1. Developmental Biology
Download icon

Mechanism and consequence of abnormal calcium homeostasis in Rett syndrome astrocytes

  1. Qiping Dong
  2. Qing Liu
  3. Ronghui Li
  4. Anxin Wang
  5. Qian Bu
  6. Kuan Hong Wang
  7. Qiang Chang  Is a corresponding author
  1. University of Wisconsin-Madison, United States
  2. National Institute of Mental Health, United States
Research Article
  • Cited 0
  • Views 1,297
  • Annotations
Cite as: eLife 2018;7:e33417 doi: 10.7554/eLife.33417

Abstract

Astrocytes play an important role in Rett syndrome (RTT) disease progression. Although the non-cell-autonomous effect of RTT astrocytes on neurons was documented, cell-autonomous phenotypes and mechanisms within RTT astrocytes are not well understood. We report that spontaneous calcium activity is abnormal in RTT astrocytes in vitro, in situ, and in vivo. Such abnormal calcium activity is mediated by calcium overload in the endoplasmic reticulum caused by abnormal store operated calcium entry, which is in part dependent on elevated expression of TRPC4. Furthermore, the abnormal calcium activity leads to excessive activation of extrasynaptic NMDA receptors (eNMDARs) on neighboring neurons and increased network excitability in Mecp2 knockout mice. Finally, both the abnormal astrocytic calcium activity and the excessive activation of eNMDARs are caused by Mecp2 deletion in astrocytes in vivo. Our findings provide evidence that abnormal calcium homeostasis is a key cell-autonomous phenotype in RTT astrocytes, and reveal its mechanism and consequence.

https://doi.org/10.7554/eLife.33417.001

Introduction

Rett syndrome (RTT) is a debilitating neurodevelopmental disorder (Hagberg, 1985) caused by mutations in the methyl-CpG binding protein 2 (MECP2) gene (Amir et al., 1999). RTT is a complex disease, as all key cell types (neurons, astrocytes, microglia, and oligodendrocytes) in the brain have been shown to contribute to the disease etiology (Ballas et al., 2009; Chao et al., 2010; Derecki et al., 2012; Fyffe et al., 2008; Lioy et al., 2011; Luikenhuis et al., 2004; Maezawa and Jin, 2010; Maezawa et al., 2009; Nguyen et al., 2013; Samaco et al., 2009; Williams et al., 2014). To fully understand the disease etiology and develop effective treatments, it is essential to define the key phenotypes, link the phenotypes with loss of MeCP2 function, and reveal the consequences of the phenotypes in each brain cell type. Moreover, it is also important to define the connection between phenotypes in different cell types.

Earlier research to understand the RTT disease etiology mostly focused on neuronal dysfunction in RTT (Chao et al., 2010; Fyffe et al., 2008; Luikenhuis et al., 2004; Samaco et al., 2009), because MeCP2 is expressed at high levels in neurons in a temporal and spatial pattern closely tracking neuronal maturation (Akbarian et al., 2001; Shahbazian et al., 2002). Not surprisingly, specific subsets of RTT-like phenotypes were observed in mouse models when Mecp2 is deleted in specific neuronal subtypes (Chao et al., 2010; Fyffe et al., 2008; Samaco et al., 2009). Collectively, these studies confirm the essential role of MeCP2 in maintaining the normal function of neurons, suggesting the loss of function or malfunction of MeCP2 in neurons as the main cause of RTT pathogenesis. Recently, a series of studies have reported that astrocytes express MeCP2, loss of MeCP2 in astrocytes causes neuronal defects in a non-cell autonomous manner, and restoring MeCP2 expression to normal level in astrocytes alone rescues some disease symptoms (Ballas et al., 2009; Lioy et al., 2011; Maezawa et al., 2009; Williams et al., 2014). Intriguingly, cell type specific deletion of Mecp2 in astrocytes alone is not sufficient to generate most of the RTT-like phenotypes in mice (Lioy et al., 2011), suggesting the loss of function or malfunction of MeCP2 in astrocytes is not the initial cause of the disease, yet is required to maintain the disease progression. Although a few studies have reported gene expression changes in Mecp2 mutant mouse astrocytes (Forbes-Lorman et al., 2014; Yasui et al., 2013), no major cell autonomous phenotype in astrocytes has been characterized in depth. Thus, identification of cell autonomous phenotypes in astrocytes not only is essential for our understanding of astrocyte dysfunction in RTT, but also will help us to determine the mechanism underlying its non-cell autonomous effects on neurons.

To identify novel cell autonomous phenotypes in RTT astrocytes, we examined intracellular calcium dynamics in astrocytes differentiated from congenic pairs of wild type and mutant RTT iPSC lines. We have used the term ‘isogenic’ to describe these RTT iPSC lines in our previous reports (Ananiev et al., 2011; Williams et al., 2014). However, given that the difference between wild type and mutant iPSC lines generated from the same RTT patients is the entire X chromosome, congenic is a more accurate term to define these lines. We observed abnormal spontaneous and pharmacologically evoked cytosolic calcium activities in cultured mutant RTT astrocytes, when compared to their congenic wild type controls. Similar phenotypes can be observed in Mecp2 knockout mouse astrocytes in vitro, in situ, and in vivo. In addition, re-expression of wild type MeCP2 is sufficient to rescue the abnormal calcium activities in both the mouse and human RTT astrocytes, further confirming that these phenotypes are caused by the loss of MeCP2 function. Moreover, we report that these abnormal calcium activities are mediated by calcium overload in the endoplasmic reticulum (ER), which is caused by abnormal TRPC4-dependent store operated calcium entry (SOCE). Although we provide no direct evidence that TRPC4 is the actual SOCE channel, both loss- and gain-of-function manipulations from our study suggest TRPC4 is upstream of the SOCE and subsequent calcium phenotypes in RTT astrocytes. Finally, we show that these abnormal calcium activities in astrocytes lead to excessive activation of extrasynaptic NMDA receptors (eNMDARs) on neighboring neurons and increased network excitability; and both the abnormal astrocytic calcium activity and the excessive activation of eNMDARs are caused by the loss of Mecp2 in astrocytes (but not neurons) in vivo. Our findings provide the first evidence that suggests abnormal calcium homeostasis as a cell autonomous phenotype in RTT astrocytes, and begin to reveal the mechanism and consequence of such a phenotype.

Results

Abnormal calcium activities in RTT astrocytes

Congenic pairs of wild type and mutant RTT iPSC lines carrying the common R294X mutation were differentiated into glial fibrillary acidic protein positive (GFAP+) astrocytes as described in our previous paper (Williams et al., 2014). On average, more than 90% of the cells were GFAP +in cultures used in our study (Figure 1—figure supplement 1 and Table 1). Using the green-colored calcium-sensitive dye Fluo-4, we monitored spontaneous cytosolic calcium (Ca2+) dynamics in live mutant (R294X-MT) and congenic wild type control (R294X-WT) astrocytes (Videos 12, Figure 1A–B). Ionophore A23187 was used as a loading control (Figure 1—figure supplement 2). A significantly higher percentage of mutant astrocytes (30 ± 3% in MT vs. 18 ± 2% in WT, n = 26 randomly selected fields in each genotype, p=0.001) showed spontaneous oscillation in cytosolic Ca2+ levels (Figure 1C). In each astrocyte showing spontaneous oscillation, we further separated and quantified the calcium transients in the soma from those in the processes (Figure 1A). In the soma, the mutant cells had a significantly higher frequency (0.74 ± 0.02/min in MT vs. 0.41 ± 0.01/min in WT, n = 408 cells in MT and n = 280 cells in WT, p<0.001; Figure 1C) and slightly higher amplitude (Table 2) than their congenic wild type controls. Similar changes were observed in the processes of those astrocytes (Table 2 and Figure 1—figure supplement 3). Moreover, the R294X mutant astrocytes had significantly higher peak amplitude (4.60 ± 0.07 in MT vs. 4.04 ± 0.07 in WT, n = 567 cells in MT and n = 811 cells in WT, p<0.001) of ATP-induced Ca2+ rise than the congenic wild type control (Figure 1D–E). The increase in ATP-induced Ca2+ rise in human RTT mutant astrocytes was unlikely to be mediated by the P2 receptors, because no difference was detected in the expression level of any P2X or P2Y receptor family members between the wild type and mutant RTT astrocytes by RNA-seq and microarray analysis (data not shown). Similar observations of altered Ca2+ activities were made in human astrocytes carrying the rare RTT-causing V247fs mutation (Figure 1—figure supplement 4). Thus, both spontaneous and pharmacologically evoked cytosolic Ca2+ activity appeared abnormal in mutant human RTT astrocytes differentiated from patient-specific iPSCs and grown in the absence of neurons.

Video 1
Spontaneous calcium activities in wild type human astrocytes
https://doi.org/10.7554/eLife.33417.002
Video 2
Spontaneous calcium activities in mutant human astrocytes
https://doi.org/10.7554/eLife.33417.003
Table 1
Quantification of the percentage of GFAP+ cell in five differentiations from human iPSCs
https://doi.org/10.7554/eLife.33417.004
stereology #12345
 WT90%94%92%89%91%
 MT90%92%94%89%92%
Table 2
Amplitudes of spontaneous Ca2+ elevations in different experiments
https://doi.org/10.7554/eLife.33417.005
Cell typeWild-typeMutant
 R294X astrocytes1.660 ± 0.01624 N = 10911.808 ± 0.01882 N = 2962
 R294X astrocytes (processes)1.467 ± 0.01601 N = 22281.722 ± 0.01994 N = 2743
 V247fs astrocytes1.686 ± 0.02453 N = 9021.816 ± 0.02128 N = 1760
 V247fs astrocytes (processes)1.560 ± 0.02224 N = 18251.640 ± 0.02447 N = 2312
 Mouse primary astrocytes1.787 ± 0.007193 N = 56941.822 ± 0.006209 N = 8389
 Astrocytes in situ2.010 ± 0.02998 N = 13672.151 ± 0.03079 N = 1997
 Astrocytes in vivo3.463 ± 0.1410 N = 2853.045 ± 0.05410 N = 1346
 Astrocytes in vivo (processes)3.283 ± 0.03640 N = 20403.007 ± 0.01546 N = 8796
Figure 1 with 4 supplements see all
Abnormal Ca2+ activities in mutant human RTT astrocytes.

(A) Pseudocolored Fluo4 fluorescence images from wild type (WT, left) and MeCP2 mutant (MT, right) astrocytes differentiated from human iPSCs. Green ellipses indicate astrocyte cell soma, while yellow rectangles indicate processes. Scale bars = 50 μm. (B) Representative ΔF/F0 traces showing the spontaneous intracellular Ca2+ activity from soma in (A). The black traces are from WT astrocytes and red traces are from MT ones. (C) Quantification of the percentage (left) of astrocytes showing spontaneous Ca2+ oscillations and the frequency (right) of such oscillations. (D) Trace of Fluo4 fluorescence changes in wild type (WT) and mutant (MT) human astrocytes stimulated by 10 μM ATP. Average traces are shown with the solid lines. (E) Quantification of the peak amplitude of the ATP-evoked Ca2+ elevations in wild type (WT) and mutant (MT) human astrocytes.

https://doi.org/10.7554/eLife.33417.006

To determine whether the observed abnormal cytosolic Ca2+ activities were caused by the loss of MeCP2 function, we infected mutant human RTT astrocytes with lentiviruses expressing either the green fluorescence protein (lenti-GFP) alone or GFP and wild type MECP2 together (lenti-MECP2/GFP). Three days after the infection, more than 90% of the cells were positive for GFP in both infection types (Figure 2A). In R294X mutant astrocytes infected with lenti-MECP2/GFP, anti-MECP2 immunoreactivity was detected in all the GFP+ cells (right, Figure 2A). Further quantification of MECP2 immunoreactivity in those cells revealed no difference from wild type astrocytes (Figure 2—figure supplement 1), suggesting the exogenous expression of MECP2 was not significant above the endogenous level. When Ca2+ activity was recorded (Figure 2B), the percentage of cells with spontaneous Ca2+ oscillation was significantly reduced by the expression of wild type MECP2, compared with expression of GFP alone, in the mutant human RTT astrocytes (20 ± 2% in MECP2/GFP vs. 36 ± 4% in GFP, n = 6 randomly selected fields in each group, p=0.006; Figure 2C). In the cells detected with spontaneous cytosolic Ca2+ oscillation, both the frequency (0.27 ± 0.06/min in MECP2/GFP vs. 0.59 ± 0.06/min in GFP, n = 21 cells in MECP2/GFP and n = 32 cells in GFP, p<0.001; Figure 2C) and amplitude (1.65 ± 0.08 in MECP2/GFP vs. 1.75 ± 0.03 in GFP, n = 62 events in MECP2/GFP and n = 209 events in GFP, p<0.001) were significantly reduced by the exogenous expression of wild type MECP2 as compared with expression of GFP alone. In addition, expression of wild type MECP2 significantly reduced the elevated level of pharmacologically evoked Ca2+ response in mutant human RTT astrocytes (3.90 ± 0.13 in MECP2/GFP vs. 4.20 ± 0.10 in GFP, n = 96 cells in MECP2/GFP and n = 123 cells in GFP, p=0.022; Figure 2D–E). Similar rescue effects were observed when the V247fs mutant astrocytes were used (Figure 2—figure supplement 2). Together, these results further support the cell autonomous nature of the abnormal Ca2+ activities in RTT mutant astrocytes.

Figure 2 with 2 supplements see all
Exogenous expression of MeCP2 rescues the abnormal calcium activities in RTT astrocytes.

(A) Representative images of mutant human RTT astrocytes infected with lentivirus expressing GFP alone (left) or with lentivirus co-expressing wild type MECP2 and GFP (right). MECP2 immunoreactivity was detected in GFAP positive cells in the right column, but not the left column. Scale bars, 50 μm. (B) Representative ΔF/F0 traces showing the spontaneous intracellular Ca2+ activity in RTT mutant astrocytes infected with lentivirus expressing either GFP alone (GFP) or co-expressing GFP and wild type MECP2 together (MECP2/GFP). (C) Quantification of the percentage (left) of astrocytes showing spontaneous Ca2+ oscillations and the frequency (middle) and amplitude (right) of such oscillations in RTT mutant astrocytes infected with lentivirus expressing either GFP alone (GFP) or co-expressing GFP and wild type MECP2 together (MECP2/GFP). (D) Trace of ATP (10 μM) evoked Fluo4 fluorescence changes in RTT mutant astrocytes infected with lentivirus expressing either GFP alone (GFP) or co-expressing GFP and wild type MECP2 together (MECP2/GFP). Average traces are shown with the solid lines. (E) Quantification of the peak amplitude of the ATP-evoked Ca2+. The bar graphs in this figure show the mean ±s.e.m. *p<0.05, **p<0.01, ***p<0.001.

https://doi.org/10.7554/eLife.33417.012

Since the phenotype of abnormal Ca2+ activities have not been reported in previous studies using RTT mouse models, we next examined whether they can be observed in Mecp2 null mouse astrocytes. Primary hippocampal astrocytes were isolated from newborn male Mecp2 null mice (Mecp2-/y) and their wild type (WT) littermates, both of which also carried the Rosa-CAG-LSL-GCaMP6s allele (http://jaxmice.jax.org/strain/024106) and the hGFAP-creERT allele (http://www.jax.org/strain/012849). The astrocytes were treated with 4-OH tamoxifen for 24 hr to induce the expression of GCaMP6s, and then imaged for GCaMP6s fluorescence as the indication of cytosolic Ca2+ activities. Similar to our findings in human RTT astrocytes, spontaneous Ca2+ oscillation (Figure 3A–B) was detected in significantly higher percentage of Mecp2 null astrocytes than in wild type controls (67 ± 5% in Mecp2-/y vs. 47 ± 6% in WT, n = 19 randomly chosen fields in both Mecp2-/y and WT, p=0.01; Figure 3C). In cells that showed spontaneous Ca2+ oscillation, significantly higher frequency (2.01 ± 0.04/min in Mecp2-/y vs. 1.64 ± 0.03/min in WT, n = 575 cells in Mecp2-/y and n = 439 cells in WT, p<0.001; Figure 3C) and slightly higher amplitude (Table 2) were observed in the soma of Mecp2 null astrocytes than in wild type astrocytes. Since the processes were very short in primary mouse astrocyte culture, spontaneous Ca2+ activity in processes was not quantified separately because it closely resembled that in the soma. As for pharmacologically evoked Ca2+ response, significantly higher amplitude was detected in primary Mecp2 null mouse astrocytes (7.25 ± 0.23 in Mecp2-/y vs. 6.54 ± 0.12 in WT, n = 153 cells in Mecp2-/y and n = 278 cells in WT, p=0.04) than in wild type astrocytes (Figure 3D–E).

Abnormal Ca2+ activities in primary astrocytes isolated from Mecp2 null mice.

(A) The pseudocolored GCaMP6s fluorescence image from astrocytes. (B) Representative ΔF/F0 traces showing the spontaneous intracellular Ca2+ activity in mouse primary astrocytes from wild type and Mecp2-/y mice. (C) Quantification of the percentage (left) of astrocytes showing spontaneous Ca2+ oscillations and the frequency (right) of such oscillations. (D) Trace of fluorescence changes in wild type and Mecp2-/y astrocytes stimulated by 10 μM ATP. Average traces are shown with the solid lines. (E) Quantification of the peak amplitude of the ATP-evoked Ca2+ elevations from wild type and Mecp2-/y astrocytes.

https://doi.org/10.7554/eLife.33417.016

Next, to explore the in situ relevance of this novel phenotype in Ca2+ dynamics, we studied spontaneous and agonist evoked Ca2+ activities in astrocytes in acute hippocampal slices prepared from 2 to 3 week-old Mecp2 null mice and their wild type littermates using Fluo4. In this series of experiments, SR101 was used to label astrocytes (Figure 4A); TTX (1 μM) was used to suppress spontaneous neuronal activity. Similar to our observation in cultured human patient-specific iPSC derived astrocytes and primary mouse astrocytes, higher frequency (0.59 ± 0.01/min in Mecp2-/y vs. 0.47 ± 0.01/min in WT, n = 337 cells in Mecp2-/y and n = 293 cells in WT, p<0.001) and slightly higher amplitude (Table 2) of spontaneous Ca2+ oscillation was observed in the soma of SR101+ cells in Mecp2 null slices than in wild type slices (Figure 4B–C). In addition, the astrocytes in Mecp2 null slices had significantly higher peak amplitude (2.75 ± 0.02 in MT vs. 2.22 ± 0.09 in WT, n = 754 cells in Mecp2-/y and n = 238 cells in WT, p<0.001) of glutamate-induced Ca2+ rise than astrocytes in wild type slices (Figure 4D).

Abnormal Ca2+ activities in astrocytes in acute brain slices prepared from Mecp2 null mice.

(A) Representative images of SR101 labeling (astrocyte marker) and Fluo4 signals in acute hippocampal slices. Note the colocalization of SR101 and Fluo4. Scale bars = 50 μm. (B) Representative ΔF/F0 traces showing the spontaneous intracellular Ca2+ activity in SR101 positive cells in acute hippocampal slices prepared from wild type and Mecp2-/y mice. (C) Quantification of the frequency of the astrocytic Ca2+ oscillations in the slice experiments. (D) Traces of Fluo4 fluorescence changes stimulated by 100 μM glutamate in SR101 positive cells in acute hippocampal slices prepared from wild type and Mecp2-/y mice. Average traces are shown with solid lines. (E) Quantification of the peak amplitude of the glutamate-evoked Ca2+ elevations. The bar graphs in this figure show the mean ±s.e.m. *p<0.05, **p<0.01, ***p<0.001.

https://doi.org/10.7554/eLife.33417.018

To further rule out potential artifacts in cultured astrocytes and in acute slices, we examined spontaneous Ca2+ activities in astrocytes of the frontal cortex in live mice carrying the GCaMP6s sensor. We mated male Rosa-CAG-LSL-GCaMP6s mice with either female wild type (Mecp2+/+) or Mecp2flox/flox mice, injected adeno-associated virus (AAV) expressing mCherry and Cre recombinase under the human GFAP promoter (AAV8-hGFAP-mCherry-Cre) into the lateral ventricles of the postnatal day one male pups from these matings to activate the expression of GCaMP6s and delete Mecp2 predominantly in astrocytes. Based on the mating scheme, male pups would be of the genotype of either Mecp2+/y;Rosa-CAG-LSL-GCaMP6s (WT) or Mecp2flox/y; Rosa-CAG-LSL-GCaMP6s (floxed). When immunostaining with MeCP2 antibody on sections prepared from 3 months old floxed mice that receivedAAV8-hGFAP-mCherry-Cre injection at birth, 100% of the GCaMP6s (anti-GFP) positive cells in the cortex were MeCP2 negative. In contrast, GCaMP6s positive cells from WT mice were MeCP2 positive (Figure 5—figure supplement 1). When GCaMP6s fluorescence were recorded in mCherry-positive cells in the frontal cortex of 2–3 months-old live WT and floxed mice injected with AAV8-hGFAP-mCherry-Cre (MT) (Video 34Figure 5A–B), higher frequency of spontaneous oscillation of GCaMP6s fluorescence was detected in both the soma (2.15 ± 0.11/min in MT vs. 1.44 ± 0.20/min in WT, n = 57 cells in MT and n = 18 cells in WT, p=0.003, Figure 5C) and processes (2.14 ± 0.04/min in MT vs. 1.69 ± 0.06/min in WT, n = 373 processes in MT and n = 110 processes in WT, p<0.001, Figure 5—figure supplement 2) of MT astrocytes. However, the amplitude of spontaneous astrocytic Ca2+ activity appeared to be slightly smaller in MT mice (Table 2). Taken together, the phenotype of increased frequency of spontaneous cytosolic Ca2+ activity was found consistently across species, as well as in vitro, in situ, and in vivo.

Video 3
Spontaneous calcium activities in astrocytes from live wild type mice
https://doi.org/10.7554/eLife.33417.020
Video 4
Spontaneous calcium activities in astrocytes from live Mecp2 mutant mice
https://doi.org/10.7554/eLife.33417.021
Figure 5 with 2 supplements see all
Abnormal spontaneous Ca2+ activities in vivo from Mecp2 null astrocytes.

(A) Representative GCaMP6s image showing two astrocytes in vivo. The somas and the processes could be clearly identified. Scale bars = 10 μm. (B) Representative ΔF/F0 traces showing the spontaneous intracellular Ca2+ activity in the soma of astrocytes in vivo from wild type and Mecp2 null astrocytes. (C) Quantification of the frequency (left) and the full width at half maximum (FWHM, right) of the astrocytic Ca2+ oscillations in vivo. The bar graphs in this figure show the mean ±s.e.m. ***p<0.001.

https://doi.org/10.7554/eLife.33417.022

Abnormal calcium load in the ER, calcium leak from the ER, baseline cytosolic calcium level, and TRPC4-dependent SOCE in RTT astrocytes

To reveal the cellular and molecular mechanisms underlying the abnormal Ca2+ activities in the RTT astrocytes, we first examined the source of Ca2+. Since the abnormal spontaneous cytosolic Ca2+ oscillation in mutant human astrocytes persisted after Ca2+ was removed completely from extracellular environment (0 mM Ca2+ plus 2 mM EGTA, Figure 6—figure supplement 1), we measured the Ca2+ load in the endoplasmic reticulum (ER), the major source of intracellular Ca2+ storage. Treating astrocytes with thapsigargin (TG, 1 μM), an irreversible inhibitor of sarco-ER Ca2+ ATPase, induced a slow elevation of Ca2+ in the cytosol, which was mediated by the leakage of Ca2+ from the ER. The amplitude of the TG-induced Ca2+ elevation was significantly higher in the mutant human RTT astrocytes than in their congenic controls (125 ± 4%. in MT vs. 100 ± 3% in WT, n = 156 cells in MT and n = 178 cells in WT, p<0.001, Figure 6A), suggesting the mutant astrocytes had more releasable Ca2+ in the ER. Moreover, when the kinetics of this Ca2+ rise was analyzed, the leakage of Ca2+ from the ER in the mutant astrocytes was much faster (time constant 54 ± 1 s in MT vs. 59 ± 1 s in WT, n = 156 cells in MT and n = 178 cells in WT, p<0.001, Figure 6B–C).

Figure 6 with 5 supplements see all
RTT astrocytes display abnormal calcium load in the ER, calcium leak from the ER, baseline cytosolic calcium level, and TRPC4-dependent SOCE.

(A) Left, average traces of Fluo4 fluorescence changes in wild type and mutant astrocytes treated with 1 μM TG to release ER calcium. Bath solution contained 0 mM Ca2+ plus 2 mM EGTA. Right, Quantification of the peak amplitude of TG induced Ca2+ elevations. (B) The rise phase (dots) shown in (A) is fitted with a single exponential curve. (C) Quantification of the time constant of the rise phase in the left panel from congenic mutant and wild type human RTT astrocytes in response to TG. (D) Representative GFP and Rhod-2 images in primary astrocytes from female Mecp2+/- ± with a GFP transgene on the wild type X chromosome. Scale bars = 50 μm. (E–F) Quantification of Rhod-2 fluorescence intensity in GFP negative and GFP positive primary astrocytes isolated from either female Mecp2+/- ± with a GFP transgene on the wild type X chromosome (E) or female Mecp2+/+ mice with a GFP transgene on the wild type X chromosome (F). (G) The average trace of Fluo4 fluorescence changes in response to extracellular Ca2+ (2 mM) after depletion of ER Ca2+ store using TG. (H) The Western blot result showing the protein level of TRPC4 in WT and MeCP2 mutant astrocytes. (I) Quantification of the Western blot results. (J) Representative traces of current density of TG-induced inward current in wild type and mutant astrocytes held at −70 mV. This current was partially blocked by TRPC4 selective antagonist, ML204. (K) Quantification of ML204 sensitive current from wild type and mutant astrocytes held at −70 mV. (L) Representative traces of current density of TG-induced inward current in mutant astrocytes infected with lentivirus expressing either GFP alone or co-expressing GFP and shTRPC4. (M) Quantification of ML204 sensitive current at −70 mV. The bar graphs in this figure show the mean ±s.e.m. *p<0.05, **p<0.01, ***p<0.001.

https://doi.org/10.7554/eLife.33417.026

To accurately compare the baseline cytosolic Ca2+ level between Mecp2 mutant and wild type astrocytes, we prepared primary mouse astrocytes from Mecp2+/- mice that carry an X chromosome-linked GFP transgene. The GFP transgene is on the same X chromosome as the wild type Mecp2 gene. Due to random X chromosome inactivation, a mixture of GFP+ and GFP- astrocytes will be present in these cultures, with the GFP+ cells expressing the wild type MeCP2 and the GFP- cells expressing no MeCP2. When the red-colored calcium-sensitive dye Rhod-2 was used to image baseline cytosolic Ca2+ level in the Mecp2 null (GFP-) and wild type (GFP+) astrocytes in the same culture (Figure 6D), significantly higher level of Rhod-2 fluorescence was detected in the Mecp2 null cells (154 ± 7% in GFP- vs. 100 ± 5% in GFP+, n = 30 in each group, p<0.0001; Figure 6E). As a control, astrocyte cultures were also prepared from Mecp2+/+ female mice that carry the same X chromosome-linked GFP transgene. No significant deference was detected in baseline cytosolic Ca2+ level between the GFP+ and GFP- cells (107 ± 8% in GFP- vs. 100 ± 6% in GFP+, n = 31 in GFP- and n = 32 in GFP+, p=0.46; Figure 6F). These results reveal Ca2+ overload in the ER, the faster Ca2+ leak rate, and the elevated baseline cytosolic Ca2+ level as significant contributors to the abnormal spontaneous and pharmacologically evoked cytosolic Ca2+ activities in the mutant RTT astrocytes.

To further define the cellular events upstream of the abnormal Ca2+ homeostasis, we measured Ca2+ influx through the store operated calcium entry (SOCE) pathway, which is the primary pathway for reloading Ca2+ into the ER in astrocytes. After depleting Ca2+ from the ER by treating astrocytes with TG in the absence of extracellular Ca2+ (0 mM Ca2+ plus 2 mM EGTA), we switched bath solution to that containing 2 mM Ca2+ and measured Ca2+ influx. Comparing to that in the congenic control, the amplitude of Ca2+ influx was significantly higher in the mutant human RTT astrocytes (2.93 ± 0.07 in MT vs. 2.41 ± 0.06 in WT, n = 156 cells in MT and n = 178 cells in WT, p<0.001; Figure 6G), suggesting an abnormal increase of SOCE in the mutant cells.

In search of the molecular mechanism underlying the increased SOCE in mutant astrocyte, we performed RNA-seq and microarray experiments to compare gene transcription profiles in the mutant astrocytes and their congenic controls. Highly relevant to the observed SOCE phenotype, the expression of the transient receptor potential cation channel, subfamily C, member 4 (TRPC4), which has been previously shown to regulate SOCE in lung vascular endothelial cells (Tiruppathi et al., 2002), was found to be significantly higher in the mutant human RTT astrocytes (~27 fold increase by RNA-seq,~14 fold increase by microarray). Western blot analysis using protein lysates from mutant human RTT astrocytes and their congenic controls also confirmed the increased expression of TRPC4 (Figure 6H–I). Immunostaining with TrpC4 antibody also detected significantly higher immunoreactivity in GFAP positive astrocytes in hippocampal sections from the Mecp2 null mouse brains as compared with that in wild type mouse brains (Figure 6—figure supplement 2). Chromatin immunoprecipitation analysis revealed significant MeCP2 occupancy at the promoter of Trpc4 (Figure 6—figure supplement 3), suggesting direct regulation of Trpc4 transcription by MeCP2. To corroborate the expression level changes at the RNA and protein level for TRPC4, we performed whole cell patch clamp recording on the astrocytes to directly assess the current mediated by TRPC4 using a TRPC4-selective antagonist, ML204 (Miller et al., 2011). Significantly higher density of ML204 sensitive inward current (at −70 mV) was detected in the R294X mutant human RTT astrocytes (−0.93 ± 0.01 pA/pF in MT vs. −0.62 ± 0.08 pA/pF in WT, n = 14 cells in MT and n = 15 cells in WT, p=0.01; Figure 6J–K). Moreover, this current was significantly reduced when mutant human RTT astrocytes were infected with a lentivirus expressing an shRNA specific against TRPC4 (−0.61 ± 0.10 pA/pF in MT infected with lenti-GFP-shTRPC4 vs. −1.04 ± 0.05 pA/pF in MT infected with lenti-GFP, n = 9 cells in each condition, p=0.006, Figure 6L–M), suggesting the current is dependent on TRPC4. The I-V curve of the ML204-sensitive current can be found in Figure 6—figure supplement 4. Finally, the phenotypes of increased Ca2+ overload in the ER, faster Ca2+ leak from the ER, increased SOCE, and increased level of TrpC4 were also observed in primary astrocytes isolated from Mecp2 null mice (Figure 6—figure supplement 5).

To determine the contribution of increased TRPC4 expression to the series of Ca2+ related phenotypes in the Mecp2 null astrocytes, we treated these cells with ML204 for 24 hr, and then measured SOCE-mediated Ca2+ elevation, Ca2+ load in the ER, and recorded spontaneous cytosolic Ca2+ activities. Comparing to untreated Mecp2 null astrocytes, ML204-treated cells showed significantly reduced amplitude of SOCE mediated Ca2+ elevation (3.15 ± 0.12 in ML204-treated group vs. 6.61 ± 0.86 in untreated, n = 18 in ML204-treated and n = 16 in untreated, p=0.0002; Figure 7A) and lower ER Ca2+ load (1.86 ± 0.06 in ML204-treated vs. 2.47 ± 0.24 in untreated, n = 14 in ML204-treated and n = 15 in untreated, p=0.02; Figure 7B). More importantly, ML204 treatment significantly reduced the percentage of Mecp2 null astrocytes showing spontaneous cytosolic Ca2+ oscillation (23 ± 3% in ML204-treated vs. 46 ± 7% in untreated, n = 9 randomly chosen fields in ML204-treated and n = 8 randomly chosen fields in untreated, p=0.03; Figure 7C) and the frequency (0.20 ± 0.01/min in ML204-treated vs. 0.31 ± 0.01/min in untreated, n = 91 cells in ML204-treated and n = 203 cells in untreated, p<0.001; Figure 7C) and amplitude (1.64 ± 0.02 in ML204-treated vs. 1.97 ± 0.02 in untreated, n = 168 events in ML204-treated and n = 597 events in untreated, p<0.001; Figure 7C) of such oscillations.

The role of TRPC4 in regulating calcium homeostasis in astrocytes.

(A) Extracellular Ca2+ -induced Fluo4 fluorescence changes after TG-induced ER Ca2+ depletion in Mecp2-/y astrocytes in the absence (control) or presence of ML204. (B) Average traces of TG-induced Fluo4 fluorescence changes in Mecp2-/y astrocytes with or without 24 hr of ML204 treatment. (C) Quantification of percentage (left) of astrocytes showing spontaneous Ca2+ oscillation and the frequency (middle) and amplitude (right) of such oscillation with or without 24 hr of ML204 treatment. (D) TRPC4 Western blot result showing decreased expression level of TRPC4 in mutant RTT human astrocytes after infected with lentivirus-shTrpc4-GFP, compared with astrocytes infected with lentivirus-GFP. (E) Quantification of the TRPC4 Western blot result. (F) Average traces of extracellular Ca2+ (2 mM)-induced Rhod-2 fluorescence changes from mutant RTT astrocytes infected with lentivirus expressing either shTRPC4/GFP or GFP alone, after depletion of ER Ca2+ store by TG pre-treatment. (G) Average traces of TG-induced Rhod-2 fluorescence changes from mutant RTT astrocytes infected with lentivirus expressing either shTRPC4/GFP or GFP alone. (H) Quantification of the frequency of spontaneous Ca2+ elevations from mutant RTT astrocytes infected with lentivirus expressing either shTRPC4/GFP or GFP alone. (I) Representative images of wild type human RTT astrocytes infected with lentivirus co-expressing TRPC4 and GFP (left) or with lentivirus expressing GFP alone (right). Note anti-TRPC4 immunoreactivity is higher in astrocytes infected with lenti-GFP/TRPC4 than in those infected with lenti-GFP. Scale bar = 50 μm. (J) Average traces of extracellular Ca2+ (2 mM)-induced Rhod-2 fluorescence changes from wild type astrocytes infected with lentivirus expressing either GFP/TRPC4 or GFP alone, after depletion of ER Ca2+ store by TG pre-treatment. (K) Average traces of TG-induced Rhod-2 fluorescence changes from wild type astrocytes infected with lentivirus expressing either GFP/TRPC4 or GFP alone. (L) Quantification of percentage (left) of astrocytes infected with lentivirus expressing either GFP/TRPC4 or GFP alone that showed spontaneous Ca2+ oscillation and the frequency (middle) and amplitude (right) of such oscillations. The bar graphs in this figure show the mean ±s.e.m. *p<0.05, **p<0.01, ***p<0.001.

https://doi.org/10.7554/eLife.33417.033

In addition to pharmacological manipulation, we infected mutant RTT human astrocytes with lentivirus expressing a shRNA specific against TRPC4 to reduce its level (Figure 7D–E). 72 hr after the infection, we measured SOCE-mediated Ca2+ elevation, ER Ca2+ load, and recorded spontaneous cytosolic Ca2+ activities. Comparing to those infected with lentivirus expressing GFP alone, the mutant astrocytes infected with lentivirus expressing shTRPC4 showed significantly reduced SOCE-mediated Ca2+ elevations (1.72 ± 0.05 in MT/Lenti-ShTRPC4-GFP vs. 2.02 ± 0.04 in MT/Lenti-GFP, n = 23 in MT/Lenti-shTRPC4 and n = 26 in MT/Lenti-GFP, p<0.0001; Figure 7F), lower ER Ca2+ load (2.37 ± 0.16 in MT/Lenti-shTRPC4-GFP vs. 2.95 ± 0.29 in MT/Lenti-GFP, n = 23 in MT/Lenti-shTRPC4 and n = 26 in MT/Lenti-GFP, p=0.04; Figure 7G), and lower frequency (0.42 ± 0.03/min in MT/Lenti-shTRPC4-GFP vs. 0.63 ± 0.03 in MT/Lenti-GFP, n = 86 cells in MT/Lenti-shTRPC4 and n = 79 cells in MT/Lenti-GFP, p<0.001, Figure 7H) of the spontaneous Ca2+ elevations. Together, the pharmacological and molecular manipulations demonstrated that elevated TRPC4 expression is required for the abnormal calcium homeostasis in MeCP2 deficient astrocytes.

Finally, we overexpressed TRPC4 in wild type human astrocytes (Figure 7I), and observed increased SOCE-mediated Ca2+ elevations (2.76. ± 0.15 in WT/GFP-TRPC4 vs. 2.36 ± 0.08 in WT/GFP, n = 52 in WT/GFP-TRPC4 and n = 55 in WT/GFP, p=0.02; Figure 7J), higher ER Ca2+ load (3.65 ± 0.23 in WT/GFP-TRPC4 vs. 2.62 ± 0.13 in WT/GFP, n = 52 in WT/GFP-TRPC4 and n = 55 in WT/GFP, p=0.0001; Figure 7K), more cells showing spontaneous Ca2+ oscillation (29 ± 5% in WT/GFP-TRPC4 vs. 16 ± 3% in WT/GFP, n = 13 in WT/GFP-TRPC4 and n = 12 in WT/GFP, p=0.03; Figure 7L), higher frequency (0.71 ± 0.04/min in WT/GFP-TRPC4 vs. 0.45 ± 0.01/min in WT/GFP, n = 74 cells in WT/GFP-TRPC4 and n = 100 cells in WT/GFP, p<0.001; Figure 7L) and amplitude (1.49 ± 0.04 in WT/GFP-TRPC4 vs. 1.22 ± 0.01 in WT/GFP, n = 523 events in WT/GFP-TRPC4 and n = 445 events in WT/GFP, p<0.001; Figure 7L) of spontaneous Ca2+ oscillations in those cells. These results strongly suggest that elevated TRPC4 expression is sufficient to cause abnormal calcium homeostasis in wild type astrocytes.

Abnormal calcium homeostasis in RTT astrocytes leads to excessive activation of extrasynaptic NMDA receptor activation in neighboring neurons and increased network excitability

Previous work has shown that Ca2+ oscillation in astrocytes could lead to activation of eNMDARs (Fellin et al., 2004). We measured activation of eNMDAR on mouse hippocampal pyramidal neurons by whole cell patch clamp recording, and detected a significantly higher frequency of a slow inward current (SIC, Figure 8A, left) in acute hippocampal slices prepared from 2 to 3 weeks old Mecp2 null mice than those from the wild type littermate control (3.88 ± 0.46/20 min in Mecp2 null vs. 1.50 ± 0.39/20 min in WT, n = 16 in Mecp2 null and n = 12 in WT, p=0.001; Figure 8A, right). The SICs were sensitive to both D-APV (selective NMDA receptor antagonist, 50 µM, Figure 8B, left) and ifenprodil (selective NR2B-containing NMDA receptor antagonist, 10 µM, Figure 8B, right), suggesting they were mediated by NR2B-containing eNMDARs. Similar increase in SIC frequency was also detected in acute hippocampal slices prepared from 6 weeks old symptomatic Mecp2 null mice (Figure 8—figure supplement 1). Although similar phenotype was recently reported (Lo et al., 2016), the underlying cause of such increased activation of eNMDAR is not well understood.

Figure 8 with 2 supplements see all
Abnormal Ca2+ activities in astrocytes lead to excessive activation of extrasynaptic NMDA receptors in neighboring neurons and increased network excitability.

(A) Left: representative traces of whole-cell patch clamp recording from CA1 pyramidal neurons in acute hippocampal slices prepared from wild type and Mecp2-/y mice showing slow inward currents (SICs). Arrows indicate all SICs. Inset: magnified view of the recording. Right: quantification of the SIC frequency in neurons from wild type and Mecp2-/y mice. (B) Quantification of the amplitude of SICs of neurons under control condition and in the presence of D-APV (left), an antagonist of NMDA receptors, or ifenprodil (right), a selective antagonist of NR2B-containing NMDA receptors. (C) Quantification of the SIC frequency in neurons adjacent to astrocytes with or without intracellular infusion of BAPTA. (D) Orthogonal (x-y view, y-z view, and x-z view) projections of z-scanning images showing immunofluorescence of NeuN and GFAP in regions around the hippocampus. The left image is from a Mecp2flox/y mouse injected with AAV-hSyn-mCherry-Cre, while the right image is from a Mecp2flox/y mouse injected with AAV-GFAP-mCherry-Cre. Note that all of the mCherry positive cells are NeuN positive in the left image, while all of the mCherry positive cells are GFAP positive in the right image. Scale bars = 50 μm. (E) Quantification of the frequency of the spontaneous Ca2+ oscillations in astrocytes from Mecp2flox/y mice injected with AAV-mCherry, AAV-hSyn-mCherry-Cre, or AAV-GFAP-mCherry-Cre. ***p<0.001 vs. AAV-mCherry. (F) Quantification of the SIC frequency from Mecp2flox/y mice injected with AAV-mCherry, AAV-hSyn-mCherry-Cre, or AAV-GFAP-mCherry-Cre. *p<0.05 vs. AAV-mCherry. (G) Representative patch clamp recordings in CA1 pyramidal neurons showing characteristic spontaneous epileptiform bursting activity in response to application of the GABAA-receptor antagonist bicuculline (3 μM) from wild-type (top) and Mecp2-/y (bottom) mice. (H) Quantification of the latency, frequency, duration and the amplitude of the epileptiform activity. The latency is defined as the time elapsed between Bic application and epileptiform activity onset. The bar graphs in this figure show the mean ±s.e.m. *p<0.05, **p<0.01, ***p<0.001.

https://doi.org/10.7554/eLife.33417.035

To determine the role of abnormal Ca2+ activities in the astrocytes in the increased frequency of SICs, membrane-impermeable BAPTA, along with a small dye that can pass through gap junctions, was injected into a single astrocyte in acute hippocampal slices prepared from the Mecp2 null mice. Thirty minutes after injection, many astrocytes became positive for the dye in the field of injection, indicating diffusion of the dye and BAPTA into astrocytes connected through gap junctions. Whole cell patch clamp recording on hippocampal pyramidal neurons within the field of these dye-labeled astrocytes (i.e. with the intracellular chelator BAPTA) revealed a significant reduction in the frequency of SIC (2.08 ± 0.51/20 min in Mecp2 null with BAPTA infusion into a single astrocyte vs. 5.33 ± 1.04/20 min in Mecp2 null with no treatment, n = 12 in each group, p=0.01; Figure 8C), suggesting that the abnormally high calcium activity in astrocytes is required for the excessive activation of eNMDARs on neighboring neurons.

To further determine the relationship between these two phenotypes, we examined calcium activities and activation of eNMDARs in astrocyte-specific and neuron-specific Mecp2 knockout mice. To generate cell type-specific Mecp2 knockout mice, we mated wild type males with Mecp2flox/flox females to generate Mecp2flox/y male pups, and injected AAV viruses (AAV8-hGFAP-mCherry-Cre, AAV8-Syn-mCherry-Cre, or AAV-mCherry) into the lateral ventricles of postnatal-day-1 old Mecp2flox/y male pups. Consistent with previous report (Kim et al., 2013), mCherry-positive cells showed widespread presence in the forebrain (including cortex and hippocampus), suggesting robust distribution of AAV in this method. Immunohistological analysis of sections from Mecp2flox/y male mice injected with AAV8-hGFAP-mCherry-Cre showed that 100% of mCherry-positive cells were positive for GFAP (right, Figure 8D, Table 3) and negative for MeCP2 (Figure 8—figure supplement 2), and that ~ 27% of GFAP-positive cells were positive for mCherry (Table 3), suggesting successful deletion of Mecp2 in about one third of the astrocytes in the forebrain. Similar analysis of sections from Mecp2flox/y male mice injected with AAV8-Syn-mCherry-Cre showed that 100% of mCherry-positive cells were positive for NeuN (left, Figure 8D, Table 3) and negative for MeCP2 (Figure 8—figure supplement 2), and that ~ 74% of NeuN-positive cells were positive for mCherry (Table 3), suggesting successful deletion of Mecp2 in about more than two thirds of the neurons in the forebrain. Further double staining with either anti-GFAP and anti-MeCP2 antibodies or anti-NeuN and anti-MeCP2 antibodies on those brain sections (Figure 8—figure supplement 2) revealed that, while almost all mCherry-positive cells were indeed negative for MeCP2 in mice receiving either AAV virus, there was a negligible number (2%) of NeuN-positive cells that were negative for MeCP2 in the hippocampus of the AAV-GFAP-mCherry-Cre injected mice, and a small number (7%) of GFAP-positive cells that were negative for MeCP2 in the hippocampus of the AAV-hSyn-mCherry-Cre injected mice (Table 4). Thus, mice with predominantly astrocyte-specific deletion of Mecp2 and predominantly neuron-specific deletion of Mecp2 were generated. Mecp2flox/y male mice injected with AAV-mCherry (no Cre recombinase) were used as wild type control for this series of experiments.

Table 3
Histological results of mice injected with AAVs (I)
https://doi.org/10.7554/eLife.33417.039
Mice injected with
AAV-GFAP-mCherry-Cre
Mice injected with
AAV-hSyn-mCherry-Cre
%GFAP + cells in all mCherry + cells1000
%NeuN + cells in all mCherry + cells0100
%mCherry + cells in all GFAP + cells28 ± 30
%mCherry + cells in all NeuN + cells074 ± 10
Table 4
Histological results of mice injected with AAVs (II)
https://doi.org/10.7554/eLife.33417.040
Mice injected with
AAV-GFAP-mCherry-Cre
Mice injected with
AAV-hSyn-mCherry-Cre
%MeCP2- cells in all mCherry + cells98 ± 0.897 ± 1.5
%MeCP2- cells in all GFAP + cells35 ± 37 ± 1.7
%MeCP2- cells in all NeuN + cells2 ± 0.175 ± 11

Abnormal spontaneous calcium activity in astrocytes (Figure 8E, n = 34 in AAV-mCherry group, n = 36 in AAV-Syn-mCherry-Cre group, and n = 47 in AAV-hGFAP-mCherry-Cre group; ANOVA: F = 14.11, p<0.0001) and excessive activation of extrasynaptic NMDA receptors in neighboring neurons (Figure 8F, n = 38 in AAV-mCherry group, n = 21 in AAV-Syn-mCherry-Cre group, and n = 30 in AAV-hGFAP-mCherry-Cre group; ANOVA: F = 4.23, p<0.018) were only observed in predominantly astrocyte-specific, but not in predominantly neuron-specific, Mecp2 knockout hippocampal slices. Results from this series of genetic manipulations strongly suggest that the abnormal calcium activities in astrocytes are caused by the loss of MeCP2 function in astrocytes, and can lead to excessive activation of extrasynaptic NMDA receptors on neighboring neurons (even when these neighboring neurons are wild type).

Beyond eNMDAR activation, several previous studies have suggested that increased astrocytic Ca2+ activity may lead to a hyperexcitable network (Gómez-Gonzalo et al., 2010; Kuchibhotla et al., 2009; Tian et al., 2005; Wetherington et al., 2008). Such a hypothesis is highly relevant to RTT, because 80–90% of RTT patients have seizures (Jian et al., 2007), and spontaneous seizures have also been reported in RTT mice (D'Cruz et al., 2010). As the first step to explore a role for abnormal astrocytic Ca2+ activity in leading to seizure in RTT, we adopted a widely used slice model of seizure (or epileptiform activity or network hyper-excitability) by adding the GABAA receptor antagonist biccucullin and increasing bath Ca2+ concentration (Borck and Jefferys, 1999; McLeod et al., 2013; Ratté et al., 2011; Tian et al., 2005).

In this model, whole cell patch clamp recording in CA1 pyramidal neurons readily detected epileptiform activity in slices from 5 to 8 week-old wild type and Mecp2-/y mice (Figure 8G). Comparing with wild type, epileptiform activity in Mecp2-/y neurons had shorter latency (4.5 ± 0.4 min in Mecp2-/y vs. 6.5 ± 0.6 min in wild type, Figure 8H), higher frequency (5.6 ± 0.4 Hz in Mecp2-/y vs. 3.5 ± 0.5 Hz in wild type, Figure 8H), longer duration (755 ± 30 ms in Mecp2-/y vs. 621 ± 39 ms in wild type, Figure 8H), and higher amplitude (364 ± 15 pA in Mecp2-/y vs. 279 ± 28 pA in wild type. Figure 8H). More importantly, intracellular infusion of membrane impermeable BAPTA into a single astrocyte significantly reduced epileptiform activity in neighboring neurons (latency: 4.5 ± 0.4 min in Mecp2-/y vs. 5.7 ± 0.4 min in Mecp2-/y with BAPTA infusion; frequency: 5.6 ± 0.4 Hz in Mecp2-/y vs. 3.8 ± 0.5 Hz in Mecp2-/y with BAPTA infusion; duration: 755 ± 30 ms in Mecp2-/y vs. 633 ± 31 ms in Mecp2-/y with BAPTA infusion. Figure 8H). 7–10 neurons from 3 to 4 mice were analyzed from each genotype/treatment in this series of experiments. Similar results were obtained recording field potentials under the same conditions described in Figure 8G–H (data not shown). Together, these data reveal a potential link between increased astrocytic Ca2+ activity and increased network excitability. Since such increased network excitability may underlie seizure-a specific RTT symptom, these data further highlight the functional significance of abnormal Ca2+ activity in RTT.

Discussion

Since the initial report that astrocytes express MeCP2 and that Mecp2 null astrocytes have non-cell autonomous influence on wild type neurons (Ballas et al., 2009), many gene expression changes (Maezawa et al., 2009; Okabe et al., 2012; Yasui et al., 2013) and a few cellular phenotypes have been identified in Mecp2 null astrocytes (Maezawa et al., 2009; Nectoux et al., 2012). However, it is not clear how these alterations contribute to RTT pathogenesis either by directly changing astrocyte functions or by indirectly changing neuronal functions. In addition, no major cell autonomous phenotype in astrocytes has been clearly defined, with its underlying mechanism and downstream consequences examined in detail.

A cell autonomous phenotype in astrocytes is inherently difficult to define in vivo, because astrocytes normally develop in the presence of neurons and intimately interact with neurons in the intact brain. To circumvent this issue, we first turned to human iPSC-differentiated astrocytes because they are differentiated and maintained in the absence of neurons. Using astrocytes differentiated from congenic pairs of human RTT patient specific iPSCs, we discovered that both the spontaneous and the pharmacologically evoked cytosolic calcium activities are abnormal in mutant RTT astrocytes. We then confirmed that similar phenotypes could be found in Mecp2 null mouse astrocytes in vitro, in situ, and in vivo. While significant difference exists across the astrocyte populations included in our study (species [human and mouse], developmental stage [embryonic and postnatal], and system [in vitro, in situ, and in vivo]), the core phenotype of abnormal astrocytic calcium activity remains consistent. As for the underlying mechanism, we provide evidence that the abnormal calcium activity is mediated by calcium overload in the endoplasmic reticulum (ER) caused by TRPC4-dependent abnormal store operated calcium entry (SOCE). Although our findings don’t directly reveal whether TRPC4 is a component of the SOCE channel, they do suggest elevated TRPC4 expression in mutant RTT astrocytes is upstream of the SOCE, ER, and cytosolic calcium phenotypes. As for the downstream consequences, we demonstrate that these abnormal calcium activities lead to excessive activation of the eNMDARs on neighboring neurons and increased network excitability. Together, our results not only identify a novel cell autonomous phenotype and its underlying mechanism in RTT astrocytes, but also reveal a functional link between the astrocyte phenotype and a novel neuronal phenotype in RTT models and network hyper-excitability that may underlie a significant RTT symptom. For future studies, we hypothesize a linear model in the following order: first, loss of MeCP2 expression in astrocytes leads to increased expression of TRPC4 in astrocytes; second, increased expression of TRPC4 in astrocytes leads to calcium overload in the ER of astrocytes; third, calcium overload in the ER of astrocytes leads to increased cytosolic calcium activity in the astrocytes; fourth, increased cytosolic calcium activity in astrocytes leads to excessive activation of eNMDARs in neighboring neurons and increased network excitability (summarized in Figure 9).

A schematic summary of the findings in our study.
https://doi.org/10.7554/eLife.33417.041

The abnormal calcium homeostasis in mutant RTT astrocytes manifests in multiple interconnected cellular events, including elevated SOCE, Ca2+ overload in the ER, elevated baseline cytosolic Ca2+ level, and abnormal spontaneous and evoked Ca2+ activities. Both molecular and pharmacological inhibition of the TRPC4 channel in mutant RTT astrocytes rescue the phenotypes of increased SOCE, ER Ca2+ overload and abnormal cytosolic Ca2+ activities. Conversely, overexpression of TRPC4 in wild type astrocytes can cause the mutant phenotypes. Together, these results strongly suggest that increased TRPC4 expression is upstream of the altered SOCE, and the subsequent ER and cytosol Ca2+ phenotypes. As most of our TRPC4-related data came from in vitro experiments, future in vivo studies are needed to ascertain the functional significance of increased TRPC4 expression in RTT disease progression. In addition, it is worth noting that other TRP channels such as TRPA1, TRPC, TRPV4 have been implicated in modulating Ca2+ dynamics in astrocytes (Ma et al., 2016; Molnár et al., 2016; Shibasaki et al., 2014; Shigetomi et al., 2013; Shigetomi et al., 2011). Yet, their expression remained unchanged in mutant human RTT astrocytes (data not shown). Of course, other deregulated genes and pathways due to the loss of MeCP2 function may also contribute to these phenotypes, providing interesting topics for future studies to further illustrate the central role of abnormal calcium homeostasis in astrocyte dysfunction in RTT. Furthermore, it is worth noting that abnormal calcium activities in astrocytes have been implicated in other neurological diseases, such as Alzheimer’s disease(Kuchibhotla et al., 2009), suggesting abnormal calcium homeostasis in astrocytes may be a common pathological event in neurological diseases. However, the molecular mechanisms underlying the abnormal calcium activities may be different in different diseases. While the P2Y1 receptor is shown to mediate the astrocytic hyperactivity in Alzheimer’s disease (Delekate et al., 2014), we have identified TRPC4 as a major contributor to the abnormal calcium homeostasis in RTT astrocytes. Since Ca2+ is a pleiotropic signal for a diverse range of cellular functions, abnormal Ca2+ homeostasis can lead to additional functional consequences in mutant RTT astrocytes. Future studies are therefore needed to define the full spectrum of cellular deficits in RTT astrocytes that are downstream of the altered Ca2+ signaling.

Up to now, a major challenge in understanding the contribution of astrocytes to RTT pathogenesis has been the missing link between astrocyte dysfunction and neuronal dysfunction and RTT symptoms. We showed that intracellular infusion of a membrane-impermeable calcium chelator, which abolishes calcium activities in astrocytes only, reduced the excessive activation of eNMDARs on neighboring neurons and rescued network hyper-excitability. Moreover, we provided genetic evidence that the loss of MeCP2 in astrocytes is both necessary and sufficient to cause the abnormal calcium activity in astrocytes and the excessive activation of eNMDARs in neighboring neurons. These results strongly suggest that this novel neuronal phenotype is dependent on a cell autonomous phenotype in the astrocytes. Thus, our findings provide a direct functional link between astrocyte dysfunction and neuronal dysfunction in RTT.

Although it remains debatable whether excessive activation of eNMDARs leads to increased neuronal synchrony and seizure in general (Angulo et al., 2004; Fellin et al., 2004; Tian et al., 2005), increased synchronous neuronal firing has recently been observed in the hippocampus of both male Mecp2-/y and female Mecp2+/- mice(Lu et al., 2016). Thus, the relationship between the abnormal astrocytic calcium activity, the excessive activation of eNMDARs, and increased neuronal synchrony is worth further studying in the context of RTT. While 80% of RTT patients have seizures (Jian et al., 2007), the cellular and molecular mechanisms underlying this symptom are not well understood. Even if the excessive activation of eNMDARs and/or the increased neuronal synchrony may not be directly linked to the seizure phenotype in RTT, our data linking the abnormal calcium activities in RTT astrocytes to increased network excitability still provide a new avenue for investigating this often difficult-to-manage symptom in RTT. At this moment, the exact molecular events linking abnormal calcium activities in astrocytes and altered neuronal and network properties remain elusive. One obvious candidate is glutamate, because previous studies have detected elevated glutamate level in primary cultures of astrocytes and microglia prepared from RTT mice (Maezawa and Jin, 2010; Okabe et al., 2012). Beyond glutamate, astrocytic calcium is known to modulate inhibitory synaptic efficacy by controlling GABA level through astrocytic GAT-3 (Shigetomi et al., 2011), and to modulate basal synaptic transmission through the release of purines (Panatier et al., 2011). Both mechanisms may help explain how abnormal calcium activities in RTT astrocytes cause increased network excitability. Therefore, more systematic approaches are needed in future studies to fully characterize the changes in factors present in RTT astrocyte conditioned medium (due to alteration in either release or uptake by astrocytes) and reveal how those changes underlie the non-cell autonomous influence on neurons.

Materials and methods

Cell lines

This study used congenic pairs of wild type and mutant Rett syndrome patient specific induced pluripotent stem cell lines carrying the R294X and the V247fs mutations. The cell lines were routinely tested to ensure that the cell lines were of human origin, were not contaminated with other cell lines, and were not contaminated with other microbial species including bacteria, fungi, or mycoplasma. Human identity testing was performed using Short Tandem Repeat (STR) testing. This test method is capable of providing a unique genetic fingerprint that is traceable back to a single human donor. In addition, STR is capable of detecting other contaminating cell types, either other human cell lines or other species, at a very low level. Isoenzyme analysis could also be performed to verify species of origin for cell lines. Testing was also routinely performed to test for microbial and fungal contamination using test methods that are typically used for testing clinical-grade cell lines. In addition, testing for mycoplasma contamination was be performed using either MycoAlert (Lonza), qPCR testing, or direct/indirect culture methods. These test methods provided a very high level of assurance that cell lines were not contaminated with other mammalian, bacterial, fungal, or mycoplasma cell lines.

In addition, for congenic pairs of iPSC lines derived from the same human RTT patient, the X chromosome inactivation status and the allelic expression of MECP2 were closely monitored to ensure correct genotype identification. These methods have been described in detail in our previous studies (Ananiev et al., 2011; Williams et al., 2014).

Human astroglial differentiation

Human astrocytes were differentiated from astroglial progenitors as previously described (Williams et al., 2014). Briefly, Terminal differentiation into astrocytes from astroglial progenitors was achieved by dissociating progenitor spheres into single cells with Accutase, and plating on coverslips pre-coated Poly-Lysine at a density of 2 × 105 cells per ml in astrocytes differentiation media: DMEM/F12 with 1% N2, 1 × NEAA, 1 × pen/strep and 2 µg/ml heparin (Sigma H3149), and CNTF (BioSensis). To facilitate cells attachment, 10% FBS (Gibco) was added. The next day, the media was replaced by fresh ADM without FBS. Differentiated progenitors were then fed every other day and used for experimentation 7 days after the start of terminal differentiation. To improve cell attachment, in some experiments, dissociated astroglial progenitor cells were plated on Matrigel-coated plates. The medium used for plating was DMEM/F12 with 1% N2, 1 × NEAA, 1 × pen/strep and 2 µg/ml heparin, supplemented with 10 ng/ml BMP4, LIF, CNTF. The medium was changed once on day 3. Astrocytes were used for experiments 7 days after the start of terminal differentiation. Stereological analysis was performed to determine the percentage of GFAP positive cells in each differentiation. Astrocyte cultures with higher than 85% GFAP positive cells were used for subsequent experiments.

Primary culture of mouse astrocytes

Mouse astrocyte cultures were prepared from postnatal day 0 (P0) male Mecp2 null and wild type littermate mice, as well as P0 female Mecp2+/- and Mecp2+/+ mice that carry an X chromosome-linked GFP transgene. After the brain of mice were removed and placed into ice-cold Hank's Balanced Salt Solution (HBSS), hippocampi were dissected and treated with 0.25% trypsin at 37°C for 30 min. Digestion was stopped by fetal bovine serum (FBS). Then the tissue was mechanically dissociated through a small fire-polished Pasteur pipette. After centrifuge, cells were re-suspended with DMEM, supplemented with penicillin/streptomycin and 10% FBS. Cells were plated into T-25 flasks and allowed to reach confluence. Contaminating neurons and microglia were then dislodged by shaking. The remaining adherent astrocytes were trypsinized, re-suspended in DMEM, and plated onto poly-D-lysine-coated coverslips at about 20,000 cells/ml. Cells were fed every 4 days by replacing the medium with fresh medium. The cells were used after 4 weeks in vitro. Stereological analysis was performed to determine the percentage of GFAP positive cells in each culture. Astrocyte cultures with higher than 90% GFAP positive cells were used for subsequent experiments.

In vitro calcium imaging

For the human iPSC derived astrocytes and some mouse primary astrocytes, intracellular Ca2+ was indicated by Fluo-4 or Rhod-2. Cells were bulk-loaded for 15 min at 37°C in artificial cerebrospinal fluid (aCSF) containing Fluo-4/AM (12.5 μg/ml), pluronic acid (0.05%), and DMSO (0.1%). After the Ca2+ indicator was loaded, cells were transferred to a chamber and Ca2+ imaging was performed. Mouse primary astrocytes from GCaMP6s mice were treated with 4-OH tamoxifen (0.5 μM, Sigma-Aldrich Inc.) for 24 hr to induce the expression of GCaMP6s. Cells were imaged using a Nikon A1 confocal microscope at room temperature. All image data were taken in the frame-scanning mode at 1 frame every 2 s (in the experiments using GCaMP6s as the Ca2+ indicator, 1 frame every 1 s). Fluo-4 or GCaMP6s was excited at 488 nm. The Ca2+ imaging data was analyzed using custom-written programs in Python. The metadata and the image data of the raw images were read with python-bioformats. Ca2+ signals were presented as relative fluorescence changes (ΔF/F0) from specified regions of interest (ROIs). ROIs were selected using an automated segmentation algorithm (Cai et al., 2016), while somata and processes were manually identified on the basis of morphology. For the traces with baseline drift, baseline correction was performed using a ‘rolling-ball’ algorithm. The peaks were detected using the algorithm developed by Matlab (findpeaks function). The frequency, amplitude and the full-width at half maximum (FWHM) were calculated and measured. Images with obvious motion were excluded for analysis. In experiments that examined spontaneous Ca2+ oscillations, the Ca2+ level was reported as F/F0 = Ft/F0. Calcium elevation events were detected with thresholds of 3 times of standard deviation of the baseline. The source code for Ca2+ events detection and automated segmentation is available in the supplementary files. Rhod-2 was used in the experiments involving viral expression of exogenous wild type MECP2, TRPC4, or shRNA against TRPC4. For the spontaneous astrocytic Ca2+ activity imaging, aCSF containing (in mM): 120 NaCl, 3 KCl, 15 HEPES, 1 MgCl2, 2 CaCl2, 20 Glucose (pH7.4) was used. For TG-induced ER Ca2+ release, aCSF with 0 mM Ca2+ and 2 mM EGTA was used, and TG was perfused to induce the astrocytic Ca2+ elevations. For SOCE, 2 mM Ca2+ containing aCSF was perfused after TG induced ER Ca2+ depletion. The spontaneous Ca2+ elevations from the soma and process of astrocytes were analyzed separately. For the analysis of Ca2+ imaging data in acute mouse brain slices, only Ca2+ elevations in soma and major process of astrocytes were included.

In vivo Ca2+ imaging

GCaMP6s mice (Ai96, 024106) were purchased from Jackson Laboratory. GCaMP6s and GCaMP6s;MeCP2flox/y mouse pups were injected with AAV-GFAP-Cre-mCherry (see Materials and methods/Adeno associated virus (AAV) injection into neonatal mice). Mice were used for Ca2+ imaging after they are 2 months old. A cranial window (3 mm x 3 mm) was installed over the M2 frontal cortex on the left hemisphere of mutant and control mice (center coordinates: 1.5 mm anterior to Bregma, 1.0 mm lateral to midline). Mice were anesthetized with isoflurane during surgery and recovered to full awakeness (>45 min post surgery) before imaging layer I astrocyte calcium signals under the 25x water immersion lens (NA = 1.05) of a multi-photon microscope (FV1000MPE, Olympus) (Cao et al., 2013Managò et al., 2016). The excitation wavelength of the two-photon laser was set at 900 nm, and the power of the laser emitted from the objective was set at 80 mW or lower. Image stacks had dimensions of 699 × 167 (width x length in µm) and the frame rate (frames per second) is 1.83. Three to five image recordings (655 s each) were made from each animal. The analysis of the in vivo Ca2+ imaging data is described in Materials and methods/In vitro Calcium imaging. The traces were low-pass filtered (cut-off frequency: 0.5 Hz) before analysis to remove noises.

Virus mediated expression of MECP2 and phenotypic analysis

Human or mouse astrocytes were infected with lentivirus encoding either GFP alone or GFP and wild type MECP2. 36 hr later, astrocytic Ca2+ imaging was performed. Appropriate MOI was used to achieve infection rates higher than 90% in human astrocytes and ~50% in mouse astrocytes based on the percentage of GFP positive cells.

Virus mediated expression of TRPC4

Human astrocytes were infected with lentivirus encoding either GFP alone or GFP and TRPC4. 10 days later, astrocytic Ca2+ imaging was performed. The infection rate was higher than 80%, according to the percentage of GFP positive cells.

Immunofluorescence

For immunohistological evaluation on brain sections, mice were killed by barbiturate overdose and perfused transcardially with phosphate-buffered saline (PBS), followed by 4% paraformaldehyde (PFA, in PBS). Brains were removed and post-fixed overnight. Then the brains were cryoprotected in buffered 30% sucrose (wt/vol) for at least 2 d. 25 μm coronal frozen sections were prepared using a cryostat microtome. For immunocytochemistry, cultures were fixed on ice with 4% PFA for 30 min. Immunostaining was performed as previously described (Li et al., 2011). Briefly, cultures were fixed on ice with 4% PFA for 30 min. After washed three times with PBS, cultures or sections were permeablized with 1% Triton X100 (Sigma, in PBS) for 30 min, blocked with 3% normal donkey serum and 0.25% Triton X100 in PBS for 90 min at room temperature (RT), and then incubated with primary antibodies overnight at 4°C. Then the cultures were incubated with the corresponding secondary antibodies for 60 min at RT, and washed five times with PBS at RT. Primary antibody dilutions were as follows: anti-GFAP (Millipore MAB3402 RRID:AB_94844, 1:500; and Dako, Z0334 RRID:AB_10013382, 1:500), anti-MeCP2 (Abcam ab50005 RRID:AB_881466, 1:500 and ab2828 RRID:AB_2143853, 1:500; and Cell Signaling Technology Cat# 3456S, RRID:AB_10828482, 1:500), anti-NeuN (Millipore MAB377 RRID:AB_2298772, 1:100), anti-S100β (Abcam Cat# ab868, RRID:AB_306716, 1:500) anti-TRPC4 (Alomone labs ACC-018 RRID:AB_2040239, 1:100; and Abcam ab84813 RRID:AB_1860087, 1:500). Secondary antibody dilutions were as follows: Alexa Fluor 568 Donkey-anti-Rabbit antibody (Thermo Fisher Scientific A10042 RRID:AB_2534017, 1:500), Alexa Fluor 488 Donkey-anti-mouse antibody (Thermo Fisher Scientific A21202 RRID:AB_141607, 1:500), and Alexa Fluor 647 Donkey-anti-mouse antibody (Thermo Fisher Scientific A-31571 RRID:AB_162542, 1:500). DAPI was used at 3 nM for counterstaining. Images were taken using a Nikon A1 confocal microscope.

Western blot analysis

Astrocytes were collected into an Eppendorf tube and directly lysed by adding 1X LDS buffer (Life technology). Samples were sonicated to facilitate cell lysis and boiled for 5 min before loading into 10% SDS-PAGE gel. Proteins were transferred onto nitrocellulose membrane (Whatman) using a semi-dry transfer system from BioRad. Membrane was first blocked with 5% milk solution for 1 hr and then incubated with anti-TRPC4 (Alomone labs, ACC-018 RRID:AB_2040239, 1:100) diluted in 3% BSA solution at 4°C overnight. After incubating with infrared dye-conjugated secondary antibody (Thermo Fisher Scientific Cat# 35518 RRID:AB_614942; # SA5-35571 RRID:AB_2556775;1:10,000) for 1 hr at room temperature, membrane was scanned in an Odyssey infrared imaging system.

RNA-Sequencing analysis

Total RNA was extracted from R294X-WT and R294X-MT iPSC-derived astrocytes using Trizol (life technology) and cleaned up with on-column DNase treatment (Qiagen). 150 ng total RNA was used to prepare sequencing library in a Mondrian SP +Workstation according to manufacturer’s instructions (Nugen Encore SP +Complete). 100 bp single-end reads sequencing for each library was performed in an Illumina Hi-Seq 2000. Reads were mapped to the human genome build hg19 using Tophat (2.0.8) and differential gene expression analysis was done using Cuffdiff (2.0.2).

Microarray analysis

Purified RNA was sent to Affymetrix for microarray analysis. HTA-2.0 chip was used in this project. Sample processing, dye labeling, hybridization, and scanning were performed by Affymetrix. Each sample was hybridized to four chips. Spike-in RNA was supplemented to samples and used as a normalization control.

Chromatin immunoprecipitation (ChIP) and quantitative PCR (qPCR) to determine MeCP2 occupancy at the Trpc4 gene promoter

Primary astrocyte culture was prepared from the forebrain of the Mecp2-Flag mice (Li et al., 2011) and control mice. ChIP was performed as previously described (Li et al., 2011). qPCR was done using a StepOne plus (Life technology). The following primers were used in the qPCR. TRPC4-P-1 forward: 5′-GCAGAGTGAGCCTGAGTCTA-3′. TRPC4-P-1 reverse: 5′-CGTGATCTCAAGACCAAGGG −3′. TRPC4-P-2 forward: 5′-TAGTATGGTTGGAGCAGGGC-3′. TRPC4-P-2 reverse: 5′-AGCTAAGTGGTGGTCAGGAC-3′. TRPC4-P-3 forward: 5′-CACCTTGGGAACGCAACTTT-3′. TRPC4-P-3 reverse: 5′-AAAACCCGCACGAAACCAG-3′. TRPC4-P-4 forward: 5′-CCCCATCGGAACTGACCA-3′. TRPC4-P-4 reverse: 5′-AGTATCCCAGATGTGAGGCC-3′.

TRPC4 knockdown experiment

The hTRPC4 shRNA sequence used in our study is identical to that used in a previous study (Zagranichnaya et al., 2005). The shRNA sequence was cloned into a lentiviral construct, pLentilox3.7. Lentiviral particles were prepared as previously described (Ananiev et al., 2011). Mutant human astrocytes were infected with lentivirus encoding either GFP alone or GFP and shTRPC4. 36 hr later, astrocytic Ca2+ imaging was performed. Appropriate MOI was used to achieve an infection rate of higher than 90% based on the percentage of GFP positive cells.

Adeno-associated virus (AAV) injection into neonatal mice

Experiments described below were performed according to protocols approved by the IACUC at University of Wisconsin-Madison. Neonatal Mecp2flox/y mice were anesthesized by hypothermia before the injection. The AAV solution was injected into the lateral ventricles using a 5 µl Hamilton syringe. The injection site was located three-fifths in the line defined between the lambda intersection of the skull and each eye. The needle was held perpendicular to the skull surface and inserted to a depth of ~3 mm. Then 1 µl of viral solution was injected into the ventricle. After injection, mice were transferred to a heated pad with perfused water at 38°C until they regained normal color and resumed movement. Three types of AAV were injected in this series of experiments: AAV8-hGFAP-mCherry-Cre (expression of fluorescence marker mCherry and the Cre recombinase under the astrocyte-specific hGFAP promoter), AAV8-Syn-mCherry-Cre (expression of mCherry and Cre under the neuron-specific human synapsin1 promoter), and AAV-mCherry. Stereological studies were performed to examine the infection rate of the AAV and the cell type specificity of mCherry expression and Cre-mediated deletion of MeCP2. Brain sections were immunostained with either anti-GFAP or anti-NeuN antibody and counterstained with DAPI; or combination of anti-GFAP and anti-MeCP2 antibodies; or combinations of anti-NeuN and anti-MeCP2 antibodies. Sections were viewed with a Zeiss photomicroscope at 20X. Stereology was performed with the StereoInvestigator software (MicroBrightField). The optimal number of counting sites was determined empirically based upon a Scheaffer CE value less than 0.30.

Acute brain slices preparation and Ca2+ imaging

Male mice at postnatal 2–3 weeks were anesthetized and coronal brain slices (400 μm) were prepared in ice-cold modified artificial cerebrospinal fluid (aCSF) (in mM: 124 NaCl, 2.5 KCl, 0.5 CaCl2, 5 MgCl2, 1.25 NaH2PO4, 26 NaHCO3, and 15 glucose) bubbled with 95%O2/5%CO2. Then the slices were incubated in normal aCSF (in mM: 124 NaCl, 2.5 KCl, 2.5 CaCl2, 1.2 MgCl2, 1.25 NaH2PO4, 26 NaHCO3, and 15 glucose) at room temperature for at least 1 hr. After incubation, the slices were bulk-loaded for 1 hr at 37°C in aCSF containing Fluo-4/AM (12.5 μg/ml), pluronic acid (0.05%), and DMSO (0.1%) saturated with 95%O2/5%CO2. Sulforhodamine 101 (SR101) was co-loaded as an astrocyte marker. Then slices were transferred to a chamber perfused with 95%O2/5%CO2 saturated aCSF. Slices were imaged using Nikon A1 confocal microscope. Fluo-4 and SR101 were excited at 488 nm and 561 nm respectively. The data acquisition protocol was the same as that used in cultured cells. The person performing the Ca2+ imaging experiments and the subsequent data analysis was blind to the genotypes of mice used in these experiments.

Electrophysiological recording

Slices were transferred to a recording chamber perfused with 95%O2/5%CO2 saturated aCSF. To record the extrasynaptic NMDA receptor current, hippocampal CA1 neurons were clamped at − 30 mV to relieve Mg2+ block. We adjusted the equilibrium potential of Cl to −30 mV by controlling the [Cl] in the pipette solution. QX-314 was added to block Na+ channels. The intracellular solution was (in mM): 95 Cs-methanesulfonate, 34 CsCl, 10 HEPES, 10 QX-314, 2.5 MgCl2, 1 CaCl2, 5 NaCl, and 10 EGTA, pH 7.2. To selectively inhibit the astrocytic Ca2+ activity, membrane-impermeable BAPTA (10 mM) was injected into the CA1 stratum radiatum astrocytes by patch clamp. To avoid a leakage of BAPTA from the pipette before the formation of whole cell patch clamp, the BAPTA solution was backfilled with the standard intracellular solution. The distinctive electrophysiological characteristics such as a linear I/V relationship was obtained to confirm the type of cells being patched.

To record store operated Ca2+ influx in astrocytes, we transferred the coverslip into a chamber fixed in an Olympus BX51WI upright microscope. The recording chamber was perfused with aCSF containing (in mM): 120 NaCl, 3 CsCl, 15 HEPES, 1 MgCl2, 2 CaCl2, 20 Glucose (pH7.4). The pipette solution contained (in mM): 130 Cs-gluconate, 10 CsCl, 10 HEPES, 1 EGTA, 4 Mg-ATP and 0.5 Li-GTP. After whole cell patch clamp formed, the SOCE currents (ISOCE) were induced by switching the perfusing solution to aCSF containing 1 μM TG. ML204 (10 μM) was used to isolate ML204 sensitive currents. The membrane capacitance was also obtained to normalize the currents. To get the I-V curve of ISOCE, a voltage ramp from 100 mV to −100 mV (for 200 ms) was applied. The raw data was acquired with a Multiclamp 700B amplifier and pClamp10.2 software (Axon Instruments, Sunnyvale, CA). All data was analyzed using Clampfit10.2. The current density at −70 mV was quantified.

For the in vitro epilepsy model, male mice at postnatal 5–8 weeks were used. The slices were transferred to a recording chamber perfused with 95%O2/5%CO2 saturated aCSF containing 5 mM KCl. Epileptiform activity was induced by switching perfused solution to aCSF containing the GABAA receptor antagonist bicuculline (Bic, 3 μM; Sigma) (McLeod et al. (2013). Reduced seizure threshold and altered network oscillatory properties in a mouse model of Rett syndrome. 231, 195–205.). Epileptiform events were detected with Clampfit 10.2 (Axon Instruments) and verified visually. The intracellular solution was (in mM): 140 K-gluconate, 7.5 KCl, 10 Hepes-K, 0.5 EGTA-K, 4 Mg-ATP and Li-GTP. The currents caused by unclamped action potentials has been truncated (Fellin et al., 2006).

Statistics analysis

All data were analyzed using the SigmaPlot 13.0 (Systat Software, Inc RRID:SCR_003210). Student's t-tests or Mann–Whitney U tests were used to make comparisons between two groups unless indicated otherwise. One-way ANOVA followed by post-hoc Holm-Sidak test was performed when there are three or more groups. Ca2+ imaging data was analyzed by two-way ANOVA. A list of the statistical methods used in each figure can be found in Table 5. Average data are shown as the mean ±SEM. p values < 0.05 were considered statistically significant. No statistical methods were used to pre-determine sample sizes. Data were collected from randomly selected fields within images. More than three independent experiments/biological replicates (for human astrocyte data set, each independent experiment was defined as one independent differentiation; for mouse astrocyte in vitro data set, each independent experiment was defined as one isolation culture from one mouse; for mouse in situ and in vivo data set, both number of cells and number of mice were described) were included in each data set.

Table 5
List of statistical methods
https://doi.org/10.7554/eLife.33417.042
FigureStatistical methods
Figure 1CMann-Whitney U Test
Figure 1EMann-Whitney U Test
Figure 2CTwo Tailed Unpaired t-Test
Mann-Whitney U Test
Mann-Whitney U Test
Figure 2EMann-Whitney U Test
Figure 3CTwo Tailed Unpaired t-Test
Mann-Whitney U Test
Figure 3EMann-Whitney U Test
Figure 4CMann-Whitney U Test
Figure 4EMann-Whitney U Test
Figure 5CTwo Tailed Unpaired t-Test
Mann-Whitney U Test
Figure 6AMann-Whitney U Test
Figure 6CMann-Whitney U Test
Figure 6ETwo Tailed Unpaired t-Test
Figure 6FTwo Tailed Unpaired t-Test
Figure 6ITwo Tailed Unpaired t-Test
Figure 6KTwo Tailed Unpaired t-Test
Figure 6MMann-Whitney U Test
Figure 7CMann-Whitney U Test
Figure 7HMann-Whitney U Test
Figure 7LTwo Tailed Unpaired t-Test
Mann-Whitney U Test
Mann-Whitney U Test
Figure 8ATwo Tailed Unpaired t-Test
Figure 8BMann-Whitney U Test
Figure 8CMann-Whitney U Test
Figure 8EOne-way ANOVA followed by post-hoc Holm-Sidak test
Figure 8FOne-way ANOVA followed by post-hoc Holm-Sidak test
Figure 8HTwo Tailed Unpaired t-Test

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49

Decision letter

  1. Beth Stevens
    Reviewing Editor; Boston Children's Hospital, Harvard Medical School, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for choosing to send your work, "Mechanism and Consequence of Abnormal Calcium Homeostasis in Rett Syndrome Astrocytes", for consideration at eLife. Your initial submission has been assessed by a Senior Editor in consultation with a member of the Board of Reviewing Editors. Although the work is of interest, we regret to inform you that the findings at this stage are too preliminary for further consideration at eLife.

Elucidating how astrocytes contribute to the pathobiology of Rett Syndrome is interesting and important question. The authors provide evidence that cell autonomous changes in astrocyte calcium signaling affect neurons in Rett model mice. However, there were several shared issues raised by reviewers that diminish enthusiasm for this paper. Specifically, more direct evidence that calcium signaling is altered in mecp2- null astrocytes in vivo is needed to strengthen their conclusions. In addition, the in vitro calcium imaging data require further controls and analyses to strengthen their conclusions. The use of Rett patient-derived IPSCs to study astrocyte dysfunction and the demonstration that similar effects are observed in mouse null astrocytes are strength; however more characterization of astrocytes and clarification of the protocols used to generate human astrocytes are needed. Reviewers raised additional concerns regarding the quality and consistency of the data that need to be addressed.

Reviewer #1:

The paper makes a compelling argument that cell autonomous changes in astrocyte calcium signaling affects neurons in RETT model mice. Key aspects of the calcium signaling studies are supported by work in human astrocyte cultures.

Overall, this is a valuable and useful study and adds to the growing body of work that shows astrocyte calcium signals are altered in neurological disease. However, several points should be addressed.

1) In most of the figures, the photomicrographs of images are rather poor and hard to make out. Replace them with better examples. If needed, adjust the gray levels so the reader can see the images and determine if what is stated in the text is supported by the data shown.

2) There are no data for evoked calcium signals in the paper. The authors have used GPCR agonist-mediated calcium signals. They should clarify this in the paper and explicitly state that evoked signals due to action potential firing were not studied.

3) I think the data for and R294X should be shown separately in a Table. I don't understand the logic of averaging data from two separate genotypes. Alternatively, they could show the data from just one genotype. Averaging across two distinct disease associated genotypes seems odd.

4) Why is it that astrocytes in culture (human and mouse) have only a small population of cells that show spontaneous calcium elevations? What is this proportion in the mouse slice studies? Why do only a small population of cells show calcium responses in WT and MT? Usually, all astrocytes in vivo show calcium signals and so the low proportion is odd. Please explain or discuss. Does this reflect an artifact of cell culture?

5) The RNAseq data are mentioned but not shown. These data seemingly exist and should be reported in the main text and figures. How else can the reader assess the quality of the data set and believe the findings with TRPC4? The RNAseq data deserve to be reported in full.

6) What is the molecular basis of the ML204-insensitive current in astrocytes? Please discuss or explain with experiments.

7). The data with hGFAP-Cre AAVs are not convincing as a way to target astrocytes selectively, especially in the hippocampus. My suggestion is to drop these data and use the space that is gained to report the RNAseq data in full. I don' think the hGFAP-Cre data add anything useful to the paper, because the logic for why this should be astrocyte specific is not clear.

Overall, a potentially valuable paper that needs another round of hard work to tighten up some aspects.

Reviewer #2:

Summary:

The authors of this manuscript attempt to prove that MeCP2 mutant astrocytes from both iPSC and mouse models display altered calcium homeostasis by a mechanism of high internal calcium store concentration and increased expression of TRPC4 channels. This leads to increased spontaneous calcium activity, increased amplitude to pharmacologically induced calcium transients, and a proposed aberrant activation of NMDA receptors. While this manuscript presents an interesting first attempt at probing calcium homeostasis in MeCP2 mutant astrocytes, there are significant and fundamental errors in both experiments and the approaches used that undermine the interpretation of results.

Essential revisions:

1) Throughout the paper, the authors claim to show in vivo evidence that supports their conclusions. There is not a single in vivo experiment in this entire data set. They consistently refer to their acute slice experiments as in vivo work: this is incorrect. Acute slice work is generally referred to as in situ, as it does not wholly replicate the in vivo environment. This is critically important when it comes to astrocytes, which are extremely sensitive to osmotic and ion changes that can occur during the process of slicing and can alter astrocyte activity.

2) There is no characterization of the iPSC cultures to determine if they are in fact astrocytes. A reference is given to a previous paper, but no characterization of the cultures used in the experiments conducted in this manuscript is provided. It is critical to include experiments where the cultures used for calcium imaging are fixed and stained for astrocyte markers.

3) Calcium imaging data: Overall, the calcium imaging data from both iPSCs, primary cultures and in situ slice experiments is not convincing. Fluo-4 is a single wavelength dye, and while normalizing to baseline can reduce issues concerning concentrations of indicator, this indicator can only be used for relative measures of calcium amplitudes and requires careful loading controls that do not appear to be presented. Use of a ratiometric dye for experiments would have been more convincing for the changes that are described. Additionally, in the entire manuscript the data presented is from somatic recordings of astrocyte transients. While for the iPSCs work this may be understandable, for the primary culture and slice work there should have been an effort to examine calcium transients in the enlarged cell body or processes of astrocytes, as the somatic activity of an astrocyte is incredibly low compared to the activity in processes (see Di Castro et al., 2011) and represent a small fraction of the total calcium activity. As the authors used a genetically encoded calcium indicator in some experiments, why not in all experiments? This would have been more convincing.

4) The frame rate for calcium imaging is too slow for an accurate recording of the spontaneous activity and is likely underestimating the frequency of spontaneous events.

5) In general, in the majority of the experiments, the n's presented for results are low for what this reviewer would expect from culture experiments where the dish contains many thousands of cells.

6) The choice of representative images, such as the immunostaining in Figure 5A, is inconsistent with the data presented. The authors use a lentivirus to reintroduce Mecp2 into their null cell populations, along with GFP. The MeCp2-null cells depicted by the GFP images seem to be more numerous and have different morphology. Furthermore, there seem to be some red signal indicating some expression of MeCP2 in these cells. This calls into question the potential health and fidelity of the cells, or the choice of images. A similar issue is found in Figure 5H: a representative staining has several cells from MeCP2-null neonates selected that express GFP and Rhod-2. However, a close examination of these areas does not show any Rhod-2, which could explain why the authors do not report any Ca transients in these regions. Again, the choice of representative images is called into question, since the representative traces depict absolutely no traces during a ~10 min period, yet their data show the frequency in this population to be 1.33 / 0.1 min-1. In general, throughout the paper the representative data seems inconsistent with the reported quantification.

7) The authors of the paper make an assumption concerning astrocyte calcium signaling that there have been no changes to the signaling pathways contributing to the generation of spontaneous signaling aside from TRPC4. This is highly unlikely given the nature of MeCP2. The major release pathway from the ER involves activation of IP3 receptors. While TRPC4 does appear altered, it may also be the case that other components of the ER release pathway may be altered as well and be the root cause of the increased spontaneous signals. A closer examination of other components of ER associated calcium signaling should be conducted. Additionally, staining for TRPC4 to show it is even expressed by astrocytes in vivo is needed to validate the model.

Reviewer #3:

The manuscript by Qiping Dong et al., provides in vitro and ex vivo data that astrocytes from Rett Syndrome patients and mouse models of the human disorder have cell-autonomous increases in cytosolic calcium oscillations. Furthermore, data suggest that increased astrocytic calcium responses results in increased activation of extrasynaptic NMDA receptors on neighboring neurons. The authors identify that increased cytosolic calcium in astrocytes is primarily derived from the endoplasmic reticulum through increased store operated calcium entry (SOCE). They also identify a potential mechanism for increased SOCE in mutant and null astroctyes via increased levels ofTRPC4. The authors postulate that this increased astrocytic calcium leads to seizures in Rett patients via increased activation of extrasynaptic NMDA receptors.

Identifying how astrocytes play a role in Rett Syndrome is interesting, timely, and therapeutically relevant. The use of Rett patient-derived IPSCs to study astrocyte dysfunction is innovative and impactful and the demonstration that similar effects are observed in mouse null astrocytes is rigorous. However, there are some major concerns regarding the quality and consistency of the data as well as the relevance for effects in vivo. See points below.

1) All data are in vitro (cultured cells) or ex vivo (slice) with no in vivo evidence that astrocytic calcium dynamics and subsequent extrasynaptic NMDA transmission are dysregulated in Rett Syndrome. While it is appreciated that not everyone is capable of imaging calcium oscillations in vivo, one relatively straight forward experiment that would start to address this point is to assess seizure susceptibility in vivo (vs. ex vivo) following astrocyte-specific knockdown or ablation of Mecp2.

2) There are multiple inconsistencies in the data reporting changes in calcium in mutant or null astrocytes versus wild-type. Often calcium dynamics are very different across figures for the same genetic or pharmacological manipulation. For example, the calcium dynamics reported in Figure 12C,D show different levels of mutant calcium oscillation and frequency than is reported in Figure 2B,C. These discrepancies need to be addressed.

3) The authors fail to clearly define their N numbers in their figures or results. It is not clear from reading the manuscript whether the N's provided refer to number of cells, number of experiments, etc. Given that the N's are relatively high, one would assume this represents cells. However, it is unclear how many independent experiments these N's represent. Data should be assessed from at least three independent experiments and the authors should sample similar numbers of cells across experiments. This information should be clearly outlined. Furthermore, while the authors list N's in the text, the authors should list this in each figure legend for clarity. The authors also need to list the stats and p-values for each experiment in the figure legend. This is inconsistent across the manuscript.

4) The authors conclude that evoked calcium activities are disrupted following either ATP or glutamate stimulation. The mutant or null astrocytes already have higher amplitude responses spontaneously (e.g. Figure 1E). If you compare the amplitudes of spontaneous responses to evoked responses, it appears that the actual magnitude of increase following stimulation is not different between wild-type and null or mutants astrocytes (e.g. Figure 1-amplitude increases from about 2 (spontaneous) to about 3.5 (+ATP) for WT and from about 3 (spontaneous) to about 4.5 (evoked) in MT). Therefore, the conclusion that evoked responses are different is not accurate and should be clarified.

5) The figures are generally very difficult to follow and could use some reorganization. For example, the authors often go back and forth between human and mouse within the same figure without clearly identifying this within the figure. In addition, a number of figures could be either combined or included as supplemental material. For example, Figure 8, Figure 9, Figure 10 and Figure 11 could all be represented along with data from Figure 7 or included as a supplemental file.

6) The images presented throughout the manuscript often lack many general cell markers that would benefit interpretation of the data. In some cases, the figure lacks any images of cultures (Figure 2). Most images do not include DAPI to label all cells and no images show staining for astrocyte-specific markers. Higher magnification images should also be added to the current low magnification images. In addition, the authors list quantification in the text but these data should be included in the figure to increase clarity.

7) The authors need to validate sufficient knockdown of genes and reduced protein in astrocytes following shRNAs. These data can be included as supplemental.

8) In Figure 13 and Figure 14, there is only quantification of mCherry labeling not loss of Mecp2 expression. The authors must quantify presence or absence of Mecp2 protein in mCherry positive and negative astrocytes and neurons. This is particularly important given that detecting Mecp2 in vivo in non-neuronal cells is notoriously difficult and typically requires amplification (Ballas et al., 2011). From the images, it is not clear that the antibody recognizes Mecp2 in uninfected astrocytes, which is important to show in order to validate loss of Mecp2 in infected astrocytes. This figure could also benefit from higher magnification images. Lastly, the authors suggest that they are creating cell-specific knockout mice. This is exaggerated as only about 27% of cells are infected.

9) The manuscript emphasizes that their heightened extrasynaptic NMDA receptor activation has never before been reported in any RTT models. This is inaccurate. There is at least one recent paper reporting increased activation of eNMDARs in RTT (Lo et al., 2016). While it is appreciated that this effect has never been attributed to astrocytes, the text should be modified to include these published findings.

10) The model provided in Figure 15 is not detailed enough to add any meaningful contribution to the manuscript as a whole and could be excluded.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "Mechanism and Consequence of Abnormal Calcium Homeostasis in Rett Syndrome Astrocytes" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

The authors have made significant improvements to their manuscript from the original submission. The authors have addressed several of the key concerns raised in the first submission. The revised manuscript is better organized, clear in its presentation of the data and the model systems used and draws well-grounded conclusions from the data. However, there are a number of additional revisions and comments that will need to be addressed before publications as summarized below.

Essential revisions:

1) Most data supporting the TRPC4 mechanism are in vitro. The only data showing some evidence of increased TRPC4 function in Mecp2 null mice are immunostaining in Figure S10. More data demonstrating TRPC4 expression changes and function in vivo/ex vivo would strengthen the study. For example, western blots for TRPC4 from acutely isolated astrocytes (mouse model) in addition cultured astrocytes would be helpful.

At minimum, please include in brief discussion emphasizing the need for future studies on the topic. This would help the reader and thoughtfully point out the limitations of the current work.

2) The authors' explanation to reviewers regarding the large variability in N's is not sufficient. In many cases the N's vary by 100s of cells. For example, Figure 1C the N=526 cells for mutant cultures and N=811 for wild-type cultures. As one reviewer previously pointed out, this could greatly affect the statistical analyses. Further, it is unclear how the authors chose a particular N prior to doing statistics and it remains unclear in the manuscript how many independent experiments each data set represents.

3) While the authors make more attempt to validate their cell-specific genetic manipulations of Mecp2, these data are still not quantified sufficiently. The data in Supplementary file 3 and Figure 8—figure supplement 2 only shows quantification of% cells with mCherry, GFAP, and NeuN expression. The% of NeuN+ and GFAP+ cells +/- MECP2 immunoreactivity after AAV injection should also be included here. Given that the goal is to achieve cell-type specific manipulation, these data are important.

4) Many immunofluorescence images are still generally low quality. For example, in Figure 6—figure supplement 2, the background staining of TRPC4 is quite a bit higher in MECP2 null animals vs. WT. The non-specific background staining should be the same in these images to ensure a more accurate comparison. This and other figures would also benefit from some higher magnification zoom-ins in addition to the low magnification images. There is also immunostaining that is very difficult to visualize in several figures. One clear example is Figure 8—figure supplement 2 (MECP2 and NeuN immunoreactivity).

https://doi.org/10.7554/eLife.33417.048

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #1:

The paper makes a compelling argument that cell autonomous changes in astrocyte calcium signaling affects neurons in RETT model mice. Key aspects of the calcium signaling studies are supported by work in human astrocyte cultures.

Overall, this is a valuable and useful study and adds to the growing body of work that shows astrocyte calcium signals are altered in neurological disease. However, several points should be addressed.

We are grateful the reviewer thinks our study is valuable and makes a compelling argument.

1) In most of the figures, the photomicrographs of images are rather poor and hard to make out. Replace them with better examples. If needed, adjust the gray levels so the reader can see the images and determine if what is stated in the text is supported by the data shown.

We have replaced all images with better ones.

2) There are no data for evoked calcium signals in the paper. The authors have used GPCR agonist-mediated calcium signals. They should clarify this in the paper and explicitly state that evoked signals due to action potential firing were not studied.

We have used the terms “pharmacologically evoked” and “agonist evoked” to specify ATP- and glutamate-evoked calcium elevation and distinguish them from those evoked by neuronal activity under physiological conditions. Furthermore, those assays involving evoked activities were only meant to be additional measure of altered calcium homeostasis, but not to imply any underlying mechanism.

3) I think the data for V247f and R294X should be shown separately in a Table. I don't understand the logic of averaging data from two separate genotypes. Alternatively, they could show the data from just one genotype. Averaging across two distinct disease associated genotypes seems odd.

We combined data from the two genotypes, because both appears to be null at the protein level. Nonetheless, we have taken the reviewer’s suggestion by presenting the R294X (a common RTT mutation) genotype in the main text and mentioned the R247fs genotype in supplementary data in our revised manuscript.

4) Why is it that astrocytes in culture (human and mouse) have only a small population of cells that show spontaneous calcium elevations? What is this proportion in the mouse slice studies? Why do only a small population of cells show calcium responses in WT and MT? Usually, all astrocytes in vivo show calcium signals and so the low proportion is odd. Please explain or discuss. Does this reflect an artifact of cell culture?

Our calcium imaging results represent a snapshot sampling of the cell population during a relatively short time window (10 minutes). The higher percentage of cells showing spontaneous calcium activity in MT means a higher chance than in WT. If the imaging period is longer, the percentage will increase in both genotypes, yet the percentage will still be higher in the MT.

5) The RNAseq data are mentioned but not shown. These data seemingly exist and should be reported in the main text and figures. How else can the reader assess the quality of the data set and believe the findings with TRPC4? The RNAseq data deserve to be reported in full.

We are committed to deposit our RNA-seq data in public database after our manuscript is accepted for publication. Since the whole RNA-seq data set is not the main focus of our study, we have decided to describe its role as the discovery approach leading our attention to TRPC4. In order to bolster the validity of our conclusion of increased TRPC4 expression, we added data from multiple independent experimental approaches, including microarray, qRT-PCR, Western blot, and immunostaining on brain sections. Results from all of those experiments consistently showed elevated TRPC4 expression in MeCP2 mutant astrocytes.

6) What is the molecular basis of the ML204-insensitive current in astrocytes? Please discuss or explain with experiments.

Depleting ER calcium leads to the activation of store operated calcium entry. In astrocytes, this calcium entry may be mediated by TRP channels and calcium release activated calcium channels (Oral1). ML204 is a newly discovered inhibitor that only blocks TRPC4 containing channels but not Oral1 channels mediated currents. Thus, we used the ML204-sensitive current as an indirect measurement of TRPC4 mediated current. An increase of ML204-sensitive current (i.e. TRPC4 current) is consistent with our findings of increased TRPC4 expression in MT astrocytes.

7) The data with hGFAP-Cre AAVs are not convincing as a way to target astrocytes selectively, especially in the hippocampus. My suggestion is to drop these data and use the space that is gained to report the RNAseq data in full. I don' think the hGFAP-Cre data add anything useful to the paper, because the logic for why this should be astrocyte specific is not clear.

In our study, there are two sets of experiments involving hGFAP-Cre AAV injection. In the first set, our analysis of phenotypes (both eNMDAR activation and calcium dynamics) was performed in acute hippocampal slices prepared from 2-3-week-old mice (Figure 8). Even if neural stem cells in the hippocampus infected with AAV express hGFAP-cre, there won't be enough time (2 weeks after injection of the AAV into postnatal day 0 pups) for those neural stem cells to differentiate into neurons and complicate our experimental results. Indeed, when we further sectioned and stained those slice, 100% of the mCherry positive cells in those sections were GFAP positive astrocytes (Figure 8 and Table 3).

The second set is our imaging of spontaneous calcium activity in live mice. Although the lapse of time after AAV administration was longer, our analysis was limited to the superficial layer (layer 1) of the frontal cortex, in which region our immunological analysis again confirmed the specificity of the AAV-cre mediated MeCP2 deletion was limited to astrocytes.

Overall, a potentially valuable paper that needs another round of hard work to tighten up some aspects.

Reviewer #2:

Essential revisions:

1) Throughout the paper, the authors claim to show in vivo evidence that supports their conclusions. There is not a single in vivo experiment in this entire data set. They consistently refer to their acute slice experiments as in vivo work: this is incorrect. Acute slice work is generally referred to as in situ, as it does not wholly replicate the in vivo environment. This is critically important when it comes to astrocytes, which are extremely sensitive to osmotic and ion changes that can occur during the process of slicing and can alter astrocyte activity.

We have revised our description of the slice results by using the phrase “in situ” instead. In addition, we have collaborated with the Wang laboratory at NIMH, an expert in live imaging, to perform calcium imaging experiment in live mice, which replicate the in vivo environment. Consistent with our findings in cultured human and mouse astrocytes and mouse brain slices, we observed increased frequency of spontaneous calcium activity in astrocytes in live mice whose Mecp2 gene was predominantly deleted in cortical astrocytes. These new data are included in our revised manuscript (Figure 5).

2) There is no characterization of the iPSC cultures to determine if they are in fact astrocytes. A reference is given to a previous paper, but no characterization of the cultures used in the experiments conducted in this manuscript is provided. It is critical to include experiments where the cultures used for calcium imaging are fixed and stained for astrocyte markers.

We have performed detailed characterization of astrocytes differentiated from congenic pairs of Rett syndrome iPSC lines in our previous publication (Williams et al., 2014), and again in our current study. We didn’t include those data in our original submission but have included them in our revised manuscript (Figure 1—figure supplement 1).

3) Calcium imaging data: Overall, the calcium imaging data from both iPSCs, primary cultures and in situ slice experiments is not convincing. Fluo-4 is a single wavelength dye, and while normalizing to baseline can reduce issues concerning concentrations of indicator, this indicator can only be used for relative measures of calcium amplitudes and requires careful loading controls that do not appear to be presented. Use of a ratiometric dye for experiments would have been more convincing for the changes that are described. Additionally, in the entire manuscript the data presented is from somatic recordings of astrocyte transients. While for the iPSCs work this may be understandable, for the primary culture and slice work there should have been an effort to examine calcium transients in the enlarged cell body or processes of astrocytes, as the somatic activity of an astrocyte is incredibly low compared to the activity in processes (see Di Castro et al., 2011) and represent a small fraction of the total calcium activity. As the authors used a genetically encoded calcium indicator in some experiments, why not in all experiments? This would have been more convincing.

In our experiments, we focus more on the relative changes of intracellular calcium concentration rather than the absolute concentration. In addition, our focus was on the comparison between the MT and WT genotypes. Finally, the frequency results should not be affected by the loading conditions.

That being said, we took the reviewer’s suggestion, and performed additional experiments while including calcium ionophore A23187 as a loading control (Figure 1—figure supplement 2) in calcium imaging experiments involving human astrocytes. More importantly, we used the GCaMP6 reporter mice in calcium imaging experiments in primary mouse astrocytes (Figure 3) and live mice (Figure 5), both of which yielded results consistent with calcium dye-based results from cultured human astrocytes (Figure 1) and acute mouse brain slices (Figure 4). Finally, we took the reviewer’s suggestion to re-analyze the calcium activity of astrocytic processes and presented them separately (Figure 1—figure supplement 3).

4) The frame rate for calcium imaging is too slow for an accurate recording of the spontaneous activity and is likely underestimating the frequency of spontaneous events.

To avoid potential phototoxicity, we did calcium imaging at the rate of one image per every 2 seconds. The kinetics of astrocytic calcium activity is far slower than this frame rate. We did the kinetics analysis and the FWHM (full width at half maximum) is ~40 Sec for soma and ~30 sec for processes. Also, there are many papers in which the authors used similar sampling rate. For GCaMP6s imaging, we used faster frame rates (mouse primary astrocytes (Figure 3): 1 frame per second; live mice (Figure 5): 1.83 frame per second).

5) In general, in the majority of the experiments, the n's presented for results are low for what this reviewer would expect from culture experiments where the dish contains many thousands of cells.

During the short imaging period (10 minutes) for each imaging field, only a percentage of the astrocytes in culture exhibited spontaneous calcium elevations. Only these cells were included in the statistical analysis. Further, being mindful that long duration of such imaging experiment may affect the health of cells and the data consistency, we limited the total amount of imaging time for each dish, thereby obtaining only a few images from a single dish. We hope the consistency in results between species and across the in vitro, in situ, and in vivo platforms will alleviate the reviewer’s concern.

6) The choice of representative images, such as the immunostaining in Figure 5A, is inconsistent with the data presented. The authors use a lentivirus to reintroduce Mecp2 into their null cell populations, along with GFP. The MeCp2-null cells depicted by the GFP images seem to be more numerous and have different morphology. Furthermore, there seem to be some red signal indicating some expression of MeCP2 in these cells. This calls into question the potential health and fidelity of the cells, or the choice of images. A similar issue is found in Figure 5H: a representative staining has several cells from MeCP2-null neonates selected that express GFP and Rhod-2. However, a close examination of these areas does not show any Rhod-2, which could explain why the authors do not report any Ca transients in these regions. Again, the choice of representative images is called into question, since the representative traces depict absolutely no traces during a ~10 min period, yet their data show the frequency in this population to be 1.33 / 0.1 min-1. In general, throughout the paper the representative data seems inconsistent with the reported quantification.

We apologize for the poor quality of the representative images, which was due to image processing during the submission of the manuscript. We have corrected that issue. In new images included in the revised manuscript, Figure 2A shows no difference in cell morphology between mutant and wild type. Finally, in the mouse astrocytes rescue experiments presented in Figure 5 of our original submission, the average of frequency was calculated using cells showing calcium transient. In another word, although all cells (cell 1-8) were used to calculate percentage of cells with calcium activity, only cells with calcium activity (cell 1, 5, 6, 7, 8) were used to calculate frequency and amplitude. Given the data are largely redundant with those from the human cells, we removed them from our revised manuscript as part of our effort to streamline the organization.

7) The authors of the paper make an assumption concerning astrocyte calcium signaling that there have been no changes to the signaling pathways contributing to the generation of spontaneous signaling aside from TRPC4. This is highly unlikely given the nature of MeCP2. The major release pathway from the ER involves activation of IP3 receptors. While TRPC4 does appear altered, it may also be the case that other components of the ER release pathway may be altered as well and be the root cause of the increased spontaneous signals. A closer examination of other components of ER associated calcium signaling should be conducted. Additionally, staining for TRPC4 to show it is even expressed by astrocytes in vivo is needed to validate the model.

We agree that additional components of the calcium homeostasis regulation may be affected, and plan to further investigate in the future. As suggested by the reviewer, we performed immunostaining of Trpc4 on brain sections from wild type and Mecp2 knockout mice and observed increased intensity of Trpc4 immunoreactivity in astrocytes in brain sections from the Mecp2 knockout mice. These new results are consistent with our observation of increased TRPC4/Trpc4 expression in cultured human and mouse astrocytes, and are included in the revised manuscript (Figure 6—figure supplement 2).

Reviewer #3:

1) All data are in vitro (cultured cells) or ex vivo (slice) with no in vivo evidence that astrocytic calcium dynamics and subsequent extrasynaptic NMDA transmission are dysregulated in Rett Syndrome. While it is appreciated that not everyone is capable of imaging calcium oscillations in vivo, one relatively straight forward experiment that would start to address this point is to assess seizure susceptibility in vivo (vs. ex vivo) following astrocyte-specific knockdown or ablation of Mecp2.

We have collaborated with the Wang laboratory at NIMH to perform calcium imaging experiment in live mice, which replicate the in vivo environment. Consistent with our findings in cultured human and mouse astrocytes and mouse brain slices, we observed increased frequency and amplitude of spontaneous calcium activity in astrocytes in astrocyte-specific Mecp2 knockout mice. These new data are included in our revised manuscript (Figure 5).

2) There are multiple inconsistencies in the data reporting changes in calcium in mutant or null astrocytes versus wild-type. Often calcium dynamics are very different across figures for the same genetic or pharmacological manipulation. For example, the calcium dynamics reported in Figure 12C,D show different levels of mutant calcium oscillation and frequency than is reported in Figure 2B,C. These discrepancies need to be addressed.

The apparent inconsistency is mostly due to poor selection of sample traces. We have included more representative sample traces in our revised manuscript.

3) The authors fail to clearly define their N numbers in their figures or results. It is not clear from reading the manuscript whether the N's provided refer to number of cells, number of experiments, etc. Given that the N's are relatively high, one would assume this represents cells. However, it is unclear how many independent experiments these N's represent. Data should be assessed from at least three independent experiments and the authors should sample similar numbers of cells across experiments. This information should be clearly outlined. Furthermore, while the authors list N's in the text, the authors should list this in each figure legend for clarity. The authors also need to list the stats and p-values for each experiment in the figure legend. This is inconsistent across the manuscript.

We will include the requested information in our revised manuscript.

4) The authors conclude that evoked calcium activities are disrupted following either ATP or glutamate stimulation. The mutant or null astrocytes already have higher amplitude responses spontaneously (e.g. Figure 1E). If you compare the amplitudes of spontaneous responses to evoked responses, it appears that the actual magnitude of increase following stimulation is not different between wild-type and null or mutants astrocytes (e.g. Figure 1-amplitude increases from about 2 (spontaneous) to about 3.5 (+ATP) for WT and from about 3 (spontaneous) to about 4.5 (evoked) in MT). Therefore, the conclusion that evoked responses are different is not accurate and should be clarified.

If the increase in spontaneous and evoked calcium elevations share the same mechanism, it is rational that the scale in difference between WT and MT in spontaneous and evoked calcium activity is similar. We respectfully disagree with the reviewer that we should compare the amplitude of pharmacologically evoked calcium transients with spontaneous ones.

5) The figures are generally very difficult to follow and could use some reorganization. For example, the authors often go back and forth between human and mouse within the same figure without clearly identifying this within the figure. In addition, a number of figures could be either combined or included as supplemental material. For example, Figure 8, Figure 9, Figure 10 and Figure 11 could all be represented along with data from Figure 7 or included as a supplemental file.

We have rearranged the figures in our revised manuscript to make it easier to follow.

6) The images presented throughout the manuscript often lack many general cell markers that would benefit interpretation of the data. In some cases, the figure lacks any images of cultures (Figure 2). Most images do not include DAPI to label all cells and no images show staining for astrocyte-specific markers. Higher magnification images should also be added to the current low magnification images. In addition, the authors list quantification in the text but these data should be included in the figure to increase clarity.

We have included the requested information in figures in our revised manuscript.

7) The authors need to validate sufficient knockdown of genes and reduced protein in astrocytes following shRNAs. These data can be included as supplemental.

We did western blot to validate shRNA knockdown efficiency and included the data in our revised manuscript (Figure 7).

8) In Figure 13 and Figure 14, there is only quantification of mCherry labeling not loss of Mecp2 expression. The authors must quantify presence or absence of Mecp2 protein in mCherry positive and negative astrocytes and neurons. This is particularly important given that detecting Mecp2 in vivo in non-neuronal cells is notoriously difficult and typically requires amplification (Ballas et al., 2011). From the images, it is not clear that the antibody recognizes Mecp2 in uninfected astrocytes, which is important to show in order to validate loss of Mecp2 in infected astrocytes. This figure could also benefit from higher magnification images. Lastly, the authors suggest that they are creating cell-specific knockout mice. This is exaggerated as only about 27% of cells are infected.

We have carried out the quantification as suggested by the reviewer and included the data in our revised manuscript (Figure 8 —figure supplement 2). As for the low infection rate, it offers strong indication of the connection between loss of MeCP2 in astrocytes and the excessive activation of eNMDARs, because the phenotype of eNMDAR activation is detectable when ~1/4 of astrocytes lost MeCP2.

9) The manuscript emphasizes that their heightened extrasynaptic NMDA receptor activation has never before been reported in any RTT models. This is inaccurate. There is at least one recent paper reporting increased activation of eNMDARs in RTT (Lo et al., 2016). While it is appreciated that this effect has never been attributed to astrocytes, the text should be modified to include these published findings.

We have cited the recent publication in our revised manuscript.

10) The model provided in Figure 15 is not detailed enough to add any meaningful contribution to the manuscript as a whole and could be excluded.

We believe Figure 15 in our original submission (Figure 9 in our revised manuscript) is a good summary of our results, and can help orient readers, and prefer to keep it unless advised otherwise by the editors.

[Editors' note: the author responses to the re-review follow.]

Essential revisions:

1) Most data supporting the TRPC4 mechanism are in vitro. The only data showing some evidence of increased TRPC4 function in Mecp2 null mice are immunostaining in Figure S10. More data demonstrating TRPC4 expression changes and function in vivo/ex vivo would strengthen the study. For example, western blots for TRPC4 from acutely isolated astrocytes (mouse model) in addition cultured astrocytes would be helpful.

At minimum, please include in brief discussion emphasizing the need for future studies on the topic. This would help the reader and thoughtfully point out the limitations of the current work.

We agree with the reviewers that most of our data on TRPC4 are in vitro. However, current methods for isolating mouse primary astrocytes involve culturing the cells for extended period (~4 weeks), so that other cell types are removed during the process. In that sense, Western blot analysis from those cells is still in vitro. In order to assess the functional relevance of Trpc4 in Rett syndrome disease progression, we have generated conditional Trpc4 knockout mice, and are breeding them with the Mecp2 knockout mice. We plan to use astrocyte-specific Cre driver to delete Trpc4 in the Mecp2 knockout mice and hope to report our findings in a future paper. As suggested by the reviewers, we have included a brief discussion emphasizing the need for future studies on the topic in the Discussion section. The exact language used is “As most of our TRPC4-related data came from in vitro experiments, future in vivo studies are needed to ascertain the functional significance of increased TRPC4 expression in RTT disease progression.”

2) The authors' explanation to reviewers regarding the large variability in N's is not sufficient. In many cases the N's vary by 100s of cells. For example, Figure 1C the N=526 cells for mutant cultures and N=811 for wild-type cultures. As one reviewer previously pointed out, this could greatly affect the statistical analyses. Further, it is unclear how the authors chose a particular N prior to doing statistics and it remains unclear in the manuscript how many independent experiments each data set represents.

We didn’t choose a particular N prior to doing statistics. We analyzed all cells with calcium activity in our experiments. The big difference in N between genotypes is caused by the biological difference between the genotypes. For the spontaneous calcium activity, because MECP2 mutant (MT) or knockout (KO) group usually have more cells showing calcium activity and have more frequent calcium events, resulting a bigger sample size N than that of the WT. Take Figure 1A-C as an example. 26 fields were randomly selected from each genotype group. Since the MT and WT cells were plated at the same density, the total number of cells imaged were about the same between the two genotypes. However, significantly more MT cells were observed to have spontaneous calcium activity than the WT cells (30% vs. 18%). When the subsequent analysis included only the cells with spontaneous calcium activity, the N for MT ended up being significantly bigger than the N for WT (408 vs. 280). This is the reason for the difference in N in all of our figures, except for those acute brain slice experiments. For the acute brain slice experiments, because the person performing the calcium imaging experiments was blind to the genotype of mice, it was impossible to make the N similar between the two genotype groups (especially when we typically used all mice in each litter, and those litters happened to have more WT mice than KO mice).

Moreover, we performed additional statistical analysis of all of existing data by randomly removing some values in the groups with bigger N to make the sample sizes the same between groups and included the results at the end of our response in a section named “Secondary Statistical Analysis”. For each of the figures with big difference in N, we performed such test three times (Random removal trials 1-3). All conclusions remained the same as before random removal of data (All data), suggesting the difference in N had no effect on our original conclusions. While it is not necessary to include such secondary analysis in our manuscript, we include it here to satisfy the reviewers.

Finally, more than three independent experiments/biological replicates (for human astrocyte data set, each independent experiment was defined as one independent differentiation; for mouse astrocyte in vitro data set, each independent experiment was defined as one isolation culture from one mouse; for mouse in situ and in vivo data set, both number of cells and number of mice were described) were included in each data set. Since many different types of data sets are presented, we chose to report the number of cells in most cases to make it easier to read, compare and interpret. The above language has been added to the Materials and methods section under the subheading of Statistics analysis.

3) While the authors make more attempt to validate their cell-specific genetic manipulations of Mecp2, these data are still not quantified sufficiently. The data in Supplementary file 3 and Figure 8—figure supplement 2 only shows quantification of% cells with mCherry, GFAP, and NeuN expression. The% of NeuN+ and GFAP+ cells +/- MECP2 immunoreactivity after AAV injection should also be included here. Given that the goal is to achieve cell-type specific manipulation, these data are important.

We originally focused on the mCherry signal in our quantification, because it was the fluorescence marker used for targeted recording. We thank the reviewers for making this good suggestion. Table 3 and Figure 8—figure supplement 2 using a new cohort of mice. In this experiment, we quantified the percentage of MeCP2 positive/negative cells in NeuN+ and GFAP+ cells, and reported the new results in Table 4. The top row of Table 4 confirmed our expectation that almost all the mCherry positive cells are negative for MeCP2 (i.e. virus-infected cells expressed the Cre recombinase and lacked MeCP2 expression). Nonetheless, there is negligible number (2%) of NeuN+ cells that are negative for MeCP2 in the hippocampus of AAV-GFAP-mCherry-Cre injected mice, and a small number (7%) of GFAP+ cells that are negative for MeCP2 in the hippocampus of AAV-hSyn-mCherry-Cre injected mice. These results have been added to the main text of the Result section. Furthermore, for more appropriate interpretation of our results, we have changed our conclusion of those experiments from “astrocyte-specific” and “neuron-specific” to “predominantly astrocyte-specific” and “predominantly neuron-specific”. The exact language used in our revised Result section is “Further double staining with either anti-GFAP and anti-MeCP2 antibodies or anti-NeuN and anti-MeCP2 antibodies on those brain sections revealed that, while almost all mCherry-positive cells were indeed negative for MeCP2 in mice receiving either AAV virus, there was a negligible number (2%) of NeuN-positive cells that were negative for MeCP2 in the hippocampus of the AAV-GFAP-mCherry-Cre injected mice, and a small number (7%) of GFAP-positive cells that were negative for MeCP2 in the hippocampus of the AAV-hSyn-mCherry-Cre injected mice. Thus, mice with predominantly astrocyte-specific deletion of Mecp2 and predominantly neuron-specific deletion of Mecp2 were generated.”

4) Many immunofluorescence images are still generally low quality. For example, in Figure 6—figure supplement 2, the background staining of TRPC4 is quite a bit higher in MECP2 null animals vs. WT. The non-specific background staining should be the same in these images to ensure a more accurate comparison. This and other figures would also benefit from some higher magnification zoom-ins in addition to the low magnification images. There is also immunostaining that is very difficult to visualize in several figures. One clear example is Figure 8—figure supplement 2 (MECP2 and NeuN immunoreactivity).

We have replaced the images in question with better quality and more representative ones (Figure 6—figure supplement 2 and Figure 8—figure supplement 2), and added higher magnification zoom-ins.

https://doi.org/10.7554/eLife.33417.049

Article and author information

Author details

  1. Qiping Dong

    Waisman Center, University of Wisconsin-Madison, Madison, United States
    Contribution
    Data curation, Formal analysis, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  2. Qing Liu

    Unit on Neural Circuits and Adaptive Behaviors, National Institute of Mental Health, Bethesda, United States
    Contribution
    Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared
  3. Ronghui Li

    Waisman Center, University of Wisconsin-Madison, Madison, United States
    Contribution
    Data curation, Investigation
    Competing interests
    No competing interests declared
    ORCID icon 0000-0001-6329-5895
  4. Anxin Wang

    Waisman Center, University of Wisconsin-Madison, Madison, United States
    Contribution
    Data curation, Investigation
    Competing interests
    No competing interests declared
  5. Qian Bu

    Waisman Center, University of Wisconsin-Madison, Madison, United States
    Contribution
    Data curation, Investigation
    Competing interests
    No competing interests declared
  6. Kuan Hong Wang

    Unit on Neural Circuits and Adaptive Behaviors, National Institute of Mental Health, Bethesda, United States
    Contribution
    Supervision, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-2249-5417
  7. Qiang Chang

    1. Waisman Center, University of Wisconsin-Madison, Madison, United States
    2. Department of Medical Genetics, University of Wisconsin-Madison, Madison, United States
    3. Department of Neurology, University of Wisconsin-Madison, Madison, United States
    Contribution
    Conceptualization, Data curation, Supervision, Funding acquisition, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    qchang@waisman.wisc.edu
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-7625-2170

Funding

National Institute of Neurological Disorders and Stroke (R21NS081484)

  • Qiang Chang

National Institute of Mental Health (ZIAMH002897)

  • Kuan Hong Wang

Eunice Kennedy Shriver National Institute of Child Health and Human Development (U54HD090256)

  • Qiang Chang

Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01HD064743)

  • Qiang Chang

National Institute of Neurological Disorders and Stroke (R56NS100024)

  • Qiang Chang

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Xiaoji Zhang for assistance with mouse work, and Dr. Dan Bolt for assistance with statistical analysis. RL was supported by a pre-doctoral fellowship from the Stem Cell and Regenerative Medicine Center at the University of Wisconsin-Madison. This work was partially supported by R21 NS081484, R56NS100024, and R01HD064743 to QC, ZIA MH002897 to KHW and QL, and U54HD090256 to the Waisman Center.

Ethics

Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled according to approved institutional animal care and use committee (IACUC) protocols (G005315) of the University of Wisconsin-Madison

Reviewing Editor

  1. Beth Stevens, Reviewing Editor, Boston Children's Hospital, Harvard Medical School, United States

Publication history

  1. Received: November 7, 2017
  2. Accepted: March 28, 2018
  3. Accepted Manuscript published: March 29, 2018 (version 1)
  4. Version of Record published: April 16, 2018 (version 2)

Copyright

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

  • 1,297
    Page views
  • 224
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Cell Biology
    2. Developmental Biology
    3. Stem Cells and Regenerative Medicine
    Girish R Mali et al.
    Research Article Updated
    1. Developmental Biology
    2. Stem Cells and Regenerative Medicine
    Wael Bazzi et al.
    Research Article