Specific pathogen-free (SPF) grade Sprague-Dawley rats weighing 250 ± 50 g were acquired from the Experimental Animal Center of Xuzhou Medical University. Rats successfully injected with the tracer into the CSF-contacting nucleus were used for observation and analysis (n = 6). All the animals were housed under a 12/12 light/dark cycle at the temperature of 22–25°C, with ad libitum access to food and water.

Rats were anesthetized with pentobarbital sodium (40 mg/kg, i.p.), and their heads were fixed on the stereotaxic instrument (Stoelting 51700, United States). Retrograde tracer CB solution (0.2 μl, 1%; Sigma, United States) was injected into the core of the CSF-contacting nucleus (Posterior to Bregma: 8.24 ± 0.18 mm, Lateral: 0.09 ± 0.01 mm, Depth: 6.45 ± 0.11 mm) (Song et al., 2019) by using Hamilton syringe with a 33G needle tip (Hamilton Company, Switzerland). The injection was applied under the microinfusion pump (KD Scientific, United States) for over 30 min periods. At the end of the injection, the microsyringe was left in place for 10–15 min before retraction.

After 7–10 days of retrograde tracer CB injections, rats were perfused transcardially and sacrificed. The whole brain and spinal cord were isolated and sectioned (40 μm thickness) coronally using a cryostat (Leica CM1900, Germany). All the sections were placed in sequence and numbered. In this study, only the brainstem and spinal cord were analyzed.

After CB immunofluorescent staining (rabbit anti-CB primary antibody diluted in 1:600, Abcam; donkey anti-rabbit Alexa Fluor 488 secondary antibody diluted in 1:200, Life Technologies), the sections were sequentially mounted on slides, counterstained with DAPI, and coverslipped. The sections of the brainstem and spinal cord were visualized using fluorescence (Leica DM6, Germany) and confocal laser microscopes (Zeiss LSM 880, Germany) for different positions at the uniform standard. The cell density of CB-positive neurons (cell number/0.2 mm² area) in each brain region was calculated using Image-Pro Plus 7.0 software. The density of CB-positive neurons was classified according to the densities: <5, 6–10, and >10, which correspond to sparse, moderate, and dense distributions, respectively.

The recombinant adenoassociated virus vector (serotype 2/9) expressed enhanced GFP (EGFP) under control of a human synapsin I promoter was used for anterograde tracing to verify the retrograde tracing results. 0.2 μl rAAV2/9-hSyn-EGFP (1.2 × 10¹² gene copies; Brainvta Company, Wuhan, China) was injected into the raphe magnus nucleus (RMg) (Posterior to Bregma: 10.92 ± 0.2 mm, Lateral: 0 mm, Depth: 10.5 ± 0.1 mm) (Paxinos and Watson, 2007). 1 μl (1 μg) Cholera toxin subunit B Alexa Fluor 555 conjugate (CB-555) (Brainvta Company, Wuhan, China) was microinjected into the lateral ventricle (Posterior to Bregma: 1.0 ± 0.1 mm, Lateral: 1.6 ± 0.1 mm, Depth: 4.0 ± 0.2 mm) (Paxinos and Watson, 2007) to label the CSF-contacting nucleus. The injection parameters were the same as above. Rats were perfused after 14–21 days of anterograde tracer rAAV2/9 injections. The axon terminals from the RMg to the CSF-contacting nucleus were captured under confocal laser microscopes (Zeiss LSM 880, Germany).

The CB-positive neurons were aligned, segmented, and registered into rat common reference atlas (Paxinos and Watson, 2007). The 3D connections were reconstructed by using Imaris software version 8.4.1 (Bitplane, United States).

SPSS 13.0 software was used for data analysis in the present study. Data are presented as mean ± SD.

Injections of the CB tracer produced dense immunofluorescence staining (green). The tracer was confined within the boundary of the CSF-contacting nucleus (Figures 1A,B), where the microsyringe needle tract can be seen to be located at the core of the CSF-contacting nucleus. Representative sections of the CSF-contacting nucleus are shown in Figures 1C,D.

After the retrograde tracer was injected into the CSF-contacting nucleus, it was transported in a retrograde manner along the axons; consequently, the neural somata projected from the brainstem and spinal cord can be detected.

The CB-positive neurons were round or fusiform in shape. Sizes of these neurons are different, and the processes are obvious (Figure 2A). In the reticular structure of the brainstem, the neurons are large and appear in polygonal shape; they have many processes, and the processes have many branches (Figures 2B,C). In the spinal cord, the neurons are fusiform, and the processes are obvious and longer (Figure 2D).

The brainstem and spinal cord projections to the CSF-contacting nucleus can be identified by CB labeling.

In the midbrain, CB-positive neuron can be found in the precommissural nucleus (PrC), periaqueductal gray (PAG), nucleus of Darkschewitsch (DK), medial accessory oculomotor nucleus (MA3), pre-Edinger-Westphal nucleus (PrEW), Edinger-Westphal nucleus (EW), rostral linear nucleus of the raphe (RLi), interfascicular nucleus (IF), interstitial nucleus of Cajal (InC), red nucleus (R), ventral tegmental area (VTA), substantia nigra (SN), medial pretectal nucleus (MPT), posterior pretectal nucleus (PPT), olivary pretectal nucleus (OPT), nucleus of the optic tract (OT), posterior thalamic nuclear group (Po), interpeduncular nucleus (IP), supraoculomotor cap (Su3C), supraoculomotor periaqueductal gray (Su3), oculomotor nucleus (3N), trochlear nucleus (4N), mesencephalic reticular formation (mRt), superior colliculus (SC), inferior colliculus (IC), caudal linear nucleus of the raphe (CLi), subbrachial nucleus (SubB), and retrorubral field (RRF) (Figures 3, 4).

Among them, the PrC, PAG, MA3, IP, Su3C, Su3, 3N, 4N, and SubB have dense projections; the PrEW, EW, IF, SN, MPT, PPT, OPT, OT, and mRt send moderate projections; and the DK, RLi, InC, R, VTA, Po, SC, IC, CLi, and RRF send sparse projections to the CSF-contacting nucleus (Figures 3, 4).

In the pons, CB-positive neuron can be found in the pontine nuclei (Pn), dorsal raphe nucleus (DR), median raphe nucleus (MnR), paramedian raphe nucleus (PMnR), isthmic reticular formation (isRt), pedunculopontine tegmental nucleus (PTg), subpeduncular tegmental nucleus (SPTg), microcellular tegmental nucleus (MiTg), precuneiform area (PrCnF), cuneiform nucleus (CnF), laterodorsal tegmental nucleus (LDTg), medial paralemniscial nucleus (MPL), nucleus of the lateral lemniscus (LL), A7 noradrenaline cells (A7), central gray (CG), pontine reticular nucleus oral part (PnO), pontine reticular nucleus caudal part (PnC), pontine reticular nucleus ventral part (PnV), Barrington’s nucleus (Bar), subcoeruleus nucleus (SubC), lateral parabrachial nucleus (LPB), medial parabrachial nucleus (MPB), Kolliker Fuse nucleus (KF), supratrigeminal nucleus (Su5), motor nucleus of trigeminal nerve (Mo5), medioventral periolivary nucleus (MVPO), lateroventral periolivary nucleus (LVPO), locus coeruleus (LC), superior vestibular nucleus (SuVe), medial vestibular nucleus parvicellular part (MVePC), medial vestibular nucleus magnocellular part (MVeMC), lateral vestibular nucleus (LVe), spinal vestibular nucleus (SpVe), nucleus of origin of efferents of the vestibular nerve (EVe), supragenual nucleus (SGe), ventral cochlear nucleus (VC), dorsal cochlear nucleus (DC), lateral superior olive (LSO), superior paraolivary nucleus (SPO), nucleus of the trapezoid body (Tz), abducens nucleus (6N), paraabducens nucleus (Pa6), raphe interpositus nucleus (RIP), raphe pallidus nucleus (RPa), dorsal paragigantocellular nucleus (DPGi), gigantocellular reticular nucleus (Gi), gigantocellular reticular nucleus alpha part (GiA), gigantocellular reticular nucleus ventral part (GiV), raphe magnus nucleus (RMg), facial nucleus (7N), A5 noradrenaline cells (A5), intermediate reticular nucleus alpha part (IRtA), parvicellular reticular nucleus alpha part (PCRtA), parapyramidal nucleus (PPy), lateral paragigantocellular nucleus alpha part (LPGiA), and lateral paragigantocellular nucleus external part (LPGiE) (Figure 5).

Among them, the Pn, DR, MnR, CnF, LDTg, CG, Bar, LPB, KF, LC, MVePC, EVe, SGe, DC, and 6N have strong projections; the PMnR, PTg, MPB, MVPO, SuVe, MVeMC, VC, TZ, RPa, RMg, LPGiA, and LPGiE have moderate projections; and the isRt, SPTg, MiTg, PrCnF, MPL, LL, A7, PnO, PnC, PnV, SubC, Su5, Mo5, LVPO, LVe, LSO, SPO, Pa6, RIP, DPGi, Gi, GiA, GiV, 7N, A5, IRtA, PCRtA, and PPy have sparse projections to the CSF-contacting nucleus (Figure 5).

In the medulla oblongata and spinal cord, CB-positive neurons can be found in the prepositus nucleus (Pr), solitary nucleus (Sol), raphe obscurus nucleus (Rob), epifascicular nucleus (EF), rostroventrolateral reticular nucleus (RVL), caudoventrolateral reticular nucleus (CVL), spinal trigeminal nucleus (Sp5), ambiguus nucleus (Amb), Botzinger complex (Bo), nucleus of Roller (Ro), external cuneate nucleus (ECu), cuneate nucleus (Cu), inferior olive (IO), dorsal motor nucleus of vagus (10N), hypoglossal nucleus (12N), rostral ventral respiratory group (RVRG), paratrigeminal nucleus (Pa5), lateral reticular nucleus (LRt), gracile nucleus (Gr), area postrema (AP), A1 noradrenaline cells (A1), A2 noradrenaline cells (A2), accessory nerve nucleus (11N), intermediate reticular nucleus (IRt), parvicellular reticular nucleus (PCRt), medullary reticular nucleus dorsal part (MdD), medullary reticular nucleus ventral part (MdV), and spinal cord (Figures 6–10).

Among them, the Pr, Sol, RVL, CVL, Amb, 10N, 12N, AP, and A1 have strong projections; the Rob, Sp5, Bo, IO, LRt, A2, 11N, and IRt have moderate projections; and EF, Ro, ECu, Cu, RVRG, Pa5, Gr, PCRt, MdD, MdV, and spinal cord have sparse projections to the CSF-contacting nucleus (Figures 6–10).

In summary, CB-positive neurons were distributed in 4 functional areas including 112 sub-regions in the brainstem and spinal cord. However, their distribution ranged from sparse to dense in each functional region. Other parts did not contain CB-positive neurons. In addition, the retrograde results were further verified by the anterograde method. After the anterograde tracer rAAV2/9 injection into the RMg, large number of axon terminals can be detected in the CSF-contacting nucleus (Figure 11).

The distributions of CB-positive neurons throughout the brainstem and spinal cord were three dimensionally reconstructed. The densities of the connections were obvious in the 3D view. The red areas indicated dense connections (PrC, PAG, MA3, IP, Su3C, Su3, 3N, 4M, SubB, Pn, DR, MnR, CnF, LDTg, CG, Bar, LPB, KF, LC, MVePC, EVe, SGe, DC, 6N, Pr, Sol, RVL, CVL, Amb, 10N, 12N, AP, and A1); green areas indicated moderate connections (PrEW, EW, IF, SN, MPT, PPT, OPT, OT, mRt, PMnR, PTg, MPB, MVPO, SuVe, MVeMC, VC, Tz, RPa, RMg, LPGiA, LPGiE, ROb, Sp5, Bo, IO, LRt, A2, 11N, and IRt); and blue areas indicated sparse connections (DK, RLi, InC, R, VTA, Po, SC, IC, CLi, RRF, isRt, SPTg, MiTg, MPL, LL, A7, PnO, PnC, PnV, SubC, Su5, Mo5, LVPO, LVe, SpVe, LSO, SPO, Pa6, RIP, DPGi, Gi, GiA, GiV, 7N, A5, IRtA, PCRtA, PPy, EF, Ro, ECu, Cu, RVRG, Pa5, Gr, PCRt, MdD, MdV, and spinal cord) to the CSF-contacting nucleus (Figure 12).

In the brainstem and spinal cord, the CB-positive neurons were found in 112 sub-regions. The amount of projection from these regions to the CSF-contacting nucleus is shown in Figure 13.

The pain descending inhibitory system is located in the brainstem. This system can send fibers in a top-down manner to the dorsal spinal horn to inhibit pain transmission (Millan, 2002). The system comprises of the PAG, DR, rostral-ventromedial medulla (RVM), and LC (Reynolds, 1969; Francois et al., 2017; Torralba et al., 2018; Zhang et al., 2018). The system also sends dense connections to the CSF-contacting nucleus, which might be a novel pathway in pain modulation.

Apart from the descending inhibitory system, other regions (CnF, PTg, A5, A7, LRt, MdD, and spinal cord) (Zemlan and Behbehani, 1988; Dugast et al., 2003; Mravec et al., 2012; Huma et al., 2015; Francois et al., 2017; de Oliveira et al., 2018) can also regulate nociception and send projections to the CSF-contacting nucleus. For example, deep brain stimulation of the CnF neurons is known to produce antinociceptive effect (Zemlan and Behbehani, 1988). Lesion in the PTg impair the antinociceptive effects (de Oliveira et al., 2018).

In the brainstem, the Bar, CnF, Gi, LPB, MPB, PTg, 10N, A1, Amb, Bo, RVL, RVRG, and Sol are important for visceral activity modulation; they have extensive connections with the CSF-contacting nucleus. The Bar engages in modulating the bladder functions (Blanco et al., 2014). Manipulation of the neuronal functions in the CnF, Gi, PTg, A1, Amb, and RVL has great effects on cardiovascular responses (Chan and Chan, 1983; Elliott et al., 1985; Balan Junior et al., 2004; Topchiy et al., 2010; Shafei et al., 2012; Brailoiu et al., 2017). The Bo and RVRG are responsible for respiratory control. The Bo has been described to be involved in the compensatory responses to hypoxia (Alberto et al., 2014), while neurons of the RVRG can directly project to the phrenic motor neurons and discharge an inspiratory pattern (Hayakawa et al., 2004).

The LPB and MPB are located around the superior cerebellar peduncle (scp). They can receive various visceral inputs from the Sol and spinal cord and can relay the visceral information to other regions of the brain (Palmiter, 2018). The Sol is one of the most important sites involved in visceral modulation, and it is regarded as the entry point of afferent signals from various visceral mechanosensors, chemosensors, and C-fibers (Kotmanova et al., 2018; You et al., 2018). The 10N can integrate information and control the visceral activities (Browning et al., 2005; Cruz et al., 2007).

The CSF-contacting nucleus receives broad projections along the brainstem reticular formation—a region of the brain that plays a key role in regulating sleep and arousal (Moruzzi and Magoun, 1949). The PnO is a part of the ascending reticular activating system, and it regulates rapid eye movement sleep (REMS) and wakefulness (Nunez et al., 2010; Silvia et al., 2013; Vanini and Baghdoyan, 2013). The Gi has projections to the midbrain reticular formation and intralaminar thalamic nuclei, which help modulate arousal (Martin et al., 2011). During the REMS, activity of the LRt neurons changes significantly (Stettner et al., 2013).

Apart from the brainstem reticular formation, other regions of the brain that innervate the CSF-contacting nucleus, such as the DR and LC, can also modulate sleep and arousal. Pharmacological intervention the DR can significantly change REMS (Monti and Jantos, 2018). The LC is a key arousal node, and it is pivotal for modulating consciousness and effects of general anesthetics (Du et al., 2018).

The CSF-contacting nucleus receives projections from the EW, IP, PAG, DR, LC, and MnR, which might be associated with emotion. The EW is a region implicated in stress responses and different aspects of stress-related behaviors (Xu et al., 2010, 2014). Lesion-based studies demonstrated that the IP is involved in regulating anxiety and fear (Mclaughlin et al., 2017). The PAG is regarded as an integrated site for coordinating rapid responses to aversive stimuli, such as those provoking panic, anxiety, or fear (Lowery-Gionta et al., 2018). The DR and MnR contain the major serotonergic (5-HT) populations (Pollak Dorocic et al., 2014). Both of the nuclei and 5-HT are involved in anxiety and depression, and they are crucial and central drug targets (Lanzenberger et al., 2012; An et al., 2013, 2016; Lopez Hill et al., 2013; Campos et al., 2018; de Souza et al., 2018; Donner et al., 2018). The LC is the major noradrenergic system with regard to stress and major depressive disorder (Chandley and Ordway, 2012; Fan et al., 2018).

The CSF-contacting nucleus receives input from the EW, IP, LDTg, and LPGi, which might be associated with drug addiction. After oral self-administration of ethanol, the EW shows significantly elevated c-Fos expression (Bachtell et al., 1999; Giardino et al., 2011). Pharmacological studies show that the IP mediates behavioral responses to nicotine reinforcement, aversion, and withdrawal (Ramiro et al., 2009; Antolin-Fontes et al., 2015; Morton et al., 2018). The LDTg is involved in drug reward processing and reinforcement in the development of addiction and abuse (Kaneda, 2018). The LPGi has been shown to play an important critical role in the development of opiate tolerance and dependence (Ahmadi-Soleimani et al., 2017).

In this study, we used tract-tracing method to reveal the CSF-contacting nucleus input patterns from the brainstem and spinal cord. The unique morphological feature of the CSF-contacting nucleus is that the somata are located in the brain parenchyma, which can receive inputs from the above-mentioned areas; the processes can form synaptic and non-synaptic connections with non-CSF-contacting neurons, CSF, or plasma. For one aspect, it might form the brainstem and spinal cord→CSF-contacting nucleus→non-CSF-contacting neurons circuits and participate in the regulation of life activities via neuron-neuron crosstalk. For another aspect, it might be involved in the brainstem and spinal cord→CSF-contacting nucleus→CSF/plasma circuit and modulate physiological functions via neuron-fluid interactions. According to the connection regularities of CSF-contacting nucleus from the brainstem and spinal cord, we preliminary conclude that the CSF-contacting nucleus involved in modulation pain, visceral activity, sleep and arousal, emotion, and drug addiction, etc. The current study provides morphological evidence for unveiling the significance of the CSF-contacting nucleus.