MRL/lpr and MRL^+/+ Macrophage DNA Synthesis in the Absence and the Presence of Colony-Stimulating Factor-1 and Granulocyte-Macrophage Colony-Stimulating Factor¹

The MRL-MpJ-lpr/lpr (abbreviated MRL/lpr) mouse has been widely employed as an autoimmune model of systemic lupus erythematosus and rheumatoid arthritis (1, 2). The spontaneous disease in mice presents as massive lymphadenopathy, splenomegaly, arthritis, and glomerulonephritis in addition to the presence of rheumatoid factor, circulating immune complexes, and autoantibodies to DNA and type II collagen. The lpr gene has recently been identified as a mutation in the Fas apoptosis gene and appears to promote the dysregulation of lymphocyte functions (3). However, the lpr mutation alone neither initiates nor causes murine lupus-like disease, since kidneys remain relatively normal in non-MRL background strains with the mutation (4, 5). Also, MRL^+/+ mice, which express normal Fas protein, uniformly acquire lupus-like disease, although it is relatively mild and of slow progression (3, 6). Thus it would seem that defects other than those in apoptosis are responsible for the initiation and maintenance of the lupus manifestations (6).

Most studies of the autoimmunity in MRL/lpr mice have focused on the prominent and unusual T lymphocytes. However, reports have suggested the importance of macrophages, since, for example, there are increased numbers of them in the kidney, liver, and spleen (4, 5), and extramedullary monocytopoiesis is sustained throughout life (7). Abnormalities in several functions in MRL/lpr, and sometimes in MRL^+/+, macrophages have been observed (8, 9, 10, 11, 12, 13, 14, 15). However, there are disagreements as to whether there is inherent inflammatory cytokine production (6, 12, 13, 15). There is also debate as to whether there are innate defects in the autoimmune macrophages or whether the abnormalities result from the exposure of the cells to the autoimmune milieu (4, 5, 6, 12).

During an inflammatory reaction there is increasing evidence for local proliferation of resident and/or newly arrived monocytes/macrophages contributing to their accumulation in tissues (16, 17, 18). Histopathologic examination of the arthritic joints in MRL/lpr mice revealed erosion and destruction of articular cartilage and bone largely associated with proliferating synovial cells, originally termed transformed mesenchymal cells, in a manner very similar to that described in rheumatoid arthritis lesions (2). However, these cells were later shown to have macrophage markers (19, 20).

Macrophage CSF (M-CSF)³ or CSF-1 and granulocyte-macrophage CSF (GM-CSF) are important cytokines for the development and function of cells of the monocyte/macrophage lineage (21). Evidence has been presented that in the MRL/lpr mouse CSF-1 and/or GM-CSF in the kidney are responsible for the accumulation of macrophages in this organ and can initiate renal tissue destruction (4, 5, 22, 23). Elevated circulating CSF-1 levels are also found in these mice (4). Both CSF-1 and GM-CSF have been viewed as inflammatory cytokines (24, 25, 26).

The murine peritoneum is often used as a convenient sterile site to explore the characteristics and involvement of macrophage lineage cells in the development of an inflammatory reaction. Under steady state conditions, resident peritoneal macrophages can be derived locally in the peritoneum (27, 28), and from the op/op (CSF-1-deficient) mouse, this production would appear to be dependent upon the presence of CSF-1 (29). Soon after injection of an irritant, such as thioglycollate medium (TM), predominantly neutrophils followed by macrophages appear in the cavity (30). It has been shown in many studies that elicited murine peritoneal macrophages can proliferate in vitro but require the presence of added CSF-1 (31, 32, 33).

In support of the possibility that there may be innate defects in the autoimmune macrophages (6, 8, 9, 10, 11, 12, 13, 14, 15) (see above), it has been reported that bone marrow-derived macrophages (BMM) from MRL background mice (i.e., both MRL/lpr and MRL^+/+) respond more readily to CSF-1 than BMM from control strains and that modulation of CSF-1 receptor expression is more rapid (34); it was suggested that this altered response of MRL macrophages to CSF-1 is responsible for the notable accumulation of macrophages in the kidneys, is dictated by the background, and may be a valuable predictor of the tempo of renal damage (34). In an accompanying study it was found that TNF-α enhanced CSF-1-induced BMM proliferation in MRL/lpr mice but not in other strains, including MRL^+/+ (35); based on this finding and others it was proposed that the simultaneous expression of TNF-α and CSF-1 in the MRL/lpr kidney fosters macrophage accumulation, which is responsible for the rapid tempo of autoimmune renal injury in MRL/lpr mice (35). We had shown earlier that exogenous TNF-α was one of a series of G₁ phase inhibitors of CSF-1-stimulated proliferation of BMM from other mouse strains (36, 37, 38, 39).

Given the above discussion, we examined in vitro the DNA synthesis response of MRL/lpr and MRL^+/+ peritoneal exudate macrophages and BMM to both CSF-1 and GM-CSF. Interestingly and atypically, the MRL peritoneal exudate macrophages underwent DNA synthesis in the absence of added exogenous CSF; they also responded very well to GM-CSF. Both responses appeared to be independent of the presence of endogenous CSF-1 in the cultures. In contrast to the literature, we found no obvious differences in the proliferative response to CSF-1 of MRL BMM and BMM from other strains and no obvious differences in CSF-1 receptor (or ligand) turnover; in addition, we found that exogenous TNF-α and other agents suppress CSF-1-stimulated MRL/lpr BMM DNA synthesis, as we had found previously with BMM from other mice.

Male MRL/MpJ^+/+ (MRL^+/+) and MRL/MpJ-lpr/lpr (MRL-lpr) mice were purchased from The Jackson Laboratory (Bar Harbor, ME), bred, and used between 8 and 12 wk of age. Equivalent C57BL/6, CBA, and BALB/c mice were obtained from the Walter and Eliza Hall Institute (Parkville, Australia). Other mice of the same age were from the following sources: C57BL/6, C3H/HeJ, and C3H/lpr, Walter and Eliza Hall Institute; CBA and BALB/c, Monash University (Clayton, Australia); DBA/1 (Animal Resources Center, Canning Vale, Australia), (129/OLA × C57BL/6)F₁, Ludwig Institute for Cancer Research (Parkville, Australia); (C3H × C57BL/6)F₁, J. Proietto, University of Melbourne, Department of Medicine (Parkville, Australia)); and (129/SV × C57BL/6)F₁, P. Carmeliet (Leuven, Belgium).

Peritoneal cells were washed from the peritoneal cavity of mice by lavage with 5 ml of ice-cold, sterile PBS (Trace BioSciences, Castle Hill Australia). Peritoneal exudate cells were elicited by i.p. injection of 1 ml of Brewer’s TM (Difco, Detroit, MI), and cells were harvested in the same way at 4 days after injection.

BMM were generated as adherent cells from their nonadherent progenitors in bone marrow by a protocol similar to that we have used previously (40). However, to ensure comparable initial cell numbers, nonadherent precursors, generated after 3 days from CSF-1-treated bone marrow cells in RPMI/10% heat-inactivated FBS, were plated into nontissue culture-treated plastic dishes (10-cm diameter; Disposable Products, Adelaide, Australia) at 1.2 × 10⁶ cells/dish. After a further 3 days of culture in CSF-1-containing medium (40), the adherent BMM were aspirated from the surface, plated at 7.5 × 10⁴ cells/well in 24-well tissue culture plates (Nunc, Naperville, IL) in the same medium, and allowed to adhere for another 24 h. Cells were prepared for experiments by washing twice with PBS and reculturing in growth medium in the absence of CSF-1. BMM were growth arrested for 20 h under these conditions before use (40).

DNA synthesis in peritoneal macrophages and BMM was measured as the incorporation of [methyl-³H]TdR as described previously (41). Briefly, peritoneal cells (2 × 10⁵ or 4 × 10⁵ cells/well) were allowed to adhere to a 24-well plate (Nunc) in 0.5 ml of α-MEM plus 10% FCS for 2–3 h at 37°C. Triplicate cultures were prepared for each treatment. Nonadherent cells were removed by three washes with PBS. Unless otherwise indicated, each well was replenished with 0.5 ml of α-MEM plus 10% FCS in the presence or the absence of 5,000 U/ml of human CSF-1 or 10,000 U/ml of murine GM-CSF. [³H]TdR (sp. act., 82Ci/mmol; Amersham, U.K.) was added to each culture well at a concentration of 2.5 μCi/ml. After 4–5 days of culture, uptake was stopped by washing with PBS, the cells were solubilized in 0.2 M NaOH, and the incorporation of radioactivity into TCA-precipitable material was recovered using an Inotech cell harvester (Berthold Australia, Bundoora, Australia) and measured by beta scintillation counting. A long labeling protocol was used because of the lag period before the cells enter S phase (32). For BMM, [³H]TdR incorporation was measured over different time periods as indicated.

Macrophages were collected by scraping gently, and viable cells were counted in a hemocytometer using trypan blue exclusion.

Attached cells were scraped loose using a cell scraper, resuspended in cold PBS containing 2% FCS (staining buffer), and centrifuged at 250 × g for 10 min. Cells were resuspended in 50 μl of staining buffer and left for 30 min on ice before staining. For single-color immunofluorescence a 10-min incubation with FCS at 4°C preceded the addition of 50 μl of the anti-Mac-1 mAb (American Type Culture Collection, Manassas, VA) or 5 μg/ml of the isotype (IgG) control. After 20 min at 4°C, cells were washed twice with staining buffer then incubated with 50 μl of phycoerythrin-conjugated affinity-purified F(ab₂)′ donkey anti-rat IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) for 20 min at 4°C. Stained cells were then washed, resuspended, and analyzed immediately on a FACScaliber (Becton Dickinson, Mountain View, CA).

For the two-color immunofluorescence staining, cells were prepared, incubated with the anti-Mac-1 mAb, and washed as described above, then resuspended in 100 μl of 0.5% paraformaldehyde and left for 4 h at room temperature. Following this, 25 μl of 5% Tween 20 in PBS were added, and the cells were left overnight at 4°C. After washing twice in PBS, 10 μl of DNase I (Sigma, St. Louis, MO) were added, and the cells were incubated with 50 μl of the phycoerythrin-conjugated anti-rat IgG and 10 μl of FITC-conjugated anti-BrdUrd mAb (Becton Dickinson, Mountain View, CA). Included as part of the controls for immunofluorescence staining were cells grown in the absence of BrdUrd (see above). After incubation for 60 min at room temperature, cells were analyzed as described above.

Macrophage monolayers were washed with staining buffer, then preincubated for 30 min at 37°C with 1 ml of the same solution. The fluorescent latex Fluospheres (L-5281 carboxylate-modified, 1.0-μm diameter, 2% solids; Molecular Probes, Eugene, OR) were first precoated with 1% BSA before 10 μl of the 0.5% BSA-coated bead solution was added to the cells. The macrophages were then incubated with latex beads in the solution for 60 min at room temperature. Control cultures with no addition of latex beads were also included. After washing five times with cold PBS to eliminate uningested freely suspended particles, the medium was changed to 1 ml of PBS containing 0.25% trypsin, and the cells were incubated for 2.5 h at 37°C. Incubation with trypsin has been reported to remove uningested particles attached to the macrophage membrane surface (42). The cells were fixed with the addition of 1 ml of 1% glutaraldehyde (0.5% final concentration) and were left overnight at 4°C. Following this the cells were scraped as described above, centrifuged at 250 × g for 20 min, and resuspended in PBS before analysis by flow cytometry.

For the measurement of surface-bound CSF-1 receptor, quiescent BMM were prepared as described above, but plated at 3 × 10⁵ cells/well in six-well plates (Falcon, Becton Dickinson Labware, Mountain View, CA). CSF-1 was iodinated using chloramine-T as described previously (43) to between 1 and 2 mol of bound iodine/mol and was used within 2–3 days. The growth-arrested cells were incubated with CSF-1 (5000 U/ml) for 1 or 24 h at 37°C, and surface receptor numbers were measured as follows. Cultures were cooled rapidly on ice, and surface-bound CSF-1 was removed by a 60-s wash with PBS, pH 4. Triplicate wells were incubated at 4°C for 30 min in binding buffer (α-MEM, 0.02% NaN₃, and 0.2% BSA) with 0.15 nM ¹²⁵I-CSF-1 (2.2 × 10⁸ cpm/μg) in the presence or the absence of 50-fold excess of unlabeled CSF-1. After incubation, unbound CSF-1 was removed by washing three times in ice-cold binding buffer and three times in PBS; bound CSF-1 was estimated using a United Technologies Packard Crystal gamma counter following solubilization in 1.5 ml of 50 mM Tris (pH 7.5) and 0.5% SDS.

Quiescent cells (∼2 × 10⁶) were treated as described in the text for 18 h, and total RNA was prepared by a published method (44) and analyzed by Northern blotting. Equal loading and transfer of RNA were confirmed by UV illumination of both gels and membranes following ethidium bromide staining. Probing was with a ³²P-labeled c-fms cDNA probe followed by membrane exposure to Kodak XAR autoradiographic film (Kodak, Rochester, NY) at −70°C with intensifying screens (36).

Internalization during a single uptake of ¹²⁵I-CSF-1 was followed in a temperature-shift experiment as described (45). Quiescent BMM, cultured to a density of 1 × 10⁶ cells/well, were cooled on ice for 30 min. Cultures were then incubated at 4°C with ¹²⁵I-CSF-1 (300 pM) for a further 60 min. Unbound ¹²⁵I-CSF-1 was removed by washing in ice-cold RPMI/1% heat-inactivated RPMI/FBS and then the cultures were shifted to 37°C by the addition of 1 ml of prewarmed medium for the times indicated. The uptake of ligand was stopped at the times indicated by transfer of cultures to 4°C and washing twice in 2 ml of buffer (α-MEM, 0.2% BSA, and 0.02% sodium azide). Receptor-bound ¹²⁵I-CSF-1 was determined as radioactivity released following incubation of cells at pH 4 for 60 s. The internalized ¹²⁵I-CSF-1 was determined following solubilization in 1.5 ml of lysis buffer (50 mM Tris-Cl, pH 7.5, and 0.5% SDS). Cells at the zero time point were maintained at 4°C following the removal of unbound ¹²⁵I-CSF-1, and then cell surface receptor-associated ¹²⁵I-CSF-1 was measured. The internalized ligand at a given time point is represented as a percentage of the total ¹²⁵I-CSF-1 present at the time.

Degradation of ¹²⁵I-CSF-1 was determined as described previously (45). Quiescent BMM were cooled on ice, and ¹²⁵I-labeled CSF-1 (100 pM) was allowed to bind for 60 min. Following the removal of unbound ligand, cells were replaced with medium at 37°C. The degradation of ¹²⁵I-CSF-1 was estimated by counting the TCA-soluble fraction of the medium. Cells at the zero time point were replaced with medium at 4°C, and the radioactivity in the lysate was estimated to determine the total surface-associated counts before internalization. The degradation is represented as a proportion of the initial surface-associated ¹²⁵I-CSF-1.

The following were obtained as gifts: recombinant human CSF-1 (sp. act., 7 × 10⁷ U/mg; Chiron, Emeryville, CA); recombinant murine GM-CSF (sp. act., 2 × 10⁷ units/mg) (Sandoz, Vienna, Austria); recombinant murine TNF-α (sp. act., 1.2 × 10⁷ U/mg (Boehringer Ingelheim, Sydney, Australia); neutralizing anti-murine c-Fms hybridoma (AFS-98) (S. I. Nishikawa, Kyoto, Japan) (46); c-fms cDNA (D. Hume, Brisbane, Australia). The following were purchased: α-MEM (Trace BioSciences), RPMI (ICN-Flow Laboratories, Sydney, Australia), and FCS (CSL, Parkville, Australia).

All other reagents were analytical grade. All practical precautions for minimizing endotoxin contamination were taken. Solutions were made in pyrogen-free water (Delta West, Bentley, Western Australia) and endotoxin levels were routinely monitored by Limulus lysate tests (CSL, Parkville, Australia), with the minimum detectable level being 0.01 ng/ml.

In the steady state the numbers of peritoneal cells (∼6 × 10⁶/mouse) in the peritoneal cavities of MRL mice (i.e., MRL/lpr and MRL^+/+ mice) were similar to those in matched C57BL/6 mice. Many studies with several mouse strains have shown that only a low proportion of resident peritoneal macrophages from an unstimulated peritoneal cavity can proliferate in vitro in response to CSF-1 (32, 41). We found in several experiments (n = 10) that in the absence of CSF-1, both MRL/lpr and C57BL/6 resident peritoneal macrophages had low [³H]TdR incorporation over a 5-day culture period, as a measure of DNA synthesis (data not shown); both cell populations had similar responses to CSF-1 (5,000 U/ml). They also had weak and similar responses to GM-CSF (10,000 U/ml, which was approximately 10% of the value of the CSF-1 response; data not shown).

Injection of TM gave rise to increased numbers of MRL peritoneal cells (maximum, 15–20 × 10⁶ cells/mouse) at the 96 h examination, which were lower than the maximum numbers obtainable from the injected C57BL/6 cavity (20–40 × 10⁶ cells/mouse). Many studies with several mouse strains have shown that peritoneal exudate macrophages, usually elicited by TM injection, require CSF-1 for proliferation in vitro (31, 32, 33, 47, 48, 49). However, it is shown in Table I for a representative experiment that both MRL/lpr and MRL^+/+ adherent peritoneal cells (i.e., macrophages; see below) from a 4-day TM-induced exudate underwent a relatively high level of DNA synthesis in the absence of added CSF-1 compared with the equivalent C57BL/6 response (12- to 13-fold increase). As expected (31, 32, 33, 47, 48, 49), the exudate macrophages from each strain responded quite well to added CSF-1. The number of adherent cells at the start of the experiment was comparable for each mouse strain. The effect of GM-CSF on peritoneal macrophage proliferation in vitro has not been widely examined, although in one report with C3H/HeJ TM-elicited peritoneal macrophages it appeared to result in less proliferation than CSF-1 (50). It can also be seen in Table I that the extent of [³H]TdR incorporation in response to GM-CSF was much higher for the MRL peritoneal cells than for the C57BL/6 counterparts.

The data in Table I for the exudate cells were confirmed in several experiments (n = 13), and no significant differences were observed among the various responses for cells from each MRL strain. Also, it was found, in general, that the extent of [³H]TdR incorporation in response to CSF-1 was similar in the three mouse strains. If the data from the several experiments (n = 13) were calculated with the CSF-1-stimulated [³H]TdR incorporation arbitrarily given a value of 100% in an individual experiment, it was found for the TM-elicited MRL/lpr macrophages that the mean ± SEM relative percentage was 34 ± 9% in the absence of CSF-1 and was 138 ± 11% in the presence of GM-CSF; the corresponding values for the C57BL/6 macrophages were 3 ± 0.3 and 32 ± 6%, respectively. When the absolute proportion of cells entering S phase was examined by autoradiography, the results paralleled the [³H]TdR incorporation data, with 17–40% of the non-CSF-treated MRL macrophages (4-day exudate) registering positive over a 5-day period in vitro; the corresponding percentage for the C57BL/6 macrophages was <5% (data not shown). For the same MRL macrophages and with the same assay, CSF-1 and GM-CSF both stimulated 50–70% of the macrophages to enter the cell cycle. The above data were confirmed by flow cytometry and BrdUrd incorporation (see below). We found that the MRL cells that responded in the absence of added CSF appeared in the cavity within 16 h after TM injection (data not shown).

We then examined the respective DNA synthesis responses of peritoneal exudate macrophages (4-day, TM-induced exudate) from a number of mouse strains to gain some idea of the generality or otherwise of our observations. Macrophages from the following strains were analyzed: CBA, BALB/c, DBA/1, C3H/HeJ, (129/OLA × C57BL/6)F₁, (C3H × C57BL/6)F₁, and (129/SV × C57BL/6)F₁. As expected from the literature (31, 32, 33, 47, 48, 49), the DNA synthesis levels in the absence of CSF were uniformly low and similar to those for C57BL/6 macrophages, as were the dramatic responses to CSF-1; if calculated as a proportion of the CSF-1 responses, they were always <5% (data not shown). The responses to GM-CSF varied more than the CSF-1 responses, with the range being 50–100% of the latter values (data not shown).

For the above experiments the macrophage cultures always contained FCS. We therefore determined in additional experiments the serum dependence of the high basal response of the MRL/lpr macrophages. We see in Fig. 1 that, in fact, 5% FCS resulted in an even higher stimulation than 10% FCS, with responsiveness even occurring in serum-free medium.

As mentioned, we assumed above that the MRL-adherent peritoneal exudate cells responding abnormally in the absence of CSF-1 were macrophage-type cells and not another cell type. However, it was thought necessary to provide evidence for this. The adherence, size, and morphologic appearance of the cell populations in culture were macrophage-like. Also, 50–70% of the MRL-adherent exudate cells generally entered cell cycle in response to the macrophage-specific growth factor, CSF-1, and in some experiments 40% entered cell cycle in the absence of added CSF (see above). Further evidence supporting the macrophage nature of the MRL exudate cells proliferating in the absence of CSF was obtained by investigating immunoreactivity to Mac-1. Immunohistochemistry (data not shown) and flow cytometry revealed that >90% of the adherent peritoneal exudate cells were positive for the Mac-1 Ag both before and after 4 days of culture. Flow cytometric data for a MRL/lpr peritoneal exudate culture (4-day culture) are given in Fig. 2,A. Data were similar for the three mouse strains examined (MRL/lpr, MRL^+/+, and C57BL/6) and were constant over the three culture conditions, i.e., in the absence or the presence of either CSF. In addition, >90% of the cultured MRL/lpr and C57BL/6 adherent peritoneal exudate cells under all conditions were able to phagocytose FITC-labeled latex beads, confirming their macrophage nature (data for the MRL/lpr cells are presented in Fig. 2 B).

The analysis of BrdUrd⁺ cells in the peritoneal exudate cultures confirmed the heightened responses observed in the MRL cultures in the absence of added CSF shown earlier (Table I), and two-color immunofluorescence analysis revealed that the majority (≥80%) of cells undergoing DNA synthesis expressed Mac-1 (Fig. 2,C); this proportion of BrdUrd⁺ cells that was also Mac-1⁺ was confirmed in a total of five experiments. There appeared to be no difference in the intensity of Mac-1 expression when comparing Mac-1⁺BrdUrd⁻ and Mac-1⁺BrdUrd⁺ subpopulations (Fig. 2 C), indicating that Mac-1 expression by adherent peritoneal exudate macrophages is not altered during the cell cycle.

It is possible that endogenous CSF-1 production in the MRL peritoneal exudate cultures may be responsible for the enhanced proliferative response of the macrophages in the absence of added CSF and the relatively high response to GM-CSF. In this context, it has been reported that GM-CSF can induce CSF-1 in monocyte cultures (51) and can synergize with low concentrations of exogenous CSF-1 to stimulate murine macrophage proliferation (50). To investigate this issue, the proliferative responses were conducted in the presence of blocking Ab to c-Fms (CSF-1 receptor) (46). As shown in Table II, this Ab effectively blocked the DNA synthesis response to CSF-1 in the MRL/lpr macrophages, reducing the response to that obtained in FCS alone. However, in the experiment for which data are presented, the responses in the absence of a CSF and also to GM-CSF were not significantly reduced (in some experiments the response to GM-CSF was lowered by ∼10%). As a confirmation of these findings, when BrdUrd⁺ incorporation and flow cytometry were used to monitor DNA synthesis (as in Fig. 2 C), anti-c-Fms Ab had no effect on the percentage of BrdUrd⁺ cells in the FCS-containing and GM-CSF-treated cultures and again reduced the CSF-1 response to the basal level seen in its absence (data not shown). No discernible change in Mac-1 expression was noted. Assuming that the action of any endogenous CSF-1 with its receptor would also be susceptible to inhibition, it would appear that this mechanism is not the major one operating to account for the heightened response of MRL macrophages in the absence of added CSF as well as the degree of the response to GM-CSF; however, these negative data cannot be considered conclusive, and additional approaches are needed to establish the mechanism.

The altered regulation of DNA synthesis reported above for MRL peritoneal exudate macrophages could be inherent to the macrophage lineage cells themselves and/or could be due to an altered in vivo milieu (4, 5, 6, 12). One way to examine these possibilities is to exploit BMM that are derived in vitro from bone marrow progenitors after culture in CSF-1 (40). This approach has been used previously to examine whether there were any inherent abnormalities in both motheaten (52) and MRL (34, 35) macrophages. Unlike TM-elicited peritoneal macrophages, BMM from many mouse strains have been shown in several studies to require the presence of the exogenous CSF-1 for survival; they enter S phase rapidly (10–12 h) after CSF-1 addition to noncycling cells (53, 54). We determined whether quiescent MRL BMM, namely cells from which the CSF-1 used for their outgrowth from progenitor cells had been previously removed for 20 h (see Materials and Methods), would proliferate in the absence of added CSF. We found that these MRL/lpr BMM incorporated more [³H]TdR, albeit at a low level, than the corresponding C57BL/6 BMM over the next 22 h (data not shown); the BMM numbers at the time of the [³H]TdR addition were similar for the two groups. However, when the [³H]TdR pulse was delayed a further 20 h, the degree of DNA synthesis decreased abruptly (the cells begin to die; see below). MRL^+/+ BMM showed similar [³H]TdR incorporation as MRL/lpr BMM, while the values for the BMM from several strains (C3H/HeJ, C3H/lpr, CBA, (129/OLA × C57BL/6)F₁, (C3H × C57BL/6)F₁, and (129/SV × C57BL/6)F₁) were comparable to those for the C57BL/6 BMM (data not shown). These data indicate that MRL macrophages, which are derived in vitro from precursors, for some reason stay in cycle longer once CSF-1 is removed; however, this difference is quite transient.

Since it has been reported that the spontaneous apoptosis in many cells (3), including human monocytes and macrophages in vitro (55), is mediated by the interaction of Fas and Fas ligand, that the lpr mutation in Fas can reverse this (1), and that hemopoietic cells (56), including BMM (53),⁴ appear to die by apoptosis in the absence of growth factors, we tested whether MRL/lpr BMM survival upon CSF-1 removal was relatively enhanced. We see in Fig. 3 that MRL/lpr BMM viable cell numbers under the particular culture conditions used are not maintained any better than C3H/HeJ BMM after CSF-1 removal over the period examined. Similar findings were made when MRL^+/+, C3H/HeJ, and C3H/lpr BMM were compared, with no obvious enhancement attributable to the lpr mutation (data not shown).

It should be noted in Fig. 3 that the proliferative response to an optimal CSF-1 concentration, as judged by cell number, was no different for MRL/lpr BMM compared with C3H/HeJ BMM. Since Moore et al. (34) reported that both MRL/lpr and MRL^+/+ BMM showed enhanced [³H]TdR incorporation compared with C3H/HeJ BMM at all CSF-1 doses tested (but without a shift in the dose-response curve), we decided to examine this question in more detail. In Fig. 4 A we present a CSF-1 dose response for MRL/lpr and C3H/HeJ BMM under conditions similar to those used previously (34), namely by plating a constant number of BMM, removing the CSF-1 for a period, and treating them with CSF-1 for 48 h before an 18-h pulse with [³H]TdR. As can be seen, we also found a similar dose-response curve for the two BMM populations but, contrary to what has been reported (34), found no evidence for any enhanced DNA synthesis for the MRL/lpr BMM. Under these conditions MRL^+/+ BMM did not show any enhanced response either, and C57BL/6 BMM gave similar results in four experiments (data not shown). The protocol that we usually use to study CSF-1-stimulated BMM DNA synthesis is to treat quiescent BMM, grown previously in the same 24-well plates, with CSF-1 for 20 h and to pulse with [³H]TdR for 2 h, i.e., by a protocol designed to measure cells entering S phase for the first time after addition of CSF-1 (53, 54). We found in a total of 11 experiments that MRL/lpr, MRL^+/+, C3H/HeJ, and C3H/lpr BMM showed similar CSF-1 dose responses under these conditions as well (data not shown).

It was also reported (34) that a kinetic analysis revealed that enhanced [³H]TdR incorporation to an optimal CSF-1 dose was consistently increased 1.5- to 3-fold in both MRL/lpr and MRL^+/+ BMM vs C3H/HeJ BMM at 48 and 72 h after stimulation; in contrast, at 24 h, the responses to CSF-1 were similar in three BMM populations (34). In Fig. 4 B we present such a kinetic analysis for MRL/lpr, MRL^+/+, and C3H/HeJ BMM, again conducted with a similar protocol (34), i.e., 18-h pulses with a constant number of quiescent BMM, treated with optimal CSF-1 for different time periods. Again, our findings differ, in that we found no enhanced [³H]TdR incorporation into MRL BMM at any of the time points. If we administered 2-h pulses after 24, 48, and 72 h of CSF-1 treatment, no differences were noted in the rate of [³H]TdR incorporation for the three BMM populations or for C3H/lpr BMM (data not shown).

To explain the increased sensitivity to CSF-1 of their MRL BMM, Moore et al. (34) provided evidence that the down-modulation of CSF-1 receptor after MRL BMM exposure to CSF-1 was restored by 18 h, in contrast to control BMM exposure, at both the protein and mRNA levels; they therefore concluded that MRL BMM would have more CSF-1 receptors available to bind CSF-1, which may be responsible for the enhanced proliferation in response to CSF-1. In Table III, we indicate that in our hands there was no difference in the down-modulation of surface receptor number following additions of ligand to MRL BMM (i.e., MRL/lpr and MRL^+/+ BMM) compared with C3H/HeJ BMM, with no evidence of recovery of surface receptor numbers by 20 h. This uniform down-modulation of CSF-1 receptor (c-Fms) protein was found also at the level of mRNA by Northern analysis, since c-fms mRNA expression after 18-h treatment with CSF-1 was no higher in MRL/lpr BMM than in C3H/HeJ BMM at the 18 h point (Fig. 5); the protocol adopted for the experiment for which data are presented in Fig. 5 was previously published (34). In addition to the lack of difference in the degree of disappearance and reappearance of ligand binding sites over the 20-h period, we illustrate in Fig. 6,A that the rate at which surface-bound CSF-1 is internalized is similar for MRL/lpr, MRL^+/+, C3H/HeJ, and C3H/lpr BMM; likewise, there is also a similar rate at which surface-bound CSF-1 is degraded for MRL/lpr, MRL^+/+, and C3H/HeJ BMM (Fig. 6 B).

In studies mostly with BMM from CBA or C3H/HeJ mice we showed previously that a number of agents, including TNF-α, suppressed CSF-1-stimulated DNA synthesis (36, 37, 38, 39). However, it was recently reported that TNF-α enhanced CSF-1-stimulated DNA synthesis specifically in MRL/lpr BMM but not in BMM from MRL^+/+ mice and other strains (35). Since the in vitro conditions adopted for the BMM studies reported previously (35) differed slightly from those we had used before (36, 37, 38, 39), we tested the effects of TNF-α on MRL/lpr BMM under the conditions employed previously (35), namely a 48-h culture followed by an 18-h [³H]TdR incorporation period in the presence of an optimal CSF-1 concentration. In Table IV we demonstrate that TNF-α (100 and 1000 U/ml) did not enhance the CSF-1-stimulated [³H]TdR incorporation in MRL/lpr BMM. However, at 1000 U/ml it suppressed the CSF-1-stimulated DNA synthesis in MRL/lpr and C57BL/6 BMM (Table IV) and in MRL^+/+ BMM (data not shown). We also did not find any potentiation of the DNA synthesis response to suboptimal CSF-1 concentrations (160 and 320 U/ml). If the effect of TNF-α was monitored in a 20-h experiment with a 2-h [³H]TdR pulse, the conditions under which we had previously analyzed the suppressive effects on BMM DNA synthesis (36, 37, 38, 39), then TNF-α (1000 U/ml) was again inhibitory for these BMM (data not shown). TNF-α-mediated suppression also occurred with C3H/lpr and C3H/HeJ BMM under all conditions. We also found that doses of TNF-α even as low as 1 U/ml did not have a stimulatory effect. Other G₁ phase inhibitors of CSF-1-stimulated BMM DNA synthesis, namely LPS, IFN-γ, IFN-α, and 8-bromo-cAMP (36, 37, 38, 39), again suppressed CSF-1-stimulated MRL BMM DNA synthesis in 20-h cultures (data not shown).

We have used the inflamed peritoneal cavity as a model system to study the properties of MRL macrophages that appear at a site of inflammation. Numerous studies with mice from diverse strains have shown that the majority of elicited peritoneal macrophages, often in response to TM injection, can proliferate in vitro but require a source of CSF-1 to do so; in other words, FCS is insufficient (31, 32, 33, 47, 48, 49). We confirmed these earlier findings. In contrast, we report that many TM-elicited MRL (both MRL/lpr and MRL^+/+) macrophages are unusual in that they do not require the presence of exogenous CSF-1 to commence DNA synthesis. This abnormality is independent of the lpr mutation, since both strains behaved similarly. The macrophage-like nature of the proliferating cells was confirmed by their morphology, adherence, ability to phagocytose, and Mac-1 expression. The reason(s) for this difference in MRL macrophages is unknown. The lack of inhibition of the basal DNA synthesis by anti-CSF-1 receptor Ab, although not a conclusive finding, is consistent with this difference in the MRL macrophages not being due to endogenously produced CSF-1. The adhered MRL peritoneal macrophages may possess some inherent change (for example, a dysregulated signal transduction pathway(s)), since FCS was not absolutely necessary but potentiated the serum-free response; however, it is also possible that another mitogen is being elaborated in the cultures. The relationship and significance of the small basal and transient DNA synthesis in the in vitro-derived MRL BMM are presently unknown.

Our findings with the peritoneal exudate macrophages may help to explain the increased macrophage accumulation in a number of organs in MRL mice, probably present there as part of an autoimmune (inflammatory) reaction. It should also be noted that there are numerous examples in the literature where macrophages have been shown to proliferate as part of an inflammatory response (see, e.g., Refs. 16–18); in the particular case of the MRL/lpr mouse, proliferating synovial joint cells with macrophage markers have been observed (2, 19, 20). Evidence has been presented for the reasonable conclusion that in the MRL/lpr mouse CSF-1-dependent proliferation in the kidney is responsible for the accumulation of macrophages in this organ and the initiation of renal tissue destruction (4, 5, 22, 23). From our findings it could mean that CSF-1 may not be needed for the above in vivo findings with the MRL/lpr mouse. A consequence of macrophage accumulation as a result of local proliferation would be more cells with increased potential for inflammatory mediator production (e.g., cytokines), which may contribute to the autoimmune milieu.

We also showed that more MRL exudate macrophages also responded in vitro at least equally as well to GM-CSF as to CSF-1; the CSF-1 response was similar to that for macrophages from the other mouse strains examined, while the GM-CSF response was better than that for macrophages from some of these other strains. The effects of GM-CSF on the proliferation of murine peritoneal macrophages have not been widely studied, although it appeared in one report for C3H/HeJ peritoneal macrophages from 3-day TM-induced exudates that GM-CSF was a weaker mitogen than CSF-1 (50). The DNA synthesis response of MRL peritoneal exudate macrophages to GM-CSF did not appear to be dependent on the presence of endogenous CSF-1; it should be noted that GM-CSF can induce CSF-1 in human monocyte cultures (51). Of potential relevance, Müller et al. (7) reported an increased DNA synthesis response to GM-CSF of nonadherent cells from the spleen, bone marrow, and liver of MRL/lpr mice compared with the response of C3H/HeN cells. MRL^+/+ mice were not examined. GM-CSF has been considered by some, including us, to be a proinflammatory cytokine (24, 25, 26) and has been found at sites of inflammation (25). GM-CSF has been considered to have a role in MRL/lpr pathology, since gene transfer of GM-CSF into the MRL/lpr kidney initiates severe local renal injury (22). Our findings could also mean that the accumulation and proliferation of macrophages in MRL tissues mentioned previously (2, 4, 5, 19, 20, 22, 23) may be related to their responsiveness to GM-CSF.

Unlike BMM from motheaten mice (52), we could not find any enhanced proliferative response of MRL (MRL/lpr and MRL^+/+) BMM to CSF-1 at any CSF-1 concentration when monitored by [³H]TdR incorporation under various conditions or by cell number. In other words, we could not confirm the data of Moore et al. (34) with MRL BMM even though we have used similar experimental conditions as part of our study. We also could not confirm that CSF-1 receptor expression on MRL BMM returned more rapidly to baseline levels after down-regulation by CSF-1 at the level of either mRNA or protein (34). The reasons for the discrepancies are unclear. Of possible relevance, we have reported before that variations in endogenous IFN-α/β levels in CSF-1-treated BMM cultures from experiment to experiment can dramatically influence the degree of the proliferative response even for cells of the same mouse strain and can lead to inconsistent CSF-1 proliferative responses (57); also, BMM use CSF-1 rather rapidly (45), and to avoid this potential artifact we ensured in our studies that CSF-1 levels were maintained to sustain the proliferative response. From our data, at least, an enhanced in vitro proliferative response of MRL BMM to CSF-1 cannot be used to explain the accumulation of macrophages in the kidneys of MRL/lpr mice, as has been previously suggested (34). We also measured the rate of CSF-1 internalization and degradation by MRL BMM and did not find any differences with respect to these parameters for other BMM.

Interestingly, we did not find any differences in the survival of MRL/lpr BMM upon removal of CSF-1 under our particular culture conditions (Fig. 3). Since hemopoietic cells (56), including BMM (53) (see Footnote 4), appear to die by apoptosis in the absence of growth factors, our data suggest that there is no inherent deficiency in the apoptotic rate of CSF-1-starved MRL/lpr BMM compared with those in other strains. Since from these results the lpr mutation does not enhance survival, the Fas pathway would not appear to be a major pathway involved in cell death in this experimental system, in contrast to what has been reported for human monocytes and macrophages (55).

We also reported above for MRL mice, as we have found previously using BMM prepared from other mouse strains (36, 37, 38, 39), that TNF-α inhibited CSF-1-stimulated BMM DNA synthesis over a wide range of conditions, an observation that is not in agreement with a recent report (35). The reason for this discrepancy is again unclear. We have indicated before that the inhibitory effect of exogenous TNF-α on CSF-1-stimulated BMM DNA synthesis was to a significant extent due to the action of endogenous IFN-αβ production in response to this cytokine (57). The enhanced stimulation by TNF-α of the CSF-1-induced MRL/lpr BMM proliferation in vitro has been used as a model system to suggest that simultaneous expression of TNF-α and CSF-1 in the MRL/lpr kidney fosters macrophage accumulation (35). Our in vitro data suggest that MRL/lpr BMM cannot be used to support this concept. Therefore, at least by the criteria examined, namely CSF-1 responsiveness, CSF-1 receptor dynamics, TNF-α sensitivity of the CSF-1 proliferative response, and apoptotic rate in the absence of CSF-1, MRL BMM, which are an in vitro-derived cell population, do not appear to be inherently abnormal. We propose that other model systems besides CSF-1-stimulated BMM are therefore needed to explain the altered properties of MRL/lpr macrophages (6, 8, 9, 10, 11, 12, 13, 14, 15) and to establish the putative role of CSF-1 in MRL/lpr disease (4, 5).

In the studies reported here we found no obvious differences between the MRL/lpr and MRL^+/+ macrophages; as mentioned in the introduction, MRL^+/+ mice, which express intact Fas and Fas ligand proteins, uniformly acquire lupus-like disease, although one that is relatively mild and slow to progress (3, 6). These and other findings have indicated that defective events other than apoptosis are responsible for the initiation and maintenance of lupus (6). Some macrophage abnormalities have been reported to be specific for the MRL/lpr strain, i.e., absent from the MRL^+/+ strain, while others have been found to be shared between MRL strains (6, 8, 9, 10, 11, 12, 13, 14, 15); the enhanced DNA synthesis responses of MRL peritoneal exudate macrophages in the absence of an added CSF described above belong to the latter category. These unusual proliferative responses of the elicited MRL macrophages presumably reflect differences in the circulating macrophage lineage cells (monocytes) that enter the inflamed site and, in turn, may reflect changes in hemopoietic organs such as spleen and bone marrow. There is evidence that a postnatal expansion of macrophage precursors may be a characteristic of murine systemic lupus erythematosus in MRL/lpr, NZB/NZW, and BXSB mice (7, 58). Our findings may be reflecting these differences; as discussed, whether they reflect inherent cellular changes or are a result of cell exposure to the autoimmune milieu await clarification (4, 5, 6, 12). Since MRL pathology increases with age, studies to explore the peritoneal macrophage changes as MRL mice develop would be of interest as would be analogous studies with other autoimmune mice.

We thank R. Sallay for typing the manuscript.