A novel role of HLA class I in the pathology of medulloblastoma

Medulloblastoma tissue microarrays were generated as previously described [37]. Formalin-fixed, paraffin embedded tissue blocks were obtained from both the Armed Forces Institute of Pathology and Children's National Medical Center. More than 95% of tissues were obtained for diagnostic purposes, hence patients underwent no prior therapy. A neuropathologist (MS) marked two representative tumor core regions within each tissue block. The cores measuring 0.6 mm in diameter and 3–4 mm in height were isolated. A de-identified microarray was generated using 4–8 μm sections cut from the core blocks including: 139 medulloblastomas (47 classic, 40 anaplastic, 25 desmoplastic and 27 with unclassified histology), 21 primitive neuroectodermal tumors, 10 small cell carcinomas, 5 atypical teratoid rhabdoid tumors, 3 oat cell lung carcinomas, one each of ependymoblastoma and lymphoma, and 20 tissues of brain metastases from tumors of various origin. IRB approval was obtained for the construction and analysis of these tissue microarrays and all other tumor specimens investigated.

DAOY, D283 and D556 medulloblastoma cell lines were maintained at 37°C in a humidified atmosphere containing 5% CO₂ in RPMI media supplemented with 10% Fetal bovine serum, 2 mM L-glutamine, 1 mM 2-mercaptoethanol, 100 U/ml penicillin and 100 μg/ml streptomycin.

Polyclonal rabbit anti-human β2m and monoclonal mouse anti-human CD45 (clones 2B11 + PD7/26) were purchased from DakoCytomation, Carpinteria, CA. The hybridoma secreting HC-10 antibody, specific for HLA heavy chain epitope revealed in the absence of β2m and peptide [38], was kindly provided by Dr. Pan Zheng (University of Michigan School of Medicine, Ann Arbor, MI). Anti- TAP1 (NOB-1) and TAP2 (NOB-2) antibodies [39] were provided by Dr. S. Ferrone (Hillman Cancer Center, University of Pittsburgh Cancer Institute, Pittsburgh, PA). The W6/32 monoclonal antibody is specific for HLA A, B and C and specifically to the combined epitope contributed by the α2 and α3 domains of the heavy chain and β2m [40‐42]. mAb rabbit anti-human GAPDH, rabbit anti-human phospho-p44/p42 MAP kinase (Thr202/Tyr204), mAb rabbit anti-human p44/p42 MAP kinase, rabbit anti-human phospho-AKT (Ser473), rabbit anti-human AKT, mAb mouse anti-human phospho-EGF receptor (Tyr1068), rabbit anti-human EGF receptor, rabbit anti-human phospho-PDGFRβ (Tyr751) were purchased from Cell Signaling (Beverly, MA), anti-c-myc from Abcam (Cambridge, MA), rabbit anti-human PDGFR- β (Santa Cruz) and anti-rabbit IgG HRP-linked antibody (Cell Signaling). HC-10 and W6/32 antibodies were partially purified from hybridoma supernatants using the centriplus YM-100 (Millipore) regenerated cellulose filters that have a molecular weight cutoff of 100,000 Da. Antibody concentration after purification was 2 mg/ml.

Slides containing tissue arrays were deparaffinized and rehydrated. The primary Abs HC-10, TAP1 and TAP2 required an additional heat-induced epitope retrieval step (Target Retrieval Solution pH 9 or Target Retrieval Solution, pH6.1, DakoCytomation, Carpinteria, CA). Slides were blocked with biotin (Biotin Blocking System, DakoCytomation) and processed using the 3-step streptavidin-biotin-immunoperoxidase staining system (LSABTM2+ System-HRP, DakoCytomation, Carpinteria, CA). Diaminobenzidine (DAB) was the chromogenic substrate. Mayers hematoxylin was used as a counterstain and followed with an ammonia wash. Slides were mounted using an aqueous mounting medium (Faramount, DakoCytomation). Controls consisted of parallel sections without primary antibody. Stainings for heavy chain, β2m, TAP1, TAP2 and CD45 were assigned scores of 0–3 [43] based on the percentage of stained cells (0- no staining; 1- less than 10% cells stained; 2- 10–50% cells stained; 3- >50% cells positive). The staining for c-myc was not restricted to particular cells, but was rather present or absent throughout the tumor. Hence, c-myc scores were assigned 0 (absent staining), 1 (weak diffuse staining) or 2 (strong diffuse staining). Each score was an average of the two samples graded for each individual tumor in the array. Each slide was evaluated by three independent graders including two neuropathologists. The images were acquired using the following equipment: Microscope – Carl Zeiss Axioskop; Lenses- Achroplan 40× and Achroplan 20×; Camera – Carl Zeiss Axio Cam HRC; Acquisition software- Axio Vision Rel. 4.3.

Cells were incubated with either 5% FBS in PBS (controls) or W6/32 or HC-10 supernatant for 1 hour. After 3 rinses, cells were incubated with PE labeled donkey anti-mouse IgG (H+L) (eBiosciences, San Diego, CA) for 30 minutes at 4°C. Cells were rinsed 3 times and fixed in cytofix (BD Sciences, San Diego, CA).

Total cellular RNA, obtained from 10 frozen medulloblastoma specimens, was subjected to the SuperScript First-strand synthesis system for RT-PCR (Invitrogen, Carlsbad, CA) to generate the cDNA for the PCR reactions. RT-PCR were performed using 35 cycles of 30 seconds at 94°C, 1 minute at the annealing temperature, and 1 minute at 72°C (with the exception for HLA-A and HLA-B where the last cycle was for 1 minute 30 seconds at 72°C). The annealing temperatures were 57°C for TAP2 and β2m, 58°C for β-actin and CD45, 59°C for TAP1 and 60°C for HLA-A and HLA-B. The primer sequences were: 5'-GAGACATCTTGGAACTGGAC-3' and 5'-CTCTGAGTGAGAATCTGAGC-3' (forward and reverse, TAP1), 5'-GTACAACACCCGCCATCAG-3' and 5'-GGACGTAGGGTAAACGTCAGC-3' (TAP2), 5'-CTCGCGCTACTCTCTCTT-3' and 5'-AAGACCAGTCCTTGCTGA-3' (β2m), 5'-TATAGTCGACCACCCGGACTCAGAATCTCCT-3' and 5'-ATATGGATCCATCTCAGTCCCTCACAAGA-3' (HLA-A), 5'-TATAGTCGACCACCCGGACTCAGAGTCTCCT-3' and 5'-ATATGGATCCATCTCAGTCCCTCACAAGA-3' (HLA-B), 5'-ACCTGTACGCCAACACAGTG-3' and 5'-GCCATGCCAATCTCATCTT-3' (β-Actin), 5'-CTGAAGGAGACCATTGGTGA and 5'-GGTACTGGTACACAGTTCGA-3' (CD45). The reactions were loaded onto a 1% agarose gel and products were visualized by ethidium bromide staining.

DAOY and D283 cells were cultured in serum-free media for 24 hrs at 37°C. Cells were washed twice and treated with serum-free RPMI with or without human β2m (Lee Biosolutions, Inc, St. Louis, Missouri) or monoclonal antibodies for indicated times, at 37°C. Signaling was halted by the addition of ice cold PBS and cells were lysed with 1× cell lysis buffer (Cell Signaling, Danvers, Massachussetts) supplemented with PhosSTOP phosphatase inhibitor cocktail and complete mini protease inhibitor cocktail (Roche, Indianapolis, Indiana). Lysates were centrifuged at 14,000 rpm for 15 minutes to remove cell debris, boiled for 5 minutes in loading buffer and separated by a 4–12% Bis-Tris Gel (Invitrogen, Carlsbad, California). Membranes were blocked in 5% milk in TBST for 1 hour at room temperature. Primary and secondary antibodies were diluted in 3% BSA in TBST and incubated with the membrane at 4°C overnight and at room temperature for 1 hour, respectively. Signal was detected using the enhanced chemiluminescence system (Pierce, Rockford, IL). Densitometry quantification of the bands was performed using Quantity One software (Bio-Rad, Hercules, CA), according to manufacturer's instructions. The Band 1/Band 2 ratio was obtained by dividing differences between intensity units observed in the square areas containing specific bands and an identical blank area drawn in the immediate vicinity of the band, according to the following formula: Band 1/Band 2 ratio= (Band 1-blank)/(Band 2 -blank).

Wound scratch assay was used to evaluate tumor cell migration [44, 45]. DAOY cells were plated in a 60 mm dish at 40% confluency and grown overnight. After 24 hours the cells were scraped down the middle of the plate with a 200 μl pipette tip to induce the resemblance of a wound and washed twice with serum-free media. Cells were cultured with serum free media with or without 2 μg/ml β2m for 24 hours. Images of the wound were taken at 0 and 24 hours. The area of the wound was calculated using the Axiovision system by tracing the cleared space. Area was measured in μM2. The ratio of 24 hours to 0 hours was calculated for each sample.

Expression of MHC class I in medulloblastoma has been reported in two studies with conflicting results [46, 47]. We therefore examined the MHC class I expression in our collection of medulloblastoma specimens. The expression of the essential components of the MHC class I antigen-processing machinery: HLA heavy chains (A and B), β2m, TAP1 and TAP2 was first evaluated using RNA from ten frozen medulloblastoma specimens. β2m was expressed in all samples, while heavy chains and TAP subunits were detected in seven samples (Fig. 1A). An essential element of this analysis was the assignment of MHC class I expression to the tumor cells and not to the leukocytes infiltrating the medulloblastoma tissues. CD45 detection was used to evaluate contamination by leukocyte-derived RNA. CD45 was undetectable in four samples. Three of the CD45-negative samples (M66, M63 and M68) expressed β2m and not the other three components, indicating that the majority of medulloblastomas fail to express MHC class I. The fourth sample (M69) expressed all four components.

Because of relatively frequent leukocyte infiltration, the ability to discriminate RNA from tumor versus stromal host cells was low in many medulloblastoma samples. IHC analysis of medulloblastoma tissues is advantageous in this respect, because areas of leukocyte infiltration and MHC class I expression can be directly visualized and compared to tumor cell expression in each individual sample. We therefore analyzed HLA heavy chains, β2m, TAP1, TAP2 and CD45 intracellular and/or cell surface expression using medulloblastoma tissue microarrays. Representative staining patterns for HLA heavy chains are shown in Fig. 1B. Of the 106 evaluable specimens 87% showed absent or faint heavy chain positivity (56% scored 0 and 31% scored 1), while scores 2 and 3 were observed in 5% and 8% of tissues, respectively (Table 1). To gain a better molecular understanding of MHC class I expression, medulloblastoma arrays were stained with antibodies to TAP subunits and β2m (Table 1). In agreement with the mRNA analysis, the majority of medulloblastoma tissues expressed high levels of β2m protein. Surprisingly, similarly high levels of TAP1 were seen, while staining of TAP2 was intermediate (Table 1). The 14 samples with scores 2 or 3 for staining with HC-10 antibody also displayed high levels of TAP1, TAP2 and β2m (except for TAP2 in two samples). Therefore, we conclude that there is a small subset of medulloblastomas that may express all components required for functional MHC class I molecules, whereas at least one essential component (heavy chain or TAP2) is missing in most tumor samples.

To determine whether expression of the components of MHC class I processing machinery could be ascribed to infiltrating leukocytes, we stained the medulloblastoma microarray with anti-CD45 antibody (Fig. 1C). Of the 103 evaluable medulloblastoma specimens, 58.3% were negative for infiltration with 39.8% receiving a score of 1 and 0.95% a score of 2 and 3 each (Table 1). The positive staining samples showed leukocytes present in the tumor, but few that had migrated away from the blood vessels into the tissue. The low levels of infiltration appear to be indicative of the expected poor immune response associated with medulloblastomas. Of the 14 MHC class I-high samples, 6 were negative for leukocyte infiltration and 6 received a score of 1. Scores of 2 and 3 were noted in one sample each. Therefore, contamination by infiltrating peripheral white blood cells cannot account for positive heavy chain, TAP1, TAP2 and β2m staining in most of these cases and MHC class I expression appears to be genuinely derived from medulloblastoma cells.

Anaplastic histopathology [48, 49] and c-myc expression [49‐51] are negative prognostic markers for medulloblastoma. To test whether MHC class I expression had any prognostic value for medulloblastomas, we evaluated the distribution of histological subtypes and expression of c-myc in specimens with scores 2 or higher for heavy chain versus those that scored less than 2. The reason for the cut-off at score 2 is that score 1 was given if only up to 10% cells in the specimen stained positive. In contrast to the diffuse character of the anaplasia and c-myc expression, we considered that the impact, if any, of so few positive cells on the overall histology of the tumor could not have been significant.

Of the 106 samples evaluable for HC-10 staining, 31 were anaplastic, 26 classic, 23 desmoplastic and 26 histologically unclassified (without evidence of diffuse anaplasia). Of the fourteen MHC class I-high medulloblastomas 8 were anaplastic, 3 classic, 1 desmoplastic and 2 unclassified. Similar findings were observed when c-myc expression was considered. Of the 88 samples that were evaluable for both HC-10 and c-myc, 59 scored 0, 23 scored 1 and 6 scored 2 for c-myc expression. In contrast, in the MHC class I-high population there were 5 c-myc negative samples, and 6 and 2 with scores 1 or 2 respectively. The frequency of anaplastic versus any other histological subtype, or c-myc negative versus c-myc positive specimens in the MHC class I-high versus MHC class I-low or -negative medulloblastoma specimens is significantly different (Fig. 2), with p values being 0.0251 for histopathology (n = 105) and 0.0257 for c-myc expression (n = 88).

The levels of open and closed conformer were respectively analyzed using the HC-10 antibody that detects denatured heavy chains [38], and W6/32 that binds to the combined epitope contributed by the α2 and α3 domains of the heavy chain and β2m, dependent on the presence of peptides in the peptide-binding groove [40‐42]. The levels of open conformers were clearly detectable in DAOY and to a lesser degree in D556 medulloblastoma cell lines that display relatively high levels of MHC class I antigens (Fig. 3A). In contrast, open conformers were virtually non-existent in MHC class I low D283 cell line. To determine whether β2m can bind to open conformers we evaluated the effect of HC-10 antibody on β2m binding to DAOY cells (Fig. 3B). We focused on DAOY cells because they displayed the highest levels of open conformers. An increase in the levels of β2m observed following addition of exogenous β2m was largely prevented by HC-10, but not W6/32 antibody, suggesting that most of the exogenous β2m binds to open forms of HLA class I. However, residual β2m bound even in the presence of HC-10, likely reflecting its specificity. Molecular HLA class I typing of DAOY cells revealed a genotype consisting of A*0101, A*0201, B*0703, B*5701, Cw*0602 and Cw*0702 (data not shown), of which only A*0101 and A*0201 are not recognized by HC-10 [38]. Finally, we performed flow cytometry to determine whether binding of exogenous β2m modulates the ratio of open to closed cell surface conformers. As predicted, addition of exogenous β2m to DAOY cells reduced the relative levels of open, and increased the levels of closed conformers (Fig. 3B), suggesting that binding of exogenous β2m can alter the balance between the open and closed conformers.

We next examined whether altering balance between the open and closed MHC class I conformers may contribute to signal modulation. We therefore evaluated the impact of exogenous β2m on phosphorylation of ERK1/2 and AKT that are downstream arms of many receptor pathways. Increased levels of phospho-ERK1/2 were found 15–30 minutes following addition of exogenous β2m to DAOY cells (Fig. 4A), while the levels of phospho-AKT remained largely unaltered. The optimal increase in ERK1/2 phosphorylation was achieved with 2 μg/ml β2m (Fig. 4B), which is well within the range of physiological concentration in human serum [26, 27, 29]. Further, the effect was partially inhibited by anti-β2m antibodies despite their 10-fold molar deficit (Fig. 4C). Because there is a constitutive low level of ERK1/2 phosphorylation in the absence of any treatment, we refer to the effect of β2m as modulation, rather than induction of ERK1/2 phosphorylation. Consistent with the available levels of open conformers, increased levels of pERK1/2 were seen in D556 cells, albeit with a slightly delayed kinetics (Fig. 4D), but not in MHC class I deficient D283 cells (Fig. 4E). We next tested whether HC-10 antibody could prevent β2m -induced ERK1/2 phosphorylation. We were surprised to see that adding HC-10 itself modified ERK1/2 activation in a similar manner as β2m (Fig. 5). This was not the case with W6/32 antibody that recognizes closed HLA class I conformers. Therefore, two independent open conformer ligands enhance ERK1/2 phosphorylation in DAOY cells.

Given the previously established association of higher expression of β2m with metastatic disease in medulloblastoma [35], we wondered whether engagement of open conformers might affect the migration characteristics of medulloblastoma cells in a wound scratch assay. DAOY cells were grown into near confluence when a wound was created in the center of the monolayer, and the ability of neighboring cells to migrate into the denuded area within 24 hours was determined. While control DAOY cells showed marginal ability to invade, the presence of β2m clearly induced significant migratory activity (Fig. 6A). Evaluation of the surface area of the scratch at 0 and 24 hours time points indicated that untreated cells recolonized only 6%, while the β2m- and HC-10-treated recovered 24% and 22% of the wound area, respectively (Fig. 6B). The migration was inhibited in the presence of 100 μm PD98059, pharmacologic inhibitor of upstream member of the ERK1/2 activation pathway [52]. Thus, the ERK1/2 activation enhanced by the engagement of open HLA class I conformers may contribute to higher migratory capacity of medulloblastoma cells.