Phosphorylated tau exhibits antimicrobial activity capable of neutralizing herpes simplex virus 1 infectivity in human neurons

Abstract

Tau is a microtubule-associated cytoskeletal protein, which, when hyperphosphorylated and aggregated, can result in a myriad of different tauopathies, including Alzheimer's disease (AD). We previously showed that the principal component of senile plaques, amyloid beta (Aβ), is an antimicrobial peptide capable of binding and entrapping microbial pathogens. Here we show that tau is hyperphosphorylated in neurons in response to viral infection and can neutralize herpes simplex virus 1 (HSV-1) infectivity by directly binding to viral capsids. Our data suggest that the 'pathogenic' characteristics of tau hyperphosphorylation, microtubule destabilization and aggregation are part of an antiviral response, in which tau serves as a host defense protein in the innate immune system of the brain. The combined antimicrobial activities of Aβ and phosphorylated tau resulting in Aβ plaques and neurofibrillary tangles, along with neuroinflammation, suggest that AD neuropathology may have evolved as an orchestrated innate immune host defense response to microbial infection in the brain.

Fig. 1. Phosphorylation of tau is key to inhibiting HSV-1 infection and viral plaque formation in 2D human neuronal cell culture.

2D ReNcell VM cultures were preincubated with dilutions of synthetic 2N3R tau, 2N4R tau or 2N4R GSK-3β p-tau followed by HSV-1 infection to assay the antimicrobial protective properties of tau. After HSV-1 infection, whole wells were imaged by confocal microscopy and analyzed for red fluorescence. a, Images comparing HSV-1 plaque formation in the absence or presence of 2N4R GSK-3β p-tau (1.25 µg ml⁻¹). b−f, Whole-well images from the 2N4R GSK-3β p-tau (scale of 2.5 µg ml⁻¹ to 0.625 µg ml⁻¹) pretreated conditions were analyzed using Nikon Elements GA3 to compare the number of HSV-1 plaques (F_{3, 54} = 7.069, **P = 0.0015) (b), the number of HSV-1 single-cell infections (F_{3, 54} = 2.754, *P = 0.0291) (c), the ratio of plaques to single-cell infections (F_{3, 54} = 6.428, *P = 0.0169) (d), the size of individual HSV-1 plaques by area (*P = 0.0260) (e), and the average size of HSV-1 plaques by area distributed into quadrants (F_{3, 1,376} = 1.859, **P = 0.0022) (f). In addition, whole well images from the 2N3R and 2N4R tau conditions (1.25 µg ml⁻¹) were analyzed using Nikon Elements General Analysis 3 (GA3) to compare the number of HSV-1 plaques (g) and the number of HSV-1 single-cell infections (h). Box plots are representative of ±s.e.m. (n = 11) depicting median and interquartile range, with whiskers denoting variability according to Tukey’s method. Statistical mean comparisons were calculated by one-way ANOVA using Dunnett’s multiple comparisons test (b−d), one-way ANOVA using Tukey’s multiple comparisons test (g,h), two-tailed Kolmogorov−Smirnov test (e) for plaque size distribution and two-way ANOVA using Sidak’s multiple comparisons test (f). Source data

Fig. 2. Tau binding to HSV-1 viral capsid proteins is potentiated by microtubule-binding repeats and phosphorylation.

Dilutions of synthetic tau isoforms were incubated with heat-immobilized HSV-1 capsid or whole virion in indirect and competitive ELISAs to measure tau−virus binding affinity. a, 2N4R GSK-3β p-tau binding between HSV-1 isolated capsid or whole virion was compared after normalization of available binding sites assessed using VP21/VP22a antibody (****P < 0.0001). b, Dilutions of 2N4R GSK-3β p-tau (scale of 5 µg ml⁻¹ to 0.3125 µg ml⁻¹) were incubated with immobilized HSV-1 capsids (F_{4, 55} = 43.18, ****P < 0.0001). c, The binding affinity of 2N4R GSK-3β p-tau to HSV-1 capsids was compared to 2N3R, 50/50 mix of 2N3R/2N4R and 2N4R tau (F_{3, 44} = 8.996, *P = 0.267 ((2N4R GSK-3β p-tau versus 2N3R tau, ***P = 0.0001), (2N4R GSK-3β p-tau versus 2N3R/2N4R tau, ***P = 0.0007)). d,e, Tau adhesion inhibition was assessed by preincubation of immobilized HSV-1 capsids with an anti-VP21/VP22a antibody (F_{8, 32} = 7.071, ****P < 0.0001) (d) or an anti-pUL48-VP16 antibody (F_{8, 32} = 3.978, ***P = 0.004, ****P < 0.0001) (e) before exposure to 2N4R GSK-3β p-tau. f, Tau adhesion inhibition was repeated with the anti-VP21/VP22a antibody using a mannose-incubated 2N4R GSK-3β p-tau. Box plots are representative of ±s.e.m. (n = 12) depicting median and interquartile range, with whiskers denoting variability according to Tukey’s method. Statistical significance was calculated by two-tailed Mann−Whitney test (a) and one-way ANOVA using Tukey’s multiple comparisons test (b−f). Source data

Fig. 3. HSV-1 capsids bind to intraneuronal tau and stimulate aggregation.

Co-incubated isolated HSV-1 capsids and 2N4R GSK-3β p-tau were analyzed by TEM. a,b, Co-incubated HSV-1 capsid (arrows) and 2N4R GSK-3β p-tau displayed amorphous tau aggregates (triangles) and fibrillar tau structures (*) emanating from viral capsid surfaces. c, 3D ReNcell VM cultures infected with HSV-1 were sliced and probed with anti-p-tau-Au nanoparticles. d, Inside the cell nucleus, HSV-1 capsids were identified with anti-p-tau-Au binding on their surface (arrows). Micrographs are representative of data from multiple discrete imaging fields (n = 3). Source data

Fig. 4. HSV-1 induces intraneuronal aggregation of p-tau and neuritic dystrophy.

3D ReNcell VM cultures were infected with HSV-1 to characterize tau’s change in cellular distribution and levels in response to viral infection. As shown in a, 3.5−4-week-old 3D ReNcell VM cultures were infected with a replication-deficient HSV-1 (deletion of UL28 protein) for 24 hours, immunoprobed with anti-p-tau (PHF1) and anti-HSV1 labeled with a fluorescent secondary antibody and analyzed for neurons (GFP), HSV-1 (568) and p-tau (647) fluorescence by confocal microscopy. b,c, Fluorescent image captures from 384 wells over four experiments were compared by GA3 software in Nikon Elements for p-tau fluorescent neurites (triangles) (***P = 0.0002) (b) and cell bodies (arrows) (*P = 0.0022) (c) between uninfected and infected wells. d, Confocal images showing 3D ReNcell VM cultures that were infected with HSV-1 for 24 hours, immunoprobed with anti-p-tau (PHF1) labeled with a fluorescent secondary antibody and immunostained with thioflavin S. e, Pearson’s correlation coefficient (error bars represent 95% confidence interval) was derived from analyzing uninfected and infected wells for co-localization of thioflavin S (green), HSV-1 (RFP) and p-tau (647) fluorescence. f−k, 3.5−4-week-old 3D ReNcell VM cultures were infected with serial dilutions of HSV-1 for 24 hours, and cell lysates were separated into soluble and insoluble tau fractions for analysis by MSD Multi-Spot Phospho (Thr 231)/Total Tau Assay. f, Comparison of cellular soluble p-tau at different viral loads (F_{3, 96} = 129.6, **P = 0.0027, ****P < 0.0001). g, Comparison of cellular insoluble p-tau at different viral loads. h, Comparison of the ratio between cellular insoluble and soluble p-tau at different viral loads (F_3,96 = 26.54, ****P < 0.0001). i, Comparison of cellular soluble total tau at different viral loads (F_{3, 96} = 9.251, **P = 0.0040, ***P = 0.0001, ****P < 0.0001). j, Comparison of cellular insoluble total tau at different viral loads (F_{3, 89} = 6.810, **P = 0.0049, ***P = 0.0002). k, Comparison of the ratio between cellular insoluble and soluble total tau at different viral loads (F_{3, 89} = 7.469, **P = 0.0057, ****P < 0.0001). Box plots are representative of ±s.e.m. ((f−h, n = 25), (i−k, n = 19)) depicting median and interquartile range, with whiskers denoting variability according to Tukey’s method. Statistical mean comparisons were calculated by two-tailed unpaired t-tests (b,c) and one-way ANOVA using Dunnett’s multiple comparisons test (f−k). Source data

Fig. 5. HSV-1 promotes the release of p-tau from infected neurons and the accumulation of p-tau in uninfected neurons adjacent to infection.

ReNcell VM cultures were infected with HSV-1 for 24 hours to characterize p-tau’s extracellular release and changes in proximity to viral infection. a−d, 3.5−4-week-old 3D ReNcell VM cultures were infected with serial dilutions of HSV-1 for 24 hours, and cell media and cell lysates were analyzed by MSD Multi-Spot Phospho (Thr 231)/Total Tau Assay for soluble p-tau and total tau. a, Comparison of cell media soluble p-tau at different viral loads (F_{3, 54} = 7.521 ((uninfected versus 2.55, ***P = 0.0005), (uninfected versus 5.10, ***P = 0.0004)). b, Comparison of cell media soluble total tau at different viral loads (F_{3, 55} = 10.49, **P = 0.0011). c, Comparison of the ratio between cell media p-tau and total tau at different viral loads (F_{3, 54} = 19.94, ****P < 0.0001). d, Comparison of the ratio between cell lysate soluble p-tau and soluble total tau at different viral loads (F_{3, 96} = 202.5, ****P < 0.0001). e, 2D ReNcell VM cultures in microfluidic devices were infected in the left chamber with HSV-1 for 48 hours, immunoprobed with anti-p-tau (PHF1) labeled with a fluorescent secondary antibody and analyzed for neurons (GFP), HSV-1 (RFP) and p-tau (647) fluorescence by confocal microscopy. Fluorescence signals for ReNcell VM, HSV-1 and p-tau were imaged for infected and uninfected conditions. f, GA3 analysis of HSV-1-positive neurons compared intracellular p-tau fluorescence among infected neurons (triangles), uninfected neurons proximal to infected neurons (*) and uninfected neurons not proximal to infected neurons (arrows) (F_{2, 3,288} = 45.35, ****P < 0.0001). g, Individual neuronal p-tau fluorescence intensity was compared to total proximal HSV-1−RFP fluorescence intensity (R² = 0.7912). Box plots are representative of ±s.e.m. ((a−c, n = 13), (d, n = 25)) depicting median and interquartile range, with whiskers denoting variability according to Tukey’s method. Statistical mean comparisons were calculated by one-way ANOVA using Dunnett’s multiple comparisons test (a−d), one-way ANOVA using Tukey’s multiple comparisons test (f) and simple linear regression (g). Source data

Fig. 6. Exogenous p-tau induces phosphorylation for essential antiviral activity.

ReNcell VM cultures were infected with HSV-1 to determine the degree of immune response. a, One-week-old 2D ReNcell VM uninfected and infected cultures were immunoprobed with L-azidohomoalanine (AHA) and anti-p-tau (PHF1) and then labeled with fluorescent secondary antibodies and analyzed for neurons (GFP), p-tau (594) and AHA (647) fluorescence by confocal microscopy. b, Distribution plot of individual AHA and p-tau pixel intensities after infection. c, Pearson’s correlation coefficient (error bars represent 95% confidence interval) of images of internalized AHA and PHF1 in ReNcell VM cells. d, Fluorescent image captures from nine wells over three experiments were compared by GA3 software in Nikon Elements for p-tau fluorescence (****P < 0.0001) between azide-positive and azide-negative cells. e, One-week-old 2D ReNcell VM cultures were preincubated with dilutions of a GSK-3β inhibitor (scale of 50 nM to 25 nM) for 24 hours followed by a 24-hour HSV-1 infection and imaged by confocal microscopy for red fluorescence. Whole-well images were analyzed using Nikon Elements GA3 to compare the number of HSV-1 plaques between conditions (F_{3, 53} = 2.806, *P = 0.0268). f,h, Conditioned media collected from 1-week-old ReNcell VM uninfected (f) and 1-hour HSV-1 infected (h) (F_{2, 2} = 23.85, P** = 0.0025) cultures pretreated with 2N4R GSK-3β p-tau (1.25 µg ml⁻¹) were run on a proinflammatory cytokine MSD. g, One-week-old 2D ReNcell VM cultures were preincubated with 2N4R GSK-3β p-tau (1.25 µg ml⁻¹) and anti-IFNγ antibody (10 µg ml⁻¹) for 24 hours followed by a 24-hour HSV-1 infection and imaged by confocal microscopy for red fluorescence. Whole-well images were analyzed using Nikon Elements GA3 to compare the number of HSV-1 plaques between conditions (F_{3, 56}= 4.863, *P = 0.0386). Box plots are representative of ±s.e.m. ((e, n = 11), (f, n = 6), (g, n = 15), (h, n = 3)) depicting median and interquartile range, with whiskers denoting variability according to Tukey’s method. Statistical mean comparisons were calculated by two-tailed unpaired Welch’s t-test (d), one-way ANOVA using Dunnett’s multiple comparisons test (e), two-way ANOVA using Tukey’s multiple comparisons test (f), one-way ANOVA using Tukey’s multiple comparisons test (g) and two-tailed unpaired t-test (h). Source data

Fig. 7. p-tau and HSV-1 co-localize inside microglia.

Mature microglia were added to 3D ReNcell VM cultures prior to infection with HSV-1 to analyze the interaction among p-tau, HSV-1 and microglia. Uninfected and infected microglia−ReNcell VM cultures were immunoprobed with anti-p-tau (PHF1) and anti-IBA1 labeled with fluorescent secondary antibodies and analyzed for microglia (405), neurons (GFP), HSV-1 (RFP) and p-tau (647) fluorescence by confocal microscopy. Co-localization of HSV-1 (red) and p-tau (purple) inside of microglia (blue) was observed (arrows). Panels are representative of multiple image fields (n = 5).

Extended Data Fig. 1. GSK-3ꞵ treated 2N4R tau, but not non-phosphorylated tau, reduces HSV1 plaque formation in human cell culture model.

2D ReNcell VM cultures were pre-incubated with dilutions of synthetic 2N3R tau, 2N4R tau, or 2N4R GSK-3β phosphorylated tau (p-tau) followed by infection with HSV1 to assay tau’s antimicrobial protective properties. After HSV1 infection, whole wells were imaged by confocal microscopy and analyzed for red fluorescence. (a) Whole well images comparing HSV1 plaque formation in the absence or presence of 2N4R GSK-3β p-tau (1.25 µg/mL). (b-g) Whole well images from the 2N3R or 2N4R tau pre-treated conditions (scale of 2.5 to 0.625 µg/mL) were analyzed using Nikon Elements GA3 Analysis to compare (b, e) the number of HSV1 plaques, (c, f) the number of HSV1 single cell infections, and (d, g) the ratio of plaques to single cell infections. Box plots are representative of ±SEM ([B-D, n=11], [E-G, n=12]) depicting median and IQR with whiskers denoting variability according to Tukey’s method. Statistical mean comparisons were calculated by one-way ANOVA using Dunnett’s multiple comparison test (b-f). Source data

Extended Data Fig. 2. Anti-capsid antibodies and tau isoforms have a higher binding affinity to HSV1 capsid than whole virion.

Anti-capsid antibodies and synthetic tau isoforms were incubated with heat-immobilized HSV1 capsid or whole virion in an indirect ELISA to measure tau-virus binding affinity differences between isolated viral capsids and whole virions with intact envelopes. (a) Anti-capsid antibodies VP21/22a and ICP5 viral binding were compared between isolated HSV1 capsids and whole virion (F_{(1, 20)}=14.29, *P=0.0136, ****P<0.0001). (b) Synthetic tau isoforms 2N3R, 50/50 mix of 2N3R/2N4R, 2N4R, and synthetic 2N4R GSK-3β p-tau binding affinities were compared between isolated HSV1 capsids and whole virion normalized against VP21/22a antibody binding (F_{(3, 88)}=4.957, ****P<0.0001). Box plots are representative of ±SEM ([A, n=6], [B, n=12]) depicting median and IQR with whiskers denoting variability according to Tukey’s method. Statistical significance was calculated by two-way ANOVA using Šídák’s multiple comparisons test (a, b). Source data

Extended Data Fig. 3. 2N3R tau isoform has the lowest binding affinity to HSV1 compared to other tau isoforms.

Dilutions of synthetic tau isoforms were incubated with heat-immobilized HSV1 capsid or whole virion in an indirect ELISA to measure tau-virus binding affinity. (a) Dilutions of 2N3R tau (scale of 5 to 0.3125 μg/mL) are incubated with immobilized HSV1 capsids (F_{(4, 55)}=33.42, *P=0.0232, ****P<0.0001). (b) Dilutions of a 50/50 mixture of 2N3R/2N4R tau (scale of 5 to 0.3125 μg/mL) are incubated with immobilized HSV1 capsids (F_{(4, 55)}=30.21, **P=0.0071, ****P<0.0001). (c) Dilutions of 2N4R tau (scale of 5 to 0.3125 μg/mL) are incubated with immobilized HSV1 capsids (F_{(4, 55)}=11.50, **P=0.0011, ****P<0.0001). (d) 2N4R GSK-3β p-tau binding to HSV1 was compared to 2N3R, 50/50 mix of 2N3R/2N4R, and 2N4R tau (F_{(3, 44)}=6.913, *P=0.0211, [2N4R tau vs. 2N3R tau, **P=0.0025], [2N4R GSK-3β p-tau vs. 2N3R tau, **P=0.0012]) (e) 2N4R GSK-3β p-tau and scrambled-sequence LL-37 are incubated with immobilized HSV1 capsids (F_{(3, 16)}=2.299, *P=0.0422, ***P=0.0002, ****P<0.0001). Box plots are representative of ±SEM (n=12) depicting median and IQR with whiskers denoting variability according to Tukey’s method. Statistical significance was calculated by one-way ANOVA using Tukey’s multiple comparisons test (a-d) and two-way ANOVA using Šídák’s multiple comparisons test (e). Source data

Extended Data Fig. 4. Site-specific anti-capsid antibodies less effectively inhibit tau binding to HSV1.

Synthetic 2N3R tau and 2N4R GSK-3β p-tau isoforms were incubated with heat-immobilized HSV1 capsid after pre-incubation with anti-capsid protein antibodies in an indirect ELISA to examine inhibition of tau-capsid binding. (a) 2N3R tau binding to HSV1 capsid was pretreated with anti-ICP5 antibody. (b) 2N4R GSK-3β p-tau binding to HSV1 capsid was pretreated with anti-RS1-ICP4 antibody (F_{(8, 32)}=13.59, [0 vs. 1.0 μg/mL, **P=0.0064], [0 vs. 5.0 μg/mL, **P=0.0056]). (c) 2N4R GSK-3β p-tau binding to HSV1 capsid was pretreated with anti-pUL25 antibody. Box plots are representative of ±SEM (n=12) depicting median and IQR with whiskers denoting variability according to Tukey’s method. Statistical significance was calculated by two-tailed Mann-Whitney test (a) and one-way ANOVA using Dunnett’s multiple comparison test (b, c). Source data

Extended Data Fig. 5. Cellular total tau fluorescence is increased in HSV1 infected neurons but not in adjacent uninfected neurons.

ReNcell VM cultures were infected with HSV1 for 24 hours to characterize total tau’s distribution changes from infection and in proximity to viral infection. (a) 2D ReNcell VM cultures in microfluidic devices were infected in the left chamber with HSV1 for 48 hours, immunoprobed with anti-total tau labeled with a fluorescent secondary antibody, and analyzed for total tau (405), neurons (GFP), and HSV1 (RFP) fluorescence by confocal microscopy. Fluorescence signals were compared by General Analysis 3 (GA3). (b) GA3 of HSV1-positive neurons compared intracellular p-tau fluorescence between infected neurons, uninfected neurons proximal to infected neurons, and uninfected neurons not proximal to infected neurons (F_{(2, 3321)}=166.3, ****P<0.0001). (c) Individual neuronal total tau fluorescent intensity was compared to total proximal HSV1-RFP fluorescent intensity (R²=1.350e-006). (d) Fluorescent image captures from 384 wells over 4 experiments were compared for total tau fluorescence in cell bodies (***P=0.0003) between uninfected and infected wells. Statistical mean comparisons were calculated by one-way ANOVA (b), simple linear regression (c), and two-tailed paired t-test (d). Images are representative of experiments repeated in triplicate. Source data

Extended Data Fig. 6. The phosphorylation of tau from an HSV1 infection occurs at multiple phosphorylation sites.

3D ReNcell VM cultures were infected with HSV1 to characterize tau’s change in cellular distribution and levels to viral infection. 96 well plates were immunoprobed with anti-p-tau antibodies AT180 and PHF-1 labeled with a fluorescent secondary antibody. Images were collected via confocal microscopy and immunofluorescent representative images show neurons (GFP), HSV1 (RFP), and p-tau (647). Images are representative of experiments repeated in triplicate. Source data

Extended Data Fig. 7. Cell death is not increased by HSV1 24 hours post infection.

Conditioned media from 24 hour HSV1 infected 3D ReNcell VM cultures were analyzed via a cytotoxicity assay to characterize culture cell death. Box plots are representative of ±SEM (n=30) depicting median and IQR with whiskers denoting variability according to Tukey’s method. Statistical significance was calculated by one-way ANOVA using Dunnett’s multiple comparison test. Source data

Extended Data Fig. 8. HSV1 infectivity and plaque sizes increase following progressive GSK-3β inhibition.

1 week old 2D ReNcell VM cultures were pre-incubated with a GSK-3β inhibitor for 24 hours followed by a 24 hour HSV1 infection to assess reduced kinase activity on viral replication. Post-infection whole wells were imaged by confocal microscopy and analyzed for red fluorescence. Images were analyzed using Nikon Elements GA3 Analysis to compare (a) the number of HSV1 single cell infections (F_{(3, 53)}=5.204, **P=0.0012), (b) the ratio of plaques to single cell infections (F_{(3, 53)}=4.173, **P=0.0046), and (c) the size of individual HSV1 plaques by area (****P<0.0001). Bars represent the distribution of replicate wells (n=11), with whiskers denoting variability according to Tukey’s method. Statistical mean comparisons were calculated by one-way ANOVA using Dunnett’s multiple comparisons test (a, b) and two-tailed Kolmogorov-Smirnov test (c). Source data

Extended Data Fig. 9. IFNγ concentration is effectively reduced by neutralizing anti-IFNγ antibodies, but is not altered by p-tau pre-incubation.

2D ReNcell VM cultures were pre-incubated with 2N4R GSK-3β p-tau or an IFNγ antibody followed by a short-term HSV1 infection to observe effects on extracellular IFNγ levels. (a) IFNγ concentrations in conditioned media were measured following a 24 hour pre-incubation of IFNγ antibody and again following HSV1 infection. (F_{(2, 2)}=7.093, P***<0.0010). (b) IFNγ concentrations in conditioned media were measured following a 24 h pre-incubation of 2N4R GSK-3β p-tau and again following HSV1 infection. Box plots are representative of ±SEM ([A, n=3], [B, n=4]) depicting median and IQR with whiskers denoting variability according to Tukey’s method. Statistical mean comparison was calculated by two-tailed unpaired t-test (a) and two-tailed Welch’s t-test (b). Source data

Extended Data Fig. 10. p-tau localizes in dystrophic neurites with HSV1 and around infected nuclei during viral replication.

3D ReNcell VM cultures were infected with HSV1 to characterize tau’s co-localization with HSV1. (a, b) 3D ReNcell VM cultures were infected with HSV1 for 24 h, immunoprobed with anti-p-tau (PHF-1) labeled with a fluorescent secondary antibody, and analyzed for neurons (GFP), HSV1 (RFP), and p-tau (647) fluorescence by confocal microscopy. (a) Dystrophic neurites, identified by punctate linear formations originating from a cell soma, contain p-tau and HSV1 fluorescence. (b) Increased punctate p-tau fluorescence observed in neuron soma adjacent to HSV1-positive cells and in cells with nuclear replication compartments (arrows). Panels are representative of data from multiple discrete imaging fields across experiments repeated in triplicate.