Summary.We quantified tissue inhibitor of metalloproteinase (TIMP)-1 and TIMP-2 in serum and plasma in normal control subjects and patients with a low or high platelet count, using one-step sandwich enzyme immunoassays. The serum levels of TIMP-1 and TIMP-2 were 101 . 1 Ϯ 13 . 3 ng/ml, and 82 . 7 Ϯ 26 . 3 ng/ml, respectively, in normal subjects. In patients with an elevated platelet count, such as in essential thrombocytosis, polycythaemia vera, and myelofibrosis, serum levels of TIMP-1 and TIMP-2 were 351 . 6 Ϯ 200 . 9 ng/ml and 148 . 9 Ϯ 84 . 0 ng/ml, respectively. Serum levels of TIMP-1 and TIMP-2 in patients with a low platelet count, such as in aplastic anaemia and idiopathic thrombocytopenic purpura, were 57 . 2 Ϯ 25 . 8 ng/ml and 19 . 7 Ϯ 7 . 68 ng/ml, respectively. The serum level of TIMP-1 was significantly correlated with the platelet count in all subjects. The correlation between the serum level of TIMP-2 and the platelet count was not as strong. The level of TIMP-1 in platelet-depleted plasma was not correlated with the platelet count.Immunohistochemical staining using monoclonal antibodies against TIMP-1 and TIMP-2 showed that megakaryocytes and platelets were positive for both TIMP-1 and TIMP-2, confirming that they are rich sources of TIMPs. TIMP-1 and TIMP-2 stimulated the proliferation of bone marrow fibroblasts, although their effect was less potent than that of TGF-b and PDGF.Erythroleukaemia and megakaryoblastic cell lines showed the highest secretion of TIMP-1 among the leukaemia cell lines examined. There was no lineage specificity for TIMP-2 secretion. These results suggest that TIMPs released from megakaryocytes or from local platelet coagulation may be important in the development of bone marrow fibrosis.

Midkine (MK) was originally cloned as a product of a retinoic acid-responsive gene. The rationale for studying MK expression is based on previous reports showing that it transforms 3T3 cells, and that it acts as an autocrine growth factor in Wilm's tumors, and that its overexpression has been associated with worse outcome in bladder carcinoma. Besides bladder carcinoma, its expression was reported in various solid tumors. We investigated the expression of MK protein and/or MK gene in biopsied specimens from 40 patients with primary malignant lymphoma, 21 with Hodgkin's disease (HD) and 19 with non-Hodgkin's lymphoma (NHL). Reed-Sternberg (R-S) cells were stained positive in 10 of 16 HD cases evaluated by immunohistochemical method, whereas 18 of 19 NHL cases did not stain, and one B-cell NHL stained weakly positive. Immunostaining analysis was extended to established cell lines and to normal lymphocytes with or without lectin stimulation or with EB virus transformation. Among hematopoietic cells examined, erythro- or megakaryoblastic leukemia cell lines (K562, MEG-01 and UT7) were positive, while normal lymphocytes (except the EB virus-transformed one) and most myeloid and lymphoid cell lines (except Raji cells) were negative. On the contrary, solid tumor cell lines showed high and strongly positive staining including cell lines derived from of lung gastric, colon, and a pancreatic cancer. Using semi-quantitative reverse transcription polymerase chain reaction (RT-PCR), which is suitable for the detection of weakly expressed mRNA, the relative ratio of MK mRNA to beta-actin mRNA of samples was measured and compared in cases where RNA was available. The mean values of relative ratio (MK/beta-actin) of HD were almost twice as those of NHL samples, peripheral blood T cells, and spleen B cells. Our findings showed that MK is expressed in Reed-Sternberg cells of HD, and that MK might play a role in the pathogenesis of HD.