This study was approved by the Institutional Review Board of Shandong University and written informed consent was obtained from all patients, and the Hospital Ethics Committee approved the experiments.

Human glioma cell lines (U87 and U251) were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Both U87 and U251 cells had been recently authenticated based on cross-species checks, DNA authentication and quarantine. The cell lines were grown in Dulbecco's modified Eagle's medium (DMEM; HyClone, Logan, UT, USA) supplemented with 10% fetal bovine serum (Gibco BRL, Gaithersburg, MD, USA) in a humidified incubator with 5% CO₂ at 37°C. Six normal brain tissues were collected from patients undergoing internal decompression surgery following severe traumatic brain injury. Twenty-five human glioma tissues, including 12 low-grade glioma tissues (5 grade-I and 7 grade-II tumours) and 13 high-grade glioma tissues (3 grade-III and 10 grade-IV tumours) were obtained from the Department of Neurosurgery, Qilu Hospital of Shandong University.

Recombinant human MIF (rhMIF; Peprotech, Inc., Rocky Hill, NJ, USA), Y27632 (Selleck Chemicals, Houston, TX, USA), IL-4, GM-CSF (both from Peprotech, Inc.), LPS (Sigma-Aldrich), CCL21 (Peprotech, Inc.) and DAPI (Beyotime Institute of Biotechnology, Shanghai, China) were obtained from indicated companies. siRNA negative control and siMIF were designed and purchased from GenePharma (Shanghai, China). The siRNA sequences used were the following: siRNA negative control forward, 5′-UUCUCCGAACGUGUCACGUTT-3′ and reverse, 5′-ACGUGACACGUUCGGAGAATT-3′; siMIF forward, 5′-CCGAUGUUCAUCGUAAACATT-3′ and reverse, 5′-UGUUUACGAUGAACAUCGGTT-3′. Cell transfection experiments were performed with nucleic acids using Lipofectamine 2000 (Invitrogen Life Technologies, Carlsbad, CA, USA) according to the manufacturer's instructions.

The activity of RhoA and Rac1 GTPases in cell lysates from control and MIF-treated glioblastoma cells was assessed using a G-LISA^® assay to measure the GTP-bound form of RhoA and Rac1 (Cytoskeleton, Inc., Denver, CO, USA). Briefly, glioblastoma cells were incubated with MIF (100 ng/ml, 30 min). Cell lysates were incubated on RhoA and Rac1 GTPase affinity plates and color developed using HRP detection reagent mixture. Samples were read on a microplate reader (Bio-Rad, Berkeley, CA, USA).

qPCR was conducted to measure the expression levels of DCs. Total RNA was isolated using RNAiso Plus (Takara). Total RNA (0.5–1 μg) was reverse-transcribed with a ReverTra Ace qPCR RT kit (FSQ-101, Toyobo) according to the manufacturer's protocol to synthesize cDNA. Real-time PCR was performed using a SYBR Premix Ex Taq^TM kit (Toyobo) with specific primers. The primers used were the following: CD80 forward, 5′-AAACTCGCATCTACTGGCAAA-3′ and reverse, 5′-GGTTCTTGTACTCGGGCCATA-3′; CD83 forward, 5′-AAGGGGCAAAATGGTTCTTTCG-3′ and reverse, 5′-GCA CCTGTATGTCCCCGAG-3′; CD86 forward, 5′-CTGCTCAT CTATACACGGTTACC-3′ and reverse, 5′-GGAAACGTCGT ACAGTTCTGTG-3′. The reactions were performed by using a LightCycler 2.0 Instrument (Roche Applied Science, Mannheim, Germany). mRNA levels were normalized to GAPDH. All data for each sample were collected in triplicate. The fold changes were calculated by relative quantification (2^−ΔΔCt).

To obtain U251 and U87 GFP-LC3 stable cell lines, we cloned LC3 inserts to the plenty-N-GFP vectors and generated lentiviruses by GenePharma. U87 and U251 cells were infected with the lentiviruses and selected stable clones with G418 (Sangon Biotech Co., Ltd., Shanghai, China). U87 and U251 GFP-LC3 stable cell lines were transfected with siMIF or treated with rhMIF and fixed in 4% paraformaldehyde. GFP-LC3 puncta formation assay was determined by capturing images using Olympus microscope (DP72; Olympus, Tokyo, Japan). Cells with ≥5 puncta were considered as GFP-LC3 puncta-positive cells. The percentage of GFP-LC3 puncta-positive cells was quantified by counting 200 GFP-LC3 stable cells.

Glioblastoma cells and DCs plated on 24-well plate were grown for 24 h and then cultured in serum-free medium for 16 h. After incubating with or without rhMIF (100 ng/ml, 24 h), glioblastoma cells were washed once with phosphate-buffered saline (PBS) and DCs were dropped on a glass slide. Then they were fixed with 4% formaldehyde in PBS for 10 min. The fixed cells were washed twice with PBS and permeabilized with 0.3% Triton X-100 in PBS. Cells were then stained with rhodamine-phalloidin (Cytoskeleton, Inc.) in PBS and cell nuclei were counterstained with DAPI. Images were captured by Olympus microscope.

Total protein was extracted from tissues and cells using RIPA buffer (Beyotime Institute of Biotechnology) with 1% phenylmethyl sulfonyl fluoride, and protein concentration was determined by the BCA method (Beyotime Institute of Biotechnology). Proteins were separated using 10–15% SDS-PAGE and transferred onto polyvinylidene difluoride membranes (Millipore, USA). The membranes were blocked by 5% skim milk blocking buffer for 1 h and then incubated in the primary antibodies at 4°C overnight. After washing with TBST, the blots were incubated with horseradish peroxidase-conjugated secondary antibodies at room temperature for 1 h. Finally, protein bands were visualized by enhanced chemiluminescence (ECL) (Millipore) and detected using an ECL detection system (Thermo Fisher Scientific, Inc., Beijing, China) and quantified with Quantity One software. The following primary antibodies were used: rabbit anti-LC3B, p62 and p-MYPT1 were purchased from Cell Signaling Technology, Inc. (Danvers, MA, USA); rabbit anti-ROCK1 was purchased from Abcam (Cambridge, UK); rabbit anti-MYPT1 was purchased from ProteinTech Group, Inc. (Wuhan, China); rabbit anti-GAPDH was purchased from Goodhere Biotechnology Co., Ltd. (Hangzhou, China). The relative integrated density values were measured based on the GADPH protein as the control.

Paraffin-embedded samples were sliced and mounted on microscopic slides. Rabbit anti-MIF (1:200 dilutions), anti-LC3B (1:200 dilutions) and anti-CD1a (ZSGB-Bio, Beijing, China) antibodies were used as the primary antibodies. Heat-induced epitope retrieval was performed with a microwave in 10 mmol/l citric acid buffer at pH 7.2. The samples were incubated with the antibody overnight in a humidified chamber at 4°C followed by incubation with a horseradish peroxidase-conjugated secondary antibody (ZSGB-Bio). Finally, 3,3′-diaminobenzidine tetrahydrochloride (DAB) was used to reveal the signal. The total immunostaining score was estimated using both the percentage of positively stained tumour cells and the staining intensity. The percentage positivity was scored as ‘0’ (<5%, negative), ‘1’ (5–25%, sporadic), ‘2’ (25–50%, focal), or ‘3’ (>50%, diffuse). The staining intensity was scored as ‘0’ (no staining), ‘1’ (weakly stained), ‘2’ (moderately stained), or ‘3’ (strongly stained). Both the percentage of positive cells and the staining intensity were evaluated under double-blind conditions. The immunostaining score was calculated as the percentage positive score multiplied by the staining intensity score and ranged from 0 to 9.

DC migration was assessed as previously described (32). Briefly, the lower chamber of 24 Transwell plates with polycarbonate membranes and 5-μm pore size (Corning, Inc.) was filled with 200 μl of RPMI-1640 media containing 10% FBS and CCL21 (100 ng/ml). Next, iDC and mature DC (mDC) (1×10⁵/100 μl medium) were seeded in the upper chamber, and plates were incubated for 2 h at 37°C. Cells in the lower chamber were counted. Glioblastoma cell migration was evaluated using a Transwell chamber (8-μm pore size; Corning, Inc.). Cells that did migrate to the lower surface were fixed and stained with eosin solution and counted under a microscope (Olympus). Five random views were used to count the cells, and the independent experiments were repeated three times.

Glioblastoma cells (500) were seeded in 6-well plates and cultured for 4 weeks at DMEM containing 10% FBS with indicated treatments. Then the colonies were washed three times with PBS, and fixed with 75% ethanol for 10 min, dried and stained with 0.1% crystal violet solution (Beyotime Institute of Biotechnology) for 10 min. Images were taken and the colonies were counted under the light microscope.

PBMCs were isolated from leukocyte-enriched buffy coats of healthy volunteers by centrifugation with Ficoll-Paque Plus (Sigma-Aldrich). Monocytes were enriched from PBMCs by positive selection using anti-CD14-conjugated magnetic MicroBeads (Miltenyi Biotec). Monocytes were cultured at 1×10⁶/ml in complete RPMI medium containing 1,000 U/ml GM-CSF and 500 U/ml IL-4 under 37°C, 5% humidified CO₂ for 5 days to generate iDCs. In the middle of the 5 days, half of the medium was replaced by fresh medium containing cytokines. To induce maturation, 1 μg/ml LPS was added in iDCs for another 2 days. To investigate the influence of MIF on DC maturation, 100 ng/ml rhMIF were added during or after the maturation.

Data analyses were conducted with SPSS 16.0 (SPSS, Inc., Chicago, IL, USA) and GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA, USA). Data were analyzed using one-way ANOVA, Student's two-tailed t-test. Data are presented as the mean ± standard deviation (SD) of three independent experiments, followed by Dunnett's test for multiple comparisons of the means. All tests were two-tailed, and p<0.05 was considered statistically significant.

We first investigated whether RhoA-ROCK1 activity was enhanced by MIF in glioblastoma cells. G-LISA assays were performed to measure the Rho GTPase activity. The RhoA GTPase activity increased following stimulation with rhMIF, but Rac1 GTPase was not affected in glioblastoma cells (Fig. 1A). Next, three GBM cell lines were treated with rhMIF, and actin filaments were detected by rhodamine-phalloidin staining. Actin polymerization was strongly enhanced in GBM cells treated with rhMIF as there was increased F-actin stress fibre formation compared to that of the control, indicating that ROCK1 was activated by rhMIF (Fig. 1B). To further confirm the activity of ROCK1, the phosphorylation of MYPT1 was assessed by western blot analysis. MIF significantly increased the phosphorylation of the MYPT1 in a dose-dependent manner (Fig. 1C). In contrast, the expression of ROCK1 was unchanged following rhMIF stimulation, suggesting that ROCK1 was regulated only at the enzyme level. These data demonstrated that MIF activates the canonical RhoA-ROCK1 pathway.

To explore the role of MIF in autophagic activity in glioblastoma, we performed LC3 conversion and GFP-LC3 puncta formation assays in both U87 and U251 cell lines. siMIF used to neutralize the level of endogenous MIF was transfected into U87 and U251 cells. The expression of LC3B-II decreased, suggesting that MIF knockdown suppressed autophagy in glioblastoma cells (Fig. 2A). We also examined the localization of GFP-LC3 by fluorescence microscopy in siMIF-transfected U87 and U251 cells stably expressing the GFP-LC3 fusion protein. There was a significant decrease in GFP-LC3 puncta in siMIF-transfected cells compared with that in the negative control cells (Fig. 2B). Both cell lines were stimulated by rhMIF, and autophagy was enhanced, as indicated by the increased expression of LC3B-II and percentage of GFP-LC3 puncta-positive cells (Fig. 2A and B). However, the effect of rhMIF on autophagy of glioblastoma cells seemed not to be quite in a dose-dependent manner.

Finally, to assess the clinical relevance of the above observations, we investigated the expression of MIF and LC3B in glioma specimens. Glioma tissues with strong MIF immunohistochemical signals had high expression of LC3B and vice versa (Fig. 2C). The linear regression analysis indicated the positive correlation of MIF with LC3B. These results demonstrated a positive relationship between MIF and autophagy in glioma.

Although several studies have identified ROCK1 as a regulator of autophagy, whether MIF enhances autophagy via ROCK1 is unknown. Therefore, we used Y27632, an inhibitor of ROCK1 activity, and siROCK1 to confirm the role of ROCK1 in MIF-induced autophagy. Western blot analyses were performed to detect autophagy. The increased expression of LC3B-II induced by MIF was significantly reversed by Y27632 and ROCK1 knockdown (Fig. 3B). In addition, U87 cells stably expressing GFP-LC3 were treated with MIF and then fixed to assess the actin filaments by rhodamine-phalloidin staining. The F-actin stress fibre formation was localized in GFP-LC3 puncta-positive cells, suggesting that glioblastoma cells with high ROCK1 activity had increased autophagy (Fig. 3A).

Furthermore, we explored whether ROCK1 activity contributes to increased migration and colony formation induced by MIF. The migration of U87, U251 and T98 cells was measured by Transwell assays. rhMIF promoted the migration of all three cell lines, and as expected, Y27632 suppressed the MIF-induced migration (Fig. 3D). Y27632 partly blocked the MIF-induced colony formation (Fig. 3C). These data demonstrated that the activity of ROCK1 played a crucial role in MIF-induced tumourigenesis.

A previous report identified DCs in brain tissues (33). To determine whether DCs infiltrated glioma tissues, we used CD1a, a marker of DCs, to detect DCs in glioma specimens by immunohistochemistry (34). Few CD1a-positive cells were found in glioma specimens, while in normal brain tissues, there were no CD1a-positive cells (Fig. 4A). This suggested that DCs could infiltrate gliomas through blood-brain barrier (BBB) disruption or an unknown mechanism. Moreover, in serial sections of glioma tissue, CD1a-positive cells were localized in regions with high levels of MIF (Fig. 4B). These data indicated that MIF may suppress the migration of DCs to impede initiation of immune responses.

To study the effects of MIF on DCs, we generated iDCs and mDCs from PBMCs of healthy individuals (Fig. 5A). The cellular morphology of the iDCs and mDCs was observed (Fig. 5B). The effect of MIF on migration of DCs was assessed using Transwell assays. The migration of iDCs and mDCs towards CCL21 was significantly decreased in MIF-treated DCs compared to that of the controls (Fig. 5C). Furthermore, to verify whether MIF regulates actin rearrangement in DCs, rhodamine-phalloidin staining was performed. There were strong phalloidin-stained puncta in MIF-treated DCs, suggesting a deficiency of the intact F-actin ring (Fig. 5D).

To further characterize the immunological impact of MIF on DCs, we studied its effects on DC maturation and the expression of co-stimulatory molecules. iDCs were cultured with MIF or supernatants from siMIF-transfected or negative control U87 cells for 3 days in the presence of LPS, which promoted maturation (Fig. 6A). Subsequently, the expression of the maturation markers CD83, CD80 and CD86 was quantified by qPCR. In the presence of MIF, the two co-stimulatory molecule markers, but not CD83, were significantly decreased compared to that of the no MIF treatment, indicating that MIF had an inhibitory role in the maturation of iDCs (Fig. 6C). Similarly, cultures of iDCs with supernatant from siMIF-transfected U87 cells showed slightly increased expression of CD80 (Fig. 6C), while increased expression of CD83 and CD86 was not detected (data not shown). Moreover, we cultured the mDCs in the presence or absence of MIF for 24 h, and the expression of the markers was also downregulated (Fig. 6B). These data demonstrated that MIF plays an inhibitory role in DC maturation and function.