Recombinant human HMGB1 protein (Sigma-Aldrich, St. Louis, MO, USA) was commercially purchased. A concentration of 25 ng/ml protein was used in the experiments detailed below. The Rap1 inhibitor (Cell Signaling Technology, Inc., Danvers, MA, USA), which is a transferase inhibitor, was used at a concentration of 5 µM. The reagents listed above were handled and used according to the manufacturer's protocols.

MSCs (Cyagen Biosciences Inc., Guangzhou, China) were commercially purchased. Adherent cells were trypsinized using 0.25% trypsin (Cyagen Biosciences Inc.) for ~30 sec and passaged after the cell confluence had reached ~80%, and the cells at passages 3 to 5 were used in the experiments detailed below. Typically, these cells exhibited the capacity of differentiation into osteoblasts, adipoblasts and chondrocytes under specific inductive conditions.

Cell migration was performed with Transwell chambers (pore size, 8-µm diameter; Corning Costar, Inc., Corning, NY, USA). Complete™ medium (Cyagen Biosciences Inc.) containing 0.1% fetal calf serum (FCS; Cyagen Biosciences Inc.) was added into the wells of a 24-well plate, and subsequently, serum-starved MSCs (1×10⁵) suspended in a volume of 100 µl Complete™ medium containing 0.1% FCS were added into the upper chamber. Prior to the addition of HMGB1, the transwell plate (with MSCs in the upper chamber and medium containing 0.1% FCS only in the lower chamber), was first incubated at 37°C for 1 h. Following the addition of HMGB1, the plate was subsequently incubated at 37°C for 3 h, followed by membrane fixation with 4% paraformaldehyde (Beyotime Institute of Biotechnology, Haimen, China) and staining with 0.1% crystal violet (Beyotime Institute of Biotechnology). The membrane was subsequently washed, and the cells on the underside of the membrane were observed under a light microscope (Leica DMI/LM; Leica Microsystems GmbH, Wetzlar, Germany). Numbers of cells were counted in five to ten random fields for each membrane.

Cell migration was determined using a scratch assay. The cells were cultivated to 90% confluence on 12-well plates. The groups were as follows: Control, without HMGB1; treatment, MSCs cultured with 25 ng/ml HMGB1; supernatant, MSCs pretreated with 25 ng/ml HMGB1 and following 48 h, the supernatant of the MSCs was extracted to culture a fresh batch of MSCs. Subsequently, cell scrapers (Corning Costar Inc.) were used to scratch the confluent cells. The extent of cellular growth was observed at 0 and 48 h. All the experiments were repeated in triplicate.

Cells were harvested and lysed in radioimmunoprecipitation assay buffer containing proteinase inhibitors (Cyagen Biosciences Inc.). Following measurement of the protein concentration using a BCA kit (Beyotime Institute of Biotechnology), protein samples were separated using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and the proteins were transferred onto a polyvinylidene difluoride (PVDF) membrane by western blotting. The membranes were blocked in 5% skimmed milk for 1 h, and incubated with the following antibodies: Rabbit anti-human GTP-Rap1 (Active Rap1 Detection kit, cat. no. CST8818; dilution, 1:500) from Cell Signaling Technology, Inc., rabbit anti-human Rap1 (cat. no. ab47234; Abcam, Cambridge, MA, USA; dilution, 1:500) and rabbit anti-human β-actin (cat. no. sc-130301; Santa Cruz Biotechnology, Inc., Dallas, TX, USA), separately at 4°C overnight. After an incubation with goat anti-rabbit peroxidase-linked secondary antibody (cat. no. 31210; Thermo Fisher Scientific, Inc., dilution, 1:5,000) at 25°C for 3 h, immunoreactive proteins were visualized using an enhanced chemiluminescence (ECL) reagent (Thermo Fisher Scientific, Inc.). Relative quantification of bands in the western blot was performed using ImageJ software [National Institutes of Health (NIH), Bethesda, MA, USA).

Soluble proteins in the medium of the stromal cell lines were measured using the Human Cytokine Array G1000 (AAH-CYT-G1000; RayBiotech, Inc., Norcross, GA, USA), according to the manufacturer's protocol. These arrays are able to detect up to 120 proteins. Stromal cells were plated 3 days prior to the experiment in Dulbecco's modified Eagle's medium (DMEM; Cyagen Biosciences Inc.) containing 10% FBS, and were 75–90% confluent when the cell lysis solution was collected and filtered. Medium containing 10% FBS was also hybridized to the arrays, and used subsequently for normalization. Ten technical and biological replicates were performed, and these demonstrated a very high correlation (correlation coefficient >0.9; data not shown). Hybridization was performed overnight at 4°C. All slides were scanned using a GenePix^® 4000B Microarray Scanner (Axon Instruments, Inc., Union City, CA, USA) and analysed using the software GenePix^® Pro 6.0. The F532 median - 2B532 score was used, and averaged across triplicates on each array. The results were subsequently normalized using internal controls, and the values for cytokines in clear medium containing 10% FBS were subtracted.

Statistical significance was performed using the two-tailed Student's t-test, assuming equal variances. The chi-squared test was used to compare rates. P<0.05, P<0.01 and P<0.001 were taken to indicate statistically significant values.

To investigate the effect of HMGB1 on the migration of MSCs, a scratch assay was performed (Fig. 1). The MSCs of all groups moved towards the blank area to a certain extent after 48 h, indicating a certain level of migrational ability. Compared with the control group, MSCs cultured with 25 ng/ml HMGB1 significantly surpassed the baseline (P<0.05), indicating that the migrational ability of MSCs was increased on stimulation with HMGB1 (Fig. 1B and C). In another experiment, MSCs were pretreated with 25 ng/ml HMGB1. After 48 h, the supernatant of the MSCs was extracted to culture a new batch of MSCs (Fig. 1D), and a scratch assay was performed. A greater number of the MSCs from the second treatment crossed the baseline compared with that observed in the group of MSCs directly cultured with HMGB1 (P<0.05). This confirmed that HMGB1 could significantly increase the migrational ability of MSCs (Fig. 1E).

To determine the influence of HMGB1 stimulation on human MSCs at the protein level, a Quantibody^® array was performed. The treatment group was stimulated with 25 ng/ml HMGB1 and compared with non-stimulated cultures (Fig. 2).

Underlying our search criteria of a differential synthesis (>2.0-fold change or <0.5-fold change), the bioinformatics data analysis identified that seven cytokines were differentially synthesized between the two groups. Increased cytokine synthesis (relative level, >1.5-fold change) of bone morphogenetic protein-4 (BMP-4), neurotrophin-3 (NT-3), LIGHT [or tumor necrosis factor superfamily member 14 (TNFSF14)], monocyte chemoattractant protein 4 (MCP-4), Dtk and macrophage inflammatory protein-1β (MIP-1β) following induction with HMGB1 was measured (Fig. 3). A reduced level of cytokine synthesis (relative level, <0.5-fold change) was observed for β-nerve growth factor (β-NGF; Fig. 3A and B).

Hierarchical clustering of those seven cytokines with normalized cytokine synthesis values disclosed two distinct groups: The HMGB1-treated group and the non-stimulated control group. Six differentially secreted cytokines were visualized as being induced, and one cytokine as being repressed. Notably, among the seven cytokines, three of them were identified as being involved in the nuclear factor-κB (NF-κB) signaling pathway, namely, MCP-4, MIP-1β and LIGHT. Subsequently, two cytokines were identified as being involved in the neurotrophic factor-mediated Trk receptor signaling pathway: NT-3 and β-NGF. Seven differentially secreted cytokines were also revealed to be associated with the response to an external stimulus (7), cell migration (6), localization of the cell (6), regulation of programmed cell death (5) and the immune system process (5), according to Gene Ontology using the Database for Annotation, Visualization and Integrated Discovery (DAVID). As expected, numerous differentially synthesized cytokines were associated with more than one biological process.

Underlying our search criteria of a differential secretion, bioinformatics data analysis resulted in the selection of three cytokines (relative level, >2.5-fold control). These were visualized as being induced between the two groups. Increased levels of cytokine secretion were measured for chemokine ligand 4 (CCL4), CCL13 and TNFSF14 following induction with HMGB1. The detected differentially secreted cytokines serve roles in other molecular and cellular functions, including cellular growth and proliferation, cell death, cell morphology and cellular development. On the basis of the three differentially secreted cytokines, the cytokine-cytokine receptor interaction pathway (0460hsa in the Kyoto Encyclopedia of Genes and Genomes database) was the dominant pathway that was influenced by HMGB1. Thus, the secretion of CCL5 and CCL26 was induced (Fig. 3C).

The present study identified that six cytokines, of the differentially synthesized cytokines, are involved in cell migration based on the results of the Quantibody^® array. Subsequently, among the six cytokines, β-NGF was identified as being involved in the Rap1 signaling pathway. Therefore, the Rap1 signaling pathway was proposed to be significant to the current study. However, a limitation of the study is that the array was not repeated three times. Thus, to verify the hypothesis that the Rap1 signaling pathway of MSCs is also activated by HMGB1 stimulation, western blot analysis was performed. Compared with the expression of GTP-Rap1 in the control group, that in MSCs increased significantly following treatment with 25 ng/ml HMGB1. By contrast, the expression of GTP-Rap1 declined significantly following treatment with the Rap1 inhibitor. These results indicated that HMGB1 was able to activate the Rap1 signaling pathway in MSCs. However, the Rap1 inhibitor demonstrated marked inhibition of the Rap1 signaling pathway activated by HMGB1 (Fig. 4A).

To investigate whether activation of the Rap1 signaling pathway could increase the mobility of MSCs, a transwell migration assay was performed. The number of MSCs migrating towards the other side of the membrane increased markedly following treatment with HMGB1. However, following treatment with the Rap1 inhibitor, the number of migrating cells was markedly reduced compared with that in the group that was not treated with the inhibitor. This result suggested that activation and inhibition of the Rap1 signaling pathway significantly enhanced and suppressed the migration of MSCs, respectively (Fig. 4B–E).