INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

According to the inflammatory hypothesis of Alzheimer's disease (AD),¹ chronic cerebral inflammation results in injury to neurons, contributing over time to cognitive decline. Neuronal injury is hypothesized to result from the direct effects of inflammatory effectors such as cytokines and activated complement, or indirect effects such as increased production of neurotoxic reactive oxygen and nitrogen species in response to cytokines or other inflammatory stimuli (1, 2). This hypothesis is supported by epidemiological studies, which indicate that anti-inflammatory medications may protect against AD (3-6). In the present study, we demonstrate that macrophage colony-stimulating factor (M-CSF), a cytokine which is increased in the central nervous system in AD (7), dramatically augments -amyloid peptide (AP)-induced production of pro-inflammatory interleukin-1 (IL-1), interleukin-6 (IL-6), and nitric oxide (NO) by microglial cells.

Microglia are likely to have a pivotal role in inflammatory neuronal injury in AD. As intrinsic immune effector cells of the brain, microglia are potent mediators of cerebral inflammation in a variety of disease states (8, 9). AP induces cultured microglia to produce agents with the potential to directly or indirectly injure neurons, including inflammatory and chemotactic cytokines (10, 11), nitric oxide (12-14), and reactive oxygen species (12, 15). However, previously reported AP-induced increases in microglial production of these factors have been of limited magnitude, on the order of only 2-5-fold greater than control levels. It is difficult to reconcile this weak in vitro microglial response to AP with the hypothesis that AP activation of microglia is important in AD pathophysiology.

One reason for the limited response of cultured microglia to AP may be that important costimulatory agents present in AD brain have not been taken into consideration in prior reports. The extracellular environment surrounding neuritic plaques in AD brain is rich in a variety of pro-inflammatory agents including cytokines (2), which are likely to augment the effects of AP on microglia. It has been shown that interferon-, phorbol ester, and lipopolysaccharide all augment the effects of AP on microglia and monocytic cells (13-16). However, none of these augmenting stimuli has a physiologic role in AD. Results showing large synergistic increases in AP-induced microglial activity in cultures cotreated with these agents may have no direct relevance to AD.

We examined the effect of M-CSF (also called colony-stimulating factor 1 (CSF-1)) on AP-induced cytokine and NO production by cultured microglia. M-CSF is an important regulator of mononuclear phagocyte development and function throughout the body (17). In the brain, M-CSF is expressed by neurons, astrocytes, and endothelial cells (7, 18-22), where it induces proliferation, migration, and activation of microglia (23-26). M-CSF treatment of microglia also induces increased expression of macrophage scavenger receptors (7), which mediate microglial interactions with AP (27, 28). AP binds to neuronal receptors for advanced glycation end products to increase neuronal M-CSF expression (7), which causes further microglial activation. In AD brain, there is increased immunoreactivity for the M-CSF receptor (c-fms) on microglia (29), neurons in AD show labeling with M-CSF antibodies, and M-CSF levels in AD cerebrospinal fluid are 5-fold greater than in controls (7). Thus, M-CSF represents a potent microglial activator relevant to AD pathophysiology.

We hypothesized that in AD, M-CSF activates microglia to augment AP-induced production of inflammatory cytokines and NO, which in turn promote additional inflammation and may directly injure nerve cells. To test this hypothesis, we examined the effects of combined M-CSF and AP treatment on production of interleukin-1, interleukin-6, and NO by the BV-2 immortalized murine microglial cell line.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

-Amyloid Peptides-- Synthetic AP 1-40 and AP 40-1 were purchased from Bachem California (Torrance, CA). Peptides were aggregated by resuspending at 2 mg/ml in endotoxin-free water (Sigma), holding at 4 °C for 60 h, incubating at 37 °C for 8 h with gentle mixing every 2 h, and then storing at 4 °C until use.

Cytokines-- Recombinant mouse M-CSF, recombinant mouse granulocyte macrophage-CSF (GM-CSF), and recombinant mouse interleukin-3 (IL-3) were purchased from R & D (Minneapolis, MN). Cytokines were resuspended in sterile tissue culture-grade phosphate-buffered saline (Sigma) with 0.1% tissue culture-grade bovine serum albumin (Sigma), aliquoted, and stored at 80 °C until ready for use.

Cell Culture-- The BV-2 immortalized microglial cell line was cultured as described previously (30). BV-2 cells were detached from the substrate by gentle pipetting and reseeded at 1 × 10⁵ cells in 500 µl of medium per well in a 48-well tissue culture dish. After an additional 24 h in culture, cells were used for experimentation by washing two times in serum-free medium and then applying fresh serum-free medium containing AP and/or M-CSF. All treatments were performed for 24 h, after which conditioned media were collected, centrifuged at 600 × g at 4 °C for 10 min, and stored at 80 °C until ready for analysis. Each experiment included triplicate cultures for each treatment condition, and each experiment was replicated on separate occasions a minimum of two additional times.

Cell Counting-- After harvesting conditioned medium for cytokine or NO assays, BV-2 cells were detached by trypsinization and resuspended in fresh medium. Aliquots of cells from each well were counted three times in a hemocytometer using trypan blue exclusion, and counts were averaged. For each treatment condition, triplicate wells were counted, and values were averaged. All cytokine and NO results were adjusted for the number of viable BV-2 cells present for each treatment condition.

IL-1 and IL-6 ELISA-- Mouse IL-1 and IL-6 in conditioned media were determined using ELISA kits according to the manufacturer's instructions (Endogen, Woburn MA). Each sample was assayed in duplicate, and values from duplicates were averaged. Means for each treatment condition were calculated, along with standard errors. To increase signal intensity, poly-horseradish peroxidase (RDI, Flanders, NJ) was substituted for streptavidin-horseradish peroxidase in the IL-1 ELISA. Absorbency at 450 nm was determined using a Molecular Devices (Sunnyvale, CA) microplate reader. Data were analyzed using the SOFTmax 2.32 program (Molecular Devices).

Nitrite Assay-- Nitrite, an end product of NO oxidation, was used as an indicator of NO production by microglial cells (31). Nitrite in conditioned media was determined using the Griess assay according to the manufacturer's instructions (Promega). Absorbency was determined at 550 nm using a Dynatech Laboratories MR700 microplate reader (Dynex, West Sussex, UK).

Reverse Transcription and Polymerase Chain Reaction (RT-PCR) for Inducible Nitric Oxide Synthase mRNA-- RT-PCR was used to determine the effects of M-CSF and AP on inducible nitric oxide synthase (iNOS) mRNA in BV-2 cells. Total RNA was extracted from BV-2 cells using the Trizol reagent (Life Technologies, Inc.) according to the manufacturer's instructions. Reverse transcription was performed using 1 µg of total RNA and Superscript II RNase H reverse transcriptase (Life Technologies, Inc.) primed with random hexamers according to the manufacturer's instructions. PCR was performed on cDNA using primers for mouse iNOS (32) and 28 cycles of PCR amplification consisting of 94 °C for 30 s, 57 °C for 30 s, and 72 °C for 45 s. To control for differences in total RNA concentration among samples, mRNA levels for mouse hypoxanthine phosphoribosyl transferase were determined with RT-PCR as described previously (33). PCR products were visualized on 2.5% agarose gels with ethidium bromide staining.

M-CSF Receptor Blocking-- To demonstrate specificity of the M-CSF effect, BV-2 cells were reacted for 1 h with a monoclonal blocking antibody against the mouse M-CSF receptor, c-fms, at a concentration of 20 µg/ml (gift from Drs. R. Shadduck and G. Gilmore). This reagent has been shown previously to specifically block the effects of mouse M-CSF on macrophages (34, 35). Sister cultures were reacted with a subclass-matched mouse IgG₁  control antibody (Sigma) also at 20 µg/ml. After 1 h, medium was removed, and the cells were treated with 22 µM AP 1-40 or 50 ng/ml M-CSF, or 22 µM AP 1-40 plus 50 ng/ml M-CSF, with or without the c-fms antibody or the control antibody. After 24 h, conditioned media were harvested and cleared by centrifugation, and IL-1 was quantified using ELISA as described above.

GM-CSF Receptor Phenotyping-- To demonstrate the expression of GM-CSF receptor  and  subunits by BV-2 cells, RT-PCR was performed on BV-2 cell total mRNA. For the  subunit, primers were designed using the Genbank cDNA sequence MUSCLNYSIM (36) for the mouse GM-CSF low affinity receptor subunit. The forward primer was a 21-mer, which spanned nucleotides 508-528, whereas the reverse primer was a 21-mer spanning nucleotides 931-951. Thirty-five cycles of PCR amplification were performed consisting of 95 °C for 1 min, 61 °C for 1 min, and 72 °C for 2 min, 20 s. This resulted in a PCR product of 444 bp. For the  subunit (AI2CB cDNA), the primers and PCR protocol of Fung et al. (37) were used, resulting in a PCR product of 325 bp. PCR products were visualized on 1.5% agarose gels using ethidium bromide staining. As a positive control for GM-CSF receptor expression, total RNA harvested from mouse bone marrow cells, which had been stimulated with 50 ng/ml GM-CSF for 24 h, was subjected to the same RT-PCR phenotyping protocol as the BV-2 cell RNA.

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

The BV-2 immortalized microglial cell line has been extensively characterized and has many of the features of primary microglia (38, 39), but it is devoid of immunologically active cells such as astrocytes commonly found in primary microglial cultures. Further, BV-2 cells express receptors for advanced glycation end products, which bind AP and induce signal transduction, and BV-2 cells treated with M-CSF show chemotaxis and other indications of activation (7, 40).

In the present study, treatment of BV-2 microglial cells for 24 h with 11 µM AP 1-40 resulted in an increase in IL-1 production of about three times control levels (Fig. 1). However, when BV-2 cells were simultaneously treated with M-CSF (25 or 50 ng/ml) and 11 µM AP 1-40, there was a large increase in IL-1 production (approximately 70 times control levels in the experiment illustrated in Fig. 1; these results were replicated in three other independent experiments). M-CSF alone had little effect on BV-2 IL-1 production. A similar augmenting effect of AP 1-40 and M-CSF on IL-1 production was obtained with a AP concentration of 22 µM (Fig. 4). AP 40-1, a reverse sequence control peptide that was prepared in the same manner as AP 1-40, had little effect either alone or in combination with M-CSF. The augmenting effect of M-CSF on AP-induced IL-1 production by BV-2 cells was inhibited by a monoclonal antibody to the mouse M-CSF receptor, c-fms (Table I), but not by a subclass-matched control antibody.

View larger version (13K): [in this window] [in a new window]   Fig. 1.   M-CSF augments AP-induced IL-1 expression by BV-2 microglia. BV-2 cells were treated for 24 h with serum-free medium alone, 11 µM AP 40-1, 11 µM AP 1-40, 50 ng/ml M-CSF, 50 ng/ml M-CSF plus 11 µM AP 40-1, or 50 ng/ml M-CSF plus 11 µM AP 1-40. Mouse IL-1 in conditioned medium was quantified using ELISA. Results are expressed as the mean ratio of IL-1 in conditioned media from treated cells to that in medium from sister control cultures (with standard error of the mean). All values were adjusted for number of viable cells present in each culture well. Actual mean IL-1 concentration for the M-CSF plus AP 1-40 treatment was 42.9 pg/ml.

View this table: [in this window] [in a new window]   Table I M-CSF receptor (c-fms) blocking antibody inhibits M-CSF augmentation of AP-induced microglial IL-1 expression BV-2 microglia were pretreated in triplicate for 1 h with a blocking monoclonal antibody against the M-CSF receptor (c-fms) or a subclass-matched control antibody. Fresh medium was then applied containing AP 1-40, M-CSF, or AP plus M-CSF, with or without the control or blocking antibodies. After 24 h, conditioned media were harvested for IL-1 ELISA. The c-fms antibody resulted in an approximately 50% reduction in M-CSF augmentation of AP-induced microglial IL-1 expression. Data area expressed as mean ratio of IL-1 in treatment medium to that in control medium, with standard error.

Simultaneous treatment of BV-2 cells with M-CSF and AP 1-40 also induced a very large increase in IL-6 production (Table II). Treatment of BV-2 cells with M-CSF alone or AP alone resulted in modest increases in mouse IL-6 in conditioned media. However, the combination of M-CSF and AP 1-40 (22 µM) resulted in an increase in IL-6 production by BV-2 cells that was over 200-fold greater than control values.

View this table: [in this window] [in a new window]   Table II M-CSF augments AP-induced IL-6 production by BV-2 cells BV-2 microglia were treated in triplicate for 24 h with medium alone or medium containing 22 µM AP 1-40, 50 ng/ml M-CSF, or 22 µM AP plus 50 ng/ml M-CSF. IL-6 in conditioned media was quantified using ELISA. Combined treatment with AP plus M-CSF resulted in a larger increase in IL-6 in medium than did either agent alone. These results were replicated in two other independent experiments. Results are expressed as mean ratio of IL-6 in treated medium to that in control medium with standard error.

M-CSF also augmented AP effects on NO (nitrite) production. Treatment of BV-2 cells with AP 1-40 (11 µM) or M-CSF (50 ng/ml) alone had little effect on nitrite in conditioned medium (Fig. 2). In contrast, simultaneous treatment of BV-2 cells with AP 1-40 and M-CSF resulted in nitrite levels in conditioned medium that were about 30-fold greater than control values. The control peptide AP 40-1, either alone or in combination with M-CSF, had little effect on nitrite in conditioned medium. The augmenting effect of combined AP and M-CSF treatment on microglial NO production was also detected at the mRNA level. Treatment with 22 µM AP 1-40 in combination with 50 ng/ml M-CSF for 18 h resulted in a larger increase in iNOS mRNA than either agent alone (Fig. 3). The control peptide AP 40-1 did not augment M-CSF effects on iNOS expression.

View larger version (11K): [in this window] [in a new window]   Fig. 2.   M-CSF augments AP-induced NO (nitrite) production by BV-2 cells. BV-2 cells were treated under the same conditions as in Fig. 1. Nitrite in conditioned medium was quantified using the Griess assay. Results are expressed as the mean ratio of nitrite in conditioned media from treated cells to that in medium from sister control cultures (with standard error of the mean). All values were adjusted for number of viable cells present in each culture well. Actual mean nitrite concentration for the M-CSF plus AP 1-40 treatment was 28.1 µM.

View larger version (46K): [in this window] [in a new window]   Fig. 3.   Inducible nitric oxide synthase mRNA is increased by combined treatment of BV-2 cells with AP and M-CSF. RT-PCR products are visualized in 2.5% agarose gel with ethidium bromide. The 200-bp position is indicated. Cells were treated for 18 h and then harvested for total mRNA. A, the 230-bp RT-PCR product derived from mouse-inducible nitric oxide synthase (iNOS) mRNA; B, the 178-bp product derived from mouse hypoxanthine phosphoribosyl transferase. Lane 1, control; lane 2, 22 µM AP 1-40; lane 3, AP 40-1; lane 4, 50 ng/ml M-CSF; lane 5, 22 µM AP 1-40 plus 50 ng/ml M-CSF; lane 6, 22 µM AP 40-1 plus 50 ng/ml M-CSF. Whereas all conditions show approximately equal levels of hypoxanthine phosphoribosyl transferase mRNA, the AP 1-40 plus M-CSF treatment shows a large increase in iNOS mRNA.

To further test for the specificity of M-CSF in augmenting AP effects on microglia, BV-2 cells were treated with the hematopoietic cytokine GM-CSF alone or in combination with AP 1-40 (22 µM). Unlike M-CSF, GM-CSF did not augment AP-induced IL-1 expression (Fig. 4). Likewise, cotreatment of BV-2 cells with AP and the microglial activator IL-3 did not result in an increase in IL-1 production (data not shown). Although GM-CSF did not augment AP-induced cytokine secretion by BV-2 cells, this was not because of a lack of GM-CSF receptors. RT-PCR phenotyping of BV-2 cells showed the expression of mRNAs for both the  and  subunits of the GM-CSF receptor (Fig. 5).

View larger version (14K): [in this window] [in a new window]   Fig. 4.   GM-CSF does not augment AP-induced IL-1 expression by BV-2 microglia. BV-2 cells were treated for 24 h with serum-free medium alone, 22 µM AP 1-40 plus 50 ng/ml M-CSF, or 22 µM AP plus 10, 100, or 1000 units/ml GM-CSF. Results are expressed as the mean ratio of IL-1 in conditioned media from treated cells to that in medium from sister control cultures (with standard error of the mean). All values were adjusted for number of viable cells present in each culture well. Actual mean IL-1 concentration for the M-CSF plus AP 1-40 treatment was 37.8 pg/ml.

View larger version (40K): [in this window] [in a new window]   Fig. 5.   BV-2 cells express the GM-CSF receptor  and  subunits. RT-PCR products were visualized in 1.5% agarose gel with ethidium bromide. RT-PCR reactions for the  or  subunits of the mouse GM-CSF receptor were performed on total RNA extracted from BV-2 cells or from GM-CSF stimulated mouse bone marrow cells. Lane 1, 444-bp PCR product derived from GM-CSF receptor  subunit mRNA in BV-2 cells; lane 2, 325-bp PCR product derived from GM-CSF receptor  subunit mRNA in BV-2 cells; lane 3, PCR product derived from GM-CSF receptor  subunit mRNA in mouse bone marrow cells; lane 4, PCR product derived from GM-CSF receptor  subunit mRNA in mouse bone marrow cells; lane 5, 300-, 400-, and 500-bp markers.

Neither M-CSF nor GM-CSF alone or in combination with AP resulted in proliferation of BV-2 cells at the doses, cell density, and treatment duration used in the present study. At the end of a representative 24-h experiment, the mean number of control cells was 1.7 × 10⁵/ml (S.E. = 0.2), whereas for 50 ng/ml M-CSF, the mean number was 1.5 × 10⁵/ml (S.E. = 0.2), and for 1000 units/ml GM-CSF, the mean number was 1.3 × 10⁵/ml (S.E. = 0.2). For M-CSF plus 11 µM AP 1-40, the mean number of cells was 1.2 × 10⁵/ml (S.E. = 0.2), whereas for GM-CSF plus AP 1-40, the mean number was 1.4 × 10⁵/ml (S.E. = 0.2).

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

The results presented here suggest that M-CSF augments AP-induced microglial inflammatory cytokine and NO production in AD. Simultaneous treatment of BV-2 microglia with M-CSF and AP resulted in large synergistic increases in IL-1, IL-6, and NO in conditioned media, which were greater than increases due to either agent alone. Interleukin-1 is expressed early in AD (41, 42) primarily by microglia (43), and through autocrine and paracrine mechanisms could further augment microglia-mediated inflammation and neuronal injury in AD. Interleukin-6, another microglial cytokine (44), is also increased in AD brain (45) and in the serum of AD patients (46). Increased IL-6 expression may induce inflammatory changes, which injure neurons (47). There is evidence that NO, an important inflammatory effector produced by rodent and human microglia (48), is present at increased levels in AD brain, resulting in nitration of proteins and other abnormal cellular changes (49).

Our results and the results of prior studies indicate that in AD there is a self-amplifying pathophysiologic cascade involving microglia, astrocytes, and neurons and the key AD cytokines M-CSF, IL-1, and IL-6 (Fig. 6). M-CSF levels are increased in the cerebrospinal fluid of AD patients, and M-CSF antibodies label neurons in AD brain (7). Further, expression of the M-CSF receptor c-fms is increased on microglia in AD brain (29), which may sensitize these cells to M-CSF effects. Astrocytes, an important source of M-CSF in the brain, can be induced to secrete M-CSF by IL-1 (20). We hypothesize that in AD, AP induces microglia to secrete small amounts of IL-1, as our results and the results of others indicate (10, 16, 50). IL-1 then induces astrocytes to express M-CSF, which augments AP-induced expression of IL-1 by microglia, resulting in further M-CSF expression by astrocytes. In addition, microglial IL-1 self-activates microglia via autocrine and paracrine effects. Neurons themselves may also secrete M-CSF in response to AP (7), which may further activate microglia.

View larger version (23K): [in this window] [in a new window]   Fig. 6.   Model for a self-amplifying pathophysiologic cascade in Alzheimer's disease. We propose that in AD M-CSF augments AP-induced microglial secretion of IL-1, IL-6, NO, and reactive oxygen species (ROS), which injure neurons. Microglial IL-1 and IL-6 activate astrocytes, which in turn produce M-CSF that further activates microglia. Activated astrocytes also produce NO and IL-6, which directly injures neurons. Neurons may also be a source of M-CSF, which activates microglia.

IL-6 promotes astrogliosis (51) and activates microglia (47). Increased IL-6 found in AD brain could come from microglia, astrocytes, or both. Our results suggest that M-CSF and AP would induce microglial IL-1 and IL-6 production in AD. IL-1 causes astrocytes to express IL-6 (52), so microglial IL-1 induced by M-CSF and AP would promote astroglial IL-6 expression. Through pro-inflammatory effects, IL-6 is thought to contribute to neurodegeneration in AD (47, 53).

Although AP alone may increase microglial NO production (13, 54), in the presence of M-CSF AP-induced microglial NO production is dramatically augmented. Contrary to prior findings, recent evidence indicates that human microglia can produce NO (48), so results obtained with murine cells are likely to closely model the human system. Microglial NO, either directly or through its highly toxic derivative peroxynitrite, would injure neurons in AD (14, 49, 55). NO may also induce additional IL-1 expression (56, 57), which in turn would promote astroglial M-CSF expression, ultimately resulting in further AP-induced NO production. Astrocytes, too, produce NO, and IL-1 can induce astrocyte iNOS (58, 59). Thus, in AD, microglial IL-1 induced by AP and M-CSF would augment NO neurotoxicity by activating astrocyte iNOS. Finally, microglia are likely to generate neurotoxic reactive oxygen species in response to AP (12).

In contrast to M-CSF, the hematopoietic cytokines GM-CSF, present in the brain in AD (60), and IL-3 did not augment AP-induced microglial cytokine and NO expression in our studies. Both GM-CSF and IL-3 can induce microglial activation (61). Thus, the synergistic effect of M-CSF and AP cannot be due to nonspecific microglial activation. Whereas GM-CSF and IL-3 share a common receptor subunit (_c) and elements of signal transduction (62), the M-CSF receptor, c-fms, is distinct (63). Indeed, our results indicate that blockade of c-fms attenuates M-CSF augmentation of AP effects on microglia. Differences in receptor function and signal transduction between M-CSF and other hematopoietic cytokines may account for the unique effects of M-CSF on AP-treated microglia. The absence of an augmenting effect of GM-CSF cannot be the result of a receptor deficiency, as BV-2 cells were shown to express mRNA for both subunits of the GM-CSF receptor. M-CSF augmentation of AP effects is not secondary to proliferation, as neither M-CSF nor GM-CSF induced BV-2 proliferation at the doses, cell density, and treatment duration we employed.

In conclusion, our results indicate that M-CSF may have an important role in the pathophysiology of AD by augmenting the microglial response to AP. Further, these results support the hypothesis that inflammatory effectors are an integral part of neuropathologic change in AD rather than being nonspecific signs of brain injury. Future studies should further clarify the relative roles of astrocytes and neurons in generating M-CSF in AD, fully phenotype microglia activated by combined M-CSF and AP treatment, and determine the effects of microglia activated by AP and M-CSF on neurons.