Macrophage colony stimulating factor is associated with excretion of amyloid-β peptides from cerebrospinal fluid to peripheral blood


Dr Masayasu Okochi MD, Department of Integrated Medicine, Division of Internal Medicine, Osaka University Graduate School of Medicine, D3, Yamada-oka 2-2, Suita, Osaka 565-0871, Japan. Email:


Background:  The process of aggregation of brain amyloid-β peptides (Aβ) is thought to be associated with the pathogenesis of Alzheimer's disease (AD). Amyloid-β peptides are produced by sequential endoproteolysis by β-site amyloid-β protein precursor-cleaving enzyme (BACE) followed by presenilin (PS)/γ-secretase. There are several species of Aβ due to cleavage diversity of PS/γ-secretase. The predominant species in human cerebrospinal fluid (CSF) or plasma is Aβ40, whereas Aβ42 is much more aggregatable and accumulated in senile plaques. The level of Aβ in the brain is determined by the balance between the generation and clearance of Aβ, including transport across the brain–blood barrier (BBB). Although the processes of Aβ generation and degradation have been studied in some detail, knowledge of the Aβ transport process across the BBB is limited. So far, low-density lipoprotein receptor-related protein (LRP1), P-glycoprotein (P-gp), and insulin-like growth factor-1 (IGF-1) have been identified to modify the excretion of brain Aβ to the blood.

Methods:  To investigate whether macrophage colony stimulating factor (M-CSF) has a role in the Aβ transport process, human Aβ was injected into the lateral ventricle of the brain of M-CSF-deficient (op/op) mice. Then, plasma and brain Aβ levels were measured by ELISA to determine the time-course of Aβ movement from the brain to the plasma.

Result:  When human Aβ40 was injected into mouse lateral ventricles, the efflux of Aβ from the CSF to the blood was transiently decreased and delayed in M-CSF-deficient mice. Moreover, endogenous plasma Aβ40 levels were lower in M-CSF-deficient mice.

Conclusion:  The results indicate that M-CSF deficiency impairs excretion of human-type Aβ40 from the CSF to blood. We propose that M-CSF may be a novel factor that facilitates the excretion of Aβ from the CSF to the blood via the BBB.


Accumulation of amyloid-β peptides (Aβ) in senile plaques in the brain is one of the pathological features of Alzheimer's disease (AD)1,2 The Aβ are produced by sequential endoproteolysis by β-site amyloid-β protein precursor (APP)-cleaving enzyme (BACE), which sheds β-amyloid protein precursor (βAPP) at the extracellular domain, and by presenilin (PS)/γ-secretase, which cleaves in the middle of the transmembrane domain. Increases in Aβ levels in the brain may trigger or enhance aggregation of Aβ in the brain, which is associated with the pathogenesis of AD.2 Therefore, Aβ clearance from the brain is a key step in reducing brain Aβ levels.3 However, our knowledge of Aβ clearance from the brain is limited. Low-density lipoprotein receptor-related protein 1 (LRP1)4–6 and P-glycoprotein (P-gp)7 promote brain Aβ efflux at the blood–brain barrier (BBB), whereas insulin-like growth factor-1 (IGF-1) reduces brain Aβ levels by enhancing the permeability of Aβ carrier proteins, such as albumin and transthyretin.8

Macrophage colony stimulating factor (M-CSF) regulates the growth, differentiation, and function of mononuclear phagocytes.9 Interestingly, several recent reports indicate a possible association between M-CSF and the pathological processes in AD. First, plasma levels of M-CSF in the early stages of AD is lower,10 whereas M-CSF levels in the cerebrospinal fluid (CSF) are higher.11 Second, although still controversial, in M-CSF deficient (op/op) mice, senile plaque-like mouse Aβ deposition has been reported.12–14

In the present study, we used M-CSF-deficient mice to investigate whether M-CSF is associated with Aβ excretion from the CSF to the blood via the BBB. We injected human Aβ into the lateral ventricles of the brain and monitored Aβ levels in the brain and peripheral blood. The results indicate that M-CSF may be a novel factor that facilitates Aβ40 excretion from CSF to the blood.



B6C3Fe-a/a-op/+ male and female mice, as a breeding pair, were obtained from Jackson Laboratory (Bar Harbor, ME, USA). The M-CSF–/– (op/op) and M-CSF+/– genotypes were raised and confirmed by polymerase chain reaction.15 The M-CSF deficiency was judged by the failed eruption of incisors around postnatal Day 10. The M-CSF–/– mice were fed a granulated diet (CE-2, CLEA JAPAN) after weaning.12 All animals were treated ethically as specified by the Ethics Committee of Osaka University Graduate School of Medicine.

Synthetic Aβ preparation for injection

Human Aβ1–40 purchased from Peptide Institute of Japan (Osaka, Japan). The peptide was freshly solubilized in 10 mmol/L phosphate buffer and pH adjusted to 7.4 with NaOH. The Aβ solution was sonicated for 15 min on ice.

Intracerebroventricular and intravenous injections of Aβ

A stainless-steel 30 gauge cannula was placed stereotaxically into the lateral ventricle. Coordinates for the tip of the cannula were 0.75 mm posterior, 1.8 mm lateral to the bregma, and 2.5 mm depth from the surface of the skull of mice anesthetized with 60 mg/kg, i.p., sodium pentobarbital. One the cannula had been place, 1 μg/2 µL fresh human Aβ was injected into the lateral ventricles on both sides over a period of 1 min. In addition, fresh 1 μg hAβ1–40 was injected as a 100 µL bolus i.v. into the tail vein of mice. Blood was collected from the eye veins into heparinized microhematocrit tubes at indicated times after injection. The amount of Aβ1–40 in the brain after i.c.v. injection was analyzed at 45 min and 2 h. Brains were extracted in 1% CHAPS as described below and 1% CHAPS-soluble hAβ1–40 quantified by ELISA.

Sandwich ELISA for Aβ

All Aβ was quantified using an ELISA kit (Wako Pure Chemical Industries, Osaka, Japan). To determine plasma and brain levels of human Aβ after injection, a human Aβ1–40 (292–62301) ELISA kit was used. To determined endogenous mouse Aβ levels, a human/rat Aβ40 (294–62501) ELISA kit was used.


The following antibodies against Aβ were used. Polyclonal, forAβ40: AB5074P (anti-34–40 of Aβ40; Chemicon, Temecula, CA, USA)16 and FCA3340 (anti-33–40 of Aβ40; EMD Biosciences, San Diego, CA, USA); for Aβ42: AB5078P (anti-37–42 of Aβ42; Chemicon)12,13 and FCA3542 (anti-35–42 of Aβ42; EMD Biosciences); forAβ40/42: AB5076 (anti-40/42; Chemicon);17 for mouse Aβ: NE1012 (anti-mouse 3–16 of Aβ40; EMD Biosciences); monoclonal, 4G8 (anti-17–24).

Brain Aβ extraction

Mouse brains were frozen in liquid nitrogen and stored at −80°C. Brains were extracted sequentially by 1% 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS; Sigma, St Louis, MO, USA),18 2% sodium dodecyl sulfate (SDS), and formic acid (FA). Brains were homogenized 15 times up–down using a Teflon homogenizer with 1% CHAPS (2.5 times as much volume) in 50 mmol/L Tris-buffered saline (TBS) with protease inhibitors (complete protease inhibitor cocktail, one tablet in 50 mL solution; Roche, Basel, Switzerland) and ultracentrifuged at 100 000 g for 15 min at 4°C. The 1% CHAPS-soluble supernatant was used to measure Aβ by ELISA. The 1% CHAPS-insoluble pellet was homogenized in 2% SDS in TBS and formic acid and extracted by 1% CHAPS/TBS.19 The formic acid-soluble fractions were dried using a speed-vac (Thermo, Waltham, MA, USA), mixed with 2× SDS sample buffer, and boiled for 5 min at 100°C.


Brains were fixed in 4% paraformaldehyde and rinsed in 0.1 mol/L phosphate-buffed saline (PBS), pH 7.6, overnight at 4°C. Specimens were then embedded in paraffin and cut into 10 μm sections. Paraffin sections were deparaffinized, rehydrated, and quenched for endogenous peroxidase activity in 3% H2O2 for 5 min, then pretreated with formic acid for 1 min. After blocking with 1% normal serum from the species in which the secondary antibody was raised, sections incubated with a primary antibody overnight at 4°C and biotinylated secondary antibody for 1 h at room temperature. Sections were then washed in PBS with 0.02% Tween 20 and incubated for 30 min at room temperature with avidin-biotinylated horseradish peroxidase complex (Elite ABC Elite; Vector Laboratories, Burlingame, CA, USA). Labeling was visualized by diaminobenzidine (DAB) staining.

For Congo red staining,20 sections were deparaffinized, rehydrated, preimmersed in 80% ethanol containing 4% sodium chloride for 30 min, and then immersed in 80% ethanol containing 0.2% Congo red for 30 min at room temperature. Sections were coverslipped and observed under a microscope with a reflex filter.

For Thioflavin-S staining (, sections were deparaffinized, rehydrated, couterstained in Mayer's hematoxylin for 5 min, rinsed in water for 5 min, rinsed in double-distilled H2O, and stained with 1% Thioflavin-S (Sigma) for 5 min, followed by differentiation in 70% ethanol for 5 min. Sections were then mounted in glycerin jelly.


The Aβ was separated by Tris-glycine (Invitrogen, Carlsbad, CA, USA) SDS–polyacrylamide gel electrophoresis (SDS-PAGE), as described previously,21–23 after samples had been transferred to a nitrolocellurose membrane and the membrane boiled in PBS for 10 min before being immunobloted for chemiluminescence.24


Levels of Aβ in the blood and brain were analyzed by Mann–Whitney's U non-parametric test. To calculate Aβ kinetics in the plasma, a moment analysis model was used. Statistical software was downloaded from


Effects of M-CSF deficiency on plasma Aβ40 levels after i.c.v. injection

There has been some controversy as to whether M-CSF can induce deposition of Aβ in mouse brain.12–14 To determine whether M-CSF is associated with Aβ metabolism in the central nervous system, we investigated how exogenously injected Aβ is metabolized in M-CSF-deficient mice. Human Aβ40 (hAβ), the predominant Aβ species, was microinjected (1 μg in 4 μL buffer) into the lateral ventricles. Levels of hAβ excreted into the peripheral blood, as well as those remaining in the brain, were then determined. Following i.c.v. injection, blood was collected at 5, 30, 60, 120, and 240 min, and plasma hAβ levels determined using an ELISA kit (Fig. 1). Surprisingly, the maximum plasma levels of hAβ in M-CSF–/– mice were much lower than those in M-CSF+/+ mice. Plasma hAβ levels in M-CSF–/– mice reached a maximum approximately 60 min after injection, whereas levels in wild-type mice were maximum at approximately 30 min after injection (Fig. 1a). Moreover, the area under the concentration and time curve (AUC), which indicates the amount of hAβ excreted to the plasma, was significantly lower in M-CSF–/– mice compared with wild-type mice (Fig. 1b). Thus, after i.c.v. hAβ injection, the total and maximum levels of hAβ excreted from the CSF into plasma is much lower in M-CSF–/– mice than in wild-type mice. In addition, the peak in plasma hAβ levels is delayed in M-CSF–/– mice.

Figure 1.

(a) Plasma levels of human (h) Aβ40 after i.c.v. injection of amyloid-β peptide (Aβ) into the lateral ventricles of 3-month-old M-CSF–/– (▪) and M-CSF+/+ (●) mice and (b) the area under the curve (AUC) after injection. The amound of hAβ excreted to the plasma was significantly less in M-CSF–/– compared with M-CSF+/+ mice (7.3 ± 3.1 vs 21.3 ± 10.8 ng/mL per h, respectively; *< 0.01). Data are the mean±SEM (n = 5–8).

Degradation of hAβ in plasma in M-CSF–/– mice

To rule out the possibility that processes of Aβ degradation are accelerated in M-CSF–/– mice, we injected hAβ (1 μg in 100 μL buffer) into the tail vein of both M-CSF–/– and M-CSF+/+ mice, collected blood 5, 30, 60, 120, and 240 min, and determined plasma hAβ levels (Fig. 2). As shown in Fig. 2, there was no difference in the speed or amount of Aβ degraded in M-CSF–/–and M-CSF+/+ mice. The half-life of the injected hAβ in plasma was similar in M-CSF–/– and M-CSF+/+ mice (0.368 h and 0.372 h, respectively). These results clearly indicate that lower plasma hAβ levels after i.c.v. injection of hAβ into M-CSF–/– mice were not due to accelerated degradation of hAβ. The results suggest that M-CSF deficiency may cause impaired excretion of hAβ from the CSF to plasma, which results in lower hAβ levels in plasma after i.c.v. injection into the lateral ventricles.

Figure 2.

There was no difference in the rate of degradation of intravenously injected human (h) Aβ40 in 3-month-old M-CSF–/– (▪) and M-CSF+/+ (●) mice. Data are the mean±SEM (n = 5–7).

Brain levels of hAβ in M-CSF–/– and wild-type mice

Next, we investigated hAβ levels in brains of M-CSF–/– mice following i.c.v. injection. Mice were killed 45 or 120 min after injection and brain hAβ levels were determined in the CHAPS soluble fraction. As shown in Fig. 3, hAβ in M-CSF–/– mouse brain was higher than in wild-type mouse brain 45 min after injection. However, 120 min after injection, there was no longer any difference between the two groups. These results suggest that: (i) because of impaired excretion of hAβ from the CSF to the plasma, M-CSF deficiency causes a transient increase in hAβ levels in the brain after injection; and (ii) although the rate of excretion is slower under conditions of M-CSF deficiency, any hAβ that has been injected is effectively excreted from the CSF eventually.

Figure 3.

Brain human (h) Aβ40 levels in M-CSF–/– (□) and M-CSF+/+ (▪) mice after i.c.v. injection. At 45 min after injection, there was a significant difference in brain hAβ40 levels between M-CSF–/– and M-CSF+/+ mice (6.45 ± 1.31 vs 3.51 ± 1.90 pmol/g, respectively; *< 0.01). Data are the mean±SEM (n = 4 per group).

Effect of M-CSF deficiency on accumulation of Aβ in mouse brain

We considered the possibility that the impairment of Aβ excretion from brain of M-CSF–/– mice may result in the accumulation of endogenous mouse Aβ (mAβ) in the brain. To address this, we extracted a 1% CHAPS/TBS-soluble fraction from 30-day-old mouse brains and measured Aβ by ELISA (Wako). As shown in Fig. 4a, Aβ levels in M-CSF–/– brains were no higher than those in the wild-type brain. Because our results differ from those of previous studies, which showed that mAβ was deposited in M-CSF–/– mice,12,13 we performed biochemical and immunohistochemical analysis of M-CSF–/– mice. Most of the Aβ in senile plaques is highly insoluble and was recovered in the 70% formic acid-soluble fraction in the 2% SDS-insoluble fraction. First, we extracted the formic acid fraction from brain tissues of 30-day-old M-CSF–/– and wild-type mice, as well as from a patient with AD, and performed SDS-PAGE followed by western blotting (Fig. 4b). As shown, Aβ was not detected in the formic acid fraction of brains from either M-CSF–/– or wild-type mice, whereas a marked Aβ band was detected in the formic acid fraction of brain tissue from the AD patient (Fig. 4b).

Figure 4.

(a) Concentrations of mouse Aβ40 (mAβ) in brains from 30-day-old M-CSF–/– (n = 8) and M-CSF+/+ (n = 8) mice, as well as 3-month-old βAPP/PS1 transgenic mice (n = 12). (b) Amyloid-β peptide (Aβ) in the formic acid (FA) fraction of brain tissues from 30-day-old M-CSF–/– (n = 4) and M-CSF+/+ (n = 4) mice, as well as from a human Alzheimer's disease (AD) patient. The synthetic mouse Aβ42 peptide was loaded in the 10th lane. Asterisks indicate Aβ larger than the monomer. (c) Immunohistochemical, Congo red, and Thioflavin-S staining of senile plaques in brain sections from M-CSF–/–, M-CSF+/+, and βAPP/PS1 transgenic mice.

Immunohistochemical studies (Fig. 4c) were performed using anti-Aβ antibodies, which identify Aβ40 or 42, 40/42, and rodent Aβ specifically. Notably, although staining a considerable number of amyloid plaques in brain sections of the βAPP/PS1 transgenic mouse,25 none of these antibodies immunostained amyloid plaque-like Aβ deposition in brain sections of 30-day-old M-CSF–/– mice. Cerebrovascular Aβ deposition was not observed in brain sections from M-CSF–/– mice (data not shown).

In addition, brain sections were stained using Congo red and Thioflavin-S, which bind to the amyloidogenic β-sheet structure (Fig. 4c). Although clearly detected in the βAPP/PS1 transgenic mouse brain, fibrillogenic Aβ was not observed in the M-CSF–/– brain. Collectively, the results indicate that M-CSF deficiency does not cause accumulation of endogenous Aβ in the mouse brain.

Therefore, our biochemical results indicate that endogenous mouse-type Aβ is not accumulated in the M-CSF–/– mouse brain, which is consistent with the results reported previously.14

Effects of M-CSF deficiency on plasma hAβ levels

A deficiency of M-CSF may cause impaired hAβ excretion from the CSF to the blood. The results of the present study suggested to us that a defect in Aβ metabolism caused by M-CSF deficiency may have an impact on levels of plasma Aβ. As shown in Fig. 5a,b, plasma levels of Aβ in M-CSF–/– mice were significantly lower than those of wild-type mice, regardless of age (1 or 3 months). As expected, the CHAPS/TBS-soluble fractions of brains from wild-type and M-CSF–/– mice contained similar levels of Aβ40 (Fig. 5c,d).

Figure 5.

Concentrations of mouse (m) Aβ40 in the plasma of (a) 30-day-old and (b) 3–4-month-old M-CSF–/–or M-CSF+/+ mice. Plasma levels in 30-day-old and 3–4-month-old M-CSF–/– mice (39 ± 22 and 58 ± 19 pmol/L, respectively) were significantly lower than plasma concentrations in age-matched M-CSF+/+ mice (101 ± 21 and 107 ± 15 pmol/L, respectively; *< 0.001). Data show the mean (n = 12–21). Concentrations of mAβ40 levels in the CHAPS/TBS-soluble fractions of (c) 30-day-old and (d) 3–4-month-old M-CSF–/–or M-CSF+/+ mice measured by ELISA. Data show the mean (n = 6–8).

Collectively, we found that: (i) M-CSF deficiency may cause impaired Aβ40 excretion from the CSF to the blood; and (ii) plasma Aβ40 in M-CSF–/– mice is lower than in wild-type mice.


We have discovered that M-CSF is associated with human Aβ40 excretion from the CSF to the blood. When hAβ was injected into the lateral ventricles of the brain in M-CSF-deficient mice, the appearance of the transient peak of hAβ in the plasma was delayed and the total amount of hAβ in the plasma was decreased. In addition, levels of hAβ after i.c.v. injection into the brains of M-CSF–/– mice are transiently higher than those in wild-type mice. Because degradation of hAβ in the plasma is not accelerated in M-CSF–/– mice, these changes may be caused by impaired excretion of hAβ from the CSF to the plasma. Importantly, we found that M-CSF deficiency causes a decrease in plasma Aβ40 levels, but not in levels in the brain. Our data conclusively suggest that M-CSF may facilitate Aβ40 excretion from the CSF to the plasma via the BBB.

Our biochemical and immunohistochemical studies have shown that endogenous mAβ is neither deposited nor even increased in the brain of M-CSF–/– mice, which favors the previous finding that no Aβ deposition was observed in brains of M-CSF–/– (op/op) mice.14 Thus, these mice are not a useful model in which to study the process of endogenous Aβ deposition. However, we found that mAβ40 levels in the plasma are decreased in M-CSF–/– mice. So far, such a change has not been reported for experimental mouse models of other factors associated with Aβ excretion, such as P-gp, IGF-1, and LRP-1. Therefore, M-CSF–/– mice may be a useful model in which to investigate the process of Aβ excretion from the CSF to the blood via the BBB.

Both LRP14,5 and P-gp7 exist on brain capillaries and clear Aβ from the brain. Recently, it was reported that soluble circulating LRP plays a role in the clearance of Aβ from the brain.6 Insulin-like growth factor-1 reduces brain Aβ levels by enhancing the permeability of the Aβ carrier proteins albumin and transthyretin.8 It remains to be elucidated how M-CSF causes the excretion of Aβ from the CSF to the plasma.

Macrophage colony stimulating factor, also known as CSF-1, plays an important role in the proliferation and differentiation of macrophages; thus, the number of tissue macrophages, including microglia, in the brain is reduced in M-CSF–/– mice.26,27 Microglia are considered to play a role in the clearance of brain Aβ.28 Therefore, a higher level of brain Aβ in M-CSF–/– mice brain after i.c.v. injection of hAβ may be explained, in part, by a reduction in the number of microglia. There are at least three isoforms of M-CSF.9 One isoform exists on the cell surface and is involved in local regulation. The other two isoforms are secreted proteoglycan or glycoprotein. The M-CSF–/– mice lack all three isoforms.9 Further studies are necessary to reveal whether all or only certain isoforms have are active in the transport of Aβ from the CSF to the blood via the BBB.


The authors acknowledge funding from the Program for the Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (05-26; to MT, MO, and ST), Grants-in-Aid for Scientific Research on Priority Areas–Advanced Brain Science Project (to MO) and KAKEN-HI from the Ministry of Education, Culture, Sports, Science and Technology (to MT, MO, and ST), and Grants-in-aid from the Japanese Ministry of Health, Labor and Welfare (to MT, MO).