All reagents were from Aladdin (Aladdin, China), if not otherwise noted. Poly (lactic-co-glycolic acid) (PLGA, LA:GA = 80:20, Mw = 200,000) was prepared via ring-opening co-polymerization of L-lactide (LA) and glycolide (GA) in our laboratory (Yan et al., 2019).

The synthesis of oleic acid grafted Fe₃O₄ (Fe₃O₄-OA) NPs was slightly modified according to the method previously published by our laboratory (Hao et al., 2019). In brief, iron (III) chloride (10.8 g, 40 mM) and sodium oleate (36.5 g, 120 mM) were blended in a mixture of 60 ml distilled water, 80 ml ethanol, and 140 ml n-hexane, in a round bottom flask. Then, the reaction solution was heated to 70°C for a 4 h reflux with violent magnetic stirring. The brown hexane phase was separated from the solution by cooling and washed three times with 30 ml distilled water. Then, the hexane phase was transferred to a clean round-bottomed flask and dried in a rotary vacuum evaporator with water bath heating. Subsequently, 6.4 ml oleic acid and 253 ml 1-octadecene were added to the red-brown viscous product. The mixing solution was then ultrasonicated for 30 min to complete dissolution. Finally, a glass capillary was added, and slender branch pipes were connected to the mouth of the flask. After holding at 100°C for 1 h, the reaction mixture was heated to 320°C and held again for 1 h. After cooling down to room temperature, the black products were collected using a magnet, washed three times with 100 ml of petroleum ether, and precipitated with 300 ml of ethanol. The obtained product was dried at room temperature for standby application.

Preparation of Fe₃O₄-OA-PLGA microspheres was performed using the high-voltage electrostatic (HVE) technique (Yan et al., 2021). Briefly, PLGA (1 g) and Fe₃O₄-OA NPs (50 mg) were fully dissolved in dichloromethane (DCM, 10 ml). The solution was loaded into a 2 ml syringe with a 27 G needle. The needle tip and the ethanol replacement solution were linked to the positive and negative electrodes of the HVE droplet generator, respectively. The syringe was advanced at a uniform speed. Microspheres were collected through filtration via sieves and washed successively with ethanol solution and distilled water to remove the excess residual organic solvent.

To prepare pDA/Fe₃O₄-OA-PLGA microspheres, the obtained Fe₃O₄-OA-PLGA microspheres were immersed in dopamine solution (2 mg ml⁻¹, 10 mM tris, pH = 8.5) for 2 h at room temperature. Then, the pDA/Fe₃O₄-OA-PLGA microspheres were incubated in di-amino-PEG (Mw = 2,000, ToYongbio, China) solution (200 mg ml⁻¹, 10 mM tris, pH = 8.5) at 40°C for 4 h. The modified microspheres were washed with distilled water and termed PEG/pDA/Fe₃O₄-OA-PLGA. In order to enhance the selective adhesion of MSCs, the surface of the microspheres was functionalized with the E7 peptide (GLBiochem, China). The modification of peptide was conducted in the following two ways. 1) The heterobifunctional crosslinker 4-(N-Maleimidomethyl) cyclohexane-1-carboxylic acid 3-sulpho-N-hydroxysuccinimide ester sodium salt (sulfo-SMCC) was used to bind the peptide to the di-amino-PEG molecule on the surface of the microspheres. Sulfo-SMCC was dissolved in distilled water and diluted to 1 mg ml⁻¹ with phosphate buffered saline-ethylene diamine tetraacetic acid (PBS-EDTA) coupling buffer. The sulfo-SMCC solution was pipetted into the PEG/pDA/Fe₃O₄-OA-PLGA microspheres, followed by incubation for 1 h at room temperature. The microspheres were then covered with E7 peptide solution (2 mg ml⁻¹, coupling buffer) and incubated overnight at 4°C to obtain E7/PEG/pDA/Fe₃O₄-OA-PLGA. 2) For E7/pDA/Fe₃O₄-OA-PLGA microspheres, pDA/Fe₃O₄-OA-PLGA microspheres were incubated with E7 peptide solution (2 mg ml⁻¹, 10 mM tris, pH = 8.5) overnight at 4°C. If used for cell experiments, the above processes were completed under sterile conditions after the alcohol disinfection of Fe₃O₄-OA-PLGA microspheres. Partial microspheres were dried under vacuum for further characterization.

The chemical structure of Fe₃O₄-OA NPs and PEG-modified microspheres was determined by a Perkin Elmer 2000 FTIR instrument (Perkin Elmer, Germany). A Tecnai G2S-Twin transmission electron microscope (FEI, United States) was used to record the size and shape of the Fe₃O₄-OA NPs. The magnetic properties were determined in a Quantum Design-MPMS-XL7 vibrating sample magnetometer (VSM) system (United States) at 300 K. A Philips XL30 ESEM-FEG instrument (Philips, Japan) was used to observe the surface morphology of microspheres. Elemental mapping was carried out by an XL-30W/TMP scanning electron microscope (Philips, Japan). Static water contact angles were measured using a VCA 2000 contact angle system (AST, United States). The percentage of grafting was measured using a TGA500 thermogravimetry analyzer (TA Instruments, United States). Each sample was heated from room temperature to 600°C at 10°C min⁻¹.

MSCs were isolated from bone marrow of the tibias and femurs of Sprague-Dawley (SD) rats (100 g, male), as reported previously (Soleimani and Nadri, 2009). The experimental protocol was approved by the Institutional Animal Care and Use Committee of Jilin University School of Pharmaceutical Science. The extracted cells were seeded in a 100-mm cell culture dish (Nest, China) and placed in a humidified incubator at 37°C with 5% CO₂. The culture medium was changed every 3 days until the cells reached 80% confluency for subculture. All experiments were carried out during passages 3–5. NIH3T3 and RAW 264.7 cell lines were purchased from ATCC and cultured under standard conditions.

MSCs were seeded at a density of 2 × 10⁴ cells well⁻¹ in 48-well (containing 30 µl microspheres) for 1, 3, and 7 days. At each predetermined time point, 500 μl of fresh medium containing 30 μl of CCK-8 (Solarbio, China) was used to replace the medium, and the plate was incubated for 2 h at 37°C. After incubation, the solution (100 μl) in each well was pipetted into a new 96-well plate and measured the absorbance at 450 nm using an Infinite M 200 microplate reader (Tecan, Switzerland). For observing the morphology of cells seeded on each microsphere, double staining of 1,1′-dioctadecyl-3,3,3′,3′tetramethylindocarbocyanine perchlorate (DiI) (Meilun Biological, China) and 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI) (Invitrogen, United States) was performed after 3 days. In order to assess the cell area fraction on various microspheres, the image analysis software ImageJ was used.

In order to evaluate the cell-adhesion-resistant effect of PEG/pDA/Fe₃O₄-OA-PLGA microspheres, NIH3T3 cells at a density of 2 × 10⁴ cells well⁻¹ were seeded on microspheres in a 48-well plate and cultured for 12 h. Cell adhesion was observed by fluorescence microscopy with DAPI staining.

MSCs, NIH3T3, and RAW 264.7 cells were seeded on the different microspheres in 48-well plates at a density of 2 × 10⁴ cells per well. After incubation for 6 h, the microspheres were washed with PBS, respectively stained with DiI, Calcein-AM (Sigma, Germany), and Hoechst 33342 (Beyotime, China), and viewed under a fluorescent microscope.

MSCs cultured on E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres for 7 days were digested by trypsin digestion for subsequent experiments. An appropriate amount of cells was used to prepare a single-cell suspension, and monoclonal antibodies CD29, CD90, and CD34 (Invitrogen, United States) were added separately to each tube. After incubation at room temperature for 30 min in the dark, the cells were washed with PBS three times to remove the unbound antibodies and detected by flow cytometry. Osteogenic, adipogenic (Puhebio, China), and chondrogenic (Cyagen, China) induction mediums were used instead of basic medium for culture, respectively, to evaluate the differentiation potential of the expanded MSCs. To confirm osteogenesis, cells were stained with the alkaline phosphatase (ALP) assay kit on day 3 of culture and alizarin red on day 14 of culture. Adipogenic differentiation was detected by Oil Red O staining on the 21st day. Alcian blue staining was used to detect the formation of cartilage following 28 days.

Statistical analysis was conducted using analysis of variance (ANOVA, one-way, Origin 8.0, United States). The data were presented as means ± standard deviations (SD). In each experiment, triplicate samples were evaluated. Significant differences were defined at p < 0.05.

The dispersion stability of Fe₃O₄-OA NPs was first examined in chloroform. After standing for 50 h, Fe₃O₄ NPs (Supplementary Figure S2A) were precipitated, while Fe₃O₄-OA NPs (Supplementary Figure S2B) remained well dispersed: given that Fe₃O₄ NPs are poorly dispersed in organic solvents, this was probably due to magnetostatic (magnet dipole-dipole) interactions between magnetic particles (Maity and Agrawal, 2007). However, the mutual repulsion of OA groups on the surface of Fe₃O₄-OA NPs can weaken this phenomenon. This facilitated the preparation of homogeneous magnetic composites. When a magnet was applied, Fe₃O₄-OA NPs were gathered to the side close to the magnet (Supplementary Figure S2C), indicating that they had good magnetic responsiveness. Based on these findings, it could be preliminarily concluded that Fe₃O₄-OA NPs were successfully synthesized. Fe₃O₄-OA NPs were characterized using TEM. Figure 1A shows TEM images of Fe₃O₄-OA NPs, where it can be seen that the NPs were polygonal and homogeneous in size. TEM images were further analyzed by ImageJ, in which at least 100 NPs were measured (Figure 1B). The average size of the NPs was about 15.03 ± 2.47 nm. The size of the individual iron oxide NPs was typical of single domain superparamagnetic iron oxides (Teja et al., 2009), as verified in subsequent magnetic measurements. The magnetization versus the applied field curves was shown in Figure 1C. At T = 300 K, neither remanence nor coercivity could be detected for the sample with saturation magnetization of 51.4 emu g⁻¹, indicating that superparamagnetic behavior existed. Superparamagnetic iron oxide nanoparticles are of increasing interest in the biomedical field due to their unique small size, physical properties, low toxicity, and biocompatibility (Neuberger et al., 2005; Hao et al., 2021). As shown in Figure 1D, stretching vibration of the Fe-O bonds located at about 578 cm⁻¹ was observed in the spectra. In the spectrum of Fe₃O₄-OA NPs, the peak at 2,849 cm⁻¹ was due to the symmetrical stretching vibration of methylene groups, while the peak at 2,920 cm⁻¹ was attributed to the asymmetric stretching vibration of methylene groups. These bands are a trait of unsaturated hydrocarbon chains of OA. It was found that the two peaks of the sample were concentrated at 1,512 cm⁻¹ and 1,417 cm⁻¹, which corresponded to the vibration modes of asymmetric and symmetric carboxylate groups. The stretching vibration of the carbon-carbon double bond in sodium oleate was responsible for the peak at 1,656 cm⁻¹. This indicated that the carboxyl group of oleic acid was chemisorbed on the surface. The weight loss shown by the TGA curve indicated that the grafting of OA on Fe₃O₄-OA NPs was 21.247 wt.%. The peak of the DTA curve for oleic acid in Figure 1E was at 270.66°C, while the peak of the DTA curve for Fe₃O₄-OA was at 349.91°C. This weight loss was related to the decomposition of OA covalently bonded to the surface of the NPs (Mahdavi et al., 2013).

Macroscopic observation showed that the Fe₃O₄-OA-PLGA microspheres (right) were darker in color than the Fe₃O₄-PLGA microspheres (left) (Supplementary Figure S3A). This was caused by the presence of oleic acid on the surface of Fe₃O₄-OA NPs. The alkane chains on the surface of Fe₃O₄-OA NPs improved the interfacial compatibility with the hydrophobic polymer matrix (Okassa et al., 2007; Hao et al., 2019). Therefore, the dispersion of Fe₃O₄-OA NPs in PLGA solution was more uniform than that of Fe₃O₄ NPs, and the NPs were not easy to precipitate and clog the needle. As seen from the optical microscope, these microspheres were round and regular (Figure 2A), with a diameter of about 273.07 ± 19.09 μm (Figure 2B). The surface morphology of different microspheres was observed by SEM (Figure 2C). Because they were formed by solvent extraction, the microspheres had well-formed sphericity and a porous surface. After pDA modification, the microspheres acquired a rough surface, although their spherical shape was preserved. The saturation magnetization was 2.5 emu g⁻¹ for microspheres (Figure 2D). The magnetic field had a significant effect on the distribution of microspheres: in the absence of a magnet (Supplementary Figure S3B), the microspheres in the water gradually sank because of gravity. After applying the magnet (Supplementary Figure S3C), the microspheres were quickly attracted to the vial wall near the magnet.

PEG is a versatile polymer that is non-toxic and FDA-approved for use in biomedical research applications, including tissue engineering and drug discovery (Raina et al., 2021). Because of the cell-repulsive and non-adhesive properties of PEG, di-amino-PEG is used here as a microsphere surface modifier for facile conjugation and subsequent activation of PEG molecules on the microsphere surface. PEG can act as an anti-adhesive barrier or binding to peptide to promote cell adhesion. Cells cannot adhere to PEG, which helps facilitate contact between the cells and the peptide.

Figure 3A shows the FTIR spectra of the microspheres after dopamine coating and PEG grafting modification. The strong signal at 1760 cm⁻¹ that appeared in all spectra could be attributed to the presence of carbonyl groups in the PLGA. The N-H bending vibration peak at 1,613 cm⁻¹ after dopamine coating, which might be associated with the amino group contained in polydopamine (pDA), could be seen in the spectra. This indicated the polymerization reaction of dopamine on the surface of the microspheres to form the pDA coating layer. From the spectra, it could be observed that the peak of 2,923 cm⁻¹ appeared after grafting PEG on the surface of pDA coated microspheres, suggesting that the tensile vibration of C-H was enhanced. The signal at 1,340 cm⁻¹ revealed the strengthening of the C-H bending vibration. This implied that PEG had been grafted on the surface of pDA-coated microspheres.

The NIH3T3 cells were inoculated on the surface of the Fe₃O₄-OA-PLGA, pDA/Fe₃O₄-OA-PLGA, and PEG/pDA/Fe₃O₄-OA-PLGA microspheres. After incubation for 12 h, the adherent cells were fixed and stained with DAPI nuclear dye, then observed and statistically analyzed. Fluorescence photographs (Figure 3B) showed that NIH3T3 cells could adhere to the pDA/Fe₃O₄-OA-PLGA and Fe₃O₄-OA-PLGA microspheres, but they could not adhere well on the PEG/pDA/Fe₃O₄-OA-PLGA microspheres. Statistical analysis (Figure 3C) showed that cell adhesion on the PEG/pDA/Fe₃O₄-OA-PLGA microspheres was significantly lower than that on the pDA-modified and control surfaces. These results suggest that PEG molecules grafted on the microspheres formed a surface that was more resistant to cell adhesion than the untreated microspheres.

In this study, selective adhesion to MSCs was achieved by grafting E7 peptide on PEGylated magnetic microspheres. The activation of grafted di-amino-PEG molecules on the microsphere surface depended on their free terminal amine groups. First, we attached sulfo-SMCC to the di-amino-PEG molecule on the surface of the microspheres. Then, the E7 peptide was ligated to the sulfo-SMCC molecule on the surface of the PEG/pDA/Fe₃O₄-OA-PLGA microspheres via the Cys residue at the C-terminal end of the peptide. The E7/pDA/Fe₃O₄-OA-PLGA microspheres were modified with E7 by pDA coating. Since only the peptide contained sulfur elements, the mapping of sulfur elements was used to characterize the grafting of the peptide. As seen in Figure 4A, the surfaces of E7/pDA/Fe₃O₄-OA-PLGA microspheres and E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres were uniformly covered with elemental sulfur. This result indicated that the E7 peptide was successfully immobilized on the magnetic microspheres.

The water contact angle was 79.76 ± 1.52° for Fe₃O₄-OA-PLGA microspheres, 57.46 ± 1.27° for pDA/Fe₃O₄-OA-PLGA microspheres, and 45.79 ± 0.81° for E7/pDA/Fe₃O₄-OA-PLGA microspheres (Figure 4B). Applying pDA and E7 to the microsphere surface decreased the water contact angle, indicating that they improved the hydrophilicity. This was also expected, because pDA and E7 comprise many hydrophilic functional groups, such as amino and hydroxyl groups. The surface of PEG/pDA/Fe₃O₄-OA-PLGA and E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres became more hydrophilic, with a contact angle of only about 39.13 ± 0.74° and 39.26 ± 0.87°, possibly due to the presence of PEG with excellent hydrophilicity on the modified surface (Lee et al., 1995).

The organic coating content on the prepared microsphere surface was calculated using the TGA method (Figure 4C). The mass loss values for Fe₃O₄-OA-PLGA, pDA/Fe₃O₄-OA-PLGA, PEG/pDA/Fe₃O₄-OA-PLGA, E7/pDA/Fe₃O₄-OA-PLGA, and E7/PEG/pDA/Fe₃O₄-OA-PLGA were at 91.038, 91.858, 97.021, 92.957, and 97.725 wt.%, respectively. E7 grafting of E7/pDA/Fe₃O₄-OA-PLGA and E7/PEG/pDA/Fe₃O₄-OA-PLGA was calculated to be 1.099 wt.% and 0.704 wt.%, respectively, which might be related to the coupling efficiency of sulfo-SMCC.

Because cells could not adhere to the PEG/pDA/Fe₃O₄-OA-PLGA microspheres, the PEG/pDA/Fe₃O₄-OA-PLGA microspheres were not evaluated in subsequent experiments.

The E7 peptide, which was bound on the surface of the microspheres, was initially shown to have a specific affinity for MSCs relative to NIH3T3 (fibroblasts) and RAW 264.7 cells (macrophages). Cell adhesion was detected 6 h after inoculation, as shown in Figure 6. The images revealed that the number of cells adhering to the Fe₃O₄-OA-PLGA surface was lower than that on the pDA/Fe₃O₄-OA-PLGA surface, regardless of cell type. The addition of dopamine coating on the microspheres increased non-specific cell adhesion, but the cell proportions did not change (Figure 6A). As previously reported (Madhurakkat Perikamana et al., 2015), pDA imparts a positive charge to the material surface, allowing the material to bind more readily to integral protein receptors of the cell membrane, thus improving cell adhesion. On the surface of E7/pDA/Fe₃O₄-OA-PLGA, the adhesion of NIH3T3 and RAW 264.7 cells was not higher than that for pDA/Fe₃O₄-OA-PLGA, whereas the number of adherent MSCs was increased. On E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres, the cells of MSCs were further increased, while that of NIH3T3 and RAW 264.7 cells were further reduced. Statistical analysis showed that at 6 h post-inoculation (Figure 6B), the proportion of adherent MSCs on the surface of E7/PEG/pDA/Fe₃O₄-OA-PLGA was more than two-fold higher than that of NIH3T3 and RAW 264.7 cells. These results indicated that the grafted PEG molecules and E7 peptide effectively resisted the non-specific adhesion of cells.

The selective adhesion of MSCs to the E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres was further verified through co-culturing MSCs with NIH3T3 and RAW 264.7 cells under static conditions (Figure 7A). The results showed that the total number of cells on the surface of Fe₃O₄-OA-PLGA microspheres was low, and the proportions of the three types of cells were comparable. A large number of adherent MSCs (red), NIH3T3 (green), and RAW 264.7 cells (blue) could be observed on pDA/Fe₃O₄-OA-PLGA microspheres. With the addition of E7 peptide, more red cells were observed on the E7/pDA/Fe₃O₄-OA-PLGA microspheres, indicating enhanced selectivity for MSCs adhesion. However, E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres exhibited more red cells and fewer green and blue cells than unmodified Fe₃O₄-OA-PLGA microspheres. This suggested that the PEG antifouling layer effectively blocked the non-specific adhesion of cells. Accordingly, the proportions of cells on the different microspheres were statistically analyzed (Figure 7B). The percentages of MSCs, NIH3T3, and RAW 264.7 cells cultured on Fe₃O₄-OA-PLGA were 33.19 ± 5.32, 39.32 ± 3.11, and 27.49 ± 3.51%, respectively. Similar to separate culture, when cells were co-cultured on the pDA/Fe₃O₄-OA-PLGA microspheres, there was no significant change in the proportion of each cell type, although there was an increase in the number of each cell type. With respect to E7/pDA/Fe₃O₄-OA-PLGA microspheres, the percentage of MSCs increased from 32.4 ± 3.58% to 49.9 ± 5.27%. Furthermore, on E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres, the percentage of MSCs was increased to 76.39 ± 5.67%. The above results led us to conclude that the selective adhesion of MSCs to E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres was significantly enhanced under static culture conditions. Compared with the situation when the 3 cells were inoculated separately, on the surface of E7/pDA/Fe₃O₄-OA-PLGA and E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres grafted with E7 peptide, the proportions of NIH3T3 and RAW 264.7 cells were decreased, which might be due to the increasing selective affinity of microspheres for MSCs, which caused more adherence of MSCs, and competitively reduced the adherence of NIH3T3 and RAW 264.7 cells.

Observing the morphology of the three types of cells on the microspheres (Figure 7C), corresponding to Figure 7A, the E7-grafted microspheres exhibited more adhered MSCs. This also suggests that E7 might influence the subsequent proliferation of MSCs by affecting their early adhesion (Figure 5A).

Because microspheres are often cultured in a dynamic environment as cell microcarriers, the selective capture of MSCs by the obtained microspheres under dynamic conditions was further investigated (Figure 8A). The total number of adherent cells on the surface of each microsphere decreased compared with that in static culture. Presumably, shear forces in the system may cause adhesion to take longer to achieve (Ning et al., 2013). After modification with dopamine, the number of cells on the pDA/Fe₃O₄-OA-PLGA microspheres increased (versus the Fe₃O₄-OA-PLGA microspheres group). Unsurprisingly, modification of the E7 peptide resulted in more MSC adhesion, as under static culture conditions. On the E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres, the adherent numbers of NIH3T3 and RAW 264.7 cells were also decreased. These results showed that the PEG antifouling layer could also inhibit non-specific cell adhesion under dynamic conditions. Accordingly, statistical analysis of cell proportions was performed (Figure 8B). The percentages of MSCs, NIH3T3, and RAW 264.7 cells cultured on Fe₃O₄-OA-PLGA microspheres were 34.76 ± 3.77, 31.35 ± 2.94, and 33.893 ± 2.71%, respectively. On pDA/Fe₃O₄-OA-PLGA microspheres, the proportions of cells were 37.37 ± 2.94, 29.36 ± 3.10, and 33.27 ± 1.29%. For E7/pDA/Fe₃O₄-OA-PLGA microspheres, MSCs increased to 52.16 ± 4.22%, while NIH3T3 and RAW 264.7 decreased to 19.75 ± 2.59% and 28.09 ± 1.58%. The proportion of MSCs on E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres further increased to 77.19 ± 3.05%. The results showed that the E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres were equally selective for MSCs under dynamic conditions compared with that in static culture.

MSCs can offer promising therapeutic potential for many diseases (Patel et al., 2013). Because of the limited fraction of MSCs collected from tissues, in vitro expansion of MSCs is required; nevertheless, loss of stemness and undesired differentiation of MSCs during in vitro culture can diminish their efficiency (Saei Arezoumand et al., 2017). Therefore, their stemness qualities must be preserved in vitro (McKee et al., 2017). To investigate the stemness retention and differentiation potential of MSCs expanded on the surface of E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres, the cells were inoculated for an additional 7 days and harvested; then, flow cytometric analysis and induced differentiation were carried out (Figure 9). The positive expression indicators CD29 and CD90 and the negative expression indicator CD34 by MSCs were taken as identification reference indicators. Flow cytometry (Figure 9A) showed that the cells did not express CD34, but did express CD29 and CD90. This was consistent with the surface marker characteristics of MSCs (Harting et al., 2008). ALP activity was a sign of early osteogenic differentiation in MSCs (Weinreb et al., 1990). Figure 9B showed positive ALP staining. Mineral deposition has been shown to be a late marker of osteogenesis, and calcium deposition can be measured using alizarin red staining (Liu et al., 2016). Significant calcium deposition (Figure 9C) was found in MSCs inoculated on E7/PEG/pDA/Fe₃O₄-OA-PLGA microspheres. With regard to lipogenic differentiation, oil red O staining (Figure 9D) showed the existence of lipid vacuoles. Alcian blue staining (Figure 9E) confirmed chondrogenic morphogenesis. The results demonstrated that cells growing on the microspheres still retained the ability to differentiate. Furthermore, we found that the number of MSCs captured on E7/PEG/pDA/Fe₃O₄-OA-PLGA under dynamic conditions was statistically significantly lower than the number of MSCs captured under static conditions (Supplementary Table S1). Consistent with the preceding results, it is possible that shear forces affected early cell adhesion (Ning et al., 2013), and such shear forces might, to some extent, also mimic the actual stress environment in vivo (Zhang et al., 2010).

Additionally, microspheres can be used not only in vitro, but also in vivo as supporting material and carriers for cell growth and delivery (Hiraoka et al., 2006; Kang et al., 2009; Kamali et al., 2019). Nevertheless, the non-specific capture of fibrocytes and inflammation-related cells may lead to fibrosis because of the complexity and diversity of the cells participating in the tissue repair procedure (Tabata, 2009). Hence, in order to achieve tissue regeneration without fibrosis, selective capture of MSCs by the E7-modified microspheres in the presence of fibrocytes and inflammatory cells is crucial. MSCs can selectively adhere to such microspheres, reducing the potential for fibrosis and inflammation caused by fibroblasts and inflammatory cells.