REVIEW article

Front. Mol. Neurosci., 29 November 2022

Sec. Neuroplasticity and Development

Volume 15 - 2022 | https://doi.org/10.3389/fnmol.2022.1072475

Transcription factors regulating the specification of brainstem respiratory neurons

  • 1. The Brainstem Group, Institute for Cell Biology and Neurobiology, Charité Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Berlin, Germany

  • 2. Functional Genoarchitecture and Neurobiology Groups, Biomedical Research Institute of Murcia (IMIB-Arrixaca), Murcia, Spain

  • 3. Department of Human Anatomy and Psychobiology, Faculty of Medicine, University of Murcia, Murcia, Spain

  • 4. Developmental Biology/Signal Transduction, Max Delbrück Center for Molecular Medicine, Berlin, Germany

Abstract

Breathing (or respiration) is an unconscious and complex motor behavior which neuronal drive emerges from the brainstem. In simplistic terms, respiratory motor activity comprises two phases, inspiration (uptake of oxygen, O2) and expiration (release of carbon dioxide, CO2). Breathing is not rigid, but instead highly adaptable to external and internal physiological demands of the organism. The neurons that generate, monitor, and adjust breathing patterns locate to two major brainstem structures, the pons and medulla oblongata. Extensive research over the last three decades has begun to identify the developmental origins of most brainstem neurons that control different aspects of breathing. This research has also elucidated the transcriptional control that secures the specification of brainstem respiratory neurons. In this review, we aim to summarize our current knowledge on the transcriptional regulation that operates during the specification of respiratory neurons, and we will highlight the cell lineages that contribute to the central respiratory circuit. Lastly, we will discuss on genetic disturbances altering transcription factor regulation and their impact in hypoventilation disorders in humans.

Introduction

In vertebrates, the developing brainstem generates an enormous diversity of neuron types that control bodily homeostasis and process multiple modalities of sensory information (; ; Venkatraman et al., 2017). These neurons vary not only in their morphological, chemical, and electrophysiological properties, but also in their connectivity patterns that allow them to form elaborate circuits, such as those required to generate, monitor, and adjust breathing to meet various external and internal physiological demands. How this neuronal diversity emerges during development and how it contributes to functional circuits has been intensively investigated by many generations of neuroscientists, as well as by cellular, molecular, and developmental biologists.

The work of these brainstem enthusiasts has already revealed that neuronal diversity in this brain region depends on the temporal and spatial patterning of local neural progenitors (, ; ; ; van der Heijden and Zoghbi, 2020; ). This patterning is first achieved by early diffusible cues that impose an anterior-posterior identity, and subsequently by other morphogens that provide a distinctive dorsal-ventral molecular signature to progenitor cells. This means that distinct progenitor cells can be distinguished primarily based on their differential expression of numerous transcription factors that commit them to generate specific neuron types. Most transcription factors normally expressed in progenitor cells are later silenced in their progeny, which allows for the postmitotic maturation of the differentiated neurons. At the same time, each neuron type can also be characterized by the expression of particular sets of transcription factors that form part of their cell physiology and identity.

In this review, we aim to summarize our current knowledge on the transcriptional programs that allow for the speciation of brainstem respiratory neurons. We will first present a general overview of the anterior-posterior and dorsal-ventral patterning of the developing brainstem as an entry point to understand its neuronal diversity. Next, we will focus on the specification of three large groups of neurons that play key roles in respiration: (i) the pontine groups, (ii) the dorsal medullary respiratory column, and (iii) the ventral medullary respiratory column. These neurons locate to the pons and the medulla oblongata where they perform a variety of functions, such as respiratory rhythm generation, respiratory modulation, and tissue gas monitoring. Lastly, we will discuss some genetic disturbances that affect breathing and respiratory neuron specification.

Anterior to posterior patterning of the developing brainstem

Brainstem development is an evolutionary conserved process that begins with the specification of the mesencephalon (midbrain) and rhombencephalon (hindbrain) by the isthmic organizer. This organizer is located at the midbrain-hindbrain border and produces diffusible morphogens (particularly Wnt1 and Fgf8 ligands) that act directly on the neighboring nervous tissue to impose midbrain and hindbrain cell fates (Zervas et al., 2004; ; ; ; ). Various transcription factors are differentially expressed anterior and posterior to the isthmic organizer, such as the antagonistic homeodomain factors Otx2 (anterior) and Gbx2 (posterior), whose expression defines the rostral and caudal regions of the developing central nervous system (Figure 1A; , ; ; ; ; ; ). Alterations in the expression of Otx2 and Gbx2 are catastrophic for early brainstem development. For instance, the ablation of Otx2 and its closely related family member Otx1 results in the loss of midbrain tissue, which becomes re-specified into more rostral hindbrain-like regions, such as the cerebellum (; Suda et al., 1997; ). Similarly, the misexpression of Gbx2 disrupts the correct positioning of the isthmic organizer and the development of rostral hindbrain (; Wassarman et al., 1997; ; ; Waters and Lewandoski, 2006).

FIGURE 1

that was published under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution, provided that the original author and source are credited (https://creativecommons.org/licenses/by/4.0/).

The regulation of both Otx2 and Gbx2 largely depends on a complex molecular network that centers on the instructive signals of Fgf8 (Figure 1B; ; ; ; ; ; ). About embryonic (E) 7.5 in mice, Fgf8 expression is activated by the diffusible ligand Fgf4 (from the notochord) that directly induces the presumptive midbrain tissue to express the transcription factor En1, which in turns activates Fgf8 expression in the isthmic organizer (Shamim et al., 1999). Two Fgf8 isoforms appear to differentially act on the specification of the midbrain and rostral hindbrain. Fgf8a safeguards midbrain identities, while Fgf8b directs rostral hindbrain development. It is important to note that the expression of both isoforms is indispensable for the development of the midbrain and the hindbrain (; , ; Sato et al., 2001; ). Notably, the duration of Fgf8 expression in the isthmic organizer is crucial, as available evidence shows that its sustained expression is required to restrict Otx2 rostral to the isthmic organizer, while simultaneously maintaining Gbx2 caudal to it (; ; Sato and Joyner, 2009). In addition to controlling the expression patterns of Otx2 and Gbx2, Fgf8 induces the expression of the homeodomain transcription factor Lmx1b, which primary function is to stabilize Wnt1 expression in the isthmic organizer. In this regard, several studies show that the concomitant ablation of Lmx1b and its family member Lmx1a severely alters the specification of the hindbrain, which adopts a “spinal cord-like” fate in Lmx1a and Lmx1b double mutant mice (; Su et al., 2014; ; ). Thus, the establishment of the midbrain–hindbrain border, a prerequisite for brainstem development, relies on a complex molecular network of various transcription factors and signaling cascades (briefly summarized in Figure 1B).

Soon after the establishment of the isthmic organizer, the hindbrain undergoes a series of morphological changes that transiently divide it into seven or eight smaller segments called rhombomeres, from which the cerebellum, pons and medulla oblongata emerge (; ). In mice, these segments are recognizable at E8.5, whereas in humans they appear by E29 (Figure 1A; ). Each of these rhombomeres develops a specific set of cellular and molecular features that distinguishes them from the adjacent nervous tissue. Although still unclear, recent anatomical studies indicate that some of rhombomeres might be further regionalized according to the expression of some patterning genes (Tomas-Roca et al., 2016; Watson et al., 2017, 2019; ). An interesting trait in the “rhombomerization” of the hindbrain is the differential expression of the Hox superfamily of transcription factors, whose expression creates molecular codes that coincides with the morphological borders of each rhombomere (Figure 2; ; ; ; ). One should note that these transcriptional codes are not limited to the hindbrain, as some members of the Hox family are differentially expressed in the developing spinal cord, from which the characteristic cervical, thoracic, lumbar, sacral and coccygeal levels emerge (). In addition to Hox genes, several other transcription factors show rhombomeric specific expression patterns during early hindbrain development, such as Pax2 in rhombomere 1, Meis2 in rhombomeres 2 and 3, Egr2 (formely known as Krox20) in rhombomeres 3 and 5, and MafB (previously called Kreisler) in rhombomeres 5 and 6 (Figure 2; Seitanidou et al., 1997; Rowitch et al., 1999; , ; Voiculescu et al., 2001; ; ; ; ; Stedman et al., 2009). The timely expression of hox genes, as well as the above mentioned transcription factors, is essential for the correct development of each rhombomere and significantly depends on active derivatives of vitamin A, such as retinoic acid (). Indeed, dietary deficiencies in vitamin A produce gross disturbances in hindbrain “rhombomerization” that can lead to the complete absence of caudal rhombomeres, as manifested by the loss of specific genes and neuron types normally produced in these regions (; ; ; ). Thus, early brainstem development relies on the function of diffusible ligands (i.e., Fgf8 and retinoids) that determine its anterior-posterior identity.

FIGURE 2

. (B) A sagittal section of the mouse brainstem stained with antibodies against mCherry and counterstained with DAPI. The section is taken from Egr2Cre/+;H2BmCherry/+ mice at birth (P0). In these mice, only r3 and r5 neural derivatives express the nuclear mCherry protein after Cre mediated recombination. This figure is adapted from our previous publication in Handbook of Clinical Neurology, Chapter 5, entitled Early development of the breathing network, published by Elsevier Books. The license number 5392080585902 between Hernandez-Miranda, Charite Universitätsmedizin Berlin and Elsevier allows us to reuse it in a journal/magazine.

Dorsal to ventral patterning of the developing brainstem

Once the brainstem acquires its anterior-posterior identity, a second series of diffusible cues further pattern the identity of progenitor cells along its dorsal-ventral axis. During this patterning, the ventralizing Sonic hedgehog (produced by the floor plate), as well as the dorsalizing bone morphogenetic proteins and Wnt ligands (secreted by the roof plate) create concentration gradients that differentially signal onto brainstem progenitor cells (Roelink et al., 1995; ; , ; ; Ulloa and Marti, 2010; ). These gradients create a great diversity of molecularly distinct progenitor domains that vary depending on their spatial distance from the signaling source (Figure 3 and Table 1).

FIGURE 3

in Handbook of Clinical Neurology, Chapter 5, entitled Early development of the breathing network, published by Elsevier Books. The license number 5392080585902 between Hernandez-Miranda, Charite Universitätsmedizin Berlin and Elsevier allows us the reuse of it in a journal/magazine.

TABLE 1

Early progenitor domainsbHLH transcription factorsHomeodomain transcription factors
dA1Olig3, Atoh1Pax3, Msx1
dA2Olig3, Ngn1, Ngn2Pax3, Pax7low, Msx1
dA3Olig3, Ascl1, Ngn2Pax3, Pax6, Pax7, Gsx2
dA4Olig3, Ascl1, Ngn2, Ptf1a
dB1Ascl1, Ngn2, Ptf1aPax3, Pax6, Pax7, Gsx1/2
dB2Phox2b
dB3Ascl1Pax3, Pax6, Pax7, Gsx1/2, Dbx2
dB4Ngn1, Ngn2Pax3, Pax6, Pax7, Dbx2
V0Ngn1, Ngn2Dbx1, Dbx2, Pax6, Pax7
V1Ngn1, Ngn2Dbx2, Pax6, Nkx6.2
V2Ngn1, Ngn2Dbx2, Pax6, Nkx6.1, Nkx6.2
MNsOlig2Pax6
MNvAscl1Phox2b (early), Nkx2.2, Nkx2.9

Transcription factors expressed in progenitor domains of the developing hindbrain.

Depending on the rhombomere, six to eight progenitor domains can be distinguished in the dorsal (also known as the alar plate) aspect of the developing hindbrain, while four to five progenitor domains can be identified in its ventral (or basal) plate (Figure 3). Outstandingly, there exist great resemblance between the developing hindbrain and the spinal cord, in terms of progenitor domains that locate to their dorsal-ventral axis, which share similar gene expression patterns, illustrating common developmental programs that occur between these two nervous system regions (, ; ; ; van der Heijden and Zoghbi, 2020; ). In the dorsal-most part of the alar plate, the combinatorial expression of the basic Helix-loop-Helix (bHLH) transcription factor Olig3 with other bHLH genes distinguishes four progenitor domains that give rise to dorsal (d) class A neurons: dA1, dA2, dA3, and dA4 (Figure 3 and Tables 1, 2 for a list of genes expressed in these progenitors and postmitotic neurons) (; Zechner et al., 2007; ; Storm et al., 2009). Ventral to class A progenitors, the alar plate contains four more progenitor domains that collectively generate class B neurons that express and depend on the homeodomain factor Lbx1 for their proper specification (; ; ; Sieber et al., 2007; ). The combinatorial expression of Lbx1 with additional homeodomain transcription factors demarcates class B neurons into four types: dB1, dB2, dB3, and dB4 (Figure 3 and Tables 1, 2). Like in the alar plate, progenitor cells of the basal plate exhibit specific molecular codes of transcription factor expression that impose distinctive identities to at least five major neuron types: ventral (V) 0, V1, and V2 interneurons as well as somatic motor and vicero/branchio motor neurons (Figure 3 and Tables 1, 2; , ; ; ; van der Heijden and Zoghbi, 2020).

TABLE 2

Early born neuron typesTranscription factors
dA1Pou4f1, Barh1, Lhx2, Lhx9, Evx1
dA2Pou4f1, Lhx1, Lhx5, Foxp2
dA3Tlx3, Phox2b, Lmx1b
dA4Foxd3, Foxp2
dB1Lbx1, Pax2, Lhx1, Lhx5
dB2Lbx1, Phox2b, Atoh1
dB3Lbx1, Tlx3, Lmx1b, Prrxl1,
dB4Lbx1, Pax2, Lhx1, Lhx5, Wt1, bHLHb5, Dmrt3
V0Evx1, Pax2, Lhx1/5
V1En1, Pax2, Lhx1/5
V2Chx10, Sox14, Sox21
PMNsIsl1/2
PMNvPhox2b, Isl1/2 (visceral motor neurons)
Gata2, Gata3, Lmx1b and Pet1 (Raphe neurons)

Transcription factors expressed in neuronal cell types emerging from the developing hindbrain.

Thus, complex networks of signaling cues pattern the developing brainstem along its anterior-posterior and dorsal-ventral axes to form a molecular grid of longitudinally and transversely distinct progenitor domains, each displaying a particular molecular code of transcription factor expression that singles out the specification of particular neuron types. In the following sections we will discuss the current knowledge of the transcriptional codes that safeguard the specification of pontine and medullary neurons that form the central respiratory circuits in the hindbrain, which collectively generate, monitor and modulate breathing (Figure 4).

FIGURE 4

in Handbook of Clinical Neurology, Chapter 5, entitled Early development of the breathing network, published by Elsevier Books. The license number 5392080585902 between Hernandez-Miranda, Charite Universitätsmedizin Berlin and Elsevier allows us the reuse of it in a journal/magazine.

Development of the anterior respiratory groups

These respiratory groups locate to the pons and include two major structures: (i) the dorso-lateral parabrachial complex and its associated Kölliker-Fuse nucleus (for simplicity here shortened to parabrachial/Kölliker-Fuse complex) that surrounds the cerebellar peduncle, and (ii) the intertrigeminal (also known as the peritrigeminal) region that surrounds the trigeminal motor nucleus. The parabrachial/Kölliker-Fuse complex is key in the transition phase between inspiration and expiration. It is classically considered the major component of the pontine pneumotaxic center that controls the amount of air inspired in each breath by providing an off-switch for inspiration (; ; Song et al., 2006). The function of the intertrigeminal region is yet to be defined, although available evidence suggests that this region might be an anti-apneic breathing center (Radulovacki et al., 2003, ,; Stoiljkovic et al., 2009; van der Heijden and Zoghbi, 2018).

Using conditional mutagenesis and lineage-tracing experiments, the group of Huda Y. Zoghbi has studied the development of these pontine groups to great effect. These studies show that both the parabrachial/Kölliker-Fuse complex and intertrigeminal region depend on the bHLH transcription factor Atoh1 (formerly called Math1) for their development, as ablation of Atoh1 results in the absence of these pontine groups in mice (Wang et al., 2005; Rose et al., 2009a,b). Recently, Van der Heijden and Zoghbi traced the rhombomeric origin of the parabrachial/Kölliker-Fuse complex to rhombomere 1, whereas they identified the intertrigeminal region to derive from rhombomere 2 (van der Heijden and Zoghbi, 2018). One should recall that rhombomere 1 development depends on the correct expression of En1, whereas Hoxa2 is the most rostral Hox gene expressed in the developing hindbrain and delimits the border between rhombomeres 1 and 2 (Figures 1, 2). The importance of neurons derived from En1-expressing cells to respiration has not been specifically investigated, but an early report showed that En1 null mutant mice die at birth and present with severe malformations of the cerebellum, midbrain and the parabrachial/Kölliker-Fuse complex (Wurst and Bally-Cuif, 2001). first explored the function of Hoxa2-expressing neurons in respiration in the 2000s, whose work unveiled marked respiratory phenotypes and neonatal death in Hoxa2 null mutant mice. However, the loss of numerous brainstem structures in these mutant mice precluded the identification of the particular respiratory neurons lost by the constitutive ablations of En1 or Hoxa2.

In the more recent van der Heijden and Zoghbi study, the ablation of Atoh1 from En1-expressing cells, using En1Cre/+;Atoh1LacZ/Flox mice, leads to the absence of a recognizable parabrachial/Kölliker-Fuse complex (van der Heijden and Zoghbi, 2018). Physiologically, the absence of this respiratory group does not affect basal respiratory parameters, although spontaneous apneas and frequent sighing behavior is observed in En1Cre/+;Atoh1LacZ/Flox mice. Notably, the specific elimination of the parabrachial/Kölliker-Fuse complex impairs respiratory chemoreflexes to hypoxia (low oxygen) and hypercarbia (high carbon dioxide). Despite the fact the parabrachial/Kölliker-Fuse complex does not sense changes in blood gases by itself, it is known to form reciprocal connections with the nucleus tractus solitarius, a center known to mediate respiratory chemoreflexes (, ; ). Thus, the loss of communication between the nucleus tractus solitarius and the parabrachial/Kölliker-Fuse complex might account for the impaired chemoreflexes observed in En1Cre/+;Atoh1LacZ/Flox mutants. In addition, van der Heijden and Zoghbi restricted the ablation of Atoh1 to rhombomere 2 by using a transgenic mouse line that specifically expresses Cre in rhombomere 2 derived cells (Hoxa2:CreTG;Atoh1LacZ/Flox mice). In doing so, these scientists anatomically demonstrated the aberrant migration and defective location of intertrigeminal neurons in their conditional mutants. Physiologically, these animals show sigh-induced spontaneous apneas and smaller respiratory tidal volumes that led to a mild hypoventilation phenotype, but otherwise they are fully capable to respond to hypoxic and hypercarbic chemoreflexes. Thus, the anterior-posterior origin of the parabrachial/Kölliker-Fuse complex and intertrigeminal region has been assigned to rhombomeres 1 and 2, respectively.

During development, Atoh1 is transiently expressed (E10.5-E13.5) in progenitor cells of the dA1 domain that encompasses rhombomere 1 (also known as the rostral or upper rhombic lip) and rhombomeres 2–7/8 (known as the caudal or lower rhombic lip) (Figure 3 and Table 1; ; ; ; Storm et al., 2009; ). Therefore, the study of van der Heijden and Zoghbi (2018) indicates that the upper rhombic lip is the bona fide origin of the parabrachial/Kölliker-Fuse complex. The progenitor domain that generates the intertrigeminal region remains to be conclusively identified. Given that intertrigeminal neurons express the homeodomain factors Lbx1 and Phox2b, in addition to Atoh1, the source of intertrigeminal neurons could be: the dA1 (Atoh1 +) or the dB2 (Phox2b +) progenitor domain in rhombomere 2 (Wang et al., 2005; ; Rose et al., 2009a,b; ; Ruffault et al., 2015; ,). Hence, either dA1-derived neurons activate Phox2b and Lbx1 expression or dB2-derived (Phox2b+/Lbx1+) neurons switch on the expression of Atoh1. Current evidence suggests that the latter option is the correct, as retrotrapezoid neurons, a sub-population of dB2 neurons produced in rhombomere 3 or 5, originates from Phox2b + (dB2) progenitors whose progeny subsequently express Lbx1 and Atoh1 (; ; ).

Development of the dorsal medullary respiratory column

This respiratory column contains neurons known to be critical for the regulation of inspiratory activity. Most neurons forming this respiratory column reside within the nucleus tractus solitarius (nTS, bilaterally located in the dorsal medulla oblongata), although a few neurons belonging to this column can be found in the reticular formation adjacent to the nTS. The nTS extends from the caudal level of the facial motor nucleus (at the border between rhombomeres 6/7) to the cervical spinal cord. It represents the primary entry site of peripheral viscerosensory information into the central nervous system (Figure 5A; ; Travagli, 2007; ). This nucleus contains second order sensory neurons that further process and relay this information to other brain regions in the brainstem (i.e., parabrachial/Kölliker-Fuse complex), forebrain and spinal cord (, ; ). The intermediate (at the level of the area postrema) and caudal regions of the nTS receive cardiovascular and respiratory viscerosensory afferents, while the rostral to intermediate nTS primarily receives digestive information from the vagal and glossopharyngeal nerves (; Vangiersbergen et al., 1992; Zoccal et al., 2014; ). With respect to respiration, the intermediate nTS receives afferent information from slowly adapting pulmonary stretch receptors, whereas the caudal nTS receives afferent information from the peripheral chemoreceptors (that is the carotid bodies) and from rapidly adapting pulmonary stretch receptors (; , ; ; ; ; Zhao et al., 2022). Furthermore, the nTS harbors several groups of premotor neurons that can control, for instance, the laryngeal and expiratory motor activity used in breathing-associated behaviors, such as in vocalization ().

FIGURE 5

in Handbook of Clinical Neurology, Chapter 5, entitled Early development of the breathing network, published by Elsevier Books. The license number 5392080585902 between Hernandez-Miranda, Charite Universitätsmedizin Berlin and Elsevier allows us to reuse it in a journal/magazine. The primary data used in this figure was published in under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution, provided that the original author and source are credited (https://creativecommons.org/licenses/by/4.0/).

Histological studies traced the origin of excitatory nTS neurons to the dA3 progenitor domain (located between rhombomeres 4 and 7/8; Figure 3). This progenitor domain shares molecular traits with a progenitor domain in the spinal cord called dl3 (; Storm et al., 2009; ). Indeed, neuronal derivatives from these progenitor domains are excitatory and seem to only vary in the expression of the transcription factors Phox2b (in dA3 neurons) and Isl1 (in dI3 neurons), but otherwise they co-express the transcription factors Pou4f1, Tlx3, Prrxl1, and Lmx1b (Figures 3, 5B; ; , ; ; ; ; ). For over a quarter century the group of Christo Goridis and Jean-François Brunet has characterized the critical roles of Phox2a and Phox2b in visceral nervous system development. These homeobox transcription factors are necessary for the development of central and peripheral noradrenergic neurons, parasympathetic and sympathetic ganglia, branchial, and visceral motor neurons, as well as primary and secondary viscerosensory neurons (; , , ,,; ; ; ; ). Phox2b is critical for the specification of excitatory nTS neurons, and its mutation precludes the formation of this brainstem center in mice (; ; ). In addition, most excitatory nTS neurons co-express the transcription factor Tlx3 (previously known as Rnx) that seems to stabilize and maintain the expression of Phox2b in nTS neurons (; ; Storm et al., 2009; ).

In the developing hindbrain, the precise identity of dA3 progenitor cells is determined by the co-expression of the bHLH transcription factors Olig3, Ascl1 and Ngn2 (Figure 5B; ; ; Storm et al., 2009; ). In addition to excitatory nTS neurons, dA3 progenitors also generate other excitatory neurons that include: area postrema neurons (associated with vomiting reflexes) and caudal (nor)adrenergic neurons (the baroreflex-associated A1 and A2 groups) (; ; ; Storm et al., 2009; Zhang et al., 2021). Progenitors in the dA3 domain generate these neuron types in a temporal order in which (nor)adrenergic neurons are generated first, followed by excitatory nTS neurons and lastly area postrema neurons (). Although most nTS neurons are excitatory, a substantial amount of inhibitory neurons also reside within this nucleus. Inhibitory nTS neurons depend on Lbx1 and appear to derive from the dB1 or dB2 progenitor domains (). The rhombomeric origin of the nTS has not been directly investigated, incidental evidence, however, suggests that it primarily originates from rhombomeres 7 & 8, as the ablation of the transcription factor MafB, which severely affects the development of rhombomeres 5 and 6, does not significantly interfere with nTS development ().

Development of the ventral medullary respiratory column

Several anatomical and physiological distinct groups of respiratory-related neurons have been identified in the ventral medulla oblongata, which form the ventral medullary respiratory column, these groups include: (a) the retrotrapezoid/parafacial nucleus, (b) the Bötzinger complex, (c) the preBötzinger complex, (d) the PiCo complex (dorsal to the Bötzinger and preBötzinger complexes), (e) the rostral ventral and caudal respiratory groups, as well as (f) the raphe obscurus (Figure 4). Neurons within these groups can either project onto respiratory premotor neurons or themselves act as premotor neurons to regulate the motoric behavior associated with breathing. Interestingly, three of these groups exhibit intrinsic rhythmic activity during prenatal development: (i) the parafacial nucleus, (ii) the preBötzinger complex and (iii) a newly identified PostInspiratory COmplex (PiCo), and are believed to be central for the generation of the respiratory rhythm (Smith et al., 1991; ; ; ; Ramirez and Baertsch, 2018). Except for the recently identified PiCo complex, the development of all other ventral respiratory groups has been studied.

Development of the retrotrapezoid/parafacial nucleus

The retrotrapezoid and parafacial nuclei are two small groups of excitatory neurons that are located ventro-laterally to the facial motor nucleus. Whether these groups are genuinely distinct remains to be conclusively determined. However, recent evidence seems to suggest that they are indeed physiologically distinct neuronal populations (, , ; ; Zoccal et al., 2018). From an anatomical point of view, retrotrapezoid neurons locate ventrally to the facial motor nucleus and are long known to contain central respiratory chemoreceptor neurons; that is, acid-activated neurons that maintain constant levels of arterial PCO2 (Figures 6A,B,F). For more details on the physiology of retrotrapezoid neurons and other central respiratory chemoreceptor neurons, we refer to the excellent work of , , , and . On the other hand, neurons of the parafacial nucleus are located lateral to the facial motor nucleus. They are thought to control active expiration, a particular type of expiration that is produced when high metabolic demands induce an increase in respiration (; Thoby-Brisson et al., 2009; ; , ; ; ).

FIGURE 6

in Handbook of Clinical Neurology, Chapter 5, entitled Early development of the breathing network, published by Elsevier Books. The license number 5392080585902 between Hernandez-Miranda, Charite Universitätsmedizin Berlin and Elsevier allows us to reuse it in a journal/magazine. The primary data used in this figure was published in and under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution, provided that the original author and source are credited (https://creativecommons.org/licenses/by/4.0/).

In 2009, identified that virtually all retrotrapezoid and parafacial neurons originate from Egr2-expressing cells (that is from rhombomeres 3 and/or 5; see Figure 2). Furthermore, they identified their distinctive molecular signature, namely their co-expression of Phox2b, Lbx1, and Atoh1 (Figure 6B). Ablation of each of these genes leads to the anatomical absence (Phox2b and Lbx1 mutant mice) or aberrant location (Atoh1 mutants) of retrotrapezoid and parafacial neurons, as well as to the loss of the hypercarbic reflex (the natural acceleration of breathing in response to increasing PCO2 levels) and neonatal death (Wang et al., 2005; ; ; Rose et al., 2009a,b; ). The dB2 progenitor domain generates retrotrapezoid and parafacial neurons and is exclusively located between rhombomeres 2 and 6 (Figures 3, 6C). dB2 progenitor cells express Phox2b and their postmitotic progeny co-express Lbx1 in addition to Phox2b (Figure 6C). A small number of dB2 (Lbx1 + /Phox2b +) neurons migrate ventrally toward the facial motor nucleus and activate the expression of Atoh1 during their migration (Figures 6D,E). The expression of Atoh1 seems to be essential for the migration and maturation of retrotrapezoid and parafacial nucleus neurons (; Ruffault et al., 2015; ). As retrotrapezoid and parafacial neurons mature throughout postnatal life, and for reasons yet to be identified, these neurons silence Lbx1 and Atoh1, but retain Phox2b. Several studies show that interfering with the specification of retrotrapezoid and parafacial neurons does not compromise neonatal or postnatal survival in mice. However, the loss of these neurons leads to a range of hypoventilation behaviors and the loss of the hypercarbic reflex in neonatal mice (Ramanantsoa et al., 2011; ; Ruffault et al., 2015; ). Notably, transgenic mice lacking retrotrapezoid and parafacial neurons are able to recover some of the ventilatory responses to hypercarbia in adult life (Ramanantsoa et al., 2011; ; Ruffault et al., 2015; ). How these animals gain the ability to respond to hypercarbia in adulthood is currently unknown, but this indicates that other chemoreceptor cells, either in the central or peripheral nervous system, can compensate for the loss of retrotrapezoid and parafacial neurons in the adult life.

Development of the Bötzinger complex

The major inhibitory component of the ventral respiratory column is the Bötzinger complex that uses GABA and glycine as its primary neurotransmitters (Shao and Feldman, 1997; Schreihofer et al., 1999; Song et al., 2001). Classic studies showed that the electrophysiological and pharmacological activation of Bötzinger neurons strongly inhibits inspiration and that Bötzinger neurons display decrementing postinspiratory or augmenting firing patterns during expiration (; ). These neurons mutually interact with the preBötzinger complex to regulate the respiratory rhythm (; Tian et al., 1998; ,,). The Bötzinger complex innervates all other brainstem respiratory neurons and projects to spinal premotor and motor phrenic neurons (; ; ; Tian et al., 1998; Smith et al., 2007).

The precise rhombomeric origin and progenitor domain from which the Bötzinger complex develops has not yet been fully investigated. However, circumstantial evidence might indicate that this respiratory group develops from the Lbx1-lineage. Indeed, a lineage tracing study by revealed that virtually all GABAergic and glycinergic neurons found in the anatomical region where the Bötzinger complex resides have a history of Lbx1 expression, ablation of which results in the absence of GABAergic and glycinergic neurons in this area (). Mature “Bötzinger” neurons lose the expression of Lbx1 but can be recognized by the expression of Pax2, GABA and other glycinergic markers (). As abovementioned, Lbx1 is key for the specification of four distinct neuron types, two of which express Pax2 and are GABAergic and glycinergic in nature: dB1 and dB4 neurons (Figure 3 and Table 2). In Lbx1 null mutant mice, Pax2 expression seems to be uniquely lost from the dB1 domain (that extends from rhombomeres 2–7/8), which might suggest this region as the possible source of Bötzinger neurons (). However, more research is necessary to clearly define the developmental origin of the Bötzinger complex.

Development of the preBötzinger complex

In the early 1990’s, the work of Jeffrey C. Smith and Jack L. Feldman identified the preBötzinger complex as essential for generating the respiratory rhythm in mammals (Figures 7A,B; Smith et al., 1991). Later studies uncover some molecular markers such as the neurokinin 1 receptor and somatostatin to be expressed by preBötzinger neurons (Smith et al., 1991, 2007; , ). How the preBötzinger complex generates the respiratory rhythm is currently the subject of intense investigation and outside the scope of this review, but we recommend the reader the excellent reviews by the groups of , Ramirez and Baertsch (2018), Rubin and Smith (2019), Ramirez et al. (2022) and Smith (2022).

FIGURE 7

in Handbook of Clinical Neurology, Chapter 5, entitled Early development of the breathing network, published by Elsevier Books. The license number 5392080585902 between Hernandez-Miranda, Charite Universitätsmedizin Berlin and Elsevier allows us to reuse it in a journal/magazine. The primary data used in this figure was published in under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution, provided that the original author and source are credited (https://creativecommons.org/licenses/by/4.0/).

In 2010, two independent groups defined the V0 progenitor domain (in the basal plate) as the source of the preBötzinger complex (; ). V0 progenitor cells share molecular traits with the p0 progenitor domain in the ventral spinal cord and are characterized by the expression of Dbx1 and Dbx2, the former is, however, unique for V0 and p0 progenitor cells (Figure 7C and Table 1; , ). Both and used Dbx1LacZ mice, which express the reporter protein beta-galactosidase under the control of the Dbx1 promoter, to lineage trace the preBötzinger complex and demonstrated that all the excitatory and rhythmically active neurons in this region derived from the V0 progenitor domain. Neurons emanating from this domain express the transcription factor Evx1, which distinguishes them from the surrounding dB4 and V1 neurons that express Lbx1 and En1, respectively (Figure 7C and Table 2; , ). Interestingly, V0 neuronal derivatives destined to populate the preBötzinger complex are correctly generated and reach the preBötzinger complex area in Dbx1 null mutant (Dbx1LacZ/LacZ) mice, but these cells do not exhibit intrinsic rhythmicity and fail to adopt their mature molecular identity, as determined by the lack of neurokinin 1 receptor or somatostatin expression in these cells ().

Expression of the transcription factor Pax7 molecularly distinguishes two V0 progenitor subdomains: V0 dorsal (V0D; Pax7 +) and V0 ventral (V0V; Pax7-) (Figure 7C). In the authors walked an extra mile and ablated Dbx1 from the V0D domain, using Pax7Cre;Dbx1LacZ/Flox mice, to elucidate which of these two subdomains generate the preBötzinger complex. This elegant experiment found an anatomically, molecularly and physiologically intact preBötzinger complex, demonstrating that this respiratory center emerges from the V0V subdomain. More recent studies have exploited this developmental knowledge to selectively activate, silence, or ablate Dbx1-derived preBötzinger neurons, resulting in respiratory changes that enhance, depress, or halt breathing in adult mice, respectively (Wang et al., 2014; Vann et al., 2016, 2018).

Even though the anterior-posterior origin of the preBötzinger complex has not yet been identified, an early study indicates that it emerges from the most posterior rhombomeres and partly from rhombomere 6 (MafB +). In this context, a study by showed that MafB null mutant mice rarely breathe at birth and display limited phrenic nerve activity (). Histologically, MafB null mutants have a significant, but not complete, loss of preBötzinger neurons (). Since Egr2 null mutant mice display breathing rhythmicity and can survive for about a day after birth (), it is likely that V0 progenitors in rhombomere 6 generate a fraction of preBötzinger neurons that is complemented by V0 progenitors in rhombomeres 7/8.

Development of the rostral and caudal ventral respiratory groups

The preBötzinger complex controls breathing by activating premotor neurons that in turn regulate multiple brainstem (e.g., hypoglossal or vagal) and spinal cord (phrenic) motor neurons. Caudal to the preBötzinger complex, two sets of respiratory premotor neurons are associated with inspiratory and expiatory motor activity: the rostral (rVRG) and caudal (cVRG) ventral respiratory groups, respectively (Smith et al., 2013). A recent study from the group of Gilles Fortin identified that excitatory rVRG neurons have a history of Dbx1 expression and, as such, to originate from the V0 progenitor domain (Wu et al., 2017). Even though both preBötzinger neurons and rVRG neurons express Slc17a6 (vGlut2) and Pax2, only the former expresses neurokinin 1 receptor and somatostatin (; ; Wu et al., 2017). These molecular differences could be explained by their distinct rhombomeric origins and/or by their generation in distinct V0 progenitor subdomains. The development of cVRG neurons has not been yet addressed, but some Dbx1-derivaties neurons can be observed caudal to the rVRG, suggesting that cVRG neurons are also generated from the V0 progenitor domain (; Wu et al., 2017). An alternative source of cVRG neurons could be the dB4 domain, whose derivatives co-express Lbx1 and Wt1 (Table 2). Indeed, a recent study reported the presence of GABAergic (Wt1 +) neurons in the cVRG region (Schnerwitzki et al., 2020).

Development of respiratory serotonergic neurons

The monoamine neurotransmitter serotonin has long been implicated in the control of respiration. Brain serotonin is produced by nine distinct groups of cells (called raphe nuclei), all of which are in the brainstem. Among the distinct raphe nuclei, several lines of research indicate that the midline located raphe obscurus in the medulla oblongata is an important component of the central respiratory chemoreceptor circuit (). First, the en masse inhibition of serotonergic neurons (Ray et al., 2011), or the targeted inhibition of raphe obscurus neurons (), significantly impairs the chemoreflex to hypercarbia in mice. Second, the optogenetic activation of these raphe neurons accelerates respiration in conscious and anesthetized rodents (). Lastly, mice genetically engineered to lack raphe neurons display dulled chemoreflexes to hypercarbia (, ; ).

The ventral-most progenitor domain (termed as pMN; Figure 3) in the developing hindbrain initially generates motor neurons, and later all brainstem serotonergic neurons. Due to its proximity to the floor plate, the pMN domain is under the direct influence of Sonic Hedgehog signaling (; ). Molecularly, this progenitor domain is subdivided into a dorsal subdomain (pMNs; expressing the transcription factors Pax6 and Olig2) and a ventral subdomain (pMNv; expressing the transcription factors Nkx2.2, Nkx2.9, and Phox2b) (Figure 3 and Table 1). Detailed histological and genetic analyses showed that the pMNs subdomain generates somatic motor neurons (i.e., hypoglossal motor neurons), whereas the pMNv subdomain generates branchial (e.g., facial motor neurons) and visceral (i.e., nucleus ambiguous) motor neurons (; ). Pioneer studies by Briscoe and colleagues showed that pMNv progenitors first generate branchio/viscero motor neurons before E11.5 in mice, and then serotonergic neurons (; ). One should note that except for rhombomere 4, the pMNv progenitor domain in all other rhombomeres contributes to raphe neurons. pMNv progenitor cells of rhombomere 4 are known to generate a large group of branchial (facial) motor neurons and to have an unusual prolonged expression of Phox2b between E9.5 to E12.5, which seems to be attributable to its incapacity to generate serotonergic neurons (). Indeed, the silencing of Phox2b expression seems to be a molecular switch in the transition of pMNv progenitor cells from first generating branchio/visceromotor neurons to later generating serotonergic neurons in the other rhombomeres (; , ). In support of this, analysis of Nkx2.2 null mutant mice revealed an unusual extended expression of Phox2b within the pMNv domain, which leads to the overproduction of branchio/visceromotor neurons and the absence of serotonergic neurons (; , ). Conversely, the ablation of Phox2b results in the early generation of raphe neurons at the expense of branchio/visceromotor cells ().

The bHLH transcription factor Ascl1 is critical for development of peripheral (i.e., enteric nervous system) and central (raphe) serotonergic cells (; ). In the pMNv domain, Ascl1 is co-expressed with Phox2b during the genesis of branchio/visceromotor neurons and is retained by these progenitors throughout the specification of raphe cells (). Mutation of Ascl1 does not affect Phox2b expression nor the development of branchio/visceromotor cells, but severely interferes with the specification of raphe neurons, which are completely absent in Ascl1 null mutant mice (). The maturation of raphe neurons is regulated by several other transcription factors, such as Gata2, Gata3, Lmx1b and Pet1, of which the null mutation of Lmx1b or Pet1 results in the total loss or a severe decrease (> 70%) of raphe cells, respectively (; ; ; ; ).

Cell lineages contributing to respiratory and non-respiratory neurons

From a developmental point of view, most brainstem respiratory neurons emerge from a few molecularly defined cell-lineages and progenitor domains: (i) an Atoh1-lineage that contributes to the development of the parabrachial/Kölliker-Fuse complex (dA1, in rhombomere 1) and the intertrigeminal region (either dA1 or dB2, in rhombomere 2); (ii) an Olig3/Phox2b/Tlx3-lineage (dA3, in rhombomere 7/8) that generates the dorsal medullary respiratory column (nTS); iii) a Phox2b/Lbx1/Atoh1-lineage (dB2, in rhombomere 3 and/or 5) that generates the retrotrapezoid/parafacial nuclei; (iv) an Lbx1-lineage (presumably dB1, unknown rhombomeric origin) that produces the Bötzinger complex; v) a Dbx1-lineage (V0) that gives rise to the preBötzinger complex (in rhombomeres 6-7/8) and the rVRG (and possibly the cVRG, in rhombomere 7/8) groups of premotor neurons; as well as (vi) an Nkx2.2/Ascl1/Lmx1b-lineage that produces raphe serotonergic neurons (pMNv, across rhombomeres).

One should not forget, however, that each of these progenitor domains produce a much greater diversity of neuron types than just respiratory neurons. A good example of this is the dA1 (Atoh1 +) domain in rhombomere 1 (upper rhombic lip). This domain generates: in addition to the parabrachial/Kölliker-Fuse complex, all excitatory deep cerebellar neurons, all cerebellar granule cell progenitors, as well as all cerebellar and cochlear unipolar brush cells (; ; Ray and Dymecki, 2009; ; van der Heijden and Zoghbi, 2018, 2020; ; , ; ). How these progenitors generate such a vast array of neuron types is currently being investigated and appears to depend on the temporal expression of transcription factors that act as selector genes. In this context, the co-expression of Atoh1 with Olig3 is critical for deep cerebellar neuron development, whereas the co-expression of Atoh1 with Neurod1 is essential for granule cell progenitor specification and cerebellar and cochlear unipolar brush cell development (; ; ; ; ; ). The selector gene for the specification of parabrachial/Kölliker-Fuse complex is presently unknown.

It might not be surprising that across rhombomeres, each progenitor domain generates different neuron types. Nonetheless, an interesting trait of these spatially segregated progenitor domains is that their shared expression of transcription factors might instruct their progeny to synaptically connect and form functional circuits. For instance, both Atoh1 + /Olig3 + (dA1) progenitors in rhombomere 7 or the pdI1 progenitors in the spinal cord (equivalent to dA1) produce second relay neurons that project to the cerebellum and synapse onto granule cells and deep cerebellar neurons that derive from rhombomere 1 dA1 (Atoh1 + /Olig3 +) progenitors (; ; Storm et al., 2009; ; , ). Another example of this molecular logic could be the inferior olive-Purkinje cell circuit. Indeed, inferior olive cells that derive from Olig3 + /Ptf1a + (dA4 in rhombomere 7) progenitors send axons that synapse onto Purkinje cells that emerge from Olig3 + /Ptf1a + progenitors in the cerebellar ventricular zone in rhombomere 1 (; Storm et al., 2009; ; , ). To which extent these shared transcriptional codes allow for the interconnection of the distinct neurons that form the brainstem respiratory circuit is presently unknown. However, emerging evidence shows similar developmental strategies, i.e., preBötzinger complex neurons that connect with the rVRG are both derivatives of the V0 domain (; ; Wu et al., 2017). An even more conspicuous case is the Phox2b-lineage that generates virtually all neurons that form the central and peripheral visceral nervous system (; , , , , , , ; ; ; ; ; ). The question of how these developmental strategies emerged during evolution remains to be explored.

Conclusion

The enormous knowledge gained during the last three decades of brainstem development research now allows us to further dissect the function of respiratory neurons with unprecedented detail. Several experimental and theoretical approaches have recently taken advantage of the developmental trajectories of respiratory neurons to explore the complex character of the respiratory rhythm generator or to elucidate different components of the central chemoreceptor circuit (; Ray et al., 2011; ; Wang et al., 2014; Vann et al., 2016, 2018).

Many aspects of how respiratory neuron diversity emerges during development remain to be elucidated. The identification of how the brainstem generates respiratory neurons is critical to understand this complex behavior and essential for the development of new therapeutic approaches for the management of respiratory diseases. In this context, recent studies on the genetic disturbances causing congenital respiratory syndromes are currently steering our views into both the development and function of respiratory neurons. A clear example is the study of congenital central hypoventilation syndrome (CCHS, also known as Ondine’s curse; OMIM 209880). Although rare (1 in 200,000 live births), this disorder is life threatening and characterized by slow, apneic and shallow breathing (hypoventilation) while awake and respiratory arrest during sleep (Weese-Mayer et al., 2008, 2017; Sivan et al., 2019; ). Frequently, CCHS patients also present with blunted responses to hypercabia and have abnormal levels of PCO2. Early genetic studies identified de novo mutations in PHOX2B as the most prevalent cause of CCHS (). Two types of PHOX2B mutations that cause CCHS have been identified: (i) polyalanine repeat expansions, and (ii) non-polyalanine repeat expansions that are more prevalent in severe cases of CCHS (; Zhou et al., 2021). CCHS patients with PHOX2B mutations frequently manifest Hirschsprung’s disease, revealing that genetic disturbances on PHOX2B can simultaneously alter the development of both the central and peripheral nervous systems (Weese-Mayer et al., 2008, 2017; Sivan et al., 2019; ). PHOX2B is a central factor in the development and function of the visceral nervous system, whose mutation results in midterm fetal lethality (). Interestingly, the insertion of a frequent poly-alanine PHOX2B mutation (called PHOX2B+7ala) into the murine genome has shown that this aberrant expansion only affects a subset of Phox2b functions, as only one Phox2b-dependent neuron type (the retrotrapezoid nucleus) does not develop correctly in Phox2b+7ala mutant mice (). More recently, the characterization of a CCHS disease-causing frameshift mutation in LBX1 has further revealed that this respiratory disorder originates from the misspecification of dB2 neurons in mice, and that this is caused by a lack of cooperativity between PHOX2B and LBX1 ().

For many years, the size and location of the brainstem was a major impediment to comprehend its physiology. With the advent of new technologies such as single-cell transcriptomics, monosynaptic viral tracing as well as opto- and chemo- genetic tools, several of the long-lasting obstacles associated with the identification and modulation of specific breathing behaviors are now amenable for scientific exploration. There is no doubt that the years to come will foster and propel our understanding of this elementary and humble animal behavior in ways that we can only now imagine.

Statements

Author contributions

YX, KC, AA, EL, and LH-M reviewed the literature. LH-M wrote the original draft and edited it with the input from all authors. All authors contributed to the article and approved the submitted final version.

Funding

Work in the Hernandez-Miranda’s laboratory is supported by the Fritz-Thyssen-Stiftung (grant no. 10.20.1.004MN) and Deutsche Forschungsgemeinschaft (grant no. 450241946), both granted to LH-M.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

Summary

Keywords

transcription factors, brainstem development, progenitor domains, neuronal specification, respiratory neurons

Citation

Xia Y, Cui K, Alonso A, Lowenstein ED and Hernandez-Miranda LR (2022) Transcription factors regulating the specification of brainstem respiratory neurons. Front. Mol. Neurosci. 15:1072475. doi: 10.3389/fnmol.2022.1072475

Received

17 October 2022

Accepted

14 November 2022

Published

29 November 2022

Volume

15 - 2022

Edited by

Estela Maris Muñoz, CONICET Dr. Mario H. Burgos Institute of Histology and Embryology (IHEM), Argentina

Reviewed by

Flavio S. J. De Souza, CONICET Institute of Physiology, Molecular Biology and Neurosciences (IFIBYNE), Argentina; Mitsuhiro Hashimoto, Fukushima Medical University, Japan

Updates

Copyright

*Correspondence: Luis R. Hernandez-Miranda,

This article was submitted to Neuroplasticity and Development, a section of the journal Frontiers in Molecular Neuroscience

Disclaimer

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

Outline

Figures

Cite article

Copy to clipboard


Export citation file


Share article

Article metrics