Cell-specific deletion of C1qa identifies microglia as the dominant source of C1q in mouse brain
© The Author(s). 2017
Received: 9 December 2016
Accepted: 7 February 2017
Published: 6 March 2017
The complement cascade not only provides protection from infection but can also mediate destructive inflammation. Complement is also involved in elimination of neuronal synapses which is essential for proper development, but can be detrimental during aging and disease. C1q, required for several of these complement-mediated activities, is present in the neuropil, microglia, and a subset of interneurons in the brain.
To identify the source(s) of C1q in the brain, the C1qa gene was selectively inactivated in the microglia or Thy-1+ neurons in both wild type mice and a mouse model of Alzheimer’s disease (AD), and C1q synthesis assessed by immunohistochemistry, QPCR, and western blot analysis.
While C1q expression in the brain was unaffected after inactivation of C1qa in Thy-1+ neurons, the brains of C1qa FL/FL :Cx3cr1 CreERT2 mice in which C1qa was ablated in microglia were devoid of C1q with the exception of limited C1q in subsets of interneurons. Surprisingly, this loss of C1q occurred even in the absence of tamoxifen by 1 month of age, demonstrating that Cre activity is tamoxifen-independent in microglia in Cx3cr1 CreERT2/WganJ mice. C1q expression in C1qa FL/FL : Cx3cr1 CreERT2/WganJ mice continued to decline and remained almost completely absent through aging and in AD model mice. No difference in C1q was detected in the liver or kidney from C1qa FL/FL : Cx3cr1 CreERT2/WganJ mice relative to controls, and C1qa FL/FL : Cx3cr1 CreERT2/WganJ mice had minimal, if any, reduction in plasma C1q.
Thus, microglia, but not neurons or peripheral sources, are the dominant source of C1q in the brain. While demonstrating that the Cx3cr1 CreERT2/WganJ deleter cannot be used for adult-induced deletion of genes in microglia, the model described here enables further investigation of physiological roles of C1q in the brain and identification of therapeutic targets for the selective control of complement-mediated activities contributing to neurodegenerative disorders.
KeywordsC1q Alzheimer’s Expression Microglia Mouse model Complement Conditional knockout
The complement system provides rapid recognition and response to danger that threatens the host. The complement cascade is regulated by the local environment  and synergizes with other responses to infection or stress or both [2–4]. Complement activation is carefully controlled, and imbalanced activity is associated with many degenerative diseases including rheumatoid arthritis, nephritis, traumatic brain injury, and age-related macular degeneration (reviewed in [5, 6]). The complement system also functions during development with complement-dependent synapse pruning being an integral neurodevelopmental process, at least in the brain regions such as the retinogeniculate system [7, 8]. While beneficial during development, recent reports suggest that unregulated complement-mediated synapse pruning is associated with behavioral alterations in mouse models of AD , aging , frontotemporal dementia , and virus infection . Further delineation of the molecules in the pathway that can be most strategically targeted for therapeutic intervention in such disorders requires precise information about unintended consequences of inhibition of the entire pathways and clarification of the sources of various complement components that would enable more selective targeting to prevent or slow neurodegenerative diseases .
The complement system consists of over 30 independent proteins. C1q is the recognition component of C1, the classical complement pathway multi-subunit complex. C1q is a hexamer of trimers composed of three distinct polypeptide chains, A, B, and C , with each chain required for proper assembly of C1q . When complexed with C1r and C1s, the enzymatic components of the C1 complex, C1q binding to an activator initiates a cascade of enzymatic reactions resulting in the opsonization of the activating substance, generation of the activation peptides C3a and C5a, and formation of a membranolytic pore. Binding of C3a and C5a to their cellular receptors contributes to inflammation . However, C1q, independent of C1r and C1s, can enhance clearance of apoptotic cells and cellular debris, downregulate proinflammatory cytokine expression by phagocytes in vitro [17–19], and protect neurons from nutrient stress and fibrillar amyloid-induced damage [20, 21] consistent with a role in maintaining homeostasis. Indeed, synthesis of C1q is increased in response to a variety of neuronal injuries (reviewed in ).
To analyze the contribution of C1q to Alzheimer’s disease (AD) pathology, previous studies used a constitutive C1qa knockout in two mouse models of AD that demonstrated a ~50% decrease in glial activation markers and protection of neuronal integrity . Subsequent work treating mouse models of AD with a C5a receptor antagonist resulted in a 50–70% decrease in pathological markers suggesting that the detrimental effects of complement activation in the brain were predominantly mediated by the proinflammatory C5a fragment generated upon the activation of the complement cascade through the cleavage of C5 (to C5a and C5b) . We have previously shown that C1q synthesis is increased at an early age in AD mouse models (i.e., in the absence of fibrillar amyloid plaques), without a similar rise in C1r and C1s, which are induced only much later when the deposition of fibrillar amyloid plaques occurs . Thus, with differential synthesis of complement proteins in the brain, C1q could have beneficial effects during neurodevelopment as a component of C1, be directly neuroprotective early in disease processes independent of other complement activities, and contribute to complement-mediated inflammatory events accelerating progression of cognitive loss at later stages of disease, i.e., when C1r, C1s and other complement pathway proteins are present.
To further delineate the physiological roles of C1q in the brain, including the influence of age, injury, and/or the cellular source of C1q, new models enabling defined cellular and temporal control of C1q expression are needed. Specifically, generation of mice in which C1q is present for the beneficial and protective activities during development and early stages of injury, but which could be ablated at later ages to suppress detrimental activities resulting from activation of the complement cascade would be useful. In addition, while there is evidence of C1q mRNA synthesis in the brain [24–26], the cellular source of C1q within the brain (neurons or microglia), and the contribution of the central nervous system (CNS) versus periphery to the production of functional C1q protein in the brain with aging and stages of disease have remained unclear. The ability to regulate detrimental responses in the brain while retaining the protective functions of the complement system in the peripheral would be beneficial particularly for long-term treatment of chronic diseases. These issues can be investigated using mice in which Cre recombinase activity can be induced in a time and cell-type specific manner. Here, we characterized C1q expression in mice homozygous for a loxP-flanked C1qa chain gene crossed to mice containing tamoxifen-inducible cre (Cre-ERT2) under control of the Cx3cr1, Thy1, or Rosa26 promoters. The results, while revealing a caveat in the use of Cx3cr1 CreERT2 , demonstrate that microglia, not neurons or peripheral sources, are the dominant source of C1q in a healthy brain at all ages up to 10 months and in an Alzheimer’s disease transgenic model.
Generation of mouse models
All animal experimental procedures were approved by the Institutional Animal Care and Use Committee of University of California, Irvine, and performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals. The C1qa tm1a(EUCOMM)Wtsi “reporter-tagged insertional with conditional potential” allele (hereafter referred to as C1qa GT/+ ) is a gene-trap allele in which a Frt’d splice-acceptor–lacZ–polyadenylylation (polyA) cassette and PGK-neo/G418 resistance cassette is inserted within intron 2 of C1qa (see https://www.mousephenotype.org/data/genes/MGI:88223) (obtained from Jackson Laboratory, Bar Harbor, ME). This allele is predicted to be null for C1qa function. To analyze lacZ (β-galactosidase; β-gal) reporter gene expression from the targeted allele of C1qa, mice with a C1qa GT/+ allele were crossed with B6.FVB-Tg(Stra8-cre)/1Reb/LguJ mice (Jackson, stock #017490). This removes the PGK-neo cassette that is flanked by loxP sites, to avoid possible interference of the PGK-neo cassette with transcription of the C1qa-encoded lacZ reporter gene. Offspring with the C1qa tm1b(EUCOMM)Wtsi “reporter-tagged deletion” allele, hereafter referred to as C1qa GT-neo/+ , retain one wild type C1qa allele to prevent developmental abnormalities that may result from a complete deletion of C1q during development. To generate mice with a cre-conditional (“floxed”) allele, C1qa GT/+ were crossed with B6.Cg-Tg(ACTFLPe)9205Dym/J (Jackson, stock # 005703) mice to produce mice with a C1qa tm1c(EUCOMM)Wtsi allele, hereafter referred to as C1qa FL/+ . In the C1qa FL/+ allele, the coding sequence in exon 3 is flanked by two loxP sites, one within intron 2 and the other in the 3′ untranslated region of exon 3. To generate mice for tissue- and temporal-specific knockout of C1qa, C1qa FL/FL mice were crossed with B6.129P2(Cg)-Cx3cr1 tm2.1(cre/ERT2)Litt /WganJ, (Jackson, stock #021160) mice, here designated as Cx3cr1 CreERT2/WganJ or Cx3cr1 CreERT2 , which co-expresses Cre-ER fusion protein and EYFP in monocytes, dendritic cells, NK cells, and brain microglia , or Tg(Thy1-cre/ERT2,-EYFP)HGfng/PyngJ, (Jackson, stock #012708), here referred to as Thy1 CreERT2 which expresses Cre-ERT2 fusion protein and EYFP in Thy1-expressing neurons, or B6.129-Gt(ROSA)26Sor tm1(cre/ERT2)Tyj /J, (Jackson, stock #008463), here designated as Rosa26 CreERT2 mice which expresses Cre-ERT2 fusion protein ubiquitously. All mice were on the C57BL6/J background (i.e., Nnt −/− ). Progeny from each cross were backcrossed with C1qa FL/FL mice to generate C1qa FL/FL :Cx3cr1 CreERT2 , C1qa FL/FL :Thy1 CreERT2 , and C1qa FL/FL :Rosa26 CreERT2 animals. In addition, Cx3cr1 CreERT2 were crossed to B6;129S6-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J (Jackson, stock #007908, also called Ai14(RCL-tdT)-D), which we refer to as ROSA26-STOP-tdTomato, to generate ROSA26-STOP-tdTomato +/−: Cx3cr1 CreERT2 mice. As a sensitive cell lineage reporter of Cre activity, cells with a ROSA26-STOP-tdTomato allele do not express tdTomato until after the floxed stop cassette has been excised by Cre recombinase activity. Arctic48 mice were obtained from Dr. Lennart Mucke (Gladstone Institute, San Francisco, CA, USA). Constitutive C1qa knockout mice  were originally a gift from Dr. Marina Botto, Imperial College London.
Induction of Cre activity with tamoxifen
Tamoxifen (T5648; Sigma, St. Louis, MO) was dissolved in 5% ethanol/corn oil at final concentration of 50 mg/ml. Mice were treated with tamoxifen at 0.2 mg/g body weight or vehicle control by oral gavage once a day for five consecutive days . Animals were bled, and tissue was harvested at 3–56 days after treatment as noted.
Collection of tissue
Tissue was processed as previously described . Briefly, mice were anesthetized with isoflurane, perfused with PBS, and the brains were collected and fixed for 24 h in 4% paraformaldehyde/PBS. To detect C1q in interneurons, mice were perfused with PBS, followed by 4% paraformaldehyde/PBS for 2 min. Tissue was then post-fixed for only 2 h in 4% paraformaldehyde/PBS at 4 °C. For some mice, the liver and kidney were also collected.
Immunostaining procedures were performed as described . Briefly, 40-μm brain or 30-μm liver and kidney were incubated overnight at 4 °C with primary antibodies or corresponding control IgG. Primary antibodies were detected with Alexa555 or Alexa488 labeled secondary antibodies (Invitrogen, Carlsbad, CA). Antibodies used were anti-mouse C1q (rabbit monoclonal, clone 27.1) tissue culture supernatant (previously generated and characterized in collaboration with the Barres lab , anti-Cre recombinase (mouse monoclonal ascites, clone 2D8, 1:1000, Millipore, Temecula, CA), anti GAD67 clone (mouse monoclonal clone 1G10.2, 1:700, Millipore ), anti-parvalbumin (mouse monoclonal, clone PARV-19, 1:300, Sigma), anti-somatostatin (rat monoclonal, clone YC7, 1:200, Millipore), anti-F4/80 (rat monoclonal clone CI:A3-1, 0.5 ug/ml, AbD Serotec), anti-CD31 (rat monoclonal, clone SZ31, 1 ug/ml, Dianova, GmbH, Germany), and anti-Iba-1 (rabbit polyclonal, 1 ug/ml, Wako, Richmond, VA). For colocalization experiments of C1q with other markers, primary antibodies were incubated simultaneously (with anti-Cre recombinase, F4/80, CD31, somatostatin, or parvalbumin) or sequentially (after anti-GAD67) followed by secondary antibodies. For immunoperoxidase staining of C1q, CD45 or amyloid ß, sections were pretreated with 3%H2O2/10%MeOH/Tris-buffered saline (TBS), pH 7.4 to block endoperoxidase. After blocking with 2%BSA/10% normal goat serum/0.1%Triton/TBS, sections were incubated with anti-mouse CD45 (goat polyclonal 1 ug/ml, R&D) or anti-amyloid ß (#1536, 1:1,000, gift of Dr. Cooper ) in blocking solution or anti-mouse C1q clone 27.1, overnight at 4 °C. After incubation with primary antibody, sections were incubated with the corresponding biotinylated secondary antibodies (Vector Labs, Burlingame, CA) (1 h, RT) followed by ABC (Vector Labs) (1 h, RT) and developed with DAB (3,3′-diaminobenzidine) (Vector Labs) following manufacturer instructions. Tissue was dehydrated and mounted with DePeX (BDH Laboratory Supplies, Poole, England). Immunostaining was analyzed using a Zeiss Axiovert-200 inverted microscope (Zeiss, Thornwood, NY) and images acquired with a Zeiss Axiocam high-resolution digital camera (1300x1030 pixel) using Axiovision 4.6 software. Confocal images were acquired using a Zeiss LSM 700 or 780 confocal microscopes and Zen software. Images of staining were acquired with the same exposure times and camera settings when comparing the same brain area in the different genotypes. The percentage of C1q-positive microglia (yellow) over total microglia (green + yellow) per animal was obtained by scoring the number of yellow and green microglia per field and averaging all five fields per section (two sections per animal). For quantitative analysis of C1q in the molecular layer of hippocampus, images were analyzed using ImageJ software. Four regions of interest (ROI) (squares) were defined randomly within the molecular layer area and the mean pixel intensity per ROI was determined. The mean intensity of each animal was obtained by averaging all ROI mean intensities in the sections studied (two–three sections per animal). For CD45 and Aß, images were analyzed using Axiovision 4.6 sofware. Percentage of immunopositive area (immunopositive area/total image area times 100) was determined by averaging several images per section (n = 2–3 sections per animal) that covered the hippocampal area. One way ANOVA statistical analysis was used to assess the differences in C1q mean intensity, plaque area, and glial reactivity.
Detection of lacZ (ß-gal) expression by X-Gal histochemistry was done using a staining kit (Life Technologies, Carlsbad, CA). Sections were incubated with a final concentration of 1 mg/ml X-gal, 4 mM potassium ferricyanide, 4 mM potassium ferrocyanide, and 2 mM magnesium chloride, with or without 25 μM sodium deoxycholate at RT or 37 °C for different times (between 30 min and 24 h). Detection of ß-gal by immunostaining was performed using a rabbit polyclonal anti-ß-galactosidase (0.1–10 ug/ml) (Life Technologies) or chicken polyclonal anti-ß galactosidase (1:100–1:5000, Abcam, Cambridge, MA).
Microglial isolation, sorting, and immunocytochemistry
Primary neonatal microglia were isolated as described  from postnatal day 2 pups from a C1qa FL/FL :Cx3cr1 CreERT2 animal crossed with C1qa FL/FL . Cells were stained with an Allophycocyanin (APC)-conjugated anti-mouse CD11b. Viable cells were gated according to size and granularity (based on forward and side scatter properties), followed by gates for YFP and CD11b (FACS Aria II). The sorted populations, (1) YFP+, CD11b+; (2) YFP−, CD11b+; and (3) YFP−, CD11b−, were plated on poly-l-lysine coated coverslips for 1 h. After blocking with 2% BSA/0.1% Triton/PBS for 30 min and incubating with or without rabbit monoclonal anti-C1q antibody (clone 27.1) for 1 h at RT, followed by Alexa 555 goat anti-rabbit antibody (1 h, RT), coverslips were mounted with antifade reagent containing DAPI (Prolong gold, Invitrogen). No APC or Alexa 555 signal was detected when the primary anti-C1q antibody was omitted.
Plasma was collected into EDTA (10 mM final) immediately preceding perfusions by cardiac puncture. The blood was withdrawn from the left ventricle with a 25-gauge needle and transferred to a prechilled glass tube on ice. After centrifugation at 2000×g for 10 min at 4 °C, the plasma was removed and immediately distributed into small aliquots and stored at −80 °C. Two microliters of mouse plasma diluted in GVB (4.8 mM sodium barbital, 0.1% gelatin, 142 mM NaCl, 1 mM MgCl2, and 0.15 mM CaCl2 at pH 7.3) was combined with human C1q-deficient serum (300 μl). Eighty microliters of sheep erythrocytes (Colorado Serum Co., Denver, CO) previously sensitized with rabbit anti-sheep erythrocyte immunoglobulin (EA) (5 × 108 EA/ml) was added and incubated at 37 °C for 30 min. After addition of 1.6 ml ice-cold GVB, reactions were centrifuged at 800×g at 4 °C for 3 min and the optical density at 412 nm of the supernatants were obtained as an index of released hemoglobin. The functional activity of mouse C1q was determined by analysis of hemolytic activity derived from the means of duplicate serial dilutions of each sample as previously described .
Each frozen half-brain or hippocampus was pulverized, aliquoted, and stored at −80 °C. Brain powder was solubilized by homogenization in ten volumes of Tris-buffered saline (TBS, pH 7.4) containing protease inhibitor cocktail solution (Complete mini, Roche), 1 mM EDTA, and 2% SDS using motor pestle for 5 s twice on ice. After centrifugation at 18,400×g for 30 min at 4 °C, protein concentration in the supernatants was determined with BCA protein assay (Pierce, Rockford, IL). Sixty μg of brain protein or 1 ul plasma were subjected to SDS-polyacrylamide gel electrophoresis (10%) under reducing conditions. Gels were transferred at 4 °C in Tris-glycine, SDS, and 10% methanol transfer buffer onto polyvinylidenedifluoride (PVDF, Immobilon-P, Millipore) membrane at 300 mA for 2 h. PVDF membrane was blocked 1 h at RT in 5% nonfat dry milk in TBS-Tween. Rabbit anti-mouse C1q (1151)  or mouse anti-ß-actin (Sigma) in either 5 or 3% milk respectively was applied overnight at 4 °C or 2 h at RT. After washing, HRP-conjugated secondary antibodies diluted 1:5000 (Jackson Labs), were added and incubated for 1 h at RT. The blots were developed using ECL 2 or ECL (Pierce) and analyzed using a Nikon D700 digital camera and the ImageJ software as described .
RNA extraction and qRT-PCR
Total RNA from pulverized mouse brain (5–10 mg) was extracted using the Illustra RNAspin Mini Isolation Kit (GE Healthcare). cDNA synthesis was performed using SuperScript III reverse transcriptase (Life Technologies) according to the manufacturer’s protocol. Quantitative RT-PCR was performed using the iCycler iQ and the iQ5 software (Bio-Rad) with the maxima SYBR/Green Master Mix (Thermo Fisher Scientific). The mouse primers for C1qa , C1qb, and C1qc genes  and Hprt primers were obtained from Eurofins MWG Operon (Huntsville, AL). Hprt primers were designed using primer-blast (http://www.ncbi.nlm.nih.gov/tools/primer-blast/), with the forward sequence (5′-3′): AGCCTAAGATGAGCGCAAGT and reverse sequence (5′-3′): ATCAAAAGTCTGGGGACGCA. The relative mRNA levels were determined as follows: mRNA levels = 2-ΔCt and ΔCt = (CtTarget−CtHPRT). Ct values represent the number of cycles at which fluorescence signals were detected .
C1q in the brain is colocalized with microglia, neuropil, and interneurons but not Thy1+ neurons.
Initial attempts to use the lacZ reporter detection in the C1q “reporter-first” gene-targeted mouse failed to provide a signal in the brain (Additional file 1), and thus, the monoclonal anti-C1q antibody clone 27.1 was used to analyze C1q expression. As reported , this antibody stained microglia, interneurons, and was particularly strong in the molecular layer of hippocampus (Additional file 1: Figure S1A–C). As a negative control for antibody specificity, tissue from a homozygous constitutive C1qa knockout animal was routinely stained in parallel and showed complete lack of reactivity with this antibody (Additional file 1: Figure S1D).
C1q is unchanged in the brains of C1q FL/FL :Thy1 CreERT2 mice treated with tamoxifen, but is significantly reduced in the brain of C1q FL/FL :Cx3cr1 CreERT2 mice even without tamoxifen treatment.
While analyzing the impact of deleting the C1qa gene in microglia on total brain C1q protein, C1q staining in C1qa FL/FL :Cx3cr1 CreERT2 adult animals was seen to be very low or absent, even at the earliest timepoint after tamoxifen treatment (n = 15, data not shown). Surprisingly, vehicle treated (n = 9, data not shown) and the untreated C1qa FL/FL :Cx3cr1 CreERT2 animals (Fig. 1c, f) also lacked C1q reactivity throughout the brain as shown in both the molecular layer of the hippocampus (Fig. 1c) and in the cortex (Fig. 1f) in contrast to WT (Fig. 1a, d) or C1qa FL/FL littermate mice (Fig. 1b, e). This dramatic decline was confirmed by western blot of brain extracts in 1-and 2-month-old animals (Fig. 1g, h). C1q was detected in a subset of interneurons (GAD67+) in the C1qa FL/FL :Cx3cr1 CreERT2 mice, although the intensity was decreased in comparison to interneurons in the littermates lacking Cx3cr1 CreERT2 (Additional file 1: Figure S4A, right panel).
As complement-mediated synapse pruning has been demonstrated as a normal process in neurodevelopment, it is important to demonstrate if the C1qa gene is deleted or still present in neonatal microglia in C1qa FL/FL :Cx3cr1 CreERT2 mice to perform this beneficial function. Cells from the brains of postnatal day 2 litters containing both C1qa FL/FL :Cx3cr1 CreERT2 (YFP+) and C1qa FL/FL (YFP-) pups were sorted for CD11b (i.e., microglia) and YFP (indicating the presence of Cre). C1q staining was clearly detected in both CD11b+ populations (Fig. 4q, r) with an equal percentage of CD11b+ cells (microglia) expressing C1q (96%) in YFP+ (C1qa FL/FL :Cx3cr1 CreERT2 ) and YFP− (C1qa FL/FL ) microglia demonstrating that Cre deletion of the C1qa gene had not yet occurred in microglia from neonatal mice. All CD11b-negative cells lacked C1q reactivity as expected (Fig. 4s).
Upregulated C1q expression in the brain from aged mice and an Alzheimer’s disease mouse model is also abrogated in C1q FL/FL :Cx3cr1 CreERT2 mice independent of tamoxifen treatment.
C1q in the liver, kidney, and plasma is unaffected in C1qa FL/FL : Cx3cr1 CreERT2 mice.
Cre activity from the Cx3cr1 CreERT2 allele is “leaky” in the absence of tamoxifen in long-lived microglia.
Similar to the Thy1 CreERT2 mice, C1qa FL/FL mice crossed to Rosa CreERT2 (ubiquitously expressed Cre-ERT2) had no difference in the level of C1q in the brain in the absence of tamoxifen at any ages tested (2–6 m). Importantly, C1q was reduced in the brain after tamoxifen treatment of C1qa FL/FL : Rosa CreERT2/+ mice (Additional file 1: Figure S6). These results suggest the robust CreERT2 expression driven by the Cx3cr1 promoter compared to the relatively weak promoter of the Rosa26 locus that is responsible for the tamoxifen-independent gene deletion, and not an inherent property of C1qa FL/FL . To further test this prediction, we looked for evidence of Cx3cr1 CreERT2 -mediated loxP recombination using mice containing a different floxed sequence. The ROSA26-STOP-tdTomato allele has a “floxed” stop cassette upstream of a coding sequence for tdTomato, such that cells express tdTomato only after the stop cassette has been removed by Cre activity. Tamoxifen-independent tdTomato fluorescence was observed in microglia in ROSA26-STOP-tdTomato crossed to Cx3cr1 CreERT2 mice by 1 and 2 months of age, consistent with leakage of the recombinase activity to the nucleus in these cells (Additional file 1: Figure S7). The percentage of YFP-positive cells that were also tdTomato-positive cells was 60.15% +/−SE 2.6 (n = 2) at 1 month and 73.6%, +/−SE 0.9 (n = 2) at 2 months, compared to 0% of Rosa26-STOP-tdTomato mice littermate controls lacking the Cx3cr1 CreERT2 . Together, these results demonstrate that the tamoxifen-independent recombination of loxP sites mediated by the Cx3cr1 CreERT2 allele is not specific to the C1qa gene, and that this Cx3cr1 CreERT2 , but not all CreERT2 deleter mice show C1q gene ablation in microglia in the absence of the inducer tamoxifen. That is, Cx3cr1 CreERT2WgaJn is “leaky”, with Cre activity accessing the nucleus in the absence of tamoxifen, which compromises its use in studies involving postnatal temporal induction of Cre deletion of floxed genes.
Cell-specific Cre-recombinase-mediated ablation of the C1qa gene was used to demonstrate that microglia, but not Thy1+ neurons or peripheral blood, are the dominant source of C1q in mouse brain beyond one month of age. The data also indicate that the enhanced expression of C1q with aging and as seen in AD models is also predominantly, if not exclusively, microglial. Upon deletion of microglial C1qa, C1q is only detected at very low levels in subpopulations of interneurons in the brain. Given the known functions of C1q in clearing apoptotic cells while suppressing proinflammatory cytokine production (reviewed in ) and signaling direct neuroprotection , a dominant microglial synthesis of C1q is consistent with a reparative and protective role for microglia and the importance of an anti-inflammatory environment to maintain normal brain function. Yet, when infection or damage increases such that other complement proteins are produced, as occurs in neurodegenerative diseases, more powerful effector mechanisms can be enabled, including activation of the complement cascade with potential collateral progressive inflammatory polarization of microglia with damaging consequences.
The site of synthesis of complement proteins was traditionally believed to be in the myeloid compartment of the liver . Petry and colleagues demonstrated that wild type bone marrow transplanted into a constitutive C1q knockout mouse could completely restore the plasma compartment of C1q . However, in the past two decades there have been multiple reports of increases in C1q synthesis upon injury, particularly in the nervous system (reviewed in ) but also in other tissues [50, 51]. The upregulation of mRNA for complement components in the brain with age and in neurodegenerative diseases such as AD has been documented in both human [26, 52] and mouse models [53, 54] of disease. Evidence for both microglial and neuronal synthesis have been presented [7, 24, 55, 56]. Here, when C1qa FL/FL mice were crossed with animals expressing CreERT2 under control of the Cx3cr1 locus (Cx3cr1 CreERT2 ), within 1 month of age, most of the brain C1q was eliminated, while no alteration in C1q protein was detected in the brain after tamoxifen treatment of mice containing a transgene expressing a tamoxifen-inducible Cre ERT2 under the neuronal Thy1 promoter.
Surprisingly, the deletion of the C1qa gene in C1qa FL/FL : Cx3cr1 CreERT2 mice occurred even in the absence of the tamoxifen suggesting that the CreERT2 gains access to the nucleus independently of tamoxifen to cause the deletion. Similar tamoxifen-independent Cre-recombinase activity was demonstrated with a different floxed sequence in microglia from ROSA26-STOP-tdTomato:Cx3cr1 CreERT2 (Additional file 1: Figure S7). Thus, these data suggest that the Cx3cr1 CreERT2WganJ mouse does not maintain normal expression of floxed genes in microglia in the absence of tamoxifen, thereby limiting its use for inducible deletion of microglial genes. The original report describing this Cx3cr1-driven Cre deleter investigated only young Cx3cr1 CreERT2 mice treated with tamoxifen .
Effect of Promoter Specific CreERT2 Expression
Effect on target gene expressionb
Interestingly, recent studies have shown that activation of the C1 complex results in the covalent tagging of synapses with the activation fragments C3b and iC3b, marking them for elimination by microglia via the complement receptor CR3 . While initially discovered as critical for proper brain development (reviewed in ), overzealous or inappropriate “pruning” of synapses has now been associated with loss of function or alterations in behavior in mouse models of frontotemporal dementia , epilepsy , AD , and West Nile Virus infection . However, it is also well supported that the activation of the complement cascade can lead to the generation of C5a which can recruit glia, promote their proinflammatory polarization (reviewed in ), and could also be a cause of the loss of phenotypic functions  noted in these systems. Future studies will be important to determine the relative contribution of the increased synapse elimination versus complement activation leading to C5a-enhanced inflammation/injury to these behavioral impairments. In addition, since evidence of localized neuronal C1q has been reported, it remains to be tested if specific deletion of microglial C1q will abrogate complement-dependent synapse pruning or if local synthesis of C1q may be induced in neurons during development, infection or other disease states which we did not assess here, with physiological consequences.
The data presented here show that microglia, not peripheral blood or other CNS cells, are the overwhelming source of C1q in the murine brain from postnatal through the aging adult (at least though 10 months of age) and in one mouse model of Alzheimer’s disease, where C1q levels are normally dramatically increased with age and amyloid accumulation. Although these data demonstrate that loss of C1q in microglia occurs after 1 month of age with the Cx3cr1 CreERT2/WganJ strain  independent of tamoxifen and thus temporal induction is not possible, this cell-specific deleter strain of mice may be valuable to determine if the cellular source of C1q (and indeed any microglia-expressed gene) in vivo changes in various models of neurological disorders and will enable subsequent investigation of the beneficial or detrimental contribution of CNS C1q to the progression of neurodegenerative diseases, including cognitive decline. We also show that it is possible to deplete the brain of C1q selectively without depleting C1q in the blood, thus enabling the delineation of C1q-mediated events as being of CNS rather than peripheral origin. These results provide a characterized framework for further investigation of beneficial roles as well as detrimental effects of complement in the CNS (reviewed in ), clarification of the contribution of peripheral complement to neurodegenerative diseases [13, 61], determination of what controls the induction of C1q synthesis in the brain, and subsequent design of selective and specific therapeutic interventions for complement-mediated events in neurodegenerative diseases .
Central nervous system
The authors thank Dr. E. Monuki (UCI) for providing tissue from BRE-gal mice expressing BRE-gal lacZ reporter gene, Dr. L. Mucke (Gladstone Institute, San Francisco, CA, USA) for the Arctic48 breeding pair. Dr. S. Gandhi (UCI) for the Ai14(RCL-tdT)-D breeding pair mice and antibodies, and Dr. C. Hughes (UCI) for antibodies. The authors also thank Phuong (Remy) Vu, Eric Nguyen, Purnika Selvan, and Bethany McGonnigal (UCI) for excellent technical assistance.
This work was supported by NIH NIA grant AG 00538 and NIH NIA T32 AG000096.
Availability of data and materials
The relevant raw data are freely available.
MIF, SC, and AJT designed the experiments and overall project. MXH performed microglia isolation and contributed to Fig. 4. MIF, MJF, LM, and PS performed the immunohistochemistry, SC preformed western blots, hemolytic titers, and qPCR. GRM advised in the animal construction/breeding. SC, MJF, and MXH performed all the animal work and strain identification. MIF, SC, GRM, and AJT wrote the manuscript, and all authors reviewed the manuscript. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Consent for publication
The care and euthanization of animals were approved by the University of California, Irvine Institutional Animal Care and Use Committee, which is in agreement with the NIH Guide for the Care and Use of Laboratory Animals.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Sjoberg AP, Trouw LA, Blom AM. Complement activation and inhibition: a delicate balance. Trends Immunol. 2009;30(2):83–90.View ArticlePubMedGoogle Scholar
- Hajishengallis G, Lambris JD. Crosstalk pathways between Toll-like receptors and the complement system. Trends Immunol. 2010;31(4):154–63.View ArticlePubMedPubMed CentralGoogle Scholar
- Abe T, Hosur KB, Hajishengallis E, Reis ES, Ricklin D, Lambris JD, Hajishengallis G. Local complement-targeted intervention in periodontitis: proof-of-concept using a C5a receptor (CD88) antagonist. J Immunol. 2012;189(11):5442–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Brekke OL, Waage C, Christiansen D, Fure H, Qu H, Lambris JD, Osterud B, Nielsen EW, Mollnes TE. The effects of selective complement and CD14 inhibition on the E. coli-induced tissue factor mRNA upregulation, monocyte tissue factor expression, and tissue factor functional activity in human whole blood. Adv Exp Med Biol. 2013;734:123–36.View ArticleGoogle Scholar
- Alexander JJ, Anderson AJ, Barnum SR, Stevens B, Tenner AJ. The complement cascade: Yin-Yang in neuroinflammation—neuro-protection and -degeneration. J Neurochem. 2008;107(5):1169–87.View ArticlePubMedPubMed CentralGoogle Scholar
- Ricklin D, Lambris JD. Complement in immune and inflammatory disorders: pathophysiological mechanisms. J Immunol. 2013;190(8):3831–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Stevens B, Allen NJ, Vazquez LE, Howell GR, Christopherson KS, Nouri N, Micheva KD, Mehalow AK, Huberman AD, Stafford B, et al. The classical complement cascade mediates CNS synapse elimination. Cell. 2007;131(6):1164–78.View ArticlePubMedGoogle Scholar
- Schafer DP, Lehrman EK, Kautzman AG, Koyama R, Mardinly AR, Yamasaki R, Ransohoff RM, Greenberg ME, Barres BA, Stevens B. Microglia sculpt postnatal neural circuits in an activity and complement-dependent manner. Neuron. 2012;74(4):691–705.View ArticlePubMedPubMed CentralGoogle Scholar
- Hong S, Beja-Glasser VF, Nfonoyim BM, Frouin A, Li S, Ramakrishnan S, Merry KM, Shi Q, Rosenthal A, Barres BA, et al. Complement and microglia mediate early synapse loss in Alzheimer mouse models. Science. 2016;352:712–6.View ArticlePubMedPubMed CentralGoogle Scholar
- Shi Q, Colodner KJ, Matousek SB, Merry K, Hong S, Kenison JE, Frost JL, Le KX, Li S, Dodart JC, et al. Complement C3-deficient mice fail to display age-related hippocampal decline. J Neurosci. 2015;35(38):13029–42.View ArticlePubMedGoogle Scholar
- Lui H, Zhang J, Makinson SR, Cahill MK, Kelley KW, Huang HY, Shang Y, Oldham MC, Martens LH, Gao F, et al. Progranulin deficiency promotes circuit-specific synaptic pruning by microglia via complement activation. Cell. 2016;165(4):921–35.View ArticlePubMedGoogle Scholar
- Vasek MJ, Garber C, Dorsey D, Durrant DM, Bollman B, Soung A, Yu J, Perez-Torres C, Frouin A, Wilton DK, et al. A complement-microglial axis drives synapse loss during virus-induced memory impairment. Nature. 2016;534(7608):538–43.View ArticlePubMedGoogle Scholar
- Perry VH, Teeling J. Microglia and macrophages of the central nervous system: the contribution of microglia priming and systemic inflammation to chronic neurodegeneration. Semin Immunopathol. 2013;35(5):601–12.View ArticlePubMedPubMed CentralGoogle Scholar
- Reid KBM, Porter RR. Subunit composition and structure of subcomponent C1q of the first component of human complement. Biochem J. 1976;155:19–23.View ArticlePubMedPubMed CentralGoogle Scholar
- Botto M, Dell’agnola C, Bygrave AE, Thompson EM, Cook HT, Petry F, Loos M, Pandolfi PP, Walport MJ. Homozygous C1q deficiency causes glomerulonephritis associated with multiple apoptotic bodies. Nat Genet. 1998;19(1):56–9.View ArticlePubMedGoogle Scholar
- Klos A, Tenner AJ, Johswich KO, Ager RR, Reis ES, Kohl J. The role of the anaphylatoxins in health and disease. Mol Immunol. 2009;46(14):2753–66.View ArticlePubMedPubMed CentralGoogle Scholar
- Benoit ME, Clarke EV, Morgado P, Fraser DA, Tenner AJ. Complement protein C1q directs macrophage polarization and limits inflammasome activity during the uptake of apoptotic cells. J Immunol. 2012;188(11):5682–93.View ArticlePubMedPubMed CentralGoogle Scholar
- Roumenina LT, Sene D, Radanova M, Blouin J, Halbwachs-Mecarelli L, Dragon-Durey MA, Fridman WH, Fremeaux-Bacchi V. Functional complement C1q abnormality leads to impaired immune complexes and apoptotic cell clearance. J Immunol. 2011;187(8):4369–73.View ArticlePubMedGoogle Scholar
- Fraser DA, Pisalyaput K, Tenner AJ. C1q enhances microglial clearance of apoptotic neurons and neuronal blebs, and modulates subsequent inflammatory cytokine production. J Neurochem. 2010;112(3):733–43.View ArticlePubMedGoogle Scholar
- Benoit ME, Tenner AJ. Complement protein C1q-mediated neuroprotection is correlated with regulation of neuronal gene and microRNA expression. J Neurosci. 2011;31(9):3459–69.View ArticlePubMedPubMed CentralGoogle Scholar
- Benoit ME, Hernandez MX, Dinh ML, Benavente F, Vasquez O, Tenner AJ. C1q-induced LRP1B and GPR6 proteins expressed early in Alzheimer disease mouse models, are essential for the C1q-mediated protection against amyloid-beta neurotoxicity. J Biol Chem. 2013;288(1):654–65.View ArticlePubMedGoogle Scholar
- Fonseca MI, Zhou J, Botto M, Tenner AJ. Absence of C1q leads to less neuropathology in transgenic mouse models of Alzheimer’s disease. J Neurosci. 2004;24(29):6457–65.View ArticlePubMedGoogle Scholar
- Fonseca MI, Ager RR, Chu SH, Yazan O, Sanderson SD, LaFerla FM, Taylor SM, Woodruff TM, Tenner AJ. Treatment with a C5aR antagonist decreases pathology and enhances behavioral performance in murine models of Alzheimer’s disease. J Immunol. 2009;183(2):1375–83.View ArticlePubMedPubMed CentralGoogle Scholar
- Rozovsky I, Morgan TE, Willoughby DA, Dugichi-Djordjevich MM, Pasinetti GM, Johnson SA, Finch CE. Selective expression of clusterin (SGP-2) and complement C1qB and C4 during responses to neurotoxins in vivo and in vitro. Neuroscience. 1994;62(3):741–58.View ArticlePubMedGoogle Scholar
- Fan R, Tenner AJ. Complement C1q expression induced by Abeta in rat hippocampal organotypic slice cultures. Exp Neurol. 2004;185(2):241–53.View ArticlePubMedGoogle Scholar
- Cribbs DH, Berchtold NC, Perreau V, Coleman PD, Rogers J, Tenner AJ, Cotman CW. Extensive innate immune gene activation accompanies brain aging, increasing vulnerability to cognitive decline and neurodegeneration: a microarray study. J Neuroinflammation. 2012;9(1):179.View ArticlePubMedPubMed CentralGoogle Scholar
- Parkhurst CN, Yang G, Ninan I, Savas JN, Yates III JR, Lafaille JJ, Hempstead BL, Littman DR, Gan WB. Microglia promote learning-dependent synapse formation through brain-derived neurotrophic factor. Cell. 2013;155(7):1596–609.View ArticlePubMedPubMed CentralGoogle Scholar
- Harris JA, Hirokawa KE, Sorensen SA, Gu H, Mills M, Ng LL, Bohn P, Mortrud M, Ouellette B, Kidney J, et al. Anatomical characterization of Cre driver mice for neural circuit mapping and manipulation. Front Neural Circuits. 2014;8:76.View ArticlePubMedPubMed CentralGoogle Scholar
- Stephan AH, Madison DV, Mateos JM, Fraser DA, Lovelett EA, Coutellier L, Kim L, Tsai HH, Huang EJ, Rowitch DH, et al. A dramatic increase of C1q protein in the CNS during normal aging. J Neurosci. 2013;33(33):13460–74.View ArticlePubMedPubMed CentralGoogle Scholar
- Webster SD, Galvan MD, Ferran E, Garzon-Rodriguez W, Glabe CG, Tenner AJ. Antibody-mediated phagocytosis of the amyloid beta-peptide in microglia is differentially modulated by C1q. J Immunol. 2001;166(12):7496–503.View ArticlePubMedGoogle Scholar
- Li M, Pisalyaput K, Galvan M, Tenner AJ. Macrophage colony stimulatory factor and interferon-gamma trigger distinct mechanisms for augmentation of beta-amyloid-induced microglia-mediated neurotoxicity. J Neurochem. 2004;91(3):623–33.View ArticlePubMedGoogle Scholar
- Tenner AJ, Lesavre PH, Cooper NR. Purification and radiolabeling of human C1q. J Immunol. 1981;127(2):648–53.PubMedGoogle Scholar
- Huang J, Kim LJ, Mealey R, Marsh Jr HC, Zhang Y, Tenner AJ, Connolly Jr ES, Pinsky DJ. Neuronal protection in stroke by an sLex-glycosylated complement inhibitory protein. Science. 1999;285(5427):595–9.View ArticlePubMedGoogle Scholar
- Khoury MK, Parker I, Aswad DW. Acquisition of chemiluminescent signals from immunoblots with a digital single-lens reflex camera. Anal Biochem. 2010;397(1):129–31.View ArticlePubMedGoogle Scholar
- Miura-Shimura Y, Nakamura K, Ohtsuji M, Tomita H, Jiang Y, Abe M, Zhang D, Hamano Y, Tsuda H, Hashimoto H, et al. C1q regulatory region polymorphism down-regulating murine C1q protein levels with linkage to lupus nephritis. J Immunol. 2002;169(3):1334–9.View ArticlePubMedGoogle Scholar
- Schmittgen TD, Livak KJ. Analyzing real-time PCR data by the comparative C(T) method. Nat Protoc. 2008;3(6):1101–8.View ArticlePubMedGoogle Scholar
- Young P, Qiu L, Wang D, Zhao S, Gross J, Feng G. Single-neuron labeling with inducible Cre-mediated knockout in transgenic mice. Nat Neurosci. 2008;11(6):721–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Jung S, Aliberti J, Graemmel P, Sunshine MJ, Kreutzberg GW, Sher A, Littman DR. Analysis of fractalkine receptor CX(3)CR1 function by targeted deletion and green fluorescent protein reporter gene insertion. Mol Cell Biol. 2000;20(11):4106–14.View ArticlePubMedPubMed CentralGoogle Scholar
- Feil R, Wagner J, Metzger D, Chambon P. Regulation of Cre recombinase activity by mutated estrogen receptor ligand-binding domains. Biochem Biophys Res Commun. 1997;237(3):752–7.View ArticlePubMedGoogle Scholar
- Fonseca MI, Chu SH, Berci AM, Benoit ME, Peters DG, Kimura Y, Tenner AJ. Contribution of complement activation pathways to neuropathology differs among mouse models of Alzheimer’s disease. J Neuroinflammation. 2011;8(1):4.View ArticlePubMedPubMed CentralGoogle Scholar
- Cheng IH, Palop JJ, Esposito LA, Bien-Ly N, Yan F, Mucke L. Aggressive amyloidosis in mice expressing human amyloid peptides with the Arctic mutation. Nat Med. 2004;10(11):1190–2.View ArticlePubMedGoogle Scholar
- Nilsberth C, Westlind-Danielsson A, Eckman CB, Condron MM, Axelman K, Forsell C, Stenh C, Luthman J, Teplow DB, Younkin SG, et al. The ‘Arctic’ APP mutation (E693G) causes Alzheimer’s disease by enhanced Abeta protofibril formation. Nat Neurosci. 2001;4(9):887–93.View ArticlePubMedGoogle Scholar
- Ikarashi M, Nakashima H, Kinoshita M, Sato A, Nakashima M, Miyazaki H, Nishiyama K, Yamamoto J, Seki S. Distinct development and functions of resident and recruited liver Kupffer cells/macrophages. J Leukoc Biol. 2013;94(6):1325–36.View ArticlePubMedGoogle Scholar
- Shaner NC, Steinbach PA, Tsien RY. A guide to choosing fluorescent proteins. Nat Methods. 2005;2(12):905–9.View ArticlePubMedGoogle Scholar
- Yona S, Kim KW, Wolf Y, Mildner A, Varol D, Breker M, Strauss-Ayali D, Viukov S, Guilliams M, Misharin A, et al. Fate mapping reveals origins and dynamics of monocytes and tissue macrophages under homeostasis. Immunity. 2013;38(1):79–91.View ArticlePubMedGoogle Scholar
- Bohlson SS, O’Conner SD, Hulsebus HJ, Ho MM, Fraser DA. Complement, c1q, and c1q-related molecules regulate macrophage polarization. Front Immunol. 2014;5:402.View ArticlePubMedPubMed CentralGoogle Scholar
- Armbrust T, Nordmann B, Kreissig M, Ramadori G. C1Q synthesis by tissue mononuclear phagocytes from normal and from damaged rat liver: up-regulation by dexamethasone, down-regulation by interferon gamma, and lipopolysaccharide. Hepatology. 1997;26:98–106.PubMedGoogle Scholar
- Petry F, Botto M, Holtappels R, Walport MJ, Loos M. Reconstitution of the complement function in C1q-deficient (C1qa−/−) mice with wild-type bone marrow cells. J Immunol. 2001;167(7):4033–7.View ArticlePubMedGoogle Scholar
- Veerhuis R, Nielsen HM, Tenner AJ. Complement in the brain. Mol Immunol. 2011;48:1592–603.View ArticlePubMedPubMed CentralGoogle Scholar
- Bulla R, Tripodo C, Rami D, Ling GS, Agostinis C, Guarnotta C, Zorzet S, Durigutto P, Botto M, Tedesco F. C1q acts in the tumour microenvironment as a cancer-promoting factor independently of complement activation. Nat Commun. 2016;7:10346.View ArticlePubMedPubMed CentralGoogle Scholar
- Bossi F, Tripodo C, Rizzi L, Bulla R, Agostinis C, Guarnotta C, Munaut C, Baldassarre G, Papa G, Zorzet S, et al. C1q as a unique player in angiogenesis with therapeutic implication in wound healing. Proc Natl Acad Sci U S A. 2014;111(11):4209–14.View ArticlePubMedPubMed CentralGoogle Scholar
- Yasojima K, Schwab C, McGeer EG, McGeer PL. Up-regulated production and activation of the complement system in Alzheimer’s disease brain. Am J Pathol. 1999;154(3):927–36.View ArticlePubMedPubMed CentralGoogle Scholar
- Zhou J, Fonseca MI, Pisalyaput K, Tenner AJ. Complement C3 and C4 expression in C1q sufficient and deficient mouse models of Alzheimer’s disease. J Neurochem. 2008;106(5):2080–92.View ArticlePubMedPubMed CentralGoogle Scholar
- Bodea LG, Wang Y, Linnartz-Gerlach B, Kopatz J, Sinkkonen L, Musgrove R, Kaoma T, Muller A, Vallar L, Di Monte DA, et al. Neurodegeneration by activation of the microglial complement-phagosome pathway. J Neurosci. 2014;34(25):8546–56.View ArticlePubMedGoogle Scholar
- Bialas AR, Stevens B. TGF-beta signaling regulates neuronal C1q expression and developmental synaptic refinement. Nat Neurosci. 2013;16(12):1773–82.View ArticlePubMedPubMed CentralGoogle Scholar
- Shen Y, Li R, McGeer EG, McGeer PL. Neuronal expression of mRNAs for complement proteins of the classical pathway in Alzheimer brain. Brain Res. 1997;769:391–5.View ArticlePubMedGoogle Scholar
- Chow LM, Zhang J, Baker SJ. Inducible Cre recombinase activity in mouse mature astrocytes and adult neural precursor cells. Transgenic Res. 2008;17(5):919–28.View ArticlePubMedPubMed CentralGoogle Scholar
- Stephan AH, Barres BA, Stevens B. The complement system: an unexpected role in synaptic pruning during development and disease. Annu Rev Neurosci. 2012;35:369–89.View ArticlePubMedGoogle Scholar
- Chu Y, Jin X, Parada I, Pesic A, Stevens B, Barres B, Prince DA. Enhanced synaptic connectivity and epilepsy in C1q knockout mice. Proc Natl Acad Sci U S A. 2010;107(17):7975–80.View ArticlePubMedPubMed CentralGoogle Scholar
- Wyss-Coray T, Rogers J. Inflammation in Alzheimer disease-a brief review of the basic science and clinical literature. Cold Spring Harb Perspect Med. 2012;2(1):a006346.View ArticlePubMedPubMed CentralGoogle Scholar
- Ide M, Harris M, Stevens A, Sussams R, Hopkins V, Culliford D, Fuller J, Ibbett P, Raybould R, Thomas R, et al. Periodontitis and cognitive decline in Alzheimer’s disease. PLoS One. 2016;11(3):e0151081.View ArticlePubMedPubMed CentralGoogle Scholar
- Howell GR, Macnicoll KH, Braine CE, Soto I, Macalinao DG, Sousa GL, John SW. Combinatorial targeting of early pathways profoundly inhibits neurodegeneration in a mouse model of glaucoma. Neurobiol Dis. 2014;71:44–52.View ArticlePubMedPubMed CentralGoogle Scholar