Increasing brain serotonin corrects CO₂ chemosensitivity in methyl-CpG-binding protein 2 (Mecp2)-deficient mice

Abstract

Mice deficient in the transcription factor methyl-CpG-binding protein 2 (Mecp2), a mouse model of Rett syndrome, display reduced CO₂ chemosensitivity, which may contribute to their breathing abnormalities. In addition, patients with Rett syndrome and male mice that are null for Mecp2 show reduced levels of brain serotonin (5-HT). Serotonin is known to play a role in central chemosensitivity, and we hypothesized that increasing the availability of 5-HT in this mouse model would improve their respiratory response to CO₂. Here we determined the apnoeic threshold in heterozygous Mecp2-deficient female mice and examined the effects of blocking 5-HT reuptake on the CO₂ response in Mecp2-null male mice. Studies were performed in B6.129P2(C)-Mecp2^τm1.1Bird null males and heterozygous females. In an in situ preparation, seven of eight Mecp2-deficient heterozygous females showed arrest of phrenic nerve activity when arterial CO₂ was lowered to 3%, whereas the wild-types maintained phrenic nerve amplitude at 53 ± 3% of maximal. In vivo plethysmography studies were used to determine CO₂ chemosensitivity in null males. These mice were exposed sequentially to 1, 3 and 5% CO₂. The percentage increase in minute ventilation in response to increased inspired CO₂ was less in Mecp2^−/y than in Mecp2^+/y mice. Pretreatment with citalopram, a selective 5-HT reuptake inhibitor (2.5 mg kg⁻¹ I.P.), 40 min prior to CO₂ exposure, in Mecp2^−/y mice resulted in an improvement in CO₂ chemosensitivity to wild-type levels. These results suggest that decreased 5-HT in Mecp2-deficient mice reduces CO₂ chemosensitivity, and restoring 5-HT levels can reverse this effect.

A recent study showed that CO₂ chemosensitivity is depressed in male mice with deletion of the transcription factor methyl-CpG-binding protein 2 (Mecp2; Zhang et al. 2011). MECP2 is mutated in the autism spectrum disorder Rett syndrome (RTT), and the phenotype includes abnormal respiration characterized by hyperventilation followed by apnoea (Southall et al. 1988; Smeets et al. 2006). An elevated CO₂ threshold in RTT would contribute to apnoea, and significant hypocapnia has been reported in affected individuals (Southall et al. 1988). Zhang and co-authors reported a blunted CO₂ response, relative to wild-type (WT) mice, at 1, 2 and 3% CO₂, but no difference at 6 and 9%. Earlier reports had not shown a difference in the respiratory response to CO₂ between Mecp2-deficient and WT mice (Bissonnette & Knopp, 2006; Voituron et al. 2009). Bissonnette & Knopp (2006) used CO₂ concentrations of 3, 5 and 7%, with the balance being oxygen. The hyperoxia would blunt contributions from peripheral chemoreceptors that are important to the integrated CO₂ response (Forster et al. 2000; Blain et al. 2010; Ramanantsoa et al. 2011). Voituron and co-authors (2009) used a single CO₂ level of 4% in air. In view of the findings of Zhang et al. (2011), which showed no difference in response to CO₂ at higher levels (i.e. 6 and 9%), this single 4% exposure may have failed to show the depressed respiratory response in Mecp2-null mice that is present at 1 and 3% CO₂.

Subjects with RTT and mouse models have deficits in serotonin (5-HT). 5-Hydroxyindoleacetic acid levels are low in the cerebrospinal fluid of individuals with RTT (Samaco et al. 2009). Brain 5-HT levels in Mecp2-null males, while initially normal, are depressed at postnatal day 42 and thereafter (Ide et al. 2005; Viemari et al. 2005; Panayotis et al. 2011). In addition, a recent study has shown that a serotonin 1a receptor agonist reduces apnoea, corrects the irregular breathing pattern and prolongs survival in Mecp2-null males (Abdala et al. 2010). The importance of 5-HT in CO₂ chemosensitivity is well documented (reviewed by Hilaire et al. 2010; Hodges & Richerson, 2010a,b). The goal of the present study was to determine whether increasing brain 5-HT will correct the blunted CO₂ response in Mecp2-deficient mice.

Methods

Ethical approval

The experiments were performed according to the UK Home Office Animals (Scientific Procedures) Act (1986) and were in agreement with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH publication no. 85-23, revised 1996).

Animals

Studies were performed with B6.129P2(C)-Mecp2^τm1.1Bird mice (Jackson Laboratory, Bar Harbor, ME, USA; stock no. 003890) and their WT littermates. Heterozygous females were between 4.0 and 18.7 months of age and null males between postnatal day 40 and 72 at the time of study. Heterozygous female mice were used because they are thought to better recapitulate the disease in female human patients. Mecp2-null males were chosen for the citalopram protocol because they have been shown to have depressed brain 5-HT concentrations (Ide et al. 2005; Viemari et al. 2005; Panayotis et al. 2011).

In situ studies

We used an in situ decerebrated ‘working heart–brainstem preparation’ of the mouse (Paton, 1996), which is a modified version of cardiopulmonary bypass and allows direct recordings to be made from the phrenic nerve (PN), as well as the ability to manipulate the arterial CO₂ concentration precisely. This preparation has previously been used to define an apnoeic threshold in juvenile rats (St-John & Paton, 2000) and is not confounded by depressant action of anaesthetic agents or changes in sleep/arousal states. Briefly, female heterozygous Mecp2^−/+ mice were heparinized (1500 IU I.P.) and subsequently anaesthetized deeply with halothane until loss of their paw withdrawal reflex. Mice were decerebrated, then bisected subdiaphragmatically and the head and thorax immersed in ice-cold carbogenated artificial cerebrospinal fluid. The thoracic phrenic nerve was cut distally. Preparations were moved to a recording chamber. A double-lumen catheter (Braintree Scientific, Braintree, MA, USA) was inserted into the descending aorta for retrograde perfusion by a peristaltic roller pump (505Du; Watson Marlow, Falmouth, UK). The perfusion solution consisted of carbogenated modified Ringer solution at 32°C with an oncotic agent replacing serum albumin (Ficoll 1.25%). The second lumen of the catheter was used to monitor aortic perfusion pressure. The baseline perfusate flow was preset between 18 and 20 ml min⁻¹.

The level of CO₂ within the perfusate was monitored using an in-line gas analyser (FM2 CO₂ analyser; P.K. Morgan, Sequim, WA, USA), and the mix of CO₂ to oxygen was altered manually to achieve different levels of CO₂ ranging from 3 to 9% in 3% incremental steps (balance oxygen). Maximal phrenic amplitude was obtained with 7 or 9% CO₂ in the perfusate.

Arterial pressure, ECG and phrenic nerve bursts (via a suction electrode) were recorded in Spike2 software (Cambridge Electronic Design, Cambridge, UK) on a PC. Data were saved to disk and analysed offline.

Plethysmography studies

Respiratory frequency, tidal volume (V_T) and their product, minute ventilation (V̇_E), were determined in a body plethysmograph (Mortola & Noworaj, 1983; Bissonnette & Knopp, 2006). Briefly, individual unanesthetized animals were placed in a 65 ml chamber with their head exposed through a close-fitting hole in Parafilm.^® A pneumotachograph (Mortola & Noworaj, 1983) was connected to the chamber and a differential pressure transducer (Model PT5A; Grass Instrument Co., West Warwick, RI, USA). The pressure signal was integrated to give tidal volume. Volume changes were calibrated by injecting known amounts of air into the chamber. The analog signal from the transducer was amplified, converted to digital, displayed on a monitor, and stored to computer disk for later analysis. The studies were begun after the animal was given time to become adjusted to the chamber. A cone was fitted over the animal’s head that allowed inspired gases to be delivered. The gas was rapidly (5 s) changed in succession from air to 1, 3 and then 5% CO₂, with the balance being air. The last 2 min of each 5 min exposure to air and to CO₂ was used for analysis (as used previously by Davis et al. 2006). Segments of the record that contained apnoea were not used to calculate frequency. The minute ventilation in response to CO₂ is reported as the percentage increase relative to the control. Recorded data were analysed with custom functions in Chart v5.5.6 (AD Instruments Inc., Colorado Springs, CO, USA). In order to determine the effect of raising brain 5-HT in Mecp2-null males, citalopram (2.5 mg kg⁻¹, I.P.) was administered 40 min prior to a CO₂ study. This dose and time interval were based on a microdialysis study that showed effective increases in 5-HT in the prefrontal cortex of mice (Calcagno et al. 2007).

Analysis

Results are reported as means ± SEM. Comparisons between Mecp2^−/+ and WT phrenic amplitude in the in situ studies were made with Student’s unpaired t test. Comparisons between Mecp2^−/y mice and WT and between control and citalopram treatment were made with one-way repeated-measures ANOVA and the Bonferroni method used for post hoc analysis. A P value of <0.05 was used for significance.

Results

Effect of lowering CO₂ in the in situ preparation

Previous studies of CO₂ chemosensitivity in Mecp2-deficient mice have used additions to inspired gases (Bissonnette & Knopp, 2006; Voituron et al. 2009; Zhang et al. 2011) that preclude an estimate of apnoeic threshold. Arterial perfusion in the in situ preparation allowed us to lower the CO₂ concentration to a level that resulted in phrenic apnoea (see representative trace in Fig. 1A). Six of seven Mecp2-deficient heterozygous females (Mecp2^−/+) showed arrest of phrenic nerve activity when CO₂ was lowered to 3% (Fig. 1A and B). The seventh animal had a phrenic amplitude that was 38.6% of its maximal value. In two Mecp2^−/+ mice, CO₂ was lowered to only 4%. One became apnoeic, and in the other the phrenic amplitude fell to 55.9% of maximal. In the Mecp2^−/+ mice that developed apnoea, phrenic activity returned in 126 ± 42 s after the perfusate was set back to 6% CO₂. In contrast to Mecp2^−/+ mice, phrenic amplitude was sustained at 53 ± 3% of maximal in WT females perfused with 3% CO₂ (n = 7). Intermediate CO₂ levels resulted in non-significant decreases in phrenic amplitude in Mecp2^−/+ mice compared with WT; 72.2 ± 7.5 versus 81.2 ± 2.4% of maximal at 6% CO₂ and 62.5 ± 7.6 versus 71.5 ± 4.7% at 5% CO₂.

Figure 1. The apnoeic CO₂ threshold is higher in Mecp2^+/− mice in situ.

A, representative phrenic nerve (PN) traces for wild-type and Mecp2^−/+ female mice. Recordings were made in situ at baseline (6% arterial CO₂) and 3% CO₂ and show both raw and integrated data. The Mecp2^+/− mice become apnoeic at 3% CO₂, whereas the WT do not. Note the different time scales. B, summary of PN response to changes in arterial CO₂ for wild-type (n = 5–7) and Mecp2^−/+ mice (n = 5–7). Six of seven Mecp2^−/+ mice were apnoeic at 3% CO₂. In two Mecp2^−/+ mice, CO₂ was only lowered to 4%. One became apnoeic, and phrenic amplitude decreased to 55.9% of maximal in the other animal. Values are means ± SEM. ^* P = <0.001, Student’s unpaired t test.

Carbon dioxide chemosensitivity in Mecp2-null males in vivo

Baseline respiratory parameters were similar in Mecp2-null males (Mecp2^−/y; n = 9) and WT (n = 10; frequency, 250 ± 11 versus 242 ± 11 breaths min⁻¹; V_T, 5.0 ± 0.6 versus 5.5 ± 0.6 μl g⁻¹; and V_E, 1.23 ± 0.17 versus 1.29 ± 0.09 ml g⁻¹) as was age at time of study (48.8 ± 2.2 versus 47.9 ± 2.1 days; ANOVA). The response to CO₂ in Mecp2^−/y mice was significantly depressed at all three levels of inspired gas (P = 0.013 at 1%, P = 0.02 at 3% and P = 0.049 at 5% CO₂; ANOVA; Fig. 2A).

Figure 2. Summary of CO₂ chemosensitivity.

A, increase in minute ventilation in wild-type (n = 10) compared with Mecp2-null mice (n = 9). B, effect of citalopram (2.5 mg kg⁻¹) on CO₂ response in Mecp2-null mice (n = 6); ^* P ≤ 0.05, ^{* *} P ≤ 0.01 (ANOVA).

Effect of citalopram on CO₂ chemosensitivity in male mice

Forty minutes after administering the 5-HT reuptake blocker citalopram (2.5 mg kg⁻¹. I P.), respiratory frequency was increased from 242 ± 11 to 284 ± 9 breaths min⁻¹ (P = 0.007, Student’s paired t test) in WT mice (Table 1). There was a tendency for V̇_E also to increase (1.58 ± 0.08 versus 1.29 ± 0.09 ml g⁻¹; P = 0.054) in WT. The reuptake antagonist did not significantly affect baseline respiratory parameters in Mecp2^−/y mice (frequency, 250 ± 11 versus 224 ± 22 breaths min⁻¹; V̇_E, 5.0 ± 0.6 versus 6.6 ± 1.2 μl g⁻¹; and V_T, 1.23 ± 0.17 versus 1.39 ± 0.17 ml g⁻¹; n = 6). This dose of the 5-HT reuptake blocker significantly increased the CO₂ response at all levels of inspired gas in Mecp2^−/y mice (P = 0.04 at 1%, P = 0.003 at 3% and P = 0.002 at 3%; ANOVA; Fig. 2B). The increases in minute ventilation that resulted after pretreatment with citalopram restored CO₂ chemosensitivity in Mecp2^−/y mice to that of WT, but baseline ventilation was unaffected. In the baseline study prior to administration of citalopram, increases in both respiratory frequency and tidal volume contributed to the V̇_E response to CO₂ (Fig. 3B). After inhibition of 5-HT reuptake, V_T alone accounted for the increase in V̇_E (Fig. 3). Indeed, frequency decreased at 3 and 5% CO₂ compared with the corresponding CO₂ response during baseline conditions (Fig. 3B). In contrast to the effects of citralopram in Mecp2-deficient males, the drug did not significantly change CO₂ responses in WT mice (n = 8; data not shown).

Table 1.

Baseline respiratory parameters

Strain	Age (days)	Weight (g)	Treatment	Number	Frequency (breaths min−1)	Tidal volume (μl g−1)	Minute ventilation (ml g−1)
Mecp2+/y	47.9 ± 2.1	21.5 ± 0.4	Control	10	242 ± 11	5.5 ± 0.6	1.29 ± 0.09
			Citalopram	8	284 ± 9*	5.6 ± 0.4	1.58 ± 0.08
Mecp2−y	48.4 ± 2.2	20.1 ± 1.4	Control	9	251 ± 11	5.0 ± 0.6	1.23 ± 0.17
			Citalopram	6	224 ± 22	6.6 ± 1.2	1.39 ± 0.17

Values are means ± SEM.

^*P = 0.007 compared with control (Student’s paired t test).

Figure 3. Respiratory pattern in Mecp2-null male mice before and after citalopram.

A, representative traces from a single mouse showing respiratory pattern during baseline and during 3% CO₂ inspired gas. The top two traces were obtained before and the bottom two traces after administration of citalopram (2.5 mg kg⁻¹ I.P.) to block 5-HT reuptake. Calibration bars are the same for all panels. B, summary of effects of citalopram on respiratory frequency (left panel) and tidal volume (right panel) in Mecp2^−/y mice (n = 6). ^* P ≤ 0.05, ^{* *} P ≤ 0.01 (ANOVA).

Discussion

This report shows that arrest of activity in the phrenic nerve of heterozygous Mecp2-deficient mice occurs at a CO₂ concentration that sustains discharge in wild-type animals at half their maximal amplitude. In addition, the results reveal that a depressed 5-HT concentration in Mecp2-null male mice underlies their blunted CO₂ chemosensitivity. The results add to those of Zhang et al. (2011), who studied Mecp2-null males, by demonstrating an altered CO₂ apnoeic threshold in heterozygous females. Their previous work showed that 5–7 days of treatment with the noradrenaline reuptake blocker desipramine partly (50–70%) restored sensitivity to WT levels. Here we add serotonin to the neuromodulators whose deficiency contributes to the blunted CO₂ response in Mecp2-deficient mice.

The Mecp2-null males exposed to 1, 3 and 5% inspired CO₂ in air showed depressed ventilatory response at all levels. The heterozygous female mice with arterial perfused artificial cerebrospinal fluid at 5 and 6% CO₂ in balance oxygen displayed a reduced phrenic activity compared with WT, but the result was not significant. The difference most probably lies in the oxygen concentrations used. With hyperoxia, the peripheral carotid body ventilator drive (oxygen sensor) will be depressed, but CO₂ will still have a peripheral stimulatory effect. This peripheral CO₂ drive is important for setting the apnoeic threshold, as seen when the perfusate is lowered to 3% CO₂. The results suggest that peripheral CO₂ contributes less to elevated carbon dioxide responses.

Most previous studies examining the role of 5-HT have concentrated on the effects of lowering its level on chemosensitivity. The hypercapnic ventilatory response was decreased by 50% in mice with near-complete absence of central 5-HT neurons (Hodges et al. 2008). This marked lack of 5-HT neurons was achieved by using a conditional knockout mouse in which Lmx1b is deleted in Pet-1-expressing cells. A lesser effect was observed in Pet-1-null animals (Hodges et al. 2011). These animals had 73% fewer tryptophan hydroxylase-immunoreactive neurons in mid-line raphe and 50% fewer in ventrolateral medulla compared with WT mice. At 7% inspired CO₂, minute ventilation was 40% depressed in male mice. There was no significant effect on CO₂ sensitivity at 3 or 4%. In addition, female mice lacking Pet-1 did not show a depressed CO₂ response. The effect of loss of 5-HT transporter has been examined (Li & Nattie, 2008). These mice develop with an elevated brain 5-HT environment and, as adults, have decreased brain 5-HT content. Minute ventilation in 5% CO₂ was 50% less in males that lacked the 5-HT transporter, but there was no effect in females (Li & Nattie, 2008). Recently, the confounding effects of germline removal of 5-HT genes has been eliminated by a genetic strategy that allowed ligand-induced acute hyperpolarization of 5-HT neurons (Ray et al. 2011). Baseline ventilation was not affected by the ligand. The V̇_E, however, was depressed by 47% compared with control mice on exposure to 5% CO₂. Depletion of 5-HT may also affect the interaction between peripheral and central CO₂ chemosensitivity. Depletion of 5-HT neurons that ranged from 37 to 68% at various caudal–rostral levels in the nucleus raphe obscurus of rats when combined with carotid body removal depressed the response to 7% CO₂ by 31.2%. This depression was greater than the sum of 5-HT depletion in raphe obscurus alone and carotid body removal alone (da Silva et al. 2011). Conversely, Depuy et al. (2011) recently showed that selective optogenetic stimulation of serotoninergic raphe obscurus neurons increases ventilation and potentiates ventilatory responses to high CO₂ in mice.

Our observation that an increase in central 5-HT in mice whose serotonin levels are low restores CO₂ chemosensitivity is not singular. In the animals with near-complete absence of 5-HT neurons discussed above (Hodges et al. 2008), the authors used an intracerebroventricular infusion of 5-HT for 2 days. On the second day of infusing 1 and 5 mM 5-HT, the CO₂ response was restored to WT levels. It was estimated that these infusions effected a cerebrospinal fluid 5-HT concentration of 55–275 μM. This estimate is considerably higher than the 0.25 nM found with microdialysis studies of extracellular 5-HT in medulla and prefrontal cortex (Taylor & Basbaum, 1995; Calcagno et al. 2007). Following I.P. injection of citalopram (1.25 mg kg⁻¹) in C57BL/6J mice, prefrontal cortex 5-HT increased by 50%, and after 5.0 mg kg⁻¹ it increased twofold (Calcagno et al. 2007). The intermediate dose used (2.5 mg kg⁻¹) in the present study in Mecp2-null males corrected the CO₂ sensitivity to WT levels.

In normal rats, microdialysis of the 5-HT reuptake inhibitor fluoxetine into raphe magnus augmented the response to 7% CO₂ (Taylor et al. 2004). There was no effect at 1 day of treatment, and changes were present at 7 and 15 days. Additionally, systemic fluoxetine, administered via osmotic minipump, did not alter the response to CO₂. These results are consistent with our findings in WT mice, where acute systemic 5-HT reuptake inhibition did not affect CO₂ chemosensitivity.

In Mecp2-null male mice pretreated with citalopram to increase brain 5-HT, the increase in V̇_E with added inspiratory CO₂ was entirely due to increases in V_T (Fig. 3). In man, this pattern, an increase in tidal volume and no change in frequency, was observed during pressure-support ventilation (Georgopoulos et al. 1997). The hyperventilation significantly lowered the end-tidal CO₂. Assuming that the pattern had the same effect in mice, this would result in lower arterial CO₂ in the postcitalopram study compared with the baseline study, where the response to CO₂ showed increases in both frequency and in V_T.

Recent studies have shown that both astrocytes (Lioy et al. 2011) and microglia (Derecki et al. 2012) play important roles in generating apnoea and breath cycle irregularity in mouse models of RTT. These cell types express 5-HT receptors (Whitaker-Azmitia & Azmitia, 1994; Krabbe et al. 2012); therefore, glia may contribute to the depressed CO₂ chemosensitivity seen in Mecp2-deficient mice.

Reports in Mecp2-deficient mice that established depressed brain 5-HT have not completely elucidated the underlying mechanisms (Ide et al. 2005; Viemari et al. 2005; Panayotis et al. 2011). The number of 5-HT neurons is not reduced (Viemari et al. 2005), indicating that synthesis, secretion or turnover must be affected. The reported increase in 5-hydroxyindoleacetic acid:5-HT ratio (Ide et al. 2005; Viemari et al. 2005; Panayotis et al. 2011) is consistent with an elevated turnover. Recently, Nguyen et al. (2012) measured a number of synaptic proteins in male mice whose Mecp2 protein was markedly depleted at 5 weeks postnatal. Two of the synaptic proteins that were depressed may contribute to the low level of 5-HT in brain of Mecp2-deficient mice. Synapsin 1 has been shown to regulate 5-HT release (Browning et al. 1985) and calcium, calmodulin-dependent protein kinase II phosphorylation of tryptophan hydroxylase 2 increases the activity of this enzyme that is rate limiting for 5-HT production (Kuhn et al. 2007).

The present findings that there is an elevated apnoeic threshold in heterozygous Mecp2-deficient female mice and that an agent which blocks 5-HT reuptake corrects defective CO₂ chemosensitivity have clinical implications. Hypocapnia secondary to hyperventilation has been documented in RTT subjects (Southall et al. 1988; Smeets et al. 2006) and, in one case, supplemental CO₂ was reported to be beneficial (Smeets et al. 2006). In addition, one of two patients with low cerebrospinal fluid 5-HT metabolite was clinically improved when treated with the reuptake blocker fluoxetine, although breathing was not mentioned (Temudo et al. 2009). A separate single case report found marked reduction in hyperventilation/apnoeic episodes following treatment with fluoxetine (Gökben et al. 2012). Although the underlying mechanisms generating rhythm irregularity, apnoeas and low CO₂ sensitivity are likely to be separate, there is strong evidence that low CO₂ sensitivity could exacerbate periods of hyperventilation followed by apnoeas (Khoo, 2000).

Our findings also have special significance for the critical care of Rett syndrome patients suffering from acute respiratory distress. Patients on the more severe end of the spectrum often experience periods of severe recurrent apnoeas (or breath-holds) that result in low oxygen saturation (<60%), with some requiring hospital admission and ventilatory support. Currently, there are no standardized critical care guidelines for treating low oxygen saturation due to severe episodes of apnoeas in Rett patients. Our data indicate that interventions to increase oxygen saturation that may also result in even mild lowering of partial pressures of CO₂ may further exacerbate the episodes of breath holding.

In summary, we have shown that heterozygous Mecp2-deficient female mice have a hypocapnic threshold that exceeds that of wild-type mice. In addition, we have confirmed that Mecp2-null males have a depressed response to CO₂ and have extended this finding to show that elevation of brain 5-HT corrects their CO₂ chemosensitivity.

New Findings.

• What is the central question of this study?

  Mice deficient in methyl-CpG-binding protein 2 (Mecp2; a model of Rett syndrome) display reduced CO₂ chemosensitivity, which may contribute to their breathing abnormalities. Patients and mice show reduced levels of brain 5-HT, which is important in central chemosensitivity. It is not known whether increasing 5-HT in this mouse model would improve their respiratory response to CO₂.

• What is the main finding and its importance?

  We show, for the first time, that females heterozygous for the Mecp2 mutation have a higher hypocapnic threshold for apnoeas versus wild-type mice and that increasing 5-HT in Mecp2-null males, with citalopram, can correct CO₂ chemosensitivity, a finding that may be important clinically.