Male Sprague-Dawley (SD) rats aged 3–4 weeks were purchased from the Experimental Animal Center of Chongqing Medical University, China. MSCs were isolated using the adherent culture method as described (Cui et al., 2019). Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM)/F12 (Gibco; Thermo Fisher Scientific, Waltham, MA, United States) supplemented with 10% fetal bovine serum (FBS; Gibco) in 5% CO₂ at 37°C and 95% humidity and subcultured when the cell growth confluence was 70–80%. C6 rat glioma cells were cultured under the same conditions as MSCs. To identify the phenotype of MSCs, the expression of MSC surface antigens at passage 4 was detected via flow cytometry. Cells were incubated with phycoerythrin (PE)-conjugated monoclonal anti-rat CD29 (BD Biosciences, Franklin Lakes, NJ, United States), CD45 and CD34 (Abcam, Cambridge, MA, United States), and allophycocyanin (APC)-conjugated CD71 (BioLegend, San Diego, CA, United States), and percp1-conjugated CD90 (BioLegend) antibodies at 37°C for 30 min and subsequently analyzed by a flow cytometer (FACS Canto Plus; BD Biosciences).

Mesenchymal stem cells were seeded on 35 mm confocal dishes overnight, and the confluency was 30–50% when the cells were transfected. Small interfering RNAs (siRNAs) were designed and synthesized by Biomics Biotechnologies Inc. (Jiangsu, China) and are listed below.

Cells were transfected with siRNA targeting clathrin heavy chain (si-CHC) or dynamin2 (si-DNM2) or with negative control siRNA (si-NC) using Lipofectamine RNAiMAX (Thermo Fisher Scientific, Waltham, MA, United States) (Table 1). The siRNAs were diluted in Opti-MEM (Thermo Fisher Scientific), added to the cells and mixed gently at room temperature for 20 min, and then serum-free DMEM/F12 (without antibiotic) was added according to the manufacturer’s protocols. Four hours later, the medium was removed and replaced with fresh DMEM/F12 containing 10% FBS. The concentrations of si-CHC and si-DNM2 used in our experiments were 50 and 100 nM, respectively. After 72 h, live cell internalization confocal assays were conducted as described. The knockdown effects were measured with western blot or RT-qPCR experiments as described after 48–72 h of transfection.

Rat recombinant S100B (Cloud-Clone Corp., Wuhan, China) was dissolved in 10 mM phosphate-buffered saline (PBS) to 1 mg/ml, followed by fluorescent labeling with an Alexa Fluor488 Microscale Protein Labeling Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. To separate the labeled protein from unreacted dye, the conjugate was purified following the steps in the instructions.

Fluorescently labeled S100B (S100B-Alexa488; 500 ng) was subjected to 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions. The gel was fixed in 30% ethyl alcohol (v/v) and 10% ethanoic acid (v/v) for 30 min, washed with ddH2O, and imaged on a Gel Doc^TM XR+ (Bio-Rad, Hercules, CA, United States) gel imager. A 532/28 filter and Blue Epi Illumination were used to excite the Alexa Fluor488 dye.

Confocal images were acquired using the A1R laser confocal microscope (Nikon, Tokyo, Japan). A certain number of cells of similar size in each group were randomly selected for capturing. To compare the internalized vesicles and the mean fluorescence intensity (MFI) in each experiment, cells in each group were captured and analyzed under the constant channel parameters (for capturing), diameter and contrast in the same binary layer (for analysis). Pearson’s correlation coefficient and Mander’s overlap coefficient (MOC) was calculated for evaluating colocalization index. Consecutive Z-stack images were captured with the Nikon N-SIM, reconstructed into three-dimensional (3D) images, and processed into dynamic movies. Excitation for the Alexa488, the Alexa 555/TMR/Lysotracker Red, and Dylight649 chromophores or dye was produced by a 488-, 561-, and 638-nm laser, respectively. All images were captured under a 100× oil objective, denoised and analyzed with the NIS-Elements AR software (version 5. 01. 00; Nikon).

For S100B-Alexa488 and Dextran-TMR uptake assays, when the confluence was 50–60%, cells on each confocal dish were incubated with S100B-Alexa488 (0.1 μM) and 10 kDa Dextran-TMR (0.05 mg/ml, dextran conjugated to tetramethylrhodamine, Thermo Fisher Scientific) simultaneously at 37°C for different time periods (0.5, 1, 2, 6, 12, and 24 h) after 2 h of starvation. For Lyso-Tracker staining, MSCs were incubated with S100B-Alexa488 (0.1 μM) at 37°C for for aforesaid time periods. Then the cells were subjected to fresh medium containing 75 nM Lyso-Tracker Red (Beyotime, Shanghai, China) for 30 min. For endocytic inhibitor assays, the DyLight649-coupled ChromPure Rat Transferrin (15 μg/ml, Jackson ImmunoResearch, West Grove, PA, United States) and the Alexa Fluor 555-coupled cholera toxin subunit B (CTB, 0.5 μg/ml, Thermo Fisher Scientific) are used. MSCs were subjected to S100B-Alexa488 (0.1 μM) and Transferrin-Dylight649 or CTB-Alexa555 for 2 h at 37°C after preincubation with inhibitors as described. For genetic intervention experiments, MSCs were subjected to S100B-Alexa488 (0.1 μM) and Transferrin-Dylight649 (15 μg/ml) for 6 h following 72 h of siRNA transfection. At the end of all internalization assays, cells were gently washed with PBS three times to remove unconjugated dyes. Culture medium without red phenol (Gibco) was added to sustain cell viability, and live cells were subjected to confocal system, images were captured and analyzed with NIS Elements AR software as described above.

The endocytosis inhibitors used in our experiments were pitstop-2 (Abcam), dyngo-4a (Selleck, Houston, TX, United States), nystatin (Selleck), and MβCD (Selleck) dissolved in dimethyl sulfoxide (DMSO) or water. MSCs were seeded on confocal dishes overnight, grown to 70% confluency, the preincubation time and concentrations used are pitstop-2 (5 μM) for 30 min, dyngo-4a (20 μM) for 15 min, nystatin (30 μM) for 1 h and MβCD (2 mg/ml) for 1 h, respectively. After pretreatment with inhibitors, cells were subjected to S100B-Alexa488 and Transferrin-Dylight649 or CTB-Alexa555 as described above without washing out the inhibitors. An equal volume of 0.2% DMSO as an inhibitor diluent was used as a control, except for MβCD inhibitor experiments.

Total protein was prepared with a Whole Cell Lysate Extraction Kit (KeyGEN, Nanjing, China), a BCA Assay Kit (KeyGEN) was used to detect the concentration, and equal amounts of protein were separated on SDS-PAGE gels. The specific primary antibodies were rabbit anti-rat CHC and DNM2 (1:1000; Bimake, Shanghai, China), S100B (1:1000, Abcam), and mouse anti-rat GAPDH (1:1000; Zhongshan Golden Bridge, Beijing, China). The membranes were incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit or goat anti-mouse secondary antibody (Zhongshan Golden Bridge). GAPDH was used as a loading control, the protein bands were detected by ECL (Millipore, Billerica, MA, United States), and the intensities were quantified with Quantity One software (version 4.6.2; Bio-Rad, Hercules, CA, United States).

Total RNA of cells after 48 h of transfection was extracted with TRIzol reagent (TaKaRa, Dalian, China) and then used for cDNA synthesis using a PrimeScript RT Reagent Kit (TaKaRa). Quantitative real-time PCR was conducted using a SYBR Green Dye Kit (TaKaRa) with gene-specific primers. The primers purchased from TSINGKE Biological Technology (Beijing, China) were as follows: CHC, forward, 5′-GGCGAGTATCAGGCAGCAGTTG-3′, and reverse, 5′-CCCATCTACACAGGCAAAGCAGAC-3′. DNM2, forward, 5′-CGAGGATGGAGCACAAGAGAACAC-3′, and reverse, 5′-GCCACATAGGAGTCCACCAAGTTG-3′. β-actin (forward, 5′-GGAGATTACTGCCCTGGCTCCTA-3′; and reverse, 5′-GACTCATCGTACTCCTGCTTGCTG-3′) was used as the housekeeping gene to normalize mRNA expression. A real-time QuantStudio 3 system (Thermo Fisher Scientific) was used, and the relative gene expression was calculated by the 2^–ΔΔCt method.

Mesenchymal stem cells were grown overnight in a 96-well plate. When the cell confluence reached 50–60%, cells in each well were washed three times with serum-free medium, and the medium was replaced with serum-free medium containing different concentrations of pitstop-2 and dyngo-4a. The cells were incubated with drugs for 2.5 h, then CCK8 reagent (ATGene, Taiwan, China) was added to the medium and allowed to react for 4 h at 37°C. A microplate reader (BioTek, Winooski, VT, United States) was used to detect the optical density values at a wavelength of 450 nm.

Data are presented as the median and interquartile range (IQR) or as the means ± SEM unless stated otherwise by GraphPad Prism 7 software (GraphPad Software Inc., La Jolla, CA, United States). Each experiment was performed at least three times. P-values were estimated by Kruskal–Wallis one-way ANOVA with Bonferroni correction or by Mann–Whitney U test unless stated otherwise. Statistical analysis was conducted with IBM SPSS software (version 25; Armonk, NY, United States).

Specific surface markers of rat MSCs at passage 4 were identified with flow cytometric assays. The cells were positive for the MSC surface markers CD29 (95.2%), CD73 (98.3%), and CD90 (98.3%) and negative for the hematopoietic stem cell surface markers CD34 (1.05%) and CD45 (1.56%) (Figure 1).

To enable tracking by fluorescence microscopy, recombinant S100B protein conjugated to AlexaFluor488 dye (S100B-Alexa488) was produced by fluorescently labeling purified recombinant rat S100B. To check the quality of labeling and avoid experimental artifacts caused by unconjugated dye, we performed SDS-PAGE followed by fluorescence imaging. The molecular weight detected upon SDS-PAGE matched the expected molecular weight of S100B (12 kDa), and S100B-Alexa488 was almost completely free of any other impurities (Figure 2A). To test the integrity of surface antigens, S100B-Alexa488 probe and C6 glioma cell lysates (positive control) were subjected to western blot assays. S100B was detected in both samples, while the internal reference GAPDH was only detected in the C6 cell lysates (Figure 2B). These results indicated that recombinant S100B protein was labeled successfully and could be recognized by the S100B antibody.

To visualize the possible endocytosis of exogenous S100B, fluorescent dextran (Dextran-TMR), a fluid-phase endocytic marker, was added to the culture medium simultaneously with S100B-Alexa488. Confocal microscopy and structured illumination microscopy (SIM) were used to track this process. To monitor the internalization kinetics of S100B-Alexa488 and Dextran-TMR in detail, we evaluated the number, MFI and colocalization level of fluorescent vesicles internalized per cell at six sequential time points.

The S100B-Alexa488 protein (in green) encountered Dextran-TMR (in red) and thus generated a colocalization signal (in yellow), indicating their strong colocalization (Figure 3A). We evaluated the colocalization level of two fluorescent probes by measuring both the Pearson’s correlation coefficient (PCC) and the MOC, the colocalization index (both the PCC and the MOC) of S100B-Alexa488 and Dextran-TMR was around 0.5 at 0.5 h, this value increased with time and reaches nearly 0.8 at the latter stage, suggesting a partial overlap between the two probes at the beginning of incubation and they have a great potential colocalized in the same organelle in the subsequent endocytic process (Figure 3B). Finally, consecutive Z-stack scanning was performed, and a 3D-reconstructed movie was obtained by using super-resolution SIM (Supplementary Video 1). The 3D movie clearly showed the spatial colocalization of the two fluorescent probes, confirming that they were routed into the same compartments, which is direct evidence that S100B-Alexa488 enters the cell through endocytosis.

Extracellular S100B-Alexa488 were captured in vesicle-like structures, visualized as numerous spots in the cytoplasm, and the MFI represents the content of S100B-Alexa488 in each vesicle. From a short time period perspective, the number and MFI of the two fluorescent probes were hardly detectable at 0.5 or 1 h. Notably, the number of S100B-Alexa488-positive vesicles was clearly detected after 2 h of incubation, and this value doubled at 6 h, peaked at 12 h and slightly decreased at 24 h (Figure 3C, left panel). The MFI of S100B-Alexa488-positive vesicles increased continuously from 0.5 h to a peak value at 24 h (Figure 3C, right panel and Figure 3A, fourth panels). Likewise, the number of Dextran-TMR-positive vesicles increased in a manner similar to that of S100B-Alexa488-positive vesicles (Figure 3D, left panel), and the MFI of it increased over a period of time and slightly decreased at 24 h (Figure 3D, right panel and Figure 3A, fourth panels). The above results showed that the number and the MFI of the two probes had roughly the same variation tendency at different time points, indicating that they have similar time-dependent internalization kinetics.

In general, the above data indicate that exogenous S100B can be efficiently taken up by MSCs simultaneously with the fluid-phase endocytic marker dextran.

To explore the association of S100B-Alexa488 with Lyso-Tracker-labeled lysosomes in the dynamic transportation process, we investigated their colocalization level and the number of fluorescent vesicles per cell using confocal microscopy and SIM. At different time points, S100B-Alexa488 accumulated within the lysosomes (Figure 4A). The colocalization index (both the PCC and the MOC) of S100B-Alexa488 and lysosomes was around 0.3–0.4 at 0.5 h, and this value increased with time, peaked at 6 h and remained stable during the subsequent incubation time, indicating that at the beginning of incubation, a fraction of S100B-Alexa488–positive vesicles reached lysosomes and after prolonged exposure, S100B-Alexa488 was highly colocalized with lysosomes (Figure 4B). In addition, the colocalization of S100B-Alexa488 and lysosomes was reconstructed into a 3D movie that clearly showed the locations of them inside the cell (Supplementary Video 2). Moreover, we observed that during 24 h of incubation, the number of S100B-Alexa488-positive vesicles increased over time (Figure 4C), which was consistent with our previously described results (Figure 3C, left panel), and the number of S100B-Alexa488-positive vesicles that were colocalized with lysosomes increased with time (Figure 4D). The accumulation of S100B-Alexa488 in lysosomes indicated that it is internalized and transported through endolysosomal pathways.

Clathrin-mediated endocytosis is a well-studied and well-characterized endocytic pathway that is regulated by a host of proteins, including clathrin and dynamin. We investigated whether S100B-Alexa488 is internalized into MSCs by CME by studying the effects of chemical inhibitors and RNA interference (RNAi) of this cellular pathway. To assess the inhibition efficiency, fluorescent transferrin (Transferrin-Dylight649), a well-established protein internalized into the cytoplasm by the CME pathway, was used as a parallel control. DMSO was used to dissolve chemical inhibitors, and we confirmed that DMSO had no adverse effect on endocytosis, as assessed by the number or MFI of S100B-Alexa488 and Transferrin-Dylight649-positive vesicles (Supplementary Figure 1A–C).

Pitstop-2 acts by blocking the initial binding of ligand to the terminal domain of clathrin, whereas dyngo-4a inhibits dynamin, a membrane scission protein that forms a helical polymer around the constricted neck of a vesicle and mediates its scission from the plasma membrane. To assess the cytotoxicity of these drugs, we used CCK8 assays to screen drug concentrations. The selected concentrations of pitstop-2 and dyngo-4a were 5 and 20 μM, respectively, which corresponded to 50% inhibition of cell viability (Figures 5G, 6G). MSCs were exposed to both S100B-Alexa488 and Transferrin-Dylight649 for 2 h and were pretreated with the drugs or DMSO (control) for the indicated time points. Both fluorescent probes were localized to the periphery and to the perinuclear cytoplasmic regions in the control group, while few fluorescent dots were detected on the plasma membrane in the pitstop-2- or dyngo-4a-treated group (Figures 5A, 6A). There was a significant reduction in the number of Transferrin-Dylight649-positive vesicles in the presence of pitstop-2 or dyngo-4a compared with control cells (Figures 5B, 6B, left panels), confirming that the CME pathway was blocked successfully at the concentrations used. These drugs clearly inhibited the internalization of S100B-Alexa488, as the number and MFI of S100B-Alexa488-positive vesicles were significantly reduced after treatment with pitstop-2 or dyngo-4a (Figures 5C,6C).

Since chemical inhibitors may have many off-target effects, to further verify the above results, siRNA-mediated knockdown was used to investigate the role of clathrin and dynamin in endocytosis. MSCs were transfected with si-CHC and si-DNM2 in the experimental group or si-NC in the control group. The knockdown efficiencies were measured at the mRNA and protein levels, showing that CHC and DNM2 were successfully downregulated (Figures 5H, 6H). The number of Transferrin-Dylight649-positive vesicles was significantly decreased in cells transfected with si-CHC or si-DNM2 compared to those transfected with si-NC (Figures 5E, 6E, left panels), confirming that the CME pathway was inhibited successfully by the specific siRNAs. siRNA-mediated CHC knockdown greatly reduced the number and MFI of S100B-Alexa488-positive vesicles (Figures 5D,F). Likewise, DNM2 silencing significantly reduced the number and MFI of S100B-Alexa488-positive vesicles (Figures 6D,F). The above results suggested that knockdown of CHC and DNM2 inhibits the internalization of S100B-Alexa488.

In summary, these results indicate that inhibition of clathrin and dynamin significantly reduced the uptake of S100B-Alexa488, confirming that S100B-Alexa488 is internalized by MSCs via the CME pathway.

The above results indicated that the internalization of S100B-Alexa488 is dependent on dynamin, which is also involved in multiple CIE pathways. Since the majority of CIE pathways share a dependency on lipid rafts, and membrane cholesterol is an essential component of lipid rafts, we then investigated the role of lipid raft-mediated endocytosis, by using nystatin or MβCD, chemical inhibitors extracting or sequestering cholesterol on the cellular membranes, respectively. To assess the inhibition efficiency, fluorescent CTB (CTB-Alexa555), a maker served to label lipid rafts was used as a parallel control. Treatment with nystatin or MβCD significantly decreased CTB-Alexa555 uptake, assessed by the number and MFI of CTB-Alexa555-positive vesicles, confirming the inhibitory effect of these drugs on lipid rafts (Figures 7B,E). The number of S100B-Alexa488-positive vesicles were greatly reduced in cells treated with nystatin compared with control cells (Figures 7A,C). Likewise, there was a reduction in the number of S100B-Alexa488-positive vesicles after treatment with MβCD (Figures 7D,F). The above results indicate that the presence of cholesterol is indispensable for S100B-Alexa488 uptake, suggesting that lipid raft-mediated endocytosis is involved in the internalization of S100B-Alexa488 by MSCs.