A Transient Cell-Shielding Method for Viable MSC Delivery within Hydrophobic Scaffolds Polymerized In Situ

Abstract

Cell-based therapies have emerged as promising approaches for regenerative medicine. Hydrophobic poly(ester urethane)s offer the advantages of robust mechanical properties, cell attachment without the use of peptides, and controlled degradation by oxidative and hydrolytic mechanisms. However, the application of injectable hydrophobic polymers to cell delivery is limited by the challenges of protecting cells from reaction products and creating a macroporous architecture post-cure. We designed injectable carriers for cell delivery derived from reactive, hydrophobic polyisocyanate and polyester triol precursors. To overcome cell death caused by reaction products from in situ polymerization, we encapsulated bone marrow-derived stem cells (BMSCs) in fast-degrading, oxidized alginate beads prior to mixing with the hydrophobic precursors. Cells survived the polymerization at >70% viability, and rapid dissolution of oxidized alginate beads after the scaffold cured created interconnected macropores that facilitated cellular adhesion to the scaffold in vitro. Applying this injectable system to deliver BMSCs to rat excisional skin wounds showed that the scaffolds supported survival of transplanted cells and infiltration of host cells, which improved new tissue formation compared to both implanted, pre-formed scaffolds seeded with cells and acellular controls. Our design is the first to enable injectable delivery of settable, hydrophobic scaffolds where cell encapsulation provides a mechanism for both temporary cytoprotection during polymerization and rapid formation of macropores post-polymerization. This simple approach provides potential advantages for cell delivery relative to hydrogel technologies, which have weaker mechanical properties and require incorporation of peptides to achieve cell adhesion and degradability.

1. Introduction

Autologous and allogeneic cell-based therapies have emerged as promising approaches for regenerative medicine [1]. While direct injection of cells has limited therapeutic efficacy due to poor cell survivability [2–4], delivery of cells within a 3D matrix can improve integration with host tissue and promote healing [5]. Injectable and settable cell carriers could be advantageous as a minimally invasive surgical approach to rapid filling of complex defects followed by in situ curing to form a porous scaffold with suitable mechanical properties [6].

Lysine-derived poly(ester urethane)s (PURs) offer potential advantages as injectable carriers for local cell delivery, such as curing using non-cytotoxic catalysts [7] without the need for UV radiation [8], support of cell attachment without cell adhesion peptides [9, 10], tunable hydrolytic and oxidative degradation to non-cytotoxic breakdown products [11, 12], and adjustable mechanical properties ranging from those of soft tissue [13] to bone [9, 14]. Furthermore, macropores can be generated within PUR scaffolds by CO₂ gas foaming via the reaction of isocyanate groups with water [15]. When using these materials as acellular scaffolds, the CO₂ and heat generated by the in situ reaction is well tolerated at the biomaterial-tissue interface [7, 16] due to the relatively long length scales (>1 mm) between the material and surrounding cells (Figure 1A). However, cells encapsulated within the reactive hydrophobic polymer experience steeper CO₂ and temperature gradients due to transport of reaction products over much smaller length scales (<100 µm, Figure 1A). Furthermore, after the reaction is complete, hydrophobic polymers absorb negligible amounts of water and allow less diffusion of vital cell nutrients and wastes than swollen hydrogels. While hydrophobic biomaterials such as PUR provide a generalizable, biodegradable platform for tissue scaffolding, their use as an injectable carrier for cell delivery has not been achieved due to two primary challenges: (1) maintenance of cell viability during in situ polymerization, and (2) provision of an interconnected, macroporous structure to allow effective nutrient and waste exchange post-cure. Overcoming these key barriers was the goal of the current work in order to enable the use of injectable, settable, mechanically robust, and cell-adhesive PUR networks to fill tissue defects and to locally deliver and retain viable cells in vivo.

Figure 1. Design of injectable, settable carriers for cell delivery.

(A) For an acellular scaffold, the length scale of diffusion of reaction products is comparable to the size of the tissue defect. However, in a cellular scaffold, reaction products diffuse radially toward the encapsulated cell over a much shorter length scale (comparable to the size of the cell). (B) Schematic illustrating the design concept in which an NCO-functional prepolymer reacts with a polyester polyol in the presence of an iron acetylacetonoate (FeAA) catalyst to form a polyurethane network. Encapsulation of cells in oxidized alginate beads (green) provides temporary protection from the chemical reaction and is followed by hydrolytic degradation of the oxidized alginate to form interconnected macropores that are enhanced by the NCO-water reaction.

Achieving these goals will provide a new alternative to photopolymerizable systems that utilize cytocompatible initiators [17, 18] and water-soluble macromers [19–21] to encapsulate cells in injectable hydrogels [8]. Polyethylene glycol (PEG)-based hydrogels have generated considerable interest for localized cell delivery since they can be administered by minimally-invasive injections, set within clinically relevant working times, exhibit tissue-like structure, and induce a minimal inflammatory response [1, 22–24]. However, PEG hydrogels must be functionalized with an optimal combination of peptides that serve as integrin-binding sites for cell adhesion and peptide crosslinkers that are matrix metalloproteinase (MMP) substrates to enable cellular infiltration and cell-mediated hydrogel degradation [5, 25].

Alternative settable carriers must protect cells from reaction products prior to cure and then set in situ to form an interconnected, macroporous scaffold that supports cell adhesion and growth. In this study, we designed injectable PUR scaffolds for concurrent incorporation of macropores and cells within PUR scaffolds (Figure 1B). Through encapsulation within partially oxidized sodium alginate (o-Alg) beads, cells were protected from the PUR reaction prior to gelation. Hydrolytic degradation of the o-Alg beads within the first 1 – 2 days after gelation was anticipated to result in cell release and attachment to the scaffold. Thus, in contrast to the porogen co-encapsulation approach [26, 27], the o-Alg beads functioned both as a temporary barrier to transport of reaction products as well as a porogen. We varied bead size, timing of bead addition, and bead loading within PUR scaffolds to investigate the effects of heat and CO₂ generation on cell survivability both prior to and after gelation in vitro. In a proof-of-concept experiment, the lead-candidate formulation that produced maximal cell survivability in vitro was injected into full-thickness excisional skin wounds in Sprague-Dawley rats to evaluate the potential of the injectable PUR cell carrier for wound repair and restoration.

2. Materials and Methods

2.1. Materials

The sodium salt of alginic acid (Alg, viscosity = 20 – 40 cPs) was supplied by Sigma Aldrich (St. Louis, MO). Acros Organics supplied calcium chloride and glycerol. αMEM and DMEM were supplied by GIBCO. Fetal bovine serum (FBS) was purchased from Thermo Scientific. Penicillin/streptomycin (P/S), trypsin EDTA and Amphotericin B were obtained from Corning Cellgro. Live/Dead kits for mammalian cells were supplied by Life Technologies. Glycolide and D,L-lactide were purchased from Polysciences (Warrington, PA). Lysine triisocyanate-poly(ethylene glycol) (LTI-PEG) prepolymer was supplied by Medtronic, Inc, and hexamethylene diisocyanate trimer (HDIt) was supplied by Bayer Material Science. Iron acetylacetonate (FeAA) was supplied by Sigma-Aldrich. ε-caprolactone was dried over anhydrous MgSO₄, and all other materials were used as received.

2.2. Cell culture

MC3T3 cells (ATCC) were cultured in a complete medium of αMEM with 10% FBS and 1% P/S. Primary rat bone marrow mesenchymal stem cells (BMSCs) were maintained in DMEM with 10% FBS, 1% P/S, and 0.1% Amphotericin B (Sigma). BMSCs were generated from pooled bone marrow from 4 male Sprague-Dawley rats. Rat femora and tibiae were removed after sacrificing and bone marrow flushed with BMSC culture medium. After centrifuging, cell pellets were suspended in BMSC medium and plated in T75 tissue culture flasks. Three days after seeding, floating cells were removed and culture medium refreshed every other day. BMSCs were differentiated to osteoblasts and adipocytes (Figure S1) to confirm their pluripotency.

2.3. Preparation of partially oxidized alginate

Partially oxidized sodium alginate (o-Alg) degrades significantly faster than untreated Alg [28]. Furthermore, o-Alg has been reported to induce negligible inflammation and oxidative stress responses in cells [29], and it does not react with PUR to generate radicals (Figure S2). These properties underscore its potential utility as a temporary barrier to shield cells from harmful reaction products. An aqueous solution of sodium periodate (2.0 mM) was mixed with 1 w/v% solution of sodium alginate (Alg) and reacted in the dark for 24 h at ambient temperature. Two drops of ethylene glycol were added to stop oxidation. The resultant solution was precipitated in ethanol (2:1 v/v ethanol/water) and sodium chloride (6.25 g/L). Precipitates were dissolved in distilled water to the original volume, precipitated in ethanol solution, and dried under vacuum at room temperature. After drying, partially oxidized sodium alginate (o-Alg) was dissolved in distilled water, filtered, and lyophilized [28]. A concentration of 4 w/v% of o-Alg was utilized to generate hydrogel beads.

2.4. Encapsulation of cells in alginate beads

Cells (10⁵ cells/mL) were encapsulated in calcium alginate hydrogel by pumping the sodium alginate solution (1 w/v% for Alg and 4 w/v% for o-Alg) through a nozzle (0.35µm diameter) into a 100 mM calcium chloride crosslinking solution. An electronic bead maker (Nisco, VAR V1) was used to control bead size over the range 300 – 2000 µm by adjusting the potential difference between the nozzle and gelling agent solution [30]. Alginate bead size was measured by light microscopy.

2.5. Synthesis and characterization of polyurethane scaffolds

A polyester triol (900 g/mol) was synthesized from a glycerol starter and a backbone comprising 70 wt% ε-caprolactone, 20 wt% glycolide, and 10 wt% D,L-lactide as described previously [31]. An isocyanate (NCO)-terminated prepolymer (21,000 cP, NCO:OH equivalent ratio = 3.0:1.0, 21% NCO [32]) was synthesized by adding polyethylene glycol (PEG, 200 g/mol) drop-wise to lysine triisocyanate (LTI). Polyurethane (PUR) scaffolds were synthesized by reactive liquid molding of the prepolymer with a hardener component comprising the polyester triol, iron catalyst (5% iron acetylacetonate (FeAA) in dipropylene glycol), and alginate beads. The reactivity of the LTI-PEG prepolymer was measured by using ATR-FTIR (Bruker, Billerica, MA) to quantify the disappearance of the NCO peak [7]. Rheological properties of the scaffolds during curing process were measured with a parallel plate AR 2000ex rheometer in dynamic mode (New Castle, DE) to determine the working time (crossover point of storage moduli (G’) and loss moduli (G’’)).

For porosity, permeability, and mechanical measurements, scaffolds were prepared using o-Alg beads followed by 48 h incubation in PBS to dissolve the o-Alg and vacuum drying. Pore morphology and size distribution were determined by SEM (Hitachi, Finchampstead, UK). Porosity was calculated from mass and volume measurements of cylindrical scaffold cores (ρ_PUR = 1.27 g cm⁻³) [15]. Young’s modulus was determined from the slope of the stress-strain curve from compression tests performed using a TA Instruments Dynamic Mechanical Analyzer Q1000 (New Castle, DE). The flow rate of air through the scaffold was measured using a flowmeter and the permeability calculated as:

$$k=Q\frac{L}{A}\left(\frac{\mu }{\mathrm{\Delta }P}\right)$$ (1)

where Q = volumetric air flow rate, L and A are the scaffold thickness and cross-sectional area, μ is the viscosity of air, and ΔP is the pressure drop across the scaffold [33].

2.6. Effects of bead size on survival of encapsulated cells during polymerization

The ability of cells to survive the polymerization was evaluated at 30 min after mixing of the polyisocyanate and polyester triol components. At this early time point, conduction of heat and diffusion of CO₂ into the beads, which occur on the time scale of minutes, were anticipated to be the primary regulators of cell survival. Alg beads (500 – 2000 µm) containing cells were added to the LTI-PEG prepolymer, polyester triol, and catalyst (0.26 wt% FeAA) at a loading of 50%. Beads were removed from the scaffolds at 30 min post-mixing using forceps, washed with Dulbecco’s Phosphate Buffered Saline (DPBS, Corning, Corning, NY), and stained with the Cytotoxicity Kit (Live/Dead® Viability/Cytotoxicity Kit for mammalian cells, Invitrogen). An inverted confocal microscope (Zeiss LSM 510) was used to capture a series of Z-stack images of the 3D beads. The number of live (N_live, green) and the total number of cells (N_total, the number of all stained cells, including both live and also yellow, orange, and red dead or damaged cells) in each image were counted. Cell viability was calculated as [34]:

$$\%\text{Viability}=\frac{N_{\mathit{\text{live}}}}{N_{\mathit{\text{total}}}}\times 100\%$$ (2)

2.7. Survival of encapsulated cells at early time points prior to gelation

Preliminary experiments revealed evidence of acute cell death when cells were encapsulated in 500 µm beads, presumably due to transport of PUR reaction products into the beads. Therefore, the polyisocyanate composition (LTI-PEG or HDIt), timing of 500 µm bead addition (0, 3, or 6 min delay), and catalyst concentration (0, 0.26, or 0.52 wt% FeAA) were varied to control the amount of cell exposure to heat and CO₂. The study design is summarized in Table 1. The viability of cells embedded in non-reactive controls with no catalyst (L-0C-0) was also measured to decouple any effects of chemical toxicity from loss of cell viability due to reaction-generated heat and CO₂. Beads containing cells were removed from the cured scaffolds at 10 min using forceps and analyzed for cell viability (V₁₀) as described above.

Table 1.

Experimental conditions for measurement of early-stage (10 min) cell survivability. Values of n_CO2,10 (CO₂ generated by the reaction at 10 min) and Q₁₀ (heat generated by the reaction at 10 min) were calculated from the PUR reaction kinetics model. Alginate beads were removed from the scaffolds 10 min after the start of mixing and cell viability measured (V₁₀).

Treatment Group	Isocyanate	FeAA catalyst wt%	Delay min	nCO2,10 mmol cm−2	Q10 J cm−2	Rate Const. g eq−1 min−1	kG/kW
L-0C-0	LTI-PEG	0%	0	0	0	N/A	N/A
L-5C-0	LTI-PEG	0.26%	0	0.107	0.392	kG = 12.1 kB = 1.9	6.4
L-5C-3	LTI-PEG	0.26%	3	0.085	0.380
L-5C-6	LTI-PEG	0.26%	6	0.063	0.367
H-5C-0	HDIt	0.26%	0	0.044	0.317	kG = 8.2 kB = 0.61	13.4
H-10C-0	HDIt	0.52%	0	0.036	0.655	kG = 31.4 kB = 0.68	46.2

2.8. Survival of encapsulated cells at later time points after gelation

At later time points after gelation, both exposure to reaction products as well as the permeability of the scaffold, which controls transport of nutrients and wastes into the scaffold, could limit cell survivability. Therefore, cell survivability was investigated as a function of bead loading (50 or 70% 500 µm Alg beads) and timing of bead addition (0 or 3 min delay) at 10 min (prior to gelation), 30 min (at gelation), and 3 h (after gelation) post-mixing. At each time point, Alg beads containing cells were removed from the cured scaffolds and stained with an Apoptotic & Necrotic Cell Differentiation kit (PromoCell GmbH). Apoptotic cells were identified with fluorescein- (FITC, green) labeled Annexin V, necrotic cells were identified with a positively charged nucleic acid probe Ethidium homodimer III (EthD-III, red), and Hoechst 33342 (blue) was used to identify the total number of cells. An inverted fluorescence microscope (Olympus CKX41) was used to identify healthy cells (blue) as well as cells entering apoptosis (blue and green) or necrosis (blue, green, and red). The percentage of cells entering necrosis or apoptosis was calculated using Eq. (2). In another test group, the porosity and permeability of the scaffolds were measured (except for the 70% immediate addition group, which were too friable to be tested).

2.9. Culture of cellularized PUR scaffolds

Rat BMSCs were stained with a cytoplasmic dye (VyBrant^® CFDA SE Cell Tracer Kit, Life Technologies, per the manufacturer’s guidelines), encapsulated in Alg or o-Alg beads, embedded in PUR scaffolds, and cultured in 48-well tissue culture plates for 1, 4, and 7 days. Scaffolds were rinsed with DPBS and fixed with 5% glutaraldehyde or 2% OsO₄ solution before vacuum drying for SEM imaging. A subset of scaffolds was also sectioned (30µm) for microscopic imaging to observe cell viability and attachment to the scaffolds, which were cut open to expose the interior. Scaffolds were fixed in 10% formalin for 15 minutes, washed in PBS, and dehydrated through increasing alcohol concentrations (50–100%). Materials were air-dried and mounted to a specimen stub using carbon tape. Samples were sputter-coated with gold (108 Auto Sputter Coater; TedPella, Redding CA) and viewed via SEM (Carl Zeiss VP-40; Oberkochen, Germany). The ability of the MSCs to retain pluripotency after embedding in the scaffolds was determined by measuring adipogenic and osteogenic differentiation. Scaffolds were maintained in growth, adipogenic, or osteogenic media for up to 21 days and stained with Oil Red O or Alizarin Red S. After staining, dyes were dissolved in appropriate solvents (100% isopropanol for Oil Red O and 5% SDS for Alizarin Red S) and absorbances of the solutions were read on a plate reader (OD 490nm for Oil Red O and OD 570nm for Alizarin Red S). Absorbances were compared to stained scaffolds cultured in growth media.

2.10. In vivo cutaneous repair in rats

All surgical and care procedures were carried out under aseptic conditions per an approved Institutional Animal Care and Use Committee (IACUC) protocol. Scaffolds (n=4) that contained encapsulated male rat BMSCs were injected into 10 mm full-thickness excisional wounds in the dorsal skin of adult female Sprague-Dawley rats and allowed to cure for 15 min [16]. BMSCs encapsulated in Alg beads and embedded in PUR scaffolds did not integrate with host tissue and were consequently ejected from the wound bed. Therefore, only scaffolds containing o-Alg beads were evaluated. Injectable scaffolds containing no cells (Inj group) and implanted, pre-formed scaffolds seeded with cells (Impl+BMSC group) were both evaluated as controls compared to injectable scaffolds with cells (Inj+BMSC). Rats were euthanized 4d and 7d after surgery and the wounds harvested for histology and qRT-PCR. RNA from each sample was isolated and purified by RNeasy Mini Kit (Qiagen). cDNA synthesis was carried out from purified total RNA using iScript™ Reverse Transcription Supermix (Biorad). RT-PCR amplified for rat SRY gene (5′ -CATCGAAGGGTTAAAGTGCCA-3 ′, 5′ -ATAGTGTGTAGGTTGTTGTCC-3 ′) was measured to track the fate of delivered cells. Gomori’s trichrome staining and Ki67 and collagen IV immunostaining were performed on tissue sections for tissue infiltration, cell proliferation and angiogenesis analysis, respectively. The ROI (region of interest) for quantitative analysis of tissue infiltration comprised a rectangle centered between the midpoint and the edge of the excisional wound. The reactivity was expressed as the percentage of area occupied by immunoreactive cells.

2.11. Statistical analysis

The statistical significance between experimental groups was determined by a two-factor ANOVA. Graphs show mean ± standard deviation. p≤0.05 was considered statistically significant.

3. Results

3.1 Reactivity and settability

Injectable reactive liquid precursors that cure in situ to form a solid scaffold should ideally be amenable to flow through a small-bore needle and set within clinically relevant gelation times to form a polymer network with suitable mechanical properties [35–37]. The gel point approximates the working time available for injection, since beyond the gel point the mixture is no longer flowable. For a prepolymer with functionality of 4 and a polyol with functionality of 3, the gel point occurs at ξ_GP = 38% conversion of the reactive NCO groups [38], which was achieved at 25.5 min when cell-containing beads were immediately added to the reactive PUR mixture (Figure 2A). Since CO₂ generated during the polymerization might harm the cell-loaded beads added to the reactive mixture, the effects of delaying the addition of the beads were also investigated. When addition of beads to the reactive PUR was delayed for 3 min, the gel point decreased to 19.5 min (Figure 2A). The working time can also be determined by the G’-G” crossover point (Figure 2B, C), at which point the storage modulus (G’) equals the loss modulus (G”). For delayed addition, the crossover point occurred at 22 min, which is comparable to that determined from chemical reaction kinetics (Figure 2A). In contrast, for immediate addition, the crossover point occurred at 33 min (both G’ and G” decreased with time at early time points due to significant volumetric expansion of the scaffold as a result of CO₂ generation (Figure 2C)).

Figure 2. Handling properties of injectable and settable PUR scaffolds.

(A) Overall NCO conversion for immediate and delayed (3 min) addition of alginate beads. The gel point (working time t_w) occurred at 38% NCO conversion (dashed line), which corresponded to 19.5 min for delayed addition and 25.5 min for immediate addition. (B–C) Storage (G’) and loss (G”) moduli versus time for delayed (3 min, B) and immediate (0 min, C) addition of alginate beads. The value of t_w is defined as the G’-G” crossover point (22 min for delayed addition).

3.2 Effects of bead size on cell survival at gelation

MC3T3 cells were encapsulated in Alg beads that were immediately mixed (0 min delay, 50% loading) with the reactive PUR. Beads were harvested from the scaffolds after 30 min and stained for live and dead cells. For 500 – 2000-µm diameter beads not embedded in PUR, the viability of encapsulated MC3T3 cells exceeded 95% and was independent of bead size (Figure 3A,C). However, when embedded in the reactive PUR (Figure 4B), cell viability decreased with decreasing bead size (Figure 3C). These observations suggest that transport of heat and/or CO₂ generated by the PUR reaction reduced cell survival prior to gelation.

Figure 3. Effects of bead size on survival of MC3T3 cells encapsulated in Alg beads and embedded in injectable PUR scaffolds at early time points (10 min post-mixing).

(A) Confocal images show viable (green) cells in 500 µm beads. (B) Viability decreases when viable (green) MC3T3 cells encapsulated in Alg are immediately embedded in PUR scaffolds prepared from LTI-PEG (L-5C-0 group, Table 1). Immediate embedding of Alg beads in PUR scaffolds resulted in significant cell death (yellow, orange, and red cells) near the surface of the beads. (C) The viability of encapsulated cells immediately embedded in PUR scaffolds correlated with bead size, suggesting that transport of reaction products into the beads was responsible for the observed cytoxicity.

Figure 4. Effects of heat and CO₂ released by the PUR reaction on viability of MSCs encapsulated in Alg beads and embedded in a reactive hydrophobic polymer at early time points.

(A) Plot of the moles CO₂ generated by the PUR reaction (n_CO2, calculated from the reaction kinetics model) as a function of time for up to 10 min. (B) Plot of the heat generated (Q, calculated from the reaction kinetics model) as a function of time for up to 10 min. (C) Contour plot showing V₁₀ as a function of CO₂ (n_CO2,10) and heat (Q₁₀) generated at 10 min. Red stars represent the data points and the surface was plotted from the fit to the experimental data shown on the plot.

3.3 Effects of delayed addition of MSCs on acute cell survivability at early time points

As shown in Figure 3, survival of MC3T3 cells encapsulated in 500 µm Alg beads was only 30%, presumably due to exposure of cells to PUR reaction products. The timing of bead addition (0, 3, or 6 min delay), the isocyanate composition (LTI-PEG or HDIt), and the catalyst concentration (0, 0.26, or 0.52 wt% FeAA) were varied to control the amount of heat (Figure 4A) and CO₂ (Figure 4B) generated by the PUR gelling (k_G) and blowing (k_B) reactions. The NCO groups in the polyisocyanate (R₁-NCO) react with hydroxyl groups (OH) in the polyester triol (R₂-OH) by the gelling reaction or in water (W) by the blowing reaction:

(3)

The amounts of heat (q_G or q_B) and CO₂ generated by Reaction (3) at 10 min were calculated from a PUR reaction kinetics model [7] (Q₁₀ and n_CO2,10 listed in Table 1). Cell viability was measured at 10 min (V₁₀) to test the hypothesis that transport of heat and/or CO₂ is the primary cause of acute cell death prior to gelation. The rates of the second-order gelling (r_G) and blowing (r_B) reactions are given by:

$$r_G=k_G{C}_{NCO}{C}_{OH,P}$$ (4)

$$r_B=k_B{C}_{NCO}{C}_{OH,W}$$

where C_NCO is the concentration of NCO groups in the prepolymer (eq g⁻¹) and C_OH is the concentration of OH groups (eq g⁻¹) in the polyester triol (P) or water (W). The specific reaction rates k_G and k_B (Table 1) were calculated from kinetic experiments in which the polyisocyanate was reacted with either the polyester triol (k_G) or water (k_B) and the disappearance of the NCO peak monitored by ATR-FTIR over time [7, 39].

The concentration profiles of each component were calculated as a function of time by modeling the system as a constant-volume isothermal batch reactor, since the increase in temperature in the bulk scaffold was <15°C [15]. The equivalent balance equations for polyester triol and water were solved C_OH,P and C_OH,W using the ode45 function in MATLAB:

$$\frac{d{C}_{OH,P}}{dt}=-r_G{M}_{PUR},t=0,C_{OH,P}=C_{OH,P0}$$

$$\frac{d{C}_{OH,W}}{dt}=-r_B{M}_{PUR},t=0,C_{OH,W}=C_{OH,W0}$$ (5)

where M_PUR is the mass of the PUR component (polyisocyanate and polyester triol) and C_{OH, P0} and C_OH,W0 denote the initial concentrations (eq g⁻¹) of polyester triol and water, respectively (details of how these parameters were determined are described in the Supplemental Information).

The heat generated by the gelling and blowing reactions as a function of time was normalized by the total alginate (A) bead area (Q, J cm⁻²). The CO₂ generated by the blowing reaction was also normalized by the alginate bead area (mmol CO₂ cm⁻²):

$$Q=\frac{\mathrm{\Delta }{H}_{Rx}{a}_A{\rho }_A(1-x_A)}{3x_A}(\frac{C_{OH,P0}-C_{OH,P}}{f_P}=\frac{C_{OH,W0}-C_{OH,W}}{f_W})$$

$$n_{c{o}_2}=\frac{a_A{\rho }_A(1-x_A)}{3x_A}\left(\frac{C_{OH,W0}-C_{OH,W}}{f_W}\right)$$ (6)

where f is the functionality (eq mol⁻¹), ΔH_Rx = 80 kJ mol⁻¹ is the heat of reaction [40], a_A is the radius of the alginate beads, ρ_A = 1.601 g cm⁻³ is the density of alginate, and x_A is the weight fraction of alginate beads in the scaffold. The values of Q and n_CO2 are plotted versus time in Figure 4A–B. The amounts of heat (Q₁₀) and CO₂ (n_CO2,10) generated at 10 minutes are listed in Table 1. The effects of Q₁₀ and n_CO2,10 on cell viability (V₁₀) are shown in the contour plot in Figure 4C. The values of V₁₀ were fit to the following equation to generate the contour plot:

$$V_{10}=87.7-0.0883\text{exp}(56.35n_{c{o}_2,10})-1.34\text{exp}(3.76Q_{10})$$ (7)

Mixing the beads with a non-reactive PUR mixture (LTI-PEG or HDIt) reduced the viability to 88%, which is about 10% less than that measured for the beads alone. Thus, for the region bounded by Q₁₀ < 0.4 J cm⁻² and n_CO2,10 < 0.08 mmo cm⁻², the effects of the chemical reaction on cell viability were negligible. However, outside this range, V₁₀ decreased exponentially with Q^*₁₀ and n^*_CO2,10. Taken together, these data indicate that CO₂ diffusion and heat conduction into the beads contributed to acute cell death prior to gelation.

3.4 Effects of permeability on cell survival at later time points after gelation

After gelation (20 – 30 min, Figure 1), the reactive PUR cures to form an elastomeric scaffold. Permeability (eq (1)) and porosity are key parameters controlling the rate of transport of nutrients into the scaffold. Thus, the effects of bead loading and the timing of bead addition on the porosity, permeability, and mechanical properties of the PUR scaffolds were investigated. SEM images comparing scaffolds prepared by 3 min delayed addition of o-Alg beads at 50 wt versus 70 wt% (Figure 5A–B) supported this hypothesis and showed that pore connectivity increased with bead loading. As bead loading increased from 50 and 70 wt% (3 min delay), the increase in porosity was not significant (78 – 82%), but the air permeability increased five-fold (p<0.05, Figure 5C) to values comparable to those reported for open-pore PUR foams with similar densities [33]. In contrast, when the beads were added immediately (0 min delay), neither permeability nor porosity increased with bead loading (Figure 5D). This observation suggests that CO₂ gas foaming controlled porosity and permeability when the beads were immediately added to the PUR. The elastic modulus (E^*) of scaffolds prepared by immediate or delayed addition of beads followed the predicted scaling with porosity ε (Figure 5E) [41]:

$$E^{*}=E_s{\left(\frac{1-{\rho }_s\epsilon }{{\rho }_s}\right)}^2$$ (8)

where the density of the bulk polymer ρ_s = 1.27 g cm⁻³ and the modulus of the bulk polymer E_s = 2.5 MPa.

Figure 5. Effects of bead loading and timing of bead addition on PUR scaffold properties.

(A–B) Representative SEM images of scaffolds fabricated with (A) 50 wt% o-Alg beads and (B) 70 wt% o-Alg beads. (C, D) Porosity and permeability of PUR scaffolds as a function of o-Alg bead loading for (C) delayed (3 min) and (D) immediate addition. (E) The elastic modulus of the scaffolds prepared by delayed and intermediate addition of beads decreased with bulk porosity.

The data in Figures 2 – 4 point to transport of heat or CO₂ into the Alg beads as a key factor contributing to acute cell death prior to gelation. After gelation, cells may undergo apoptosis or necrosis due to the continuing effects of the chemical reaction and/or hindered transport into the interior of the scaffold. To investigate the relative contributions of the chemical reaction and scaffold permeability to cell survivability, Alg beads were removed from the cured scaffolds at 10 min (prior to the gel point), 30 min (at the gel point), and 3 h (after the gel point) using forceps and stained to identify apoptosis and necrosis. Plots of the percentage of cells undergoing apoptosis (Figure 6A) or necrosis (Figure 6B) versus time reveal that the number of cells entering apoptosis or necrosis did not change substantially versus time at 70 wt% bead loading. In contrast, at 50 wt% loading, >45% of the cells entered apoptosis or necrosis at 30 min post-mixing. For the immediate addition group, % apoptosis (or necrosis) decreased slightly at 3 h, while for the delayed addition group % apoptosis (or necrosis) continued to increase. As shown in the contour plots (Figure 6C–D), the percentage of cells entering apoptosis or necrosis increased with increasing reaction time and decreasing permeability. As anticipated, permeability exhibited only a modest effect on % apoptosis (or necrosis) at 10 min post-mixing, since the scaffold had not yet formed. However, at 3 h % apoptosis increased with permeability, approaching 50% at the lowest permeability. These observations point to both chemical reaction products and scaffold permeability as key factors limiting cell survivability.

Figure 6. Effects of bead loading and timing of bead addition on cell viability at late time points.

The percentage of cells undergoing apoptosis (A) or necrosis (B) was measured at 10 min, 30 min, and 3 h as a function of bead loading (50 or 70 wt% Alg beads) and timing of addition (immediate or delayed for 3 min). (C–D) Contour plots showing the percentage of cells undergoing apoptosis (C) or necrosis (D) as a function of reaction time (10 min, 30 min, or 3 h) and scaffold permeability. Cell survivability decreases with increasing reaction time and decreasing permeability.

3.5 Long-term culture of encapsulated cells in vitro

The ability of BMSCs to attach to the scaffold in vitro was investigated for both Alg and o-Alg beads for the conditions at which cell survivability was highest: 70% loading and 3 min delayed addition. The beads exhibited a folded surface morphology (Figure S4A) and nanoscale mesh size (19 ± 3 nm for Alg and 65 ± 3 nm for o-Alg beads [42, 43]) consistent with findings from previous studies [44]. After encapsulation, cells exhibited a rounded morphology (Figure S4B), since there are no adhesive ligands to facilitate attachment to the Alg. After 7 days in culture, cells (stained green with a cytoplasmic dye) remained clustered within the Alg beads, with few cells appearing adjacent or adherent to the PUR scaffold (stained blue, Figure 7A), which suggests that cells could not escape the slow-degrading, nanostructured mesh. In contrast, PUR scaffolds embedded with o-Alg beads showed evidence of cell release from the beads and increasing numbers of cells lining the PUR surface with time. Similarly, SEM analysis showed that cells were attached to the surface of PUR scaffolds embedded with o-Alg beads (Figure 7B). Thus, loading the scaffold with 70% o-Alg beads not only increased permeability and pore interconnectivity (Figures 6B–C), but also supported release of cells from the beads and consequent attachment to the scaffold. These observations are consistent with the notion that o-Alg is a temporary protective shield that degrades within 1 – 2 days (Figure S5), releasing the cells so they can attach to the PUR scaffold. To determine whether MSCs retained their pluripotency after the reaction, scaffolds with o-Alg beads were cultured in growth, adipogenic, or osteogenic medium for 21 days. Compared to cells cultured in growth medium, cells cultured in adipogenic medium showed higher Oil Red O absorbance, while cells cultured in osteogenic medium showed higher alizarin red dye absorbance (Figure 7C). Thus, after exposure to the chemical reaction, BMSCs retained their potential to differentiate to adipocytes or osteoblasts.

Figure 7. In vitro culture of BMSCs on injectable PUR scaffolds in vitro.

(A) Representative histological sections stained with the cytoplasmic dye carboxyfluorescein diacetate (CFDA, green) and DAPI (blue) of PUR scaffolds loaded with 70 wt% 500 µm Alg or o-Alg beads show viable rat BMSCs at days 1, 4, and 7. Cells are stained green and the scaffold is stained blue. (B) Representative SEM images of PUR scaffolds loaded with 70 wt% 500 µm o-Alg beads showed cells attached to the scaffold after 7 days, while few attached cells were observed for scaffolds loaded with Alg beads. (C) Osteogenic (measured by Alizarin red absorption) and adipogenic (measured by Oil Red O absorption) differentiation of BMSCs encapsulated in polyurethane foams.

3.6 In vivo delivery of BMSCs encapsulated in injectable PUR scaffolds

To investigate the ability of the cells to survive the injection and generate new extracellular matrix in vivo, a proof-of-concept experiment was performed in full-thickness excisional skin wounds in Sprague-Dawley (SD) rats [16]. BMSCs from male SD rats were delivered to wounds in female rats, and SRY (sex determining region Y, Sox9) immunohistochemical staining was performed to track transplanted cells. PUR scaffolds embedded with Alg beads with or without cells were extruded from the wounds after 7 days (data not shown) due to persistence of Alg and consequent low porosity. In contrast, transplanted BMSCs (10⁵ cells/ml equivalent to 2×10⁴ cells/scaffold) encapsulated in o-Alg and embedded in injectable (Inj+BMSC) or implantable (Impl+BMSC) PUR scaffolds survived for at least 7 days (Figure 8A,B). High-magnification (20×, Figure 9A,B) images from trichrome staining revealed degradation of o-Alg to form new pores throughout the scaffold (PUR, light gray), while some fragments of o-Alg (A, green acellular material) remained. New extracellular matrix (green tint, M) was deposited as early as day 4. The Inj+BMSC group showed significantly more deposition of new extracellular matrix at both time points compared to the Inj (injectable with no cells) or the Impl+BMSC (implanted scaffold seeded with cells) groups (Figure 9C). To investigate the mechanism by which transplanted BMSCs enhanced deposition of new matrix, we measured the prevalence of Ki67⁺ proliferating cells and deposition of collagen IV (a marker of angiogenesis) in Inj and Inj+BMSC scaffolds by immunohistochemical staining. Inj+BMSC scaffolds showed significantly more Ki67⁺ proliferating cells (Figure 9D) and increased collagen IV accumulation (Figure 9E) compared to Inj scaffolds. Taken together, these observations indicate that transplanted BMSCs not only survived the chemical reaction, but also stimulated cell proliferation and angiogenesis after transplantation in vivo.

Figure 8. Rat BMSCs encapsulated in 500 µm o-Alg beads embedded in PUR scaffolds survive transplantation for up to 7 days in a rat excisional wound model.

(A) SRY (sex determining region Y, Sox9) immunohistochemical staining revealed the presence of male donor rat BMSCs in wounds on female rats (black arrows) at day 7 (40× magnification). (B) qRT-PCR measurements of SRY expression show that cells survived for up to 7 days in implanted (Impl+BMSC) and injected (Inj+BMSC) scaffolds.

Figure 9. Rat BMSCs encapsulated in 500 µm o-Alg beads embedded in PUR scaffolds enhance deposition of new extracellular matrix in a rat excisional wound model.

(A–B) High-magnification (20×) images of histological sections 7 days after injection of PUR scaffolds without (A) or with (B) 10⁵ rat BMSCs/ml encapsulated in o-Alg beads into 10-mm excisional wounds in rats. Local cell delivery increased deposition of new extracellular matrix (M). O-Alg beads (A) degraded to form macropores (P), resulting in infiltration of cells and ingrowth of granulation tissue along the surface of the residual polyurethane (PUR) scaffold. (C) Histomorphometric analysis showed that Inj+BMSC scaffolds supported significantly greater ingrowth of extracellular matrix at days 4 and 7 compared to the injected acellular (Inj) and cellular implant (Impl+BMSC) controls. (D–E) Ki67⁺ proliferating cells (D) and collagen IV (E) are higher in Inj+BMSC scaffolds at days 4 and 7 compared to the acellular Inj control. * denotes significant differences between the blank and BMSC groups, p < 0.05.

4 Discussion

In this study, we designed injectable PUR scaffolds for local transplantation of viable cells for tissue repair and restoration by encapsulating cells in degradable o-Alg beads prior to embedding in the reactive polymer. In contrast to hydrogels that utilize water-soluble initiators [17, 18] and macromers [19–21] to facilitate cell encapsulation from aqueous suspensions, direct encapsulation of cells in reactive hydrophobic polymers is confounded by their low (<5%) swelling in water and generation of chemical by-products and heat [7]. Two factors limited cell survivability in vitro: (1) generation of CO₂ and heat by the chemical reaction prior to gelation, and (2) permeability of the scaffolds after gelation. Delayed (3 min) addition of the o-Alg beads at a loading of 70% balanced the requirements for minimal exposure of cells to reaction products, high permeability for transport of nutrients and wastes, and mechanical integrity of the scaffolds. Under these conditions, PUR scaffolds injected with encapsulated BMSCs promoted increased extracellular matrix deposition in vivo compared to both injected acellular scaffolds and implanted scaffolds seeded with BMSCs, and they did so without biofunctionalization of the scaffold with expensive peptides, growth factors, or other biologics.

Encapsulation of cells in Alg beads of sufficient size provided a barrier to diffusion of CO₂ and heat prior to gelation (10 min). This observation is consistent with a previous study reporting that acellular PUR scaffolds reach the reaction exotherm at 3 min post-mixing [15]. While Alg protected the cells from the chemical reaction prior to gelation, the persistence of Alg after gelation hindered attachment of cells to the scaffolds in vitro (Figure 7A–B) and tissue ingrowth in vivo. These observations are in agreement with a previous study reporting that cells encapsulated in Alg beads and embedded in a CPC failed to release from beads after 14 days in culture [34]. Thus, slow dissolution of Alg beads precludes the formation of interconnected macropores (>10 µm) [36, 45]. Partial oxidation to o-Alg renders it susceptible to hydrolysis [28, 46], which has prompted the use of o-Alg as a degradable carrier for MSCs. Delivery of human adipose stem cells from o-Alg hydrogels with a degradation time of ~40 days promoted generation of new adipose tissue in mice [47]. In another study, MSCs encapsulated in o-Alg beads and mixed with a calcium phosphate cement (CPC) provided mechanical protection during mixing [48, 49]. However, the utility of o-Alg as a temporary barrier to diffusion of harmful chemical reaction products, which was the subject of the present study, has not been systematically investigated.

Considering the optimal pore size of 90 – 360 µm reported for cellular infiltration and new tissue ingrowth [50], the diameter of the o-Alg beads was initially targeted at ~350 µm. However, when 500 µm beads were immediately mixed with the reactive polymer, only 30% of the cells survived at 10 min (Figure 3B–C). As shown in Figure 4C, generation of both CO₂ and heat outside the region bounded by Q₁₀ > 0.4 J cm⁻² and n_CO2,10 > 0.08 mmol cm⁻² (calculated at 10 min from the chemical kinetics) resulted in excessive cell death. Delayed addition of o-Alg beads reduced CO₂ generation below 0.08 mmol cm⁻², thereby increasing acute survivability of cells encapsulated in 500 µm beads to levels exceeding 80% (Figure 4C). These observations are consistent with a previous study reporting that the viability of cells encapsulated in fibrin-alginate beads embedded in an injectable CPC decreased as the concentration of NaHCO₃ (reacting with citric acid to produce CO₂) increased from 15 to 30% [49]. Considering that the bicarbonate-citric acid reaction is endothermic, cell death in this previous study was likely caused by CO₂.

An important unanswered question is whether the cells die in response to a cumulative increase in temperature (or CO₂ concentration) or the rate at which these parameters are changing. Quantifying the relative contributions of CO₂ and heat generation to cell death both prior to and after gelation requires solution of the unsteady state heat conduction and CO₂ diffusion equations for both Alg (A) and polymer (PUR) phases [51]:

$$\frac{\partial T_A}{\partial t}={\alpha }_A{\nabla }^2{T}_A$$

$$\frac{\partial T_{PUR}}{\partial t}={\alpha }_{PUR}{\nabla }^2{T}_{PUR}+\frac{Q(t)}{{\rho }_{PUR}{C}_{P,PUR}}$$

$$\frac{\partial c_{c{o}_2,A}}{\partial t}=D_{c{o}_2,A}{\nabla }^2{c}_{c{o}_2,A}$$

$$\frac{\partial c_{c{o}_2,PUR}}{\partial t}=D_{c{o}_2,PUR}{\nabla }^2{c}_{c{o}_2,PUR}+r_{c{o}_2,PUR}$$ (9)

where α = κ/ρC_p is the thermal diffusivity, κ is the thermal conductivity, C_p is the heat capacity, Q(t) is the heat generated by the chemical reaction (Eq. 6), c_CO2 is the concentration of carbon dioxide, and D_CO2 is the diffusivity of CO₂. Both D_CO2,A and α_A are anticipated to increase with increasing mesh size, resulting in steeper temperature and CO₂ gradients in the beads. The exact solution of the unsteady state heat conduction and CO₂ diffusion equations is outside the scope of this study, but several observations can be made from the apoptosis/necrosis kinetic data (Figure 6C–D). The percentage of cells entering apoptosis or necrosis increased from 10 min to 3 h for all scaffolds, including highly permeable (>2×10⁻¹⁰ m²) PUR scaffolds with minimal transport limitations, suggesting that the cells did not recover from the initial exposure to heat and CO₂. Furthermore, the percentages of apoptotic and necrotic cells were comparable at all time points and permeabilities. The majority of damaged cells stained positive for both apoptotic and necrotic markers, which further confirms that cells did not recover from the initial exposure to reaction products. Finally, cell survival after gelation (30 min – 3 h) improved dramatically in highly permeable (>2×10⁻¹⁰ m²) scaffolds. These observations suggest that exposure to heat and CO₂ regulates cell survival prior to gelation, while scaffold permeability controls cell survival after gelation.

Potential hurdles to clinical translation of the injectable hydrophobic scaffold approach include the requirements of specialized encapsulation equipment and delayed addition of the o-Alg beads, which increases the duration of the surgical procedure. In this study, the maximum delay for bead addition was 6 min, beyond which time the beads could not be uniformly mixed with the reactive PUR. The maximum delay is determined by the working time, which was targeted at 20 min to be consistent with the handling properties reported for calcium phosphate cements [37], a clinically relevant class of injectable and settable biomaterials. As an alternative to delayed addition, the rate of CO₂ generation can be controlled by tuning the gel:blow (k_G/k_B) ratio. For the LTI-PEG prepolymer and FeAA catalyst used in this study, the gel:blow ratio was 6.4 (Table 1), which is substantially greater than the value of ~0.05 reported for a triethylene diamine catalyst [7] but not large enough to obviate the need for delayed addition of the beads. In contrast, HDIt exhibited a gel:blow ratio of 13.4 at the lowest catalyst level (Table 1), which was sufficiently high that delayed addition of the beads was not required to achieve high viability. These observations point to the gel:blow ratio as a key parameter for maintaining high cell survivability without delayed addition of the beads. The adverse effects of the polymerization on cell survivability could be further reduced by slowing the gelling reaction, which would decrease the rate of heat generation. However, the advantageous effects of slowing the gelling reaction on cell survival at early time points must be balanced against the potentially adverse effects of longer in situ setting times on both the handling properties as well as the mechanical stability of the PUR scaffold after injection.

In a proof-of-concept in vivo experiment, injected scaffolds showed comparable cell survival to implanted scaffolds (Figure 8B). However, the Inj+BMSC group showed significantly more new granulation tissue compared to both the Impl+BMSC and Inj (no cells) groups (Figure 9C). To investigate the mechanism by which transplanted BMSCs enhanced healing, we measured the number of Ki67⁺ proliferating cells and deposition of collagen IV, a basement membrane protein that marks capillary endothelium in granulation tissue that forms within embedded scaffolds. The area% of Ki67⁺ cells was significantly higher in the Inj+BMSC group compared to the Inj group on days 4 and 7 (Figure 9D). Furthermore, the area% collagen IV was significantly higher in the Inj+BMSC group on days 7 and 14, despite the fact that the transplanted cells survived for only 7 days. These observations are consistent with the notion of trophic activity, by which MSCs influence healing by the secretion of growth factors and cytokines that stimulate proliferation of tissue-intrinsic progenitor cells as well as angiogenesis [52, 53]. Using this adaptable and versatile PUR carrier, BMSCs can be encapsulated in o-Alg beads directly after harvesting, mixed with the reactive polymer, and injected into defects of varying sizes and complex shapes as a site-directed therapeutic [54].

5 Conclusion

Injectable PUR scaffolds embedded with bone marrow-derived MSCs encapsulated in o-Alg were designed to promote peripheral tissue infiltration in rat subcutaneous wound model. MSCs were encapsulated in o-Alg before the PUR reaction to enhance cell survivability. After incorporation, o-Alg beads subsequently degraded to form interconnected macropores that supported cellular migration, proliferation, and deposition of new extracellular matrix in vitro and in vivo. These properties underscore the potential utility of PUR scaffolds as a versatile, clinically relevant, and functionally-significant injectable cell delivery system for regenerative medicine applications.