Granulocyte–macrophage colony-stimulating factor regulates cytokine production in cultured macrophages through CD14-dependent and -independent mechanisms (original) (raw)

Abstract

Granulocyte–macrophage colony-stimulating factor (GM-CSF) has multiple effects on the antigen phenotype and function of macrophages. In this study we investigated the effect of GM-CSF on cytokine production by macrophages. We found that GM-CSF may modify the tumour necrosis factor-α (TNF-α) and interleukin-6 (IL-6) response to lipopolysaccharide (LPS) through two different mechanisms. Relatively early in culture, GM-CSF increases the amount of cytokines synthesized by responding cells; this effect appears to be unrelated to modulation of CD14 expression and LPS-binding capacity. After prolonged incubation, GM-CSF up-regulates both CD14 expression and LPS-binding capacity, and the frequency of cytokine-producing cells. Release of CD14 in the culture supernatant was decreased in the presence of GM-CSF, suggesting that a reduced shedding was responsible for the effect of GM-CSF on CD14 expression. Enhancement of cytokine production was also observed in GM-CSF-treated macrophages after stimulation by phorbol 12-myristate 13-acetate (PMA), thus indicating that GM-CSF affects both CD14-dependent and -independent cytokine production. Finally, GM-CSF did not modulate the LPS- and PMA-induced production of IL-10 and IL-12. We conclude that GM-CSF may play a role in manipulating the activation-induced expression of pro-inflammatory cytokines by macrophages. Enhanced production of these cytokines could play an important role in the pathogenesis of Gram-negative septic shock syndrome and in defence against infectious agents.

INTRODUCTION

Production of cytokines by cells of the monocyte/macrophage lineage is an important action of the immune system. Among the various cytokines produced by monocyte/macrophages, tumour necrosis factor-α (TNF-α) and interleukin-6 (IL-6) are induced by engagement of the CD14 receptor by lipopolysaccharide (LPS) and exert mostly pro-inflammatory activity.1 The induction of TNF-α and IL-6 by LPS is a critical host defence response, however, an excessive activation may result in Gram-negative septic shock syndrome in Gram-negative bacterial infections.2,3 On the other hand, monocyte/macrophages also release immunoregulatory cytokines, such as IL-10 and IL-12, which are critically involved in the regulation of the T helper type 1 (Th1)/Th2 characterization of the lymphocyte response.4 Granulocyte–macrophage colony-stimulating factor (GM-CSF) is a member of the family of colony-stimulating factors involved in proliferation and differentiation of myeloid precursor.5–7 Monocytes and macrophages are known to have receptors for GM-CSF,8 and GM-CSF has multiple effects on the antigen phenotype and function of monocyte/macrophages. Incubation of human peripheral blood monocytes and macrophages with GM-CSF increases the expression of surface antigens like CD40,9 CD1a,b,c10 and human leucocyte antigen (HLA) -DR.11 Conflicting reports have been published on its influence on CD14 expression. In addition, GM-CSF stimulates the secretion of IL-8 and macrophage-colony stimulating factor (M-CSF)12 and induces the mRNA for TNF-α.13

In this study we have analysed the effect of GM-CSF stimulation on pro-inflammatory and immunoregulatory cytokine production by monocyte/macrophages. We found that GM-CSF is able to modulate the activation-induced production of TNF-α and IL-6 with both CD14-dependent and -independent mechanisms.

MATERIALS AND METHODS

Compounds

Recombinant GM-CSF was obtained from Sandoz Research Institute (East Hanover, NJ) and contains 5·4 × 106 chronic myelogenous leukaemia units per milligram of glycoprotein. Recombinant M-CSF was kindly provided by Genetic Institute (Cambridge, MA). The product concentration was 0·78 mg/ml, and the specific activity was 1·9 × 106 units/mg of protein (where one unit equals half maximal stimulation in the Murine Bone Marrow Colony Assay). LPS and fluorescein isothiocyanate (FITC) -conjugated LPS from Escherichia coli 0111/B4 were purchased from Sigma Chemical Co. St Louis, MO. Phorbol 12-myristate 13-acetate (PMA) was purchased from Calbiochem-Novabiochem INTL, La Jolla, CA.

Cells

Peripheral blood obtained from healthy donors was enriched for peripheral blood mononuclear cells by centrifugation over Ficoll–Hypaque. The peripheral blood mononuclear cells were then further enriched for monocytes by elutriation as previously described.14 Cells obtained by this method are > 90% monocytes as determined by fluorescence-activated cell sorter (FACS) analysis. The cells were cultured in RPMI-1640 medium supplemented with 20% heat-inactivated fetal calf serum, 2 mm l-glutamine, 50 U/ml penicillin, 50 µg/ml streptomycin, referred to as complete medium, with or without 100 U/ml GM-CSF, at 37° in a humidified atmosphere of 5% CO2 in air, in 48-well plates (Costar, Cambridge, MA) at a concentration of 5 × 105 cells/well/ml.

Limulus amoebocyte lysate test

All the compounds and media used in this study were analysed for endotoxin contamination by the limulus amoebocyte lysate test (QCL-1000, BioWhittaker, Inc., Walkersville, MD). All the samples analysed were found to be free of endotoxin contamination (less than 0·1 EU/ml).

Assessment of cell viabilty

The number of live cells was determined by trypan blue dye assay. Propidium iodide staining and flow cytometric analysis were employed to quantify the percentage of necrotic and apoptotic cells. Briefly, at given time-points, the cells were washed twice in phosphate-buffered saline (PBS), detached by gentle scraping and collected by low-speed centrifugation. The centrifuged cell pellet was gently resuspended in either 0·5 ml of PBS containing 2 µg/ml propidium iodide (Sigma) (for the quantification of necrotic cells) or was fixed in 70% ethanol for 24 hr and then resuspended in 0·5 ml of hypotonic fluorochrome solution containing 10 µg/ml RNase (Sigma) and 50 µg/ml propidium iodide (for the quantification of apoptotic cells). Propidium iodide fluorescence of individual cells was measured using a FACScan flow cytometer (Becton and Dickinson, Mountain View, CA). The red fluorescence due to propidium iodide staining of the DNA was registered on a logarithmic scale. In non-fixed samples propidium iodide penetrates only necrotic cells, resulting in a peak of bright red fluorescence. In fixed samples apoptotic cells give rise to a typical hypodiploid peak showing a low DNA red fluorescence. The forward- and side-scatter of particles were simultaneously measured. Cell debris was excluded from analysis by appropiately raising the forward-scatter threshold. At least 104 cells of each sample were analysed.

Assessment of cytokine production in supernatants of macrophage cultures

At days 3 and 11, the cultures were washed and refed with complete medium that may or may not contain 100 ng/ml LPS, or 50 ng/ml PMA. After 24 hr incubation, 500 µl of the supernatant from each sample were collected and stored at −80°.

Enzyme-linked immunosorbent assay (ELISA)

Commercially available sandwich ELISA kits (R & D Systems Minneapolis, MN) were used to determine the concentration of TNF-α, IL-6, IL-10 and IL-12. The detection limits of these ELISAs are 15·6 pg/ml, 3·13 pg/ml, 7·8 pg/ml and 7·8 pg/ml, respectively. According to the manufacturer's specifications, these ELISAs are specific for the relative interleukin. The sCD14 ELISA was bought from IBL, Hamburg, Germany, the detection limit of this ELISA is 5·5 ng/ml. All the samples were determined in duplicate, in a single analytical set. The intra-series variation coefficient was <15%.

FACS analysis

To determine CD14 expression, at given time-points the cells were washed twice in PBS, detached by gentle scraping, collected by centrifugation and stained for 30 min with R-phycoerythrin-Cyanine-5-conjugated (Cy) anti-CD14 monoclonal antibody (Immunotech, Marseille, France). After three washes with PBS, the cells were resuspended for 10 min in PBS containing 4% paraformaldehyde. After two additional washes in PBS, the cells were analysed by FACS.

Binding of FITC-LPS to macrophages was carried out as follows: 2 × 105 cells were incubated for 30 min at 37° with 0·5 µg/106 cells of FITC-LPS in 50 µl final volume of complete medium. After incubation the cells were washed twice in PBS, detached by gentle scraping, collected by centrifugation and stained for 30 min with R-phycoerythrin-Cyanine-5-conjugated (Cy) anti-CD14 monoclonal antibody (Immunotech). After three washes in PBS, the cells were resuspended for 10 min in PBS containing 4% paraformaldehyde. After two additional washes in PBS, the cells were analysed by FACS.

For LPS-specific cytokine responses, 1 × 106 macrophages were cultured in 24-well plates (Corning Costar Corp., Cambridge, MA) containing 2 ml of complete medium with or without GM-CSF. At different time-points, macrophages were washed and refed with complete medium containing 100 ng/ml LPS. Thirty minutes later, 1 µg/ml of the protein transport inhibitor Brefeldin A was added. Eighteen hours after Brefeldin A addition, the cells were washed twice in PBS, detached by gentle scraping, collected by centrifugation and and fixed in paraformaldehyde 4% in PBS for 5 min at 37°. After two further washes, the cells were resuspended for 30 min at room temperature in 30 µl PBS containing 0·1% saponin, 1% bovine serum albumin and 0·5 µg/106 cells of the following monoclonal antibodies: Phycoerythrin (PE) -conjugated mouse anti-IL-6 (PharMingen) and FITC mouse anti-human TNF-α. Paired isotype-specific control antibodies (PharMingen) were run with each sample. As a last step, the cells were washed twice in PBS containing 0·01% saponin. FACS analysis was performed using a FACScan flow cytometer (Becton Dickinson). Five thousand cells were computed in list mode and analysed using the FACScan research software (Becton Dickinson). Macrophages were differentiated from lymphocytes and dead cells on the basis of forward angle and 90° scatter.

Statistics

Student's _t_-test was used to analyse the data.

RESULTS

Effect of GM-CSF on macrophage viability

We first evaluated the effect of GM-CSF on macrophage viability as compared to either control cells or macrophages cultured in the presence of M-CSF (M-CSF-macrophages), a macrophage lineage-specific growth factor that, in a similar manner to GM-CSF, improves viability of cultured macrophages. Data on macrophage viability are summarized in Table 1. At day 3 of culture, the number of viable cells and the percentage of necrotic and apoptotic cells was not different between macrophages cultured with GM-CSF (GM-CSF-macrophages), M-CSF-macrophages and control cells. Conversely, at day 11, the number of viable cells ranged up to 0·4-to 1-fold higher for M-CSF-macrophages compared with GM-CSF-macrophages and to 0·2–0·4-fold higher for GM-CSF-macrophages as compared to control cells. Consistently, the percentage of necrotic cells was greater in control macrophages as compared to growth factor-treated cells. Even at day 11, no significant apoptosis was detected in both control and growth factor-treated macrophages.

Table 1.

Effect of GM-CSF and M-CSF on macrophage viability at different point in time

Growth factor Live cells Necrosis Apoptosis
Day 3
None 435 000 11 <5
GM-CSF 451 000 9 <5
M-CSF 468 000 3 <5
Day 11
None 317 000 18 <5
GM-CSF 416 000 11 <5
M-CSF 455 000 2 <5

GM-CSF modulated activation-induced cytokine production by macrophages

Macrophages were cultured in the presence or not of GM-CSF and M-CSF and stimulated at day 3 or 11 of culture with either 0·1 µg/ml LPS or 20 ng/ml PMA. Cytokine production was quantified by ELISA testing of culture supernatants. At each time-point the amount of cytokine found in the supernatants was normalized for the number of live cells (as determined by trypan blue dye analysis, see also Table 1).

At both day 3 and day 11, the LPS- and PMA-induced production of TNF-α and IL-6 by GM-CSF-macrophages was significantly higher (P <0·05) than that produced by untreated cells (Fig. 1). However, at day 11 of culture, LPS was able to induce only moderate TNF-α and IL-6 production in control cells (<10 times that induced in 3-day-old macrophages), whereas the PMA-induced cytokine production was superimposable on that observed in 3-day-old cultures. In agreement with previously reported data,15 we found that M-CSF increased the cytokine response to LPS stimulation with respect to control cells. However, at both day 3 and day 11, the TNF-α response to LPS was significantly lower in M-CSF-macrophages as compared with GM-CSF-macrophages.

Figure 1.

Figure 1

Cytokine production in 3-day-old (a) and 11-day-old (b) macrophages stimulated by LPS or PMA and cultured with or without GM-CSF and M-CSF. The amount of the different cytokines is shown as the mean of three experiments, with three different donors, each carried out in triplicate. Cytokine production has been calculated as follows: pg of cytokine per ml of supernatants/(no. of live cells in the well/1 × 105). The error bars represent the SD. No intracellular sequestration of cytokines was detected in either control or GM-CSF-macrophages by ELISA testing of cells lysed by repeated cycles of freezing and thawing (data not shown).

In the absence of LPS and PMA stimulation, the production of TNF-α and IL-6 by control cells, GM-CSF- and M-CSF-macrophages was below the detection limits of the ELISAs at either day 3 or 11 of culture (Fig. 1). Furthermore, no modulation of the spontaneous or LPS- and PMA-induced production of IL-10 and IL-12 was observed in GM-CSF-macrophages as compared with control cells (Table 2).

Table 2.

Levels in supernatants of IL-10 and IL-12 (pg/ml; mean ± SD) in control and GM-CSF macrophages stimulated or not with LPS and PMA

Control GM-CSF
Day 3 Day II Day 3 Day II
Cytokine LPS PMA LPS PMA LPS PMA LPS PMA
IL-10 39 ± 16 66 ± 29 <7·8 45 ± 31 41 ± 24 72 ± 39 <7·8 39 ± 18
IL-12 28 ± 21 36 ± 15 <7·8 25 ± 17 23 ± 16 30 ± 19 <7·8 19 ± 9

Flow cytometric assay of activation-induced macrophage cytokine response

To confirm and expand the data obtained with ELISA testing, control and GM-CSF-macrophages were analysed by FACS for LPS-induced cytokine intracellular production. Figure 2 shows that at day 3 of culture the percentage of cells with intracellular TNF-α and IL-6 was equivalent in control cells and GM-CSF-macrophages. In contrast, at day 11, the percentage of cells that stained positive for TNF-α and IL-6 after LPS stimulation was significantly higher in GM-CSF-macrophages than in control cells. Both at day 3 and day 11, GM-CSF significantly increased the cytokine intensity of fluorescence of LPS-responding macrophages (P <0·05) compared to control cells (Fig. 3). Thus, the enhancement mediated by GM-CSF in 3-day-old macrophages was restricted to the amount of cytokines produced by responding cells, whereas, in 11-day-old cultures, GM-CSF affected both the responder frequencies as well as the amount of cytokines produced by responding cells.

Figure 2.

Figure 2

Effect of GM-CSF on the frequency of LPS-responding macrophages. The figure demonstrates a representative flow cytometric analysis of control and GM-CSF-macrophages stimulated by LPS and analysed for the frequency of intracellular TNF-α and IL-6 staining. The data are displayed as dot plots. The quadrants were set according to the negative controls (less than 1% of the isotype control cells appeared positive). Five thousand cells were gated and analysed for each sample. The data refer to a typical experiment out of three performed with similar results.

Figure 3.

Figure 3

Effect of GM-CSF on the cytokine intensity of fluorescence of LPS-responding cells. The results are presented as the mean ratio ± SD of the cytokine intensity of fluorescence (arbitrary units) observed among cells responding to LPS. Data are calculated on 1000 events additionally gated on the responding fraction (i.e. the positive cells in the dot plots of Fig. 2).

As CD14 is known as a receptor for LPS we analysed whether the ability of GM-CSF to prime macrophages for an enhanced cytokine response was due to an increased expression of CD14 or not. As shown in Table 3, both the percentage of CD14-positive macrophages and the intensity of CD14 expression per cell were comparable in 3-day-old control and GM-CSF-macrophages. A spontaneous decrease of the number of CD14-positive cells and of CD14 intensity of expression was observed at day 11 in control cells, but not in GM-CSF-macrophages. CD14 and cytokine co-staining also revealed that both at day 3 and at day 11 of culture, most of the cells that showed intracellular cytokine production also stained positive for CD14 expression (data not shown).

Table 3.

Cytofluorimetric evaluation of CD14 expression in control and GM-CSF-macrophages at different time-points in culture

Day Control* GM-CSF
3 87 ± 11 (88 ± 9) 85 ± 14 (85 ± 11)
11 19 ± 7 (21 ± 3) 75 ± 11 (74 ± 18)

Effect of GM-CSF on LPS binding to macrophages

We then measured the binding of FITC-LPS to control and GM-CSF-macrophages. At day 3 of culture, the percentage macrophages able to bind FITC-LPS was >80% in either control and GM-CSF-macrophages (P >0·05). Similarly, the intensity of FITC-LPS binding was statistically not different between the two treatment groups. Indeed, the mean fluorescence intensity (arbitrary units) was 50·7 ± 7·9 for control macrophages and 49·6 ± 6·1 for GM-CSF-macrophages (Fig. 4). Conversely, at day 11 of culture GM-CSF-macrophages showed an increase in both the percentage of LPS-FITC-positive cells (87·2 ± 16·4% versus 16·3 ± 8%, P <0·05) and the intensity of FITC-LPS binding (43·8 versus 10·5, P <0·05, arbitrary units of mean fluorescence intensity), as compared to untreated cells (Fig. 4).

Figure 4.

Figure 4

Cytofluorimetric analysis of the binding of FITC-LPS on macrophages cultured with or without GM-CSF. The percentage of CD14-positive cells was calculated by straight channel integration, with the integration channel set so that less than 1% of the isotype control cells appeared positive. The histograms shown in this figure are from data obtained from a single experiment of three which had similar results. Five thousand cells were analysed for each sample. The black histograms represent the controls (aspecific, isotype-matched IgG antibodies), the white histograms represent the FITC-LPS-stained macrophages.

Effect of GM-CSF on the release of CD14 from the monocytes

Finally, we investigated whether GM-CSF blocks the cleavage of the CD14 molecule from the macrophage surface. Soluble CD14 was determined by ELISA of control and GM-CSF-macrophage supernatants at days 3 and 11 of culture (Fig. 5). Spontaneous release of soluble CD14 was observed in control macrophages. This release was significantly reduced by GM-CSF treatment (P <0·05). These findings suggest that GM-CSF-induced reduction of CD14 shedding may be responsible for the effect of GM-CSF on CD14 expression

Figure 5.

Figure 5

Release of soluble CD14 by control and GM-CSF-macrophages. Data represent the mean ± SD of three experiments with three different donors. The means of sCD14 secretion of GM-CSF-macrophages versus control cells are significantly different at days 3 and 11 (P <0·05).

DISCUSSION

The present study focuses on the effect of GM-CSF on cytokine production by macrophages. The evidence we have produced indicates that GM-CSF may modify the TNF-α and IL-6 response of macrophages to LPS by CD14-dependent and -independent mechanisms. At an early time-point in culture, GM-CSF primes macrophages for greater LPS-induced cytokine production without affecting CD14 expression and LPS-binding capacity. After prolonged incubation, GM-CSF primes macrophages for an augmented LPS response by up-regulating both CD14 expression and LPS-binding capacity. The effect of GM-CSF on cytokine production was also observed in PMA-stimulated macrophages, thus indicating that GM-CSF broadly modulates activation-induced cytokine production in these cells.

Although CD14 has been found to be the functionally most important receptor for LPS on monocytes/macrophages, other receptors, such as CD11 and other membrane structures, have been proposed to be involved in the cytokine response of macrophages to LPS.16 Therefore, we cannot exclude that LPS receptors other than CD14 may be covertly affected by GM-CSF and may mediate the priming effect of this growth factor on LPS-induced cytokine production, particularly early in culture when no modification of CD14 can be detected in GM-CSF-macrophages with respect to control cells.

Contrasting results have been reported by several investigators on the effect of GM-CSF on CD14 expression and pro-inflammatory cytokine production by macrophages. Hart et al.17 reported that long-term incubation with GM-CSF primes macrophages for an enhanced LPS-induced cytokine secretion, however, they found no change in CD14 expression. The same results, regarding CD14 expression, were obtained by two other authors.18,19 Also, down-regulation of CD14 expression after short-term exposure to GM-CSF was reported by Hogasen et al. and Kruger et al.20,21 These conflicting results can in part be explained by the different methods of cytokine detection and monocyte isolation. In this study, we have employed a recently developed flow cytometric assay capable of accurately quantifying LPS-responsive macrophages by detecting cytokine synthesis at the single-cell level.22,23 This offered a novel opportunity to study the triggering characteristics of these cells. Previous work examining the cytokine response of macrophages to LPS, although extensive, has been limited by reliance on bulk culture assay techniques which do not distinguish changes in response frequency from changes in response intensity. Moreover, bystander activation is a theoretically possible mechanism for inflating the number of responding cells in bulk culture experimental systems. This would not be expected with Brefeldin A-treated cultures and flow cytometric assay. Brefeldin A is a relatively non-toxic, but potent, inhibitor of cellular transport that efficiently prevents potentially stimulatory cytokines or cell surface adhesion molecules from being secreted or transported to the cell surface in these cultures.24 In addition, in most studies monocytes were obtained by adherence to plastic, a procedure based on contact activation of monocytes. By this method, only the most mature and activated monocyte subpopulations can be recovered. To avoid such a possible bias, we separated monocytes by elutriation, a process which provides unactivated cells that possibly represent the whole blood monocyte pool. Variable responses to LPS have been described in monocyte/macrophages from different sources, due to differences in the signal transduction machinery that mediates LPS activity.25 Therefore, it is not surprising that macrophages derived from adherence-activated monocytes may display divergent patterns of cytokine production with respect to cultures from elutriated monocytes. Interestingly, our data are more consistent with those from Kreutz et al. who used monocytes isolated by elutriation.26

In contrast to TNF-α and IL-6, neither the LPS- nor the PMA-stimulated IL-10 and IL-12 secretion was modified by GM-CSF treatment. Both IL-10 and IL-12 are known as immunoregulatory cytokines which play a different role with respect to TNF-α and IL-6 in inflammatory reactions. IL-10 may actually inhibit the production of TNF-α and IL-6,27 whereas IL-12 potentiates T-cell function and delayed-type hypersensitivity reactions.4 This dichotomic effect on pro-inflammatory and immunoregulatory cytokines suggests that GM-CSF selectively modulates macrophage activity toward an inflammatory type. In agreement with this hypothesis, a recent report showed that GM-CSF may reduce the IL-10 response of macrophages to LPS stimulation.26

GM-CSF is currently employed for the reversal of neutropenia associated with cytotoxic chemotherapy, and haematopoietic stem cell transplantation.28–30 Side-effects include fever, chills, myalgia, headache, decreased appetite and nausea. As we have just shown, GM-CSF primes macrophages for an enhanced TNF-α and IL-6 response to LPS. LPS production, in the context of Gram-negative bacterial sepsis, is a major clinical problem in patients undergoing GM-CSF therapy. Thus, induction of TNF-α and IL-6 may contribute to the adverse effects observed in patients receiving GM-CSF. With regard to this, it has been reported that monocytes from patients with Felty's syndrome treated with GM-CSF have increased secretion of TNF-α_ex vivo_.31

The data presented here also raise the possibility that GM-CSF may contribute to the pathogenesis of septic shock in vivo. Activation of cells of the monocyte/macrophage lineage by LPS, resulting in production of cytokines, particularly TNF-α and IL-6, is thought to underlie endotoxic shock during severe acute infections.32,33 GM-CSF may represent an endogenous enhancer of LPS by transforming a given dose of LPS into a lethal stimulus capable of causing sepsis. In agreement with this, it has been recently demonstrated that subcutaneous administration of GM-CSF to mice enhanced LPS-induced TNF-α and IL-6 production in the circulation.34 Furthermore, it has been found that GM-CSF potentiates TNF-α release and hepatotoxicity induced by a subtoxic dose of LPS in galactosamine-sensitized mice.3

The induction of cytokine expression in cells of the monocyte/macrophage lineage by LPS is a critical host defence response. Therefore, GM-CSF-induced enhancement of LPS activity may, paradoxically, also serve a beneficial purpose by enabling the host to respond quickly to relatively low amounts of LPS and thereby activating antibacterial defences. In this regard, GM-CSF has proved able to increase the killing of certain intracellular pathogens by macrophages by up-regulating the nitric oxide response of these cells to LPS.35,36 Furthermore, GM-CSF administration after intensive chemotherapy restored the normal responsiveness of macrophages to LPS37 and efficiently primed macrophages to respond to LPS with increased secretion of TNF-α in experimentally immunocompromised mice.38

In conclusion, our results show that GM-CSF may play a role in manipulating the activation-induced expression of pro-inflammatory cytokines by macrophages. Enhanced production of these cytokines could play an important role in the pathogenesis of Gram-negative septic shock syndrome and in defence against infectious agents.

REFERENCES