|Home | About | Journals | Submit | Contact Us | Français|
Alzheimer Disease (AD), a progressive neurodegenerative disease characterized by the accumulation of amyloid-β protein and neuronal loss, is the leading cause of age-related dementia in the world today. The disease is also associated with neuroinflammation, robust activation of astrocytes and microglia and evidence of activation of the complement system, localized with both fibrillar amyloid-β (fAβ) plaques and tangles. The observations are consistent with a complement dependent component of AD progression. We have previously shown that inhibition of the major complement receptor for C5a (CD88) with the C5a receptor antagonist (PMX205) results in a significant reduction in pathology in two mouse models of AD. To further characterize the role of complement in AD related neuroinflammation, we examined the age and disease associated expression of CD88 in brain of transgenic mouse models of AD and the influence of PMX205 on the presence of various complement activation products using flow cytometry, western blot and immunohistochemistry. CD88 was found to be upregulated in microglia, in the immediate vicinity of amyloid plaques. While thioflavine plaque load and glial recruitment is significantly reduced after treatment with PMX205, C1q remains co-localized with fAβ plaques and C3 is still expressed by the recruited astrocytes. Thus, with PMX205, potentially beneficial activities of these early complement components may remain intact, while detrimental activities resulting from C5a-CD88 interaction are inhibited. This further supports the targeted inhibition of specific complement mediated activities as an approach for AD therapy.
Over the past two decades, the most prominent hypothesis addressing the causal factor behind Alzheimer Disease (AD) development has been the amyloid cascade hypothesis that states that the amyloid-β (Aβ) peptide, the primary component of AD plaques, is what initiates neuronal dysfunction in AD (Selkoe and Schenk 2003). Consistent with this hypothesis, Aβ accumulation in AD is also associated with neurofibrillary tangles, extensive synaptic and neuronal loss and increased inflammation. Although there has been experimental observations (Oddo et al. 2003) consistent with Aβ accumulation as necessary for AD onset, studies conducted in human patients (Terry et al. 1991), and supported using transgenic mouse models (Blurton-Jones et al. 2009) suggest that Aβ alone is not sufficient for both the cellular and cognitive loss observed in the disease. It is likely that multiple factors, acting both intra-and extracellularly, contribute to AD and that Aβ is one component in a series of physiological cascades necessary for the transition from cognitively normal to the impairment of AD type dementia (Pimplikar 2009;Hardy 2009). One physiological cascade initiated in response to increased Aβ deposition that results in recruitment of inflammatory elements of the innate immune response is the complement cascade (McGeer and McGeer 2002). Evidence of an inflammatory response to Aβ deposition has been accumulating since the 1980’s (Eikelenboom and Stam 1982;Eikelenboom et al. 1989) and multiple investigators have attempted to define the role such inflammation plays in disease development (Bonifati and Kishore 2007).
In contrast to non-demented elderly individuals, who may also contain pools of Aβ deposits and some low level inflammation, brains of AD patients have fibrillar Aβ plaques (fAβ) and these plaques show extensive deposition of components of the complement system (Afagh et al. 1996;Zanjani et al. 2005). As part of the innate immune response, the complement system is a powerful mediator of inflammation with effector functions ranging from identifying pathogenic materials to orchestrating their destruction (Seelen et al. 2005). The system is composed of over 30 fluid phase and cell bound components, most of which have been demonstrated to be produced in the AD brain (Strohmeyer et al. 2000;Yasojima et al. 1999). C1q, the recognition component of the classical complement pathway, has been shown to interact with fibrillar β-sheet rich Aβ, reviewed in (Tenner and Fonseca 2006). In addition, evidence of alternative pathway mediated complement activation has also been shown in vitro (Bradt et al. 1998) and in mouse models (Zhou et al. 2008;Maier et al. 2008). Once complement activation occurs, the downstream activation products C3a and C5a, together known as the anaphylatoxins, recruit and activate resident phagocytes, including microglia and astrocytes, to the site of initiation (Yao et al. 1990), via the interaction of these molecules with their cell surface receptors (Nataf et al. 1998). It has been hypothesized that C5a, in vitro, can increase the fAβ induction of proinflammatory cytokines in microglia, as shown in the mouse microglia cell line BV2 (O’Barr and Cooper 2000). In vivo, the response to C5a may be further enhanced through the synergistic activation of the classical C5a receptor (CD88) with other immune receptors such as P2YR (Flaherty et al. 2008) and TLR4 (Hawlisch et al. 2005;Patel et al. 2008;Zhang et al. 2007). The possible synergistic interaction between CD88 and TLR4 may be of particular interest in light of the recent reports of Aβ binding to TLR4 (Fassbender et al. 2004;Walter et al. 2007).
We have previously shown that treatment with the antagonist PMX205, which blocks C5a mediated CD88 signaling results in decreased glial activation and Aβ plaque load, increased synaptophysin reactivity in CA3 area of the hippocampus, and improved behavioral performance in the Tg2576 mouse model (Fonseca et al. 2009). Since there are conflicting reports of C5a receptor modulation and function in the CNS (Woodruff et al. 2009), and to further investigate the effects of complement associated pathological changes in these mouse models of human AD, we assessed the presence, localization and regulation of CD88 and the complement components C1q, C3 and C4 in untreated and PMX205 treated AD mouse models.
All cell culture media and supplements were purchased from Invitrogen, (Carlsbad, CA). Antibodies used were: rat anti-mouse CD88 (clone 10/92, 10μg/ml, Serotec, Raleigh, NC), rabbit anti-mouse CD88 (C1150-32, 2 μg/ml, BD Pharmingen, San Jose, CA), rabbit anti-Iba-1(5μg/ml, Wako, Richmond, VA), rabbit anti-bovine GFAP (4 μg/ml, Dako,Carpinteria, CA), rat anti-mouse C3 (clone 11H9, 5μg/ml, Cell Sciences, Canton, MA), goat anti mouse CD45 (10ug/ml, R&D,Minneapolis,MN), rabbit anti-mouse C1q (1151, 4μg/ml) (Fan and Tenner 2004) and rat anti mouse C4 (clone16D2, 5ug/ml,Cell Sciences). The isotype control rat IgG was obtained from Serotec.
Tg2576 mice which express the human APP695 gene containing the Swedish mutation (KM670/671NL) under the control of hamster prion promoter (Hsiao et al. 1996) were maintained by backcrossing to the B6/SJL strain (The Jackson Laboratory, Bar Harbor, ME). Arc48 (Cheng et al. 2004), Arctic-mutant hAPP mice (from Dr. Lennart Mucke, Gladstone Institute, San Francisco, CA), were backcrossed onto the C57BL/6J strain (Jackson Laboratory).
C5a receptor knockout mice, generated by targeted deletion of the C5a receptor gene (CD88-/-) and backcrossed 10 generations to C57BL/6J mice were provided by Dr. Rick Wetsel (Hollmann et al. 2008). Initial IHC and WB experiments were performed using the CD88-/- mouse (on BALB/C background) generated by Gerard (Hopken et al. 1996) and obtained from Jackson. Non-transgenic (B6/SJL) or (C57BL/6J) littermate mice were used as controls. Animal experimental procedures were pre-approved by the University of California at Irvine Institutional Animal Care and Use Committee.
Cortical and hippocampal tissue from pups, postnatal day 2-4, from either C57/B6 or CD88-/- mothers was dissociated by trituration and trypsinization and subsequently cultured in Dulbecco’s Modified Eagle Medium (DMEM) complete media (containing 10% Fetal Bovine Serum and 1% Pen-Strep) on poly-L-lysine, 20μg/ml, (Sigma Aldrich, St. Louis, MO) coated flasks as previously described (Li et al. 2004). Microglia were obtained after culturing for intervals of 7–10 days followed by shaking flasks at 140 r.p.m. and 37°C for 90 min. For neuron primary cultures, brains from embryonic day 15 or 16 C57/B6 or CD88-/- mice were placed in Hank’s balanced salt solution free of calcium and magnesium (CMF) and cleaned of meninges. Cortices were dissected out and placed in 0.125% of trypsin/EDTA in CMF for 7 min at 37°C. Cortical tissues was then resuspended in DMEM complete media and dissociated by trituration using flame polished siliconized Pasteur pipettes. Viable cells, quantified by trypan blue exclusion, were plated at 7.0 × 105 cells per well, in 12 well plates, in DMEM complete media on poly L-lysine (10μg/ml) coated 12-well plates (Costar, Cambridge, MA, USA) for 1 hour. Media was subsequently changed (100%) to serum-free neurobasal medium supplemented with B27 (NB/B27), and cells grown 7–14 days in before use, with a 50% media change at 4 and 8 days.
Mouse hemi-brains, frozen immediately after transcardial perfusion with PBS and removal, were first pulverized and then homogenized with 10μl/mg 2% SDS (in DI water) with protease inhibitor cocktail, per manufactures instructions (Roche Diagnostics, Cat# 118836153001, Indianapolis, IA). The lysate was then centrifuged at 15,000 × g at 4° C for 15 min and the supernatant collected. The protein concentration was determined using BCA protein assay (Pierce, Rockford, IL, USA). Primary mouse microglia or neurons were washed in sterile phosphate buffer saline (PBS) followed by lysis in 2% SDS (in DI water) with protease inhibitor cocktail (Roche Diagnostics) at a concentration of 1 × 106 cells/ml. Lysates were centrifuged at 15,000 g at 4° C for 15 min and the supernatants collected. The protein concentration was determined using BCA protein assay (Pierce). All samples were diluted in sample buffer (50mM Tris-C1 pH 6.8, 2% SDS, 0.1% bromophenol blue (Sigma-Aldrich) and 10% glycerol (EMD Chemicals, Gibbstown, NJ)) and separated under non-reduced or reduced (100mM DTT (Sigma-Aldrich,) conditions on 10% SDS-polyacrylamide gels. Proteins were then transferred to polyvinylidene difluoride (PDVF) membranes (Millipore Corporation, Bedford, MA, USA) using a 315mA current for 1½ hrs. The membranes were first washed with T-TBS (50 mmol/L Tris pH 7.4, 150 mmol/L NaCl, and 0.1%Tween), followed by blocking in 3% milk-TTBS overnight at 4° C. The membranes were then probed with anti-CD88 antibodies in 3% milk-T-TBS for 2 h at room temp, followed by three washes with T-TBS and incubation with HRP-conjugated anti-rabbit or anti-rat IgG (Jackson Immunoresearch, West Grove, PA) in 1.5% milk-T-TBS for 1½ hrs at room temp. The blot was developed with ECL Plus western blotting detection system (Amersham, Buckinghampshire, UK).
Mouse microglia, obtained from culture, were first washed in HBSS buffer (1.3mM CaC12, 5mM KCl, 0.3mM KH2PO4, 0.5mM MgCl2, 0.4mM MgSO4, 138mM NaCl, 4mM NaHCO3, 0.3mM Na2HPO4, and 5.6mM D-glucose pH 7.4) by centrifugation at 540 × g for 7 min at room temp. The cells were then resuspended to 2 × 105 cells in FACS buffer (HBSS, 0.2% BSA, and 0.2% sodium azide) and incubated with primary Abs (1μg of clone 10/92, Serotec or 0.8μg of C1150-32, BD Pharmingen) for 1 hr on ice with light tapping every 10 min. Cells were then washed twice in ice-cold FACS buffer and incubated with 2μl of FITC-conjugated secondary donkey anti-rat IgG (Jackson ImmunoResearch) for 30 min on ice with light tapping every 10 min. Evaluation of cells was performed using standard flow cytometry methods on a FACSCalibur (BD Biosciences, San Jose, CA). Flow cytometric data was analyzed with FlowJo software (Tree Star, Ashland, OR).
Mice were anesthetized with a mixture of ketamine/xylazine (67/27mg/kg) and perfused with phosphate buffered saline (PBS). After dissection half brain was immediately frozen in liquid nitrogen and the other half fixed by placement in 4% paraformaldehyde in PBS for 24 hours. Fixed tissue was stored in PBS/0.02% Na azide at 4°C until use. Immunohistochemical procedures were done as previously described on coronal, 40 μm vibratome sections (Fonseca et al. 2004). Briefly, fixed sections were stained using: 1) the avidin-biotin-peroxidase (ABC) (Vector) staining procedure: After antigen retrieval (Antigen unmasking solution, Vector), endogenous peroxidase was blocked by 3% H2O2/10% Methanol in TBS and non specific binding was blocked by incubation with 2%BSA 10% Normal Goat sera 0.1% Triton in TBS. Sections were incubated with primary antibodies overnight in blocking solution followed by the corresponding biotinylated secondary antibodies (1h RT) and ABC (1h RT). Staining was visualized with DAB. 2) Fluorescent immunostaining: A step of blocking with 2%BSA/0.1%Triton/TBS (with or without normal sera) preceded the incubation with the indicated primary antibodies. Detection was done using the species specific corresponding biotinylated secondary antibodies followed by streptavidin (SA)-CY3 (Jackson Immunoresearch) or SA-Alexa555 (Invitrogen: Molecular Probes, Carlsbad, CA). Double fluorescent immunolabeling was done incubating simultaneously with the primary antibodies to CD88/GFAP or C3/GFAP and detecting CD88 or C3 with biotinylated anti rat IgG followed by SAAlexa555 (or SACy3) and detecting GFAP with Alexa 488 (Invitrogen: Molecular Probes) or FITC (Jackson Immunoresearch) labeled anti-rabbit IgG. For CD88/Iba-1 or CD88/CD45 the procedure was done sequentially: first, incubation with anti-CD88 and detection as indicated above followed by overnight incubation with Iba-1 and detection with Alexa 488 anti-rabbit IgG or after incubation and detection of CD88, overnight incubation with CD45 and detection with biotinylated anti goat IgG (Vector) followed by SA FITC (Jackson Immunoresearch). For CD88/CD45 co-localization, before starting with the second primary antibody a blocking step with streptavidin/ biotin blocker (Vector) was performed in order to avoid cross reactivities. For all single labeling and co-localizations controls omitting each one of the primary antibodies or including normal IgGs were included. Fibrillar amyloid was labelled with 1% thioflavine. Slides were mounted with Vectashield (Vector). Immunostaining was observed with a Zeiss Axiovert-200 inverted microscope (Carl Zeiss, Thornwood, NY). Images were acquired with a Zeiss Axiocam high-resolution digital color camera (1300×1030 pixels) using Axiovision 4.6 software. Some of the images were taken using a Zeiss LSM 710 META Laser Scanning Confocal Microscope.
Microglia cells (CD88+/+ or CD88-/-) were cultured (20,000 cell/ml) on glass coverslips precoated for 1hr with 20μg/ml poly L-lysine for 3days and then fixed with 4% formaldehyde in PBS. Neurons (CD88+/+ or CD88-/-) (50,000-75,000 cells/ml) were cultured for 7 (immature) or 14 days (mature) on glass coverslips coated with poly-L-lysine (1mg/ml). Fixed cells were incubated with blocking solution (2%BSA/PBS) for 1h followed by rat anti-mouse CD88 (clone 10/92 Serotec) at 1 or 5 μg/ml or rat IgG at the same concentrations for 1 h at RT. After washing with PBS cells were incubated with biotinylated anti-rat IgG (1:200) 1h, RT followed by SA-Cy3 or SA -Alexa555 (1:200) for 60’RT. Coverslips were mounted with Vectashield (Vector).
RNA from cortical tissue, snap frozen in liquid nitrogen, or from primary mouse microglia or neurons, was purified using the RNeasy Mini Kit (Qiagen), according to manufacturer’s recommendations. An additional DNase treatment was performed using an RNase-free DNase (Qiagen). RNA (250ng) was then transcribed into cDNA by the MMLV-RT reverse transcriptase (RT) kit (Invitrogen), according to manufacturer’s recommendations. Primers for GAPDH were designed using the primer3 tool (http://frodo.wi.mit.edu/cgibin/primer3/primer_www.cgi). Primers for CD88 were previously reported in (Van et al. 2000).
Flow cytometric analysis of primary microglia isolated from neonates of C57BL/6 but not CD88-/- mice were reactive with clone 10/92 anti-CD88, demonstrating that the antibody recognizes surface expressed microglial CD88 (Figure 1A). Similarly, Western blot analysis of cell lysates verified the presence of a band at 45kDa in wild type but not CD88-/- microglial cultures (Figure 1B). In a third assay, primary microglia and neurons were plated on coverslips, fixed and stained with clone 10/92 antibody. As observed in the above experiments, the clone 10/92 anti CD88 antibody reacted with wild type (Figure 1F, see arrow heads), but not CD88-/- microglia (Figure 1D). In contrast, cultured neurons from both wild type and CD88-/- mice were negative for reactivity with this antibody (data not shown). No 45kDa band was detected in cell extracts from neurons cultured from wild type or CD88-/- neurons for either 7 days (immature, synaptophysin negative, Figure 1B) or 14 days (mature, synaptophysin positive, data not shown). In similar analysis with primary cultured microglia, the polyclonal C1150-32 antibody (BD) reacted only with microglia cultured from wild type but not CD88-/- mice by flow cytometric and Western blot analysis. However, C1150-32 antibody did not react by IHC with wild type microglia plated on coverslips and also failed to detect a 45kDa band in cultured neuronal cell extracts (data not shown). To determine if the lack of immunoreactivity in these wild type cultured neurons was due to cell type specific epitopes, both microglia and neurons were tested for the expression of CD88 mRNA in the absence of any stimulatory factors. Only microglia from wild type mice were found to express CD88 mRNA (Figure 1G). CD88 mRNA was not detected in either 7day immature (Figure 1G) or 14 day mature (data not shown) primary neuron cultures.
In a first attempt to assess CD88 in murine brain, immunohistochemical analysis of brain tissue was performed utilizing polyclonal rabbit anti-mouse CD88-peptide antibodies from BD Pharmingen (C1150-32). Surprisingly, widespread neuronal staining in the brains of all the mice tested was detected although reactivity was stronger in Tg2576 mice (data not shown). However, staining was also seen in CD88-/- brain of the same age, in contrast to the lack of reactivity of isolated primary neurons from both wild type and CD88-/- embryos with C1150-32 antibody as described above and in Figure 1. In addition, Western blot analysis of brain extracts from both wild type and CD88-/- mice, contain a band around 45kDa, the predicted size of mouse CD88 (data not shown). The positive staining observed in the CD88-/- mice in these two assays suggests that these antibodies can cross react with a protein other than CD88 within brain tissue.
The monoclonal antibody against mouse CD88 (clone 10/92, Serotec) was also tested by IHC in brain sections of WT, CD88-/- and Tg2576 mice. Although some light neuronal background staining above the “no primary antibody” control was seen in wild type, CD88-/- and Tg2576 brain as observed with the C1150-32 antibody, in contrast to the C1150-32 antibody cells surrounding plaque structures in Tg2576 and Arc48 mice, that morphologically resembled activated microglia, were dominantly stained with clone 10/92 antibody (Figure 2C, arrowheads and Figure 2G) when compared with the neuronal staining. No such microglial labeling was detected in WT or CD88-/- mice (Figure 2A,B). Importantly, while staining was prominent in Arctic mice that were sufficient for CD88, no reactivity was seen in CD88 knockout Arctic mice (Figure 2H-I), even though Iba-1 positive microglia surrounding plaques were present in the CD88 knock out Arctic mice (data not shown).
These results suggest that the dominant epitope recognized by the clone 10/92 anti-CD88 antibody was CD88 expressed on microglia. However, Western blot analysis of brain extracts using clone 10/92 showed a 45kDa band from CD88-/- mice as well as wild type and Tg2576 transgenic mice (Figure 2E). To determine if CD88 in some form in the brain was recognized by these antibodies, brain extracts from wild type mice were immunoprecipitated with the BD (C1150-32) rabbit anti-CD88 or control IgG and the immunoprecipitants were subjected to SDS-PAGE, transferred to PVDF, and the blots probed with clone 10/92 anti-CD88. Figure 2F demonstrates that clone 10/92 anti-CD88 antibody detected a 45kDa band in the C1150-32 anti-CD88 immunoprecipitants from wild type mice but not CD88-/- brain tissue. No such band was immunoprecipitated by rabbit control IgG from the wild type mice nor CD88-/- mice. This result suggests that the clone 10/92 (Serotec) and the C1150-32 (BD Pharmingen) antibodies recognize distinct CD88 epitopes, and that the reactivity seen in the Western blots of whole brain extracts of CD88-/- mice represents background or cross reactivity with other molecules present in brain tissue that were eliminated by immunoprecipititation of the CD88 receptor from total brain lysates.
Having defined the specificity of the anti-CD88 clone 10/92 antibody, the expression of CD88 in normal mice and the influence of age on that expression was investigated in wild type and AD mouse models (i.e. mice that express mutated human APP in an age dependent fashion). Two models were used to distinguish differences due to age and pathology. As has been previously reported (Hsiao et al. 1996), Tg2576 mice start to show a few fAβ plaques, as represented by thioflavine reactivity, within the neocortex at 10 months (Figure 3A). While these plaques are absent at early time points (3 months), by 15 months, thioflavine positive plaques were plentiful in the neocortex, as well as within the hippocampal formation (data not shown). In contrast, Arc48 mice develop fAβ plaques at a much faster rate, with fibrillar plaques evident by 2 months of age within the hippocampal formation, and increasing substantially in neocortex and hippocampus at 6 and 13 months of age (Figure 3B) and as reported (Cheng et al. 2004). Both Tg2576 (Figure 3A) and Arc48 (Figure 3B) mice demonstrated increases in total CD88 expression that correlated with increases in plaque size and number. Reactivity was primarily localized to cells surrounding the fAβ. In contrast, wild type mice contain no thioflavine reactivity and no CD88 expression by microglia (data not shown).
In order to define the cell type responsible for the increased CD88 reactivity observed surrounding fAβ plaques, double immunofluorescent staining was performed in 15 month Tg2576 and 6 or 9 month Arc48 mice. Antibodies against CD88 and glial fibrillary acidic protein (GFAP), as a marker for reactive astrocytes, revealed that astrocytes were not the source of CD88 expression surrounding plaques in either the Arc48 mice (Figure 4A, ,5B)5B) or Tg2576 mice (data not shown). In contrast, double labeling for CD88 and Iba-1, as a marker for microglia, revealed co-localization of CD88 and Iba-1 in the area surrounding thioflavine (fibrillar) Aβ plaques in both AD mouse models, Tg2576 (Figure 4B) and Arc48 (Figure 5A). Iba-1 immunostained microglia distant from the plaque region were not reactive with anti-CD88 (data not shown) and indeed, not all microglia surrounding the plaques were observed to be positive for CD88 Figure 4B). Interestingly, confocal analysis of CD88/Iba-1 co-localization suggests that CD88 is polarized to the microglial processes that are in close proximity to the plaque (Figure 5A). Double staining with anti-CD88 and the microglial activation marker CD45 demonstrated that CD88 also co-localizes with CD45-positive microglia (Figure 4C).
We have previously demonstrated that Tg2576 mice treated with the CD88 antagonist PMX205, from 12 to 15 months of age resulted in significant decreases in several pathological markers associated with these transgenic mice. However, while there are fewer plaques in the antagonist-treated animals and correspondingly fewer microglia, the CD88-reactivity of microglia surrounding the Aβ plaques in PMX205-treated animals was similar to the untreated Tg2576 mice (Figure 6).
Co-localization of C1q with thioflavine in brain of Tg2576 mice treated for 12 weeks (from 12-15m) with PMX205 in the drinking water (20μg/ml) showed that C1q deposition correlates with plaque area (thioflavine) and that this correlation is not altered by PMX205 treatment (Figure 7). Similarly, immunostaining with an antibody against C4 (that recognizes C4, C4b and C4d) showed that C4 association with plaques and oligodendrocytes surrounding plaques was evident in the PMX205-treated animals, again correlating with fibrillar plaque load (data not shown).
Assessing a previously reported correlate of pathology in this Tg2576 model (Zhou et al. 2008), co-localization of C3 staining with GFAP expression (Figure 8) was detected in a subset of C3 positive astrocytes associated with the plaques in both untreated and PMX205-treated mice. As mentioned above, the extent of reactive glia was diminished by treatment in parallel with plaque size and number.
CD88 was the first identified receptor for the complement activation fragment C5a, a potent immune and inflammatory mediator. Recent work from our lab, suggests that signaling through CD88 may play a detrimental neuroinflammatory role in AD. After defining the specificity of a monoclonal rat anti-mouse CD88 clone 10/92 antibody using CNS derived cells from wild type and CD88-/- mice, we demonstrated increased expression of the receptor primarily in the vicinity of thioflavine positive Aβ plaques in murine transgenic models of AD. CD88 immunoreactivity was associated with Iba-1 and CD45 positive cells, indicative of microglia, and localized predominantly to the processes of plaque associated microglia and polarized toward the Aβ plaque. These observations are consistent with the proposed chemotactic response to C5a and the potential for costimulation of CD88 and receptors for plaque associated components (Zhang et al. 2007;Walter et al. 2007;Jana et al. 2008). It was also determined that in mice treated with the specific CD88 antagonist, PMX205, microglia surrounding plaques continue to express CD88 without noticeable up- or down-regulation of receptor density. Finally, no alterations in C1q association with plaques or C3 induction in astrocytes were detected after prolonged treatment with PMX205, suggesting that any beneficial functions of these early complement components may remain undisturbed due to their continued production during treatment with the CD88 antagonist.
Thus far, there have been conflicting results reported regarding CD88 expression in the AD brain. O’Barr et al, using a polyclonal rabbit anti-human CD88 antibody, directed against the N-terminus of CD88, reported comparable CD88 expression in neuronal cell bodies in both normal and AD human brain (O’Barr et al. 2001). In contrast, using both rabbit polyclonal and mouse monoclonal anti-human CD88 antibodies, Farkas and colleagues reported that neuronal CD88 expression was reduced in the human AD brain (Farkas et al. 2003). The reduction was restricted to neurons, and observed in a broad range of brain areas, although reactivity was concentrated around Aβ plaques (at the time hypothesized to be due to dystrophic neuritis, not microglia). No analysis of mRNA was reported in this study. Similar expression of CD88 protein was reported in non-transgenic and Tg2576 mice, using a polyclonal rabbit anti-mouse CD88 antibody directed against the N-terminus, and neuronal CD88 mRNA expression was also reported using in situ hybridization analysis in wild type mice (O’Barr et al. 2001). During our initial experiments comparing CD88 protein levels in Tg2576, Arc48 mice, and non-transgenic mice using a polyclonal rabbit anti-mouse CD88 antibody also developed against the N-terminus of murine CD88 (C1150-32, BD Pharmingen), we observed an increase in neuronal CD88 in the Tg2576 mice, relative to our non-transgenic controls (data not shown). However, control analysis in CD88-/- brain tissue revealed a similar neuronal staining pattern to that found in our non-transgenic mice. Thus, no precise conclusion as to the identity of the increased staining with the C1150-32 antibody in our Tg2576 mice could be made. Reports of other inconsistencies in the protein expression pattern for CD88 detected with both polyclonal and monoclonal antibodies directed against the N-terminus, have been previously noted (Kiafard et al. 2007). In that study, the authors present data demonstrating no antibody reactivity in mouse renal tubular cells using clone 10/92, which had previously been reported to express CD88 using a polyclonal anti-CD88 antibody (Bao et al. 2005). Cells of myeloid lineage, such as macrophages, however, were found to express CD88 using the clone 10/92 antibody. While the above reports did not validate the specificity of their staining using CD88-/- mice, in our study here, the specificity of the monoclonal anti-CD88 clone 10/92 antibody was demonstrated by positive reactivity in flow cytometric, western blot analysis and immunocytochemistry on murine primary microglia cells derived from wild type mice, but lacking in microglia derived from CD88-/- mice. Using the clone 10/92 we also showed intense staining on microglia in the area of fAβ plaques in AD mouse models, in line with previous reports demonstrating antibody reactivity in cells of myeloid lineage (Kiafard et al. 2007;Tschernig et al. 2007). Further evidence for microglial CD88 protein expression, the C1150-32 anti-CD88 antibody was also able to distinguish between wild type and CD88-/- microglial CD88 expression by flow cytometric and Western blot analysis, but had no reactivity in immunocytochemistry. RT-PCR analysis, verified that our cultured microglia were also expressing CD88 mRNA, which was absent in CD88-/- microglia and in both wild type and CD88-/- immature and mature cultured cortical neurons. One explanation for the lack of CD88 mRNA observed in our neuronal cultures, in contrast to that shown by others, could be a lack of extracellular signals, perhaps coming from surrounding glia, necessary for CD88 mRNA expression. Although mRNA for CD88 has been demonstrated in neurons in vivo, the levels observed in the absence of CNS insult, have been generally very low (Paradisis et al. 1998;Stahel et al. 1997). We did observe increased CD88 mRNA in our AD mice (data not shown) using RT-PCR of whole brain extracts, however, in situ hybridization will be necessary to determine if neurons contribute to the increase in mRNA in these animal model of neurodegeneration. Lack of specificity of antibodies against another CNS receptor (a7nAcR) was also observed using knock out mice (Moser et al. 2007;Herber et al. 2004), demonstrating that this situation is not unique to CD88 and stresses the importance of validating immunoreactivity results using knockout animals where possible.
Previous studies have shown that recombinant human C5a is chemotactic for both microglia and astrocytes in vitro (Yao et al. 1990). In this study clone 10/92 anti-CD88 antibody did not immunolabel astrocytes in either wild type or AD mice. This is surprising given our previous studies in a rat model of amyotrophic lateral sclerosis, which showed strong expression of CD88 on astrocytes from SOD1G93A transgenic rats using an anti-rat CD88 monoclonal antibody (Woodruff et al. 2008) as well as the results of others demonstrating astrocyte expression of CD88 (Gasque et al. 1997;Lacy et al. 1995). The lack of anti-CD88 reactivity on astrocytes (as well as neurons, in culture) could be a difference in conformation of the receptor between those cell types and microglia, although there have been no reported cell specific differences in posttranslational modification of CD88. Alternatively, since CD88 has been reported to associate with other receptors, astrocytes could lack a co-receptor that may associate with CD88 in microglia or express a surface molecule that masks the CD88 epitope recognized by the clone 10/92 anti-CD88 antibody. These results taken together with the existence of a second C5a receptor, C5L2, suggest that caution is needed when using antibody reactivity to assess CD88 protein levels in the murine CNS or interpreting previous reports based on antibody reactivity or C5a-induced responses. Alternative detection methods, such as gene silencing or in situ hybridization, are necessary unless specificity of the antibody being used is demonstrated using CD88-/- tissue or cells. Furthermore, additional experimental approaches (novel specific anti-CD88 antibodies) are needed to further establish CD88 expression in AD models.
Previously, we reported that the administration of PMX205 resulted in reduced GFAP and CD45 reactivity (markers for activated astrocytes and microglia, respectively) as well as decreased plaque and insoluble Aβ42 loads in two distinct AD mouse models (Fonseca et al. 2009). The cyclic hexapeptide C5a receptor antagonist was administered at an age at which fAβ deposition is known to occur, and thus, at a time when C5a is likely generated by complement activation. The initial interaction of C5a with CD88 and Aβ with TLR2 (or 4) could then synergistically induce TNF-α (Jana et al. 2008), which can be directly neurotoxic, and may also upregulate CD88 on glia. This enhanced CD88 signaling would lead to greater TNF-α production and thus acceleration of pathology. However, other possibilities exist. For example, if a form of CD88 is upregulated on neurons as a result of the accumulating injury; PMX205 may also be directly neuroprotective (Farkas et al. 1998) and/or result in decreased amyloid peptide production or tau phosphorylation which would prevent/slow the progression of cognitive decline. Further study will be needed to resolve the precise mechanism of PMX205 action in this and other neurodegenerative diseases.
In any event, the specific inhibition of CD88 signaling would nevertheless allow the progression of other previously demonstrated and hypothesized protective effects of complement activation up-stream of C5a generation (Pisalyaput and Tenner 2008;Wyss-Coray et al. 2002;Maier et al. 2008). When in complex with the proenzymes C1r and C1s in C1, C1q binds fAβ and thus activates the classical complement cascade (Rogers et al. 1992), which can in turn generate C5a. However, C1q alone has been demonstrated to have multiple effects in other disease states, including the ability to enhance phagocytosis of apoptotic cells (Botto et al. 1998) and suppress proinflammatory cytokines (Fraser et al. 2009) and the ability to provide direct protection to injured neurons (Pisalyaput and Tenner 2008). Since C1q is associated with thioflavine positive fAβ plaques even following PMX205 treatment, our detection of its continued association with plaques in the presence of PMX205 demonstrates that it is possible that C1q may continue to play a protective role in AD by mediating clearance of cellular debris and preventing neuronal damage. Similarly, it has been suggested that C3 cleavage products may facilitate removal of aggregated Aβ, as inhibition of C3 cleavage through the over-expression of the complement receptor related protein y (Crry) or complete C3 genetic deletion in AD mouse models resulted in a substantial increases in Aβ deposition (Maier et al. 2008;Wyss-Coray et al. 2002). In this study, we found that C3 production by astrocytes, which is elevated in AD mouse models (Zhou et al. 2008), is not substantially altered relative to the thioflavine plaque load with PMX205 treatment, and therefore, C3 as well as C1q may continue to provide neuroprotective functions during PMX205 treatment.
In summary, we have observed microglial upregulation of CD88 in response to amyloid deposition, a localization of receptor reactivity in microglia surrounding Aβ plaques and polarization of the receptor to microglial processes in closest proximity to the plaques. In addition, our data further supports the targeting of the anaphylatoxin receptor CD88 as a therapeutic approach to the treatment of AD, since detrimental activities are inhibited, while potentially protective activities of complement are preserved.
This work was supported by NIH grants NS35144 and AG 00538. The authors thank Naseem Rowther and Alice Berci for excellent technical assistance, Dr. Lennart Mucke, Gladstone Institute, for the Arctic48 mouse and Dr. Rick Wetsel (Univ. of Texas Medical School, Houston) for the C5aR-/- mice.