Research Paper Mediators of Inflammation, 7, 195–199 (1998)

The stimulation of granulocyte macrophage-colony stimulating factor (GM-CSF) by interleukin-1 (IL-1) has been shown to be counteracted in different mesenchymal cell systems by cyclic adenosine monophosphate (cAMP) agonists. The aim of this study was the evaluation of different cAMP agonists on GM-CSF expression in human bone marrow stromal cells. Incubation of secondary haematopoietic progenitor cell deprived human stromal cell cultures with IL-1 or TNF-alpha induced GM-CSF protein expression in culture supernatants and GM-CSF-mRNA in adherent stromal cells. The coincubation with 8-bromo-cAMP (8BrcAMP), a water soluble cAMP analogue, inhibited this GM-CSF stimulation at the protein and the mRNA level. This effect was dose dependent with a maximal inhibition of about 65% occurring at a 8BrcAMP concentration of 0.75 mM. In addition to 8BrcAMP, other cAMP agonists such as dibutyryl-cAMP, forskolin, pertussis toxin, or prostaglandin E2 (PGE2) had the same inhibitory effect on GM-CSF stimulation by IL-1. Coincubation with the cyclooxygenase inhibitor indomethacin had no significant influence on GM-CSF expression in stromal cells. Our results provide evidence that the previously described inhibitory effect of cAMP agonist PGE2 on haematopoietic progenitor cells in vivo is, at least in part, mediated by modulating the expression of GM-CSF in bone marrow stromal cells.


Introduction
The bone marrow stroma provides a microenvironment for proliferation and differentiation of haematopoietic stem and progenitor cells. Stromal cell cytokine production plays a pivotal role in controlling the development of sufficient mature blood cells under normal demands or after stimulation by inflammatory processes. 1 GM-CSF is a potent growth factor for a variety of haematopoietic progenitor cells. It is produced by T-lymphocytes, macrophages and a variety of mesenchymal cells. 2 Long-term bone marrow Dexter type cultures support the prolonged proliferation of haematopoietic progenitor cells, thereby providing an in vitro model for studies aimed at elucidating the mechanisms involved in haematopoietic regulation. The adherent layer of Dexter cultures is composed of mesenchymal bone marrow stromal cells. 3 The regulation of grow th factor production in the adherent layer of Dex ter type long-term bone marrow cultures closely resembles the in vivo situation. Multiple cytokines have been found in the supernatant of Dexter cultures, including GM-CSF, G-CSF, M-CSF, interleukin-6, and leukaemia inhibitory factor (LIF). GM-CSF expression in mesenchymal cells is induced by inflammatory cytokines like IL-1 or TNF-a and is controlled at both the transcriptional and posttranscriptional levels. 4 -6 In this study, we used human stromal cell cultures to evaluate the effects of cAMP on the expression of GM-CSF, induced by inflammatory cytokines. We demonstrated that a variety of cAMP agonists inhibit the production of GM-CSF at the protein and mRNA level. These data indicate that the inhibitory effect of cAMP agonists like PGE 2 on haematopoietic progenitor cells are, at least in part, mediated by modulating the expression of growth factors like GM-CSF in the bone marrow microenvironment.

Stromal cell culture
Adherent bone marrow stromal cell cultures were essentially obtained as previously described. 7 Bone marrow mononuclear cells, separated by centrifugation over Ficoll-Hypaque, were incubated at a cell density of 1 3 10 6 /ml in 25 cm 2 tissue flasks at 33°C in culture medium consisting of RPMI 1640 supplemented with 10% fetal calf serum (FCS), 10% horse serum, 1.0 m M hydrocortisone (Sigma Chemicals), and the additives as described. 7 In weekly intervals, cultures were fed by replacing 75% of the culture medium. When cultures were covered more than 80% by adherent cells, the primary cultures were treated with trypsin-EDTA (Biochrom, Berlin, Germany). Detached stromal cells were pooled and expanded in new culture flasks at a surface ratio of 1:5 in 75 cm 2 tissue flasks. The adherent cells were incubated under the same culture conditions until the cultures became confluent again. By this culture method, homogeneous stromal cell layers were obtained that were morphologically and functionally comparable in different culture flasks. These cultures were devoid of all haematopoietic cells, including macrophages. Hydrocortisone was removed from these cultures at least 3 days before use for RNA and protein analysis. GM-CSF was measured in the culture supernatants of stromal cells after incubation for 24 h w ith various factors, as indicated in the Results section. For RNA analysis, the factors, as indicated in the Results section, were added simultaneously to parallel culture flasks from the same donor. After 6 h the cells were lysed directly in the culture flask with guanidium-isothiocyanate lysis buffer and processed as described below. The content of a 75 cm 2 tissue flask yielded 10-20 m g total RNA.

Northern blot analysis
Total cytoplasmic RNA was prepared using the single step method of guadinium/phenol-chloroform extraction as described earlier. 8 Ten to 15 m g of RNA (depending on the least yield obtained in each experiment) were subjected to electrophoresis on a 1% agarose-formaldehyde gel and transferred onto nylon membrane (Hybond-N, Amersham Buchler). Blots were hybridized to a 32 P-labelled cDNA probes using random primer DNA labelling kit (Boehringer), washed and exposed to Cronex-4 autoradiography films (DuPont) at -70°C. Human GM-CSF cDNA, an 800-bp EcoRI fragment cloned into P91023(B)-vector, was obtained from American Type Culture Collection (Rockville, MD).

Enzyme-linked immunoabsorbent assay (ELISA)
Concentrations of GM-CSF were measured using Quantikine Human Cytokine Immunoassays (R&D Systems, Minneapolis, MN) according to manufacturer's instructions. Sensitivity of the test for human GM-CSF was 7.8 pg/ml.

Stimulated GM-CSF expression in human bone marrow stromal cells is inhibited by 8BrcAMP
Expression of GM-CSF in human bone marrow stromal cells was assessed at the protein level in culture supernatants by a sensitive ELISA and at the mRNA level by Northern blot analysis. In unstimulated cultures no or very low levels of GM-CSF mRNA were detected. After incubation with IL-1 (100 U/ml) or TNF-a (500 U/ml) for 6 h, significant GM-CSF-mRNA expression was induced. IL-1 was shown to have a stronger stimulatory effect than TNF-a (Figs 1 and 2B). These RNA-data were confirmed at the protein level. IL-1 and TNF-a incubation for 24-h induced GM-CSF secretion into stromal cell culture supernatants w ith IL-1 again being the stronger stimulus. Because the absolute amount of GM-CSF expression varied between ex periments, the protein level in culture supernatants from pooled experiments is depicted in Fig. 2A relative to the maximal stimulation.
The addition of the cAMP agonist 8BrcAMP to stromal cell cultures, stimulated by IL-1, TNF-a or the combination of IL-1 and TNF-a , resulted in sig-nificantly reduced GM-CSF mRNA expression. Similarly, the GM-CSF protein secretion, induced by IL-1 or TNF-a , was inhibited with 8BrcAMP (0.5 mM) coincubation by 65.5 ± 2.9% or 61.6 ± 6.1%, respectively. These results were obtained by pooling data from three independent experiments (Figs 1 and  2). Dose dependence of the modulatory effect of 8BrcAMP on GM-CSF secretion was evaluated at concentrations betw een 0.1 and 1.0 mM. Significant inhibition of IL-1 stimulated GM-CSF secretion was observed at 8BrcAMP concentrations of 0.1 to 0.25 mM and the maximal inhibition was seen at concentrations between 0.75 and 1 mM (Fig. 3).

Different cAMP agonists inhibit GM-CSF expression in stimulated human bone marrow stromal cells
In order to evaluate w hether the results obtained by 8BrcAMP incubation of human bone marrow stromal cells on GM-CSF expression could be generalized to other cAMP agonists we tested a panel of substances. IL-1 stimulated GM-CSF protein secretion into stromal cell supernatants was inhibited by all investigated cAMP agonists including forskolin at 10 m M (81.6 ± 5.8 inhibition), dibutyryl-cAMP at 1 mM (66.1 ± 8.1% inhibition), pertussis toxin at 1 m g/ml (59.0 ± 4.6% inhibition), and PGE 2 1 m g/ml (36.4 ± 14% inhibition). inhibitory influence on GM-CSF protein or mRNA expression (Fig. 4).

Discussion
The presented data show that GM-CSF is produced in adherent human bone marrow stromal cell cultures after stimulation w ith inflammatory cytokines IL-1 and TNF-a . This effect has previously been reported by several other groups. 9 -11 Some authors also described the endogenous expression of GM-CSF in the adherent layer of human Dexter-type cultures without external cytokine stimulation. GM-CSF mRNA could until now unequivocally only be demonstrated by PCR technique in stromal cells cultures. 9,10, 12 Charbord et al. 13 showed the inhibition of GM-CSF activity in complete Dexter-type cultures by neutralizing monoclonal antibodies that suppressed 70% of the endogenous CFU-GM. In our culture system, we could not demonstrate significant production of GM-CSF in human stromal cell cultures by a sensitive ELISA. This fact could be due to the deprivation of haematopoietic cells. These cells are likely to ex ert a paracrine stimulatory effect on adherent mesenchymal stromal cells and explain the constitutive expression of GM-CSF in complete Dexter-type cultures. Both IL-1 and TNF-a were able to induce GM-CSF in bone marrow stromal cell cultures, with IL-1 being the more potent inducer. These results are in accordance w ith published data. 11 In a variety of investigated systems, it has been shown that incubation with cAMP agonists causes a differential effect on the production of cytokines. While some cytokines, such as interleukin-6 and leukaemia inhibitory factor, are induced by increased intracellular cAMP concentrations, the expression of other cytokines is inhibited. 14 -16 To the latter group belong macrophage-colony stimulating factor, IL-1, platelet derived growth factor, and GM-CSF. [17][18][19] Patil and Borch reported on the inhibitory role of PGE 2 on GM-CSF production in human fibroblasts. This effect was mediated by increased intracellular cAMP levels. Our group has described a similar finding of PGE 2 and other cAMP agonists inhibiting GM-CSF expression in a murine bone marrow stromal cell line. 20,21 The present data proves the same inhibitory effect of increased intracellular cAMP concentration in human bone marrow stromal cells at the protein and mRNA level. This inhibition was dose-dependent as shown by increasing suppression of IL-1-induced GM-CSF protein production by the water-soluble cAMP analogue 8BrcAMP. Maximal inhibition of about 60% was obtained at a concentration of 0.75 mM. The inhibitory action of 8BrcAMP was demonstrated on GM-CSF expression induced by IL-1, TNF-a , or the combination of IL-1 and TNF-a . The interference with the stimulatory action of different cytokines indicates that cAMP interacts w ith components that are either shared among different signalling pathways or act downstream from the convergence point of these pathways. However in earlier ex periments on the murine bone marrow stromal cell line +/+-1.LDA11 we demonstrated that IL-1 and TNF-a incubation had no influence on intracellular cAMP levels. 20 In addition to the water soluble cAMP analogues 8BrcAMP and dibutyryl-cAMP other cAMP agonists