Sixteen preterm lambs were born vaginally on gestational day 133 (term = 147 days) as previously described (Boudaa et al., 2013). Results on the effects of HB on bottle-feeding efficiency and nutritive swallowing-breathing coordination obtained in the same preterm lambs have previously been published (Bourgoin-Heck et al., 2015). Premature labor was induced by mifepristone (8 mg/kg), preceded by betamethasone (12 mg × 2). A unique preterm lamb model for moderate HB was designed for the purposes of the study (Bourgoin-Heck et al., 2015). Briefly at birth, lambs were randomly assigned to either the HB group or control group. On postnatal day 5, HB was induced by an intravenous infusion of 20 mg/kg of the bilirubin solution, stabilized with albumin and diluted in Ringer's Lactate buffer (Hansen et al., 1992; Mesner et al., 2008; Bourgoin-Heck et al., 2015), whereas control lambs received a bilirubin-free solution. Bilirubinemia was maintained between 150 and 250 μmol/L for a period of 17 h.

Neurological consequences of HB were characterized based on the “Lamb Acute Bilirubin Encephalopathy score” as well as on consistent electroencephalogram changes, as previously described (Bourgoin-Heck et al., 2015). Accordingly, the presence of acute bilirubin encephalopathy was assessed according to published guidelines for the detection of HB in newborns (American Academy of Pediatrics Subcommittee on Hyperbilirubinemia, 2004). Clinical examinations (consciousness, muscle tone, and audition) were repeated every 2 h throughout bilirubin infusion, then twice daily until the end of the experiment. Detailed results of the LABE score are presented for every lamb in Table 1 of the digital supplement of Bourgoin-Heck et al. (2015). In addition to the clinical detection of acute bilirubin encephalopathy, an electroencephalogram (EEG) was recorded for 6 h during bilirubin infusion and again 6 h at D3, in particular for detecting alterations in sleep organization or epileptiform discharges in HB lambs.

Finally, daily monitoring included vital signs (heart rate, respiratory rate, pulse oximetry, rectal temperature), weight and blood samples for albumin, glucose, bilirubin, and arterial blood gases.

Animals were chronically instrumented on postnatal day 5 and thereafter, and included pairs of needle-electrodes inserted subcutaneously: (i) above the occipital cortex for electroencephalogram (EEG); (ii) near to the right eye socket for electrooculogram (EOG); and (iii) on both sides of the thorax for electrocardiogram (ECG). In addition, the following were implanted: (iv) catheters into both jugular veins (IV infusion and injection; blood sampling); (v) a 5 Fr naso-pharyngeal catheter with its tip positioned 0.5 cm above the posterior border of the soft palate [induction of laryngeal chemoreflexes (LCR)]; (vi) elastic bands around the thorax and abdomen for respiratory inductance plethysmography (respiratory movements; NIMS, Miami Beach FL, USA); (vii) a Masimo pulse oximeter sensor at the tail base [arterial oxygen hemoglobin saturation (SpO₂)]. Lastly, in order to assess the response to hypoxia, a custom-built nasal mask molded to fit the lamb's muzzle (Samson et al., 2005) was connected to a pneumotachograph (flow transducer 47304A Hewlett-Packard, Hewlett-Packard Company, Palo-Alto, CA, USA) and a 2-way system which provided either medical air or a gas mixture of 10% O₂ in air. Arterial blood gases were also measured in two lambs from each group during hypoxia from an arterial brachial catheter surgically inserted under local anesthesia, prior to recordings.

Leads from all sensors were connected to our custom-built radio telemetry transmitters (Samson et al., 2011). All signals were continuously recorded using AcqKnowledge software (version 4.1, Biopac Systems Canada Inc., Montreal, Qc, Canada) and the lambs were filmed using a web-cam.

All preterm lambs were housed and cared for with their dams in our animal quarters until the experimental day. Polysomnographic recordings were first performed on postnatal day 5, named experimental day 0 (D0), to assess the immediate effects of moderate HB on baseline cardiac and respiratory activity, as well as on cardiac and respiratory responses to several chemical stimuli, including laryngeal and pulmonary chemoreflexes and hypoxia (see below and Figure 1). Recordings were repeated 72 h after HB induction (D3) to assess the delayed effects of moderate HB.

During baseline respiration, breathing frequency of HB lambs was lower than controls on both D0 and D3 (Table 1). In addition, the time spent in apnea was significantly increased in the HB group on D0 (Table 1). No other differences were observed for baseline cardiac or respiratory activity.

Hematoxylin-eosin stained brain sections from the three HB animals euthanized at the peak of hyperbilirubinemia (D0) showed notable bilirubin deposits in the leptomeninges throughout the entire brain, including the brainstem (Figure 3B). While minimal bilirubin deposition was present in brain parenchyma, Alzheimer type II astrocytes [characterized by enlarged, vesicular, and at times empty-looking, nuclei with marginated chromatin and scanty cytoplasm (Ellison et al., 2013)] were noted throughout the brain, including the brainstem (Figure 3A).

Histopathological examination of the eosin-hematoxylin stained brain sections in the 11 preterm lambs euthanized at D3 showed no bilirubin deposition in any of the HB lambs. In addition, compared to controls, there were only a few foci of slight astrogliosis in the cerebral hemispheres and cerebellum in all HB lambs, as well as in the brainstem in two HB lambs.

Compared to previous animal models of neonatal HB (Ahlfors et al., 1986; Brann et al., 1987; Park et al., 2001; Mesner et al., 2008; Ye et al., 2012), our preterm ovine model of moderate HB has several unique features. Indeed, in order to be clinically relevant, we induced a moderate (150–250 μmol/L) and sustained (17 h) HB in a (late) preterm animal. In contrast, previous studies reported data from more severe (from 220 to 1300 μmoles/L) and shorter duration HB (ranging from one injection to 4 h infusion) (Ahlfors et al., 1986; Brann et al., 1987; Park et al., 2001; Mesner et al., 2008; Ye et al., 2012). To our knowledge, there has been only one previous study on “moderate” HB (100–200 μmoles/L), but with considerably shorter HB duration (2 h) (Roger et al., 1995). The clinical relevance of our model is further shown by the obvious clinical manifestations (jaundice and neurological dysfunction), together with EEG signs of acute bilirubin-encephalopathy observed in all HB lambs. All the aforementioned abnormalities were reversed after bilirubin normalization, as reported in newborn infants during acute bilirubin-induced neurological dysfunction (American Academy of Pediatrics Subcommittee on Hyperbilirubinemia, 2004).

Observations of decreased baseline respiratory rate and increased time spent in apnea in HB lambs are in agreement with several clinical studies (Johnson et al., 2009; Amin et al., 2014; Amin and Wang, 2015). Such HB-related respiratory inhibition could potentially be mediated, at least in part, by prostaglandin E₂ binding to its respective receptors in respiratory regions of the brainstem (Hofstetter et al., 2007).

The present results also reveal that moderate HB transiently blunted the cardiorespiratory inhibition normally observed during both laryngeal and pulmonary chemoreflexes. Such observations were somewhat unexpected, given that various postnatal stresses have been shown to increase apneas and bradycardias observed in the LCR (Lindgren et al., 1996; Frøen et al., 2000; Xia et al., 2008, 2010; St-Hilaire et al., 2010; Carreau et al., 2011). Nevertheless, the blunting of LCR observed in HB lambs in the present study may be non-beneficial, due to the heightened risk of aspiration as a result of the lack of efficient glottal closure. Unfortunately, this cannot be ascertained, for the design of the present study did not include the assessment of tracheal aspiration. The HB-related alteration in cardiorespiratory activity was further confirmed by the present observations in HB lambs, namely that (i) the cardiorespiratory inhibition normally triggered by pulmonary C-fiber stimulation was blunted, and (ii) the heart rate was increased in response to acute hypoxia. The above results expand previous data obtained with severe HB in rat pups showing an altered respiratory response to hypoxia and hypercapnia (Mesner et al., 2008).

In the three preterm lambs euthanized at the peak of HB, bilirubin deposition was observed only within the leptomeninges overlying the entire brain. This is somewhat at odds with a previous report in rat pups with severe HB levels, in which bilirubin deposition was noted in the brainstem itself (Mesner et al., 2008). Such differences may be related to the lower HB level and/or to a lower permeability of the blood brain barrier in preterm lambs compared to rat pups. In addition, acute HB was associated herein with Alzheimer type II astrocytes throughout the entire brain of the three lambs, including in the brainstem. Alzheimer type II astrocytes have been reported in various acute central nervous system injuries, such as stroke or trauma. They may represent dysfunctional astrocytes (Adams and Parker, 2011) and are thought to represent glial cells in the process of transformation into reactive astrocytes (Panickar and Norenberg, 2005). Of interest, the presence of Alzheimer type II astrocytes in the cerebral cortex (Figure 3A), indicative of brain injury, was noted in concomitance with the monotonous EEG observed in all HB lambs.

In addition to the above, astrocyte and microglial activation in the NTS were detected 3 days following acute HB in preterm lambs. Cerebral glial activation, encompassing both astrocytes and microglia, has been previously reported in response to bilirubin (reviewed in Fernandes and Brites, 2009). The present observation, to the best of our knowledge, is the first report of glial activation in the NTS in response to HB. These findings provide further evidence that even a moderate HB in preterm lambs can induce an inflammatory response in the NTS region devoted to central processing of cardiorespiratory afferent inputs (Kubin et al., 2006). Both activated microglia and astrocytes could alter neuronal function such as synaptic transmission, including that involved in cardiovascular and respiratory reflexes in the NTS (Andresen and Paton, 2011). The results herein on HB-induced alterations in laryngeal and pulmonary chemoreflexes, as well as in the cardiac response to hypoxia, are in keeping with this hypothesis. Our findings are in agreement with a previous study showing abnormal ventilatory response to CO₂ and hypoxia in rat pups with severe HB (Mesner et al., 2008). Of note, astrocytic damage in the NTS has previously been shown to interfere with cardiovascular reflex transmission by attenuating the arterial baroreflex, the arterial chemoreflex, and the von Bezold–Jarisch cardiopulmonary reflex (Lin et al., 2013). In addition, migroglial activation in the NTS was previously implicated in the impaired cardiovascular regulation observed in diabetic rats (Rana et al., 2014). Of note, the absence of a detectable bilirubin deposition in the brain parenchyma may suggest that HB in our preterm lambs exerts its effects on glial cells via the release of certain mediators by cells forming the blood brain barrier, including astrocytes (Sofroniew, 2015).

As stated earlier, our ovine model provides unique characteristics (preterm birth, both moderate and sustained hyperbilirubinemia), which are of particular relevance for assessing the physiological consequences of hyperbilirubinemia in human newborns. One limitation may be that, contrary to humans, preterm newborn lambs (for yet unknown reasons) do not naturally develop hyperbilirubinemia.

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.