Introduction
Despite the long-held view of the central nervous system (CNS) as an immune-privileged site, the importance of local and systemic immunological responses in neurodegenerative brain diseases like Alzheimer’s disease has been widely recognized in the past few years [
1]. In Alzheimer’s disease, accumulation of misfolded amyloid β-peptide (Aβ) peptides plays a central role in disease pathogenesis and has been associated with a wide range of adverse reactions ultimately leading to neuronal dysfunction and neurodegeneration [
2,
3]. Local immune responses to cerebral Aβ aggregates are a pathological hallmark of Alzheimer’s disease involving activation of cellular components of the brain’s innate immunity, such as microglia and astrocytes, and release of various immune molecules and inflammatory mediators, including cytokines, acute phase proteins and factors of the complement cascade, among others [
4‐
7]. This activation of the brain’s innate immune system appears to be a double-edged sword as it has been related to both beneficial and detrimental consequences [
8]. Microglia, the resident phagocytes of the CNS, for example, are able to bind to Aβ via various cell-surface receptors and promote phagocytosis and clearance of Aβ aggregates [
5]. Their chronic activation in Alzheimer’s disease, however, has also been associated with disturbed clearance functions and an increased production of pro-inflammatory and neurotoxic factors, which may further promote neuronal injury and exacerbate neuronal loss [
9].
In contrast to the activation of the brain’s innate immune system, the role of the adaptive immunity in cerebral β-amyloidosis appears to be much less prominent and is still incompletely understood. Consistent with a role of the humoral adaptive immunity in Alzheimer’s disease pathogenesis, misfolded Aβ peptides have been shown to induce specific B cell-mediated immune responses, with naturally occurring autoantibodies against Aβ being found in the cerebrospinal fluid (CSF) and blood of affected patients, as well as, healthy individuals [
10‐
17]. Interestingly, these autoantibodies seem to share conformation-specific binding epitopes and the ability to promote microglia-mediated clearance of β-amyloid plaques, as well as to lower neurotoxicity [
12,
18‐
20]. In support of these findings, Britschgi and colleagues demonstrated that naturally occurring autoantibodies against a broad range of amyloidogenic peptides can display neuroprotective activities [
21], with similar findings being reported by other investigators [
22,
23]. Apart from the involvement of B cell-mediated humoral immune responses, several authors have begun addressing the role of the cellular adaptive immunity in Alzheimer’s disease. Previous neuropathological analyses repeatedly confirmed the presence of T cells in post-mortem Alzheimer’s disease brains [
14,
24‐
29]. Moreover, Aβ-reactive T cells were readily detectable in the peripheral blood of Alzheimer’s disease patients and healthy elderly subjects, in whom their proportion was significantly increased in comparison to middle-aged control individuals [
30]. Analysis of the profile of specific lymphocyte subsets led to ambiguous findings. Richart-Salzburger, for example, reported a significant decrease of CD3
+ CD8
+ and CD19
+ lymphocytes in the peripheral blood of Alzheimer’s disease patients, suggesting a general decline of the adaptive immunity in disease pathogenesis [
31]. Other authors, in contrast, observed significant increases in activated CD4
+ and CD8
+ T cell subsets in peripheral blood [
32‐
35]. Recently, Lueg et al. also assessed alterations of T cell subsets in the intrathecal compartment of cognitively impaired and cognitively intact elderly subjects as well as healthy younger control individuals and found a significantly increased proportion of CD8
+ lymphocytes in the CSF of patients with Alzheimer’s disease – but not in patients suffering from other dementias – which correlated with cognitive deficits [
33]. Altogether, these results provided further evidence for an involvement of the adaptive immunity in Alzheimer’s disease.
In this study, we addressed the role of a congenital deficiency of the adaptive immune system on β-amyloid-related pathology
in vivo and generated B and T cell-ablated APP transgenic mice by crossbreeding the PSAPP mice, en established mouse model of Alzheimer’s disease [
36], with the recombination activating gene-2 knockout (Rag2 ko) mice lacking functional B and T lymphocytes [
37]. In a second experiment, the effects of an acquired deficiency of the adaptive immunity were studied in lethally irradiated aged PSAPP mice reconstituted with Rag2 ko bone marrow. Our results from both experimental paradigms demonstrate an impact of functional adaptive immune cells on cerebral β-amyloid deposition
in vivo and suggest an increased Aβ clearance, likely mediated by microglial cells, as a possible mechanism underlying reduced Aβ pathology in the absence of functional B and T cells.
Materials and methods
Animals and tissue processing
The Rag2 ko mouse lines [
37] and the wildtype (WT) mouse line expressing the Ptprc
a (CD45.1 or Ly5.1) allele (Model 4007) were purchased from Taconic (Ry, Denmark). The PSAPP transgenic mice express a chimeric mouse/human amyloid precursor protein (Mo/HuAPP695swe) and a mutant human presenilin 1 (PS1-dE9) and were purchased from the Jackson Laboratory (Maine, USA) [
36,
38]. All mice were on the C57Bl/6 (H2
b) genetic background. Rag2 ko mice (Model RAGN12; immunophenotype H2
b (C57B/6), CD45.2) were crossed with the PSAPP mice (immunophenotype H2
b (C57B/6), CD45.2) to generate the four genotypes PSAPP/Rag2 ko, PSAPP, Rag2 ko and WT in the F2 generation. For adoptive bone marrow (BM) transfer experiments, Rag2 ko donor mice expressing the immunophenotype H2
b (C57B/6), CD45.1 were used (Model 461). All mice were housed in groups (two to five animals per cage in 396 × 215 × 172 mm cages (Indulab, Gams, Switzerland)) under specific pathogen free (SPF) conditions at 22 °C ±1 °C and a 12 h light–dark cycle beginning at 7:00 AM. The health status of animals was monitored twice a week except for the BM-transferred mice, which were monitored every second day during the first two weeks after irradiation. No major adverse events were observed. Water and food were provided
ad libitum except for the time of cognitive testing. Environmental enrichment included bedding, one cardboard mouse house and a paper tissue as nesting material.
For all biochemical and histological analyses gender-balanced groups of mice were used. Mice were deeply anesthetized with ketamine/xylazine. Blood samples were collected from the right ventricle in ethylenediaminetetraacetic acid (EDTA)-coated tubes (Sarstedt, Sevelen, Switzerland) and centrifuged for plasma preparation at 2000 g for 10 min at 4 °C. Following the collection of blood, mice were transcardially perfused with ice-cold phosphate buffered saline (PBS) and their brains rapidly removed. Brains were dissected into two hemispheres with the left hemisphere being placed in 4 % (w/v) paraformaldehyde (PFA) in PBS, post-fixed overnight in the same medium at 4 °C and transferred into a 30 % sucrose solution (in PBS) for 72 h (cryoprotection). The right hemisphere was snap frozen and stored at −80 °C for later biochemical analysis.
All animal experiments were performed in compliance with Swiss law for care and use of animals (“455.163 Tierversuchsverordnung” 2010) and were approved by the veterinarian authorities of the Canton of Zurich (license numbers 079/2010 and 102/2013). All sections of this report adhere to the ARRIVE guidelines for reporting animal research [
39]. A complete ARRIVE guidelines checklist is included.
Protein extracts and Western blotting
Homogenization of brain tissue was performed using a glass Teflon homogenizer in fivefold wet weight amount of buffer A (100 mM Tris–HCl, 150 mM NaCl, Phosphatase Inhibitor Cocktails 2 + 3 (Sigma-Aldrich, Buchs, Switzerland) and Complete Protease Inhibitor Cocktail (Roche Diagnostics, Rotkreuz, Switzerland)). Supernatants were collected (=Tris fraction) after centrifugation at 100’000 g for 1 h. The pellets were rehomogenized in buffer A containing 2 % sodium dodecyl sulfate (SDS). Centrifugation was repeated and supernatants again were again collected (=SDS fraction). The resulting pellets were dissolved in 70 % formic acid (FA), sonicated two times for 30 s at 30 % power, ultracentrifuged at 100’000 g for 30 min, supernatants extracted, lyophilized and reconstituted in buffer A for further analysis. Extracts were separated by SDS-polyacrylamide gel electrophoresis, blotted onto nitrocellulose, blocked in Tris-buffered saline containing 5 % milk for 1 h at room temperature. Blots were incubated with the primary antibody overnight at 4 °C, then visualized via peroxidase-conjugated antibody and ECL reaction (Amersham Biosciences, Otelfingen, Switzerland). The primary rabbit polyclonal antibody anti-C-terminal APP (Sigma-Aldrich Chemie GmbH, Buchs, Switzerland) was used at a 1:2000 dilution. A mouse monoclonal anti-GAPDH antibody (Meridian Life Science, Memphis, USA) was used as internal loading control and for normalization of densitometric analyses of the immunoreactive bands (dilution 1:10’000). Image J software (rsb.info.nih.gov/ij/) was used for quantifications of the immunoreactive bands by densitometry of the scanned films under conditions of non-saturated signal.
Meso Scale Discovery (MSD) analysis
Aβ40 and Aβ42 levels were measured in the above-mentioned brain homogenate fractions and plasma using an MSD 3plex multi-SPOT Aβ human kit (Gaithersburg, MD, USA) as previously described [
40]. Plates were measured on an MSD SECTOR Imager 600 plate reader (Gaithersburg, MD, USA) after adding the MSD Read Buffer T (Gaithersburg, MD, USA). The MSD DISCOVERY WORKBENCH software (Version 3.0.17) (Gaithersburg, MD, USA) with Data Analysis Toolbox was used to calculate sample concentrations by comparing them against a standard curve.
Histological analysis
PFA-fixed and cryoprotected hemibrains were cut continuously into 35 μm thick coronal slices at −20 °C using a microtome (Leica Jung HN40, Leica Microsystems AG, Heerbrugg, Switzerland) and stored at −20 °C in an anti-freeze solution (phosphate buffer 0.50 M in MilliQ water: ethyleneglycol: glycerol = 1.3:1:1) until use. For each staining six equidistant sections per mouse were analyzed from positions −1.94 mm to +1.70 mm from the bregma [
41]. For immunohistochemistry, sections were first washed with PBS + 0.2 % Triton X-100, blocked with appropriate normal sera for 1 h and incubated with the primary antibody overnight at 4 °C. The following primary antibodies were used: Glial fibrillary acidic protein (GFAP; 1:400 dilution; rabbit polyclonal; Sigma-Aldrich Chemie GmbH, Buchs, Switzerland), Ionized calcium binding adaptor molecule 1 (Iba1; 1:500 dilution; rabbit polyclonal; Wako Chemicals, Osaka, Japan) and anti-β amyloid 1–16 (clone 6E10; 1:500 dilution; mouse monoclonal; BioLegend, Fell, Germany). Secondary antibodies were obtained from Jackson Immunoresearch Laboratories (West Grove, PA, USA). Thioflavin S staining was performed according to the standard protocols as previously described [
42]. 4′,6-Diamidin-2-phenylindol (DAPI; Sigma-Aldrich Chemie GmbH, Buchs, Switzerland) was used as counterstaining agent. For quantitative image analysis, the cortex (including the cingulate, motor and somatosensory cerebral cortex) was analyzed on six equidistant coronal hemibrain sections from positions −1.94 mm to +1.70 mm from the bregma [
41]. Mean values per mouse were calculated and included into statistical analysis. Average plaque numbers refer to single hemibrain sections. Quantitative image analysis was performed blinded using ImageJ software (rsb.info.nih.gov/ij/). For specific microglial analyses around amyloid plaques serial confocal Z-stack images, comprising an average of 30 to 40 cortical brain sections at 0.26 μm intervals, were acquired using a confocal laser scanning microscope (Leica TCS SP8, Leica Microsystems AG, Heerbrugg, Switzerland). On average, 12 cortical Thioflavin S-positive amyloid plaques per mouse were analyzed. Plaques of an average size between 20 and 30 μm were randomly selected by the blinded investigator between cortical layers II and V on six consecutive equidistant sections (2 plaques per section) from positions −1.94 mm to +1.70 mm from the bregma [
41]. Counting of microglia was done in a diameter of 100 μm around the center of the amyloid plaque as previously described [
43].
Adoptive bone marrow transfer
For the generation of bone marrow (BM) chimeras, 12 month-old PSAPP transgenic recipient mice (CD45.2; 12 females and 7 males) were lethally irradiated with two doses of 600 rads each (second irradiation after 12 h). 3 month-old Rag2 ko (Model 461, Taconic, Ry, Denmark) and age-matched WT control donor mice (Model 4001, Taconic, Ry, Denmark) expressing the CD45.1 allele were sacrificed by CO2 inhalation, and bones (including femur, tibia and pelvis) were flushed with sterile PBS to obtain BM stem cells. A total of 5 x 106 cells in 0.2 ml sterile Hank’s Buffered Salt Solution (HBSS, Life Technologies, Zug, Switzerland) were injected i.v. per animal in a randomized and gender-balanced fashion (10 PSAPP mice (including 6 females and 4 males) received Rag2 ko BM; 9 PSAPP mice (including 6 females and 3 males) received WT BM). BM reconstitution of recipient mice was carried out with BM from gender-matched donors. Autoclaved drinking water supplemented with antibiotics was provided for two weeks (2 g/l Neomycin sulphate; Sigma Aldrich). Reconstitution efficiency with CD45.1 positive cells was assessed on the day of sacrifice (6 months later) by flow cytometry of EDTA whole blood samples.
Flow cytometry
Ammonium-Chloride-Potassium (ACK) lysis of red blood cells was performed on EDTA blood samples followed by HBSS and FACS buffer (HBSS, 2 % fetal calf serum (FCS), 10 mM EDTA, 0.01 % NaN3) washing steps. After using a live/dead exclusion dye (Molecular Probes, Eugene, OR, USA) an Fc blocking step (anti-mouse CD16/32, TruStain fcX™, Biolegend, CA, USA) was used to reduce non-specific binding. The cells were then stained for 15 min at 4 °C with surface marker-specific antibodies: anti-CD45 (PE conjugated, rat anti-mouse, clone 30-F11, BioLegend, Fell, Germany), anti-CD45.1 (PE conjugated, mouse anti-mouse, clone A20, BioLegend, Fell, Germany), anti-CD45-2 (FITC conjugated, mouse anti-mouse, clone 104, BioLegend, Fell, Germany), anti-CD11b (APC or APC-Cy7 conjugated, rat anti-mouse/human, clone M1/70, BioLegend, Fell, Germany), anti-CD4 (FITC or APC conjugated, rat anti-mouse, clone RM4-5, BioLegend, Fell, Germany), anti-Ly6C (FITC conjugated, rat anti-mouse, clone AL-21, BD Biosciences, Allschwil, Switzerland), anti-Ly6G (APC-conjugated, rat anti-mouse, clone 1A8, BioLegend, Fell, Germany) and anti-B220/CD45R (PerCP-Cy5.5 or APC-Cy7 conjugated, rat anti-mouse/human, clone RA3-6B2, BioLegend, Fell, Germany). After additional washing with FACS buffer samples were analyzed blinded using a BD LSRFortessa™ cell analyser (BD Biosciences, Allschwil, Switzerland). Single stainings and isotype control antibodies were included as controls.
Cognitive-behavioral testing
A battery of validated tests was used to assess motoric and cognitive behavior of the mice. For all cognitive tests, the state-of-the-art ANY-maze Video Tracking System software (Stoelting Co., USA) was used as an automated system to track animal movements and completion of test specific tasks. All animals were tested naïve in a random order. The tested groups were balanced for gender. The experimenter was blinded during the time of testing.
Open Field
The Open Field test measures locomotor activity and anxiety. Mice were placed in the centre of a white box (49 x 49 x 33 cm) with indirect illumination (22 Lux) and their movements were tracked for 15 min. Distance travelled, average speed and time spent in the central (inner square 33 x 33 cm) versus peripheral zones of the open field were calculated using the ANY-maze software [
44].
Novel Object Recognition Test (NORT)
This test measures memory retention and recognition memory in mice and was performed in accordance to published protocols [
45]. The NORT has been designed as a two-trial task including a sample trial (ST) and a test trial (TT) and was performed following the Open Field test with a delay of 5 min in the same box. During ST, two similar objects were placed in the box, 34 cm apart (symmetrically). Each animal was allowed to explore the box for 5 min. The animals were considered to be exploring the object when the head of the animal was within a defined area of 12 cm in diameter around the object. The objects sizes ranged within diameters from 3–4 cm, so the animal was considered exploring the object when the animals head was near the object as close as 4 cm. After ST, mice were immediately returned to their holding cages and following a 10 min delay TT was carried out. One of the objects used during ST was replaced by a novel one differing in form and color before each mouse was placed back into the same box as during ST. The mice were then allowed to freely explore for 5 min and the time spent exploring each object was recorded. The ratio was calculated by dividing measures of the new object versus the old one.
Y-Maze
The Y-Maze was used to evaluate spatial working memory in mice and was carried out according to published protocols [
40]. The Y-Maze consists of three horizontal arms (40 cm long, 11 cm wide and 21 cm high) symmetrically disposed at 120° to each other. Light was adjusted to 28 Lux. The mice were placed at one end of the first arm and then allowed to freely explore for 5 min. The mice were tracked recording arm entries and exploration time in each arm. The total number of arm entries was measured as an index of locomotor activity to rule out the interference of changes in motility with the parameters of learning and memory [
46,
47]. The percentage alternation was calculated as the ratio of actual to possible alternations (defined as the total number of arm entries −2) x 100 %.
Statistical analysis
The primary experimental outcome of our study included Aβ pathology (amyloid plaque pathology, Aβ levels) and glial pathology. Secondary outcome measures included cognitive performance and APP processing. Data analysis was performed using the SPSS software version 19.0 (16 M, Zurich, Switzerland). Extreme values according to SPSS explorative data analysis were excluded from further analysis. Tests for normal distribution were performed before statistical testing. According to the results of the Shapiro-Wilk- and the Kolmogorov-Smirnov-Test for normality, either two-sample Student’s t-test or two-tailed Mann–Whitney-U-Test for two sample groups or ONEWAY ANOVA or the Kruskal-Wallis-Test for multiple group comparisons (followed by post-hoc Bonferroni or Mann–Whitney-U-Tests with Bonferroni correction) were performed. For NORT a Wilcoxon-Test was performed to test if the ratio differed significantly from 1. For Y maze a one-sample t-test was performed to test if the percentage of correct alternations was significantly above 50. A p-value <0.05 was considered statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001 indicate significantly different performances. Error bars represent SEM.
Discussion
The ageing process is a major risk factor for sporadic Alzheimer’s disease and impacts negatively on an individual’s capacity to respond to immune challenges, a condition termed immunosenescence [
51]. Human immunosenescence is accompanied by a deterioration of B and T cell-mediated adaptive immune responses, while innate immunity is rather preserved or even upregulated [
51]. An even more pronounced reduction in systemic T and B cells has previously been reported by some authors in patients with Alzheimer’s disease [
31], suggesting a “premature immunosenescence” phenotype specifically affecting adaptive immunity in the pathogenesis of the disease [
52]. In the present study, we applied two different experimental paradigms to investigate the contribution of the adaptive immune system to cerebral β-amyloid pathology in an established APP transgenic mouse model of Alzheimer’s disease. The consequences of a congenital deficiency of the adaptive immunity were assessed in the PSAPP/Rag2 ko crossbred mice which were compared to immunocompetent PSAPP mice. To confirm the absence of T and B cells in this model and exclude potential effects of APP/mutant presenilin 1 on systemic immune compartments in the PSAPP mice [
53], FACS analyses of whole blood as well as other peripheral immune organs (data not shown) were performed. Notably, no significant alterations of peripheral lymphocytes and myeloid cells were observed between PSAPP transgenic and WT control mice, thus excluding major systemic immune alterations in this model. Histological and biochemical analyses revealed a reduced β-amyloid pathology along with decreased brain Aβ levels in aged PSAPP/Rag2 ko mice in comparison to the PSAPP mice after onset of amyloid plaque deposition. This decrease in Aβ pathology was associated with an increase in plasma Aβ levels and a relative increase in Iba1-positive microglial cells, which was not observed in 3 month-old pre-plaque PSAPP/Rag2 ko mice, suggesting that the reduction in brain Aβ in the absence of functional B and T cells only occurred in the presence of Aβ plaque deposits and was likely mediated by amyloid-reactive microglia. In line with these findings, similar results were obtained in our second model system of an acquired adaptive immune deficiency: aged PSAPP mice reconstituted with Rag2 ko BM cells showed a reduced β-amyloid load, decreased brain Aβ and elevated plasma Aβ levels along with increased microgliosis and elevated numbers of plaque-associated Aβ-phagocytosing microglia. In contrast to our findings in the PSAPP/Rag2 ko crossbred mice, reconstitution of PSAPP transgenic mice with Rag2 ko BM resulted in milder effects on Aβ pathology, as revealed by a reduced 6E10-stained Aβ load, but no change in Thioflavin S-stained compact fibrillar Aβ deposits, decreased brain Aβ levels exclusively in the detergent insoluble protein fraction and similar levels of β-amyloid-associated astrogliosis as compared to the WT BM-recipient PSAPP controls. In our opinion, the observed milder effects on Aβ pathology may at least partly be explained by the incomplete ablation of T and B cells in the BM transfer paradigm (versus a complete ablation of lymphocytes in the crossbreeding approach). Moreover, additional aspects, e.g. the older age of PSAPP transgenic mice in the BM transfer experiment and/or shorter follow-up periods (6 months versus 8 months in the crossbreeding approach) have to be taken into account.
In our experimental set up Rag2 ko mice with a combined T and B cell deficiency were used. This approach allowed us to simulate the conditions of “premature immunosenescence” in a mouse model of cerebral β-amyloidosis and investigate the effects of a general adaptive immune deficiency on Aβ pathology. Based on this experimental design, however, we were not able to delineate a specific lymphocyte subpopulation accounting for the effects on microglia and cerebral amyloid deposition. Although we cannot exclude that a lack of B cell-mediated humoral immunity might have contributed to the observed effects, this possibility appears to be less likely in our view for several reasons. On the one hand, B cells, unlike T cells, have not been detectable in brain parenchyma of Alzheimer’s disease patients [
54] nor in brains of PSAPP transgenic mice, as revealed by immunohistochemistry and brain FACS analysis (Ferretti et al., submitted). On the other hand, based on several previous studies suggesting a role for naturally occurring Aβ autoantibodies in restricting brain amyloidosis [
12,
18‐
20], a lack of Aβ-reactive B cells would probably be expected to result in exacerbated rather than reduced amyloid pathology. In line with this notion, progressive age-related amyloid plaque deposition was previously associated with a specific reduction of naturally occurring serum Aβ autoantibodies in the Tg2576 APP transgenic mouse model of Alzheimer’s disease [
55]. In contrast to B cells, T cells have been repeatedly detected as infiltrating brain tissue both in human Alzheimer’s disease brains and in animal models of the disease [
24‐
29,
56,
57] Ferretti et al., submitted). Although the exact role of T lymphocytes in the disease process remains incompletely understood, both proinflammatory and neurotoxic as well as antiinflammatory and neurotrophic mechanisms have been attributed to their presence in brain parenchyma [
58]. In certain rat and APP transgenic mouse models of Alzheimer’s disease, for example, specific effector CD4
+ T cells secreting IFN-γ, IL-17 or IL-22 were linked to increased neuroinflammation and exacerbation of disease progression [
59,
60]. In other animal studies, however, IFN-γ-dependent targeting of Aβ-specific Th1-polarized CD4
+ cells to amyloid plaques was shown to reduce amyloid pathology without promoting toxic neuroinflammation [
61,
62].
A role for the adaptive immune system in neuroinflammation and brain pathology has also been proposed in other (non-Alzheimer’s disease) neurodegenerative conditions. Removal of CD4
+ T lymphocytes, but not CD8
+ lymphocytes, for example, was previously associated with reduced MPTP-induced neuronal cell loss in a mouse model of Parkinson’s disease [
63]. In contrast, the same T cell population was shown to provide neuroprotection in an animal model of familial amyotrophic lateral sclerosis (ALS) [
64,
65]. In this work, mSOD1
G93A/Rag2 ko mice and CD4
+ T cell-ablated mSOD1
G93A/CD4 ko mice showed accelerated disease progression and increased motoneuron degeneration, which was associated with increased mRNA levels for proinflammatory cytokines and NOX2, and decreased levels of trophic factors and glutamate transporters [
64]. Interestingly, bone marrow reconstitution of mSOD1
G93A/Rag2 ko mice with T cells led to a prolonged survival and suppressed cytotoxicity [
64]. These beneficial effects were accompanied by the induction of an M2 protective microglial phenotype, which was mainly attributed to the presence of CD4
+CD25
+FoxP3
+ regulatory T lymphocytes (Tregs) [
65]. Suppressive effects on both the adaptive and the innate arm of the immune system have been well documented for Tregs [
66‐
68]. Their lack in the Rag2-deficient PSAPP transgenic mice may therefore, at least in part, account for the observed increased neuroinflammatory phenotype and microglia-mediated phagocytosis of Aβ aggregrates. This possibility is further supported by recent findings in the 5XFAD AD mouse model of AD, in which transient depletion of Foxp3
+ Tregs, or pharmacological inhibition of their activity, was similarly shown to mitigate amyloid plaque clearance [
69]. Tregs secrete various anti-inflammatory cytokines (such as IL-4, IL-10 and IL-13) and have been shown to promote alternatively activated (M2) responses and attenuate proinflammatory (M1) microglial responses [
68,
70,
71]. Notably, microglial polarization towards the M1 response has been associated with reduced amyloid pathology [
72,
73], which would also be in line with our findings in the T and B cell-ablated PSAPP mice. Despite significantly reduced Aβ levels, an increased proinflammatory phenotype in the absence of functional adaptive immune cells may be a double-edged sword. In fact, proinflammatory microglial responses have also been associated with detrimental consequences such as increased tau pathology and exacerbated neurotoxicity [
73]. In our study, the beneficial effects on Aβ pathology were not accompanied by a reversal of hippocampal-dependent memory deficits in the PSAPP/Rag2 ko mice. Non-transgenic Rag2 ko mice, however, were equally impaired on two different hippocampal-dependent tasks in comparison to the WT control, suggesting additional Aβ-independent effects of the T and B cell ablation on cognitive performance. In line with these findings, a properly functioning adaptive immune system has previously been shown to be involved in maintaining cognitive health and hippocampal memory formation in several immunocomprised mouse models lacking functional T cells [
48‐
50]. These effects were proposed to be mainly mediated by non-CNS specific CD4
+ T cells playing a role in brain plasticity and promoting neurogenesis in a BDNF-dependent manner in the adult hippocampus [
48‐
50].
Competing interests
Rebecca Derungs, Maria Teresa Ferretti, Christoph Gericke, Luka Kulic, Roger M. Nitsch, Claudia Späni, Tobias Suter, Tobias Welt and Fabian Wirth have no competing interests to declare.
Authors’ contributions
LK, CS and RMN designed experiments. LK, CS, TS, RD, MTF, TW, FW, CG performed experiments. LK, CS, TS and MTF analyzed data. LK and CS wrote the manuscript. All authors read and approved the final manuscript.