Contribution of endogenous antibodies to learning deficits and astrocytosis in human P301S mutant tau transgenic mice

Antibodies have been explored extensively as a potential therapeutic for Alzheimer’s disease, where amyloid-β (Aβ) peptides and the tau protein deposit in patient brains. While the major focus of antibody-based therapy development was on Aβ, arguably with limited success in clinical trials, targeting tau has become an emerging strategy, possibly extending therapies to dementias with isolated tau pathology. Interestingly, low titres of autoantibodies to pathological tau have been described in humans and transgenic mouse models, but their pathophysiological relevance remained elusive. Here, we used two independent approaches to deplete the B-cell lineage and hence antibody formation in human P301S mutant tau transgenic mice, TAU58/2. TAU58/2 mice were either crossed with the B-cell-deficient Ighm knockout line (muMT−/−) or treated with anti-CD20 antibodies that target B-cell precursors. In both models, B-cell depletion significantly reduced astrocytosis in TAU58/2 mice. Only when B-cells were absent throughout life, in TAU58/2.muMT−/− mice, were spatial learning deficits moderately aggravated while motor performance improved as compared to B-cell-competent TAU58/2 mice. This was associated with changes in brain region-specific tau solubility. No other relevant behavioural or neuropathological changes were observed in TAU58/2 mice in the absence of B-cells/antibodies. Taken together, our data suggests that the presence of antibodies throughout life contributes to astrocytosis in TAU58/2 mice and limits learning deficits, while other deficits and neuropathological changes appear to be independent of the presence of B-cells/antibodies.

Scientific RepoRtS | (2020) 10:13845 | https://doi.org/10.1038/s41598-020-70845-x www.nature.com/scientificreports/ of dementia [14][15][16] . This includes the TAU58/2 line more recently developed by us; TAU58/2 mice express human P301S mutant tau in neurons and present with progressive tau hyperphosphorylation and NFT pathology, as well as early-onset motor, behavioural and learning deficits 17 . Using such transgenic mouse models, tau has been extensively targeted pre-clinically in both active and passive immunization approaches, showing efficacy in reducing tau pathology and in some studies improving cognitive and behavioural deficits 18,19 . Interestingly, both non-immunized tau transgenic mice and human AD/ FTD patients showed baseline titres of autoantibodies to pathological tau 18,20 . Furthermore, autoantibodies to tau have been commonly found in the serum and cerebrospinal fluid of healthy adults as well as children 21,22 . The relevance of autoantibodies in the context of tau pathology is unknown.
Here, we crossed TAU58/2 mice with B-cell-deficient muMT −/− mice or depleted B-cells in young TAU58/2 mice using anti-CD20 antibodies to explore the role of the adaptive immune system, specifically antibodies and attributable neuroinflammation, in the context of tau pathology. This augmented learning deficits of TAU58/2 mice while astrocytosis was reduced, suggesting a contribution of antibodies to disease onset and progression. This knowledge may improve understanding of the connection between antibodies and tau in disease.

Results
Genetic B-cell depletion from birth augmented learning deficits in tau transgenic mice. To determine a possible contribution of antibodies to phenotype and pathology of TAU58/2 mice, we first crossed them with muMT −/− mice (Fig. 1a). muMT −/− mice have a targeted disruption in transmembrane region of the Ighm gene encoding the heavy chain segment of IgM antibodies which are expressed on the surface membrane at the pre-B lymphocyte stage of cell development 23 . This membrane expression is required for progression from the pre-B lymphocyte stage to immature, mature and antibody-secreting plasma cell stages 24 . Therefore, muMT −/− homozygous mice are unable to produce B lymphocytes and antibodies. First, we confirmed the absence of antibodies and B-cells in the resulting TAU58/2.muMT −/− mice compared to TAU58/2.muMT +/+ littermates. Staining of brain sections with antibodies to mouse IgG showed virtually no labelling in TAU58/2. muMT −/− brains, compared with intense staining of TAU58/2.muMT +/+ tissue (Fig. 1b). Fluorescence-activated cell sorting (FACS) of spleens from TAU58/2.muMT −/− confirmed a lack of IgM + /B220 high lymphocytes, suggesting arrest of B lymphocyte development at the pre-B lymphocyte phase in these mice (Fig. 1c,d). Taken together, TAU58/2.muMT −/− mice lacked endogenous antibodies and B lymphocytes.
We have determined that mature TAU58/2 mice show delayed spatial learning in the Morris water maze (MWM) (Przybyla M et al., unpublished). To assess possible effects of the lack of antibodies on spatial learning, 8-month-old TAU58/2.muMT −/− and TAU58/2.muMT +/+ mice and non-transgenic muMT −/− and muMT +/+ littermates were subjected to the MWM paradigm. TAU58/2.muMT −/− mice showed significantly delayed learning after 6 days of MWM testing when compared with TAU58/2.muMT +/+ littermates (Fig. 2a-c). Both non-transgenic muMT −/− and muMT +/+ controls showed similarly accelerated learning compared to TAU58/2.muMT +/+ mice, consistent with intact spatial learning. Probe trials after removal of the hidden escape platform showed a trend towards reduced time TAU58/2.muMT −/− mice spent in the target quadrant and increased time in the opposite quadrant of the MWM as compared with controls, but differences did not reach significance (Fig. 2d). There was no overt difference in the average distance swum between groups (Fig. 2e). Taken together, TAU58/2. muMT −/− mice presented with delayed learning in the MWM task.
In contrast to TAU58/2.muMT −/− mice, anti CD20-treated TAU58/2 mice showed no changes to the spatial learning deficits observed in control IgG-treated TAU58/2 mice when compared to anti CD20-or control IgGtreated non-transgenic littermates (Fig. 7a-c). Similarly, anti CD20-and control IgG-treated TAU58/2 mice spent comparable times in the target quadrant (Fig. 7d). There was no overt difference in average distance swum www.nature.com/scientificreports/ between groups (Fig. 7e). Both anti-CD20-and control IgG-treated TAU58/2 mice showed similar disinhibition behaviour during EPM testing with reduced closed arm time as compared to anti-CD20-and control IgG-treated non-transgenic mice (Fig. 7f). Movement of mice during EPM testing was comparable between all groups (Fig. 7g). Finally, anti-CD20-and control IgG-treated TAU58/2 mice showed similar deficits during Rotarod testing without improvement over consecutive test days (Fig. 7h,i). Taken together, long-term B-cell depletion in mature TAU58/2 mice did not alter learning, behavioural or motor deficits.
B-cell depletion in mature tau transgenic mice reduced astrocytosis. When assessing neuropathological changes after long-term B-cell depletion in mature TAU58/2 mice, we found that, similar to TAU58/2. muMT −/− mice, anti-CD20-treated TAU58/2 mice presented with comparable labelling of GFAP + astrocytes in the amygdala as in anti-CD20-treated non-transgenic littermates, while control IgG-treated TAU58/2 mice www.nature.com/scientificreports/ showed significant astrocytosis as compared with control IgG-treated non-transgenic mice (Fig. 8a). GFAPstaining was not increased in the brainstem or hippocampus of anti-CD20-or control IgG-treated TAU58/2 mice compared to non-transgenic controls (Fig. 8b,c). Labelling of A1 and A2 astrocyte subpopulations with antibodies to C3 and S100A10 25 , respectively, did not show subtype specific changes to astrocyte numbers upon genetic B-cell depletion in anti-CD20-treated TAU58/2 compared to control amygdala, hippocampus and brainstem ( Supplementary Fig. 1). In contrast to astrocytes, there were no changes to microglial numbers in the amygdala or hippocampus of anti-CD20-or control IgG-treated TAU58/2 mice as compared to the accordingly treated non-transgenic controls (Fig. 8d-f). There was no change in microglial appearance or engulfment of NFT-like lesions, as described earlier by us for the TAU58/2 line 17 , in the amygdala and brainstem of anti-CD20-treated TAU58/2 mice (Supplementary Fig. 3). Interestingly, the number of microglia closely associated with NFT-like lesions were reduced in the hippocampus of anti-CD20-treated TAU58/2 mice compared with TAU58/2 mice that received control IgG injections ( Supplementary Fig. 3b). Taken together, absence of antibodies induced by long-term B-cell depletion in mature mice reduced astroglial activation in TAU58/2 mice and altered microglial behaviour.
Altered tau pathology upon B-cell depletion in mature tau transgenic mice. Similar to TAU58/2 crossed with muMT −/− mice, we stained brain sections of anti-CD20-and control IgG-treated TAU58/2 mice with antibodies to the phosphorylated sites S214 (pS214), S396 + S404 (PHF1) and S422 (pS422). Except for increased PHF1 phosphorylation in the hippocampus, phosphorylation of tau at all these sites were not significantly changed in either the amygdala, hippocampus or brainstem of anti-CD20-treated TAU58/2 mice as www.nature.com/scientificreports/ compared to control IgG-treated TAU58/2 mice (Fig. 9a-i). When assessing tau solubility biochemically, we found significantly increased levels of insoluble tau in the hippocampus of anti-CD20-treated TAU58/2 mice as compared to controls, with concomitant significant reduction of tau phosphorylation at all tested sited in the insoluble fractions (Fig. 10a,b). Only S422 phosphorylation was significantly reduced in the soluble and intermediate soluble fractions, with other sites and tau levels remaining similar in anti-CD20-treated TAU58/2 hippocampus as compared with controls. For comparison, soluble and insoluble soluble tau levels were significantly reduced, while levels of tau phosphorylation at S214 in soluble fractions and all sites in the intermediate soluble fractions of cortical tissue were significantly increased in anti-CD20-treated TAU58/2 mice as compared to controls (Fig. 10c,d). In brainstem extracts, insoluble tau levels were significantly lower in anti-CD20-treated TAU58/2 mice as compared to controls, with unchanged phosphorylation (Fig. 10e,f). Phosphorylation of tau at S214 was significantly increased in soluble and at S422 significantly decreased in soluble and intermediate soluble fractions. Taken together, absence of antibodies due to long-term depletion of B-cells in mature mice did not change the histopathological patterns of tau phosphorylation but had brain regions-specific effects on tau solubility in TAU58/2 mice.

Discussion
In the present study, we show that absence of antibodies and B-cells reduced astrocytosis in the brains of P301S tau transgenic TAU58/2 mice. Experimental reduction of immunoglobulins and plasma cells was achieved either by genetically depleting B-cells throughout the lifetime of the mice or by long-term depletion after mice had fully developed. In contrast, only lifetime depletion of B-cells modulated learning deficits in TAU58/2, with www.nature.com/scientificreports/ further augmented acquisition deficits during MWM testing but improved motor performance in TAU58/2. muMT −/− mice. TAU58/2 mice present with a variety of motor and behavioural deficits, including progressive spatial learning deficits, similar to other tau transgenic mouse models of FTD and AD 17,26,27 . Comprehensive testing of TAU58/2 mice crossed with the muMT −/− line or after long-term anti-CD20 antibody-mediated B-cell depletion showed that antibodies (or other related B-cell mediated effects) play no major role in functional deficits associated with transgenic expression of mutant human tau and tau pathology. However, the very distinct effect of genetic B-cell depletion on the spatial and motor learning deficits of TAU58/2 mice may suggest that pathological pathways that involve antibodies are limited to distinct neuronal populations or networks during the early onset of deficits. Hence, later B-cell depletion has no impact on this phenotype. An alternative explanation that cannot be fully excluded is that the genetic modification itself causes effects due to unknown neuronal functions of Ighm gene products. Such effects would however be surprisingly specific for learning, since both genetic and anti-CD20 antibody-mediated B-cell depletion have the same effect on astrocytosis.
Despite the moderate changes to the behaviour of TAU58/2 mice when B-cells were depleted throughout development and post-natal life, both genetic and antibody-mediated B-cell depletion resulted in a significant reduction of astrocytosis in the amygdala, where GFAP labelling is normally sparse in wild-type mice and markedly increased in TAU58/2 brains together with early NFT pathology 17,26 . Absence of similar changes to astrocytes in the hippocampus may be due to the marked GFAP labelling of this brain area in naïve mice, likely masking effects of B-cell depletion. Given the absence of overt changes to tau pathology in the amygdala in either TAU58/2.muMT −/− or anti-CD20 antibody-treated TAU58/2 mice, our data suggest that the pathways that lead to astrocytosis in TAU58/2 mice involve, at least in part, antibody-mediated processes. Astrocytes have been found to adopt distinct states that can either promote or limit neurodegeneration 22,25,[28][29][30][31] . To this end, staining of A1 and A2 subpopulations suggests that effects of autoantibodies are astrocyte fate-independent in TAU58/2 brains, but augmented learning deficits in TAU58/2.muMT −/− mice as compared with TAU58/2.muMT +/+ littermates together with reduced astrocytosis may suggest that antibody-mediated processes support neuroprotective states of astrocytes during onset.
Autoantibodies to pathologically phosphorylated tau have been reported in humans and tau transgenic mouse models. Their role in disease, however, is unclear. Autoantibody titres against tau were very low, which is very different from auto-immune disease 22,[29][30][31] . Furthermore, it is known that only a very small proportion of circulating antibodies can pass the blood brain barrier (BBB) 32,33 . Therefore, one may conclude that a major disease-driving role of autoantibodies detected in association with tau pathology is rather unlikely. Nevertheless, specific staining of IgG in perfused brain tissue of TAU58/2 mice that was absent in crosses with the muMT −/− line, demonstrate that antibodies do pass the BBB and tend to label areas with more abundant tau hyperphosphorylation, such as the hippocampus. We and others have previously shown that both active and passive immunization of tau transgenic mice limits tau pathology when treatment is initiated before the onset of overt tau pathology 19,34,35 or halts progression of tau pathology when vaccination commenced after the onset of neuropathological changes, including during late stages of the disease 18,19,[36][37][38] . These studies demonstrated that antibodies specific for distinct phosphorylated tau species can limit tau pathology and hold great promise for therapy development. Here, our finding of unchanged tau histopathology, particularly of the late phosphorylation sites PHF1 and pS422, but changes to tau solubility, suggest that the presence of low titre autoantibodies to pathological tau are not sufficient to limit the progressive tau hyperphosphorylation yet are sufficient to impact on tau aggregation. To achieve therapeutic effects of tau-specific antibodies, much higher titres are required, such as those found during active and passive vaccination trials in different tau transgenic mouse models 18,19,38 . The mechanisms underlying brain area-specific differences in tau solubility upon B-cell depletion and their possible contribution to the functional changes of TAU58.2 mice remain to be investigated.
Taken together, our data suggest that the adaptive immune system (i.e. B-cells and antibodies) do not play a major role in the onset and progression of tau pathology and associated functional deficits in tau transgenic mice. Autoantibodies to pathological tau may however have some subtle disease modulating effects, such as altering learning. B lymphocyte depletion. Three-week-old male TAU58/2 mice were injected intravenously fortnightly through the tail vein with 10 mg/kg mouse anti-CD20 antibody (clone5D2, isotype IgG2a) or mouse isotype control antibody (anti-ragweed IgG2a). Both antibodies were kindly provided by Genentech. Animals were humanely killed and tissue harvested at eight months of age. Treatment schedule and effective dose have previously been determined 39 . Behavioural, memory and motor testing. The following behavioral tests all used the same cohorts of mice. The CD20-treated cohort consisted of heterozygous TAU58/2 tg/wt mice (CD20 n = 12, control n = 12) Scientific RepoRtS | (2020) 10:13845 | https://doi.org/10.1038/s41598-020-70845-x www.nature.com/scientificreports/ and their wildtype littermates (CD20 n = 13, control n = 12). Rotarod and elevated plus maze was performed at 3.5 months of age and Morris Water Maze at 7 months of age. muMT −/− mice were crossed with TAU58/2 transgenic mice to obtain a cohort composed of TAU58/2 tg/wt × muMT −/− (n = 12), TAU58/2 tg/wt × muMT +/+ (n = 8), TAU58/2 wt/wt × muMT −/− (n = 13) and TAU58/2 wt/wt × muMT +/+ (n = 13) mice. Rotarod and elevated plus maze were performed at 4 months of age and Morris Water Maze at 8 months of age. The experimenter was blinded to genotype and/or treatment whilst conducting the behavioural tests.

Methods
Elevated Plus Maze. Anxiety/disinhibition in the mice was assessed as previously described 40 . Briefly, the Elevated Plus maze (Stoelting Co.) consisted of two open and two closed arms (each 35 cm × 5.5 cm), as well as a central platform (5.5 cm × 5.5 cm), elevated 60 cm above the ground. Mice were acclimatized to the room for 1 h prior to testing, then placed on the center platform facing an open arm and recorded for 5 min from above. Videos were analysed using the AnyMaze software.
Rotarod. Motor performance of mice was determined using a Rotarod (Ugo Basile) in acceleration mode (5-60 rpm) over 120 s for three consecutive days, as previously reported 17 . The longest time each mouse remained on the rotating wheel out of 5 attempts per session was recorded for each day.
Morris water maze. Assessment of spatial learning and memory was performed as previously reported 41 . Briefly, after mice were acclimatized to the procedure rooms (1 h prior to experiments each day) they were placed into a 1.2 m diameter tank filled with opaque water and a submerged invisible 10 cm diameter escape platform opposite the entry quadrant. The tank was surrounded by 4 different cues displayed for each quadrant. During four subsequent trials each day, all mice were given 60 s to locate the escape platform. Only mice that failed to find the platform were thereafter guided to the platform. All mice remained on the platform for 60 s after reaching it and before being returned to their home cage. Starting positions were varied between trial within the quadrant opposite the platform. These acquisition trials were carried out for 6 and 7 consecutive days for muMT −/− crossed and CD20-treated mice, respectively, until control mice reached the criterion of less than 20 s for trial completion. On the following day, the platform was removed, and mice placed into the tank for 30 s to explore (= probe trial). One day later, the platform was equipped with a visible flag, and cues were removed from around the tank, to ensure that all mice had normal vision. Videos were analyzed using the AnyMaze tracking software (Stoelting).
Flow Cytometry. Following perfusion, spleens were harvested and homogenised into a single cell suspension in phosphate buffered saline (PBS) using a 100 µm Falcon cell-strainer. Red blood cells were lysed using BD FACS Lysing Solution (cat#349202) and the remaining cells blocked with FACS buffer (phosphate buffered saline with 2% fetal bovine serum) and stained using fluorochrome-conjugated antibodies against IgM (IgM-APC, cat#550676, BD Biosciences) and CD45R/B220 (CD45R/B220-PE, cat#553090, BD Biosciences). Cells were run on a BD LSRFortessa SORP (BD Biosciences) flow cytometer and the data analysed with Flow Jo Version 10 software (FlowJo, LLC/Treestar). n = 6 for each genotype and treatment. Representative results are shown in figures.
Histology. Mice were anesthetized and transcardially perfused with cold phosphate buffered saline. Brains were removed and one hemisphere was immersion-fixed in 4% paraformaldehyde for immunohistochemical analysis. Fixed brains were processed using an automated tissue system (Excelsior, Thermo, USA), embedded in paraffin and coronally sectioned at the level of the mid-hippocampus at 3 µm using a microtome (Thermo, USA) for immunofluorescence staining. Microscopy was performed with an Olympus BX51 (Olympus, USA) epifluorescence microscope equipped with an Olympus DP70 software colour camera for representative images. Slides were scanned with an AxioScan (Zeiss) for entire brain region counts using ImageJ software (Fiji). All staining was done in Sequenza staining racks for standardization using previously reported protocols 42 . Following 1 h in blocking buffer (3% (v/v) goat serum (Sigma), 2% (w/v) bovine serum albumin (Sigma) in phosphate buffered saline), slides were incubated with primary antibody in blocking buffer at 4 °C overnight. The following primary antibodies were used: Tau13 (sc-21796, Santa Cruz), PHF-1 (pS396/pS404 tau, P. Davies, Albert Einstein College of Medicine, New York, NY), pS422 tau (ab71495, Abcam), pS214 tau (ab170892, Abcam), GFAP (G9269 Sigma), IBA-1 (019-19741, Wako), C3 (ab11862, abcam) and S100A10 (PA5-95505, Invitrogen). Alexa Fluor secondary antibodies (Invitrogen) were used for detection and counterstaining was done with DAPI (Molecular Probes). For staining of endogenous IgG in the brain, TAU58/2.muMT −/− and TAU58/2.muMT +/+ mice were perfused as above and after fixation were sectioned at 50 μm using a vibratome (Campden, UK). Following permeabilization with 0.05% NP-40 for 10 min and treatment with 3% hydrogen peroxide for 1 h, sections were blocked for 1 h with blocking buffer and then incubated with anti-mouse IgG biotin-coupled secondary antibodies (Sigma) in blocking buffer. The ABC-HRP detection kit (Vector) using ImmPACT DAB (Vector) was used to visualise staining prior to mounting with DPX Mountant for histology (Sigma). Sequential tissue extraction. Sequential extraction of brain tissue from different brain areas with buffers of increasing stringency was carried out as previously described in detail by us 19 . Western blotting. Western blotting was carried out using established protocols 43 . Primary antibodies were against Tau13 (sc-21796, Santa Cruz), PHF-1 (pS396/pS404 tau, P. Davies, Albert Einstein College of Medicine, New York, NY), pS422 tau (ab71495, Abcam), pS214 tau (ab170892, Abcam) and Gapdh (MAB374, Merck-Mil-Scientific RepoRtS | (2020) 10:13845 | https://doi.org/10.1038/s41598-020-70845-x www.nature.com/scientificreports/ lipore). All primary antibodies were visualized with HRP-coupled species-specific secondary antibodies (Santa Cruz) and an HRP substrate (Biorad), using a Chemidoc CCD-camera detector system (Biorad). Quantification of blots was done with Image J FIJI (NIH) and band intensities were normalized to Gapdh band intensities, and tau phosphorylation signals further normalized for human tau levels per fraction. Full size blots presented in Figs. 5 and 10 are shown in Supplementary Fig. 4 and 5.
Statistical analysis. All statistical analysis was done using the Graphpad Prism 8.0 software (GraphPad, USA) using either Student's t-tests for comparison of two data sets, analysis of variance (ANOVA) for comparison of more than two data sets or two-way ANOVA for comparison across time. N-numbers are provided in each figure. P < 0.05 values were considered significant. All values are presented as mean ± standard error of the mean.