Key Brainstem Structures Activated during Hypoxic Exposure in One-day-old Mice Highlight Characteristics for Modeling Breathing Network in Premature Infants

We mapped and characterized changes in the activity of brainstem cell groups under hypoxia in one-day-old newborn mice, an animal model in which the central nervous system at birth is particularly immature. The classical biphasic respiratory response characterized by transient hyperventilation, followed by severe ventilation decline, was associated with increased c-FOS immunoreactivity in brainstem cell groups: the nucleus of the solitary tract, ventral reticular nucleus of the medulla, retrotrapezoid/parafacial region, parapyramidal group, raphe magnus nucleus, lateral, and medial parabrachial nucleus, and dorsal subcoeruleus nucleus. In contrast, the hypoglossal nucleus displayed decreased c-FOS immunoreactivity. There were fewer or no activated catecholaminergic cells activated in the medulla oblongata, whereas ~45% of the c-FOS-positive cells in the dorsal subcoeruleus were co-labeled. Approximately 30% of the c-FOS-positive cells in the parapyramidal group were serotoninergic, whereas only a small portion were labeled for serotonin in the raphe magnus nucleus. None of the c-FOS-positive cells in the retrotrapezoid/parafacial region were co-labeled for PHOX2B. Thus, the hypoxia-activated brainstem neuronal network of one-day-old mice is characterized by (i) the activation of catecholaminergic cells of the dorsal subcoeruleus nucleus, a structure implicated in the strong depressive pontine influence previously reported in the fetus but not in newborns, (ii) the weak activation of catecholaminergic cells of the ventral reticular nucleus of the medulla, an area involved in hypoxic hyperventilation, and (iii) the absence of PHOX2B-positive cells activated in the retrotrapezoid/parafacial region. Based on these results, one-day-old mice could highlight characteristics for modeling the breathing network of premature infants.

We mapped and characterized changes in the activity of brainstem cell groups under hypoxia in one-day-old newborn mice, an animal model in which the central nervous system at birth is particularly immature. The classical biphasic respiratory response characterized by transient hyperventilation, followed by severe ventilation decline, was associated with increased c-FOS immunoreactivity in brainstem cell groups: the nucleus of the solitary tract, ventral reticular nucleus of the medulla, retrotrapezoid/parafacial region, parapyramidal group, raphe magnus nucleus, lateral, and medial parabrachial nucleus, and dorsal subcoeruleus nucleus. In contrast, the hypoglossal nucleus displayed decreased c-FOS immunoreactivity. There were fewer or no activated catecholaminergic cells activated in the medulla oblongata, whereas ∼45% of the c-FOS-positive cells in the dorsal subcoeruleus were co-labeled. Approximately 30% of the c-FOS-positive cells in the parapyramidal group were serotoninergic, whereas only a small portion were labeled for serotonin in the raphe magnus nucleus. None of the c-FOS-positive cells in the retrotrapezoid/parafacial region were co-labeled for PHOX2B. Thus, the hypoxia-activated brainstem neuronal network of one-day-old mice is characterized by (i) the activation of catecholaminergic cells of the dorsal subcoeruleus nucleus, a structure implicated in the strong depressive pontine influence previously reported in the fetus but not in newborns, (ii) the weak activation of catecholaminergic cells of the ventral reticular nucleus of the medulla, an area involved in hypoxic hyperventilation, and (iii) the absence of PHOX2B-positive cells activated in the retrotrapezoid/parafacial region. Based on these results, one-day-old mice could highlight characteristics for modeling the breathing network of premature infants.
Keywords: c-FOS, catecholamine, hypoxic ventilatory depression, serotonin, PHOX2B, subcoeruleus nucleus INTRODUCTION Human infants, particularly premature infants, display frequent episodes of apnea, and bradypnea, and thus very common episodes of hypoxia during the postnatal period (Bryan et al., 1986;Carroll and Agarwal, 2010;Teppema and Dahan, 2010;Mathew, 2011). In newborn mammals, hypoxia elicits a biphasic respiratory response, characterized by a transient increase followed by a severe decline of ventilation called the hypoxic ventilatory depression (HVD; Neubauer et al., 1990;Gozal and Gaultier, 2001;Carroll and Agarwal, 2010;Teppema and Dahan, 2010). The hypoxemia resulting from HVD, especially when frequently repeated, may negatively affect cardiovascular and neurocognitive functions, neurocognitive outcome, and longterm quality of life. Dysfunction of the hypoxic ventilatory response (HVR) is suspected in respiratory diseases such as Sudden Infant Death Syndrome (SIDS). SIDS victims typically experience a centrally mediated life-threatening apnea, possibly related to exaggerated HVD, due to the activation of a defense mechanism of the fetus that limits its O 2 consumption by stopping the respiratory activity in utero under hypoxemia (Poets et al., 1999;Hunt, 2001;Kinney et al., 2011;Lavezzi, 2015).
The magnitude of the initial hyperventilation in most mammalian species increases with maturation, whereas the magnitude of the late decline is greatest in newborns and decreases with maturity (Carroll and Agarwal, 2010;Teppema and Dahan, 2010). Preterm infants with a birth weight <1500 g, exhibit no initial hyperventilation (Alvaro et al., 1992;Mathew, 2011). Hyperventilation mainly results from the activation of structures of the medulla oblongata by excitatory inputs coming from peripheral chemoreceptors (Vizek et al., 1987;Finley and Katz, 1992;Waldrop and Porter, 1995;Blessing et al., 1999;Carroll and Agarwal, 2010;Teppema and Dahan, 2010): cells of the commissural and medial parts of the nucleus of the solitary tract (cNTS and mNTS) and the ventral reticular nucleus of the medulla (VLM), many of which are catecholaminergic Millhorn, 1991, 1994;Hirooka et al., 1997). Second order projections transmit the message to other neuronal populations, such as the CO 2activated neurons of the retrotrapezoid/parafacial respiratory group region (RTN/pFRG) shown to be PHOX2B-positive Guyenet and Bayliss, 2015), and neurons of the lateral parabrachial nucleus (lPB; Hayward and Felder, 1995). Central mechanisms contribute to a decline in ventilation in parallel with peripheral chemoreceptor activation; if the initial hyperventilation fails to restore sufficient PO 2 in arterial blood, the central mechanisms reduce the metabolic demand of the respiratory musculature by lowering ventilation (Neubauer et al., 1990;Carroll and Agarwal, 2010;Teppema and Dahan, 2010). Various mechanisms underlying the hypoxic decline in ventilation have been proposed including: (i) hyperperfusion of medullary CO 2 -sensitive areas (Neubauer et al., 1990); (ii) hypometabolism leading to a drop in CO 2 production (Mortola, 2004); (iii) release of neuromodulators such as adenosine (Runold et al., 1989;Neubauer et al., 1990;Kawai et al., 1995) or serotonin (5-HT) (Herman et al., 1999;Richter et al., 1999); and (iv) intrinsic activation of cells in the medulla oblonga by low PO 2 (Nolan and Waldrop, 1993;Voituron et al., 2006Voituron et al., , 2011. Here, we examined hypoxia-induced changes in the activity of brainstem neuronal populations in one-day-old mice to characterize the neuronal brainstem component of the HVR encountered under hypoxia in newborn mammals, in particular in premature mammals. Our working hypothesis was that oneday-old mice constitute a pertinent model because their central nervous system at birth is immature relative to other newborn mammals, such as humans and rats, with respect to their neuroanatomy, neurogenesis, gliogenesis, myelinisation, and molecular and biochemical dynamics in telencephalic regions (Teppema and Dahan, 2010;Darnall et al., 2016;Mallard and Vexler, 2015). Cats and rats have been used for many years in studies to identify the underlying mechanisms of the hypoxic ventilatory response, but much work has focused on mice over the past decade because of the increasing availability of genetic mouse models (Gaultier et al., 2003). Cell populations involved in the HVR of mice, particularly at birth, are not fully known. Thus, we performed an expanded analysis of changes in c-FOS expression under hypoxia in brainstem areas related to respiratory control. Dual labeling allowed us to characterize various activated c-FOS-positive cell populations by assessing their catecholaminergic, serotoninergic, or PHOX2B immunoreactivites.

MATERIALS AND METHODS
All experiments, approved by the Charles Darwin Ethics Committee for Animal Experimentation (Ce5/2011/05), were carried out in accordance with Directive 2010/63/EU of the European Parliament and of the Council of 22 September, 2010 and French law (2013/118). All efforts were made to minimize the number of animals used and their suffering. Animals were maintained on a 12-h light-dark cycle with free access to food and water.

Recording of Respiratory Variables
Changes in ventilatory variables were non-invasively measured during the apnea-free period using a home-made whole body plethysmograph (Bartlett and Tenney, 1970) on 12 newborn mice. Animals were placed in an experimental chamber (20 ml) in which they freely moved. The chamber was connected to a reference box of the same size equipped with a temperature sensor (Newport Electronic, Santa Ana, CA, USA) that permits temperature control. The temperature of the animals was measured before and after each experiment via an oral probe. During recording sessions, the chamber was sealed and air flow to the chamber interrupted for 20 s (2 and 25 min after the beginning of hypoxia), leading to a change in pressure because the volume of the chamber was fixed. Between two recording sessions, the chamber was ventilated either with a humidified normoxic mixture (21% O 2 , 79% N 2 ) or a hypoxic mixture (11% O 2 , 89% N 2 ) heated to a temperature of ∼31 • C, the thermoneutral zone (Gordon, 1993), using an external heat source. The pressure change induced by the respiratory flow was assessed by connecting a differential pressure transducer (Valydine DP 45, Northridge, CA, USA) using an adaptation of the previously described barometric method (Bartlett and Tenney, 1970). The pressure signal was digitized through a Spike 2 data analysis system (Cambridge Electronic Design, Cambridge, UK). Measurement of the ventilatory variables was made on this signal i.e., respiratory frequency (f R ), tidal volume (V T ), minute ventilation (V E ), and the number of apneas per min (a ventilatory pause longer than twice the duration of the preceding respiratory cycles; Matrot et al., 2005;Menuet et al., 2011). The V T was calculated using Drorbaugh and Fenn's equation (Drorbaugh and Fenn, 1955).

Respiratory Variables Observation
Prior to each experimental session, newborn mice were exposed to normoxic conditions for 20 min to become habituated to the chamber. Recordings were made under normoxic conditions for a further 15 min to define the control values. During the test period, the chamber was either flushed with the normoxic mixture (control group; n = 4) or the hypoxic mixture (hypoxic group; n = 8). Use of the control group ensured that the long period of retention in the recording chamber did not induce changes in the respiratory variables. At the end of the test period, newborn mice were removed from the chamber and the buccal temperature was taken. The respiratory variables were measured during the apnea-free periods during the hypoxic test (2 and 25 min after the beginning of hypoxia) and expressed as the percentage of control values. For the V T calculation under hypoxia, the buccal temperature used was that measured before mice were placed in the recording chamber for the value at 2 min of hypoxia and that measured immediately at the end of the test for the value at 25 min of hypoxia.
Each recording session lasted ∼20 s; a calibration volume of 100 µl was injected into the plethysmographic chamber with a Hamilton syringe toward the end of each recording session. Each newborn mouse was exposed only one time to the protocol. Recordings were made during 20-s periods when the chamber was sealed, and not continuously. Values are presented as mean ± standard error of the mean (SEM). D'Agostino-Pearson omnibus normality test was realized to assess the distribution of the data. The hypoxic ventilatory response was evaluated by one-way ANOVA for repeated measures followed by a Tukey's multiple comparisons test (GraphPad Prism5 San Diego California USA). Differences were considered to be significant if p < 0.05.

Induction of c-FOS
The detection of c-FOS requires minimizing manipulations that could induce changes of cell activity unrelated to the studied stimulus and a sufficiently long induction period to induce detectable changes in c-fos expression (Herdegen and Leah, 1998;Perrin-Terrin et al., 2016). The protocol used for inducing c-FOS expression was similar to that used for analyzing the effect of hypoxia on respiratory variables with two exceptions: (i) the duration of the test period was 90 min (to facilitate c-FOS induction as c-FOS protein has a half-life of 90-100 min; Herdegen and Leah, 1998) and (ii) the absence of a recovery period to exclude the possibility that changes of neuronal activity revealed by c-FOS protein detection might be related to posthypoxic neuronal mechanisms (Morris et al., 2003). At the end of the hypoxic period, newborn mice were placed under deep cold anesthesia (Danneman and Mandrell, 1997) and the brainstems removed.

Immunohistological Procedures
Immunohistochemical analysis for c-FOS was carried out in mouse brainstems exposed to either normoxia or hypoxia (n = 19) to identify hypoxia-induced changes in cell activity. Brainstems were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) for 48 h at 4 • C . They were then cryoprotected for 48 h in 30% sucrose in 0.1 M PBS and stored at −20 • C for subsequent use. Standard immunohistochemical procedures were used to locate c-FOS on 40 µm-thick coronal free-floating sections obtained using a cryostat (Leica CM 1510S; Bodineau et al., 2011;Voituron et al., 2011).
The detection of c-FOS was coupled with that of tyrosine hydroxylase (TH), 5-HT, or PHOX2B to characterize the cells displaying changes in activity revealed by c-FOS analysis.
The manufacturer verified the specificity of primary antibodies in all cases and in addition, control sections were processed in parallel, but with the omission of the primary or secondary antibodies; we observed no labeling in the absence of the primary or secondary antibodies.
Sections were mounted in sequential caudo-rostral order on silane-treated slides, air-dried, dehydrated with absolute alcohol, cleared with xylene, and coverslipped using Depex.
Sections were mounted in sequential caudo-rostral order on silane-treated slides as for the single immunohistochemical detection of c-FOS. They were air-dried, dehydrated with absolute alcohol, cleared with xylene, and coverslipped using Entellan R (VWR International S.A.S).

Quantitative Analysis of the Effect of Hypoxia on the Number of c-FOS-Positive Cells and Their Characterization
Sections were examined under a light and fluorescence microscope (Leica DM 2000, Leica Microsystems, Heidelberg, Germany). We analyzed c-FOS-positive cells in brainstem structures related to respiratory control: the A5 region (A5), lPB and medial parabrachial nucleus (mPB), Kölliker-Fuse nucleus (KF), locus coeruleus nucleus (LC), subcoeruleus nucleus (SubC; the dorsal and ventral parts were separated at the level of the ventral boundary of the trigeminal motor nucleus i.e., dSubC and vSubC) at the pontine level and the VLM, hypoglossal nucleus (12N), dorsal motor nucleus of vagus (10N), cNTS, mNTS, and ventrolateral part of the nucleus of the solitary tract (vlNTS), raphe magnus, obscurus, and pallidus nuclei (RMg, ROb, and RPa), facial nucleus (n7), and the ventral medullary surface. Definitions of boundaries of these structures were made according to the mouse brain atlas (Paxinos and Franklin, 2001;Paxinos et al., 2007). The VLM is a neuronal column ventral to the ambiguus nucleus including the A1C1 group of neurons and extending from the pyramidal decussation to the caudal edge of the facial nucleus. We made a distinction between the caudal part of the VLM (cVLM; from the pyramidal decussation to the caudal edge of the lateral paragigantocellulaire nucleus) and the rostral part of the VLM (rVLM; from the caudal edge of the lateral paragigantocellulaire nucleus to the caudal edge of the facial nucleus) using standard landmarks as previously described . The pre-Bötzinger complex is located in the caudal part of the rVLM. We also distinguished near the ventral surface of the medulla: (i) the lateral and medial part of the RTN/pFRG i.e., the lateral RTN/pFRG and medial RTN/pFRG and (ii) a more medial area at the lateral edge of the pyramidal tract, the parapyramidal group (PP) (Berquin et al., 2000a;Stornetta et al., 2005;Voituron et al., 2011), based on previously published data (Voituron et al., 2006Huckstepp et al., 2015). We localized all of these structures with the aid of numerous ventral, dorsal, and lateral landmarks (such as those indicated in Figures 2-5) to delimit the entire volume of each structure.
The distribution of c-FOS cells was plotted onto drawings with the aid of a drawing tube attached to the microscope to illustrate their distribution (Figure 2). c-FOS and doublelabeled cells were also photographed with a digital camera (Leica DFC450C, Leica Microsystems, Heidelberg, Germany). C-FOS counts were performed by eye using a counting grid in the eyepiece of the microscope at x400 to count all immunolabeled cells by varying the micrometer of the microscope, which was essential for tissue sections of 40 µm. For dual labeling, the counts were performed either by eye (at x400 except for the locus coeruleus, the pallidus, and obscurus raphe nuclei where x1000 magnification was used due to the high density of TH and 5-HT in labeled cells) or using images obtained with a digital camera (c-FOS/PHOX2B). We compared the total number of cells under normoxia and hypoxia for each analyzed area ( Table 1). We analyzed the differences between the mean numbers of neurons obtained under normoxia or hypoxia using GraphPad (GraphPad Prism5 San Diego California USA) and used the Mann-Whitney tests to determine significance. Differences were considered to be significant if p < 0.05.
Hypoxia Induced the Classical Biphasic Respiratory Response Observed in Newborns: An Early and Transient Increase Followed by a Severe Decline in Ventilation The respiratory variables of unrestrained one-day-old mice maintained in the recording chamber under normoxia remained stable throughout the recording period (n = 4). Baseline values were defined after 20 min of adaptation in the chamber and the following values of the respiratory variables expressed as the percentage baseline. At the end of the recording period, the mean f R and mean V T were 106.9 ± 14.3% and 120.9 ± 7.6% of control values, respectively, leading to a meanV E of 131.2 ± 12.4% of control values. This observation shows the stability of the respiratory variables during the period of retention in the recording chamber.
The mean f R was 166.1 ± 12.5 breaths·min −1 , the mean V T 35.6 ± 9.6 µl·g −1 , and the meanV E 5.4 ± 1.2 ml·g −1 ·min −1 during the normoxic control period in the unrestrained one-dayold mice that were subsequently submitted to hypoxia (n = 8). The meanV E tended to peak at 158.1 ± 29.4% of control values 2 min after the onset of hypoxic exposure (Figure 1). It then decreased significantly dropping below control (p < 0.04) and below the high values observed at the onset of hypoxia (p < 0.03) to 68.9 ± 10.0% of control values 25 min after the onset of hypoxic exposure (Figure 1). Changes inV E were related to the respective effects of hypoxia on both f R and V T . The mean f R was at 115.4 ± 19.0% of control values 2 min after the onset of hypoxic exposure and at 86.5 ± 7.5% of control values 25 min after the onset of the hypoxic test. The mean V T was at 138.8 ± 18.6% of control values 2 min after the onset of hypoxic exposure and at 73.7 ± 15.8% of control values 25 min after the onset of hypoxic exposure.

Brainstem Areas that Participate in Respiratory Control Have a Low Number of c-FOS-Positive Cells under Normoxic Conditions
One-day-old mice maintained under normoxic conditions (control animals) had a relatively small number of c-FOS-positive cells both in respiratory related areas of the medulla oblongata: NTS, VLM, RTN/pFRG, PP, 10N, 12N, and medullary raphe nuclei (Table 1;  FIGURE 1 | One-day-old mice displayed a ventilatory depression in response to hypoxia. Average percent change inV E from baseline at 2 min (peak HVR, black bar) and 25 min (white bar) of hypoxic exposure in one-day-old mice (n = 8). *p < 0.05 vs. baseline, # p < 0.05 vs. peak HVR.

Nucleus of the solitary tract
The three analyzed subdivisions of the NTS i.e., cNTS, mNTS, and vlNTS, especially the mNTS, had significantly more c-FOS positive cells under hypoxic than normoxic conditions (Table 1; Figures 2B,H; +148, +235, and +205%, respectively). Virtually none of the c-FOS-positive cells were immunolabeled for TH (Figures 3A,D); we observed no dually labeled cells in the cNTS and vlNTS and only 0.3% in the mNTS. In addition, some of the c-FOS-positive cells of the cNTS and mNTS were also immunoreactive for PHOX2B: 18 and 25%, respectively.

Dorsal motor nucleus of vagus
We observed a higher of number of c-FOS-positive cells in the 10N under hypoxia than normoxia, but the difference was not significant ( Table 1).

Ventrolateral reticular nucleus of the medulla
The cVLM, but not the rVLM, which encompasses the pre-Bötzinger complex, had more c-FOS-positive cells under hypoxic than normoxic conditions (Table 1; Figures 2A,B; +340%). Only a few of the hypoxic-c-FOS-positive cells of the cVLM were also immunoreactive for TH (Figures 3B,E; 5%).

Parapyramidal group
The group of cells adjacent to the RTN/pFRG, called the PP, also had more c-FOS-positive cells under hypoxia than normoxia (Table 1; Figures 2A-D,K,F; +315%). A substantial portion of these cells, 30%, were also immunoreactive for 5-HT (Figures 5A,C,E).

Medullary raphe nuclei
Among medullary raphe nuclei, the RMg displayed more c-FOSpositive cells under hypoxic than normoxic conditions (Table 1; Figures 2C,D; +139%). Only a small portion of RMg hypoxicc-FOS-positive cells, 15%, were also immunoreactive for 5-HT (Figures 5B,D,F). In contrast, there were fewer c-FOS-positive cells under hypoxia than normoxia in the ROb but the difference was not significant ( Table 1; (Table 1; Figures 2B,J).

Parabrachial and Kölliker fuse nuclei
In the PB complex, both the lPB and mPB showed substantially more c-FOS-positive cells under hypoxia than normoxia (

Locus coeruleus and subcoeruleus nucleus
There were significantly more c-FOS-positive cells in the dSubC under hypoxic conditions (  Figures 2E,F). Among the hypoxic-c-FOS-positive cells of the dSubC, 45% were also immunoreactive for TH (Figures 3C,F,G). Additionally, there was no difference in the number of c-FOS-positive cells in the LC (Table 1; Figures 2E,F).

A5 region
We observed no difference in the number of c-FOS-positive cells in A5 between hypoxic and normoxic conditions (

DISCUSSION
We described and characterized the brainstem neuronal network activated by hypoxia in one-day-old mice by analyzing c-FOS protein levels by immunohistochemistry. Our main finding is that one-day-old mice displayed activation of catecholaminergic cells of the dSubC, an area implicated in the strong depressive pontine influence under hypoxia previously reported in the fetus, but not in newborns or adults of other mammalian species (Breen et al., 1997;Teppema et al., 1997;Berquin et al., 2000a,b;Bodineau and Larnicol, 2001). Our results also revealed that the hypoxia-activated brainstem neuronal network of oneday-old mice is characterized by weak or absent activation of cells in areas involved in hypoxic hyperventilation i.e., catecholaminergic neurons of the VLM and PHOX2B cells of the retrotrapezoid/parafacial region Millhorn, 1991, 1994;Takakura et al., 2006) as suggested by the scarcity or

Whole Body Flow Barometric Plethysmography in One-day-old Mice
We measured respiratory variables using whole-body flow barometric plethysmography in freely moving newborn mice to limit confinement-related stress. This is a commonly used technique to measure pulmonary ventilation in unrestrained and non-anesthetized animals that consists of recording the changes of pressure in a chamber caused by breathing (Tree et al., 2014). As the inspired gas is warmed and humidified from the ambient to the pulmonary values, the total pressure in the recording chamber increases; the opposite occurs during expiration. The accuracy of V T calculations based on wholebody plethysmography data of small animals has been questioned (Enhorning et al., 1998;Mortola and Frappell, 1998). The body temperature, humidity, and temperature of the recording chamber are difficult to control due to the small size of the devices used for such small animals, leading to highly variable absolute V T andV E values. These values should thus be viewed with caution. This has led researchers to base their analysis on the changing values of V T andV E as a percentage of control values. Furthermore, small mammals, particularly newborns, develop hypoxic hypothermia due to hypometabolism and increased heat loss. In our experiments, the temperature of the animal was evaluated before and after each experiment via an oral probe because of the difficulty to measure body temperature continuously with our home-made plethysmograph. Thus, our plethysmographic data should be taken with caution. Indeed, the plethysmography was used primarily to simply assure the presence of a biphasic respiratory response in one day old mice, as classically described in other newborns (Simakajornboon et al., 2004;Bairam et al., 2013). In addition to these considerations, the pups were isolated from their mother for a long time during the plethysmographic recordings. We observed no significant modification of the respiratory variables in animals maintained under normoxia despite this long separation that may have led to physiological changes. This indicates that changes in the respiratory variables developed by newborn mice submitted to hypoxic conditions are effectively due to the reduction in O 2 , and not their long retention.
Neurons of the Subcoeruleus Nucleus, an Area Implicated in the Strong Depressive Pontine Influence under Hypoxia in the Fetus, Are Activated by Hypoxia in One-day-old Mice Analysis of c-FOS protein revealed increased activity in neurons of the mPB and dSubC. Such an increase has not been reported in other newborn rodents (see the Figure 3 in Berquin et al., 2000b). In the fetus of sheep, the participation of cells located in the dorsolateral part of the pons to HVD, more precisely those in the mPB, KF, and SubC, has already been discussed in the literature (Gluckman and Johnston, 1987;Walker, 1995;Breen et al., 1997;Nitsos and Walker, 1999;Walker et al., 2000;Teppema and Dahan, 2010). Hypoxia induces an increase in c-FOS expression in neurons of the mPB and SubC in fetal, but not newborn, sheep (Breen et al., 1997;Nitsos and Walker, 1999). The hypoxia-activated cells of the SubC have been proposed to be O 2 sensors and to strongly inhibit breathing in fetal sheep, but to lose this effect after birth because of their inhibition by peripheral chemoreceptors (Breen et al., 1997). The present observed increase in c-FOS-positive cells in the dSubC, of which a significant portion are catecholaminergic, may be indicative of activated cells with similar properties to those identified in the fetus by Walker and collaborators (Walker, 1995;Breen et al., 1997;Nitsos and Walker, 1999). In such a model, they may be involved in HVD in one-day-old mice and constitute a noninhibited fetal mechanism at birth different from that of other studied newborn mammals, such as newborn rats, which are more mature at birth than mice (Gauda, 2006). This feature of one-day-old mice may make this model particularly suitable for mechanistic studies on the occurrence of excessive HVD in premature newborns or even on the occurrence of SIDS. Indeed, SIDS has been hypothesized to result from the awakening of a defense mechanism of the fetus consisting of a strong depressive response to hypoxemia that limits O 2 consumption (Lavezzi, 2015). A recent publication by Lavezzi and collaborators emphasized hypoplasia and neurochemical alterations of KF neurons underlying the pathogenetic mechanisms of SIDS (Lavezzi, 2015). Our present data and previously published work in fetal and newborn sheep (Breen et al., 1997) underscore the interest in searching for possible alterations in the integrity of dSubC neurons in the brainstems of infants who died of SIDS.
In rats, SubC and mPB have been shown to contain premotoneurons that innervate hypoglossal motoneurons involved in maintaining upper airway patency during breathing (Dobbins and Feldman, 1995;Fay and Norgren, 1997). The presently observed decrease in c-FOS protein levels in the hypoglossal nucleus cells may be linked to an inhibitory drive coming from the dSubC and mPB during hypoxia. This hypothetic inhibitory pathway may contribute to HVD by reducing the airway opening when activated under hypoxia. Noradrenergic cells of the dSubC innervate the hypoglossal nucleus, of which the motoneurons are activated by noradrenalin (Aldes et al., 1992;Fenik et al., 2008;Funk et al., 2011). If dSubC cells are involved in the decrease in the expression of c-FOS in the hypoglossal nucleus, they would not be those immunoreactive for TH. Further experiments are necessary to determine the pharmacological phenotype of the c-FOS-positive, but TH-negative, cells of the dSubC.

Hypoxia in One-day-old Mice Induces an Increase in the Number of C-FOS-Positive Cells in the Nucleus Raphe Magnus, an Area Previously Described as Having a Hypoxic Respiratory Depressive Influence
We observed a large increase in the number of c-FOS-positive cells under hypoxia in the RMg. Such an increase has not been reported in newborns of other species (Breen et al., 1997;Teppema et al., 1997;Horn et al., 2000;Berquin et al., 2000b) or adults subjected to moderate hypoxia (Berquin et al., 2000a), but has been reported in adults under severe hypoxia (Erickson and Millhorn, 1994). Based on data from the literature, we suggest that hypoxia-activated RMg neurons contribute to HVD. Indeed, activation of RMg neurons attenuate the activation of NTS neurons by peripheral chemoreceptors (Perez and Ruiz, 1995) and RMg neurons exert a moderating influence on the V T under hypoxic conditions (Gargaglioni et al., 2003). Thus, activation of RMg neurons in one-day-old mice may participate in the HVD. With the exception of a few cells, the hypoxic c-FOS-positive cells that we observed were not serotoninergic. In addition to serotoninergic neurons, the RMg contains GABAergic neurons that could be involved in HVD. This is supported by the fact that iv administration of bicuculline, a GABA A antagonist, reduces the respiratory inhibition that occurs after electrical stimulation of the RMg (Cao et al., 2006).

Hypoxia Induces an Increase in c-FOS Expression in NTS and VLM, Two Structures that Participate in Hypoxic Hyperventilation, but There Is Little or No c-FOS Protein in Catecholaminergic Neurons
Hypoxia induced an increase in the number of c-FOSpositive cells in areas recognized to be involved in early hypoxic hyperventilation triggered by stimulation of peripheral chemoreceptors (see for revue Teppema and Dahan, 2010) i.e., the cNTS, mNTS, and lPB. The cNTS and mNTS constitute the major central site for the integration of inputs from peripheral chemoreceptors whereas only few afferents from peripheral chemoreceptors project toward the VLM (Finley and Katz, 1992). The two subnuclei of the NTS are the first relay between peripheral afferences and the VLM and lPB (Nunez-Abades et al., 1993;Hayward and Felder, 1995). All of these connections conceivably provide a common basis for the hypoxic increase in c-FOS expression that we observed in these structures in oneday-old mice and that has been observed in other species at several stages of development (Teppema et al., 1997;Berquin et al., 2000b).
The dual detection of c-FOS and TH revealed that no, or only very few, neurons activated by hypoxia in the cNTS, mNTS, and VLM were catecholaminergic. Previous data about the catecholaminergic character of c-FOS-positive hypoxia-activated cells in these structures made in adult mammals have showed that no, or very few, c-FOS hypoxia-activated neurons in the NTS are also catecholaminergic, whereas a large portion of c-FOS hypoxia-activated neurons in the VLM are also immunoreactive for TH (Erickson and Millhorn, 1994;Soulier et al., 1997;Teppema et al., 1997). The only data from the literature concerning this issue in newborn animals is the study of c-FOS expression in newborn rats under high-altitude conditions showing that an altitude of 8000 m induces an increase in the number of c-FOS positive cells in the NTS and VLM with a large portion of them in the VLM also immunoreactive for TH (Kaur et al., 2001). Taking into account these data, our results suggest that the effect of hypoxia on the VLM differed in one-day-old mice from other studied newborn or adult mammals. This difference may depend on the maturation state of either VLM neurons or brainstem connections. The absence of VLM catecholaminergic neuron activation in one-day-old mice could conceivably participate in HVD, and even strengthen it in this model, as they have been associated with hypoxic hyperventilation (Erickson and Millhorn, 1994;Soulier et al., 1997;Teppema et al., 1997). If this observation is likely due to the degree of immaturity of the central nervous system of the oneday-old mice as we assume (Teppema et al., 1997;Gauda, 2006;Mallard and Vexler, 2015), a similar mechanism may be involved in exacerbating the respiratory problems encountered at birth in premature infants.
The Ventral Medullary Surface Displays Increased c-FOS Expression in One-day-old mice in Serotonin-but Not PHOX2B-Positive Neurons We observed an increase in the number of c-FOS-positive cells in the RTN/pFRG. The RTN/pFRG is involved in the respiratory adaptation to both hypercapnia (Teppema et al., 1997;Nattie, 2001;Okada et al., 2002;Guyenet et al., 2005;Stornetta et al., 2006;Teppema and Dahan, 2010;Guyenet and Bayliss, 2015) and hypoxia Bodineau and Larnicol, 2001;Takakura et al., 2006;Voituron et al., 2006Voituron et al., , 2011Teppema and Dahan, 2010). The RTN/pFRG contains both CO 2 and O 2 sensor cells (Guyenet et al., 2005;Voituron et al., 2006Voituron et al., , 2011Onimaru et al., 2008;Lazarenko et al., 2009;Guyenet and Mulkey, 2010) and also integrates chemosensory inputs from peripheral chemoreceptors (Berquin et al., 2000a;Bodineau et al., 2000b;Takakura et al., 2006). The well-identified CO 2 -sensor cells of the RTN/pFRG (Lazarenko et al., 2009;Guyenet and Mulkey, 2010) are derived from neurons that express Phox2b, Atoh-1, and Egr-2, and are characterized by the presence of PHOX2B, NK1 receptors, VGLUT2, TASK-2, GPR4, and GALANIN (Weston et al., 2004;Stornetta et al., 2006;Onimaru et al., 2008;Dubreuil et al., 2009;Rose et al., 2009;Guyenet and Mulkey, 2010;Guyenet and Bayliss, 2015;Ruffault et al., 2015). In the present work, none of the c-FOS positive cells of the RTN/pFRG were PHOX2B-positive. This suggests that although the one-day-old mice displayed severe hypoventilation under hypoxia, hypoventilation may not entail an increase in CO 2 that should have been detected by the PHOX2B cells of the RTN/pFRG. Regardless of the mechanisms responsible for HVD, the concomitant hypometabolism that maintains isocapnia, which constitutes a feature of the newborn response to hypoxia (Mortola, 2004), appears to be extremely effective in one-day-old mice. The activation of PHOX2Bnegative RTN/pFRG cells may depend on intrinsic O 2 -sensing properties that we previously demonstrated by measuring c-FOS expression in brainstem spinal cord preparations from newborn rodents (Voituron et al., 2006. It is unlikely that the increase in the number of c-FOS-positive/PHOX2B-negative cells in RTN/pFRG depends on peripheral chemoreceptor inputs. Takakura and collaborators have shown that such indirect activation implicates cNTS glutamatergic neurons projecting to PHOX2B RTN/pFRG cells . This suggests possible immaturity of the connection between cNTS cells activated by peripheral chemoreceptors and RTN/pFRG PHOX2B-positive cells. We observed a greater increase in c-FOS expression in the more medial group, called PP, with 30% of the c-FOS-positive cells containing 5-HT, than for the RTN/pFRG. This group of cells has been suggested to be the positional homolog of the human medullary arcuate nucleus (Filiano and Kinney, 1992), a structure in which abnormalities have been reported in infants who died of SIDS (Filiano and Kinney, 1992;Paterson et al., 2006;Kinney et al., 2011). The increase in c-FOS protein levels induced by hypoxia in the PP is of particular interest given the possible involvement of both the arcuate nucleus and abnormalities in the respiratory response to hypoxia in SIDS (Paterson et al., 2006;Kinney et al., 2011). The PP contains GABAergic and serotoninergic neurons (Stornetta and Guyenet, 1999;Weston et al., 2004;Stornetta et al., 2005) and displays multiple sites of projections including the pre-Bötzinger complex and intermediolateral column (Holtman et al., 1990;Jansen et al., 1995). Functionally, the PP has been reported to be involved in autonomic regulation under hypoxia. Darnall and collaborators reported that the destruction of 5-HT medullary neurons, including those of the ROb and PP increase the arousal latency from sleep induced by hypoxia in newborn rats (Darnall et al., 2016). In the relatively immature one-day-old mouse (Gauda, 2006), we only observed an increase in c-FOS-and 5-HT-positive cells in the PP, but not in the ROb, where the number of c-FOS-positive cells tended to decrease. Thus, our results suggest that 5-HT PP neurons must have a critical role in arousal from sleep under hypoxia in immature newborn mammals. Our results combined with the fact that hypoplasia of this region has been shown in newborn death due to SIDS, suggests that the dysfunction of 5-HT neurons of the PP in infants, and particularly in premature infants, could result in a high risk situation due to the decrease of arousal from sleep during hypoxia.

CONCLUSION
This study significantly contributes to the knowledge of key brainstem cell populations, for which the activity is modulated during hypoxia using an animal model characterized by its immaturity relative to other mammals, the one-day-old mouse (Gauda, 2006;Gaultier et al., 2006;Teppema and Dahan, 2010;Darnall et al., 2016;Mallard and Vexler, 2015). Our results highlight changes in the activity of cell populations that may participate in the respiratory depression of this animal model i.e., the activation of catecholaminergic cells of the dSubC, an area previously implicated in a strong depressive pontine influence in the fetus but not in newborns. Also, we did not observe an increase in the number of c-FOS-positive cells commonly associated with the development of hypoxic hyperventilation, catecholaminergic cells of VLM, and PHOX2B-postive neurons of the RTN/pFRG. This was suggested by the scarcity or absence of TH-positive or PHOX2B-positive cells among the c-FOSpositive-cells. Finally, our results suggest that 5-HT neurons of the PP, shown to be involved in arousal from sleep, are the only serotoninergic medullary neurons activated by hypoxia in immature mammals. Some physiopathological conditions in which depressant and arousing mechanisms would be more or less potent might lead to life-threatening situations, especially in premature infants. In conclusion, one-day-old mice highlight characteristics to model dysfunction of the breathing network in premature infants. In the absence of data in the literature, future experiments to explore c-FOS expression in older mice, displaying a mature hypoxic ventilatory response, would help to exclude the possibility that the pattern of c-FOS expression observed in one-day-old mice is simply species specific and not due to their relative immaturity.