Cell-loaded carboxymethylcellulose microspheres sustain viability and proliferation of ATDC5 cells

Abstract

Cell-loaded carboxymethylcellulose (CMC) microspheres were generated via a flow focusing microfluidic device, with a final aim to obtain viable ATDC5 aggregates with sustained proliferation capacity. We synthesized various CMC with phenolic groups (CMC-Ph) and demonstrated that high CMC-Ph molecular weight, high CMC-Ph concentration (>0.8 g/ml) or long culturing period had obvious inhibition effect on ATDC5 proliferation, but low horseradish peroxidase concentration (HRP, <0.4 mg/ml) did not. CMC-Ph gels being obtained through HRP/H₂O₂ reaction showed an enhancing strength and decreasing break stain as the molecular weight of CMC-Ph increased, along with a decreasing gelation time. The microfluidics-based synthesis of cell-loaded microspheres with great design flexibilities was achieved using CMC-Ph with weight-average molecular weight of 1.0 × 10⁵. ATDC5 cells were viable up to 41 days of culture and proliferated gradually with increasing culture time. High cell density in CMC-Ph solution and high fetal bovine serum concentration in culture medium were prone to forming cell aggregates. Isolated cells from the microspheres showed the typical spherical morphology of undifferentiated ATDC5. Therefore, CMC-Ph microspheres might be used as cell aggregates depots to study cell-cell or cell-biomaterials functions for tissue engineering applications.

Graphical Abstract

Keywords:

• Carboxymethylcellulose
• cell-loaded microspheres
• fast gelation
• horseradish peroxidase
• ATDC5
• microfluidic

Introduction

Cells are often regulated by their integrated response to various signals that are essential to redirecting the cell behaviors in two-dimensional culture because some cells such as chondrocytes are prone to de-differentiate and lose the ability to synthesize specific markers [1,2]. ATDC5, derived from AT805 teratocarcinoma cell line, is well acknowledged as an in vitro chondrogenic model due to characteristics of easy culture and rapid proliferation [3]. More importantly, it retains chondroprogenitor cell properties and exhibits nearly the same characteristic of chondrogenesis as mesenchymal stem cell [4,5]. Chondrogenesis of ATDC5 can be induced via chondrogenic medium while insulin has been in favor of promoting proliferation and expression of endochondral ossification related genes [6].

Cell encapsulation in semipermeable membranes has attracted much attention in the field of regenerative medicine, because it provides a feasible microenvironment for cells to respond to soluble factors, extracellular matrix mediated signals and cell-cell interaction [7–10]. It mimics the morphology and physiology of the cells in living tissues and organs better than two-dimensional cultures. A variety of three-dimensional cellular spheroids with controlled differentiation can be made via cell-encapsulation to investigate cell behavior and intercellular interactions [11,12]. Cell spheroids can decline the de-differentiation of some cell lines, such as hepatocytes, to enhance liver-specific functions because of the extensive cell-cell contacts and tight junctions to mimic the morphology and ultrastructure of native liver lobule [13,14].

Carboxymethylcellulose (CMC) is a biodegradable and biocompatible macromolecule, and has been employed for cell-loaded microspheres via forming electrolyte complexes with cationic materials [15–17], though this electrolyte complex is not stable enough in vivo. CMC do not have natural binding sites or essential cues that cells can probe their extracellular surroundings to allow for complex cell-matrix interactions. In the absence of such cues, some cells can excrete their own extracellular matrix to provide adhesion sites and prevent unwanted cell stimulation [18]. The carboxylic or hydroxyl groups of CMC are highly appreciated for physisorption [19] and covalent coupling [20,21] of bioactive substance, providing the means for stable cellulose-based biointerfaces. Tyramine-modified CMC has been used to enclose mammalian cells in gel sheet via a peroxidase-catalyzed oxidative reaction [22]. The contents of phenol moieties influenced cell adhesion and proliferation [23], as well as cell differentiation because strong adhesive cell-substrate interaction could induce hepatocyte de-differentiation with an extended and spread morphology [15]. Our lab had synthesized CMC containing phenolic groups (CMC-Ph) by covalently grafting 4-hydroxybenzylamine on CMC, and produced monodispersed CMC-Ph microspheres via a coflowing microfluidic device [24]. In this work, we synthesized various CMC-Ph with different molecular weight but similar grafting percentage of phenolic groups, and studied their gel properties and cell proliferation behaviors. Cell-loaded microspheres using CMC-Ph with low-molecular weight were prepared as a model to illustrate the effect of microsphere size, cell density and medium protein on the viability and proliferation of ATDC5.

Materials and methods

Materials

1-Ethyl-3–(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), N-hydroxysulfosuccinimide (NHS), 1-hydroxybenzotriazole hydrate (HOBT), 2–(4-morpholino) ethanesulfonic acid (MES), lecithin, horseradish peroxidase (HRP, 250 units/mg), hydrogen peroxide (H₂O₂, 30% w/w) and cellulase were obtained from Qiyun Biotech (Guangzhou, China). Liquid paraffin and 4-hydroxybenzylamine were purchased from Aladdin Chemical Reagent (China) and J & K Scientific (China), respectively. All reagents were analytical grade and used as received. ATDC5 cells were provided by National Human Organization Functional Reconstruction Engineering Technology Research Center (China). CMC with weight-average molecular weight (M_w) of 1.0 × 10⁵ (CMC₁₀), 2.0 × 10⁵ (CMC₂₀) and 3.0 × 10⁵ (CMC₃₀) were purchased from Jingchun Chemical Reagent (China).

Synthesis of CMC-Ph with different molecular weight

CMC-Ph was synthesized according to our previous study [24]: CMC and 4-hydroxybenzylamine were dissolved in MES buffer (50 mM, pH 6.0) at 1.0% and 0.8% (w/v), respectively. NHS, HOBt and EDC were added into the CMC solution at a weight ratio of 1:0.26:0.68:0.70 (CMC:NHS:HOBt:EDC). The solution was magnetically stirred for 24 h and dialyzed against deionized water using an ultrafiltration membrane (molecular weight cut-off = 3500 Da) at 25 °C. Four days later, UV absorption peak at 275 nm being attributed to 4-hydroxybenzylamine was not detected in the dialyzate. The resultant polymer was enriched by a rotary evaporator (100 rpm) at 50 °C and lyophilized. CMC₁₀-Ph, CMC₂₀-Ph and CMC₃₀-Ph represented the phenolic derivatives of CMC with different molecular weight (Table 1).

Table 1. Thermal decomposition temperatures (ºC) of CMC₁₀ and CMC-ph.

Sample	T5^a	T10^b	T50^c	Tp1^d	To^e	Te^f	Te–To
CMC10	69.1	126.1	308.0	292.8	251.9	313.2	61.3
CMC10-Ph	97.6	211.7	335.7	301.0	235.3	345.6	110.3
CMC20-Ph	88.9	199.3	314.6	303.5	235.3	345.6	110.3
CMC30-Ph	134.8	209.2	324.5	309.7	235.3	355.6	120.3

^a Thermal decomposition temperature at 5% weight loss of the films.

^b Thermal decomposition temperature at 10% weight loss of the films.

^c Thermal decomposition temperature at 50% weight loss of the films.

^d The first peak temperature in DTG of the films.

^e Onset thermal decomposition temperature of the first peak in DTG of the films.

^f End thermal decomposition temperature of the first peak in DTG of the films.

Fabrication of cell-loaded microspheres

Cryopreserved fifth-passage (P5) ATDC5 cells were thawed and cultured in Dulbecco’s Modified Eagle Medium (DMEM, Gibco, Grand Island, USA) with 10% Fetal Bovine Serum (FBS, Lifei Biotech, Shanghai, China) at 37 °C and 5% CO₂. At 90% confluence, the cells were rinsed in phosphate-buffered solution (PBS, pH7.4, Gibco, Grand Island, USA) and then passaged by 0.25% trypsinase supplemented with 0.02% ethylene diamine-N,N-tertraacetic acid (Gibco, Grand Island, USA). The cell suspension was centrifuged at 1200 rpm for 5 min, and cell pellets (P6) were resuspended in DMEM supplemented 10% FBS at 10⁶–10⁷ cells/ml via vortex.

Cell-loaded microspheres were produced via a droplet breakup technique. Two immiscible co-flowing fluids were injected into a metal-channel using separate micro-syringe pumps (Baoding Longer TS-1B/W0109–1B, Baoding, China). CMC₁₀-Ph was dissolved at 10% (w/v) in DMEM supplemented with 10% FBS, and mixed with cell suspension containing 0.4 mg/ml of HRP at a volume ratio of 1:1 (the dispersed phase). The continuous liquid paraffin phase containing H₂O₂ and lecithin was prepared as follows: 2 ml of aqueous H₂O₂ was added into 400 ml of autoclaved liquid paraffin, and the mixture was magnetically stirred and centrifuged. Lecithin was dissolved at 3.0% (w/v) into the collected upper liquid. In the co-flowing microfluidic device, the dispersed phase was driven and subsequently broke into droplets at the end of the inner channel. The mixture containing the microspheres and the continuous fluid were magnetically stirred at 100–150 rpm for 20 min and centrifuged at 1200 rpm for 5 min. The collected microspheres were then washed by PBS and re-suspended in DMEM supplemented 10% FBS. 500 μl Suspension of cell-loaded microspheres was added into each well of 24-well cell culture dish, and incubated in a humidified atmosphere at 37 °C under 5% CO₂. 100 μl of fresh medium was replenished every two days.

Characterization

Graft density of phenols was calculated by measuring the absorbance at 275 nm via a Thermo (Massachusetts, USA) Evolution 300 UV-VIS spectrometer and comparing with 4-hydroxybenzylamine standard curve. The data was defined as the ratio of phenols to 100 repeat units of CMC.

Thermogravimetric (TG) analyses of CMC-Ph were carried out using a Netzsch (Wittelsbacherstrasse, Germany) 209 F1 thermogravimetric analyzer. Thermogravimetry analysis (TGA) curves and derivative thermogravimetry (DTG) curves were recorded from room temperature to 800 °C under a nitrogen atmosphere at a rate of 10 °C/min. Differential scanning calorimetry (DSC) studies were performed on a Netzsch (Wittelsbacherstrasse, Germany) 204 F1 DSC instrument under a nitrogen atmosphere. All samples were heated from -50 to 250 °C at a rate of 10 °C/min.

Gelation time of CMC-Ph was estimated by a plate-tilting method. Briefly, 5% degassed CMC-Ph solution containing 10 mM H₂O₂ and 50 U/ml HRP was added into 24-well plate, and CMC solution was used as control sample. The cell plate was leaned at a certain angle, where the control CMC solution would just overflow. The sol to gel transition was timed as the minimum time period until no flow from the well plate was observed for CMC-Ph solution.

Rheological property of CMC-Ph gels was performed by using a Malvern Kinexus rheometer (Malvern, United Kingdom) with a parallel plate. Samples were allowed to equilibrate at 25 °C before loading. For each test, CMC gels with an approximate diameter of 10 mm and height of 3 mm were transferred onto the plate and measurements were taken over the strain from 0.01% to 100% at a constant frequency of 1 Hz.

Size distribution analysis of cell-loaded CMC₁₀-Ph microspheres was performed under a fluorescence microscope (Nikon Ti-U, Tokyo, Japan), and images were taken at random for diameter measurements of empty and cell-loaded microspheres. At least 300 microspheres were analyzed for each setting.

The viabilities of ATDC5 treated by CMC-Ph or HRP solution were measured by a cell counting kit (CCK-8, Dojindo Laboratories, Kumamoto, Japan). 100 μl of ATDC5 suspension was added to each well of 96-well plates (3000 cells/well) and cultured for 24 h in a controlled atmosphere (5% CO₂, 37 °C). Once the cells were adhered, 100 μl CMC₁₀-Ph or HRP in DMEM supplemented with 10% FBS was added into each well, respectively. At 1, 3, 5 and 7 days, 100 μl of WST-8 solution was added into each well. After 2 h incubation, the plates were monitored by a plate reader photometer (Thermo 3001, Massachusetts, USA) at 450 nm wavelengths, and cells without any treatment (0% CMC₁₀-Ph solution or 0% HRP solution) and the blank plate were used as controls. Mitochondrial activity of ATDC5-loaded microsphere was studied as follows: 500 μl of microsphere suspension was added to each well of 24-well plates with Millicell^® culture supports (Millipore, Darmstadt, Germany). At predetermined time, the culture medium was removed, and 500 μl of WST-8 solution was added into each well. Absorbance of each sample was the average of five separate experiments.

Live/dead cell viability assay (Lifei Biotech, Shanghai, China) was used to visualize cell viability. The working fluid was obtained according to the manufacturer’s instruction. Cell-loaded microspheres suspension was added to each well of 24-well plates with Millicell^® culture supports, washed with PBS, and stained with the working fluid for 30 min in the dark. The fluorescence images of the microspheres were observed using a fluorescence microscope (Nikon Ti-U, Tokyo, Japan).

Statistical analysis

Results represented mean ± SD of triplicates from five separate experiments. A Student’s t-test was used for the statistical analysis. A significance level of 0.05 was used in all the statistical tests performed.

Results and discussion

Thermal properties of CMC-Ph

Thermal properties of CMC-Ph were studied via weight loss as a function of temperature (Figure 1(I)) to understand their proper operating temperature for sterilization or other thermal-related clinical uses. CMC₁₀ lost about 11.0% of total weight due to the evaporation of physically absorbed water below 113.7 °C, and the dehydrogenation of glucose within 274.9–305.9 °C led to a weight loss of 35.7% with a peak decomposition temperature at 292.8 °C. An additional 11% weight loss followed in 305.9–498.2 °C, and the residual weight percentage was 32.2% at 800 °C. A series of gases such as H₂, CO₂, CO, CH₄, C₂H₆, C₂H₄, water vapor, or trace amounts of larger gaseous organics might be generated during the pyrolysis [25,26].

Figure 1. TG and DTG curves of CMC₁₀ and CMC-Ph with different molecular weight (I) and 4-hydroxybenzylamine (II).

4-Hydroxybenzylamine (Figure 1(II)) did not show distinct dehydration or volatilization below 70 °C, but about 12% weight loss till 115 °C with two peak decomposition temperatures at 83.9 °C and 106.6 °C. Another three peak decomposition temperatures at 139.7 °C, 168.8 °C and 408 °C were shown with a residual weight percentage of 36.4% at 800 °C. In general, the phenol groups as the crosslinking sites were very sensitive to heat even below 100 °C. We measured the graft densities of phenol groups based on the results of UV absorption, and the values of CMC-Ph with different molecular weight were calculated to be 19–21, very similar among all CMC-Ph.

CMC-Ph (Figure 1(I)) showed a weight loss of 4–6% below 103 °C due to evaporation of water and thermal decomposition of 4-hydroxybenzylamine. Small peaks at ∼190 °C were also attributed to the thermal degradation of 4-hydroxybenzylamines. CMC-Ph lost about 44–53% of the total weight in 250–335 °C with a higher main peak decomposition temperature than that of CMC, because neutral polysaccharides usually showed higher thermal resistance than charged polyanions. CMC₂₀-Ph and CMC₃₀-Ph had a weight loss of 15–17% in 320–505 °C, but only 9% was shown for CMC₁₀-Ph. An additional 29% weight loss was shown in 450–640 °C for CMC₁₀-Ph due to the thermal decomposition of 4-hydroxybenzylamine, with a peak decomposition temperature at 530 °C. The residual weight percentage of CMC₁₀-Ph, CMC₂₀-Ph and CMC₃₀-Ph were 6.5%, 17.9% and 21.6%, respectively, much lower than that of CMC. The reason might relate to the Na⁺ on CMC chains that could form inorganic sub-products in the degradation reaction [27].

Thermal decomposition temperatures at 5% (T₅), 10% (T₁₀), and 50% (T₅₀) weight loss of CMC₁₀ and CMC-Ph are summarized in Table 1. T₅ of CMC₃₀-Ph was much higher than those of CMC₁₀-Ph and CMC₂₀-Ph, perhaps owing to the formation of little crosslinking structure as the molecular weight reached to 8.8 × 10⁵ [28]. A three dimension network might be formed in CMC₃₀-Ph, so that much higher temperature was needed for the evaporation of the bound water. The peak decomposition temperature shifted to a higher value with the increasing molecular weight. Onset decomposition temperature (T_o) of the main peak in DTG curve was similar among CMC-Ph, but end decomposition temperature (T_e) increased with the increasing molecular weight owing to the crosslinking of CMC chains.

DSC thermogram of CMC-Ph (Figure 2) illustrated one endothermic transition of the thermal decomposition of 4-hydroxybenzylamine in 180–210 °C. Another endothermic transition in 80–90 °C mainly belonged to water evaporation, and the peak temperature and initial temperature of the transition decreased with the increasing molecular weight of CMC-Ph. The interaction of water and CMC-Ph might be hindered by the self H-bonding between hydroxyl groups or carboxyl groups of CMC and light crosslinking of CMC₃₀-Ph. A very thinner layer of the bound water was thus formed on the top layer of the absorbed water of CMC₃₀-Ph to present a lower peak temperature.

Figure 2. DSC thermograms of CMC₁₀-Ph (a), CMC₂₀-Ph (b) and CMC₃₀-Ph (c).

Gelation properties of CMC-Ph

The gelation time of CMC-Ph decreased greatly with the increasing molecular weight (Figure 3), and CMC₃₀-Ph could gel within 23 s owing to its higher viscosity though CMC-Ph had similar contents of phenolic groups. CMC₁₀-Ph could undergo the gelation less than 102 s without the need for external stimuli. Fast gelation would ensure the localization of the hydrogel at the injection site and prevent undesirable flow to the surrounding tissues.

Figure 3. Gelation time of CMC-Ph with different molecular weight (1 mM H₂O₂, HRP 50 U/ml).

Elastic modulus (G′, storage modulus) and viscous modulus (G′′, loss modulus) of CMC-Ph gels were studied to understand the mechanism related to the microstructural changes [29,30]. G′ and G′′ became a function of the strain at a fixed frequency (Figure 4), presenting the non-linear viscoelastic properties. When G′ dominated over G′′, the materials presented the strong gel properties. If G′ was close to and just above G′′, the materials show the typical weak gel properties, which cannot hold more energy. The internal forces induced by higher strain would break the bonds, resulting in collapse of the gel microstructure at the end point. CMC₂₀-Ph showed higher initial G′ (2480 Pa) than CMC₁₀-Ph (790 Pa), mainly because the crosslinking or physical entanglement of macromolecule chains increased with the molecular weight. These two factors enabled the hydrogels to store the energy under strain. As the strain increased, it would reach a point where the gels would not be able to withstand the applied force and started to collapse. The strain of CMC-Ph at crossover G′–G′′ decreased with the increasing molecular weight, illustrating the decreasing breaking stain of CMC-Ph.

Figure 4. Storage modulus (G′) and loss modulus (G′′) against strain for CMC₁₀-Ph, CMC₂₀-Ph and CMC₃₀-Ph gels.

Cell compatibility of CMC-Ph and HRP solution

The effect of CMC₁₀-Ph concentration on the proliferation of ATDC5 is studied (Figure 5(I)). At day 1, there was significant difference between CMC₁₀-Ph of concentrations in 20–80% and the controls, and CMC₁₀-Ph with a concentration of 100% presented remarkable difference with the other samples except the one of 80%. Compared with sample without CMC₁₀-Ph (0%), sample of concentrations in 20–60% did not present remarkable differences, illustrating that the impact of low CMC₁₀-Ph concentration (<0.6 g/ml) on the cells proliferation can be ignored during short culturing period. At day 3, there were significant differences in pair comparison between samples of concentrations within 20%–100% and controls. Samples of concentrations in 20–60% also showed remarkable difference between day 1 and 3, proving that even low CMC-Ph concentration would depress the cell proliferation at longer culturing period. At day 5, the absorbance of sample groups was lower than that of controls. More importantly, ATDC5 without any treatment (0% CMC₁₀-Ph) showed dramatic decrease in cell viability, perhaps owing to contact inhibition during monolayer culturing process. The results showed that high CMC-Ph concentration or long culturing period had obvious inhibition effect on ATDC5 proliferation. The mitochondrial activity of ATDC5 treated by CMC₂₀-Ph and CMC₃₀-Ph was much lower than that of CMC₁₀-Ph even at the concentration of 20% (Figure 5(II)), showing the distinct hindrance effect of high-macromolecular weight on the proliferation of ATDC5. High solution viscosity and micro-crosslinking may disturb the transportation of nutrition and removal of metabolic wastes, and even induce additional stress on cells to influence cell viability and proliferation.

Figure 5. Mitochondrial activity of ATDC5 treated by various CMC₁₀-Ph solutions (I), CMC-Ph with different molecular weight at concentrations of 20%–60% (II), various HRP solutions at 1, 5, 7 d (III), and 0.4 mg/ml HRP solution at 20–60 min (IV). (* representing p < 0.05).

In the case of in situ gelation of a cell-biopolymer suspension, HRP was used as an oxidoreductase along with H₂O₂. Because H₂O₂ can induce cell autoxidation or mediates protein toxicity [31,32], its concentration should be controlled strictly. Sakai S et al. have detailed the dependencies of the gelation time with the concentration of enzyme and H₂O₂ [33], therefore fast gelation can be achieved at a higher HRP concentration, despite of very low concentration of H₂O₂. The proliferation of ATDC5 treated by HRP solution with various concentrations is studied (Figure 5(III)). At day 1, 0.4 mg/ml HRP sample showed significant difference with sample without HRP (0 HRP). At day 5, a distinct difference was shown between 0 HRP sample and the other samples. At day 7, only 0.6 and 1.0 mg/ml HRP samples showed significant difference with 0 HRP sample. No significant difference was shown between the samples of the same concentration at day 1 and day 7. The results showed that higher HRP concentration would influence the proliferation of ATDC5 cells within the same culturing period, but longer culture period might not influence the proliferation of ATDC5 at the same HRP condition. The effect of short culturing period (20–60 min) on the proliferation of ATDC5 treated by 0.4 mg/ml HRP is shown in Figure 5(IV), and ATDC5 culturing without HRP was used as control. All samples did not show significant difference with the control during 20–60 min of culturing. In considering the gelation time was from dozens of seconds to several minutes, low HRP concentration (<100 U/ml) would not influence the proliferation of ATDC5.

Based on the above results, we finally chose CMC₁₀-Ph as the base material for cell encapsulation to study the effect of microsphere size, cell density, and medium protein on ATDC5 viability and proliferation due to its higher cell viability than CMC₂₀-Ph and CMC₃₀-Ph. A concentration of 5% was enough for CMC₁₀-Ph hydrogelation, but a higher HRP concentration (100 U/ml) was used in microsphere production because the gelation reaction occurred once H₂O₂ entered from the continuous fluid into the dispersed fluid. Obviously, the diffusion path would influence in situ H₂O₂ concentration.

Cell-loaded microspheres via the coflowing microfluidic method

The cell-loaded micropheres were prepared via the droplets breakup technology, where the inner fluid of aqueous cell-suspending CMC₁₀-Ph broke up into segments. Within each microdroplet, a diffusion-controlled enzymatic reaction occurred to harden CMC-Ph droplet. Two inner dispersed fluid rates (Q_d) of 10 μl/min and 20 μl/min were used to study the size distribution of the cell-loaded microspheres and empty microspheres at a fixed continuous fluid rate (Q_c) of 5 ml/min. A 5% (g/ml) CMC₁₀-Ph solution containing ATDC5 cells at a density of 10⁶ cells/ml was used as the dispersed fluid (Figure 6). When extruding the cell-containing CMC₁₀-Ph solution at a flow rate of 10 μl/min, we got the cell-loaded and empty microspheres in diameters of 166 and 103 μm, respectively. If Q_d increased to 20 μl/min, the diameters increased to 189 and 137 μm, respectively for the cell-enclosed and empty microspheres. The diameter of the resultant droplets strongly correlated with the viscoelastic property of the disperse phase and the interfacial tension between the disperse-continuous phases. An increase in the flow rate of the disperse fluid resulted in larger microspheres [34]. Nonuniform distribution of ATDC5 in the dispersed fluids and cell settlement on the injector during the production period were the main reasons for the formation of the polydispersed cell-encapsulated micropsheres. Sphere size was critical to optimize conditions for cell-loaded microgels because the proliferation potential of cells were strongly correlated with size during the production of cell aggregates [35,36]. Very few larger microspheres with a diameter up to 488 μm were formed at higher flow rates, which might be confronted with insufficient oxygen and nutrient diffusion for cells in the core [37–39].

Figure 6. Size distribution of the cell-loaded (dark grey) and empty (light grey) microspheres obtained by Q_d of 10 μl/min and Q_d = 20 μl/min at a fixed Q_c of 5 ml/min.

Live/dead microscopy images of cell-loaded microspheres just after the preparation showed a large area of red spots (Figure 7). When viscous CMC₁₀-Ph liquid was extruded into the immiscible continuous liquid flowing in the same direction, cells were easily damaged by the high drag forces. As the flow rate of the dispersed fluid increased, the number of dead cells increased. However, no remarkable reduction of cell viability was shown though higher cell density increased the viscosity of the dispersed fluid. Due to the existence of the membrane gel, the enclosed cells were expected to be protected from shear forces during the subsequent purification process. The reduction in viability of the cells just after the preparation process was interpreted as a consequence of the effect of shear stresses resulting from the mixing of cells in viscous CMC₁₀-Ph solution and the subsequent flow-focusing process. Cell death may also relate to H₂O₂ that was used during the HRP-catalyzed oxidation reaction of phenols to achieve the gelation. Once the droplets left the narrow orifice, H₂O₂ began to penetrate from the liquid paraffin into the CMC₁₀-Ph solution, where peroxidases functioned as oxidoreductases to catalyze the oxidation of donors consuming H₂O₂ immediately. H₂O₂ concentration being controlled by the diffusion process would also be influenced by the size of microdroplets. An interesting phenomenon was that ATDC5 cells were more viable when the cell-encapsulated microspheres were immediately cultured in DMEM supplemented with 10%–15% FBS (Figures 8–10), suggesting that the culture medium might timely awaken ATDC5 cells from an “apparent death” state. Generally, the co-flowing microfluidics method was ideally suited to the production of cell-loaded microspheres.

Figure 7. Live/dead and corresponding bright-field microscopy images of cell-loaded microspheres just after the preparation: (a) 10⁶ cells/ml, Q_d = 10 μl/min; (b) 10⁷ cells/ml, Q_d = 10 μl/min; (c) 10⁶ cells/ml, Q_d = 20 μl/min; (d) 10⁷ cells/ml, Q_d = 20 μl/min. Scale bar represents 100 μm.

Figure 8. Live/dead micrograph of cell-loaded microspheres (Q_c = 5 ml/min, 5% CMC-Ph concentration, 10⁶ cell/ml, 10% FBS) obtained at Q_d of 10 μl/min and 20 μl/min under various culture periods. Scale bar represents 100 μm.

Figure 9. Live/dead micrograph of cell-loaded microspheres (Q_c = 5 ml/min, 5% CMC-Ph concentration, Q_d = 20 μl/min, 10% FBS) obtained at 10⁶ cell/ml and 10⁷ cell/ml under various culture periods. Scale bar represents 100 μm.

Figure 10. Live/dead micrograph of cell-loaded microspheres (Q_c = 5 ml/min, 5% CMC-Ph concentration, Q_d = 20 μl/min, 10⁷ cell/ml) obtained at 10% FBS and 15% FBS under various culture periods.

Viability of cells within the loaded microspheres

The effect of the microspheres size on cell viability is shown in Figure 8. The cells existed individually just after encapsulation, and self-aggregated to form cell assembly within several days of culture. The occupied area of the cells assembly in the microspheres increased obviously as the culture period prolonged. In the initial culture period, no significant difference on the live cells (green dots) could be seen in large microspheres (Q_d = 20 μl/min) and small microspheres (Q_d = 10 μl/min). However, large microspheres were prone to form live cell assembly in the later period, suggesting the different physical and mechanical microenvironments to cells induced by various sizes of microspheres. Because the crosslinking structures of CMC gels would be similar for both microspheres via the diffusion-controlled gelation process, mass transport of nutrients or metabolites would be similar theoretically, but large cellular aggregates would influence the transportation route. Moreover, large space would produce minor limitation to the growth of cells in the later periods.

The viability of ATDC5 cells encapsulated in microspheres at two different cell densities is shown in Figure 9. In general, more live cell aggregates were seen in microsphere with higher cell density, and cell aggregate size was positively correlated with cell density. Chondrogenic cells can form aggregates via self-assembled interactions with adjacent cells. As the distance between neighboring cells decreased, more cells would be encapsulated at higher cell densities to produce more cell-cell interactions. High cell density was also an important factor in the chondrogenic induction of stem cells [40,41] and the maintenance of differentiation state of mature chondrocytes [42], and the formed densely cellular assembly would further promote cell-cell interaction [43,44]. Moreover, the growth of the cells was strongly affected by the characteristics of the surrounding matrices. The weak adhesion of ATDC5 cells to the hydrophilic CMC gel surface would enhance cell aggregation, which might facilitate the differentiation of ATDC5 cells into chondrocytes.

Protein concentration in microspheres is essential to maintaining cell viability and growth, and ATDC5 cells have been proven to retain the properties of chondroprogenitor cells and rapidly proliferate in the presence of 5% FBS [45]. We therefore assessed in vitro cell viability in 10% and 15% FBS DMEM (Figure 10), and found that more live cells were shown in higher FBS concentration sample. The cell-loaded microspheres where the cells dispersed uniformly and formed aggregates with a spherical shape were shown at day 14. This high sphericity of cell-loaded microsphere was very important to maintain the mechanical stability toward the forces exerted by the surrounding environment, such as drag forces in the flowing culture medium and compression forces in the body.

The growth profiles of the enclosed cells were estimated by increases in the amount of a water-soluble formazan dye. We measured the absorbance of the microsphere samples in the fixed wells at different culture periods, and calculated the mitochondrial activity ratio that was normalized by the activity at day 1 (Figure 11). The mitochondrial activity gradually increased with prolonged culturing time. Proliferation profiles of cells-loaded microspheres being obtained at Q_d of 10 μl/min and ATDC5 concentration of 10⁶ cell/ml increased to 250% compared to day one, and a 1.3-fold increase in activity was further achieved for microspheres obtained at Q_d of 20 μl/min after 21 days of culturing in 10% FBS DMEM. If the cell concentration was 10⁷ cell/ml, the mitochondrial activity ratio of ATDC5 at day 21 increased nearly 170% compared with ATDC5 concentration of 10⁶ cell/ml. Another 2.6-fold increase in activity was observed for cells cultured in 15% FBS DMEM at fixing Q_d of 20 μl/min and cell concentration of 10⁷ cell/ml. Obviously, high cell density in CMC-Ph solution and FBS concentration in culture medium favored the proliferation of ATDC5 in the microspheres greatly.

Figure 11. Time-courses of mitochondrial activity ratio for encapsulated ATDC5. Each value was normalized by the activity at day 1 (n = 5).

Cell aggregates

We examined the morphology of the cell aggregates in mircospheres obtained at Q_d of 20 μl/min using ATDC5 concentration of 10⁷ cell/ml. As shown in Figure 12(a,c), live cell aggregates could be detached from the broken microspheres, caused by the shear stress of the culture medium flow after slight vibration treatment. Cells being encapsulated in microspheres participated in two different types of interactions: cell-matrix and cell-cell interactions. Both interactions were important for cells to maintain their phenotype and progress in tissue development. Morphological change of the enclosed cells was one of the direct signs of these interactions. The isolated spherical ATDC5 cells were found around the compact cell assembly (Figure 12(b,c)) and no polarity was induced by cell-matrix interactions though encapsulated cells could adhere to three dimensions matrix. Another possible reason might be the sterical hindrance of fully embedded cells in their spreading and migration due to the confinement of the surrounding matrix.

Figure 12. Representative microscopy images of cell-loaded microspheres (Q_d = 20 μl/min, 10⁷ cell/ml, 10% FBS): (a) Bright field microscopy images of cell-loaded microspheres being cultured for 20 days and vibrated slightly; (b) Bright field microscopy images of cell-loaded microspheres being cultured for 12 days, vibrated slightly and cultured for additional 1 day; (c) Live/dead micrograph of cell-loaded microspheres at day 41. Scale bar represents 100 μm.

Conclusions

CMC with different molecular weights was successfully grafted with phenolic groups. The main peak decomposition temperature of CMC-Ph was significantly higher than that of CMC, but the grafted 4-hydroxybenzylamine was much sensitive to heat even below 100 °C. As the molecular weight increased, CMC-Ph gelled within the shorter time and the gels had a decreased strain at crossover G′-G′′ and an increased storage modulus. Cell toxicity results revealed that high CMC molecular weight, high CMC-Ph concentration or long culturing period had obvious inhibition effect on ATDC5 proliferation. CMC-Ph allowed incorporation of live cells before injection and could produce live cell-loaded microsphere using the flow focusing microfluidic device under HRP/H₂O₂ reaction. Cell encapsulation was effective in promoting the proliferation of ATDC5, and various cell aggregates were obtained via controlling microspheres’ size, cell density or FBS concentration. CMC microspheres might serve as a versatile platform for size-customized spheroid formation of non-anchorage dependent cells, and may instruct cell-cell and cell-material interactions, or even define the three-dimensional arrangement of neocartilage tissue formation.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China [51572110, 31470934]; the Guangdong Natural Science Foundation of China [2016A030313085]; the Guangdong Science and Technology Program Key project of China [2014B010105007]; and the Guangdong Science and Technology Project of China [2015A020212025].