A role for glia in the progression of Rett syndrome

Abstract

Rett syndrome (RTT) is an X-chromosome-linked autism spectrum disorder caused by loss of function of the transcription factor methyl CpG-binding protein 2 (MeCP2)¹. Although MeCP2 is expressed in most tissues², loss of MeCP2 results primarily in neurological symptoms^1,3,4. Earlier studies propelled the idea that RTT is due exclusively to loss of MeCP2 function in neurons^2,4-10. While defective neurons clearly underlie the aberrant behaviors, we and others showed recently that the loss of MeCP2 from glia negatively influences neurons in a non-cell autonomous fashion^11-13. Here, we show that in globally MeCP2-deficient mice, re-expression of MeCP2 preferentially in astrocytes significantly improved locomotion and anxiety levels, restored respiratory abnormalities to a normal pattern, and greatly prolonged lifespan compared to globally null mice. Furthermore, restoration of MeCP2 in the mutant astrocytes exerted a non-cell-autonomous positive effect on mutant neurons in vivo, restoring normal dendritic morphology and increasing levels of the excitatory glutamate transporter (VGlut1). Our study shows that glia, like neurons, are integral components of the neuropathology of RTT, and supports targeting glia as a strategy for improving the associated symptoms.

Global re-expression of MeCP2 postnatally in MeCP2-deficient mice allows normal longevity, rescues motor behaviors, and improves overall health¹⁴. Because expression of MeCP2 from the neuronal tau locus in early development prevents appearance of several RTT-like symptoms⁹, neurons are likely crucial components in a rescue. However, prior in vitro studies indicate that astrocytic MeCP2 supports normal neuronal morphology^11,12. Therefore, we asked whether astrocytes might also play a role in rescuing RTT neuropathology in vivo.

To this end, we crossed mice harboring a tamoxifen (TAM)-inducible cre recombinase transgene driven by the human astrocytic glial fibrillary acidic protein (hGFAP) promotor¹⁵ (also see Ref. 16 - 18), with mice containing a cre-excisable transcriptional stop sequence in the endogenous Mecp2 gene (Mecp2^Stop)¹⁴. The progeny that inherited both alleles are referred to as Mecp2^Stop-hGFAPcreT2 mice (Supplemantary Fig. 1a). We determined the efficiency of astrocytic excision in ROSA-reporter¹⁵ and Mecp2^Stop/y-hGFAPcreT2 mice (Supplementary Fig. 1b, c, and d). The percentage of MeCP2⁺ / GFAP⁺ astrocytes was extremely high in caudal brain regions¹⁵, similar to Mecp2^+/y (Fig. 1a and Supplementary Fig. 1e). Re-expression of MeCP2 was not detected in oil-treated Mecp2^Stop/y-hGFAPcreT2 mice (Supplementary Fig. 2a). Importantly, only a very low percentage (< 5%) of excision in neurons was detected by immunolabeling, PCR analysis of the recombined stop sequence, and single cell immunofluorescence intensity measurements (Fig. 1 and Supplementary Figs. 1f and g and 3). This low percentage did not increase with age (Supplementary Figs. 1g and 4), and MeCP2 re-expression was restricted to brain (Supplementary Fig. 5c). Over-expression of MeCP2 in rescued astrocytes was not observed (Supplementary Fig. 2b).

Figure 1.

a, Efficiencies of Mecp2 re-expression. The numbers above the bars indicate total number of cells counted. b, Genomic PCR analysis of non-recombined (Stop; 4.3 kb) and recombined amplicons (1.29 kb) of FACS-sorted NeuN⁺ and NeuN^- cells from the whole brain of a TAM-treated Mecp2^Stop/y-hGFAPcreT2 mouse. Genomic DNA was prepared from 500,000 cells per group. The wild-type (1.25 kb) Mecp2 amplicon is indicated. The β-actin promoter amplicon shows that similar amounts of DNA were present in the reactions. c, Fluorescence-intensity histogram derived from individual hippocampal pyramidal neurons in tissue sections. Cy2 immunofluorescence intensities of nuclear MeCP2 protein are indicated above the line; DAPI fluorescence intensities of the same neurons are indicated below the line. ALU, arbitrary linear units. n = 3 mice per genotype and 100 cells per mouse.

To rule out that the small percentage of neurons, in combination with the low constitutive level of MeCP2 in the stop mice (Supplemental Fig. 5b and c), might mediate any behavioral changes we would measure, we systemically injected young male Mecp2^Stop/y mice with a suboptimal titer of recombinant MeCP2-AAV9 virus¹⁹ or virus lacking MeCP2. This resulted in physiological levels of MeCP2 expression in 2% to 35% of neurons, depending on the brain region (Supplementary Fig. 6a - c). Regardless of genotype, none of the treated mice showed improvement of RTT-like phenotypes compared to the control-AAV9 injected Mecp2^-/y mice (Supplementary Fig. 6d - g). Taken together, the results validate the use of the hGFAPcreT2 system for dissecting astrocytic contributions to RTT.

The average lifespan of oil-treated Mecp2^Stop/y-hGFAPcreT2 and Mecp2^Stop/y mice was only three months¹⁴, which is prolonged compared to Mecp2^-/y mice^3,4 (Supplementary Fig. 5a), and likely due to the small amount of MeCP2 protein expressed from the stop locus (Supplementary Fig. 5b and c). In contrast, nine of 11 TAM-treated Mecp2^Stop/y-hGFAPcreT2 mice were alive at 7.5 months, when seven of the nine were sacrificed for further analysis. The longest living mouse was sacrificed at 15 months. The TAM-treated MeCP2^Stop/y-hGFAPcreT2 mice were also on average 20% larger than oil-treated MeCP2^Stop/y-hGFAPcreT2 mice (Supplementary Fig. 7a). Using a previously described observational scoring system¹⁴, overall health of the TAM-treated male (Supplementary Fig. 7b) and female (Supplementary Fig. 8) mice stabilized, rather than worsened like the oil-treated controls, and TAM treatment of a highly symptomatic MeCP2^Stop/y-hGFAPcreT2 mouse reversed symptoms to nearly hGFAPcreT2 values (Supplementary Fig. 9, Videos S1 and S2).

MeCP2-deficient mice are hypoactive^1,3,4 and show altered measures of anxiety-related behaviors⁶. In the home cage (Fig. 2a and b) and open field (Fig. 2c), oil-treated Mecp2^Stop/y-hGFAPcreT2 mice traveled only ~20% the distance, and did so at ~20% the velocity, of hGFAPcreT2 control mice. TAM-treated Mecp2^Stop/y-hGFAPcreT2 mice, however, improved to ~50% the level of hGFAPcreT2 mice in both measures. Similar improvements were observed in an open field test to measure anxiety. The oil-treated Mecp2^Stop/y-hGFAPcreT2 mice spent only ~20% as much time in the center of the cage as hGFAPcreT2 mice, while TAM-treated Mecp2^Stop/y-hGFAPcreT2 mice again improved to ~50% the level of hGFAPcreT2 mice (Fig. 2d). The ratio of distance traveled in the center square to total distance was the same for all genotypes (data not shown). In the elevated zero and plus mazes, Mecp2^-/y mice consistently show decreased anxiety-related behavior^20,21. The TAM-treated Mecp2^Stop/y-hGFAPcreT2 mice were more anxious than the oil-treated Mecp2^Stop/y-hGFAPcreT2 mice in the elevated zero maze, improving up to ~50% the level of the control hGFAPcreT2 mice (Fig. 2e)

Figure 2.

a, Representative activity in a home-cage-like setting. Duration interval, 5 min. Scale bars indicate 7 inches. b, Locomotor activity histograms in a home-cage-like setting. Duration interval, 10 min. c, Locomotor activity histograms in an open field. Duration, 20 min. d, Time spent in centre of an open field. e, Time spent in open portions of an elevated zero maze. Mice aged 3-4 months. *P < 0.05, **P < 0.01, ***P < 0.001. All error bars indicate s.e.m. The number of mice analysed is indicated above each bar. b-e, Genotypes as in a.

RTT patients and mouse models have abnormal respiration^1,22 (Fig. 3a). By 12 weeks, Mecp2^Stop/y-hGFAPcreT2 mice had irregularity scores and apnea rates significantly more severe than hGFAPcreT2 controls (Fig. 3a and b and Supplementary Fig. 7c, Traces 1 and 2). In contrast, two months after TAM-treatment, the respiratory pattern in 10 of 12 Mecp2^Stop/y-hGFAPcreT2 mice were within normal range (Fig. 3b). Two mice followed over the subsequent five-month period maintained a regular breathing patterns (data not shown). In two of three TAM-treated mice, we observed complete reversal to a normal respiratory pattern (Fig. 3a and Supplementary Fig. 7c, Traces 3 and 4). The apneic frequency in the third mouse was reduced but did not completely reverse to control levels (Supplementary Fig. 7c, Trace 5). The improvement in respiration was consistent with efficient re-expression of MeCP2 in GFAP⁺ astrocytes within the preBötzinger complex of the brainstem, an area implicated in respiratory defects in RTT²³ (Supplementry Fig. 7e). Treatment of a Mecp2^Stop/y mouse with TAM did not alleviate the irregular breathing or apneic frequency (Supplementary Fig. 7c, Trace 2). Oil-treated female Mecp2^+/Stop- hGFAPcreT2 mice developed a significant number of apneas beginning at four to six months (Fig. 3c). The apneic breathing was corrected by TAM treatment (Fig. 3c), even in the most severely affected female (Supplementary Fig. 7d).

Figure 3.

a, Representative plethysmographic recordings from a female RTT patient (modified from ref. 22) and an Mecp2^Stop/y-hGFAPcreT2 mouse and control. The two middle traces are from the same Mecp2^Stop/y-hGFAPcreT2 mouse before and 62 days after TAM treatment (Supplementary Fig. 7c, trace 3). b, Respiratory irregularity scores and apnoea rates for male mice. c, Same as in b except for female mice. Mice showing at least 1 apnoea per hour were considered for apnoea rates. All error bars indicate s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001. NS, not significant. The number of mice analysed is indicated above each bar.

The brains of girls with RTT and affected mice exhibit smaller neuronal somal sizes and reduced dendritic complexity in some regions^1,4,24,25. At ~ 3.5 months of age, the somal sizes of neurons in hippocampus, cerebellum, and cortex were still smaller in TAM-treated Mecp2^Stop/y-hGFAPcreT2 mice compared to hGFAPcreT2 controls. At seven months, however, somal size was restored only in brain regions showing astrocytic re-expression of MeCP2 (Fig. 4a). Regarding dendritic complexity, the Mecp2^Stop/y and oil- treated Mecp2^Stop/y-hGFAPcreT2 mice had ~25% fewer total number of apical dendrite branches compared to controls. By 3.5 months of age, however, neurons in TAM-treated Mecp2^Stop/y-hGFAPcreT2 mice had a normal number of branches and this was sustained with further age (Fig. 4b and c). MeCP2-deficient neurons also show deficits in proteins necessary for excitatory neurotransmission, such as VGlut1^26,27. We detected ~20% fewer peri-nuclear VGlut1⁺ puncta in Mecp2^Stop/y and oil-treated Mecp2^Stop/y-hGFAPcreT2 mice compared to controls, but the levels increased to normal by three to four months of age with TAM-treatment of Mecp2^Stop/y-hGFAPcreT2 mice (Fig. 4d and e). Taken together, the anatomical findings suggest that re-expression of MeCP2 in astrocytes can, through a non-cell autonomous mechanism, positively influence components of the neurotransmission machinery in vivo.

Figure 4.

a, Somal diameters of indicated neurons. Control, hGFAPcreT2 + TAM. b, Representative traces of silver-impregnated hippocampal CA1 neurons from male mice aged 3-4 months. c, Number of silver-impregnated CA1 apical branches in male mice. Control, Mecp2^+/y + TAM. d, Representative images of Nissl-stained neurons immunolabelled for VGLUT1 from medulla oblongata. Scale bar: 10 μm (b); 2 μm (d). e, Number of VGLUT1⁺ puncta associated with neuronal cell bodies from the medulla oblongata. All error bars indicate s.e.m. ***P < 0.001. NS, not significant. The number of analysed cells is indicated above each bar.

Our results show that re-expression of MeCP2 in astrocytes ameliorates overt RTT-like phenotypes in mice. To address the complementary question of the consequences of the removal of MeCP2 from astrocytes, we crossed mice with a floxed MeCP2 allele⁴ to the same hGFAPcreT2¹⁵ line used for the rescue. Recombination efficiencies throughout the brain were again higher in caudal compared to rostral regions (Supplementary Fig. 10a and b). The knockout progeny displayed some phenotypes shared with null MeCP2 mice, such as smaller body size, clasped hindlimb posture, and irregular breathing (Supplementary Fig. 10c, d, and f), but their lifespans, locomotion (data not shown), and anxiety-related behaviors were all normal (Supplementary Fig. 10e). Further, loss of MeCP2 from astrocytes did not affect the number of CA1 apical dendritic branches (Supplementary Fig. 10g). This suggests that loss of MeCP2 in astrocytes at postnatal day 21 is unable to disrupt the already established hippocampal neuronal circuitry. In contrast, loss from and gain of MeCP2 in astrocytes resulted in a strong non-cell autonomous influence on breathing pattern. Thus, distinct neuronal-glial interactions may underlie hippocampal and hindbrain breathing circuitries.

Our results indicate that RTT involves impairments in both neurons and glia. In familial amyotrophic lateral sclerosis²⁸, neurons and glia are proposed to play different roles in the disease process, with neurons primarily initiating the disease and astrocytes primarily affecting disease progression. Our results are compatible with this model, because removal of MeCP2 just from astrocytes, at postnatal day 21, results in a subtler phenotype than the global null, and re-expression in astrocytes mainly stabilizes symptoms. Along these lines, the appearance of a subset of phenotypes after embryonic removal of MeCP2 from subsets of neurons^4,6-10,29 could be explained by causing disease initiation, and prevention of RTT-like phenotypes after MeCP2 re-expression in embryonic neurons⁹ could be interpreted as preventing disease initiation. None of these studies address whether it could take both MeCP2-deficient neurons and glia to cause disease progression, or whether other non-neuronal cell types, including other glial types, might be involved in the disease process.

In sum, although impaired neurons ultimately underlie nervous system failure in RTT, restoring MeCP2 in glia can ameliorate four consistent and robust features of mouse models of RTT: premature lethality, aberrant respiration, hypoactivity, and decreased dendritic complexity. Future studies identifying the key molecules that are restored after glial MeCP2 re-expression may provide further clues into the mechanism of recovery, thereby providing new potential targets for therapeutic intervention.

Methods summary

Male mice harboring an hGFAPcreERT2 transgene were crossed to female Mecp2^+/Stop or Mecp2^{+/Jaenisch.Flox} mice and the F1 progeny were injected with 100mg/kg tamoxifen or oil when appropriate. Mice used in astrocyte-rescue experiments were backcrossed for at least seven generations to a C57BL6 background. Mice used in astrocyte-knockout experiments were of a FVB/N/C57BL6 background. Histology was performed on transcardially perfused, frozen sections. Behavior was analyzed using CleverSystems StereoScan software. Body plethysmography was performed on unanesthetized restrained mice. Statistics were performed with Graphpad PRISM V5.0C software. Mouse maintenance, breeding, and genotyping, tamoxifen treatments, phenotype scoring, tissue preparation and immunohistochemistry, fluorescence intensity measurements, FACS, plethysmography, motor activity and anxiety assessments, western blotting, and statistics were performed as described in Supplementary Methods.