous stirring to obtain homogeneous colloidal gels. GelMA colloidal gels were further cross-linked by photopolymerization as mentioned above. The abbreviations of each hydrogel group were shown in Table 1.

Table 1.

Abbreviations Used for Different Hydrogels

Description of the composition	Mass concentration (w/v%)
	7	15
Bulk hydrogel of GelMA with UV-triggered cross-linkage	bG-7	bG-15
Colloidal gel of GelMA nanoparticles with UV-triggered cross-linkage	cG-gelMA-7	cG-gelMA-15
Colloidal gel of gelatin nanoparticles	cG-gel-7	cG-gel-15

GelMA, methacrylate gelatin; UV, ultraviolet.

Compression measurements

The mechanical properties of the hydrogels were evaluated by conventional compression tests using a universal testing machine (E43; MTS Instrument, USA). Cylindrical samples (diameter: 12 mm, height: 8 mm) were prepared according to type 5B of ISO 527–2 standard. All mechanical tests were performed at a humidity of more than 60% to prevent the water evaporation of gels. The specimens were tested at a loading rate of 1 mm/min. The compressive modulus of each group was calculated from the linear region of the stress–strain curve (0–5% strain).

Rheological measurements

The viscoelastic properties of the hydrogels were determined with a discovery hybrid rheometer (TA Instruments, USA), and all tests were performed with a flat steel plate geometry (20 mm). Oscillatory frequency sweep measurements were performed at a frequency of 0.1–100 rad/s, and the storage modulus (G′) and loss modulus (G″) were determined by oscillatory shear deformation at a constant frequency (1 Hz) and constant shear strain (γ). All experiments were performed in triplicate at 25°C.

Stress-relaxation behavior of the hydrogels was characterized by monitoring the stress induced in hydrogels upon a step strain (10%) of the samples over time (0–1000 s).

Creep properties of samples were quantitatively characterized by applying constant shear stress of 20 Pa for 60 s (creep phase) and followed by recovery at stress = 0 Pa to release the shear force for 174 s (recovery phase). The remaining strain after the release of the stress in proportion to the initial deformation can be divided into the reversible (elastic and viscoelastic) and irreversible strains.

Cell culture

RAW264.7 cell line was purchased from ATCC and cultured in α-minimum essential medium (α-MEM) containing 10% fetal bovine serum (Gibco), 100 U/mL penicillin (Gibco), and 100 mg/mL streptomycin (Gibco) at 37°C under a 5% carbon dioxide (CO₂) atmosphere. The medium was changed every 2 days. Cells were digested with 0.25% trypsin solution (Gibco) for 3 min, followed by scraping the undigested cells from the plate with a scraper (Nest, China) and collected for future experiments.

In vitro biological evaluation

Cylinder-shaped samples of bG group with a diameter of 11 mm and a height of 3 mm, and the two groups of cG-gelMA and cG-gel with a diameter of 8 mm and a height of 2 mm were placed in a 48-well plate. Subsequently, a 50 μL macrophages suspension (cell density 5 × 10⁵ cells/mL) was seeded on the top of the hydrogel sample, followed by adding an extra 500 μL cell culture medium 2 h later. The medium was refreshed every other day and further biological evaluation was performed.

Scanning electron microscope

Hydrogel samples with macrophages cultured on the surfaces were fixed using glutaraldehyde solution (2.5%) at 4°C for 10 min. After dehydration in a series of gradients of ethanol (70%, 80%, 90%, 96%, 100% for 5 min, respectively) and CO₂ critical point drying, samples were sputter-coated with gold to improve the conductivity. The cell morphology on samples was observed by scanning electron microscope (SEM; FEI, Quant 450).

DNA content

dsDNA HS Assay Kit for Qubit (Yeasen, China) assay was used to evaluate macrophages DNA content following the manufacturer's instructions. The wavelength was determined on a microplate reader with excitation at 485 nm and emission at 530 nm (Bio-Rad iMark, USA). The concentration of DNA was calculated according to the standard curve established by different concentrations of DNA. The total DNA level on the sample surface was normalized by the surface area for cell attachment, that is, the average DNA content per unit area (1 mm²).

Live/dead assay

Live/dead test was performed to evaluate the cytocompatibility of the hydrogels. Macrophages were seeded on the surface of hydrogels for 1 and 3 days. Live and dead macrophages were labeled by calcein-AM (Invitrogen, Carlsbad, CA) and propidium iodide (PI; Invitrogen), respectively. The live/dead staining was visualized by a confocal laser scanning microscope (FV3000; Olympus, Japan).

Immunofluorescence staining

Macrophages were seeded on the surface of hydrogels for 3 days. Afterward, the cells were washed twice with PBS and fixed with 4% paraformaldehyde for 15 min. Then, macrophages were incubated with the primary antibodies rabbit anti-rabbit Arginase-1 (1:200, ab91279; Abcam, Cambridge,) and rabbit anti-rabbit inducible nitric oxide synthase (iNOS) (1:500, ab178945; Abcam) for 2 h at room temperature, followed by treatment with the secondary antibody Alexa Fluor 488 and 594 labeled goat anti-rabbit (Invitrogen) in 1% bovine serum albumin for 1 h at room temperature in the dark. Subsequently, the cells were incubated with 0.5 μg/mL 4′,6-diamino-2 phenylindole (DAPI; Sigma, USA) for 5 min. The staining was further observed with a confocal laser scanning microscope (FV3000, Olympus).

Enzyme-linked immunosorbent assay

Macrophages were seeded on the surface of hydrogels of different stiffness and then cell supernatants were collected after 3 days. The inflammatory cytokines TNF-α, IL-6, and IL-1β and anti-inflammatory transforming growth factor (TGF)-β and IL-10 were detected according to the enzyme-linked immunosorbent assay kit instructions (Invitrogen).

In vivo inflammatory response

Subcutaneous implantation

Six-week-old male BALB/c mice were obtained from Dalian Medical University and approved by the Biomedical and Animal Ethics Committee of the Dalian University of Technology (2021–097). Samples of different groups with a volume of 300 mm³ (diameter: 11 mm and height: 3 mm) cylinder-shaped specimens were implanted subcutaneously in mice back randomly. The implantation procedure followed previous reports.¹³ After 10 days, the mice were euthanized by cervical dislocation. The hydrogels samples together with the surrounding tissue were excised by cutting around the area with scissors and scalpels. Then, the specimens were fixed in 4% paraformaldehyde solution for 48 h and followed by histological and immunofluorescence analysis.

Histological and immunofluorescence analysis

Samples were hydrated, dewaxed with xylene and a series of ethanol with different concentrations, and embedded in paraffin wax. Each implant was sectioned at 4 μm and stained with hematoxylin and eosin (HE), which stains nuclei with dark purple/blue and stains cell cytoplasm and the ECM with pinkish. Subsequently, we assessed the fibrous capsule formation by measuring the thickness of the fibrous capsule surrounding the hydrogel implants. To further evaluate the cell penetration into the hydrogel matrix, we used a square-shape (1 × 1 mm²) in the histological images to indicate the region of interest (ROI). The location of the ROI was placed in the middle of the rectangle-shape of the tissue sections, and the edge of the square was overlapping with the tissue-biomaterial interfaces. By further calculating the numbers of infiltrated cells in the ROI, we can obtain a semi-quantitative evaluation of the permeability of different hydrogels.

For immunofluorescence staining, tissue sections were boiled in citric acid buffer (pH = 6.0, 100 °C) to repair antigens. Then, anti-CD68 rabbit antibody (1:200, GB113109; Servicebio, China), rabbit anti-rabbit arginase antibody Arg-1 (1:200, ab91279; Abcam), and rabbit anti-rabbit iNOS antibody (1:500, ab115819; Abcam) were incubated overnight at 4 °C, and the second antibody AlexaFluor 488 and 594 labeled goat anti-rabbit (Invitrogen) were incubated at room temperature for 1 h.

Statistical analysis

The statistical analysis was performed by GraphPad Prism 5. Univariate analysis of variance followed by Tukey back testing was used to analyze the data, which was expressed as mean ± standard deviation. *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001 are considered statistically significant.

Results

Mechanical properties of hydrogels of different compositions

We characterized the mechanical properties of the hydrogels of different compositions using conventional compression tests and rheology. Before cross-linking, 7 and 15 w/v% hydrogels were prepared to obtain a rather low and high matrix elasticity. All samples showed a rather linear response without showing obvious yield points upon compressive strain up to 40%. Specifically, 7 w/v% hydrogels of bG, cG-gel and cG-gelMA groups showed compressive modulus E of 4.3 ± 1.42, 5.4 ± 0.74, and 7.9 ± 2.06 kPa, respectively; the elasticity of all hydrogels in this concentration can be considered as rather a soft matrix. In comparison, 15 w/v% hydrogels of bG, cG-gel and cG-gelMA groups showed compressive modulus E of 33.3 ± 4.59, 46.09 ± 1.72, and 48.84 ± 6.59 kPa, respectively; all samples at this concentration are rather stiffer (Fig. 1A, B).

FIG. 1.

Characterization of the mechanical properties of different hydrogels. (A) Compressive stress–strain curves and (B) compression modulus of bG, cG-gel and cG-gelMA with different concentrations (7% and 15% (w/v), n = 4). (C) Frequency dependence of storage (solid symbol, G′) and loss (open symbol, G″) modulus of bG, cG-gel and cG-gelMA with different concentrations under 0.5% strain (frequency 1 Hz, n = 4). (D) The corresponding storage modulus of gels based on frequency sweep tests. (E) Oscillatory time sweep tests to show the triggered photo-polymerization of cG-gelMA-7 and cG-gelMA-15 upon UV irradiation (0.5% strain and 1 Hz frequency), as reflected by the sharp increase of G′’ and G″ values. (F) The viscoelastic properties measured by stress-relaxation of bG, cG-gel and cG-gelMA hydrogels with different concentrations at a constant 10% stain (n = 4). (G) Quantification of timescale at which the stress is relaxed to 75% of its original value (denoted as τ0.75) from stress relaxation tests. (H) Representative creep tests of bG, cG-gel and cG-gelMA hydrogels with different concentrations. Deformation (% strain) was recorded upon applying an external shear force of 20 Pa (60 s), followed by the release of this shear force (174 s; n = 4). (I) Quantification of the creep recovery rate from the creep tests. *p < 0.05 and ***p < 0.001 determined using one-way ANOVA test. ANOVA, analysis of variance; bG, bulk GelMA; cG-gel, colloidal gel of gelatin nanoparticles; cG-gelMA, colloidal gel of GelMA nanoparticles; GelMA, methacrylate gelatin; UV, ultraviolet.

We further performed oscillatory frequency sweep tests for the hydrogels to measure the storage modulus G′ and loss modulus G″ as a function of shearing frequency (Fig. 1C). All hydrogels of different compositions of different concentrations showed typical frequency-independent behavior with the G′ value remaining rather stable regardless of the frequency. Moreover, 15 w/v% hydrogels of bG, cG-gel and cG-gelMA groups showed storage modulus G′ of 10.01 ± 4.06, 10.62 ± 6.06, and 11.38 ± 10.07 kPa, respectively (Fig. 1D). All groups showed no significant difference. A similar trend was also observed in lower concentration samples. Also, for hydrogel containing GelMA polymer or nanoparticles, the G′ values significantly increased upon UV light irradiation as evidenced by a sharp increase of gel modulus (Fig. 1E).

These data suggested that gelatin gels of the same concentration typically showed comparable matrix elasticity. Interestingly, 15 w/v% gelatin-based colloidal gels showed significantly higher E values than that of bulk GelMA hydrogels, indicating the densely packed particulate network was more elastic than the bulk polymer gels. It can be noticed that GelMA colloidal gel was expected to show higher gel elasticity as compared to pure gelatin colloidal gels due to the additional interparticle cross-linkage by the free-radical polymerization process. However, both gels showed no significant difference in network elasticity, which can be attributed to the influence on the properties of the formed nanoparticles after the methacrylate process of gelatin.

We further assessed the viscoelasticity of the hydrogels of different compositions by rheological measurements of stress relaxation and creep tests. Stress-relaxation behavior at a constant 10% strain was recorded, which showed different relaxation rates for the different groups (Fig. 1F). We quantified the relaxation rate by measuring the timescale at which the stress was relaxed to 75% of its original value, defined as τ_0.75. bG showed a slower stress-relaxation upon applying a 10% strain, which was significantly lower than that of cG-gel and cG-gelMA. Specifically, 7 w/v% hydrogels of bG, cG-gelMA and cG-gel groups showed τ_0.75 = 122.37 ± 2.37, 41.21 ± 1.21, 35.46 ± 1.46 s, respectively; 15 w/v% hydrogels of bG, cG-gelMA and cG-gel groups showed τ_0.75 = 192.49 ± 2.49, 7.08 ± 1.08, and 14.15 ± 1.15 s, respectively (Fig. 1G).

We also performed creep tests to investigate the hydrogel response upon applying the same degree of external stress. It was revealed that all hydrogels were almost instantaneously deformed up to a relatively stable value upon shearing (20 Pa), and the deformation remained constant until the release of the external force. Upon the removal of the shear force, instantaneous recovery to a constant strain was observed (Fig. 1H). In general, 7 w/v% hydrogels exhibited larger deformation than 15 w/v% ones. At the same concentration, cG-gelMA gels showed less deformation compared to cG-gel and bG hydrogels. By further quantification of the recovery of the initial strain, colloidal gels presented a high degree of recovery above 90% regardless of the gelatin concentrations, when compared to the bulk GelMA gels with a rather low recovery efficiency below 90% (Fig. 1I).

Two-dimensional culture of macrophages on hydrogel matrix

We seeded RAW264.7 cells on the surface of the hydrogel surfaces. All hydrogels showed favorable biocompatibility as reflected by the majority of cells remaining alive after in vitro culture with the materials. Interestingly, macrophages presented different degrees of cell attachment as shown by the live/dead staining. The fluorescent microscopic images of macrophages on different hydrogels revealed that bulk GelMA hydrogels were less favorable for macrophage attachment after 72 h as compared to the gelatin-based colloidal gels (Supplementary Fig. S1, supporting information). DNA content assay for the attached cells on the gel surfaces further confirmed that bulk GelMA hydrogels were not favored for macrophage attachment (Fig. 2A). This might be related to the enhanced specific surface area and porosity of the colloidal gels as compared to the much flatter surface of bulk gels.

FIG. 2.

Characterization of behavior of RAW264.7 cells on different hydrogels. (A) The DNA content of macrophages on hydrogel samples of bG, cG-gel and cG-gelMA groups of different concentrations after culturing for 3 days (n = 4). (B) Representative micrograph morphology showing the expression of the F-actin of macrophages on different hydrogels. Phalloidin (green) and DAPI (blue) were used to label F-actin and nuclei (scale bar = 10 μm, n = 4). (C) The area of the fluorescence area of F-actin was quantified by Image J Software. (D) Quantification of the depth of macrophages penetration into different hydrogels after culturing for 5 days. *p < 0.05, **p < 0.01, and ***p < 0.001 determined using one-way ANOVA test. DAPI, 4′,6-diamino-2 phenylindole.

Since previous studies have shown that matrix elasticity can affect the morphology of the attached macrophages,^18,30 we further performed F-actin staining to evaluate the morphological feature of the attached macrophages on the gels. Noticeably, we found that macrophages displayed a spreading morphology on all hydrogels, which can be attributed to the presence of cell-attachment motifs (e.g., arginine-glycine-aspartate [RGD] sequence) in gelatin molecules (Fig. 2B). We further quantitatively evaluated the cell morphology by analyzing the actin microfilaments spreading area and found that for the same hydrogel composition. Macrophages spread more on stiffer hydrogels, which was in line with previous observations that stiffer matrix favored macrophage spreading.^30,31 Moreover, the cell spreading area for macrophages on different hydrogel surfaces followed the order of cG-gelMA > cG-gel > bG (Fig. 2C), especially for gels of 15 w/v%. This might be related to the higher elastic modulus of cG-gelMA and cG-gel colloidal gels than that of bulk bG gels.

We also observed that different hydrogels showed various degrees of macrophage penetration. Macrophages cultured on different gels for 5 days were visualized by the confocal fluorescent microscope and observed that cells on colloidal gels were substantially more intended to penetrate deeper into the matrix in comparison to bulk GelMA hydrogels (Supplementary Fig. S2). Additionally, for the same hydrogel composition, cells penetrated more in the softer gel matrix. Further quantification of the depth of the penetrated cells into the gel matrix confirmed this observation (Fig. 2D).

The polarization of macrophages on different hydrogels

The polarization of macrophages cultured on different hydrogels was determined by the typical cluster of differentiation (CD) marker of iNOS for M1 and Arg-1 for the M2 phenotype. iNOS and Arg-1-positive macrophages seeded on different hydrogels were quantitatively analyzed. In general, softer hydrogels with G′ ∼5.9 kPa induced substantially fewer iNOS but significantly higher Arg-1 expression as compared to more elastic gels with G′ ∼42.7 kPa, indicating less inflammatory differentiation of macrophages on the soft matrix (Fig. 3A). With a similar gel elasticity range, the expression of iNOS and Arg-1 was rather controversial. For gels of 7 w/v% concentration, the cG-gelMA-7 group showed a significantly higher level of iNOS but less Arg-1 expression than that of the bG-7 and cG-gel-7 groups, suggesting a stronger pro-inflammation effect. Additionally, for hydrogels of 15 w/v% concentration, colloidal gels showed stronger pro-inflammatory expression than the bulk GelMA gels (Fig. 3B, C).

FIG. 3.

CD marker expression of macrophages on different hydrogels. (A) Immunofluorescent staining of the pro-inflammatory enzyme iNOS (red) and anti-inflammatory receptor Arg-1 (green) for macrophages on different hydrogels. The nuclei were stained blue (DAPI, scale bar = 20 μm; n = 4). (B, C) Quantification of the percentage of iNOS and Arg-1 in immunofluorescence images. *p < 0.05, **p < 0.01, and ***p < 0.001 determined using one-way ANOVA test. CD, cluster of differentiation; iNOS, inducible nitric oxide synthase.

We further evaluated the macrophage polarization on different hydrogels by detecting the secretion of different cytokines in the culture medium. TNF-α, IL-6, and IL-1β are representative pro-inflammatory cytokines for M1 markers, while TGF-β and IL-10 are anti-inflammatory M2 markers. Overall, the results showed that after 3 days of culture, the secretion of pro-inflammatory cytokines of macrophages on hydrogels with a concentration of 15 w/v% was significantly higher than that of 7 w/v% hydrogels (Fig. 4A–C), which might be related to the substantially different matrix elasticity between two concentrations. Specifically, at high hydrogel concentration, the secretion of pro-inflammatory cytokines including TNF-α, IL-6, and IL-1β showed the same trend of bG > cG-gelMA > cG-gel. Hence, these cytokines expression was more disordered for low hydrogel concentration, with cG-gelMA and bG groups presenting stronger pro-inflammatory responses than that of the cG-gel group.

FIG. 4.

The secretion of pro- and anti-inflammatory cytokines by macrophages on different hydrogels. The secretion of M1 macrophages-related proteins (A) TNF-α, (B) IL-6, (C) IL-1β and M2 macrophages-related proteins, (D) TGF-β, and (E) IL-10 were detected by ELISA (n = 4). *p < 0.05, **p < 0.01, and ***p < 0.001 determined using one-way ANOVA test. ELISA, enzyme-linked immunosorbent assay; IL, interleukin; TGF-β, transforming growth factor-β; TNF-α, tumor necrosis factor-α.

Further quantification of anti-inflammatory cytokines TGF-β and IL-10 showed that 7 w/v% hydrogels induced significantly higher pro-regeneration cytokines than 15 w/v% gels (Fig. 4D, E). Specifically, the secretion of TGF-β of macrophages on cG-gel-7 was significantly higher than that on cG-gelMA-7 and bG-7 (Fig. 4D). In addition, the secretion of IL-10 on cG-gelMA-15 was significantly higher than that on cG-gel-15. In addition, the IL-10 secretion of macrophages on bG-7 was significantly higher than that on cG-gelMA-7 and cG-gel-7 (Fig. 4E). These results suggested that the more obvious difference in matrix elasticity between 7 and 15 w/v% hydrogels seemed to play a more dominant role in determining macrophage polarization other than the slender difference in matrix elasticity and network morphology between gels at the same concentration.

In vivo inflammation responses and macrophage phenotypes for different hydrogels

The inflammatory response after subcutaneous implantation of different hydrogels was observed after 10 days of implantation. HE staining was performed to detect cell infiltration and fibrous capsule formation surrounding the hydrogel implants (Fig. 5A). Generally, 7 w/v% gels led to significantly thinner fibrous tissue formation when compared to 15 w/v% gels that showed substantially higher network elasticity. For 7 w/v% gels, the thickness of fibrous tissues was 43.7 ± 0.99, 20.5 ± 5.15, and 21.8 ± 0.56 μm respectively for bG, cG-gel and cG-gelMA groups. For 15 w/v% gels, the thickness of fibrous tissues for bG, cG-gel and cG-gelMA groups was 117.2 ± 8.17, 61.2 ± 4.86, and 51.7 ± 9.04 μm, respectively. Moreover, for gels of similar matrix elasticity, bulk GelMA gels resulted in a stronger inflammatory response evidenced by thicker fibrous capsule formation than the colloidal gels (Fig. 5B).

FIG. 5.

Inflammation responses for different hydrogels after subcutaneous implantation in mice. (A) Representative HE staining images of bG, cG-gel and cG-gelMA hydrogels after being implanted subcutaneously in mice for 10 days (scale bars = 100 and 50 μm, black double arrow: the depth of fibrous tissues, black square-shape: the range analysis of the density of cells infiltrating bG, cG-gel and cG-gelMA with different stiffness; n = 6). (B) Quantitative analysis of the thickness of the formed fibrous capsules surrounding the implants. (C) Semi-quantitative evaluation of the infiltrated cells into different hydrogels after 10 days implantation. A square (1 × 1 mm²) to highlight the ROI was placed in the middle of the rectangle-shaped tissue section, with the edge of the square overlapping with the tissue-biomaterial interfaces. By further calculating the numbers of infiltrated cells in the ROI, the evaluation of cell penetration and the permeability of the hydrogels can be obtained. *p < 0.05, **p < 0.01, and ***p < 0.001 determined using one-way ANOVA test. HE, hematoxylin and eosin; ROI, region of interest.

Noticeably, different hydrogels also exhibited various degrees of cell ingrowth into the gel matrix. Due to the inherent particulate feature of colloidal gels, we observed substantially more cell penetration into the inner part of the hydrogel implants as compared to the bulk gels. We also quantified the number of cells infiltrated into the implanted hydrogels by specifically defining a ROI followed by calculating the cells inside. Interestingly, we observed the degree of cell penetration into different hydrogels followed the order of cG-gel > cG-gelMA > bG, regardless of the hydrogel concentration (Fig. 5C). This suggested that the network morphology seemed to play a more dominant role in determining cell penetration than the matrix elasticity.

We also performed extra CD68 staining to verify macrophages within the infiltrated cells (Supplementary Fig. S3A). The results showed the percentage of infiltrated macrophages was related to hydrogel composition but also the stiffness values of the gels. We observed that the implantation of bG-15 gels led to more macrophage presentation as evidenced by 84% macrophages among all infiltrated cells (Supplementary Fig. S3B). Hence, most of these macrophages were located surrounding the bulk gels. As for the colloidal gels, more viscoelastic cG-gel gels characterized by faster stress-relaxation induced less macrophage presentation when compared to cG-gelMA gels. Moreover, regardless of hydrogel compositions, stiff gels all provoked more macrophage presentation than soft ones.

We further determined the phenotype of the infiltrated macrophages within the gel implants, and M1 and M2 macrophages were distinguished by iNOS and Arg-1 immunofluorescence staining. M1 macrophages were more abundantly distributed on stiff colloidal gels, while M2 macrophages were found more abundantly on soft colloidal gels (Fig. 6A). We further normalized the M1 and M2 macrophages to the total number of infiltrating cells (DAPI staining). There is no significant difference in the number of infiltrated M1 macrophages in stiff bG, cG-gel and cG-gelMA matrices, but the number of M1 macrophages in bG-15 (73.9 ± 7.5%) is significantly higher than that in bG-7 (36.4 ± 5.6%) (Fig. 6B). Correspondingly, the number of M2 macrophages for gels of the comparable G′ values was controversial with M2 polarization less pronounced for cG-gel gels at the G′ ranging at 4–8 kPa but more dominated at G′ ∼33–49 kPa (Fig. 6C).

FIG. 6.

Macrophage phenotypes reflecting the inflammation response for different hydrogels. (A) Representative immunofluorescence images of iNOS positive macrophages (green) and Arg-1 positive macrophages (red) in bG, cG-gel and cG-gelM A hydrogels. (B, C) Quantitative analysis of the number of iNOS and Arg-1 positive cells as relevant to the number of DAPI-stained cells (scale bars = 100 20 μm, n = 6). *p < 0.05, **p < 0.01, and ***p < 0.001 determined using one-way ANOVA test.

Discussion

Macrophages are key regulators of the foreign body reaction and play a vital role in the successful tissue integration with the implanted biomaterials,³² which in turn affect macrophage's behavior and functionality via providing biophysical and biochemical signals.^33,34 Specifically, the mechanosensitivity of macrophages has recently attracted increasing attention that has been shown to regulate macrophage polarization and further induced different degrees of inflammation responses in vivo.³⁵

We recently showed that bulk GelMA hydrogels of different stiffness values can mechanically affect macrophage phenotype and inflammatory response.¹⁸ These covalently cross-linked hydrogels are rather elastic in mechanical behavior and possess nanometer-sized porosity in their network, which, however, displayed poor resemblance to natural ECM that are typical of viscoelasticity and enhanced porosity. Especially, Lee et al. recently demonstrated that stress relaxation, a typical characteristic mechanical behavior of matrix viscoelasticity, can significantly promote osteogenic differentiation of mesenchymal stem cells as compared to hydrogels with slower stress relaxation.³⁶ This indicated that matrix mechanics has a profound influence on cell fate, which should be taken into account for the design of biomaterials.

We herein studied the mechanical properties of different hydrogels. Particularly, colloidal gels, which are typically composed of interconnected particulate networks dispersed in a continuous aqueous phase.³⁷ Due to the structural feature and interparticle potentials in colloidal gels, they can present viscoelastic behavior including hear-thinning or self-healing, thereby rendering them with injectability, moldability, and adaptability to local complexity.^38,39 We prepared colloidal gels composed of gelatin or GelMA nanoparticles with different stiffness values simply by adjusting gelatin concentration at 7 or 15% (w/v). Compressive tests and rheological measurements showed that despite containing covalent cross-linkages in GelMA colloidal gels, both colloidal gels exhibited comparable matrix elasticity evidenced by similar compressive moduli and storage moduli (Fig. 1A, C). It can be speculated that GelMA colloidal gels possessed stronger interparticle potential but lower volume fraction resulting from less swelling ratio of GelMA nanoparticles as compared to pure gelatin ones; this collectively led to comparable overall gel elasticity of these two types of colloidal gels.

Further stress-relaxation tests revealed that unlike covalently cross-linked bulk GelMA gels, which showed a slower stress-relaxation rate, cG-gel and cG-gelMA colloidal gels showed much faster stress-relaxation indicating they were more viscoelastic than bulk GelMA gels (Fig. 1F, G). Additionally, creep tests also demonstrated more viscoelastic features for cG-gel and cG-gelMA colloidal gels than bG bulk gels (Fig. 1H, I).

Based on the above mechanical properties of the colloidal and monolithic hydrogels, we investigated the adhesion and proliferation behavior of macrophages to the matrix. We found that monolithic GelMA hydrogels support less macrophage attachment compared with colloidal gels (Supplementary Fig. S1 and Fig. 2A), which we attributed to the significantly higher specific surface area of colloidal gels than the bulk gel for cell attachment. Further SEM images showed macrophages were attached and spread on hydrogel surfaces (Supplementary Fig. S4). Moreover, microscopic images showed that macrophages adopted an adhesive spreading morphology on all samples (from 5.9 to 42.7 kPa), and the average cell spreading area becomes significantly larger with the increasing stiffness (Fig. 2B).

Further evaluation of secreted cytokines for M1 and M2 macrophage markers demonstrated that increasing stiffness enhanced iNOS expression and decreased Arg-1 expression, indicating a more pro-inflammatory effect for a stiffer matrix. This was further confirmed by an increased TNF-α and IL-6 secretion of macrophages and decreased TGF-β and IL-10 on stiff gels (Fig. 4). These findings were consistent with a previous study that reported that increased substrate stiffness promoted macrophage extension and activation of M1-like macrophages.⁴⁰ In addition, we also found that the pro-inflammatory factors secreted by macrophages on bG were significantly more than that on cG-gel and cG-gelMA colloidal gels, which indicated that bG bulk gels could stimulate more inflammatory responses of macrophages.

Further in vivo study by subcutaneous implantation of the hydrogels in mice that compared with stiffer hydrogels induced a significantly large number of presented inflammatory-related cells and fibrous tissue formation than that of soft gels. In addition, it was noticed that bG bulk gels provoked significantly thicker fibrous capsules, which was mainly attributed to a large number of macrophage infiltration (Supplementary Fig. S3).

Given the reported different functions of M1 and M2 macrophages, we further invested macrophage phenotypes in different hydrogels by histoimmunofluorescent staining (Fig. 6A). We found that infiltrated macrophages were mainly M2 type on soft colloidal gels, evidenced by a higher percentage of Arg-1 positive cells. On the contrary, as the stiffness of gels increased, M1-type macrophages mainly infiltrated into the gel. Despite the special structural feature of colloidal gels that can provoke cell penetration and alleviate fibrosis, the matrix stiffness still had a more dominant effect on macrophage polarization and the corresponding inflammatory response in vivo. For future tissue engineering applications, a severe inflammatory response upon biomaterial implantation can influence the cells within the materials and compromise the integration between tissues and the implants.

Thus, this study provides a potential new strategy for developing hydrogel-based biomaterials that can regulate immune response, and subsequently promote tissue regeneration. In addition, we compared gelatin-based colloidal gels with bulk GelMA hydrogels, which are one of the most widely used biomaterials for tissue engineering, and delivery of biomolecules and live cells. As we observed in this study, compared to bulk GelMA gels, macrophages showed more attachment but less expression of inflammatory factors on cG-gel and cG-gelMA colloidal gels of similar matrix stiffness. All these findings suggest the superior properties of colloidal gels as relevant to the classic GelMA hydrogels, which makes these colloidal gels ideal candidates for application in regenerative medicine.

Conclusion

We herein investigated the macrophage-mediated inflammatory response induced by gelatin-based colloidal versus monolithic hydrogels. We demonstrated that the overall matrix elasticity plays a more dominant role in determining in vitro macrophage polarization with soft matrix (E: 4–8 kPa) more favored by anti-inflammatory M2 polarization and stiff matrix (E: 33–49 kPa) more likely to induce pro-inflammatory M1 polarization. More importantly, we found a significantly alleviated immune response characterized by less fibrous capsule formation for the colloidal gels when compared to bulk gels of similar matrix elasticity, which can be attributed to the enhanced permeability for cells into the colloidal gels. In general, this study provided in-depth insight into the biophysical regulator of hydrogels on macrophage behavior and related inflammatory response, which should be taken into account for the design of future implants and therapeutic strategies for immunotherapy and tissue engineering.

Authors' Contributions

Z.Z. and S.S. contributed equally to the study. Z.Z., S.S., and H.W. designed the study, analyzed the results, and wrote the article. S.S. and Z.Z. performed the experiments. All authors commented on the article.

Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported by the National Key Research and Development Program of China (Grant No. 2018YFA0703000), National Natural Science Foundation of China (Grant Nos. 31870957, 31900966), Fundamental Research Funds for the Central Universities of China (Grant No. DUT15RC [3]113), and Shenzhen Basic Research Program general project (Grant Nos. JCYJ20190808152211686 and JCYJ20190808120217133).

Supplementary Material

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Figure S4