Validation and scalability of homemade polycaprolactone macrobeads grafted with thermo‐responsive poly(N‐isopropylacrylamide) for mesenchymal stem cell expansion and harvesting

Abstract In this study, polycaprolactone (PCL) macrobeads were prepared by an oil‐in‐water (o/w) emulsion solvent evaporation method with poly(vinyl alcohol) (PVA) as an emulsifier and conjugated to poly(N‐isopropylacrylamide) (PNIPAAm) to be used as cell carriers with noninvasive cell detachment properties (thermo‐response). Following previous studies with PCL‐PNIPAAm carriers, our objectives were to confirm the successful conjugation on homemade macrobeads and to show the advantages of homemade production over commercial beads to control morphological, biological, and fluidization properties. The effects of PCL concentration on the droplet formation and of flow rate and PVA concentration on the size of the beads were demonstrated. The size of the beads, all spherical, ranged from 0.5 to 3.7 mm with four bead categories based on production parameters. The morphology and size of the beads were observed by scanning electron microscopy to show surface roughness enhancing cell attachment and proliferation compared to commercial beads. The functionalization steps with PNIPAAm were then characterized and confirmed by Fourier transform infrared spectroscopy, scanning electron microscopy, and energy dispersion spectroscopy. PNIPAAm‐grafted macrobeads allowed mesenchymal stem cells (MSCs) to spread and grow for up to 21 days. By reducing the temperature to 25°C, the MSCs were successfully detached from the PCL‐PNIPAAm beads as observed with fluorescence microscopy. Furthermore, we validated the scalability potential of both macrobeads production and conjugation with PCL, to produce easily kilograms of thermo‐responsive macrocarriers in a lab environment. This could help moving such approaches towards clinically and industrially relevant processes were cell expansion is needed at very large scale.


| INTRODUCTION
Thermo-responsive scaffolds containing poly(N-isopropylacrylamide) (PNIPAAm) are of great interest, thanks to their specific ability to release cells without physical damage upon temperature changes (Dhamecha et al., 2021;Haq et al., 2017;Kim et al., 2003). This technique has been shown not to alter cell physiology, morphology, and immunophenotype of the released cells (Zhang et al., 2015). As industrial requirements as well as clinical applications of cell-based therapeutic treatments require large numbers of cells (Baudequin et al., 2021;Rafiq & Hewitt, 2015;Want et al., 2012), these newly developed thermo-responsive polymers have been proven to hold clear promises to expand cells with improved cell purity when combined to systems with high surface/volume ratio (S. Chen et al., 2021;Hanga & Holdich, 2014;Wu et al., 2016;H. S. Yang et al., 2010). In particular, macrocarriers have high cell density per unit volume allowing for the rapid generation of large batches of cell products.
They also offer advantages for easy handling in bioreactor vessels compared to microcarriers than can stick to the wall of vessels or be lost in conduits without proper filters and concentration procedures (Nguyen, Odeleye, et al., 2019). Moreover, in macro size (around 1-5 mm in diameter), the cells grown on the surfaces of macrocarriers are less exposed to shearing forces . In comparison with flat surfaces, macrobeads could therefore be suitable for expansion scale-up and suitable cell harvesting procedures.
In a previous study, we used commercial polycaprolactone (PCL) beads coated on their surface with PNIPAAm to produce thermo-responsive macrobeads suitable for the expansion and noninvasive harvesting of mesenchymal stem cells (MSCs) without the use of generally employed proteolytic enzymes (trypsin) (Nguyen, Odeleye, et al., 2019). The size of these PCL macrobeads, which were obtained from Sigma-Aldrich, ranged from 3 to 5 mm. In this study, we aimed at having a better control on the bead properties by developing a handmade production process. The PCL beads were prepared directly in our laboratory by an oil-in-water (o/w) emulsion solvent evaporation method. By fabricating the PCL beads in our lab, we could change the size of the PCL beads as desired by controlling the emulsification technique such as the flow rate and the concentration of emulsifier. Thus, the first objective of this study was to obtain uniform macrobeads by using various flow rates of a pump and various concentrations of poly(vinyl alcohol) (PVA) as emulsifiers and to confirm their potential as cell carrier after conjugation with PNIPAAm. These uniform spherical beads were expected to produce a better thermo-response with higher surface area when compared to other shapes such as ovoid beads (Al-Hajry et al., 1999;Lee et al., 2013;Voo et al., 2015). We also assessed their superiority in terms of fluidization potential. Second, we aimed at validating that all steps could be homemade in laboratory, from PCL beads preparation to the conjugation with PNIPAAm, at laboratory scale but also at the industrial scale.

| Preparation of PCL macrobeads
PCL macrobeads were prepared using an established emulsion method (Kemala et al., 2012;Li et al., 2015), followed by the evaporation of the solvent used to liquefy the macrosphere polymer.
A syringe containing 5 ml of PCL solution was placed on a pump (0.4 ml/min) and used to form PCL/DCM solution droplets through an 18-G needle, precursors to the solidified beads. The formed PCL/ DCM droplets were collected in a glass petri dish (12-cm diameter), containing 10 ml of PVA solution, without agitation, at room temperature. The distance between the needle and the surface of the solution was 2 cm. The PVA solution was then removed until a minimal layer of solution was covering the beads, and the samples were placed in a fume hood to allow the solvent to evaporate through the aqueous phase over 3 days, thus resulting in droplet solidification and macrobead formation. Beads were finally washed at least three times with dH 2 0 before performing further experiments.

| Conjugation of PNIPAAm with PCL macrobeads
Conjugation of PNIPAAm with PCL was performed as previously described (Nguyen, Odeleye, et al., 2019). Briefly, the PCL macrobeads were immersed in NaOH 1 M solution to obtain carboxylate ions PCL-COO-then were rinsed with sterile DI water five times.
Reaction buffer was prepared by dissolving 0.12 M EDC (0.46 g) and 0.06 M NHS (0.14 g) in 20 ml of 0.05 M MES buffer solution (pH 6).
PNIPAAm-NH 2 solution was prepared by dissolving 2 g of PNIPAAm-NH 2 powder in 20 ml of deionized water.

| 3D printing
Cylindrical bioreactor and parts of set-up to perform the scale-up of the PCL beads production were prepared by 3D printing with a Form 2 printer using PreForm software (Formlabs). Parts were designed using Autodesk Inventor Professional 2018 software (Autodesk) and printed with Clear V4 resin (Formlabs).

| Characterizations
Fourier transform infrared (FTIR) spectra were acquired with an FTIR spectrometer (Bruker, Tensor 27) equipped with attenuated total reflectance (ATR, Pike). The background signal was estimated before every measurement by measuring the response of the spectrometer without any sample.
Scanning electron microscopy (SEM; Carl Zeiss Evo LS15 VP-Scanning Electron Microscope SE, BSE, VPSE, EPSE detectors, 10 kV) was used to image the surface roughness and morphology of the PCL and PCL-P macrobeads. Samples were coated with gold by sputtering before observations. Energy dispersion spectroscopy (EDS) analysis was performed with INCA X-Act X-ray (Oxford Instruments) and OIM XM 4 Hikari EBSD (EDAX) systems. For cell viability and proliferation, the CCK-8 assay (Sigma-Aldrich) was performed after 1, 7, 14, and 21 days of incubation. Cell seeding density was 5 × 10 3 cells/ml.

| Cell detachment from PCL-P
After 1 day of culture on PCL-P macrobeads, the temperature of the cell environment was reduced from 37°C to 25°C using an incubator for 1 h. The detached cells were observed and imaged by an inverted microscope (Eclipse Ti; Nikon).

| Statistical analysis
All experiments were conducted with at least three independent replicates. Statistical analysis was performed with two-way analysis of variance with Tukey's honest significant difference post hoc test using GraphPad Prism 6. A value of p < 0.05 was considered statistically significant.

| RESULTS AND DISCUSSION
By comparing commercial (PCL-Com) and homemade PCL carriers Following the validation of such a production method at low scale, we investigated the effects of the production parameters on bead morphology, the efficiency and functionality of the PNIPAAm conjugation, the cell response as well as the fluidization and scalability potential of our PCL macrobeads.

| Effect of concentration of PCL on bead formation
The influence of PCL concentration on the bead formation was probed using 10, 12, 15, and 18-w/v% solutions, as shown in Table 1 3.2 | Effect of flow rates on the beads size

| Effect of concentration of PVA on beads size
In this study, PVA was used as the emulsifier. The hydroxyl groups in PVA interacts with the water phase while the polymer chain interacts with the dichloromethane, thus making the formed emulsion more stable (Ahlin et al., 2002;Lai & Tsiang, 2004). Variations in PVA concentration and volume were expected to affect the emulsion stability resulting in modification of the size of the macrospheres (Ahlin et al., 2002;Lai & Tsiang, 2004). As shown in Table 3, increasing the PVA concentration led to a decrease in the size of the macrospheres. When the concentration of PVA was increased, more PVA molecules overlaid the surface of the droplets, providing increased protection of the droplets against coalescence which resulted in the production of smaller emulsion droplets. Since the macrobeads were formed from emulsion droplets after solvent evaporation, the size was dependent on the size of the initial emulsion droplets (Ahlin et al., 2002). Furthermore, the viscosity of the aqueous solution was relatively higher at high PVA concentrations compared to lower concentrations, which could be another factor in the separation of droplets in the emulsion from each other (Q. Yang & Owusu-Ababio, 2000). Dispersion into water to evaporate the DCM led then to the precipitation of the macrospheres into solid macrobeads.
3.4 | Size distribution, morphology, and FITR of the PCL macrobeads Figure 2 shows the morphology of the PCL macrobeads prepared in our laboratory (PCL-Oxford) compared to commercial pellets (PCL-Com).
Although the PCL-Oxford samples showed a highly spherical shape, they presented regular surface roughness and porosity as seen with SEM (patterned surface, period of around 200-400 µm). In contrast, the surface of PCL-Com beads appeared dense and smooth, without visible patterning. Porous properties are necessary to absorb and retain nutrients and medium on the surface of the beads and in turn enhance cell adhesion (Mesquita-Guimarães et al., 2020;Santos, 2012;Webster, 2006). The porous surface of PCL-Oxford beads was therefore expected to promote cell adherence and growth better than a dense surface.
Four types of beads were prepared in this study with diameters ranging from 0.5 to 3.7 mm, as shown in Table 3 Results of the FITR spectra of PCL-Oxford and PCL-Com are shown in Figure 3c. It was shown that the PCL-Oxford peaks matched all the PCL-Com peaks, confirming that the beads were made of solidified PCL and that the fabrication process did not change the chemical structure of PCL.

| Morphology of PCL and PCL-P macrobeads and optimization of incubation time
The surface morphology of PCL and PCL-P macrobeads was characterized by SEM (Figure 4). Similar to the results obtained previously directly on commercial pellets (Nguyen, Odeleye, et al., 2019), grafting  rate has to be found to avoid jeopardizing cell viability (Baudequin et al., 2021;Carpentier et al., 2011;Rauh et al., 2011).
To assess the potential of our PCL macrobeads for such applications, we used a small cylindrical bioreactor obtained with transparent 3D-printing (Figure 6a). It allowed for easy fluidization of an initially packed bed of carriers (2 cm) with easy monitoring of the maximal bed height when running medium circulation. As shown in Figure 6b, for a specific flow rate, higher height was obtained with the PCL-Oxford macrobeads compared to PCL-Com, meaning that they were more easily fluidized. Optimal exchanges could therefore be achieved for lower flow rates, providing a better balance between nutrient flow and cell survivability upon dynamic culture conditions. Moreover, this behavior was obtained regardless of the size of the PCL-Oxford macrobeads. These results suggest that the homemade carriers could be more suitable to be used in a fluidization system.

| Adhesion, proliferation and detachment of cells
PCL is food and drug administration-approved for implantation and use in tissue engineering and drug delivery systems (Baudequin et al., 2017;Bigham et al., 2021;Kwon et al., 2013;Li et al., 2015;Q. Wang et al., 2021). As shown in our previous study (Nguyen, Odeleye, et al.,  properties of this PCL source, altering the material's ability to support human embryonic stem cell proliferation (Li et al., 2015).
However, in this study, the cells grew successfully on pure PCL80k as well as PCL80k with conjugation of PNIPAAm on the surface.
As GFP was cloned into MSCs, green emission was in terms of noninvasive detachment (Nguyen, Odeleye, et al., 2019).
Successful cell detachment at Day 7 had been shown in our previous study (Nguyen, Odeleye, et al., 2019) in the same conditions on PCLcom samples, which are used here as a reference group. As Day 1 results were here similar for PCL-com and PCL-Oxford, we assumed that the same behavior would occur at longer term and we moved then further to the scale-up analysis. F I G U R E 7 Cell proliferation of mesenchymal stem cells (MSCs) seeded on PCL and PCL-P macrobeads after 3 and 7 days, stained by (a) Hoechst only (Nuclei staining in blue dot), observed by fluorescence microscopy in low magnification and by (b) Hoechst (Nuclei staining in blue dot) and green fluorescence protein, observed by fluorescence microscopy in high magnification. The large blue dot areas show bead autofluorescence. Levine et al., 1979;Nguyen, Odeleye, et al., 2019). To achieve this and to become a new gold standard such as T flasks nowadays, the potential of such macrocarriers for large scale production has to be validated. The transfer of biotechnologies from bench to bedside needs indeed to develop processes achievable at industrial scale with the same outcomes.

|
Lab-scale studies of the homemade macrobeads reported so far in this paper were performed on samples produced manually. The process shown in Figure 1 required the experimenter to move slowly the PVA bath upon macrospheres formation to avoid immediate fusion. Such an approach is therefore time consuming and can increase size variability.
Hence, we investigated the development of an automated production system based on a continuous perfusion loop of PVA ( Figure 9a). The PCL solution was then dropped continuously in the circulating PVA solution through channels specifically designed and 3D-printed (Figure 9b). Thanks to tubing length after microsphere formation, PCL macrobeads were stable enough when they reached the reservoir beaker to accumulate without fusing (inset on Figure 9c). From a full 10-ml syringe, the complete volume of PCL solution could be turned into beads collected in the beaker. The same drying and washing processes as the manual fabrication were then applied to obtain the PCL macrobeads with spherical shape.
F I G U R E 8 Cell proliferation on polycaprolactone (PCL), PCL-P, and PCL-Com surfaces. (a) CCK-8 studies showed that mesenchymal stem cells (MSCs) survived and proliferated on PCL, PCL-P, PCL-Com, and tissue culture plate as control (TCP CON) surfaces for 1, 7, 14, and 21 days (ns, no significant difference; *p < 0.05, **p < 0.01, ****p < 0.0001). MSCs detached from PCL-P by reduced temperature from 37°C to 25°C after 1 h, (b) low magnification and (c) high magnification. The size distribution of the automated PCL macrobeads batches was evaluated and compared to the manual production method. As seen on Figure 10a, the average diameter obtained after trial and error to adjust flow rate parameters was 1233 ± 144 µm with polynomial distribution, that is, similar range to the "bead 3" group used for the cell culture validation study ( Figure 3). As an advantage, the automated system developed here could produce various sizes of beads if needed by varying parameters (PVA flow rate, PCL flow rate, needle gauge). Moreover, the same PVA solution could be used at least three times to produce new batches of macrocarriers. The 3D-printed system was designed to be used as a "lid" on a 600-ml reservoir beaker ( Figure 9c) and showed good stability and easy batch replacement. Overall, this proof-of-concept step validated the potential of the PCL macrobeads to be scaled up towards industrial scale. This is particularly promising as they can be then stored in dry form after production for delivery and later use.

| Scale-up of the PCL-P conjugation process
To go further with the validation of the scaled-up potential, it had to be confirmed that the conjugation step could be performed on large batches of PCL macrobeads, that is, 1-2 kg. The preparation of small batches for developmental studies were performed in 15-ml plastic tubes hosting a few dozens of pellets in 10 ml of grafting solution, on a roller shaker. Optimal mixing had to achieve homogeneous conjugation by avoiding bead-bead contact that could create nongrafted surfaces. As multiple tubes would not be cost-and time-efficient, we investigated the use of a system offering relevant working volume and maximum product weight, easy batch replacement, low rotation speed, solvent-proof vessel, and suitable for solid/ liquid mixing. Proof-of-concept study of conjugation on a large PCL macrobeads batch (1 kg) was then performed with a rotating stainless-steel barrel at 20 rpm (Drum hoop mixer JEL RRM Mini-II; J. Engelsmann AG) for 16 h. The results of conjugation on PCL along with small scale and blank control groups are reported in Figure 10b.
Grafting was successfully performed at both scales, as shown by the amine-related peaks for N─H bending vibration at 1577 and 3400 cm −1 (Jiang et al., 2012;Takahashi et al., 1991;W. Wang et al., 2017) that did not appear on the FTIR spectra of PCL carriers maintained in water. This confirmed therefore that the conjugation process could occur in a solid/liquid mixer hosting several kilograms of PCL macrobeads and grafting solution. Altogether with the F I G U R E 10 Validation of scalability. (a) Size distribution of the macrobeads prepared with the automated system (dashed line: polynomial trend curve, MS Excel), (b) Fourier transform infrared (FTIR) spectra of PCL conjugation in a large-scale mixer (scaled-up conjugation). Low-scale conjugation (PCL in 15-ml tubes on shaker) and PCL in water (PCL control) were used as controls.
validation of automated production, cell harvesting and control of morphology, this highlighted the potential of this approach for clinical and industrial scale applications.

| CONCLUSION
The homemade PCL macrobeads produced in this study were formed by an o/w emulsion solvent evaporation method with PVA as an emulsifier. FTIR spectra confirmed that the PCL maintained its chemical structure after macrobead formation. The morphology of the homemade PCL beads was porous and the shape of the bead was spherical. By varying the PVA concentration and flow rate, the size of the beads can be controlled to obtain uniform beads. The PCL and PCL-P beads were suitable for MSC adhesion and proliferation up to 21 days and showed better trends for expansion and fluidization than commercial PCL beads used directly. By simply reducing the temperature from 37°C to 25°C for 1 h, the MSCs were detached without the need for enzyme treatment. In addition, it was shown that both macrobead production and conjugation process could be performed in lab at large scale. Although some quality testing could be done before