Peptides and Astroglia Improve the Regenerative Capacity of Alginate Gels in the Injured Spinal Cord

Fabrication of alginate-based capillary hydrogels

Anisotropic alginate-based hydrogels were manufactured as previously described with slight modifications.^5–7,37 Briefly, sodium alginate (Pronova UP MVG; FMC Biopolymer AS d/b/a NovaMatrix, Sandvika, Norway) with a guluronic acid content of 70% and dynamic viscosity of 211 mPa · s (10 g/L, 20°C) was dissolved in purified deionized water (dH₂O) at a concentration of 20 g/L and overlaid with 20 mL of an aqueous solution of 1 M Sr(NO₃)₂. Following gelation at 10°C for at least 36 h, hydrogel blocks were washed in dH₂O, chemically stabilized with the crosslinker hexamethylene di-isocyanate, immersed in acidic solution to remove complexed Sr²⁺ ions, and finally neutralized in dH₂O. In vitro stability was tested by incubation of cylindrical hydrogel specimens (5 mm diameter, 2 mm thick) in phosphate-buffered saline (PBS) at 37°C for up to 12 weeks. No disintegration of the alginate polymer or the capillary structure could be observed. Hydrogel homogeneity was confirmed by light microscope analysis after ruthenium red staining of sections cut on a microtome (VT1000S; Leica Biosystems, Nussloch, Germany).

For cell culture experiments, capillary-free hydrogels were cut into slices (1 cm × 1 cm, 300 μm thick) using a microtome. For in vivo studies, cuboid implants (2 mm height, 2 mm length, 1.3 mm width) with anisotropic capillaries parallel to the 2-mm-long edge were cut. Homogeneity of the gel matrix as well as shape and orientation of the capillaries were confirmed by light microscopy for each implant (4 × objective, BX53; Olympus) and before implantation using a surgical stereomicroscope as previously described.^7,8 Hydrogels were stored in 70% ethanol (EtOH) at 4°C and washed 3 × (1 h each) in sterile dH₂O before further use. Two different batches of hydrogels were used in in vivo experiments evaluating the effect of peptide coating and astrocyte seeding with channel diameters of 39 ± 1.6 μm and 50.2 ± 2.1 μm, respectively. Analysis of the second batch showed a mean density of 89 channels/mm ² and a free channel volume of 20%. The different channel diameters used in the two sets of experiments did not influence outcomes (see section “Results”: similar number of neurite growth into channels coated with PLO/laminin).

Surface coating of alginate-based hydrogels

Poly-d-lysine (PDL, 30–70 kDa [n = 10] or 1–4 kDa [n = 5]), poly-l-ornithine (PLO, 30–70 kDa; n = 5), or poly-d-glutamic acid (PDGa, 15–50 kDa; n = 5; all from Sigma-Aldrich, St. Louis, MO) were bound to the hydrogel surface. To quantify coating efficiency, the amount of bound poly-peptide was measured through a depletion assay. Hydrogel cuboids were incubated in 200 μL of an aqueous solution containing the coating poly-peptides (1 mg/mL each) in a 96-well enzyme-linked immunosorbent assay plate for 16 h at 37°C, 5% CO₂. Coating solutions without hydrogels were used as controls. One hundred fifty microliter of each sample were removed and the protein concentration measured by a bicinchoninic acid assay (Thermo Fisher Scientific, Schwerte, Germany). Absorption was measured at 570 nm wavelength. To determine the amount of polypeptide bound to the hydrogel, the difference in the concentration measured in solutions incubated with hydrogels and control solutions was calculated using standard curves with the appropriate polypeptide. For in vitro cell experiments, hydrogel slices with or without capillaries were coated with PLO or PLO and subsequently with laminin (PLO/laminin). Gels were incubated in 500 μL PLO/dH₂O (1 mg/mL) for 16 h at 37°C, 5% CO₂, and washed 3 × 30 min each in Dulbecco's PBS (DPBS; Life Technologies, Carlsbad, CA) under slight agitation (60 rpm). A subset was further incubated in 500 μL laminin (10 μg/mL in DPBS) for 3 h under the same conditions. Afterward, gels were rinsed 3 × 10 min each in DPBS and cells immediately plated onto hydrogels for short-term in vitro experiments. Hydrogel cuboids with capillaries for in vivo experiments were coated with PLO/laminin as described above and implanted on the same day. For implantation of cell-seeded hydrogels, hydrogel channels were filled with astrocytes (see below) and immediately implanted.

Isolation of neonatal cortical astrocytes from Fischer 344 rats

Animal experiments were carried out in accordance with German guidelines for animal care and in accordance with the European Union Directive (2010/63/EU) or approved by the Institutional Animal Care and Use Committee. Animals were anesthetized by intraperitoneal injection (5 mL/kg) of a mixture of ketamine (62.5 mg/kg; Bremer Pharma, Warburg, Germany), xylazine (3.175 mg/kg; CP-Pharma, Burgdorf, Germany), and acepromazine (0.625 mg/kg; Sanofi-Ceva GmbH, Düsseldorf, Germany) in 0.9% (v/w) NaCl and killed by decapitation.

Astrocytes were isolated from cortices of neonatal (P1) GFP-transgenic Fischer 344 (F344) rats. Following brain dissection, hemispheres were separated and transferred into 30-mm cell culture dishes containing 1 mL ice-cold Dulbecco's modified eagle medium (DMEM) (Life Technologies). Excessive noncortical brain tissue, meninges, and blood vessels were carefully removed from cortical tissue under a dissection microscope. Cortices were mechanically dissociated using thin forceps and enzymatically digested by adding 500 μL of a 1:1 mixture of Collagenase XI (1200 U/mg; Sigma-Aldrich, in Hank's-buffered salt solution (HBSS; Life Technologies)) and Dispase I (4 U/mg in dH₂O; Worthington Biochemical Corp., Lakewood, NJ) for 45 min at 37°C, 5% CO₂ under permanent mechanical agitation (125 rpm). Afterward, cells were further dissociated by adding 1 mL prewarmed culture medium (DMEM/5% fetal bovine serum [FBS]/0.5% penicillin/streptomycin [10,000 U/mL each]/1% l-glutamine [200 mM]), and gentle trituration of the tissue 4 × with a sterile, flame-polished Pasteur pipette. The cell suspension was filtered through a 70 μm nylon cell strainer (Greiner Bio-One) and spun down for 8 min at 1300 rpm. Cell pellets were resuspended in 1 mL fresh prewarmed culture medium and plated into a T-75 cell culture flask (Greiner Bio-One) containing 15 mL culture medium.

Two days later, cells were washed once with DPBS to remove cellular debris and nonadherent cells. Medium was changed 3 × a week until cells reached confluency. To remove contaminating cells like neurons, microglia, and fibroblasts, 150 μL of a 10 mg/mL β-d-arabinofuranoside/DMEM (Ara-C; Sigma-Aldrich) solution was added and flasks were shaken at 225 rpm for 6 h at 37°C, 5% CO₂. Afterward, adherent cells were washed once with DPBS, enzymatically detached using 1 × TrypLE™ Express (Life Technologies) and seeded in fresh T-75 cell culture flasks. To assure purity, probes were labeled with rabbit anti-GFAP for astroglial cells (1:1000; Dako, Jena, Germany), mouse anti-β-III-tubulin for neurons (1:1000; Promega, Madison, WI), and rabbit anti-Iba1 for macrophages/microglia (1:500; Wako Chemicals GmbH, Neuss, Germany), goat anti-Sox2 for astrocytes and progenitors (1:200; Santa Cruz), mouse anti-vimentin for astrocytes and neural progenitors (1:1000; Millipore), goat-anti S-100β for astrocytes (1:1000; Sigma-Aldrich), and rabbit anti-NFIA for immature astrocytes (1:500; Abcam). Alexa-fluorophore-conjugated donkey secondary antibodies (1:300; Life Technologies) were used to visualize immunolabeling. All in vitro and in vivo studies were carried out with cells of passage 2 with 90–95% of all cells expressing the astrocyte marker GFAP. Cells also expressed other markers of mature and immature astrocytes, including S100β, Sox2, vimentin, and NFIA (Supplementary Fig. S1 Supplementary Data are available online at www.liebertpub.com/tea). For astrocyte seeding in vivo, a suspension of syngeneic GFP-transgenic neonatal astrocytes (3–5 μL, 2 × 10 ⁵ cells/μL in DPBS with 1 mg/mL glucose, passage 2) was soaked into the hydrogel channels with a sterile filter paper. Channel filling and the absence of air bubbles within the channels were visually inspected under a surgical microscope.

Dorsal root ganglias (DRGs) were isolated from adult GFP-transgenic F344 rats. Animals were anesthetized and killed as described above. Spinal columns were dissected and cut at the rostral and caudal end to expose the vertebral canal. The spinal cord was ejected by injecting 20 mL ice-cold HBSS into the caudal end of the vertebral canal with a 10-mL syringe. The vertebral column was cut in half longitudinally along the rostrocaudal axis and DRGs were carefully dissected by pulling the ganglia out of the bone cavity and cutting off the nerve roots. DRGs were cut in half and collected in ice-cold Hibernate A medium (Invitrogen, Carlsbad, CA). Afterward, DRGs were washed once in ice-cold HBSS and digested by incubation in 500 μL of a 1:1 mixture of Collagenase XI (1200 U/mg in HBSS) and Dispase I (4 U/mg in dH₂O) at 37°C for 30 min. Cells were washed twice with 1 × DMEM/F12 (Life Technologies) with 10% FBS and mechanically dissociated by triturating with a flame-polished glass Pasteur pipette about 15 times. Pieces of meninges and tissue debris were allowed to sink and cells were immediately used for in vitro experiments.

Quantification of coating efficiency

For in vitro experiments, neonatal astrocytes of passage 2 and freshly isolated DRG cells were used. For short-term experiments, hydrogel slices were placed into six-well cell culture plates (Greiner Bio-One) containing 4 mL prewarmed culture medium (see above). In total, 4 × 10 ⁴ astrocytes (passage 2) were plated onto uncoated control gels (n = 7), gels with PLO (n = 6) or PLO/laminin coating (n = 6). Medium was carefully changed after 2 and 4 days. After 7 days, cells were fixed with 4% ice-cold paraformaldehyde (PFA) in 0.1 M phosphate buffer (PB), pH 7.4 for 30 min and washed 3 × 10 min with 0.1 M tris-buffered saline (TBS). DRG-derived cells were seeded on uncoated control gels (n = 7) and gels with PLO (n = 6) or PLO/laminin (n = 6) coating and fixed with 4% ice-cold PFA in 0.1 M PB, pH 7.4, after 48 h.

To assess long-term bioactivity of the surface-bound proteins, PLO or PLO/laminin-coated hydrogels were incubated for 24 h in 10 mL DRG cell culture medium (DMEM/F12/2% B27 supplement/0.5% penicillin/streptomycin [10,000 U/mL each]/1% l-glutamine [200 mM]; PLO and PLO/laminin; n = 3), 24 h in 10 mL DPBS +24 h in DRG cell culture medium (PLO and PLO/laminin; n = 4), 7 days in DPBS +24 h in DRG cell culture medium (PLO and PLO/laminin; n = 4), or 14 days in DPBS +24 h in DRG cell culture medium (uncoated controls, PLO, and PLO/laminin; n = 4) at 37°C, 5% CO₂. DPBS was changed every day during the incubation period. On each hydrogel, 4 × 10 ⁴ freshly dissociated cells from DRGs were seeded, cultivated for 48 h and fixed as described above.

For immunolabeling, hydrogel slices were carefully transferred into 24-well cell culture plates (Greiner Bio-One) using blunt forceps. Unspecific antibody-binding sites were blocked and the tissue permeabilized with TBS/5% donkey serum/0.25% Triton X-100 for 1 h at room temperature (RT). Primary antibodies diluted in TBS/1% donkey serum/0.25% Triton X-100 were added overnight at 4°C. Afterward, samples were washed 3 × 10 min in TBS/1% donkey serum, incubated with Alexa-fluorophore-conjugated donkey secondary antibodies (1:300; Life Technologies) together with 4′, 6-diamidino-2-phenylindole (DAPI, 0.25 μg/mL; Thermo Fisher Scientific) in TBS/1% donkey serum for 2.5 h at RT followed by three washing steps with TBS. Finally, hydrogels were mounted onto microscope glass slides and sealed with Fluoromount G (Southern Biotechnology, Birmingham, AL). The following primary antibodies were used: goat anti-GFP for GFP-transgenic cells (1:1000; Rockland, Limerick, PA), rabbit anti-GFAP for astrocytes (1:1000; Dako), and mouse anti-β-III-tubulin for neurons (1:1000; Promega). Cell densities of all cultures and neuron density of DRG cultures were assessed by counting DAPI-labeled nuclei or β-III-tubulin-labeled neurons in a randomly selected field of 1 mm ² at 100 × magnification on a fluorescence microscope (Olympus BX53) on each gel. Neurite outgrowth was assessed by manually tracing the longest β-III-tubulin-labeled neurite of each neuron using the NeuronJ plug-in for ImageJ. All analyses were carried out by an observer blinded to group identity.

To analyze whether differences in cell adhesion underlie the observed differences in cell numbers, postnatal GFP-transgenic astrocytes (1.5 × 10 ⁴ ) were seeded on 7 × 7 mm uncoated, PLO- or PLO/laminin-coated 300 μm thick hydrogel slices after three washes in PBS and fixed after 3 h (n = 3 each). After immunolabeling for GFP, slices were imaged at 200 × magnification using an inverted epifluorescent microscope (Olympus; CKX53) with attached camera (Infinity 3-1URM) and the cell number was quantified in five randomly selected fields/slice using ImageJ. To determine whether differences in cell death contribute to differences in cell numbers, wild-type astrocytes seeded on hydrogels were cultivated for 2 days and stained with a Live/Dead Cell Assay Kit (1 μM calcein AM/4 μM nuclear dye; Fisher Scientific) according to the manufacturer's instructions. Cells were immediately imaged and live (green) and dead (red) cells were counted. Assays were done in triplicate.

To analyze the distribution of laminin coating, uncoated, PLO-coated and PLO/laminin-coated gel cuboids (1.3 mm × 2 mm × 2 mm) were washed, fixed with 4% PFA, and immunolabeled for laminin (1:300; Sigma, Alexa 488-conjugated secondary antibody; 1:1000). Images were taken on an inverted fluorescent microscope (Olympus; CKX53) with attached camera (Infinity 3-1URM) using the same exposure time for all images at the same magnification. In addition, potential differences of concentration-dependent biological effects of laminin at the edge and center in PLO/laminin-coated hydrogel cuboids were analyzed. Freshly PLO/laminin-coated and uncoated unfixed hydrogel cuboids (n = 2) were cut with a tissue chopper (McBain) (300 μm thickness), the outermost slices were discarded and GFP⁺ astrocytes (2 × 10 ³ cells) were seeded immediately onto the hydrogel slices (5–6 slices) in cell culture medium. After 48 h, hydrogels and attached cells were fixed, immunolabeled for GFP, and the size of cells found at the indicated distance from the edge (100–1000 μm) of the hydrogel was measured. Data are represented as the mean of all cells measured at a certain distance.

Surgical procedures

A total of 40 adult female F344 rats (12–14 weeks, >150g; Charles River) were used for in vivo studies. National guidelines and the European Union Directive (EU/2010/63) were strictly followed for all surgical procedures and animal care. Animals were housed in standard cages with free access to water and food. Animals underwent a C5/6 lateral hemisection lesion as previously described.^7,8 Briefly, animals were anesthetized with a mixture (2.5 mL/kg) of ketamine (62.5 mg/kg), xylazine (3.175 mg/kg), and acepromazine (0.625 mg/kg) diluted in sterile 0.9% NaCl. A laminectomy at cervical level 5 (C5) was performed. For lesioning, a rostrocaudal incision was made into the dura along the midline and a block of spinal cord tissue (∼2 mm in length) was removed unilaterally using microscissors and microaspiration without further damaging the dura overlying the lesion cavity. Hydrogel scaffolds were implanted immediately postinjury into the lesion cavity using a surgical microscope to ensure that the channels were oriented parallel to the rostrocaudal axis of the spinal cord. Care was taken to minimize any deformation of the hydrogel during implantation.

To evaluate the impact of peptide/protein coating in Experiment 1, a total of 24 animals received either uncoated control hydrogels (n = 9) or hydrogels coated with PLO/laminin (n = 15) immediately after lesioning. To evaluate the effect of astrocyte seeding into hydrogel channels, a total of 10 animals received PLO/laminin-coated scaffolds seeded with neonatal GFP⁺ astrocytes, whereas PLO/laminin-coated scaffolds without cells served as control (n = 6).

The dura was covered with a thin dried agarose film (1% UltraPure™ agarose in sterile dH₂O; Life Technologies) and sealed with fibrin glue (2 μL fibrinogen, 100 mg/mL and 2 μL thrombin, 400 U/mL; Sigma-Aldrich). Muscle layers were sutured and the skin was stapled. Postoperatively, animals received subcutaneous injections of buprenorphine-HCl (0.03–0.05 mg/kg; Reckitt Benckiser, Slough, United Kingdom) twice a day for 2 days for pain relief and ampicillin (50 mg/kg; Ratiopharm GmbH, Ulm, Germany) as needed. Carprofen (4–5 mg/kg; Pfizer, Inc., New York, NY) was administered subsequently once daily as needed. Animals were fed with 2 mL high-energy nutrition drink (Fresubin; Fresenius Kabi) up to three times per day until their body weight stabilized.

Tissue processing

Four weeks postimplantation, rats were deeply anesthetized (see above) and transcardially perfused with 0.1 M PBS followed by 4% PFA in 0.1 M PB, at pH 7.4. Spinal cords were carefully dissected and subsequently postfixed in 4% PFA/0.1 M PB at RT for 1 h. Tissue was cryoprotected in 30% sucrose/0.1 M PB at 4°C for 2 days and serially cut into 30-μm-thick horizontal sections on a cryostat (Leica Biosystems). Sections were directly mounted onto glass slides (Thermo Fisher Scientific), with every 14th section placed on the same slide. For immunohistochemistry, tissue sections were labeled as described above using DAPI (0.25 μg/mL), rabbit anti-GFAP for astroglial cells (1:1000; Dako), goat anti-rabbit GFP for cells isolated from GFP-transgenic rats (1:1000; Rockland), mouse anti-βIII-tubulin for axons (1:1000; Promega), rabbit anti-serotonin for descending raphespinal axons (5-HT; 1:2000; Immunostar), rabbit anti-von Willebrand factor for endothelial cells and blood vessels (1:1000; Sigma-Aldrich), and Alexa-fluorophore-conjugated donkey secondary antibodies (Alexa-488, Alexa 594; 1:300; Life Technologies). Sections were dried and coverslipped with Fluoromount G (Southern Biotech). Sections were imaged using a XC30 camera connected to a BX53 microscope (Olympus) or a FluoView1000 confocal microscope (BX61; Olympus).

Quantification of cell-filled channels, axon number, astroglia, and microglia

To quantify host cell migration into the scaffolds, the percentage of channels that were densely filled with DAPI⁺ nuclei was quantified at a virtual line perpendicular to the rostrocaudal orientation of channels, 100 and 500 μm from the rostral and caudal edge of the hydrogel in every seventh section at 400 × magnification under epifluorescent illumination (Olympus; BX53). Rostral and caudal areas were quantified separately as there is independent axon penetration from both sides. Distances of 100 and 500 μm were chosen to represent growth over short distances and more central location in the hydrogel channels, respectively. The same distances were measured in our previous studies making the current data comparable.^7,8 For astrocyte-seeded scaffolds, survival of grafted GFP-transgenic astrocytes was assessed by quantifying the percentage of DAPI⁺/GFP⁺ channels at 100 and 500 μm from the rostral and caudal edge of the hydrogel at 400 × magnification. Channels containing only sparse, isolated DAPI-labeled nuclei and/or GFP-labeled cells were not considered positive. In addition, the total channel area filled with GFP-labeled cells was analyzed in the rostral half of the hydrogel (0–500 and 500–1000 μm) at 100 × magnification in all animals implanted with astrocyte-filled channels.

Similarly, β-III-tubulin-labeled as well as 5-HT-labeled axons within scaffold channels were analyzed using epifluorescent illumination at 200 × magnification. In all channels, neurites crossing virtual lines perpendicular to the channel axis at 100 and 500 μm from the rostral and caudal edge of the hydrogel were counted in every seventh section. For each animal, mean values from 100 and 500 μm of the rostral and caudal site were averaged for β-III-tubulin-labeled axons and analyzed separately for 5-HT-labeled axons. Results of all animals within each treatment group at each location (uncoated vs. PLO/laminin coating or PLO/laminin coating vs. PLO/laminin coating with neonatal astrocytes) were averaged for comparisons. Channel and axon quantification was performed while blinded to group identity.

The number of axons per mm ² hydrogel was calculated using the following equation: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { \rm { N } } = { \frac { \sum { \rm { axons } } } { 1 , 000 , 000 { \rm { *section \;thickness } } \left( { { \rm { \mu m } } } \right) { \rm { * } } \sum { \rm { width \;of \;the \;hydrogel \;at \;specific \;distance } } \left( { { \rm { \mu m } } } \right) } } . \end{align*} \end{document}

For quantification of GFAP and Iba1 labeling density, images of fluorescent-labeled serial spinal cord sections, rostral and caudal, to the hydrogel implantation site and of the uninjured contralateral side were taken at 200 × magnification (250 × 250 μm). Using a threshold that represents the labeling density on the uninjured side, the percent labeling density was quantified in animals that received uncoated and PLO/laminin-coated hydrogels.

Statistical analysis

All data were analyzed by an observer blinded to group identity. Data are expressed as mean ± standard error of the mean. Statistical analysis was performed using Prism 6 software (GraphPad Software, Inc., La Jolla, CA) and a significance criterion of p < 0.05 was used. Groups were compared by one-way analysis of variance (ANOVA) followed by Tukey's post hoc test or two-way ANOVA followed by Sidak's post hoc tests.

Coating efficiency of differently charged polypeptides

Cell adhesion and biocompatibility of biomaterials are influenced by surface charges and the presence of ECM. To determine whether peptide/ECM coating would affect cell survival/adhesion and neurite growth in alginate scaffolds, we first explored whether polyamino acids and peptides can be stably bound to alginate-based hydrogels. The linear polysaccharide alginate consists of homopolymeric as well as alternating blocks of α-l-guluronate and β-d-mannuronate with negatively charged carboxylic acid groups (CO₂H⁻) at neutral pH, which may crucially influence peptide binding. To analyze the effect of protein charge, alginate-based hydrogel cuboids with a dry weight of 0.5 mg were tested for the binding capacity of acidic and basic polyamino acids (Supplementary Fig. S2). While only a very small amount of acidic, negatively charged PDGa (15–50 kDa) was bound (8 ± 5 μg/hydrogel), significantly higher amounts of basic, positively charged PLO (120 ± 9 μg/hydrogel) and of high- and low-molecular weight PDL (30–70 kDa: 213 ± 15 μg/hydrogel; 1–4 kDa: 145 ± 5 μg/hydrogel) attached to the hydrogel matrix (p < 0.001). Interestingly, higher molecular weight PDL was more effective in alginate binding than low-molecular weight PDL (p < 0.01) or PLO (p < 0.001). Although PDL showed a higher binding affinity to the alginate surface, PLO was chosen for subsequent studies as previous data suggest a lower immunogenicity compared with PDL.^12,38–40

Cell adhesion and viability on surface-coated alginate-based hydrogels in vitro

To examine the influence of polyamino acid and peptide coating on cell attachment and survival, PLO or PLO in combination with the ECM component laminin were bound to 300-μm-thick hydrogel slices, whereas uncoated hydrogels served as controls (Fig. 1). Neural cells (astrocytes or DRG neurons with satellite cells) were seeded immediately following coating and short washes of the hydrogel slices. Seven days after cell seeding, the number of GFP-transgenic neonatal cortical astrocytes was 10-fold higher on coated versus uncoated hydrogels (Fig. 1E–H; p < 0.0001), although laminin coating did not provide any additional effect (Fig. 1H). Notably, in addition to differences in cell number, clear differences in cell morphology were observed in response to coating. On uncoated and PLO-coated hydrogels, neonatal astrocytes were predominantly rounded with small somata bearing no processes. With additional laminin coating, astrocytes had larger somata and developed a bi- or tripolar morphology with longer processes. Similarly, adult DRG neurons and DRG-derived glia showed pronounced cell numbers on surface-coated hydrogels (Fig. 1I–L; p < 0.0001). Neurite growth was virtually absent on uncoated controls and only sporadically observed on PLO-coated hydrogels, whereas PLO/laminin coating elicited substantial neurite outgrowth (quantified in Fig. 2). Cells did not invade the hydrogel walls but were only found on the surface. To determine whether cell attachment and/or survival were influenced by PLO/laminin coating, astrocytes were fixed 3 h or 2 days after cell seeding. Quantification of cell numbers (GFP⁺ cells) shortly after seeding indicated a significantly higher number of cells attaching to PLO/laminin-coated gels (Fig. 1I). In addition, a smaller percentage of dead cells was detected on PLO/laminin-coated hydrogel slices after 2 days of cultivation, although this difference did not reach significance (Fig. 1J; ANOVA p = 0.28). Thus, binding of PLO and especially the combination of PLO and laminin to alginate hydrogels improves cell attachment and to a smaller degree survival, and increases neurite extension.

OPEN IN VIEWER

FIG. 1.

Cell adhesion and viability on surface-modified alginate-based hydrogels. (A–D) GFP-transgenic neonatal cortical astrocytes and (E–H) adult DRG neurons and glia were seeded on (A, E) uncoated controls, (B, F) PLO- and (C, G) PLO/laminin-coated hydrogels. After 7 days, only very few rounded astrocytes are detected on uncoated control hydrogels. Cell number is increased by PLO coating. On PLO/laminin-coated hydrogels, neonatal astrocytes develop cellular processes and larger somata. (E–H) More cells and substantially more DRG neurite outgrowth are found on PLO and PLO/laminin-coated hydrogels. (D, H) Quantification of DAPI-labeled nuclei on control- and coated hydrogels after 7 days in vitro (ANOVA and Tukey's post hoc test, ****p < 0.0001). (I) Shortly (3 h) after seeding of astrocytes, significantly more cells have attached to PLO/laminin-coated hydrogels compared with uncoated controls and PLO-coated alginate hydrogels (ANOVA p < 0.01, Tukey's post hoc test *p < 0.05; **p < 0.01). (J) In addition, a reduction in cell death by PLO/laminin coating is observed 2 days after cell seeding. (K–M) Representative images of calcein-labeled (live; green) cell somata and nuclei of dead cells (red; arrows) on (K) control, (L) PLO- and (M) PLO/laminin-coated gels. Scale bars: 100 μm. DRG, dorsal root ganglia; PLO, poly-l-ornithine; DAPI, 4′, 6-diamidino-2-phenylindole; ANOVA, analysis of variance.

OPEN IN VIEWER

FIG. 2.

Long-term stability of alginate polypeptide/protein coating agents. (A–C) DAPI labeling of nuclei (blue) and neurons (β-III-tubulin—green) isolated from adult DRGs reveal increased cell survival and neurite outgrowth on PLO- and PLO/laminin-coated alginate hydrogels even after 2 weeks of preincubation and daily washing. (D) Quantification of DAPI-labeled nuclei, (E) β-III-tubulin-labeled neurons, and (F) neurite outgrowth on uncoated control, PLO-coated, and PLO/laminin-coated hydrogels after incubation in Dulbecco's phosphate-buffered saline for 1, 2, 8, and 15 days with daily washing steps before cell plating demonstrates no differences over time, but increased cell numbers and neurite growth compared with uncoated control gels (ANOVA p < 0.0001; Tukey's post hoc test, **p < 0.01, ****p < 0.0001). Scale bar in (C): 100 μm.

Long-term retention and bioactivity of surface-coated proteins in vitro

In vivo experiments with surface-coated scaffolds require stability and bioactivity of peptide/protein coating for longer periods than the short-term experiments described above. We therefore examined alginate hydrogels that were coated with PLO or PLO/laminin and incubated under physiological conditions (37°C, 5% CO₂ in DPBS) up to 15 days, including repeated daily washing steps before cells isolated from adult rat DRGs were plated onto the hydrogels (Fig. 2A–C). At all time points, the number of cells (DAPI⁺ nuclei) attached to PLO-coated and PLO/laminin-coated hydrogels was considerably higher than the number of cells found on uncoated controls (Fig. 2D). This was also true for β-III-tubulin-labeled neurons, but these differences did not reach statistical significance (ANOVA p = 0.076; Fig. 2E). Importantly, there was no significant difference in the number of attached cells that were seeded 1 or 15 days after hydrogel coating. In addition to cell attachment and survival, surface coating also affected DRG neurite outgrowth. When measuring the longest neurite 48 h after cell seeding (Fig. 2F), a slight increase in neurite length was apparent on PLO-coated hydrogels. The addition of laminin dramatically increased the average neurite length a further three to four-fold (ANOVA p < 0.0001, Tukey's post hoc test p < 0.01 comparing PLO to PLO/laminin; p < 0.0001 comparing control to PLO/laminin). Again, this effect was independent of the time that hydrogels were used to after coating. Thus, surface modification with PLO alone or in combination with laminin is stable retaining full bioactivity for at least 2 weeks in vitro.

Integration and host cell infiltration of surface-coated hydrogel scaffolds in vivo

For in vivo experiments, alginate hydrogel cubes were coated with PLO/laminin. Immunolabeling for laminin indicated higher immunoreactivity at the hydrogel edges, but also a distinct signal along the hydrogel channels that was still present at the center of the hydrogel (Supplementary Fig. S3A–D). To determine whether the laminin detected in the center of the gel was sufficient for biological effects, uncoated and PLO/laminin-coated hydrogel cubes (1.3 mm × 2 mm × 2 mm) were freshly sectioned and seeded with GFP⁺ astrocytes. Quantification of astroglial cell size at different distances from the edge of the gel indicated an overall increased cell size by PLO/laminin coating compared with uncoated controls with no difference across different distances from the edge. Taken together, the concentration gradient observed by immunolabeling does not appear to be biologically relevant at least not for the tested cells.

To characterize cellular responses in vivo, PLO/laminin-coated alginate hydrogels, which induced superior neurite growth in vitro, or uncoated hydrogels (control) were implanted into a unilateral C5/6 spinal hemisection immediately postinjury. Four weeks postimplantation, scaffolds showed no signs of degradation, remained in rostrocaudal orientation, and the capillary 3D microarchitecture remained intact (Fig. 3), consistent with previous studies.^7,8 Implants were tightly integrated into the surrounding host parenchyma without excessive hypercellularity or cavitation.

OPEN IN VIEWER

FIG. 3.

Integration and host cell infiltration into hydrogel channels. After 4 weeks, hydrogel scaffolds remain structurally intact and integrate tightly into the lesion site without excessive cavitation. DAPI labeling of nuclei (white) reveals (A) limited host cell infiltration into uncoated controls compared with (B) PLO/laminin-coated scaffolds (rostral is to the left, medial to the top). (C) Quantification of DAPI-labeled nuclei shows significantly increased cell infiltration into surface-modified hydrogel scaffolds 4 weeks postimplantation (two-way ANOVA p < 0.01 for distance, p < 0.0001 for group difference. Sidak's post hoc test *p < 0.05). (D) Independent of coating conditions, astrocytes are found in close proximity to hydrogels, but only in few instances in direct contact with the implant or within its channels. In control and PLO/laminin-coated hydrogel channels, the majority of infiltrating host cells are (E, F) Iba1⁺ microglia/macrophages that fill the capillary lumen. (G, H) Qualitatively, more Schwann cells with rostrocaudal orientation are found in channels of coated hydrogels. Scale bar in (A, B, D): 500 μm, (E–H): 30 μm.

Nuclear staining with DAPI indicated host cell migration toward and infiltration into the hydrogel channels but not into the channel walls (Fig. 3A, B). Quantification of DAPI⁺ nuclei in hydrogel areas close to host tissue (100 μm) as well as at more central regions (500 μm) of the channels revealed a significantly higher percentage of cell filling in PLO/laminin-coated scaffolds compared with uncoated controls (p < 0.05; Fig. 3C). While infiltrating cells densely clustered at the channel entrances, they were sparser at central regions of control scaffolds. In contrast, host cells filled broad portions of the channel lumen in surface-coated scaffolds. A small area of hypercellularity was present at the host–graft interface surrounded by GFAP⁺ astrocytes independent of coating conditions (Fig. 3D). Interestingly, GFAP-labeled host astrocytes rarely infiltrated the host–graft interface and did not extend processes into the channels. Within the hydrogel scaffolds, immunolabeling identified the vast majority of infiltrating cells in both groups as Iba1⁺ cells (microglia/macrophages) (Fig. 3E, F) and some Schwann cells (Fig. 3G, H). To exclude the possibility that the increased cell filling is due to increased glial activation by PLO/laminin, microglial and astroglial labeling was also measured in areas surrounding the scaffolds. No difference in GFAP and Iba1-labeling density was observed between both groups (Supplementary Fig. S4).

In vivo neurite growth into channels of coated hydrogels

To evaluate direct or indirect influences of surface coating on host axonal growth in vivo, the number of β-III-tubulin-labeled neurites entering and extending within the hydrogel channels was quantified (Fig. 4). In control as well as in PLO/laminin-coated scaffolds, axons entered the scaffolds from both the rostral and caudal host parenchyma (Fig. 4A, B). Hardly any neurites were found in channels without infiltrating host cells and the number of neurites per cell-filled (DAPI⁺) channel did not differ between implants coated with PLO/laminin and uncoated controls (Fig. 4C) at channel entrances (100 μm) or inside the hydrogel implant (500 μm). In both groups, fewer neurites per cell-filled channel were found with increasing distance from the host–graft interface and ∼50% less neurite growth was found within the central regions of the implants.

OPEN IN VIEWER

FIG. 4.

Neurite growth into microchannels of coated hydrogel scaffolds. (A, B) Single β-III-tubulin-labeled neurites and neurite bundles are found in channels of control and PLO/laminin hydrogel scaffolds (rostral is to the left; medial to the top). (C) Quantification of β-III-tubulin-labeled fibers reveals no difference in the number of neurites per DAPI⁺ cell-filled channel between groups but a significant decline toward the center (two-way ANOVA p = 0.49 for groups, p < 0.0001 for distance; Sidak's post hoc test *p < 0.05; ***p < 0.001). (D) Significantly more neurites are found at the entry zones of coated scaffolds due to a higher number of filled channels and PLO/laminin coating increases neurite numbers at the rostral entry zone (two-way ANOVA, p < 0.05 for group difference, p < 0.0001 for distance, Sidak's post hoc test *p < 0.05 comparing control and PLO/laminin at 100 μm; ^###p < 0.001 comparing 500 μm and −500 μm vs. 100 μm and −100 μm in the PLO/laminin group, ^$p < 0.05 comparing 500 μm vs. 100 μm and −100 μm and −500 μm vs. −100 μm in the control group). Scale bar in (B): 100 μm.

However, when examining the total number of neurites, significantly more neurites extended into PLO/laminin-coated scaffolds (Fig. 4D) at the rostral entry zones (p < 0.05) consistent with the higher number of cell-filled channels in this group (Fig. 3). Taken together, these data suggest that PLO/laminin coating promotes significantly higher host cell infiltration and thereby more neurite penetration into the hydrogels, but this is insufficient to sustain axonal growth for longer distances into or across the channels.

Seeding of neonatal astrocytes into surface-coated hydrogels

Even in alginate hydrogels coated with PLO/laminin, virtually no astrocytes or astrocytic processes were detected within the channels (Fig. 3D). To assess whether astroglia would provide a superior cellular substrate compared with infiltrating host cells, improve scaffold integration and increase axonal regeneration within the scaffold, we seeded syngeneic cortical astrocytes isolated from neonatal (P1) GFP-transgenic F344 rats into the channels of PLO/laminin-coated hydrogels immediately before implantation. Cell-filled scaffolds and PLO/laminin-coated scaffolds without cells as control were implanted immediately after lesioning the spinal cord. Inspection of scaffolds before transplantation showed an even distribution of astrocytes in cell-filled channels (Supplementary Fig. S5). Four weeks postinjury, scaffolds of both groups remained intact, were oriented rostrocaudally, and tightly integrated, particularly scaffolds seeded with astrocytes (Figs. 5A, B, and 6).

OPEN IN VIEWER

FIG. 5.

Host cell infiltration and survival of grafted neonatal astrocytes. Scaffolds in both groups remain intact and are tightly integrated into the host parenchyma 4 weeks postgrafting. DAPI-labeled nuclei (white) show host cell infiltration into (A) PLO/laminin-coated scaffolds in the control group and (B) slightly higher cell density in the cell transplantation group. Dashed lines indicate channel lumen. (C) Quantification of DAPI⁺ channels reveals an overall increased cell filling in astrocyte-seeded scaffolds compared with controls (two-way ANOVA p < 0.0001 for distance and p < 0.001 for group differences, Sidak's post hoc *p > 0.01 comparing groups at 500 μm). A significant decline in the % of DAPI⁺ channels is found in the central hydrogel portion in PLO/laminin (^#p < 0.05; ^##p < 0.01; ^####p < 0.001) and astrocyte-filled channels (⁺p < 0.05, ⁺⁺p < 0.01). (D) GFAP immunolabeling and (E) GFAP/GFP double labeling of a section adjacent to the one shown in (B) reveals that the majority of channels in the cell transplantation group are filled with astrocytes, which are also GFP positive and therefore graft derived. Grafted astrocytes fill the host–alginate interface and intermingle with the host spinal tissue, but do not extensively migrate in rostral or caudal direction (see also Fig. 6). (F) In the transplantation group, more than 80% of all cell-filled channels contain grafted GFP⁺ astroglia (DAPI⁺/GFP⁺ channels) independent of the distance from the lesion edge (ANOVA p = 0.4). Scale bar: 500 μm. Rostral is to the left, medial to the top.

OPEN IN VIEWER

FIG. 6.

Integration and host–graft interaction of astrocyte-seeded hydrogel scaffolds. At the (A) rostral and (B) caudal host–graft interface, processes of GFP⁺ grafted cells (green) intermingle with host astrocytes (GFAP, red). Grafted cells form a continuous glial network guiding axons (βIII-tubulin, white) into the hydrogel scaffold (arrowheads) (C) Within the channels, the transplanted GFP⁺ cells (green) organize in a dense network-like structure covering broad portions of the microchannel walls. XZ and YZ planes are shown underneath and to the right, respectively. (D) The majority of grafted cells (GFP⁺, green) coexpress the astrocyte marker GFAP (red). (E) Endothelial cells and possibly blood vessels (von Willebrand factor, red) are found throughout the entire rostrocaudal extent of the hydrogel channels and associate with astrocytes (GFAP⁺, white). Scale bars in (B, D, E): 50 μm.

Quantification of DAPI-labeled nuclei indicated a decline in cell-filled channels at the central locations in both groups (Fig. 5C, two-way ANOVA p < 0.0001 for distance) confirming our data from the first set of experiments. In addition, an overall higher percentage of cell-filled channels was apparent in hydrogels seeded with astrocytes in particular at the central locations (Fig. 5C; two-way ANOVA p < 0.001 for group differences; p < 0.05 for 500 μm distance). Assessing survival of grafted cells (percentage of GFP⁺ cell-filled channels), the vast majority of cell-filled channels contained GFP⁺ astrocytes (Figs. 5D–F and 6C, D). At all distances, 80–85% of all cell-filled channels showed GFP labeling. At the graft–host interface, GFP⁺/GFAP⁺ double-labeled cells filled the capillary lumen in a dense network-like fashion (Fig. 6A, B) and grafted cells did not migrate extensively into the host parenchyma. However, grafted astrocytes filled the host–graft interface and their processes were tight-knit with processes of host astrocytes serving as a cellular substrate for host axons extending into the channels (Fig. 6A, B, arrowheads). Toward the hydrogel center, the density of the GFP⁺/GFAP⁺ network declined, and GFP⁺ cells were occasionally organized in clusters (Fig. 5E). The total GFP⁺ channel area was 76.3 ± 3.2% in the rostral portion of the alginate gel (0–500 μm) and slightly declined to 53.7 ± 3.1% in the more central portion of the hydrogel (500–1000 μm; unpaired t-test p < 0.001). Endothelial cells and possibly thin blood vessels also entered and extended along astrocyte-containing channels (Fig. 6E). Besides grafted astrocytes and endothelial cells, more than 80% of all GFP⁺ channels displayed immunoreactivity for Iba1 identifying GFP-negative cells mainly as macrophages/microglia (data not shown) similar to our previous observations (Fig. 3).

Neurite growth into astrocyte-seeded capillary hydrogels

To analyze axonal growth responses, β-III-tubulin-labeled fibers were quantified at a distance of 100 and 500 μm from the rostral and caudal edge of PLO/laminin-coated hydrogels and PLO/laminin-coated hydrogels filled with astrocytes (Fig. 7). Overall, more neurites per DAPI⁺ channel were present in animals implanted with astrocyte-containing hydrogels, but this difference did not reach significance (Fig. 7C). As 10–20% of channels in animals that received astrocyte-filled scaffolds only contained host cells (DAPI⁺/GFP⁻; Fig. 5), we were able to determine whether extending neurites preferred astrocyte-containing channels versus channels containing only host cells within the same alginate scaffold. Interestingly, in animals grafted with astrocyte-seeded scaffolds, the number of neurites/channel was much higher in channels that contained astrocytes (DAPI⁺/GFP⁺) compared with cell-filled channels devoid of astrocytes (DAPI⁺/GFP⁻). More than 86% of all axons in these animals were found in channels containing GFP⁺ astrocytes. This was true for all distances examined indicating that astrocytes generated a preferable environment for regenerating axons. In animals with and without astrocyte seeding, neurite counts per channel declined from the rostral to the central region of the scaffold (Fig. 7C). Quantifying the entire channel lumen at the same distances, neurite density was overall higher in astrocyte-filled scaffolds (two-way ANOVA p < 0.001 for group differences), but this difference only reached significance at 100 μm within the rostral half of the scaffold 4 weeks postimplantation (Fig. 7D). Again, a decline in the number of neurites from rostral and caudal ends toward the center was found in both groups.

OPEN IN VIEWER

FIG. 7.

Neurite growth into microchannels of astrocyte-seeded hydrogel scaffolds. (A) Thin bundles and single β-III-tubulin-labeled fibers extend into channels of control animals, whereas (B) thick neurite bundles and branching neurites are found in scaffolds seeded with neonatal astrocytes. (C) A significantly higher number of neurites/channel is found in DAPI⁺/GFP⁺ channels (filled with graft-derived astrocytes) compared with DAPI⁺/GFP⁻ channels in the same animals (two-way ANOVA p < 0.0001; Sidak's post hoc test, ^#p < 0.05, ^####p < 0.0001). Within DAPI⁺/GFP⁺ channels and in animals without grafted cells, there is a decline in neurites per channel from rostral toward the center of the hydrogel (two-way ANOVA for distance p < 0.0001; Sidak's post hoc test, *p < 0.05, **p < 0.01, ***p < 0.001). (D) Scaffolds seeded with neonatal astrocytes contain more neurites than control hydrogels at all distances (two-way ANOVA p < 0.001 for overall group differences, Sidak's post hoc *p < 0.05 at 100 μm). Both groups show a decline in neurite number toward the center of the hydrogel (two-way ANOVA p < 0.0001 for distance; Sidak's post hoc: PLO/laminin: ^#p < 0.05; ^##p < 0.01; astrocyte-filled channels: ⁺⁺p < 0.01; ⁺⁺⁺⁺p < 0.0001).. Rostral is to the left, medial to the top. (E, F). Serotonergic fibers (5-HT⁺, red) grow into and along the channels of (E) control animals and (F) animals that received astrocyte-filled hydrogels from the rostral (left) side and extend beyond areas of dense GFP-labeled astrocytes. (G) Quantification of 5-HT⁺ fibers indicates a higher number in scaffolds seeded with neonatal astrocytes at the rostral entry zone, but no overall group difference per DAPI⁺ channel (two-way ANOVA p = 0.25). There are significant differences across distances (two-way ANOVA p = 0.0001), but only in the astrocyte-containing scaffolds (Sidak's post hoc ***p < 0.001). Comparing the number of 5-HT⁺ fibers extending into GFP⁺ and GFP⁻ channels in the astroglia-seeded group, a higher number per channel is found in GFP⁺ channels at 100 μm (two-way ANOVA p < 0.05 GFP⁺ vs. GFP⁻; interaction p < 0.01, distance p < 0.0001, Sidak's post hoc ^####p < 0.0001 at 100 μm). (H) Similarly, more 5-HT fibers/mm ² are detected in astrocyte-containing scaffolds, but the high variability did not result in significant group differences (two-way ANOVA p = 0.1 for group differences). With increasing distance, the number of 5-HT⁺ fibers declines in astrocyte-containing scaffolds (two-way ANOVA p = 0.0001; Sidak's post hoc ****p < 0.0001 comparing 100 μm to other distances). Scale bar in (B): 25 μm, in (F): 50 μm.

To assess whether descending fibers contributed to neurite growth into the hydrogel scaffolds, raphespinal axons were labeled with an antibody for serotonin (5-HT) (Fig. 7E, F). Descending 5-HT⁺ fibers extended toward the injury site and entered the rostral end of the scaffolds in both groups (Fig. 7E, F). In astrocyte-seeded scaffolds, serotonergic fibers extended beyond areas with dense GFP labeling indicating that grafted astrocytes did not provide a barrier for axonal extension (Fig. 7E). The largest number of 5-HT⁺ axons was found at the rostral entry zone followed by a dramatic decrease toward the central and caudal part of the scaffold. Importantly, serotonergic fibers preferred astrocyte-containing channels over channels containing only infiltrated host cells (Fig. 7G), consistent with data evaluating overall neurite growth (Fig. 7C) (Rostral 100 μm, p < 0.0001). The same result was observed when the amount of 5-HT⁺ fiber growth in the entire scaffold was evaluated: on average, nearly 3 × as many raphespinal fibers entered the scaffold in animals with astrocyte seeding compared with controls, but this difference did not reach significance due to high variability, and the number of serotonergic axons dramatically declined toward the central and caudal regions of the scaffolds (p < 0.0001) (Fig. 7H).

In summary, seeding of neonatal astrocytes into alginate-based hydrogels improved scaffold integration into the surrounding host parenchyma. While seeded astrocytes only slightly increased axonal entry into the scaffold, regrowing axons preferred an environment containing grafted astrocytes over infiltrating host cells within scaffold channels.