The costimulatory molecule CD70 is regulated by distinct molecular mechanisms and is associated with overall survival in diffuse large B-cell lymphoma.

In diffuse large B-cell lymphomas (DLBCL), a recurrent deletion of the 19p13 region has recently been described. CD70 and TNFSF9 genes are suspected tumor suppressor genes, but previous studies suggest an oncogenic role for CD70. Therefore, we studied the consequences of variation in CD70 copy number and epigenetic modifications on CD70 expression. Copy-number variation was investigated in 144 de novo DLBCL tissues by comparative genomic hybridization array and quantitative multiplex PCR. Gene expression was assessed by quantitative RT-PCR, and CD70 promoter methylation was determined by pyrosequencing. The 19p13.3.2 region was deleted in 21 (14.6%) cases, which allowed the minimal commonly deleted region of 57 Kb that exclusively includes the CD70 gene to be defined. Homozygous deletions were observed in four (2.7%) cases, and acquired single-nucleotide variations of CD70 were detected in nine (6.3%) cases. CD70 was highly expressed in both germinal centre B-cell-like (GCB) and activated B-cell-like (ABC) DLBCL compared to normal tissue, with distinct molecular mechanisms of mRNA expression regulation. A gene dosage effect was observed in the GCB subtype, whereas promoter methylation was the predominant mechanism of down regulation in the ABC subtype. However, high CD70 expression levels correlated to shorter overall survival in both the GCB (P = 0.0021) and the ABC (P =0.0158) subtypes. In conclusion, CD70 is targeted by recurrent deletions, somatic mutations and promoter hypermethylation, but its high level of expression is related to an unfavorable outcome, indicating that this molecule may constitute a potential therapeutic target in selected DLBCL.

Mapping of MYC breakpoints in 8q24 rearrangements involving non-immunoglobulin partners in B-cell lymphomas.

Chromosomal translocations joining the immunoglobulin (IG) and MYC genes have been extensively reported in Burkitt's and non-Burkitt's lymphomas but data concerning MYC rearrangements with non-IG partners are scarce. In this study, 8q24 breakpoints from 17 B-cell lymphomas involving non-IG loci were mapped by fluorescence in situ hybridization (FISH). In seven cases the breakpoint was inside a small region encompassing MYC: in one t(7;8)(p12;q24) and two t(3;8)(q27;q24), it was telomeric to MYC whereas in four cases, one t(2;8)(p15;q24) and three t(8;9)(q24;p13) it was located in a 85 kb region encompassing MYC. In these seven cases, partner regions identified by FISH contained genes known to be involved in lymphomagenesis, namely BCL6, BCL11A, PAX5 and IKAROS. Breakpoints were cloned in two t(8;9)(q24;p13), 2.5 and 7 kb downstream from MYC and several hundred kb 5' to PAX5 on chromosome 9, joining MYC to ZCCHC7 and to ZBTB5 exon 2, two genes encoding zinc-finger proteins. In these seven cases, MYC expression measured by quantitative reverse transcription-polymerase chain reaction (RT-PCR) was significantly higher when compared to that of patients without 8q24 rearrangement (P=0.006). These results suggest that these rearrangements are the consequence of a non-random process targeting MYC together with non-IG genes involved in lymphocyte differentiation and lymphoma progression.

Hyaluronan digestion and synthesis in an experimental model of metastatic tumour.

To approach the question of hyaluronan catabolism in tumours, we have selected the cancer cell line H460M, a highly metastatic cell line in the nude mouse. H460M cells release hyaluronidase in culture media at a high rate of 57 pU/cell/h, without producing hyaluronan. Hyaluronidase was measured in the H460M cell culture medium at the optimum pH 3.8, and was not found above pH 4.5, with the enzyme-linked sorbent assay technique and zymography. Tritiated hyaluronan was digested at pH 3.8 by cells or cell membranes as shown by gel permeation chromatography, but no activity was recorded at pH 7 with this technique. Hyaluronan was digested in culture medium by tumour slices, prepared from tumours developed in nude mice grafted with H460M cells, showing that hyaluronan could be digested in complex tissue at physiological pH. Culture of tumour slices with tritiated acetate resulted in the accumulation within 2 days of radioactive macromolecules in the culture medium. The radioactive macromolecular material was mostly digested by Streptomyces hyaluronidase, showing that hyaluronan was its main component and that hyaluronan synthesis occurred together with its digestion. These results demonstrate that the membrane-associated hyaluronidase of H460M cells can act in vivo, and that hyaluronan, which is synthesised by the tumour stroma, can be made soluble and reduced to a smaller size by tumour cells before being internalised and further digested.

Hyaluronectin secretion by monocytes: downregulation by IL-4 and IL-13, upregulation by IL-10.

Hyaluronectin (HN) is a component of the extracellular matrix of connective tissue and is particularly associated with tumour inflammatory and connective stroma reaction, where it co-localizes with hyaluronic acid (HA). The HN/HA ratio has been suggested to be involved in tumour aggressivity and in the atherosclerosis process. IL-10 has also been described in atherosclerotic lesions and in cancer. HN production was therefore investigated in vitro in peripheral blood monocyte cell (PBMC) cultures, with and without bacterial lipolysaccharide (LPS) or interleukins (ILs) in the medium. HN was characterized in monocytic cell cytoplasm and in culture supernatants. Anti-IL-10 antibody suppressed the LPS-stimulating effect on HN production. HN synthesis rate was greatly increased in IL-10-activated cultures while IL-4 and IL-13, two other anti-inflammatory ILs, decreased HN release. In the presence of IL-10, the IL-4 or Il-13 inhibitory effect on HN synthesis was reversed. The results support the view that intratumoral release of IL-10 by monocytes may induce local production of HN. In conjunction with the known ability of HN to bind to HA, which is a cell migration and tumour invasion facilitating factor, and to inhibit HA-induced angiogenesis, our findings suggest that HN may modulate the effect of HA on atherosclerosis, angiogenesis and cancer development.

Activation and inhibition of human cancer cell hyaluronidase by proteins.

Results regarding hyaluronidase activity in tumor extracts or cell lines are subject to variations according to the method used for the assay and, sometimes, within an assay. Hyaluronidase was assayed at pH 3.8 in the culture medium of the human cancer-derived cell lines SA87 and H460M by several techniques: HPLC, Reissig technique, ELSA technique, and zymographic analysis. The optimal pH was between 3.3 and 4 in solutions at constant 150 mM sodium concentration. The enzyme was reversibly inhibited by sodium concentration over 200 mM. The activity of purified hyaluronidase increased in the presence of low concentrations of the specific HA-binding glycoprotein hyaluronectin, or of bovine serum albumin or immunoglobulins, or of human albumin, transferrin, or hemoglobin, showing that proteins cooperate in enzyme activity. The ELSA technique showed that optimal pH was slightly lower in the presence of HN than that with BSA. The optimal BSA concentration was determined with the ELSA technique at 0.1 g/liter, and excess of either protein inhibited hyaluronidase. When measured with the Reissig technique, the activity of purified enzyme in the presence of 0.1 g/liter BSA was up to fourfold that without BSA. The cooperative effect of BSA was visualized by zymography. We conclude that the total protein content of hyaluronidase solutions must be considered to correctly interpret quantitation of the enzyme in sera or tissue extracts because protein concentrations above 200 microg/liter lead to underestimation of the enzyme.

The origin of hyaluronectin in human tumors.

The origin of tumor stroma hyaluronectin (HN), a glycoprotein that binds to hyaluronan (HA), has long remained unknown. Histological observations of human tumors suggest that tumor HN could originate from stroma fibroblasts, and in some cases from inflammatory cells. The fibroblast origin was confirmed by the discovery of HN-like antigen along with hyaluronan in culture medium of tumor-derived fibroblasts. An HA-binding protein was characterized in the culture medium of peripheral blood mononuclear cells (PBMC) in both normal subjects and tumor-bearing patients and was found to be human HN. Cultivated monocytes did not produce HA. HN was not related to the HA-binding site CD44. Sequencing of brain HN-derived peptides demonstrated that each determined peptide sequence was similar to a sequence of the proteoglycan PG-M/versican, suggesting that HN is the HA-binding moiety of the proteoglycan. One probe was synthesized from human PBMC by polymerase chain reaction with primers derived from HN sequences also found in versican. Northern blots were positive only with HN-producing cells. The main RNAs were in the 6-8 kb range, and there was a limited proportion of smaller RNA, which was compatible with the size expected from the HN molecular mass. Southern blotting of monocytes and tumor cells demonstrated that the gene was limited to a unique band. We conclude that HN, an extracellular component of brain, connective embryonic, inflammatory and tumoral tissues, is a PG-M/versican-derived molecule. Our results suggest that tumor HN, which originates from fibroblasts and monocytes of tumor stroma, is a molecular component of the host-tumor relationship and could play a role in the regulation of HA activity in oncogenesis.

[Production of hyaluronidase by cultured human tumor cells]. / Production d'hyaluronidase par les cellules cancéreuses humaines en culture.

The presence of hyaluronidase was detected at pH 3.8 in eight out of twelve human cancer cell line culture media. Eight cell lines derived from primary tumours and four from metastases. In three culture media the enzymatic activity was lower than 0.035 pU/cell/h. In five others (in a hepatoma cell line and in four metastasis-derived cell lines) the activity was higher than 0.057 pU/cell/h. A tumour-derived fibroblast culture was negative. The optimal activity was observed at a pH comprised between 3.6 and 4. Salt inhibition of hyaluronidase was reversible. The enzyme was denaturated by a 10-min heating at 70 degrees C. The enzyme was not strictly specific for hyaluronan hydrolysis but also digested chondroitin sulfates. PH20, a spermatozoid protein that has homologies with the bee venom hyaluronidase, was not expressed by cell lines tested.

Hyaluronan and hyaluronectin production in injured rat thoracic aorta.

The aim of our study was to investigate the production of hyaluronan (HA) by the intima-media during the sclerotic response to aortic injury with a catheter balloon in the rat. In addition we analyzed, for the first time in this model, the production of a glycoprotein (hyaluronectin, HN) which binds specifically to HA. HA and HN were analyzed in control (D0), 14 (D14) and 28 (D28) days after injury using biochemical and immunohistochemical techniques. Intima-media DNA content and wet weight increased significantly on D14 and declined on D28 (but remained significantly increased in comparison to controls). HA content (median in D0 = 448 ng) increased significantly on D14 (2P < 0.04) and on D28 (2P < 0.02). HN content (median in D0 = 920 ng) increased significantly on D14 (2P < 0.05) but decreased on D28 to return to the control level. On D0 the amount of HN was about 3 times higher than that of HA (median ratio HA/HN = 0.34). The ratio remained unchanged on D14 but significantly increased on D28 (2P < 0.02). HPLC and Western blotting showed no difference between HN extracted from normal aorta and HN extracted from injured aorta at D14. Different isoforms of HN were present in both cases, ranging from 400 to 45 kDa. The HA increase on D14 and D28 was not related to a change in hyaluronidase activity of aortic tissue. Immunohistochemical analysis showed at D0 a small amount of HA around arterial smooth muscle cells (ASMC) in media, at D14 more HA was localized around and between ASMC in media and neointima but at D28 it was localized mainly near the vessel lumen. HN formed all the time (D0, D14 and D28) a continuous layer localized near the vessel lumen. In vitro studies showed that production of HA and HN was stimulated when ASMC proliferate and HA at high concentrations (1-100 micrograms/ml) reduced, in a dose dependent manner, ASMC growth. In conclusion our results show that both neointima formation in vivo and ASMC proliferation in vitro correlated with increased HA and HN production. This suggests that HA and HN are probably involved in the formation of neointima. On the other hand, the finding that HA continued to increase in the aorta when neointima decreased and that high concentrations of HA reduce ASMC proliferation in culture suggest that HA might be involved in the regression of neointima.

Enzyme-linked hyaluronectin: a unique reagent for hyaluronan assay and tissue location and for hyaluronidase activity detection.

Several techniques for assaying and localizing hyaluronan (HA), all based on the affinity to hyaluronan of proteins isolated from cartilage, chondrosarcoma, or brain, have been proposed. We show here that a unique reagent, alkaline phosphatase-linked hyaluronectin, can be used to assay hyaluronan in biological fluids or tissue extracts (enzyme-linked sorbent assay method) and to characterize it in cells or tissue sections in two steps: reagent incubation and staining. Results of assays in biological fluids or tissue extracts showed a good correlation with results of the previously described technique using antibodies to detect hyaluronectin bound to a plastic microtest plate (B. Delpech et al., 1985, Anal. Biochem. 149, 555-565) for both low concentrations (< 1 mg/liter, r = 0.973, P < 0.001) and high concentrations (> 1 mg/liter, r = 0.953, P < 0.001). The interassay variations were 8.5% when the assay was performed at 4 degrees C and 18.5% at 37 degrees C. The intraassay variations under those conditions were, respectively, 14.4 and 6.5%. Tissue HA could be detected easily with the reagent, as shown in fetal tissues and in tumors. Specificity of the reaction was controlled either by blocking the reagent with an excess of hyaluronan (which was not possible with other glycosaminoglycans) or by destroying tissue hyaluronan with streptomyces hyaluronidase. Alkaline phosphatase-linked hyaluronectin was also used to assay hyaluronidase activity in several biological fluids. One-hour incubation of hyaluronidase preparations on HA-coated plates made it possible to detect as low as 1 mU bovine testis hyaluronidase and 0.1 mTRU streptomyces hyaluronidase. Four-hour incubation made it possible to detect activity in a 1/12,500 dilution of human serum.

Expression and effects of hyaluronan and of the hyaluronan-binding protein hyaluronectin in newborn rat brain glial cell cultures.

Hyaluronan (HA) is a polymerized nonsulfated extracellular matrix glycosaminoglycan that may be involved in brain development. We have tested the expression of HA and the HA-binding protein hyaluronectin (HN) in glial cell cultures from newborn rat brain. HA was secreted into the culture medium by type 1 astrocytes in the first stages of the primary cultures. The secretion was high during cell proliferation, reached a maximum when they were confluent, and then decreased. HA was not secreted at a detectable level by total O-2A lineage cell-enriched cultures. HA labeled small O-2A progenitor cells (GFA-, A2B5+, HA+), small O-2A progenitorlike (GFA-, A2B5-, HA+) cells, and type 2 astrocytes (GFA+, A2B5+, HA+), but not mature oligodendrocytes (Galc+, HA-). In contrast to HA, hyaluronectin labeled oligodendrocyte membranes (i.e., more mature cells) from day 8. A2B5+ GFA- cells were found to be either HA+ or HN+ at days 7-9, suggesting intermediary stages. The addition of HA to primary cultures and to O-2A progenitor-enriched cultures decreased significantly the increase in the number of O-2A progenitors, of mature (Galc+) oligodendrocytes proportionally to the decrease of the O-2A progenitor number, and of BrdU+ cells, suggesting that HA acts (directly or indirectly) on O-2A cell proliferation. This effect, which was seen for concentrations as low as 0.1 micrograms/ml, was HA specific and was not observed with other glycosaminoglycans. When primary cultures were performed in the presence of hyaluronidase-digested or HA-depleted (by passage on a HN column) fetal calf serum, the total number of O-2A lineage cells was dramatically increased (100%, p < 10(-4)) in comparison with control cultures in standard fetal calf serum. Platelet-derived growth factor increased the total number of O-2A lineage cells and of (Galc+) oligodendrocytes. This effect was opposed by HA dose dependently. The effect of HA was significantly inhibited by HN (30%, p < 10(-4)). HN had, however, no effect when it was added to culture in the presence of hyaluronidase in fetal calf serum, suggesting its effect was only due to its binding to HA. During cell maturation, HA disappears as HN appears. This and the fact that HA and PDGF have opposite effects suggest an effect of these factors, or of their balance, on myelination.

Localization and solubilization of hyaluronan and of the hyaluronan-binding protein hyaluronectin in human normal and arteriosclerotic arterial walls.

Hyaluronectin (HN), a hyaluronan (hyaluronic acid, HA)-binding glycoprotein isolated from human brain, was studied in normal and atherosclerotic human arteries. It can be detected and assayed in tissue samples by immunohistochemistry. In addition, its high and specific affinity for HA makes it possible to develop specific histological localization of HA using HN as a probe. We tested the presence of HN and HA in human carotid artery samples from adults and newborns. In atheroma-free arterial samples HN was found in the intima, between smooth muscle cells and in the adventitial extracellular matrix. In atherosclerotic lesions, HN was strongly expressed in the diffuse thickened intima and surrounding extracellular microcrystalline calcium deposits, and very little in the lipid core. HA was found in the same locations. The similar localizations of HN and HA shown by immunohistology and demonstration of HN-HA complexes by high pressure liquid chromatography (HPLC) suggest that they are associated in vivo.

Caffeine decreases glial cell number and increases hyaluronan secretion in newborn rat brain cultures.

Newborn rat brain astrocytes (type 1 astrocytes, O-2A progenitor cells, and O-2A progenitor-derived cells, i.e. oligodendrocytes and type 2 astrocytes) were cultivated to investigate the effect of addition of caffeine to the culture medium on glial cell development and secretion of hyaluronan (hyaluronic acid, HA). HA is a glycosaminoglycan, secreted by type 1 astrocytes especially, which is a major component of the extracellular matrix of immature brain involved in morphogenesis and differentiation. Caffeine was added to the culture medium of primary glial cell cultures at concentrations of 102 microM (20 mg/L) or 255 microM (50 mg/L), considered therapeutic and toxic levels, respectively, in human newborns. HA was measured in the culture medium by immunoenzyme assay using sheep brain hyaluronectin, a glycoprotein that exhibits a strong affinity for HA, as probe. In primary glial cell cultures, 102 microM (20 mg/L) caffeine had no visible effect on cell number or on HA secretion. At 255 microM (50 mg/L), there was a significant reduction of cell number (i.e. type 1 astrocytes, O-2A progenitor cells, and progenitor-derived cells) and a significant increase of HA secretion per cell. These results suggest that caffeine at a high concentration in brain could have a prejudicial effect on the number of proliferating glial cells (astrocytes and oligodendrocytes) and on the composition of the extracellular matrix, which could affect myelination onset.

Hyaluronan and hyaluronectin in the extracellular matrix of human brain tumour stroma.

Hyaluronan (HA) and the hyaluronan-binding glycoprotein hyaluronectin (HN) were measured in 23 gliomas and 8 meningiomas and their location was revisited in 35 tumours. A clear-cut difference was found in the HN/HA ratio values of glioblastomas (below 0.5) and that of astrocytomas (above 0.5 P < 0.001). Besides their location in the intercellular part of gliomas, HA and HN displayed a perivascular location in 1/3 astrocytomas, 17/24 glioblastomas, and 3/7 meningiomas, suggesting they could be produced also by the vascular stroma of tumours and that they would characterise the neoangiogenesis. All cultivated glioma cells tested produced HA in vitro, whereas only 1/11 cell lines produced HN, at a low level. The results obtained suggest that glioma HA and HN are produced by both cancer cells and vascular stroma cells, which contribute to the edification of the extracellular matrix. In meningiomas only the stroma would be responsible for HA and HN production.

Establishment and characterisation of a human glioma cell line.

A new cell line, CB109, has been established from a human glioblastoma multiforme. The cytoskeleton was positive for glial fibrillary acidic protein, vimentin and fibronectin. Hyaluronan (HA) and the HA-binding protein hyaluronectin (HN) were expressed in the cell cytoplasm and in the extracellular matrix of spheroids and plated cells. Hyaluronidase did not prevent spheroid formation suggesting that HA was not involved in the cell-cell adhesion. HA precoating prevented cell adherence to the plates and favoured spheroid formation. HA was secreted in relatively large amounts into the culture medium. High performance liquid chromatography demonstrated that HA was in the high molecular weight form. The rate of HN secretion by cells was very low. Basic fibroblast growth factor significantly increased the proliferation in vitro and tumour growth after grafting into nude mice. The epidermal growth factor receptor was not expressed on cultivated CB109 cells. Cytogenetic analysis showed polysomy 7, structural rearrangement of chromosome 10 short arm and a translocation 13q13-q14 without detectable alteration of the RB gene.

Characterization of a hyaluronic acid-binding protein from sheep brain comparison with human brain hyaluronectin.

1. A hyaluronic acid (HA)-binding glycoprotein from sheep brain was characterized. 2. The specific affinity for HA was shown in vitro by high performance liquid chromatography, polyacrylamide gel electrophoresis and ELISA methods. 3. The KD for high molecular weight HA was 5.4 10(-9) M at 37 degrees C and lower than 10(-10) M at 4 degrees C. 4. No link protein was found and HA molecules could bind up to 10 times their weight of the glycoprotein. 5. The specific site for interaction was the HA-derived decasaccharide HA10. 6. The protein is composed of one polypeptidic chain. Tryptophan and lysine play a prominent role in the conformation of the binding site to HA. 7. Enzyme analysis indicated that the protein different forms are due to differences in glycosylation and that N- and O-linkages coexist in the molecules. 8. Immunohistochemistry localized the glycoprotein at the nodes of Ranvier and at the periphery of neurons. The perineuronal labeling was seen around all neurons studied in the cerebellum whereas it was almost undetectable in the cerebral hemispheres. 9. HA is not saturated by hyaluronectin (HN) in the sheep nervous system. 10. The glycoprotein is largely similar to human brain HN, and different from the hyaluronate-binding protein characterized in the cartilage.

Serum hyaluronan (hyaluronic acid) in breast cancer patients.

Eighty-three women with breast cancer (57 with systemic metastasis, 26 without) were investigated for serum hyaluronan (HA) and compared to 50 patients with benign diseases of the breast. Hyaluronan was significantly increased in sera of metastatic patients compared to sera of non-metastatic patients (p less than 0.0001) and also in sera of non-metastatic patients when compared to control sera (p less than 0.01). The difference was not related to the number of metastatic sites involved. Three months after starting cytotoxic chemotherapy in metastatic patients, lower HA concentrations were observed in patients responding to chemotherapy. The initial level of serum HA had no predictive value concerning response to chemotherapy.