Tau expression is increased in the cortex of Fmr1 KO mice

We first detected and found that both Tau protein (Cor, t(16) = 2.565, p = 0.0207, unpaired t test, Fig. 1A) and Mapt mRNA (Cor, t(6) = 4.096, p = 0.0064, unpaired t test, Fig. 1B) levels were significantly increased in the cortical but not in the hippocampal regions of Fmr1 KO mice when compared to WT controls. These results suggest that Fmr1 regulates the expression of Mapt and thus tau protein levels. On the other hand, neither FMRP protein nor Fmr1 mRNA levels were altered in the cortical and hippocampal regions of Mapt KO mice (Fig. 1C, D).

Fig. 1

Tau and Mapt mRNA levels are increased in the cortex of Fmr1 KO mice. A Tau protein in the cortex and hippocampus of WT and Fmr1 KO mice (FVB background, 1.5-month-old) was detected by western blotting; and protein levels were quantified by densitometry for comparison after normalizing to those of GAPDH. WT: n = 6; Fmr1 KO: n = 6. B Mapt mRNA levels in the cortex and hippocampus of WT and Fmr1 KO mice (FVB background, 1.5-month-old) were detected by qRT-PCR and compared after normalizing to those of β-actin. WT: n = 4; Fmr1 KO: n = 4. C FMRP protein was detected by western blotting in the cortex and hippocampus of WT and Mapt KO mice (C57BL/6/SV129 background, 4-month-old); and protein levels of three major FMRP isoforms (1, 2, and 3) were individually quantified by densitometry for comparison after normalizing to those of β-actin. WT: n = 3; Mapt KO: n = 3. D Fmr1 mRNA levels in the cortex and hippocampus of WT and Mapt KO mice (C57BL/6/SV129 background, 4-month-old) were detected by qRT-PCR and compared after normalizing to those of β-actin. WT: n = 3; Mapt KO: n = 3. Unpaired t test. ns: not significant; *p < 0.05, **p < 0.01

Genetically reducing Tau attenuates autism-like behaviors in Fmr1 KO mice

We crossed Fmr1± female mice with Mapt± male mice to obtain WT, Mapt±, Fmr1−/y, and Fmr1−/y;Mapt± male offspring. These mice were studied for their behaviors at 1 month of age. In the open field test, we found that time spent in the center, entries into the center and total travel distance were not different among the four groups of mice (Fig. 2A–C). In the nest building test, Fmr1−/y mice had poorer nesting scores than those of WT mice, whereas Fmr1−/y;Mapt±mice had better nesting scores than Fmr1−/y mice (F(3,37) = 3.559, p = 0.0112 for Fmr1−/y versus WT, p = 0.0470 for Fmr1−/y;Mapt± versus Fmr1−/y, one-way ANOVA followed by Tukey’s post hoc test, Fig. 2D). In the self-grooming test, the grooming time (F(3,37) = 21.46, p < 0.0001 for Fmr1−/y versus WT, p = 0.0002 for Fmr1−/y;Mapt± versus Fmr1−/y, one-way ANOVA followed by Tukey’s post hoc test, Fig. 2E) and bout numbers (F(3,37) = 3.455, p = 0.0211 for Fmr1−/y versus WT, one-way ANOVA followed by Tukey’s post hoc test, Fig. 2F) of Fmr1−/y mice were significantly increased, whereas Tau reduction reversed the increased grooming time in Fmr1−/y mice. In the three-chamber social interaction test, none of the four groups of mice showed preference for each chamber during the habituation phase (Fig. 2G). During the social preference testing phase, although all four groups of mice interacted more with a stranger mouse (Stranger 1) than an empty cage, Fmr1−/y mice had less preference ratios to Stranger 1 than WT and Fmr1−/y;Mapt±mice (F(3,74) = 13.29, p < 0.0001 for Empty versus Stranger 1 in all four groups, p = 0.0012 for Stranger 1 in Fmr1−/y versus Stranger 1 in WT, p = 0.0497 for Stranger 1 in Fmr1−/y;Mapt± versus Stranger 1 in Fmr1−/y, two-way ANOVA followed by Bonferroni’s post hoc test, Fig. 2H). During the social novelty testing phase, Fmr1−/y mice showed no preference for a novel stranger mouse (Stranger 2) when compared to the familiar Stranger 1. While both WT and Fmr1−/y;Mapt±mice not only showed preference for Stranger 2 when compared to Stranger 1, but also had more preference ratios to Stranger 2 than Fmr1−/y mice (F(3,74) = 8.640, p < 0.0001 for Stranger 2 versus Stranger 1 in WT, Mapt±, and Fmr1−/y;Mapt± groups, p = 0.0238 for Stranger 2 in Fmr1−/y versus Stranger 2 in WT, p = 0.0161 for Stranger 2 in Fmr1−/y;Mapt± versus Stranger 2 in Fmr1−/y, two-way ANOVA followed by Bonferroni’s post hoc test, Fig. 2I). Together, these results indicate that Tau reduction can ameliorate social defects and stereotyped and repetitive behavior in Fmr1 KO mice.

Fig. 2

Genetically reducing Tau prevents autism-like behaviors in Fmr1−/y mice. A–I Fmr1 KO mice and Mapt KO mice were crossed and the offspring (1-month-old, FVB;C57BL/6/SV129 mixed background) were studied for their behaviors. In the open field test, time spent in the center (A), center entry numbers (B), and total travel distance (C) were analyzed. In the nest building test, nesting scores were analyzed (D). In the self-grooming test, grooming time (E) and bout numbers (F) were analyzed. In the three-chamber social interaction test, time spent in each chamber (G), time spent interacting with a Stranger 1 mouse and the empty cage (H), and time spent interacting with the Stranger 1 mouse and a Stranger 2 mouse (I) were analyzed. WT: n = 11; Mapt±: n = 11; Fmr1−/y: n = 9; Fmr1−/y;Mapt±: n = 10. One-way ANOVA followed by Tukey’s post hoc test for (A–F). Two-way ANOVA followed by Bonferroni’s post hoc test for G-I. ns: not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Tau reduction reverses spine abnormality in Fmr1 KO mice

In the cortex of FXS patients and Fmr1 KO mice, spine density is found to be increased and accompanied by decreased mature spines and increased immature spines [23,24,25]; and this may underlie dysregulated neuronal functions and abnormal behaviors in FXS [26]. We conducted Golgi staining and confirmed significantly increased spine density (F(3,58) = 35.84, p < 0.0001 for Fmr1−/y versus WT, p = 0.0006 for Fmr1−/y;Mapt± versus Fmr1−/y, one-way ANOVA followed by Tukey’s post hoc test, Fig. 3A, B) and numbers of immature spines, as well as decreased mature spines (F(2,9) = 155.3, p < 0.0001 for all comparisons, one-way ANOVA followed by Tukey’s post hoc test, Fig. 3A, C) in cortical neurons of Fmr1−/y mice. Importantly, these alterations were reversed in Fmr1−/y;Mapt± mice (Fig. 3A–C).

Fig. 3

Genetically reducing Tau reverses spine abnormality in Fmr1−/y mice. A Representative projection images of dendritic spines from cortical neurons from WT, Fmr1−/y, and Fmr1−/y;Mapt± mice (2.5-month-old, FVB;C57BL/6/SV129 mixed background). Mature and immature dendritic spines were indicated by red and green arrowheads, respectively. Scale bars: 5 μm. B, C Comparisons of spine density (B) and ratios of mature and immature spines (C) of cortical neurons. One-way ANOVA followed by Tukey’s post hoc test. WT: n = 4; Mapt±: n = 4; Fmr1−/y: n = 4; Fmr1−/y;Mapt±: n = 4. ***p < 0.001, ****p < 0.0001

Tau reduction reverses altered periodic activity in Fmr1 KO mice

By performing RNA sequencing using a mixture of hippocampal and cortical tissues, we identified 96 upregulated and 65 downregulated DEGs in Mapt± versus WT, 589 upregulated and 989 downregulated DEGs in Fmr1−/y versus WT, 1040 upregulated and 1259 downregulated DEGs in Fmr1−/y;Mapt± versus Mapt±, and 4 upregulated and 5 downregulated DEGs in Fmr1−/y;Mapt± versus Fmr1−/y (Fig. 4A).

Fig. 4

Genetically reducing Tau rescues circadian rhythm defect in Fmr1−/y mice. A RNA sequencing identified DEGs between mice with different genotypes (FVB;C57BL/6/SV129 mixed background, 2.5 month-old). B DEGS shared by Fmr1−/y versus WT and Fmr1−/y;Mapt± versus Fmr1−/y groups and their information. C Per1 mRNA levels in the brain of mice with different genotypes were determined by qRT-PCR for comparison. One-way ANOVA followed by Tukey’s post hoc test. WT: n = 4; Mapt±: n = 4; Fmr1−/y: n = 4; Fmr1−/y;Mapt±: n = 4. D In the autonomous wheel-running test, the activities of different groups of mice were recorded for 5 consecutive days in a constant darkness environment and compared. Two-way repeated-measures ANOVA. WT: n = 3; Mapt±: n = 3; Fmr1−/y: n = 3; Fmr1−/y;Mapt±: n = 3. ns: not significant, *p < 0.05, ***p < 0.001, ****p < 0.0001

With an assumption that genes responsible for Tau reduction-exerted protection would be those whose expressions are altered in Fmr1−/y versus WT and reversed in Fmr1−/y;Mapt± versus Fmr1−/y, we first determined DEGs that were shared by Fmr1−/y versus WT and Fmr1−/y;Mapt± versus Fmr1−/y. However, we only found 4 DEGs overlapped in the two groups as Per1, Gm52433, Shc3, and Mbp (Fig. 4B). Among the 4 DEGs, only the expressions of Per1 and Gm52433 showed opposite direction change between Fmr1−/y versus WT and Fmr1−/y;Mapt± versus Fmr1−/y, whereas the expression change directions were the same for Shc3 and Mbp in the two groups.

Per1 is an important circadian rhythm gene [27]. Since it is reported that FXS patients and animal models also exhibit abnormal circadian behavioral rhythm [28, 29], we further studied Per1 mRNA expression by qRT-PCR. Consistent with RNA sequencing data, we found that Per1 mRNA expression was decreased in Fmr1−/y mice and partially reversed by Mapt deficiency (F(3,12) = 63.12, p < 0.0001 for Fmr1−/y versus WT, p = 0.0308 for Fmr1−/y;Mapt± versus Fmr1−/y, one-way ANOVA followed by Tukey’s post hoc test, Fig. 4C). In the autonomous wheel-running test performed under constant darkness, we found that Fmr1−/y mice had more periodic activity than WT mice, especially at the time around the night-day transition (F(3,192) = 11.37, p = 0.0001 for Fmr1−/y versus WT, two-way repeated-measures ANOVA, Fig. 4D); this is implies a circadian rhythm defect in Fmr1−/y mice. While the periodic activity of Fmr1−/y;Mapt± mice were less than Fmr1−/y mice and comparable to WT mice (F(3,192) = 11.37, p = 0.0270 for Fmr1−/y;Mapt± versus Fmr1−/y, two-way repeated-measures ANOVA, Fig. 4D).

Interestingly, we noticed that Mapt± mice also had decreased Per1 mRNA expression (F(3,12) = 63.12, p = 0.0176 for Mapt± versus WT, one-way ANOVA followed by Tukey’s post hoc test, Fig. 4C) and increased periodic activity in the autonomous wheel-running test when compared to WT mice (F(3,192) = 11.37, p < 0.0001 for Mapt± versus WT, two-way repeated-measures ANOVA, Fig. 4D). Therefore, the rescuing effect of Tau reduction on autism-like phenotypes in Fmr1 KO mice is unlikely through reversing Per1 expression and periodic activity defect.

Tau reduction reverses impaired P38/MAPK signaling in Fmr1−/y mice

Since Tau deficiency minimally affected gene expression in Fmr1 KO mice, we wondered whether gene processes/pathways affected by Tau deficiency could balance those affected by Fmr1 deficiency and thereby providing the protection. By comparing the top 20 GO processes enriched with DEGs found in Fmr1−/y versus WT and those found in Mapt± versus WT (Fig. 5A, B), we found two overlapped GO processes: “response to light stimulus” and “inactivation of MAPK activity”, of which the former is related to circadian rhythm.

Fig. 5

Genetically reducing Tau rescues impaired P38/MAPK signaling in Fmr1−/y mice. A, B GO process enrichment analysis based on DEGs identified in Fmr1−/y versus WT (A) and in Mapt± versus WT (B) groups. Red ones indicate GO processes shared in (A) and (B). C–E Equal amounts of protein lysates from mouse cortical tissues were analyzed by western blotting for indicated proteins (C). Levels of phosphorylated p38 (D) and ERK (E) in different groups of mice (FVB;C57BL/6/SV129 mixed background, 2.5-month-old) were normalized to respective total protein levels for comparison. One-way ANOVA with Tukey’s post hoc test. WT: n = 6; Mapt±: n = 6; Fmr1−/y: n = 6; Fmr1−/y;Mapt±: n = 6. ns: not significant; *p < 0.05, **p < 0.01, ***p < 0.001

We then detected levels of MAPK pathway-related proteins in cortical tissues and found that levels of phosphorylated P38 were significantly decreased in Fmr1−/y mice when compared to WT mice; and this decrease was reversed in Fmr1−/y;Mapt± mice (F(3,20) = 4.689, p = 0.0174 for Fmr1−/y versus WT, p = 0.0094 for Fmr1−/y;Mapt± versus Fmr1−/y, one-way ANOVA followed by Tukey’s post hoc test, Fig. 5C, D). While P38 phosphorylation was not different between Mapt± and WT mice. The ERK signaling is another important MAPK pathway. However, although some studies reported increased ERK phosphorylation in Fmr1 KO mice [30,31,32], other work suggested no change or even decrease of ERK phosphorylation [33,34,35]. Herein, we found that ERK phosphorylation was not altered in Fmr1−/y mice when compared to WT mice but was significantly increased when Tau was genetically reduced (F(3,20) = 12.26, p = 0.0194 for Mapt± versus WT, p = 0.0002 for Fmr1−/y;Mapt± versus Fmr1−/y, one-way ANOVA followed by Tukey’s post hoc test, Fig. 5C, E). Overall, these results suggest that Tau reduction promotes MAPK signaling.

A previous study showed that Tau interacted with PTEN and Tau reduction prevented over-activation of the mTOR/PI3K/Akt signaling [8]. Although some previous studies found that the mTOR/PI3K/Akt signaling was over-activated in Fmr1 KO mice [30, 36, 37], inconsistent results were also reported [32, 38]. Herein, we observed no significant phosphorylation changes of S6, Akt, and mTOR, all of which are indicative of the mTOR/PI3K/Akt signaling activity, in Fmr1 KO mice when compared to WT mice (Additional file 1: Fig. S1A–D). Nor did we notice that Tau reduction affected the mTOR/PI3K/Akt signaling.

Tau-targeting ASO rescues autism-like phenotypes in Fmr1 KO mice

To further determine whether targeting Tau has therapeutic potential for FXS, we used osmotic pumps to release Tau-targeting ASO (ASO-Tau) and control ASO (ASO-NC) into the lateral ventricles of 1-month-old Fmr1−/y mice with an FVB background for two weeks. After another two weeks, mice were subjected to various behavioral tests (Fig. 6A). We found that downregulation of Tau by ASO-Tau (t(10) = 3.896, p = 0.0030, unpaired t test, Fig. 6L, M) had no effects on time spent in the center, entries into the center, and total travel distance of mice in the open field test (Fig. 6B–D). In the self-grooming test, ASO-Tau treatment significantly reduced self-grooming time (t(19) = 3.133, p = 0.0055, unpaired t test, Fig. 6E), though not bout numbers (Fig. 6F) of Fmr1−/y mice. In the nest building test, ASO-Tau mice achieved higher scores than ASO-NC mice (t(19) = 3.444, p = 0.0027, unpaired t test, Fig. 6G). In the three-chamber social interaction test, neither ASO-Tau nor ASO-NC mice showed preference for each chamber during the habituation phase (Fig. 6H). During the social preference testing phase, although both mice interacted more with a Stranger 1 mouse than an empty cage, ASO-Tau mice had more preference ratios to Stranger 1 than ASO-NC mice (F(1,38) = 14.55, p < 0.0001 for Stranger 1 versus Empty in all groups, p = 0.0143 for Stranger 1 in ASO-Tau versus Stranger 1 in ASO-NC, two-way ANOVA followed by Bonferroni’s post hoc test, Fig. 6I). During the social novelty testing phase, ASO-NC mice showed no preference for a novel Stranger 2 mouse when compared to the familiar Stranger 1. While ASO-Tau mice not only showed preference for Stranger 2 when compared to Stranger 1, but also had more preference ratios to Stranger 2 than ASO-NC mice (F(1,38) = 16.21, p < 0.0001 for Stranger 2 versus Stranger 1 in ASO-Tau group, p = 0.0103 for Stranger 2 in ASO-Tau versus Stranger 2 in ASO-NC, two-way ANOVA followed by Bonferroni’s post hoc test, Fig. 6J). In the autonomous wheel-running test, ASO-Tau mice showed decreased periodic activity when compared to ASO-NC mice (F(1,240) = 20.55, p < 0.0001, two-way repeated-measures ANOVA, Fig. 6K). Together, these results indicates that ASO-Tau treatment attenuates autism-like behaviors in Fmr1 KO mice.

Fig. 6

Tau-targeting ASO rescues autism-like phenotypes in Fmr1−/y mice. A ASO treatment paradigm. Fmr1−/y mice (FVB background, 1-month-old) were treated with 25 μg/d ASO-Tau or ASO-NC via intracerebroventricular infusion for 2 weeks and the catheters were then removed. After another 2 weeks, behavioral tests were performed. B–D In the open field test, time spent in the center (B), center entry numbers (C), and total travel distance (D) were analyzed. E, F In the self-grooming test, grooming time (E) and bout numbers (F) were analyzed. G In the nest building test, nesting scores of mice were analyzed. H–J In the three-chamber social interaction test, time spent in each chamber (H), time spent interacting with a Stranger 1 mouse and the empty cage (I), and time spent interacting with the Stranger 1 mouse and a Stranger 2 mouse (J) were analyzed. K In the autonomous wheel-running test, mouse activity was recorded for 5 consecutive days in a constant darkness environment and compared. For behavioral tests in B-J, ASO-NC: n = 10; ASO-Tau: n = 11. For tests in K, ASO-NC: n = 7; ASO-Tau: n = 7. L–O Equal amounts of protein lysates from 2.5-month-old mouse cortical tissues were analyzed by western blotting for indicated proteins (L). Total Tau levels were normalized to those of GAPDH for comparison (M); and levels of phosphorylated P38 (N) and ERK (O) were normalized to respective total protein levels for comparison. ASO-NC: n = 6; ASO-Tau: n = 6. Unpaired t test for (B–G, M–O). Two-way ANOVA followed by Bonferroni’s post hoc test for H-J. Two-way repeated-measures ANOVA for K. ns: not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Similar to above findings in Fmr1 KO mice with an FVB;C57BL/6/SV129 mixed background, phosphorylation levels of P38 were decreased (t(8) = 1.146, p = 0.0360, unpaired t test, Additional file 1: Fig. S2A, F), whereas phosphorylation levels of S6, Akt, and mTOR were unaltered (Additional file 1: Fig. S2A–D) in the cortex of Fmr1 KO mice with an FVB background when compared to those of WT controls. However, ERK phosphorylation was increased in the cortex of Fmr1 KO mice (t(8) = 2.779, p = 0.0240, unpaired t test, Additional file 1: Fig. S2A, E) with an FVB background but not those with an FVB;C57BL/6/SV129 mixed background. Moreover, we found that ASO-Tau treatment promoted both P38 phosphorylation (t(10) = 6.098, p = 0.0001, unpaired t test, Fig. 6L, N) and ERK phosphorylation (t(10) = 4.227, p = 0.0018, unpaired t test, Fig. 6L, O) but not the mTOR/PI3K/Akt signaling activity (Additional file 1: Fig. S3A–D) in the cortex of Fmr1−/y mice with an FVB background.