Zika virus infection induces a persistent accumulation of Alzheimer’s disease-like Tau phosphorylation in adult immunocompetent mice in association with memory and social behavior impairments

Abstract

Clinical and epidemiological data support the link between viral encephalitis and neurodegeneration but its causal mechanism remains mostly unknown. Zika virus (ZIKV) is an emerging, neurotropic flavivirus susceptible to induce cognitive impairments in infected adults. In this work we have analyzed the capacity of ZIKV to induce the accumulation of pathological phosphorylated Tau protein (pTau), a major driver of neurodegenerative disorders such as Alzheimer’s disease, throughout the infection of adult immunocompetent mice. The capacity of ZIKV to induce pTau in vivo, was analyzed in the long term in three Collaborative Cross mouse strains displaying different responses to ZIKV. The establishment and propagation of pTau was quantified up to 60 days post-infection (dpi) in correlation with the level and localization of neuronal viral infection and of microglia activation using immunofluorescence, immunohistochemistry, wide field and confocal microscopy and gene expression analysis. Strength, coordination, memory and social behavior were evaluated before and following the establishment and progression of pTau. The role of microglia on ZIKV-induced pTau was investigated by partially depleting microglial cells using PLX3397 (PLX). ZIKV infection induced a significant accumulation of pTau starting at 15 that persisted at least until 60 dpi. At 15 dpi, pTau was observed in ZIKV-infected neurons in the CA2 and CA1 regions of the hippocampus and in non-infected cortical neurons in association with neuroinflammation and social behavior alterations. At 30 dpi, pTau progressed independently of infection and inflammation, positively correlated to PLX-susceptible Apoe gene expression in association with short-term memory defects. These results shed light on how brain viral infections, which are a major concern for public health, drive pTau accumulation and propagation in link with memory impairment and social behavior alterations laying the groundwork for potential new therapeutic treatments.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s40478-026-02368-x.

Introduction

Several viruses are capable of invading the central nervous system inducing neurological manifestations including neurocognitive deficits [1–3]. Cognitive impairment as a sequel of viral infection has been described not only in the case of viruses that have an established broad neurotropic potential such as herpes simplex virus (HSV-1) [4], but also in the case of viruses considered to display less invasive potentials of the central nervous system (CNS) such as human immunodeficiency virus (HIV) that even though does not infect neurons can enter the CNS and infect glial cells [5, 6], and more recently in the context of COVID-19 infection caused by severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) [7–9]. There is also a strong concern on the neurocognitive impacts of arboviruses and particularly flavivirus infections that can spread rapidly in pandemic proportions while being able to target the CNS and induce neurocognition impairment [3, 10–12]. Memory deficit is a major trait observed in adults infected by flaviviruses such as West Nile virus (WNV) [13, 14], Japanese encephalitis virus (JEV) [15] and tick-borne encephalitis virus (TBEV) [16, 17], with memory loss often reported many years after the acute phase of infection.

A progressive memory loss is the earliest cognitive deficit associated with the development of Alzheimer’s disease (AD), the most prevalent neurodegenerative disease, which etiology remains for the most part unknown [18, 19]. The presence of common molecular and clinical traits shared by viral infections of the CNS and Alzheimer’s disease (AD) reported in the context of infections by HSV-1 [14, 20–22], HIV [23] and SARS-CoV-2 [9, 24, 25] has raised the question of the potential role of viral infections in AD’s etiology [26–28].

Zika virus (ZIKV) is an emerging neurotropic, mosquito-borne, flavivirus [29, 30], which can induce mild to severe neurological complications in infected adults [31, 32] with long-lasting neurological sequelae [33] that include cognitive [34] and olfactory [35] impairments. ZIKV-induced delayed neuropathies of the CNS, associated with cognitive disorders, have also been observed in animal models such as macaques [36, 37] and mice [38, 39]. In previous work [40], we have described the capacity of ZIKV to induce the pathological hyperphosphorylation of the neuronal Tau protein (pTau) that is a major hallmark of AD [18, 41, 42]. ZIKV induction of pTau was observed in vitro, in primary cultured neurons as well as in vivo, early after intracranial (i.c.) infection of adult immunocompetent mice [40]. In parallel, the work of Lee et al. [43] demonstrated that ZIKV infection of human brain organoids accelerated AD phenotypes triggering pTau.

Notwithstanding the interest and novelty of the observations reported in these two studies with respect to the capacity of ZIKV infection to induce a major molecular trait associated with AD, each of them had its limitations. Both works lacked information on persistency, with observations made only at 6 [40] and 14 [43] days post infection (dpi), and lacked correlation with signs of disease. The work of Lee et al. [43] on human brain organoids was done in the absence of microglial cells that play a major role in AD pathophysiology including pTau progression [44–46]. Also, our in vivo work was limited by the rarity of the events with only a limited number of ZIKV-infected brains displaying pTau at 6 dpi [40], while the work by Lee et al. [43] was limited in its genetic diversity since brain organoids analyzed were derived from only one AD patient and one healthy individual.

In the present study, we have extended our previous work with the aim to overcome these limitations. The persistency of the capacity of ZIKV infection to induce pTau in vivo, in the brain of intracranially infected adult immunocompetent mice, was analyzed here up to 60 dpi in three different mouse strains previously characterized as displaying different responses to ZIKV [47]. The establishment and progression of pTau was analyzed/quantified in correlation with the level and localization of neuronal viral infection and of microglia activation in hippocampus and cortex using immunofluorescence, immunohistochemistry, wide field and confocal microscopy and gene expression analysis. Strength, coordination, memory and social behavior tests were performed at 0, 6, 15 and 30 dpi before or following the establishment and spreading of pTau that were observed at 15 and 30 dpi respectively. Finally, the role of microglia on ZIKV-induced pTau was investigated by partially depleting microglial cells using PLX3397.

Materials and methods

Virus

The FG15 Asian ZIKV strain, isolated from a patient during a ZIKV outbreak in French Guiana in December 2015, was obtained from the Virology Laboratory of the Institut Pasteur of French Guiana. Viral stocks were prepared from the supernatant of infected C6/36 cells, clarified by centrifugation and titrated on Vero cells (ATCC, CRL-1586) by a focus-forming assay (FFA). Stocks were kept at -80 °C. All experiments with ZIKV were carried out in biosafety level 3 facilities.

Mice

Collaborative Cross (CC) CC001, CC005 and CC071 male and female mice (purchased from the Systems Genetics Core Facility, University of North Carolina and bred at the Institut Pasteur) were maintained under specific-pathogen-free conditions with a 14-h light and 10-h dark cycle and ad libitum food and water. In all experiments, mice were killed by cervical dislocation. All experimental protocols were approved by the Institut Pasteur Ethics Committee (dap190107 and dap230008) and authorized by the French Ministry of Research (#19,469 and #43,891), in compliance with French and European regulations.

Antibodies

Primary antibodies for immunofluorescence and immunohistochemistry: anti-NS2B (Genetex GTX133308); AT8 (Thermofisher, MN1020); goat anti-IBA1 (Wako, 011-27991); rabbit anti-IBA1 (Wako, 0119-19741). Secondary antibodies for immunofluorescence: anti-Rabbit Cy3 (Jackson Immunoresearch, 711-166-152); anti-mouse Cy2 (Jackson Immunoresearch, 7115-226-151); anti-Goat (Jackson Immunoresearch, Invitrogen, A21469).

Mouse ZIKV intracranial (i.c.) infection

All infection experiments were performed in a biosafety level 3 animal facility as described previously [47]. Mice (6–15 weeks old) were anesthetized by ketamine (70 mg/kg, i.p.) and xylazine (5 mg/kg, i.p.) and received analgesia with buprenorphine (0.03 mg/kg, i.p.) prior to i.c. inoculation. Mice received either 10⁵ foci forming units (FFUs) of ZIKV FG15 diluted in 0.9% NaCl (saline solution) or saline solution alone in a volume of 10 µl. Survival and signs of disease were monitored daily until euthanasia for brain collection. For microglia depletion, CC005 mice were fed with irradiated food pellets (SAFE® 150, SAFE, Augy, France) containing PLX3397 (Pexidartinib at 290 mg/kg) or not (control group) for 1 month before and continuously after ZIKV i.c. inoculation.

RT-qPCR

For extraction of total brain RNA, brain tissue samples were homogenized at 4 °C in 1 ml of Trizol reagent (Life technologies), using ceramic beads and an automated homogeniser (PreCellys). 1 μg of total brain RNA was reversely transcribed using High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturers’ recommendations using Random Primers. Identical amounts of cDNA (corresponding to the amount reversely transcribed from 5 ng of total RNA) were used for all qPCR reactions that were performed using SENSIFAST SYBR NO-ROX MIX (Bio-TechnoFix) reagents at: 95 °C 2 min, then 40 cycles at 95 °C 5 s, 60 °C 10 s, 72 °C 15 s, followed by a dissociation step. Relative quantification of mRNA expression was calculated by the ∆C_T method using Hprt1 as reference gene. Sequences (5’-3’) of primers used for qPCR analysis are listed in Table S1.

Brain immunofluorescence, immunohistochemistry and image analysis

After necropsy, the brain was removed, fixed for 48 to 72 h in 10% neutral buffered formalin, and embedded in paraffin. Sagittal (5 μM) brain sections were deparaffinized and unmasked using citrate buffer (3.75 mM citrate acid, 2.5 mM disodium phosphate, pH 6) for 15 min in a decloaking chamber. For immunofluorescence, the sections were submerged for 1 h in 3% BSA-PBS-Triton 0,2%, and the primary antibodies were incubated overnight at 4 °C in the presence of PBS-0.2% Triton. These antibodies were revealed by adding the appropriate secondary antibodies, incubated 1 h in the presence of PBS-0.2% Triton. Autofluorescence eliminator reagent (Merck, 2160) was used during 5 min to reduce autofluorescence signal. The sections were counterstained with Hoechst (Invitrogen, 33342) and mounted with Mowiol solution (Merck, 475,904). Images of brain sections were acquired using a confocal microscope Zeiss Axio ScanZ1, a Lamina slide scanner (Perkin Elmer) or Widefield Microscope Olympus BX63F and analyzed using the QuPath or ImageJ softwares. Immunohistochemistry was performed at the HistIM platform of Institut Cochin. Slides were incubated for 15 min at 95 °C in a solution of deparaffinization and antigen retrieval at pH 6 before incubation 20 min at room temperature with the primary antibodies. Detection was performed using a Leica BOND Polymer Refine Detection System DS9800. Sections were lightly counterstained with hematoxylin, then dehydrated and cleared in graded alcohol and Ottix plus (MM-France) and finally covered with glass slips. Images were acquired using a Lamina slide scanner (Perkin Elmer) and analyzed using the QuPath software. AT8+ cells present on a z-slice image acquired with either a confocal or a wide field microscope were manually counted considering as positive those cells for which an AT8+ cytoplasmic staining was clearly associated with Hoechst (DNA) staining.

Behavioral tests

Behavior tests were performed either under a biosafety cabinet (grip strength) or inside the negative pressure isolator where mice were housed.

Rotarod test

The rotarod test was performed using the Rotarod from Maze Engineers (Skokie, Illinois, USA). Mice were placed on the rod rotating at a speed of 4 revolutions/min and tested for 180 s with an acceleration of 12 revolutions/min² and a maximum speed of 40 revolutions/min. The time during which mice remained on the rod was recorded automatically. Each mouse was tested 3 times and the mean of the two highest values was kept.

Grip strength test

The grip strength test was performed using a BIO-GS3 from BIOSEB (Vitrolles, France) equipped with a bar, according to the manufacturer’s instructions. Mice were hung by the tail and allowed to grip the bar with their fore paws. They were then gently pulled back and the mechanical resistance at which they dropped the bar was recorded. Each mouse was tested 5 times successively. Outlying values (much lower than the others) were discarded and the mean of homogenous values was kept.

Novel object recognition (NOR) test

After observing that some CC strains showed signs of anxiety and did not explore much in a normally-lit box, the NOR test was performed in a 37 cm × 21 cm red plastic cage with a red lid. Light intensity in the box was adjusted to 2–5 Lux. Objects were plastic Duplo™ bricks (Lego System A/S, Billund, Denmark) approximately 2 × 2 × 2 cm, with different shapes and color (red and green). Two similar red objects were placed 20 cm apart in the middle of the test cage. One mouse was introduced at the center of the test cage and allowed to explore for 10 min. It was then removed and placed for 10 min in its home cage, while one of the two objects was replaced with a different one. The mouse was then reintroduced in the test cage for 10 min of exploration. The time spent in contact with each object (familiar and novel) was divided by the total time spent in contact with either object.

Social behavior test

Social interactions between mice were analyzed using a Live Mouse Tracker as previously described [48] with minor modifications to be installed in an isolator. For each strain, several groups of two non-infected and two ZIKV-infected mice of the same sex (males or females) were tested before infection and at 5–7, 14–16 and 29–32 days post-infection. At every time point, interactions between mice were recorded during three successive nights determined by the room light timer (9PM–7AM). No access to the LMT room was allowed during the recording. Data from nights 2 and 3 were averaged and used for statistical analysis. Thirty-five behavioral events were defined from the video recordings, as described in Maisterrena et al. [49]. For each mouse group and each parameter, the mean value of the two non-infected mice was subtracted to the individual values of all mice to improve the representation of the differences between non-infected and ZIKV-infected mice across days.

Statistics

Data were analyzed using Prism version 9.5.1 (GraphPad Software) in the case of Figs. 1, 2, 3, 4, 5, 6a and 7a–d. Data shown on Figs. 6a–d and 7e were analyzed using R version 4.4.3 and appropriate packages. All R scripts are available from XM. The statistical analysis used and the number, n, of independent experiments or mice is specified in the text or the corresponding figure legends. For grip strength, NOR (memory) and LMT parameters, effect of ZIKV infection across days was calculated using two-way ANOVA with repeated measures. Day-wise infection effect was then tested using post-hoc t-tests with Bonferroni correction. Rotarod (coordination) data were analyzed day-by-day by a logrank test.

Fig. 1.

ZIKV infection induces a persistent Tau pathology in the brain of adult immunocompetent mice. a Schematic representation of the experiment setup. 6–15 week-old male and female mice from the Collaborative Cross CC001, CC005 and CC071 strains were inoculated by i.c. injection (at the ventricular zone of left hemisphere) with either saline solution (NI) or 10⁵ FFU of ZIKV. NI and ZIKV-infected mice were necropsied at 6, 15, 30 and 60 dpi with the left hemisphere used for immunofluorescence or immunohistochemistry and the right hemisphere for RNA extraction and gene expression analysis by RT-qPCR with respect to Hprt1 used as reference gene. b The brain section of a ZIKV- or a non-infected (NI) CC005 mouse necropsied at 30 dpi was labeled with the monoclonal AT8 antibody (specific of pTau, green) with DNA labeled with Hoechst (grey). Z-projection images of the cortex obtained with a wide field microscope are shown. Yellow arrows indicate cytoplasmic perinuclear pTau and asterisks indicate granular and fibrillar neuritic pTau. Scale bar = 100 µm. c The number of neurons displaying perinuclear cytoplasmic pTau within an entire brain section (excluding the cerebellum), as determined using ImageJ software, was calculated with respect to the total area (in mm²) of the region analyzed. d The level of expression of ZIKV RNA present within the total RNA extracts of the entire right hemisphere of the brains analyzed in c was determined by RT-qPCR with respect to Hprt1 used as reference gene. d, e Dot plots show means with one dot for each brain with 7–8 brains (from 2 to 3 independent series of infections) analyzed per condition and per mouse strain. Significance with respect to 6 dpi conditions of a given strain was assessed by unpaired, non-parametric, Mann–Whitney t-test and significance between the different strains was assessed by a 2way ANOVA Tukey’s multiple comparisons test. P-value < 0.05 (*), < 0.01 (**), < 0.001 (***), < 0.0001 (****). Not significant (ns) p-values are not indicated unless < 0.15

Fig. 2.

ZIKV infection induces pTau pathology in infected (NS2B+) and non-infected (NS2B−) neurons. Brain sections of ZIKV-infected CC005 mice necropsied at 15 dpi were labeled with the monoclonal AT8 antibody (specific of pTau, green), anti-NS2B antibodies (specific of ZIKV, red) and anti-IBA1 antibodies (specific of microglia, magenta) with DNA labeled with Hoechst (grey). Z-projection images of a the hippocampus and b the cortex obtained with a confocal microscope are shown. In crop images shown in (a), yellow arrows indicate a microglial cell engulfing a ZIKV-infected AT8+ neuron. CA = cornu ammonis region of the hippocampus. Scale bars = 100 µm

Fig. 3.

ZIKV-induced pTau pathology progresses to NS2B− cortical neurons independently of infection. a Immunohistochemical analysis of the hippocampus of the left hemisphere of ZIKV-infected CC005 mice necropsied at 6, 15, 30 and 60 dpi previously analyzed in Fig. 1, using anti-NS2B antibodies (specific of ZIKV, brown) with DNA labeled in blue (hematoxylin). Representative images of brain sections obtained with a slide scanner microscope are shown. DG = dentate gyrus, CA = cornu ammonis and BR = brain number. Scale bars = 100 µm. b, c, and d The number of total, NS2B+ and AT8+ neurons present in the analyzed section of the CA of non-infected (NI) and ZIKV-infected CC005 was quantified. In b, the percentage of infected (NS2B+) neurons was calculated with respect to the average total number of neurons present in the CA region of the NI CC005 mice. In c, the quantification of total number of neurons/mm² present within the CA region throughout infection is shown. In d, the percentage of NS2B+ AT8+ neurons was calculated with respect to the total number of NS2B+ neurons present in the CA. e Quantification of the number of NS2B+ and AT8+ neurons/mm² in the section of the indicated regions of interest (ROI) of non- and ZIKV-infected CC005 mice. b–e Dot plots show means with one dot for each brain and significance assessed with respect to NI conditions by unpaired, non-parametric, Mann–Whitney t-test. P-value < 0.05 (*), < 0.01 (**) and ns = not significant. f Brain sections of ZIKV-infected CC005 mice necropsied at 30 and 60 dpi were labeled with the monoclonal AT8 antibody (specific of pTau, green), anti-NS2B antibodies (specific of ZIKV, red) and anti-IBA1 antibodies (specific of microglia, magenta) with DNA labeled with Hoechst (grey). Z-projection images of the cortex obtained with a confocal microscope are shown. Scale bars = 100 µm. g The level of expression of the Aif1 gene coding for IBA1 present within the total RNA extracts of the entire right hemisphere of the brains analyzed here was determined by RT-qPCR with respect to Hprt1 used as reference gene. Dot plot shows means with one dot for each mouse and significance assessed by unpaired parametric t-test. P-value < 0.05 (*), < 0.01 (**), < 0.0001 (****) and ns = not significant

Fig. 4.

A potential role for microglial cells in the progression of ZIKV-induced pTau pathology to NS2B− cortical neurons. Non- (NI) and ZIKV-infected CC005 were fed with control (CTRL) or PLX-containing food 1 month before and all the time throughout infection. a The number of microglial cells in the hippocampus and the cortex present in a z brain slice of CC005 mice necropsied after 45 days treatment with CTRL or PLX food was determined following immunohistochemical labeling of microglial cells of the left hemisphere with anti-IBA1 antibodies. b The level of expression the P2ry12 (specific of microglial cells), Cx3cl1 (specific of neurons) and Gfap (specific of astrocytes) genes present within the total RNA extracts of the entire right hemisphere of the brains analyzed in a was determined by RT-qPCR with respect to Hprt1 used as reference gene. a, b Dot plots show means with one dot for each brain and significance assessed by unpaired parametric t-test. P-value < 0.05 (*), < 0.01 (**) and ns = not significant. c The level of expression of ZIKV RNA and the P2ry12 (specific of microglial cells), Cx3cl1 (specific of neurons) and Gfap (specific of astrocytes) genes present within the total RNA extracts of the brains of NI and ZIKV-infected (necropsied at 15 and 30 dpi), CTRL and PLX-fed CC005 mice was determined by RT-qPCR with respect to Hprt1 used as reference gene. Dot plot shows means with one dot for each brain and significance assessed by 2way ANOVA Tukey’s multiple comparisons test. P-value < 0.05 (*), < 0.0001 (****) and ns = not significant. d Brain sections of ZIKV-infected CC005 mice fed with CTRL or PLX-containing food necropsied at 30 dpi were labeled with the monoclonal AT8 antibody (specific of pTau, green) with DNA labeled with Hoechst (grey). Z-projection images of the cortex obtained with a wide field microscope are shown. Scale bars = 100 µm. e The number of neurons displaying perinuclear cytoplasmic pTau within a z-slice of the hippocampus and the cortex of ZIKV-infected CC005 fed with CTRL or PLX-containing food and necropsied at 15 and 30 dpi, as determined using ImageJ software, was calculated with respect to the total area (in mm²) of the region analyzed. Dot plots show means with one dot for each brain and significance assessed by 2way ANOVA Sidak’s multiple comparisons test. P-value < 0.05 (*) and ns = not significant

Fig. 5.

A potential role for Apoe-expressing microglial cells in the progression of ZIKV-induced pTau pathology to NS2B− cortical neurons. a The level of expression of the indicated genes within the total RNA extracts of the entire right hemisphere of the brains analyzed in (Fig. 4e) was determined by RT-qPCR with respect to Hprt1 used as reference gene. Dot plot shows means with one dot for each mouse and significance assessed with respect to NI conditions by 2way ANOVA Sidak’s multiple comparisons test with ns = not significant. b Correlation analysis of gene expression levels and the number of AT8+ neurons measured at either 15 or 30 dpi in ZIKV-infected mice. Dots represent individual brains

Fig. 6.

ZIKV infection induces memory and social behavior defects in CC005 mice displaying high pTau content. Grip strength (a), rotarod (b) and NOR (c) tests were performed at 6, 15 and 30 dpi on NI and ZIKV-infected mice of CC071 and CC005 strains. Grip strength data were analyzed using a two-way ANOVA with repeated measures to assess overall effect of infection which p-value is shown in the upper-right corner. Pairwise t-tests with Bonferroni correction are shown for each day. Rotarod data were analyzed using day-by-day log-rank test. For the NOR test, we first compared the values of each group to 50% using a t-test to assess the preference of mice for the novel object (significance shown above each group). The p-value of the two-way ANOVA with repeated measures testing the overall effect of infection is shown in the upper-right corner. Pairwise t-tests with Bonferroni correction are shown for each day. d Selected behavioral events measured in CC005 and CC071 mice before infection (0 dpi) and at 6, 15 and 30 dpi. For each experimental mouse group and each night, the mean value of the two non-infected mice was subtracted from all values to set the mean value of NI mice to 0 across days (shown as a dashed line) and to improve the representation of the differences between non- and ZIKV-infected mice. The p-value of the two-way ANOVA with repeated measures testing the overall effect of infection is shown in the upper-right corner. Pairwise t-tests with Bonferroni correction are shown for each day. Dots represent individual mice. M (squares): males; F (circles): females. P-value > 0.05 (ns), < 0.05 (*), < 0.01 (**), < 0.001 (***), < 0.0001 (****)

Fig. 7.

A potential role for Apoe expression in memory loss induced following ZIKV infection at 30 dpi. The level of expression of a ZIKV RNA, b IFNs-I response, c inflammatory response and d DAM genes within the total RNA extracts of the entire right hemisphere of the brains of CC071 and CC005 mice was determined by RT-qPCR with respect to Hprt1 used as reference gene. Dot plot shows means with one dot for each mouse and significance assessed with respect to the NI conditions of the corresponding strain by 2way ANOVA Sidak’s multiple comparisons test. P-value < 0.05 (*), < 0.01 (**), < 0.001 (***), < 0.0001 (****). Not significant (ns) p-values are not shown. e Correlation analysis of the relative expression of Apoe and Tnf genes and ZIKV RNA and the fraction of time spent to explore the novel object measured at 30 dpi in ZIKV-infected CC005 mice. Dots represent individual brains. r = Pearson’s correlation coefficient

Results

ZIKV induces persistent pathological Tau phosphorylation in the brain of adult immunocompetent mice

The capacity of ZIKV infection to induce pathological hyperphosphorylation of the neuronal Tau protein (pTau) in the adult brain, which we have previously investigated in ZIKV-infected mice shortly, 6 days, after infection [40] was analyzed here in the long term in adult mice of three CC strains (CC001, CC005 and CC071) that we have previously identified as displaying different responses to ZIKV [47]. Indeed, during the monitoring of mortality and clinical signs previously carried out up to 14 days post-infection (dpi) [47], mortality was observed only in the case of mice from the CC071 strain and the brain histological lesions observed in CC071 and CC005 mice, were of lower intensity in the case of CC001 mice [47]. To analyze the induction of pTau in the long term, CC adult mice were inoculated with either ZIKV or saline solution by intracranial (i.c.) injection in the ventricular zone of the left hemisphere. ZIKV- and non-infected (NI, treated with saline solution alone) mice were necropsied at 6, 15, 30 and 60 days post-infection (dpi) (Fig. 1a). The left hemisphere was used for imaging by immunofluorescence (IF) or immunohistochemistry (IHC) and the right hemisphere was used for total RNA extraction and gene expression analysis. The presence of pTau was determined by IF and confocal microscopy using the monoclonal AT8 antibody that specifically binds to the pS202/pT205 Tau epitope, which is a hallmark of Alzheimer’s disease (AD) and other tauopathies [41, 50, 51]. It labels pTau at early and late stages of AD recognizing granular and dense forms of pTau present in pre and mature neurofibrillary tangles of neurons displaying Tau pathology [42].

As exemplified in a representative image of a ZIKV-infected adult mice brain incubated with the AT8 antibody at 30 dpi, ZIKV-induced pTau exhibited a dense perinuclear cytoplasmic (indicated by yellow arrows) as well as a punctiform and neuritic (indicated by yellow asterisks) distribution similar to the one observed with the AT8 antibody in post-mortem human brain tissue from AD patients [42]. Under the same conditions, no labeling with the AT8 antibody was observed in NI brains (Fig. 1b, NI). The number of AT8+ neurons present on a confocal Z-slice was quantified for the three mouse strains at 6, 15, 30 and 60 dpi throughout the entire brain section except the cerebellum and calculated with respect to the area (in mm²) of the region analyzed (Fig. 1c). Only those neurons that displayed perinuclear AT8 labeling as exemplified in Fig. 1b by yellow arrows were considered as AT8+. The total number of AT8+ neurons present in the entire left hemisphere (Fig. 1c) was compared to the total level of ZIKV RNA (Fig. 1d) quantified by RT-qPCR from total RNA collected from the entire right hemisphere of the same brains (Fig. 1d). The presence of pTau as recognized by the AT8 antibody was observed in the brain of the three CC mouse strains up to 60 dpi (Fig. 1c where each independent brain is represented by a dot). No (in the case of CC005 mice) or a very low (in the case of CC001 and CC071 mice) number of AT8+ neurons were detected at 6 dpi (Fig. 1c) even though 6 dpi corresponded to the time for which the highest ZIKV RNA content was detected in the brain of the three strains (Fig. 1d). A significant increase of AT8+ neurons occurred between 6 and 15 dpi in the brains of CC001 and CC005 mice that persisted until 60 dpi, with CC001 and CC005 mice displaying significantly higher levels of AT8+ neurons than CC071 mice (Fig. 1c). The presence of ZIKV-induced AT8+ neurons that persisted until 60 dpi remained significantly unchanged between 15 and 60 dpi in the brains of the three mouse strains regardless that the total level of ZIKV RNA in the corresponding brains significantly diminished during the same period of time (Fig. 1d). In the case of CC001 and CC005 mice, the number of AT8+ neurons significantly increased in the hippocampus (HP), the cortex (CX) as well as outside HP and CX (Fig. S1) with a significantly higher number of AT8+ neurons present in the HP and the CX compared to the region outside HP and CX (Fig. S1). Under conditions used in this work, no sex-bias was detected concerning the capacity of ZIKV infection to induce AT8+ neurons in the different brain regions analyzed (Fig. S1). The total number of AT8+ neurons at 15, 30 and 60 dpi was significantly lower in CC071 compared to CC001 and CC005 brains (Fig. 1c). This occurred even though the level of ZIKV RNA reached a similar peak at 6 dpi in the brains of the three strains. However, while the level of ZIKV RNA remained stable until 15 dpi in CC001 and CC005 brains, it significantly dropped between 6 and 15 dpi in CC071 brains (Fig. 1d).

Overall, these results show that ZIKV infection of adult mouse brains induces a persistent AT8+ Tau pathology similar to the one observed in human AD brains. The onset of ZIKV-induced pTau was delayed with respect to the peak of ZIKV RNA levels leading to a significant increase of AT8+ neurons in the case of CC001 and CC005 mice for which the level of ZIKV RNA remained stable between 6 and 15 dpi. The significant increase of the number of AT8+ neurons established at 15 dpi persisted at least until 60 dpi even though the total level of ZIKV RNA in the same brains significantly diminished, thus raising the question of the link between neuronal infection and pTau at the cellular level.

pTau is present in infected (NS2B+) and non-infected (NS2B−) neurons

In the brain of mice and humans, ZIKV exclusively infects neurons [39, 40, 52]. In order to analyze the link between neuronal infection and pTau induction, the co-localization of pTau (AT8 staining) with the viral protein NS2B was analyzed by confocal microscopy in the brain of CC005 and CC001 ZIKV-infected mice (that displayed significant high levels of AT8+ neurons throughout infection). Alongside with labeling pTau and NS2B, microglial cells were labeled with an anti-IBA1 antibody and DNA with Hoechst (Fig. 2). At 15 dpi, a strong staining with the anti-NS2B antibodies was observed in the hippocampus, predominantly in the CA2 region with few NS2B+ positive neurons also visible in the CA1 region of the hippocampus (Fig. 2a). On the same z-slice, AT8+ neurons were visible both in the CA2 and the CA1 region. In agreement with previously published results [39, 40, 53], microglial cells strongly expressing IBA1 and displaying an amoeboid morphology with large cell bodies and short thick ramifications were seen invading both the CA2 and CA1 regions of the hippocampus (Fig. 2a). A situation very different from the one observed in the hippocampus and the cortex of NI brains where microglial cells were non-invasive and displayed a highly ramified morphology with many long, thin neurites (Fig. S2a, b). The AT8 and NS2B labeling strongly co-localized in the CA2 region (Fig. 2a and crop a), indicative of pTau being induced in ZIKV-infected neurons. This was not the case in the CA1 region (Fig. 2a and crop b) where most AT8+ neurons did not express NS2B. All the AT8+ neurons appeared closely surrounded by microglial cells, both in the CA2 and the CA1 region, independently of the presence of NS2B. Occasionally, IBA1, NS2B and AT8 staining co-localized (as indicated by yellow arrows in Fig. 2 crop a and b), a situation that could be translating microglial cells phagocyting ZIKV-infected, AT8+ neurons. Engulfing by microglial cells could also observed in the case of NS2B+ AT8− cells (Fig. S2c) whereas in NI brains, no engulfment was observed even in the case of microglial cells positioned closely to neurons (Fig. S2a, b).

The presence of AT8, NS2B and IBA1 staining was also analyzed in the cortex region in the same z-slice (Fig. 2b). AT8+ neurons were visible in the cortex even though no NS2B labeling was detected in this region of the brain. Enlargement of this region confirmed the presence of AT8 in neurons devoid of NS2B (Fig. 2b, crop a), which again appeared closely surrounded by IBA1-expressing microglial cells.

The absence of AT8/NS2B co-localization observed in the CA1 region was not a specific trait of the neurons present in this region since in the case of a brain displaying high NS2B content in the CA1, most AT8 staining co-localized with NS2B (Fig. S3a) with only one AT8+ neuron not co-localizing with NS2B (Fig. S3a, yellow arrow). Also, results shown here were not specific to CC005 mice since similar results were obtained while analyzing the hippocampus and the cortex of a ZIKV-infected CC001 mouse at 15 dpi (Fig. S3b).

The observation that pTau could be present not only in ZIKV-infected (NS2B+) neurons but also in non-infected (NS2B−) neurons raised the question of the potential spreading of pTau from infected to non-infected neurons. A question that was investigated in the brain of the CC005 mice previously analyzed in Fig. 1c and d that alongside with displaying a significant increase of AT8+ neurons from 6 to 60 dpi also displayed a significantly higher peak level of ZIKV RNA at 6 dpi compared to CC001 and CC071 mice (Fig. 1d).

pTau progresses to cortical neurons in the absence of detectable ZIKV infection

In order to gain insight into the propensity of pTau to spread to non-infected cortical neurons, the corresponding distribution of NS2B+ and AT8+ neurons was quantified throughout infection in the hippocampus and the cortex in the same CC005 brains than those previously analyzed in Figs. 1 and 2. NS2B+ neurons were almost exclusively present in the hippocampus: in the CA2 region at 6 dpi and either in the CA2 or the CA1 region at 15 dpi diminishing thereafter at 30 and 60 dpi (Fig. 3a) with little or any NS2B+ neurons visible in the cortex at 6, 15, 30 and 60 dpi (Fig. S4a). The percentage of infected cells specifically present in the CA region that was significantly enhanced at 6 and 15 dpi with respect to NI conditions (Fig. 3b) strongly diminished and was no longer significant at 30 and 60 dpi (Fig. 3b). This diminution was not related to a global diminution of neurons in the CA region since the total number of neurons/mm² in the CA was not significantly affected throughout infection (Fig. 3c), but rather translated the capacity of infected neurons to inhibit viral replication after 15 dpi in agreement with the significant diminution of ZIKV RNA levels measured in the same brains by qPCR (Fig. 1d). Even though variable, the percentage of NS2B+ neurons displaying perinuclear AT8+ staining in the CA region was significantly enhanced at 15 dpi (Fig. 3d).

The quantification of NS2B+ cells in the hippocampus versus the cortex showed that while the number of NS2B+ neurons significantly increased with respect to NI conditions at 6 and 15 dpi in the hippocampus, no significant increase was observed in the cortex either at 6, 15, 30 or 60 dpi (Fig. 3e). Despite the absence of a significant induction of NS2B+ neurons in the cortex, the number of AT8+ neurons in the same brains significantly increased at 15, 30 and 60 dpi with respect to NI conditions not only in the hippocampus but also in the cortex (Fig. 3e). These results are in agreement with the hypothesis of a potential spreading of pTau towards non-infected (NS2B−) regions after the arrest of ZIKV-replication.

Active microglial cells persistently surrounded cortical AT8+ NS2B− neurons (Fig. 3f) until 60 dpi, notwithstanding a significant decrease of the expression of the Aif1 gene coding for IBA1 (as measured in total RNA extract from the right hemisphere), which translates microglia activation (Fig. 3g). The presence of AT8+ NS2B− cortical neurons surrounded by active microglial cells at late time p.i. was not a phenomenon restricted to the brain of ZIKV-infected CC005 mice but was also visible in the brain of ZIKV-infected CC001 mice (Fig. S4b).

The persistent presence of active microglial cells in the vicinity of AT8+ NS2B− cortical neurons suggested a potential role for microglial cells in the progression of pTau to NS2B− cortical neurons.

Microglial cells participate in the progression of pTau to NS2B− cortical neurons

In order to test the potential role of microglial cells in the progression of pTau to NS2B− cortical neurons, microglial cells of the CC005 mice were partially depleted by treating the mice with PLX3397 (PLX), an inhibitor of the colony stimulating factor receptor 1 (CSFR1) necessary for microglia survival in the adult brain, that was added to the food pellet. CC005 mice were treated or not with PLX, 30 days before and all the time after infection. First, the effect of PLX treatment on the number of microglial cells was assessed in non-infected mice either by IHC (left hemisphere) using an anti-IBA1 antibody or by total brain expression analysis (right hemisphere) of the P2ry12 gene that, in the brain, is specifically expressed by microglial cells [54]. Under NI conditions, PLX-treatment significantly diminished the number of microglial cells present within the hippocampus and the cortex to respectively 37 and 45% of the average number of microglial cells counted in control (Ctrl)-fed NI mice (Fig. 4a and Fig. S5). Consistently with IHC results, gene expression analysis showed that PLX treatment significantly diminished the expression of P2ry12 to 53% of the average level of P2ry12 expression measured in total Ctrl brain extracts (Fig. 4b). Under the same conditions, PLX treatment had no significant effect on the expression of the Cx3cl1 gene (predominantly expressed by neurons in the CNS [55] and slightly increased the expression of the Gfap gene predominantly expressed by astrocytes (Fig. 4b). Thus, under NI conditions, PLX treatment partially reduced by ≈50% the number of microglial cells without significantly affecting the number of neurons or astrocytes.

Second, the effect of PLX treatment was analyzed under infection conditions (Fig. 4c). Under the conditions used here, PLX treatment had no significant effect on the level of ZIKV RNA (Fig. 4c) while diminishing the level of expression of the microglial P2ry12 gene (itself not affected by ZIKV infection under Ctrl conditions) at 15 and 30 dpi without significantly affecting the expression level of the Cx3cl1 and Gfap genes (Fig. 4c). Thus, PLX treatment diminished the number of microglial cells under ZIKV-infected conditions as previously observed under NI conditions without significantly affecting brain ZIKV content or the number of neurons and astrocytes. However it must be noticed that the effect of PLX on the number of microglial cells (as translated by P2ry12 expression) in ZIKV-infected brains was only partial, becoming significant with respect to NI_Ctrl conditions only at 30 dpi (Fig. 4c). A slight (but not significant) diminution of Cxc3cl1 expression was observed at 15 dpi in Ctrl mice that was attenuated in the presence of PLX, which could correspond to the phagocytosis of ZIKV-infected neurons by microglial cells as visualized in the hippocampus of ZIKV-infected CC005 mice (Fig. 2a). However, neither the diminution of Cx3cl1 expression in Ctrl mice nor the effect of PLX at 15 dpi were statistically significant (Fig. 4c) indicating that if there was any phagocytosis of ZIKV-infected neurons at this time, it concerned only a small fraction of neurons. These results are in agreement with those shown in Fig. 3c that indicated that the total number of neurons in the CA region was not significantly affected throughout infection.

Third, the effect of the partial diminution of microglial cells on the number of AT8+ neurons was quantified in the hippocampus and the cortex of the same brains as those analyzed in Fig. 4c, following IF labeling with the AT8 antibody and wide field microscopy (Fig. 4d, e). The treatment with PLX had no effect on the number of AT8+ neurons in the hippocampus at either 15 or 30 dpi (Fig. 4e) and had no effect on the number of AT8+ neurons in the cortex at 15 dpi (Fig. 4e). However, it nearly significantly diminished the number of AT8+ cortical neurons at 30 dpi (Fig. 4e) with the significant increase of AT8+ cortical neurons observed between 15 and 30 dpi under Ctrl conditions becoming non-significant in PLX-treated mice, consistent with the hypothesis of a role for microglial cells in facilitating the progression of pTau at 30 dpi. It cannot nevertheless be excluded that the absence of an effect observed at 15 dpi could be related to the weaker effect of PLX treatment on the number of microglial cells as translated by the not significant differences on P2ry12 gene expression levels between NI_Ctrl and ZIKV_15dpi_PLX conditions (Fig. 4c).

PLX-susceptible Apoe expression is correlated with the increase of ZIKV-induced pTau at 30 dpi

ApoE is the major genetic risk factor associated with AD [56]. It is encoded by the Apoe gene whose up-regulation has been associated with microglial-dependent spreading and progression of pTau [57, 58]. Astrocytes are the main cell type expressing Apoe in the normal brain. However, under certain pathological conditions, Apoe becomes predominantly expressed by microglial cells [56]. Particularly in human AD brains, Apoe expression is down-regulated in astrocytes and up-regulated in reactive microglial cells [59]. The activation of the Apoe gene expression in microglial cells is a key element in driving homeostatic microglia towards a disease associated microglia (DAM) response through an ApoE-TREM2 pathway [60].

In order to investigate the potential role of ApoE in the microglial-dependent progression of pTau at 30 dpi, Apoe gene expression was analyzed in the brain RNA extracts of control and PLX fed mice analyzed in Fig. 4c, d and e. In the absence of PLX, the expression of Apoe that was significantly increased between NI and 15 dpi continued to significantly increase between 15 and 30 dpi (Fig. 5a) despite the diminution of the level of ZIKV RNA and Gfap expression during this period of time (Fig. 4c). While PLX treatment had no effect on the increased expression of the Apoe gene between NI and 15 dpi, it inhibited the progression of Apoe expression between 15 and 30 dpi (Fig. 5a) indicative of Apoe becoming at least partly expressed by microglial cells beyond 15 dpi. A significant diminution of the expression of the gene coding for TREM2 (Fig. 5a) that drives Apoe expression in microglial cells [60], was also observed at 30 dpi under PLX but not NI conditions. On the contrary, no specific effect of PLX treatment was observed in the case of the expression of the Clec7a gene (Fig. 5a) coding for CLEC7A, a factor that has also been associated with pTau spreading [61] nor was it the case of the C3 gene (Fig. 5a) whose up-regulation has been associated with increased tauopathy and neurodegeneration in link with microglia-dependent phagocytosis [62]. The expression levels of the Clec7a and C3 genes as well as that of major genes related to neuroinflammation (Nos2, Il1b and Tnf) were significantly enhanced between NI and 15 dpi before significantly decreasing between 15 and 30 dpi (Fig. 5a) both under Ctrl and PLX conditions, in correlation with the evolution of ZIKV RNA levels (Fig. 4c).

In agreement with a potential role for ApoE and TREM2 in facilitating the progression of pTau at 30 dpi, the expression of the Apoe and Trem2 genes appeared significantly positively correlated with the number of AT8+ neurons/mm² (hippocampus + cortex) at 30 dpi, while this was not the case at 15 dpi (Fig. 5b). Under the same conditions, no correlation was observed between the total number of AT8+ neurons and the expression of the Clec7a gene, neither at 15 nor at 30 dpi (Fig. 5b).

The PLX-susceptible expression profile of the Apoe gene and its positive correlation with the number of ZIKV-induced AT8+ cells at 30 dpi is in concordance with the hypothesis that ApoE could be playing a role in a microglial-dependent progression of pTau at 30 dpi. They suggest that the number of microglial cells displaying a DAM Apoe-expressing phenotype, which has been associated with AD pathogenesis in humans and mice [45], could be increased following brain ZIKV infection at 30 dpi.

ZIKV infection induces memory and social behavior impairments in CC005 but not in CC071 mice

The progression of pTau, which is a hallmark of tauopathies such as AD, is associated with cognitive decline, predominantly characterized by memory loss [19]. In order to investigate if the induction and progression of pTau could lead to cognitive impairments, the effect of ZIKV infection on behavioral and cognitive abilities was analyzed in CC005 mice (high pTau content) in comparison with CC071 mice that displayed a significantly lower pTau content starting from 30 dpi (Fig. 1c). For this, groups of CC071 and CC005 mice either non- or ZIKV-infected (total of 6 male and 6 female mice per strain and condition) were submitted to a series of tests, compatible with the biosafety confinement of ZIKV-infected mice, to investigate behavioral traits such as: memory (novel object recognition test, NOR) and social behavior (live mouse tracker test, LMT) as well as muscular strength (grip strength test) and balance and coordination (rotarod test). The same groups of mice were submitted to the different tests. After a period of habituation, baseline values were determined 5 days before i.c. inoculation of ZIKV or saline solution (shown as dpi 0 on graphs). Measurements were then carried out at 6, 15 and 30 dpi (Fig. 6). The influence of sex was tested by two-way ANOVA (sex and infection status) for every parameter. Non-infected CC005 female felt more often off the rotarod than males (logrank, p = 0.037) and CC005 females (non- or ZIKV-infected) showed more strength than males (two-way ANOVA, p = 0.021). However, these sex differences did not interact with the effects of ZIKV infection. Therefore, the data for males and females were pooled for all statistical analyses, but are presented with different shapes on the graphs.

Grip strength data were analyzed using a two-way ANOVA with repeated measures and revealed no overall effect of infection in CC005 and CC071 mice (p = 0.57 and p = 0.62, respectively). Pairwise t-tests with Bonferroni correction did not identify dpi-specific differences (Fig. 6a). Analysis of rotarod data using log-rank test revealed no differences between non- and ZIKV-infected mice in either strain at any time point (Fig. 6b).

The NOR test used to evaluate the effect of ZIKV infection on memory is based on the innate preference of mice for novelty. The non-infected CC071 and CC005 mice as well as the ZIKV-infected CC071 mice spent a significantly higher fraction of time exploring the novel object as compared to the known object (t-test, significance shown above each group) as expected in the case of functional short-term memory. The two-way ANOVA with repeated measures did not identify a significant overall effect of infection in CC005 and CC071 mice (p = 0.16 and p = 0.28, respectively), however a significant difference was observed on day 30 between non-infected and ZIKV-infected CC005 mice (pairwise t-tests with Bonferroni correction, p = 0.022) with the infected mice having lost the ability to distinguish a novel from a known object (Fig. 6c), translating a ZIKV-induced short-term memory loss that was not observed in CC071 mice (Fig. 6c).

The behavior of NI or ZIKV-infected CC005 and CC071 mice was individually analyzed while evolving within a group of four mice (two NI and two ZIKV-infected) of the same strain. At every time point, the activities of the group of mice were tracked for two successive nights using the LMT system [48]. Significant differences in several parameters reflecting social behaviors were observed between NI and ZIKV-infected CC005 but not CC071 mice (Fig. 6d). ZIKV-infected CC005 mice established a significantly higher number of approach contacts compared to CC005 NI mice but they also displayed a significantly higher number of contact breaks and longer time in getaway events (Fig. 6d). They also displayed an enhanced number of isolated moving events associated with significantly increased distance traveled. None of these features were modified by ZIKV infection in CC071 mice (Fig. 6d).

Contrary to memory defects that were significant at 30 dpi, social behavior defects became significant earlier after infection at 15 and even at 6 dpi in the case of getaway, distance traveled and isolated moving events (Fig. 6d). At 15 dpi, CC005 and CC071 mice displayed not significantly different levels of pTau (Fig. 1c). However, at 15 dpi, ZIKV-infected CC005 displayed significantly higher expression levels of ZIKV RNA (Fig. 7a), type I IFN (Fig. 7b), pro-inflammatory (Fig. 7c) and DAM (Fig. 7d) response genes than CC071 mice. In the case of CC071 mice, none of the genes tested here remained significantly increased with respect to NI conditions beyond 6 dpi. In the case of CC005 mice, the expression levels of ZIKV RNA, type I IFN response (Ifnb1, Irf7 and Eif2ak2) and pro-inflammatory (Tnf, Il6 and C3) genes remained significantly increased with respect to NI mice only up to 15 dpi decreasing thereafter to levels not significantly different from NI conditions. On the contrary, the expression of the DAM associated genes remained significantly increased beyond 15 dpi, up to 30 (in the case of Trem2 and to a lesser extent Clec7a) and up to 60 dpi in the case of the Apoe gene (Fig. 7d). Thus, the onset of social behavior defects at 6–15 dpi displayed by CC005 but not CC071 mice (Fig. 6d), appeared associated with a significant induction of high levels of ZIKV RNA, IFN-I and pro-inflammatory response genes maintained up to at least 15 dpi rather than with significantly higher pTau levels. On the contrary, memory impairment that was established in CC005 but not CC071 mice at 30 dpi (Fig. 6c), appeared associated with increased pTau, Apoe and Trem2 gene expression. In agreement with the hypothesis of ApoE potentially playing a role in ZIKV-induced memory loss, the level of Apoe expression measured at 30 dpi in the total brain RNA extracts of the ZIKV-infected CC005 mice submitted to the NOR test (Fig. 6c) was significantly negatively correlated with the fraction of time spent to explore the novel object displayed by these mice (Fig. 7e) indicating a positive correlation between Apoe expression and short-term memory loss. Under the same conditions, no correlation was observed between memory loss and ZIKV RNA or Tnf gene expression levels at 30 dpi (Fig. 7e).

Discussion

Results shown in this work demonstrate that brain ZIKV infection of adult immunocompetent mice leads to a persistent pathological phosphorylation of the neuronal Tau protein recognized by the AT8 antibody similar to the one depicted in human AD brains. At 6 dpi, when high levels of ZIKV RNA are expressed, only a small number of AT8+ neurons were visible in a few brains of CC071 and CC001 mice consistent with our previous results [40]. The significant increase of AT8+ neurons started at 15 dpi and remained significant at least until 60 dpi in all the brains of CC001 and CC005 mice that, contrary to CC071 mice, displayed sustained high levels of ZIKV RNA expression until 15 dpi. Differences observed between the three mouse strains studied in this work indicate that there is no link between a significantly sustained induction of pTau and mortality which was reported in the case of i.c. ZIKV-infected CC071 but not CC001 mice [47]. More recently, mortality has also been reported in the case of i.c. ZIKV-infected female but not male CC005 mice at later (17–20 dpi) times post-infection [63]. However, no effect of biological sex on ZIKV-induced pTau was observed in this work. Nevertheless, further analysis using an increased number of female and male animals is required before reaching a definitive conclusion regarding the absence or presence of a sex-bias for ZIKV-induced pTau.

Even though the presence of pTau was maintained at significantly high levels from 15 to 60 dpi in CC001 and CC005 mice, differences were observed throughout the course of infection and between brain regions concerning the link of pTau with ZIKV-infected neurons and the role of microglial cells in the induction of pTau.

In the hippocampus, AT8+ neurons were observed in the CA2 and CA1 regions that corresponded to the regions where NS2B+ ZIKV-infected neurons were detected. In previous work we have shown that in vitro, pTau is induced in the primary cultured neurons infected by ZIKV [40]. In line with these previous in vitro observations, in vivo results obtained in this work showed that most of AT8+ labeling in the hippocampus at 15 dpi occurred in NS2B+ neurons. The neurons where only AT8+ labeling was detected were positioned in regions of infection, near an AT8+ NS2B+ positive neuron and it cannot be excluded that AT8+ NS2B− neurons observed in the CA2 and CA1 regions could be infected neurons where NS2B was no longer detected. At 15 dpi, CA2 and CA1 AT8+ neurons were closely surrounded by non-infected, reactive microglial cells that displayed large, poorly ramified amoeboid-like cell bodies with some microglial cells engulfing ZIKV-infected neurons suggesting a potential phagocytic event. The diminution of the number of microglial cells following PLX treatment had no significant effect on the number of AT8+ neurons present in the hippocampus neither at 15 nor at 30 dpi. These results suggest that the induction of pTau in hippocampal neurons would occur independently of microglial cells at 15 and 30 dpi. It should however be noted that the effect of PLX on the number of microglial cells, which was only partial, was weaker at 15 than 30 dpi and thus it cannot be excluded that the lack of effect of PLX treatment at 15 could be related to a weaker effect of the treatment on the number of microglial cells. The quantification of NS2B+ cells in the CA and of total ZIKV RNA levels in the brain of CC005 mice indicate that at 30 dpi, viral replication has been for the most stopped but we cannot totally exclude that AT8+ NS2B− cells observed in the hippocampus at 30 dpi corresponded to cells infected at earlier times. We have previously shown that ZIKV infection induces the phosphorylation of the residue Tyr216 of kinase GSK3B in primary cultured neurons, enhancing the proportion of the active form of kinase GSK3B in ZIKV-infected compared to NI neurons [40]. Since GSK3B is one of the main kinases responsible for the induction of pTau [64, 65], its activation following neuronal ZIKV infection could contribute to the induction of pTau in ZIKV-infected neurons. The uptake of pTau by “healthy” neurons has been proposed as a mechanism of pTau progression [66]. The uptake of pTau present in ZIKV-infected neurons by neighboring neurons could lead to a microglia-independent neuron-to-neuron progression of pTau.

A different situation was observed in the cortex. Cortical pTau was consistently detected in NS2B− neurons, with the significant persistent increase of cortical AT8+ neurons in the cortex occurring while no significant levels of infection were detected in this region at any time post-infection. As in the case of AT8+ NS2B− neurons present in the hippocampus, we cannot exclude that the AT8+ NS2B− neurons identified in the cortex could also be ZIKV-infected neurons where NS2B could not be detected. However, this hypothesis is weakened by the fact that contrary to the hippocampus, significant numbers of NS2B+ neurons were never detected in the cortex from 6 to 60 dpi. Notwithstanding the absence of infection, AT8+ cortical neurons were also closely associated with active microglial cells all throughout infection. The presence of pTau in the cortex at 15 dpi was resistant to PLX treatment suggesting that it could be independent of microglial cells but again, as in the case of the hippocampus, this could result from the weaker effect of PLX on the number of microglial cells induced at this time.

Contrary to results obtained in the hippocampus, the partial diminution of the number of microglial cells induced following PLX treatment partially inhibited the level of pTau in the cortex at 30 dpi suggesting a role for microglial cells in the cortical progression of pTau at 30 dpi. A role for PLX-susceptible microglia in pTau spreading has been demonstrated in different mouse models in relation with microglial dependent exosome synthesis and extracellular vesicles [61, 67]. Also a role for inflammation in pTau propagation has been reported [66] that could depend not only on microglial cells but also on peripheral immune cells also potentially affected during PLX treatment [68]. The increase of cortical pTau neurons at 30 dpi occurred at a time when the expression of genes associated to the IFN-I and pro-inflammatory response under either control or PLX conditions was not significantly different form the expression levels measured in NI mice under the same conditions. The peak of the expression of these genes occurred at 15 dpi significantly dropping down from there and was not affected by PLX treatment. On the contrary, the progression of pTau at 30 dpi was associated to significantly high levels of Apoe and Trem2 gene expression that at 30 dpi were susceptible to PLX and positively correlated to pTau increase, supporting a potential role for Apoe expressing microglia in regulating cortical pTau progression at 30 dpi independently of inflammation.

The Apoe gene that codes for the main lipid transporter in the CNS is the strongest genetic factor for non-familial sporadic AD [56, 59]. In the “ApoE cascade hypothesis”, ApoE is described as potentially intervening in AD pathways related among others with membrane integrity, extracellular vesicular physiology and neuroinflammation [69]. The Apoe gene that dramatically impacts pTau neurodegeneration [56], has been identified in human cohorts as playing a central role in pTau progression [70, 71]. In mouse models, the effect of ApoE in pTau-mediated pathologies is very often described in relation with microglia, through its capacity to modulate microglia activation, priming microglia to a pro-inflammatory response [59] and driving the transcriptional neurodegenerative phenotype of microglia through an ApoE-TREM2 axis [57, 60]. Our results are not in favor of a role of inflammation in pTau progression and thus would rather suggest that the propagation of cortical pTau at 30 dpi could in contrast be related with ApoE-dependent extracellular vesicle trafficking.

All throughout this work, cellular analyzes (IHC and IF) were carried out in the left hemisphere where ZIKV was i.c. inoculated while gene expression analyzes were carried out using RNAs collected from the right hemisphere. However, even though infection was carried in the left hemisphere, ZIKV RNA was detected in the right hemisphere indicative of the diffusion of the virus from the left to the right hemisphere. Also, the kinetics of ZIKV expression level (measured in the right hemisphere, Fig. 1d) very closely followed the kinetics of infection (number of NS2B+ cells) measured in the left hemisphere (Fig. 3c), which is indicative of equal rates of infection in both hemispheres.

In summary, the presence of pTau in ZIKV-infected brains appeared as resulting from two different mechanisms, the first one being resistant to PLX administration and as such for the most independent of microglia, the second one being susceptible to PLX and thus partly dependent on microglia (Fig. 8). The PLX-resistant induction of pTau was associated with PLX-resistant high levels of ZIKV in the CA2 and CA1 regions of the hippocampus and PLX-resistant IFN-I and pro-inflammatory gene expression (Fig. 8). The PLX-susceptible progression of pTau to cortical neurons at 30 dpi, after viral resolution, was positively correlated with PLX-susceptible Apoe and Trem2 expression. While short-term memory impairment occurred at 30 dpi concomitantly with PLX-susceptible Apoe expression and cortical pTau progression, this was not the case of social behavior alterations that peaked at 15 dpi concomitantly with high ZIKV and pro-inflammatory gene levels (Fig. 8) rather than significant higher levels of pTau. The induction of social behavior alterations at 15 dpi could result from the preferential infection of the CA2 region that has been implicated in the regulation of social behavior [72]. Thus, while memory defects observed here following brain viral infection resemble those observed at early stages of AD, this is not the case of the ZIKV-induced social behavior alterations that were induced earlier after infection, before memory defects and pTau progression, contrary to AD where behavioral changes appear at late stages of the disease [18, 19]. Differences observed in this work on pTau establishment at 30 dpi in the hippocampus (PLX-resistant) and the cortex (PLX-susceptible), raise the question of the potential existence of different populations of microglial cells. Future spatial transcriptomic analyses should help to solve this question that is beyond the scope of this work. Since infection occurs only in the hippocampus, the capacity of infected neurons to differentially signal nearby microglial cells should be considered.

Fig. 8.

Brain ZIKV infection induces a persistent accumulation of pTau in the brain of adult immunocompetent mice associated with social behavior and memory impairment. Six days after i.c. ZIKV infection, the NS2B protein encoded by ZIKV is found present in the CA2 region of the hippocampus of adult immunocompetent CC005 mice associated with the induction of type I interferon (IFN-I) and pro-inflammatory responses. The PLX-resistant brain expression of ZIKV RNA and of genes of the IFN-I and pro-inflammatory response peaks at 15 dpi, decreasing from there. Following ZIKV brain infection, a PLX-resistant induction of pTau is observed in the hippocampus and the cortex starting from 15 dpi, followed at 30 dpi by a PLX-susceptible progression of pTau to cortical neurons that persists at least until 60 dpi correlated with a PLX-susceptible increase of the expression of the DAM Apoe and Trem2 genes, in correlation with the induction of significant memory impairment in CC005 mice displaying high pTau content

Viral encephalitis, for which therapeutic strategies (other than anti-HSV treatment) are lacking, is a growing concern for clinicians. By shedding light on the mechanism that drives pTau progression in link with memory impairment, which is one of the main characteristics of viral encephalitis [2, 73] as well as AD [18, 19], our work contributes to identifying potential therapeutic approaches that could attenuate virus-induced short-term memory decline as well as pTau progression.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

MCG, XM and EB conceived and supervised this study. GC, ZM and HS performed most of the experiments and analyzed data. XM, FdC and GC designed behavior equipment and analyzed data. LC, JPMO and VB performed experiments. CM, LB and FN provided support and conceptual advice. EB, XM, MCG, GC, HS and ZM designed experiments. EB wrote the paper. MCG, XM, GC and CM contributed in writing the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the following grants: Agence Nationale de la Recherche (ANR) NeuroZika (ANR-20-CE16-0017) and the French Government’s Investissement d’Avenir program, Laboratoire d’Excellence: IBEID (Integrative Biology of Emerging Infectious Diseases, ANR-10-LABX-62-IBEID) as well as grants from CNRS (Centre National de la Recherche), Inserm (Institut National de la Santé et de la Recherche Médicale) and Université Paris Cité.

Data availability

The data that support the findings of this study and the R scripts are available from the corresponding author upon reasonable request.

Declarations

Ethics approval

All animal experimental protocols were approved by the Institut Pasteur Ethics Committee (dap190107 and dap230008) and authorized by the French Ministry of Research (#19469 and #43891), in compliance with French and European regulations.

Competing interests

The authors declare that they have no competing interests.