The Pathophysiology of Rett Syndrome With a Focus on Breathing Dysfunctions

Abstract

Rett syndrome (RTT), an X-chromosome-linked neurological disorder, is characterized by serious pathophysiology, including breathing and feeding dysfunctions, and alteration of cardiorespiratory coupling, a consequence of multiple interrelated disturbances in the genetic and homeostatic regulation of central and peripheral neuronal networks, redox state, and control of inflammation. Characteristic breath-holds, obstructive sleep apnea, and aerophagia result in intermittent hypoxia, which, combined with mitochondrial dysfunction, causes oxidative stress—an important driver of the clinical presentation of RTT.

Introduction

The quest to treat and hopefully cure Rett syndrome (RTT) began more than 50 years ago with the first description of 22 girls having characteristic behavioral regressions in addition to breathing abnormalities (Ref. 167; see also Refs. 76, 153). The discovery of a causal link with the gene MECP2 was an important next milestone (6), identified in 1999. MECP2 is a critical reader of DNA methylation and a transcriptional regulator of many other genes (201); therefore, MECP2 misfunction leads to diverse and complex consequences. The gene’s locus on the X chromosome results in inactivation of one of either the mutated or unaffected MECP2 gene during development (208). Since X inactivation can be skewed, phenotypic consequences of MECP2 mutations are unpredictable. In rare cases, most mutated X chromosomes are inactivated, resulting in a very mild clinical phenotype (187). However, X-chromosome inactivation and the resultant complexity of the MECP2 mutation expression are only two of many reasons that make RTT clinically complex. RTT continues to be a largely clinical diagnosis, because MECP2 mutations are neither necessary nor sufficient to diagnose. Approximately 8% of patients who meet criteria for “classic RS” do not have MECP2 mutations. Interestingly, 50–75% of patients classified as “atypical Rett Syndrome” have a mutation in MECP2 and can be classified into three distinct syndromes according to clinical presentation and mutations in other loci (139, 154). 1) Preserved speech variant, referred to as the Zappella variant, with the majority of cases associated with MECP2 mutation, are the least compromised. The two other variants are rarely associated with MECP2 mutations: 2) Congenital variant (Rolando variant), closely linked to mutations in FOXG1, and 3) Early seizure variant (Hanefeld variant), associated with mutations in CDKL5 (148). Males with MECP2 mutations pose other classification challenges. Initially assumed to be lethal during embryonic development (106, 137), there is increasing evidence that some males succumb to their severe clinical manifestations by the age of 2 yr. Among these children, some are classified as “male Rett encephalopathy” (137). Many insights into the Rett syndrome classification as well as improved clinical characterizations have come from a large multicenter natural history study (26, 92, 95, 140, 198, 199). Nonetheless, the enormous scientific and clinical progress has so far not translated into major therapeutic breakthroughs. Several human clinical trials have failed, and available therapies are effective in some individuals but not in others (72, 92, 94, 107, 112, 143, 153, 160, 186, 188). These inconsistent responses to pharmacological intervention may be due to numerous issues, including homogeneity (or lack thereof) of the RTT cohort (clinically and in terms of MECP2 mutations), clarity and selection of outcome measures, reproducibility of results, and statistical methods—factors not unique to RTT (89, 119, 155). Yet, parents, health care professionals, and scientists have continued to hope. The existing animal models for RTT exhibit phenotypes concordant with the human clinical features (57, 91, 92). In these animal models, pharmacological targeting of different mechanisms and genetic manipulation of specific brain regions have successfully reversed at least a portion of these features, including breathing disturbances and shortened lifespan, offering a promising avenue toward translation and eventual clinical treatments (2, 41, 56, 75, 83, 91, 99, 102, 136, 146, 169, 179, 202, 204, 213, 217).

RTT is a complex disorder, with stealthy onset in early infancy and advancing deterioration with aging. Although the phenotype is consistent for most cases in girls, all confirmed with a MECP2 mutation, there is a subset of patients who, after thorough evaluation, seem to have atypical RTT features and lack a MECP2 mutation. The focus of this invited review article will be the girls with the classic phenotype, including our explanation of the altered physiology intrinsic to the condition and impacted by the recurrent intermittent hypoxemia due to the aberrant breathing. We will also discuss the implications of autonomic dysregulation and dysphagia, clinical features that are very debilitating and thus far largely resistant to intervention trials (67, 93, 96, 145, 164, 170, 215, 216). In addition, we broaden our scope to include discussion of factors that make treating human disorders generally difficult. We will emphasize that what makes RTT complex is not only the complexity of the MECP2 mutation itself but the feedback-laden nature of physiology, which includes changes in metabolic regulation and oxygen homeostasis. Although the neuronal consequences of MECP2 mutations are the major driver of the known clinical phenotypes, ultimately, pathologies induced by MECP2 mutations are not only restricted to the brain; they affect the entire physiology, resulting in an interplay that contributes to the complexity of the clinical phenotype (FIGURE 1). Harnessing this physiological understanding may lead to improved understanding and treatment of RTT.

FIGURE 1.

The dynamic nature of RTT pathophysiology

Schematic illustrating some of the hypothesized interactions that may underlie the RTT phenotype. RTT is associated with disturbances in a variety of CNS regions, resulting in a host of clinical phenotypes. These clinical presentations, in turn, have various pathophysiological consequences that themselves are dysregulated in RTT. This will further exaggerate or create additional symptoms. Important drivers of the RTT pathophysiology are mitochondrial and inflammatory dysfunctions. Targeting specific pathophysiological dysfunctions in RTT shows great therapeutic promise, yet this specific approach may not suffice to reestablish all the interacting dysregulations that are the hallmark of RTT.

The Complex Interplay Between Genotype and Clinical Phenotype

In general, every gene discovery raises numerous new questions with regard to the genotype-phenotype relationships (32, 62, 130). These relationships are typically complex and not always consistent among individuals. One important contributor to this complexity relates to the location on the X chromosome. The same mutation can confer different clinical severity due to degrees of X-chromosome inactivation or skewing. Moreover, only 10 of 18 studies on genotype-phonotype relationships have found significant correlation (78), thus making the determination of clinical prognosis difficult (77, 138). Nevertheless, from the existing data, MECP2 mutations can be classified by their locations in the following manner: 1) methyl-CpG-binding domain, 2) transcription repression domain, 3) COOH-terminal segment, which is generally associated with milder phenotypes and preserved cognitive skills, and 4) large deletions that can be further classified into early truncating mutations, which are more severe than late truncating mutations and associated with other congenital anomalies (12, 63, 77, 140, 156). In addition, secondary complications play a role and may be particularly important in contributing to the individual variations. There are many examples for the impact of secondary complications, and this can perhaps be best explained with the lessons learned from cystic fibrosis (CF). Caused by a mutation in the anion channel CFTR (8), patients die from chronic bacterial infections in the airways, neutrophil inflammation, and lung remodeling. The inflammatory processes are not directly caused by the gene mutation but by a cascade of events: a loss of HCO₃⁻ secretion leaves H⁺ secretion unchecked, resulting in the breakdown of the lung’s host defense system and chronic inflammation (131, 165, 181). In CF, targeting the inflammation brought the first significant success in a sequence of discoveries that ultimately led to remarkable success in the treatment of this devastating disease (190). This is an important lesson for RTT, since the patients’ difficulties with breathing, swallowing, and other orofacial behaviors result in potentially fatal pneumonia and lung infections (FIGURE 1). Yet, inflammation is not only a consequence but also a driver of neuronal disturbances (150, 219). In response to inflammation, microglia interact with many intracellular pathways, including those involved in homeostatic responses (38), and proinflammatory cytokines alter synaptic transmission and plasticity, as has been demonstrated for multiple sclerosis (13, 34, 114, 177, 191). Moreover, it is known that MeCP2 also regulates T-cell function and macrophage responses (42, 43, 178, 228). Thus not only microglia but also disturbances peripheral to the nervous system contribute to the dysregulation of the inflammatory responses. In the case of RTT, neuronally caused dysphagia seems to be exaggerated by inflammation of the pulmonary system (98), as will be further discussed later.

The astonishing reversal of neurological defects following the reactivation of Mecp2 expression in mice in which Mecp2 was silenced led to important insights (75). It illustrated that RTT is not a neurodegenerative disease and showed that homeostatic mechanisms are sufficiently powerful to reverse many of the imbalances and developmental processes that were severely affected by the mutation. However, homeostatic mechanisms are also powerful drivers of the clinical phenotype when MECP2 is mutated (24). Thus, when designing therapeutic strategies, homeostatic mechanisms must be considered.

Homeostatic mechanisms can occur with and without concomitant genetic changes, involving, for example, epigenetic modifications or compensatory physiological mechanisms that are non-heritable (113, 141). These mechanisms are particularly important for the brain, which needs to balance stability and plasticity. Any mutation that challenges this balance will be met by compensatory homeostatic mechanisms that have developed throughout evolution (49, 50, 53, 74, 82). Homeostatic regulation of neuronal excitability is particularly important for brain functions in health (110) and disease (44), and occur at all levels of the nervous system through genetic, molecular, cellular, network, and behavioral mechanisms (105, 211). The mechanisms that regulate synaptic transmission have been extensively studied in neocortex and hippocampus (11, 24, 100, 152, 182, 195, 203, 207). Mecp2 mutations in mice directly impair these homeostatic mechanisms (24), which may be one reason why a variety of imbalances persist throughout the CNS. Neural activity is elevated in the hippocampus (29, 157, 226). The CA3 region in particular shows diminished basal inhibitory rhythmic activity (225), and a pathological regulation of NMDA receptors contributes to impaired synaptic plasticity (15, 123). This hippocampal hyperexcitability may increase seizure susceptibility (29, 30, 68, 142, 158, 192). The neocortex of these Mecp2 mutant mice is generally hypoactive (FIGURE 1), which is partly due to the reduction in excitatory mechanisms (33, 36, 44, 45, 135, 144, 204, 221, 222). A loss of NMDA current has been demonstrated for the mPFC (176), suggesting that targeting neocortical excitability may improve various clinical symptoms (81, 94, 101). However, the role of NMDA mechanisms in the Rett phenotype is rather complex, given that chronic low doses of ketamine have beneficial effects and seem to rescue the breathing phenotypes and increase longevity in Mecp2 KO mice (149). One explanation for its beneficial effect is that ketamine is an activity-dependent NMDA receptor blocker, which likely targets networks that are hyperactive and involve NMDA receptor-dependent synaptic transmission. In pyramidal cells of the visual cortex, loss of Mecp2 leads to increased innervation by GABAergic inhibitory neurons and an increased expression of the GluN2A NMDAR subunits in the visual cortex. Thus, although Mecp2 mutation differentially affects NMDAR expression in different cortical regions, it is conceivable that targeting these receptors will ameloriate some RTT phenotypes by reestablishing the balance of excitation (94, 101, 149, 176).

Neocortical hypoactivity also involves an elevation of postsynaptic inhibition (37). Conditional knockout tools and subsequent rescue interventions reveal that parvalbumin positive inhibitory interneurons are disrupted in the Mecp2 KO mouse (16, 103), and selective deletion of Mecp2 in PV+ neurons leads to a similarly increased inhibitory tone as a global KO. The effect in PV+ neurons was attributed to a decrease in the K/Cl cotransporter KCC2. This discovery led to the screening and identification of small molecules that enhance KCC2 transcription, resulting in rescued electrophysiological and morphological neuronal properties in cultured human RTT stem cells. Moreover, systemic injection of this KCC2 transcription enhancer in Mecp2(–/y) mice rescued the breathing phenotype (197).

These manipulations reveal that targeting excitability imbalances in hippocampus and neocortex also improves respiratory dysfunctions. An intriguing possibility is the direct targeting of homeostatic mechanisms. An important mediator of homeostatic plasticity is insulin-like growth-factor 1 (IGF-1) (205). It stimulates RIT1-dependent kinase, which has been implicated in activity-dependent plasticity, and IGF1 has also been studied in Mecp2 KO mice (16, 25, 122). MECP2 activates IGF-1, which in turn rescues synaptic function in RTT mice and restores excitatory balance from hypoactivity (33, 204). RTT mice treated with IGF-1 show improved behavioral phenotypes, including breathing (33, 224). An analog of IGF-1, Trofenitide, recently has been shown to improve core symptoms in RTT patients over placebo (69), and a phase III clinical trial is underway (FIGURE 2). These core symptoms include repetitive movements, seizures, and mood dysfunction, as well as respiratory dysfunctions (69, 70, 153).

FIGURE 2.

Disruption of excitatory-inhibitory balance in Mecp2 KO (knockout) mouse model

In normal, wild-type (WT) mice, the activity of excitatory and inhibitory neurons is delicately balanced, and the overall excitability of the entire network is tightly regulated. A: network diagrams illustrate the self-regulation of excitatory-inhibitory (E-I) balance in WT mice. Excitatory neurons (blue circles), inhibitory neurons (red circles), and their synapses (lines) are shown. Circle diameter indicates activity level, and line color indicates presynaptic identity. An example spiking trace from an excitatory neuron [in this case a L5 pyramidal neuron (35)] is shown. Normally, in WT animals, E-I is balanced. In cases in which a perturbation occurs, network activity levels may change, shown here by concomitant increase in E and I activity (middle). Homeostatic plasticity mechanisms work to return the global network state to a normal level of I-E activity and balance, although individual connections and cell activities may have changed. B: in Rett Syndrome, the balance is shifted toward hypoactivity and hypoconnectivity. This is shown as a reduction in excitatory activity (blue circle diameters) and number of synapses (lines). As in A above, an example trace of spiking activity from a L5 pyramidal neuron in a Mecp2 KO mouse is shown (35). The overall E-I balance is shifted to inhibition, and the network activity is decreased in the Mecp2 KO mouse; plasticity is impaired, and the network lacks the normal ability to achieve E-I balance or respond to perturbations normally. Potential therapeutic mechanisms are listed in the green shaded boxes. Administration of IGF-1 analogs (Trofenetide) and NMDA agonists (Ketamine) both not only restore EI balance in forebrain networks but additionally alleviate a number of clinical symptoms. An increase in excitatory postsynaptic current (EPSC) size under IGF-1 administration is one potential mechanism of E-I balance restoration, shown in the middle of the figure.

Breathing Abnormalities in RTT Patients

More than 80% of patients with classic and atypical severe RTT experience breathing dysfunction during their lifespan (31, 71, 200). The proportion of children with respiratory issues varies based on age, and peaks between 6 and 11 yr of age, followed by a period of remission (peak at 11–13 yr) and relapse at any age (199). It is essential that alterations in breathing pattern be differentiated into those occurring in wakefulness or sleep. Breathing dysfunctions during wakefulness can be classified into two distinct types: 1) erratic breathing, which is characterized by episodes of hyperventilation and irregular breathing that can be shallow and/or forceful (216); and 2) breathholding, which occurs in an episodic manner (216). Defining breathholding has been challenging. Sometimes referred to as “apneas,” these breathholds should not be characterized as “central apneas,” because the respiratory drive for breathing does not simply cease. Additionally, RTT patients do have sleep disturbances (108) and a high prevalence of obstructive sleep apneas (OSAs) during sleep (5, 18, 175). But these sleep apneas differ from the more typically awake breathhold events (31). When awake, these patients take a breath and then fail to release the air. However, lung volume is not maintained by prolonged, persistent inspiratory drive. Thus these events are different from “apneusis” (59, 218). In the cohort of RTT children, the breathhold begins with an apparently normal inspiration, followed by a brief postinspiratory phase, during which the heart rate decreases, before the patient transitions into the actual breathhold phase that is accompanied by a dramatic heart rate increase. This tachycardia is absent during breathholds of healthy subjects, which are characterized by a “diving response” and bradycardia (10, 163). By contrast, the breathholds in RTT are reminiscent of Valsalva maneuvers, i.e., forced exhalations against a closed glottis. Indeed, Andreas Rett first published this phenomenon in German as “Pressatmung” to describe the characteristic pushing (“Press”) of air against the glottis among young girls with RTT. Yet, breathholds in RTT seem different from Valsalva maneuvers in healthy subjects, because tachycardia becomes uncoupled from breathing as normal breathing resumes. The heart rate returns to baseline only later, after the second or third normal breath. This is not necessarily expected if tachycardia is reflexively caused by the increased chest pressure, as typically assumed for healthy subjects. However, even reflexive responses are never simple. The increased chest pressure triggers a series of events that include changes in the pulmonary circulation and cardiac output, and a compensatory tachycardia. Thus Valsalva maneuvers vary, depending on lung volume even in healthy individuals (10, 109). Further studies aimed at directly comparing Valsalva maneuvers and breathholds in RTT and healthy subjects should help to unravel potential differences in the physiological and pathophysiological characteristics of these breathholds. The episodic nature of breathholds in RTT is particularly interesting and points to disturbances in CNS mechanisms. Breathholds are separated by inter-breathhold intervals that vary in duration. The time between two consecutive breathholds tends to be longer if the duration of the preceding breathhold was long. Heart rate measured at exhalation onset tends to be higher after long-duration breathholds compared with shorter-duration breathholds (216). Episodic tachycardia events also occur during sleep (215), but the tachycardia precedes paradoxical breaths and does not follow breathholds, like in wakefulness. Thus we hypothesize that the tachycardia events occurring during wakefulness and sleep are caused by related disturbances in the central control of cardiorespiratory coupling (31).

Mechanisms Underlying Breathing Abnormalities

Insights into the mechanisms underlying these breathing abnormalities were gained by studying Mecp2 global and targeted knockout (KO) mice. These mice show respiratory irregularities with frequent periods of prolonged respiratory cycles that are reminiscent of breathholds in RTT patients (164, 212). The mean respiratory frequency is increased in some (91) but not all Mecp2 animal models (164), and mutant mice show reduced ventilatory CO₂ sensitivity (23, 227). The central mechanisms associated with these breathing alterations are thought to include increased excitability in respiratory areas like the pontine Kölliker-Fuse nuclei (KF) (193), the nucleus tractus solitarius (nTS) (99, 101), locus coeruleus (196), and ventrolateral medulla (83, 120, 213). It has been hypothesized that the respiratory dysfunction involves the interaction between nTS-mediated afferent feedback and pontine activity (220). The nTS is a key relay that receives important peripheral sensory inputs, including O₂ and lung inflation information (FIGURE 3) (99, 164). The KF forms a group of respiratory neurons that are located in the dorsolateral pons and send excitatory inputs to neurons within the medullary ventral respiratory column. These neurons regulate postinspiratory activity in phrenic and cervical vagus nerves, and are thought to control the transition from inspiration to expiration (55, 189), control laryngeal adductors and airflow during speech, and protect lungs from toxic substances (114, 120) or aspiration (121), as described below. The KF is known to excite upper airway musculature contraction during the postinspiratory phase of expiration (55), which may involve the activation of excitatory postinspiratory neurons in the post-inspiratory complex (PiCo) (9). During the breathhold in mouse models, the postinspiratory phase of vagus activity (cVN) is restricted to the initial portion, whereas superimposed tonic discharge of hypoglossal activity (HN) extends throughout the event, and a temporal relationship also exists for the onset of tonic discharge of abdominal activity (AbN) relative to HN (3, 51). The association of tonic HN with active expiration and post-inspiratory vagal activity (2) is reminiscent of the breathhold activities seen in patients, and it is conceivable that related disturbances also contribute to loss of speech (median age 18 mo) (58, 111) and impaired swallowing (193, 194).

FIGURE 3.

The dynamic interplay between breathholds and irregular breathing in RTT

Hypothesized interactions are represented in this schematic. Episodic breathholds generated by disturbances in the Kolliker-Fuse (KF) regions of the pons lead to intermittent hypoxia, which may be a critical driver for irregular breathing. The effect of intermittent hypoxia is exaggerated by mitochondrial dysfunctions and reactive oxygen species (ROS) production. This disturbed REDOX regulation leads to carotid body hyperactivity, which, via an already disturbed nTS, will induce irregular breathing in the ventral respiratory group (medulla VRG). Right: each dot in the two graphs represents the irregularity score of an individual patient (RTT) or control subject. Note that not every patient with RTT shows irregular breathing (pink shaded box), and many RTT patients have irregularity scores similar to those of controls (green shaded boxes). Normal breathing is maintained by homeostatic mechanisms. Graphs and traces modified from Refs. 214, 216, with permission from Pediatric Pulmonology and Pediatric Research, respectively.

Despite strong evidence for the role of the pons in contributing to the breathing abnormalities in Rett syndrome, elegant genetic manipulations in rodent models indicate that disturbances in the medullary networks are critical for explaining some of the severe autonomic phenotypes and the lethality observed in MeCP2 mutant mice (213). Animals that lack MeCP2 in medullary regions developmentally derived from HoxB1 domain show early lethality, have a progressive decline in their heart rate, and on the day of death exhibit a decline in basal temperature. The death was not associated with an obvious increase in epileptiform or seizure activity, indicating that these mutants do not die of sudden death of epilepsy. Interestingly, the increased baseline respiratory frequency seen in the null mouse persisted in mice in which MeCP2 was only removed in the medullary network. This medulla-specific removal also suggested that the medullary network contributes to the abnormalities seen in response to hypoxia (213).

Further insights into the role of MeCP2 were gained by genetic manipulations that specifically targeted a smaller subset of the respiratory network located in the medulla, regions that are defined by the HoxA4 domain (83, 84). This domain encompasses portions of the NTS and the caudal portion of the respiratory network, i.e., including and adjacent to the preBötC such as the so-called rostral VRG. Restoring MeCP2 functions specifically in these medullary regions was sufficient to prevent the development of apneas, even if MeCP2 is lacking in pontine and suprapontine structures. This manipulation also increased survival compared with the full knockout mice. However, interestingly, removing MeCP2 from these caudal regions did not increase the number of apneas or irregular breathing. Although the interpretation of these specific manipulations seems complicated, these specific manipulations suggest that Mecp2 in the medullary HoxB1 domain rostral to the HoxA4 domain is critical for the development of the apneas (83). Interestingly, this area includes the RTN, the BötC, the rostral NTS, as well as the newly described PiCo area (9). The RTN and PiCo have been implicated in the generation of active expiration and postinspiration, respectively (9, 85). Disturbances in these two respiratory phases seem to be critical for the breathholds (3, 51). In conclusion, these genetic studies provide important and very specific insights into differential contributions of different medullary regions to the increased lethality and the breathing disturbances. This precise localization is an important step toward unraveling the underlying cellular and modulatory mechanisms.

Among the neurotransmitters, neuromodulators, and neurotrophins found to be deficient in the global Mecp2 KO mice are GABA (37), noradrenaline (norepinephrine) (212), serotonin (1, 212), and BDNF (90, 99, 102, 179). The deficiency in GABAergic and/or serotonin-mediated inhibition seem to be particularly important for the disturbed interactions in breathing networks described above (1, 3, 54, 193). Serotonin (5-HT) is an important regulatory neurotransmitter in the respiratory network (151), and loss of 5-HT1A-mediated inhibition could induce respiratory cycle irregularity (52). Indeed, patients with identified MECP2 mutations show low spinal fluid levels of a 5-HT metabolite (174), and low 5-HT levels progressing with development were identified in the brain of Mecp2 KO male mice (86, 212). 5-HT is also a significant contributor to ventilatory response to CO₂ (168), and global deletion or acute inhibition of serotoninergic neurons results in blunted ventilatory responses (80, 166). Yet, the impairment of GABAergic mechanisms in the KF seems to also be critically involved in RTT (2), since substances that enhance GABAergic mechanisms can also reduce breathhold events, correct breathing irregularities, and restore chemosensitivity (2, 202). Mechanistically, MECP2 might be directly connected to developmental changes in the expression of NMDA and GABA receptor subunits (173, 223), which then may cause over-excitability of post-inspiratory neurons (193).

Breathing Abnormalities and Oxidative Stress

In RTT, the majority of patients develop some respiratory dysfunction before age 4. Seventy-five percent do so by 5.6 yr of age for breathholding and 8.7 yr for hyperventilation. Although breathholding appears on average before hyperventilation (111), the development of the clinical phenotype is nonlinear and interdependent with other clinical conditions (FIGURES 1 AND 3), like the anxiety behavior that is characteristic for RTT (1). The episodic appearance of breathholds triggers a cascade of other events that will inevitably lead to intermittent hypoxia (FIGURES 1 AND 3) (47, 193, 212). Exposure to intermittent hypoxia leads to over-activity in carotid bodies (132, 134), which could aggravate hyperventilation episodes in between breathholds. This will be further exaggerated by the known reduced ability of astrocytes to sense changes in Pco₂/[H⁺] in Mecp2 mice (206). Intermittent hypoxia itself desynchronizes the central respiratory network, which causes irregularities in the respiratory rhythm (65). Respiratory irregularities are observed in Mecp2 mice (212) and Rett patients (216). Interestingly, there is large interindividual variability among RTT patients, and measuring an irregularity score reveals that the breathing of some RTT is not more irregular than in control participants (FIGURE 3). Whether homeostatic mechanisms are capable of normalizing breathing in some but not all RTT patients is an open question. It remains to be determined whether those patients with breathing irregularities also demonstrate an increased prevalence of breathholds, OSA, and intermittent hypoxia. Animal studies have shown that intermittent hypoxia causes desynchronization within the respiratory network, which increases the probability for transmission failures to the hypoglossal nucleus (64, 65). Consequently, prevalence of OSA among RTT patients who suffer from breathholding should be determined, and objective measures should be obtained for the intermittent hypoxia “dose” and timing relative to age, caused by OSA as well as the respiratory dysregulation (65). Intermittent hypoxia also leads to disturbances in sympathetic and parasympathetic control (161, 162), which likely contributes to the dysautonomia characteristic of RTT (31, 185).

The effects of intermittent hypoxia are largely mediated by reactive oxygen species (ROS) and oxidative stress (64, 97, 133, 180), which is further aggravated in RTT because of altered mitochondrial structure and enzyme activity: mitochondria are swollen, have altered respiratory rates, and show proton leakage (FIGURE 3) (20, 41, 73, 129, 150, 183, 210). Indeed, there is ample evidence for oxidative stress in human RTT patients (39, 46, 48, 184, 209). Targeting ROS with antioxidants is one therapeutic approach that could prevent some of the phenotypes associated with oxidative stress, as suggested by the use of ω-3 PUFAs in a study with 113 RTT female patients (107).

Dysphagia and Aspiration in RTT Patients

Breathing disturbances may be mechanistically linked to other disturbances typically seen in RTT. Andreas Rett observed that difficulty swallowing liquids (in German “Trinkschwäche”) were the first clinical signs described in 30% of the 22 girls before the age of 9 mo. We now know that 70% of RTT patients develop dysphagia (disordered swallow) (121, 128). As a consequence, most patients eat modified diets, with thickened liquids and puree being the easiest and safest foods for oral ingestion (4), since thin liquids have proven high risk of aspiration and solids are difficult to chew (27). Dysphagia can be caused by different pathophysiological mechanisms, and, in RTT, patients are challenged by numerous coordination issues that make eating extremely difficult (87, 128). Swallowing is a complex behavior, requiring the coordination of 26 pairs of muscles and 5 cranial nerves (17), and is accomplished in three distinct phases: oral, pharyngeal, and esophageal (FIGURE 4). The few studies that evaluated dysphagia in RTT suggest that several factors impair this first (oral) phase. Initially, there is breakdown of the food into a bolus, which involves preparatory activity, chewing and lateralization of the tongue (118), followed by propulsion, during which the tongue moves the bolus to the back of the pharyngeal wall, triggering the pharyngeal phase (118). RTT patients have several issues that affect this first phase, including malocclusion (open bite) (40, 121), difficulty chewing (4, 40, 124), poor tongue mobility (40, 121, 125, 128), involuntary movements of the tongue (40, 121), poor bolus formation (124), and poor glossopharyngeal seal. All these factors lead to complications in the subsequent pharyngeal phase of swallow, which include pharyngeal spill, pharyngeal delay, and impaired oropharyngeal clearance (128), resulting in potential aspiration. Partial deglutition (swallow) or “failed swallow” is common due to poor bolus formation and the inability to compress the bolus to the soft palate to push the bolus to the back of the pharyngeal wall (124). It is also noted that, in RTT, the pharyngeal cavity over distends (124), requiring more food/liquid to fill the pharyngeal cavity before the pharyngeal phase of swallow is initiated. Both of these responses increase oral transit time and delay pharyngeal swallow (121, 128). Schluckatmung, German for the “swallow-breath” activation of the diaphragm muscles at the onset of swallow, creates a negative trans-diaphragmatic pressure, transporting the bolus from the pharyngeal cavity across the upper esophageal sphincter and into the esophagus (159) (FIGURE 4). RTT patients also experience impairment of the third phase of swallow, i.e., the esophageal phase. Specifically, reports suggest the absence of so-called primary or secondary motility waves and presence of gastresophageal reflux (GER) (128). Aerophagia (air swallowing) (111, 126), sensory deficits, and texture intolerance (87) are also manifestations of RTT.

FIGURE 4.

Tracings and anatomic locations depicting normal and discoordinated swallow-breathing behavior

A: muscle electromyograms (EMG) traces and pressure measurements depict normal sequential swallow pattern in the healthy adult rat with the colors coordinated to different muscles throughout the aerodigestive tract. B: anatomic depiction of the aerodigestive tract involved in both breathing and swallow. C: swallow-breathing coordination in a healthy human with spirometry airflow trace in pink (downward deflection indicating inspiration, and upward indicating expiration) with zero airflow, along with surface EMG submental complex activation (orange), indicating swallow. The leading complex (mylohyoid and geniohyoid in rat) or submental complex in the human indicates the oral phase (orange) and correlates to the anatomic structure listed in orange (B). The thyroarytenoid and thyropharyngeus in blue (A) measures the pharyngeal phase of swallow listed in blue (B). The esophageal pressure, red (A) depicts the esophageal phase and is the last in the swallow sequence. Activation of the diaphragm (in pink) during swallow occurs during leading complex activation. D: bolus travel in normal swallow (red arrow) bolus travels from pharynx to esophagus. E: spirometry (pink) and respiratory inductive plethysmography bands (purple) around the chest and abdomen show swallow-breathing coordination in a patient with Rett syndrome. Normal eupnea occurs until the spoon comes in contact with the patient; lack of airflow and no movement in the chest or abdomen represent the apneic period; and swallow occurrence at the end is depicted by the flat line at the end of the feeding session, followed by normal eupnea. Dysphagic swallow, penetration (green arrow) of the bolus passes to the larynx but does not go below the vocal folds, whereas in more severe aspiration (blue) the bolus passes through the vocal folds and into the trachea leading to the lungs. D and E are modified from Ref. 125, with permission from Developmental Medicine and Child Physiology.

Mechanisms Underlying Dysphagia

From the neural perspective, there is considerable overlap between the central neurons controlling breathing and swallowing (22, 55, 60, 66, 88, 147, 159, 172), and disturbances in the coordination between these two behaviors underlies aspiration (115, 116). Yet, despite the important clinical relevance, the mechanisms leading to dysphagia in RTT are poorly understood. Swallowing depends on precise proprioceptive control. Stimulation of mechanoreceptors in the pharyngeal wall trigger peristalsis, whereas pharyngeal constrictor muscles move the bolus to the esophagus as laryngeal adductor muscles close the larynx, vocal folds, to ensure foreign material does not enter into the airway (14, 118). As the bolus approaches the upper esophageal sphincter, it relaxes, allowing the bolus to pass through entering the esophagus—marking the esophageal phase—and continues until the bolus moves through the lower esophageal sphincter and into the stomach (88). To the best of our knowledge, it is unknown whether proprioceptive control is disturbed in RTT, and little is known with regard to the central control of swallow in RTT. Swallow generation and coordination seems to involve two distinct neuronal populations that are located in the medulla, some of which seem to overlap with neurons involved in breathing (60, 66, 88, 147, 159, 172). The so-called dorsal swallow group, located in the nTS, is responsible for processing cortical and peripheral information (88). This presumed swallow central pattern generator (CPG) in the nTS is densely connected with the Kölliker-Fuse (KF) in the dorsolateral pons (19). The so-called ventral swallow group located in the ventral lateral medulla dorsal to the nucleus ambiguus is thought to distribute drive to swallow-related motoneuron pools (88). It has been demonstrated that the amplitude of spontaneous miniature and evoked EPSCs is significantly increased in the nTS (99). Thus afferent input into this nucleus will likely evoke relatively more action potential firing in RTT. This synaptic imbalance is also associated with decreased BDNF availability in the primary afferent pathway and can be rescued by exogenously applied BDNF (99). To what extent the particular nTS region characterized in this electrophysiological study also includes the circuits responsible for swallow is not entirely clear. Furthermore, the nTS and also the KF have not only been implicated in swallowing but also in other autonomic disturbances in RTT (3), including cardiorespiratory control. This interrelationship between breathing and feeding disturbances is also suggested by the finding that participants with higher BMI Z-scores were less likely to exhibit hyperventilation (80.2% vs. 87.6%) and breathholding (89.2% vs. 96.1%) (199). Thus the same synaptic pathology may contribute to various disturbances that are characteristic for RTT. Remarkably, it has been observed that the oral motor patterns of the tongue mirrored the motor patterns of the hands (124). In patients with hypotonia and normal tone, stereotypies of the hands were accompanied by tongue stereotypies (28); and patients with more rigidity/hypertonia whose hands lay motionless were also motionless in their tongue (124). These observations suggest that RTT is characterized by synaptic and possibly also modulatory imbalances that concurrently affect several rhythmogenic networks. This conclusion is consistent with the above-mentioned observation that targeting cortical imbalances has far-reaching effects on a variety of clinical phenotypes. This raises the hope that reinstating the homeostatic balance between excitation and inhibition could lead to the concurrent improvement of several of these devastating symptoms at the same time.

However, targeting neuronal mechanisms alone will likely be insufficient, given that pneumonia is one of the most frequently reported medical conditions (7) and the most common cause of death in RTT (104). To the best of our knowledge, it is not clear whether pneumonia is the result of predisposition to lung conditions or due to orally ingested versus secretion-related aspirations. This is an important clinical issue, because a G tube could prevent orally ingested aspiration but would not prevent aspirations caused by lung conditions. According to some clinical reports, it appears that aspiration has low occurrence in early stages of RTT, but occurrence may increase with increasing progression of the disorder as well as age (124). Clinically, aspirations are rated on a penetration-aspiration scale. The severity of dysphagia is assessed by characterizing bolus movement during swallowing: specifically, by assessment of whether material passes below the level of the vocal folds (aspiration) or whether material passes into the larynx but does not pass below the vocal folds (penetration) (171) (FIGURE 4). In the event that foreign material penetrates the larynx in the healthy human, coughing expels and clears the foreign material back to the oral cavity, and the material is either removed or swallowed. Across many clinical studies, 11% of patients experience aspiration when 1) they drink thin liquids and 2) they are in later stages of the disease (4, 40, 121, 124). The majority of patients experience penetration of the bolus but are able to clear the pharyngeal space appropriately, avoiding aspiration (4, 128) due to effective cough reflex (121). Indeed, in fiberoptic endoscopic evaluation of swallow (FEES), most RTT patients have a normal pharyngeal swallow mechanism (121). In contrast to these human patient data, aspiration is much more frequent (64%) in Mecp2-null mutant mice (98). One possible explanation for this difference is that rodents are unable to cough, whether healthy or diseased (21). The work in the mouse model also suggests that MECP2 deficiency induces emphysema-like structural changes in the lung. This would suggest that pneumonia could be caused by abnormalities of the pulmonary system (98). Fifty percent of Mecp2-null mice were found with inflammation of the lung (98). Since videofluoroscopy and FEES studies have not been conducted in the Mecp2 mouse model and rodents do not have the natural ability for spontaneous gastresophageal reflux (117), one assumes postmortem aspiration diagnosis, in a mouse model, is due to aspiration from swallowing. Since GER affects 39% of the patient population (61, 127), it is conceivable that RTT patients aspirate not only from swallow-related but also from GER-related events.

RTT patients also exhibit poor coordination between breathing and chewing, which results in increased air in the pharyngeal space that is swallowed with the bolus (126) and aspiration of foreign material into the lungs (125). To avoid this, RTT patients have apneic periods lasting 10–45 s during feeding (FIGURE 4). Normal respiration resumes once swallow has safely occurred and the pharynx is cleared of any material (125, 126). Both human RTT and Mecp2 mice experience aerophagia and malocclusion (40, 98, 126). Thus swallow-breathing coordination in RTT is loosely coupled and in some cases absent (125). Many RTT patients are preferential mouth breathers (121), which probably contributes to the increased prevalence of aerophagia (air swallowing) causing painful distention of the stomach and even bowel perforation (111, 126). According to the natural history study, 47.2% of Rett parents also reported a history of air-swallowing during wakefulness (199). Although aerophagia is seen in other neurological and oropharyngeal diseases, it seems to be most common with RTT (126). From the nervous system perspective, it is conceivable that aerophagia and breathholding in RTT are caused by similar pathophysiological mechanisms. Both of these behaviors occur often in the same individuals (79, 126). For example, one study reported that 17 of 20 patients who showed aerophagia also showed breathholding, and three gulped air during hyperventilation (126). Mechanistically, breathholds seem to be caused by serotonergic disturbances in the pons, in the KF in particular. It would be interesting to know whether breathholds and aerophagia have some common mechanistic roots in the brain stem, since this could have therapeutic implications.

Conclusions

Genotyping has become common practice in diagnosing RTT, whereas genetic manipulations in mouse models have become common research tools aimed at unraveling the mechanisms underlying RTT. Conditional gene knockout of specific neurons in specific brain regions, combined with rescue interventions and comparison with global knockouts, provide important insights into the contribution of specific neurons and regions to the clinical phenotype. This approach has also contributed to the identification of promising new drugs that are currently being tested in clinical trials. However, the genotype-phenotype relationship of RTT is particularly complex. The complexity is partly due to the fact that mutations are only a starting point for multiple cascades of interactive changes in physiology that include not only the brain but all organ systems. For simplicity, we chose to consider only some of the affected organs in this review. But, from these considerations, it is clear that an excitability imbalance in the neocortex indirectly contributes to changes in breathing and swallowing, and conversely breathholding causes intermittent hypoxia, which has multiple consequences that are exacerbated by concurrent disturbances in mitochondria and redox control (FIGURES 1 AND 3). Disturbances in swallowing that result in aspiration can lead to lung inflammation, which in turn triggers inflammatory responses in the nervous system and may be exaggerated by disturbances in the inflammatory control itself. It will be difficult to unravel all the details of this dynamic interplay, in particular since multiple mutations and genes can contribute to RTT. Moreover, every individual patient has a different spectrum of x-inactivation, different history, and different predispositions, thus there is great interindividual variability in the clinical presentation (e.g., FIGURE 3).

However, there are also positive aspects in this complexity. Lessons learned so far show that targeting one transmitter system, a specific neuromodulatory receptor, a growth factor, or an anti-oxidant system can have a multitude of beneficial effects. Targeting one system can alter the trajectory of the RTT phenotype in a variety of ways. Indeed, the astonishing reversal of many phenotypes by genetically replacing Mecp2 in mice illustrates the power of homeostatic mechanisms in rebalancing these complex systems (FIGURE 2), even at a late clinical stage. Thus, although gene replacement in humans is not yet a reality, it is possible to target the homeostatic mechanisms that are disturbed in RTT. Trofinetide seems to be one of those substances that raises considerable hope. Targeting homeostatic mechanisms could be particularly powerful if combined with therapeutic approaches that also control the indirect consequences of the clinical phenotype, as discussed in this review. These indirect consequences arise from oxidative stress and inflammatory imbalances.