Cell membrane-associated proteoglycans mediate extramedullar myeloid proliferation in granulomatous inflammatory reactions to schistosome eggs

In normal adults, myelopoiesis occurs in the bone marrow, where cells of the medullar stroma, locally produced or extramedullar cytokines, and extracellular matrix, form the haematopoietic bone marrow microenvironment. The interaction of both stromal and haematopoietic cells with cytokines and the associated extracellular matrix molecules controls the progressive commitment and differentiation of myeloid stem cells into mature blood cells. These controls involve both the maintenance of blood cell homeostasis and the capacity to increase production of one or several cell lines in a coordinated and strictly controlled manner. The controls of myeloid cell production and differentiation are thus necessarily complex and tightly regulated.

In contrast to this tight regulation of the bone marrow haematopoietic environment stands the fact that partial or full haematopoiesis can be observed in extramedullar loci, like spleen or liver, in many pathological situations in adult life. This is frequent in diseases where the bone marrow microenvironment is either modified, as in myelofibrosis, or occupied by new cell populations, as in neoplasias involving bones. This phenomenon can be understood as a passive dislodging of haematopoiesis into sites that have had the haematopoietic function in earlier embryonic development: it thus represents a case of myeloid metaplasia. On the other hand, in chronic tissue inflammatory reactions, a local myeloid proliferation may be observed. In schistoso- miasis, both in humans and in experimental models, pro- liferation of mono-macrophagic, neutrophil, eosinophil and megakaryocytic cell lineages, alone or in combination, have been described, in association with diffuse or granulomatous inflammatory reactions to parasites (Borojevic and Carvalho, 1981; Borojevic et al., 1981, 1983, 1989a,b; Clark et al., 1988; El-Cheikh et al., 1991). Since in schistosomiasis the general bone marrow function was not hampered, this phenomenon has to be understood as an active displacement of myelopoiesis outside the normal myelopoietic environment, controlled by specific cellular and molecular systems.

Recent studies have shown that proteoglycans of the bone marrow extracellular matrix are involved in the spatial organization of the medullar haematopoietic environment, as well as stabilization and presentation of specific growth factors to the corresponding myeloid precursors (Gallagher and Dexter, 1988; Allen et al., 1990). The same type of molecule is considered to be responsible for adhesion of cells in the bone marrow, and for control of their release into the blood circulation after their maturation (Kolset and Gallagher, 1990).

We have recently studied myelopoiesis involving both mono-macrophagic and eosinophil cell lines in hepatic granulomas, formed around schistosome eggs embolized in the intrahepatic portal venous radiculi (Borojevic et al., 1989b; El-Cheikh et al., 1991). We have also characterized glycosaminoglycans and proteoglycans that are present in granulomas in vivo, and synthesized in vitro by granulomas isolated from hepatic tissue, or by granuloma-derived connective tissue cell lines (Silva et al., 1992a,b). In the present study, we have attempted to identify the controls that determine extramedullar proliferation of selected blood-cell lineages in the specific granulomatous inflammatory microenvironment. In particular, we have tried to determine if interactions among extracellular matrix or cell membraneassociated proteoglycans and cytokines control proliferation and differentiation of myeloid precursors, in the same way as mechanisms that govern the medullar myelopoiesis.

One-month-old C3H/HeN mice, obtained from the colony bred at the Federal University of Rio de Janeiro, were infected by transcutaneous penetration of 30 cercariae of Schistosoma mansoni (BH strain, Instituto Oswaldo Cruz, Rio de Janeiro). Mice were killed by ether overdose after 90 days of infection, corresponding to the early chronic phase of the disease (Borojevic et al., 1984). Periovular granulomas were obtained under sterile conditions by mechanical disruption of livers, and subsequent separation of granulomas by repeated sedimentation, as previously described (Pellegrino and Brener, 1956). Granulomas were plated in 25 cm² tissue culture flasks (Costar, Cambridge. MA), and maintained in the “standard culture medium”: Dulbecco’s modified minimum essential medium (D-MEM) (Sigma Chemical Co., St. Louis, MO), supplemented with 10% fetal bovine serum (FBS) (Cultilab, Campinas, Brazil), and 2 g/l HEPES buffer, pH 7.4. They were incubated at 37°C in a humidified atmosphere containing 5% CO₂ for two weeks. After this period, the connective tissue cells that had spontaneously migrated from granulomas were harvested by trypsinization. They were replated by double-split when reaching early confluence; cell cultures between the 7th and the 20th passage were used in this study. All the assays were done with cells that had reached early confluence. These granuloma-derived cell lines were termed “GR cells”.

Primary cultures of skin fibroblasts were obtained from newborn C3H/HeN mice. Mice were killed, and skin was sterilized, harvested and digested by collagenase (Sigma, type IA, 1 mg/ml in balanced sat solution; BSS) for 2 h, at 37°C under continuous stirring. This treatment was followed by trypsin (0.25% in calcium- and magnesium-free BSS supplemented with 0.05% ethylenediamine tetraacetate (EDTA), 2 h at 37°C. The obtained fibrolastoid cells were cultured as GR cells.

Permanent cell lines were obtained from the Rio de Janeiro Cell Bank (BIO-RIO, Rio de Janeiro, Brazil). We used the multipotent myeloid stem cell line FDC-P1 (interleukin-3 (IL-3)- and/or granulocyte-macrophage colony stimulating factor (GM-CSF)-dependent) and the myeloid cell lines DA-1 (IL-3 and/or GM-CSF-dependent) and AD-3 (exclusively IL-3-dependent) (Dexter et al., 1980). These cell lines were routinely maintained in culture in the standard medium supplemented with supernatant from WeHi-3B cells that secrete IL-3 constitutively (Ymer et al., 1985).

Neutralizing monoclonal antibodies against murine GM-CSF were obtained from supernatants of 22E9 cells, supplied originally to the Rio de Janeiro Cell Bank by R. Coffman, DNAX Institute for Cellular and Molecular Biology, Palo Alto, CA. Supernatants were harvested from 72-h cultures of 10⁵ cells/ml; when added at 10% concentration to the standard medium, they neutralized 0.1 ng/ml recombinant murine GM-CSF. Recombinant murine GM-CSF and purified monoclonal antibody anti-IL-3 were generously supplied by the DNAX Institute for Cell Biology, Palo Alto, CA. Feeder layers of GR cells or skin fibroblasts were studied under physiological conditions, or after irradiation (4000 rad). Alternatively, confluent cell layers were rinsed with balanced salt solution (BSS), and fixed with 2% glutardialdehyde in BSS, for 5 min at 37°C (Roberts et al., 1987). Cell layers were then extensively washed with BSS to remove all the residual fixative, and used as a substratum for cultures of myeloid cell lines.

For co-culture experiments, GR cells or skin fibroblasts were plated into 35 mm plastic Petri dishes (5 × 10⁴ cells/dish), or into 24-well or 96-well culture plates (10⁴ and 5 × 10³ cells/well, respectively); they were maintained in culture until reaching early confluence (minimum 48 h). The FDC-P1, DA-1 or AD-3 cells were extensively washed in BSS to remove residual IL-3 from their culture medium, resuspended in the fresh standard culture medium, and inoculated onto GR or skin fibroblast monolayers (2×10⁴ cells/Petri dish, 10⁴ or 10³ cells/well for living feeder layers; 5 × 10⁴ or 5 × 10³ cells/well for fixed cell layers). Alternatively, the established feeder layers were covered with an approximately 2 mm thick layer of agar (GIBCO, New York, NY), to avoid direct contact of myeloid cell lines with the underlying cell layer. Agar was prepared in the standard medium, at a final concentration of 0.5%.

Adhesion of myeloid cells to the feeder layer was monitored after labelling with methionine. FDC-P1 and DA1 cells were washed and resuspended in the standard medium containing 10% supernatant of WeHi-3B cells supplemented with 4 μCi of [³⁵S]methionine/ml (specific activity 700 Ci/mmole, ICN, Irvine, CA). They were incubated for 24 h, washed twice with BSS, and incubated for 90 min in the standard medium, to remove any remaining IL-3 and [³⁵S]methionine. The labelling obtained was greater than 6 ×10⁴ cts/min/10⁵ cells. Labelled cells were washed and inoculated onto confluent GR monolayers in 24-well culture plates, as described (Roberts et al., 1987). After different periods of co-incubation, cell layers were washed three times with BSS with gentle stirring, to remove non-adherent cells. The remaining cells were dissolved in 0.5 ml 5% Triton X-100, and the radioactivity was quantified by liquid scintillation.

For long-term quantification of the proliferation of myeloid cells adherent to the feeder layer, co-cultures were gently washed with medium, and myeloid cells that remained attached were counted under a phase-contrast microscope. Ten fields (2.36 mm² each) were counted for each culture. For proliferation quantification of non-adherent cells, cultures were gently shaken, and non-adherent cells were aspirated and quantified by counting in a haemocytometer. The total proliferation of cells in co-cultures (feeder layer and overlaid cells) or proliferation of non-adherent cells in cultures over irradiated or prefixed cell layers, were quantified by [³H]thymidine (43.5 Ci/mmol, Sigma) incorporation. Cells were incubated with 0.5 μCi of [³H]thymidine/ml of culture medium for 48 h. Alternatively, viable cells were quantified by the MTT colorimetric assay for mitochondrial dehydrogenase, as described by Mosmann (1983). Results were analysed by the Mann-Whitney test.

Extracellular matrix of GR cell monolayer was obtained from cultures as described by Woods et al. (1985). Briefly, confluent cell layers were washed with BSS and incubated for 15 min at 37°C in the 25 mM Tris-HCl buffer (pH 7.5) containing 0.2% Triton X-100, 0.2 mM phenylmethylsulphonyl fluoride (PMSF), 20 mM EDTA and 5 mM benzamidine-HCl. After the dissolution of cells, the remaining adherent extracellular matrix and the non-solubilized cytoskeleton were extensively washed with BSS, and used for cell cultures. The presence of fibronectin and actin in the remaining layer was demonstrated by standard immunoperoxidase staining using monoclonal antibodies against fibronectin (F6140, Sigma) and actin (A2668, Sigma), reacted with peroxidase-conjugated anti-mouse antibodies (A4416, Sigma).

To obtain the cell membrane preparations, GR cells were cultured to confluence in 150 cm² flasks (Costar). Cells were harvested with a rubber policeman in BSS supplemented with 5% FBS, homogenized (Potter-Elvehjem) and centrifuged at 200 g (10 min), 1000 g (20 min) and 45,000 g (60 min) at 4°C. The final pellet was assayed for protein content, following the modification of Lowry’s method by Markwell et al. (1978). All the assays with cell membrane or extracellular matrix molecules were performed in 96-well culture plates, and the cell proliferation was monitored by MTT assay.

Digestion of cell layer-associated proteoglycans was performed with heparitinase I (EC 4.2.2.8) or chondroitinase AC (EC 4.2.2.5) (both from Sigma). Previous study has shown that GR cell-layerassociated proteoglycans do not contain dermatan sulphate (Silva et al., 1992b). Cells were washed with BSS, and incubated for 2 h at 37°C with 10 munits/ml of enzyme in BSS supplemented with 2 g/l HEPES (pH 7.4). After digestion, cells were washed and fixed in glutardialdehyde, as described above.

To study the association of growth factors with the cell-associated molecules, cell layers were washed and treated with 2 M NaCl solution, or with 0.5 mg/ml heparin in BSS, for 30 min. The cell layer was subsequently fixed with glutardialdehyde, and its capacity to sustain proliferation of myeloid cell lines was monitored.

Alternatively, cell layers were pre-treated with 2 M NaCl, fixed with glutardialdehyde, extensively washed with BSS supplemented with 2% bovine serum albumin (BSA), and incubated for 1 h with the standard culture medium containing 0.5 ng/ml recombinant murine GM-CSF. This GR layer was further extensively washed, and its capacity to sustain myeloid cell proliferation was monitored by the MTT assay.

The supernatant of the NaCl-treated cell layer was extensively dialysed against BSS, and assayed for stimulation of myeloid cell proliferation in liquid cultures. Inhibition of the stimulation was assayed by adding 20% of supernatant of 22E9 cells containing anti-GM-CSF antibody.

Proteoglycans of GR cells were extracted, purified and analysed and described previously (Silva et al., 1992b).

Primary GR cells were stellate or elongated, with numerous large terminally spread pseudopodia. They grew in monolayers until full confluence, but they were only weakly inhibited by cell contact. Similar to most cell lines of the smooth muscle cell lineage, they overgrew into a typical “hills and valleys” pattern when reaching hyperconfluence. Skin fibroblasts had a typical fibroblastoid morphology: they were fusiform, and in confluence they formed monolayers organized into parallel stands and whirls.

FDC-P1, DA-1 cells and AD-3 cells required for their proliferation the presence of IL-3, routinely provided by the supernatant of WeHi-3B cells. FDC-P1 and DA-1 cells also proliferated in the presence of recombinant murine GM-CSF. No proliferation was observed in any of the cell lines studied in the presence of GR cell supernatants (Fig. 1).

Conversely, when cultured in the standard medium, but plated onto a monolayer of GR cells, a sustained increase in FDC-P1 and DA1 cells was observed, at first on the feeder-layer and subsequently in the culture supernatant. The AD-3 cell line showed very low or no proliferation stimulation (Fig. 2). These results indicated that the IL-3 was not present in this experimental system, and that the stimulation of FDC-P1 and DA-1 cell proliferation was possibly due to the presence of GM-CSF. This hypothesis was further confirmed by experiments with anti-GM-CSF neutralizing monoclonal antibodies that abolished the observed proliferation stimulation associated with the GR cell layer, whilst the anti-IL-3 antibodies did not have a similar effect (Table 1). This was also confirmed in parallel research using the reverse transcriptase-polymerase chain reaction to detect the message for cytokines in GR cells (Borojevic et al., 1992). In this study we could demonstrate the constitutive expression of the mRNA for GM-CSF in GR cells, but no message for IL-3 could be detected.

Interaction between the FDC-P1 cells and feeder-layers involved both adhesion and stimulation of cell prolifera-tion. Adhesion was quantified in a short-term assay, using myeloid cells labelled with [³⁵S]methionine (Fig. 3). Significant adhesion was observed after a short period of incubation (inferior to 30 min), increasing slowly afterwards. In a long-term assay, cells adherent to the feeder-layer were counted during five days of co-culture (Fig. 4A). The specificity of myeloid cell adhesion to connective tissue cells was shown by comparing feeder-layers of skin fibroblasts and GR cells. Although both cell layers offered a better substratum than plastic, the adhesion to GR cells was significantly higher than to skin fibroblasts, indicating that the adhesion of myeloid cells to their feeder-layer is specific for the cell type. The number of FDC-P1 cells attached to the GR monolayer increased continuously from the first day on, while those plated on skin fibroblasts maintained stationary growth after the second day, indicating that GR cells, but not the skin fibroblasts feeder-layer, could provide a suitable environment for myeloid cell survival and proliferation (Fig. 4A). When FDC-P1 cells were quantified in supernatants of GR cells or skin fibroblast feederlayers co-cultured with FDC-P1 cells, a net increase was observed after three days of co-culture with GR cells; a very small increase was observed over skin fibroblast feeder-layers (Fig. 4B). In view of the results shown in the Fig. 4A, we understand the lag-phase of three days to be a period necessary to attain focal overcrowding of proliferating FDC-P1 cells on the feeder layer, with their subsequent progressive release into the supernatant.

The question of whether the FDC-P1 cells require a direct contact with the feeder-layer was addressed in cultures in the standard medium supplemented with the supernatant of GR cells, and in cultures of FDC-P1 cells over the GR cell monolayer covered by a thin layer of agar. In both cases, after 120 h of culture, proliferation of FDC-P1 cells was less than 5% of that observed in the presence of the WeHi cell supernatant, indicating that their physical contact with the feeder layer was indeed required.

The possible function of the continuous GR cell metabolic activity for myeloid cell proliferation was assessed in co-cultures of FDC-P1 cells with GR layers that had been previously irradiated or fixed with glutardialdehyde. In both cases, the stimulatory activity was decreased as compared to non-treated cell layers, but in both cases significant cell proliferation could be observed for as long as five days (Table 2). These results indicate that the proliferation-stimulating activity is stably associated with the cell layer, and can partially retain its biological activity after a smooth fixation.

The nature of the association of the growth-stimulating factor with the cell layer was studied in assays of the dis-location of the proliferation stimulating activity into the supernatant. Fig. 5A shows that treatment of feeder layers with 2 M salt buffer or heparin could impair proliferation stimulation. The stimulation could be restored by incubation of NaCl-treated and fixed GR cell layers with recombinant murine GM-CSF. Since in this experiment the glutardialdehyde-fixed GR feeder-layer had been previously saturated with bovine serum albumin, the binding of GM-CSF to the GR layer was specific.

The probable candidates for this association are the pro-teoglycans, composed of a proteic core and covalently attached glycosaminoglycans (Gallagher and Dexter., 1988; Kolset and Gallagher, 1990). In order to investigate this possibility, we treated the GR cell layer by mild trypsinization or with mucopolysaccharidases (Fig. 5A). Trypsinization abolished both the adhesion of myeloid cells to the GR layer and their proliferation. Treatment with heparitinase did not modify the adhesion, but abrogated the proliferation stimulation, while treatment with chondroitinase had no effect.

The cell proliferation stimulatory activity, dislodged from the GR cell layer with NaCl, could be subsequently recovered in the high-salt extracts, after dialysis against the standard medium. When this extract was added to cultures of FDC-P1 cells in the standard medium, it stimulated cell proliferation (Fig. 5B). Its activity could be significantly decreased (α < 0.005) by treatment with anti-GM-CSF neutralizing antibody, again identifying GM-CSF as the cytokine responsible for proliferation stimulation in this experimental series. The fact that less inhibition was obtained using neutralizing monoclonal antibodies against this extract, as compared to the recombinant GM-CSF (Table 1), may indicate that after dialysis the growth factor may associate with other molecules present in the high-salt extract of the GR cell layer, which may either protect it from the neutralizing antibody or enhance its proliferationstimulatory activity. This issue is at present under study. These data strongly suggest a non-covalent association between the growth factors and cell surface molecules produced by the GR cell layer.

Taken together, the results obtained suggest that the proliferative activity was: (a) non-covalently associated with the cell layer, since it could be dislodged by ionic or charged molecules as salt or heparin; (b) associated with heparin-like or heparan sulphate-containing molecules, since it was sensitive to heparitinase but not to chondroitinase; and (c) associated with a protein-containing molecule, since it was sensitive to mild trypsinization; (d) or that cell adhesion was necessary for proliferation stimulation, but was mediated by proteic moiety of heparan sulphate proteoglycan, or by another cell surface protein.

The association between the cell layer and the proliferation-stimulating activity could be located on cell membrane-associated molecules or on the extracellular matrix. Treatment with mild detergent can dissolve cells without removing molecules associated with the extracellular matrix, thus preserving its biological activity (Woods et al., 1985). When this treatment was applied to GR cells, the remaining extracellular matrix contained fibronectin, as well as some of the cell cytoskeletal elements, as observed by immunocytochemical reactions (not shown). When FDC-P1 cells were plated onto this extracellular matrix derived from GR cell monolayer, no proliferation was observed (Fig. 6). This cell-free matrix was also unable to provide more than a transient adherence of myeloid cells. This indicated that the adhesion of FDC-P1 cells was necessary for proliferation stimulation and that it was mediated by cell membrane-associated molecules. When cell membranes were isolated from GR monolayer and added to FDC-P1 cell culture medium, the proliferation of FDC-P1 cells was recovered. The addition of GR cell membrane preparation, at a concentration of 20 μg protein/well, could sustain FDC-P1 proliferation for more than five days (Fig. 6). These results confirmed that the proliferation-stimulating activity was associated with cell membranes.

In view of the previously cited observations that the association of cytokines with cell layers could be mediated by heparin- or heparan sulphate-containing molecules, we have analysed the cell-associated proteoglycans. The extracted radiolabelled macromolecules from the cell-associated fraction of GR cells, chromatographed on DEAE-Sephacel columns, eluted as a single peak at approximately 0.40 M NaCl. A previous study (Silva et al., 1992b) has shown that this material separated on a Sephacryl S-400 column into two peaks: the larger proteoglycan contained heparan sulphate and chondroitin sulphate, and the smaller one the chondroitin sulphate chains only. Beta-elimination of material eluted from DEAE-Sephacel yielded a broad single peak on a Sephacryl S-400 column, confirming the proteoglycan nature of the studied molecules. Total separation between heparan sulphate and chondroitin sulphate proteoglycans obtained by anion-exchange chromatography was achieved by a hydrophobic-affinity chromatography column. Agarose gel electrophoresis has shown that the unbound fraction consisted of chondroitin sulphate proteoglycans, and represented approximately 60% of the cell-associated proteoglycans. The material that remained bound to the column represented approximately 40% of the cell-associated proteoglycans. It could be recovered using the detergent gradient. This material was observed to consist of heparan sulfate proteoglycan, the protein moiety of which is responsible for its association with a hydrophobic environment, in agreement with our previous results (Silva et al., 1992b).

In view of a recent report on the ability of a heparan sulphate-containing fraction of bone marrow to stimulate maturation of a myeloid cell line (Luikart et al., 1990), we have tested the capacity of both the chondroitin sulphate- and the heparan sulphate-containing proteoglycans, obtained from the GR cell layer and separated by hydrophobic-affinity chromatography, to stimulate proliferation of FDC-P1 cells. No stimulation was observed (results not shown). These data suggest that an association among endogenous growth factors and proteoglycans was responsible for the observed phenomenon, and that proteoglycans themselves did not have a growth-promoting activity.

In the present study, we have analysed molecular mechanisms that underlie the myeloid proliferation associated with schistosome egg-induced granulomas in liver. Connective tissue cells, designated GR cells, were isolated from granulomas. They have been shown to be able to sustain a long-term proliferation of myeloid cell lines. This stimulation is endogenous, i.e. it is constitutively produced by GR cells in vitro. We were able to determine that this stimulation corresponds to GM-CSF by monitoring proliferation of growth-factor-dependent cell lines and using neutralizing monoclonal antibodies. No proliferation-stimulatory activity could be observed in GR cell supernatants, indicating that no active molecules were secreted free into the culture medium. Conversely, all the activity was associated with the cell layer, and the physical contact between the feeder layer and the proliferating myeloid cells was rigorously required for the growth stimulation. Analysis using either highly charged molecules or a high ionic force have shown that the association of the growth factor with the cell layer was non-covalent. Analysis using trypsin and specific mucopolysaccharidases has indicated that the active molecule was a heparan sulphate-containing cell membrane proteoglycan. We have also shown that GR cells do indeed have a membrane-associated molecule corresponding to this profile, but without association with the growth factor, these molecules showed no growth-promoting avtivity. Taken together, these results indicate that in extramedullar myelopoiesis associated with schistosomal granulomas, cell surface heparan sulphate proteoglycans bind and expose growth factors to the myeloid progenitors.

In general terms, the systemic myelopoiesis in bone marrow and the peripheral myelopoiesis associated with inflammatory granulomatous reactions are regulated by similar molecular systems. In both cases, cytokines responsible for myeloid cell proliferation act in the context of the local inductive microenvironment composed of cells and extracellular matrix (Nathan and Sporn, 1991).

A major difference between the two sites of myelopoiesis apparently lies in the precise localization of heparan sulphate proteoglycans that interact with cytokines. In the bone marrow, this heparan sulphate was reported to be part of the local extracellular matrix (Bentley et al., 1990; Luikart et al., 1990), and it could be substituted by a basement membrane-like artificial matrix, derived from EHS sarcoma (Roberts et al., 1988). Alternatively, intercellular non-membrane-associated matrix, isolated either from 3T3 cells or from the bone marrow stroma, was reported to be unable to sustain myeloid proliferation (Roberts et al., 1987). In granulomas, the heparan sulphate proteoglycans that bind cytokines are apparently cell-membrane-associated molecules. This difference may reflect the fundamental distinction between the two phenomena. Medullar myelopoiesis provides throughout life both the basic continuous production of myeloid cell lines to the blood circulation, and the induced myeloid response to infections or inflammation in the body. Association of cytokines with the extracellular matrix may provide the necessary stability for this myeloid cell production. Conversely, in peripheral inflammatory reactions, local myelopoiesis has to be limited, both in time and in space. It has to be strictly controlled in terms of stimulation of one or several differentiated myeloid cell lines produced in situ, and it has to be interrupted at the end of inflammation. A mechanism associated with cellmembrane-anchored molecules is amenable to simple and direct controls by the cells involved. Membrane-associated endoglycosidase with substrate-specificity for heparan sulphate proteoglycans have been described in liver cells: they operate at physiological pH and may exert direct control over membrane proteoglycans, without endocytosis and involvement of the lysosomal compartment (Gallagher et al., 1988). Several tightly controlled functions at the cell membrane level are known to be restrained by cell surface proteoglycans, such as lipoprotein-lipase turnover in adipocytes (Cisar et al., 1989), transferrin-binding in liver endothelium (Omoto et al., 1990), or free radical destruction by blood cells (Adachi and Markland, 1989).

In the present study, we have analysed only the proliferation of myeloid cell lines corresponding to the GM-CSF-controlled amplification of the mono-macrophagic cell lineage in schistosomal granulomas (Clark et al., 1988; Borojevic et al., 1989a,b). Other cell lines, like eosinophils, mast cells and lymphocytes, are also produced locally in granulomas (El-Cheikh et al., 1991; Lenzi et al., 1987). As well as the binding of endogenous GM-CSF, GR cells may also control the exogenous cytokines, produced in granulomas by other cells, or brought there via the circulation. Studies on the interaction of GR cells with other growth factors is at present under way.

This work was supported by the FINEP and CNPq, Brazil, and the Commission of the European Communities (CEC STD2 217 M-B). Marcia C. El-Cheikh and Hélio S. Dutra are gratefully thanked for help in obtaining schistosomal granulomas and granuloma-derived cell lines.