Comparative Distribution and In Vitro Activities of the Urotensin II-Related Peptides URP1 and URP2 in Zebrafish: Evidence for Their Colocalization in Spinal Cerebrospinal Fluid-Contacting Neurons

Urotensin II (UII) is an evolutionarily conserved neuropeptide initially isolated from teleost fish on the basis of its smooth muscle-contracting activity. Subsequent studies have demonstrated the occurrence of several UII-related peptides (URPs), such that the UII family is now known to include four paralogue genes called UII, URP, URP1 and URP2. These genes probably arose through the two rounds of whole genome duplication that occurred during early vertebrate evolution. URP has been identified both in tetrapods and teleosts. In contrast, URP1 and URP2 have only been observed in ray-finned and cartilaginous fishes, suggesting that both genes were lost in the tetrapod lineage. In the present study, the distribution of urp1 mRNA compared to urp2 mRNA is reported in the central nervous system of zebrafish. In the spinal cord, urp1 and urp2 mRNAs were mainly colocalized in the same cells. These cells were also shown to be GABAergic and express the gene encoding the polycystic kidney disease 2-like 1 (pkd2l1) channel, indicating that they likely correspond to cerebrospinal fluid-contacting neurons. In the hindbrain, urp1-expressing cells were found in the intermediate reticular formation and the glossopharyngeal-vagal motor nerve nuclei. We also showed that synthetic URP1 and URP2 were able to induce intracellular calcium mobilization in human UII receptor (hUT)-transfected CHO cells with similar potencies (pEC50=7.99 and 7.52, respectively) albeit at slightly lower potencies than human UII and mammalian URP (pEC50=9.44 and 8.61, respectively). The functional redundancy of URP1 and URP2 as well as the colocalization of their mRNAs in the spinal cord suggest the robustness of this peptidic system and its physiological importance in zebrafish.


Introduction
Urotensin II (UII) is a cyclic neuropeptide which was first isolated from the teleost urophysis on the basis of its spasmogenic properties [1]. Subsequent studies have shown that UII occurs in all vertebrate classes [2][3][4][5][6] where it exerts various biological effects including regulation of behaviors, neuroendocrine activities and control of cardiovascular functions [7], [8]. UIIrelated peptide (URP) is a peptide structurally related to UII which was first identified from the rodent brain [9] then in birds [10], amphibians [11] and teleosts [12]. In mammals, UII and URP have been shown to exert their action through a single receptor called UT, a member of the G protein-coupled receptor superfamily [13][14][15][16].
In teleosts, as in chondrichthyans, UII mRNA and/or peptide has been mainly found in Dahlgren cells of the caudal neurosecretory system [5,6,[17][18][19][20] but its presence has also been reported in several subdivisions of the brain [19]. In contrast, in tetrapods, the UII gene is primarily expressed in motoneurons of the brainstem and spinal cord [21,22]. For its part, URP mRNA is mainly located in motoneurons in both tetrapods [11,23,24] and teleosts (Quan et al., unpublished results). Note that in lampreys, UII was isolated from an extract of the whole brain [6] but its precise localization in the central nervous system is still unknown.
Recently, the occurrence of two additional members of the UII gene family has been reported in teleosts, called URP1 and URP2 [25][26][27][28]. It has been proposed that both genes, together with UII and URP arose through the two rounds of whole genome duplication that occurred during early vertebrate evolution [26]. In agreement with this scenario, the existence of URP1 and/or URP2 genes has recently been reported in the spotted gar (a non-teleost actinopterygian) and elephant shark (a chondrichthyan) [27,28]. In contrast, the lack of URP1 and URP2 genes in tetrapods is believed to result from their loss in this lineage specifically [26]. The primary structure of both URP1 and URP2 is exactly the same in all fish species investigated so far.
Up to now, the expression pattern of the URP1 gene has only been studied in one species, the Japanese eel (Anguilla japonica) [25] while the URP2 gene expression has solely been reported in zebrafish (Brachydanio rerio) [26]. RT-PCR revealed that in both species urp1 and urp2 genes are mainly expressed in the brainstem and spinal cord. In zebrafish, it has been shown by in situ hybridization (ISH) that the urp2 mRNA occurs in cells located along the ventral edge of the fourth ventricle and the ependymal canal. It has been suggested that these cells may correspond to cerebrospinal fluid-in cells located along the ventral edge of the fourth ventricle contacting neurons (CSF-cNs) [26].
In the present study, we report the distribution of urp1 mRNA in the central nervous system of zebrafish and compare it to that of urp2 mRNA. We demonstrate that urp1 and urp2 are mainly colocalized in the same cells in the spinal cord but not in the hindbrain. In the spinal cord, we provide evidence that cells containing both urp1 and urp2 mRNA are GABAergic and express the gene encoding the polycystic kidney disease 2-like 1 (pkd2l1) channel, indicating that they likely correspond to CSF-cNs. In the hindbrain, we show that urp1-expressing cells are located in the intermediate reticular formation and the glossopharyngeal-vagal motor nerve nuclei. Finally, we show that synthetic URP1 and URP2 are able to induce intracellular calcium mobilization in human UT-transfected Chinese hamster ovary cells (CHO) cells.

Synthesis of the riboprobes for in situ hybridization
To generate the urp1 and urp2 ISH probes, two PCR fragments of 476 and 356 base pairs, respectively, were amplified from zebrafish brain and spinal cord RACE-ready cDNA (see S1 Table for the primer sequences) then subcloned into pGEM-T easy (Promega). Digoxigeninand fluorescein-labeled probes were synthesized from the pre-linearized plasmid using SP6 or T7 RNA polymerases with the RNA Labeling Kit (Roche Diagnostics, Mannheim, Germany), according to the manufacturer's instructions. Other probes, namely islet-1 (isl1), somatostatin 1 (ss1) and gad 67 , were synthesized as previously described [33,34].

Sample preparation for in situ hybridization
Zebrafish embryos were fixed with 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer at 4°C overnight and rinsed in 0.1 M phosphate-buffered saline, 0.1% Tween-20 (PBST). Adult zebrafish were deeply anesthetized in 0.02% MS-222 (Sigma-Aldrich) and killed by decapitation. Dissected brains and spinal cords were fixed with PFA as described above for embryos. For whole-mount ISH, fixed embryos or adult tissues were stored in 100% methanol at -20°C until use. For ISH of sections, fixed tissues were cryoprotected in 15% then 30% sucrose/PBS and embedded in Tissue-Tek (Sakura, Netherlands). Frontal or para-sagittal sections of brains (10-20 μm) and spinal cords (8-16 μm) were cut at -20°C using a cryostat (CM 3050, Leica, Nanterre, France), collected on Superfrost Plus slides (O. Kindler, Freiburg, Germany), dried at room temperature for 24 h and stored at -80°C until use. Free-floating sections (40 μM) were sometimes used for spinal cord observations.

In situ hybridization procedures
Only ISH using fluorescent probes is described here, whole-mount ISH and ISH being exhaustively described elsewhere [26]. Fixed samples (whole-mount embryos and adult brain and spinal cords) were treated with 2% H 2 O 2 in 100% methanol for 20 min, rehydrated through a series of solutions with descending ethanol percentage (75, 50, 25% ethanol all in PBST, for 10 min each) and permeabilized by treatment with 10 μg/ml of proteinase K in PBST for 1-20 min, depending on the developmental stage. They were then post-fixed in 4% PFA for 20 min, washed four times in PBST for 5 min, incubated in 2mg/ml glycine for 30 min and washed again in PBST. Samples were prehybridized in hybridation buffer (50% formamide, 5X salinesodium citrate (SSC), 50 μg/ml heparin, 0.5 mg/ml yeast RNA (Sigma-Aldrich), 0.1% Tween-20) for 1 h at 65°C. Hybridization was performed overnight at 65°C in the same buffer containing the heat-denaturated digoxigenin-labeled riboprobe. Embryos were washed in a descending washing buffer (50% formamide, 5X SSC, 0.1% Tween-20) series (75, 50, 25 and 0% washing buffer all in 2X SSC for 15 min each) at 65°C. They were rinsed twice in 0.05X SSC for 30 min at the same temperature, once in 50% 0.05X SSC/50% PBST for 5 min, then twice in PBST. Following the final wash, brain and spinal cord were included in 3% agarose and sectioned at 50 μm using a vibratome (VT1000S Leica).

Image acquisition
Samples stained by BM Purple were imaged using a Leica DM 5500 B microscope connected to LAS V4.1 software. Samples stained by tyramide system amplification were imaged by the Olympus FV1000 or Zeiss LSM700 confocal microscopes using 405, 488, 543 and 633 nm laser lines. Images were processed using the ZEN (Zeiss, Marly Le Roi, France), Fiji [36] and Adobe Illustrator (Adobe Systems, Mountain View, CA, USA) softwares. conditions, 50 μl of graded concentration (10 -12 to 10 -6 M) of different peptides (4-fold final concentration) was added to the incubation medium with the built-in 8 channel pipettor at a rate of 47 μl/sec to assess their agonistic activity. After subtraction of mean fluorescence background from control wells without Fluo-4 AM, baseline was normalized to 100% and fluorescence peak values were determined for each concentration of peptide. Potency (EC 50 ) and efficacy (E max ) were calculated with the Prism 5.0 software (GraphPad Software In., La Jolla, CA, USA) using a four-parameter logistic equation to fit peak fluorescence data.

Statistical analysis
For intracellular calcium mobilization assays, results from 7-13 independent experiments were expressed as mean ± SEM of Log(EC 50 ) and plotted with box and whiskers. The normality of each data set was verified with the Shapiro-Wilk's test and a one-way ANOVA test followed by a Tukey's multiple comparison test to compare Log(EC 50 ) between each peptide. Differences were considered significant where P<0.05.

Results
Both urp1 and urp2 genes are expressed in the hindbrain and spinal cord in embryo and adult zebrafish The expression of the urp1 gene in zebrafish embryos was investigated by ISH at stages 18, 20, 22, 28, 30, 36 and 48 hours post fertilization (hpf) stages. The first urp1-expressing (urp1 + ) cells were detected at 22 hpf in the rostral half of the spinal cord (Fig. 1A). From 22 to 36 hpf, the urp1 staining expanded more caudally within the spinal cord (Fig. 1B1, 2). At any stage analyzed, urp1 + cells were exclusively located at the base of the neural tube, ventral to the central canal (Fig. 1B3,4). These cells, which were distributed in two rows along the midline, belong to the lateral floor plate (Fig. 1B4,5). At 48 hpf, the urp1 mRNA was primarily detected in a tight bilateral cluster of cells in the hindbrain, while it was hardly detectable in the spinal cord (Fig. 1C).
RT-PCR was used to determine the distribution of urp1 and urp2 mRNAs in different organs of adult zebrafish. Both mRNAs were detected in the hindbrain and the spinal cord (Fig. 2). The urp2 mRNA was also found in the middle part of the brain, as previously reported [26]. In all the others tissues tested, the expression of urp1 and urp2 genes was undetectable (Fig. 2).
In line with PCR results, ISH analysis showed the presence of urp1 + cells in the medulla oblongata (Fig. 3A, B) at the level of the intermediate reticular formation (Fig. 3A1) and the roots of the glossopharyngeal and vagal nerves (Fig. 3A2,3) [37,38]. Most caudal, a group of urp1 + cells was found at the junction between the rhombencephalon and spinal cord, at the ventral edge of the central canal (Fig. 3A4).
In the spinal cord, urp1 + cells formed a quasi-continuous column in the ventral margin along the central canal (Fig. 4A1,2) with exception to its more caudal part (data not shown). In this regard, it is noteworthy that no labeling was observed in Dahlgren cells nor in the urophysis. As shown in Fig. 4B, urp1 staining occurred in close contact to the lumen of the central canal. The distribution of the urp2 mRNA has been previously reported both in the midbrain and the spinal cord [20]. S1 Fig. highlights the occurrence of urp2 + cells at the ventral edge of the fourth ventricle.   The expression pattern of the urp1 gene in the spinal cord appeared very similar to that of the urp2 gene [26]. To assess whether the two genes are expressed in the same cells, we performed double fluorescent ISH using urp1 and urp2 antisense probes, both in embryos and adults. In 24 hpf-embryos, we observed that all stained cells contained both urp1 and urp2 mRNAs (Fig. 5A). In adults, all urp1 + cells were stained by the urp2 probe (Fig. 5B,C), while about 20% of the urp2 + cells did not contain the urp1 mRNA (Fig. 5B).
Cells containing both urp1 and urp2 mRNA in the spinal cord also express markers of the CSF-contacting neurons urp1 + cells, as urp2 + cells, were located in close contact to the central canal, indicating that they may both correspond to CSF-cNs. In zebrafish embryos, it has been shown that spinal CSF-cNs are GABAergic [39,40]. Therefore, we asked whether urp1 + and urp2 + cells may express GABAergic markers. In 24 hpf-embryos, single fluorescent ISH using either the urp1 or urp2 probe followed by IHC for GAD 65/67 showed that all urp1 + cells as well as all urp2 + cells were GAD 65/67 immunoreactive (Fig. 6A, B, arrows).
The same results were obtained in adults by triple staining experiments using both urp1 and urp2 probes combined with an anti-GAD 65/67 antibody. All urp1 + and urp2 + cells were GAD 65/67 immunoreactive (Fig. 7A, B, arrows). As mentioned above, some of the urp2 + cells did not contained the urp1 mRNA (Fig. 7A4, arrowhead).
The gene encoding for the calcium-permeable PKD2L1 channel has been recently reported as a specific marker of CSF-cNs in various vertebrate species including zebrafish [41][42][43]. To determine whether urp1 + cells express pkd2l1, we performed double fluorescent ISH using urp1 and pkd2l1 probes, both in embryos and adults. As depicted in Fig. 8A, in 24 hpf-embryos, all urp1 + cells contained pkd2l1 mRNA, but their localization was restricted to the ventral subpopulation of pkd2l1 + cells. The same results were observed in adults, since urp1 + cells represented a fraction of pkd2l1 + cells located at the ventral edge of the central canal (Fig. 8B). Likewise, all urp2 + cells were identified as pkd2l1 + both in embryo and adult (data not shown).

Hindbrain cells containing urp1 mRNA also express motoneuron markers
To better characterize the urp1 gene expression pattern in the hindbrain, we tested the colocalization of urp1 mRNA with different markers, namely isl1, ss1, ChAT and GAD. isl1 is a member of the LIM/homeobox gene family which is expressed in all postmitotic motoneurons at early stages of development [44]. Its expression pattern is particularly suitable to discriminate the different types of cranial motor nuclei in the hindbrain. In the 48 hpf-embryo, double fluorescent ISH using urp1 and isl1 probes revealed that all urp1 mRNA was exclusively present in isl1 + cells (Fig. 9A) at the level of the medial motor nucleus of the vagus [45]. Double fluorescent ISH using the urp1 and ss1 probes was also carried out, since ss1 was previously shown to be expressed in neurons of the vagal motor nucleus [33]. As depicted in Fig. 9B, all urp1    mRNA colocalized with ss1 mRNA. It is noteworthy that urp1 + cells were detected only in the most ventral part of the ss1-positive area (data not shown).
To characterize the urp1 + cells in the adult hindbrain, we performed single fluorescent ISH using the urp1 probe followed by IHC with ChAT, a marker of cholinergic neurons [38,46]. As shown in Fig. 10, urp1 + cells located in the glossopharyngeal-vagal motor nerve nuclei were weakly stained by the anti-ChAT antibody. Double fluorescent ISH using urp1 and gad 67 probes did not reveal any double-labeled cells (S2 Fig.), in contrast to what was observed in the spinal cord. Note that in the intermediate reticular formation, urp1 + cells were both ChATand gad 67 -negative (data not shown).
URP1 and URP2 are equipotent to activate the hUT As a step to elucidate the physiological actions of URP1 and URP2, we tested the in vitro activities of both peptides using transfected CHO cells expressing the human UT using a calcium mobilization assay. Synthetic URP1 and URP2 induced a robust increase in intracellular calcium in hUT-CHO cells with similar efficacies (around 200-250%) than those evoked by hUII and mURP (Fig. 11A). Of note, URP1 and URP2 were equipotent to activate the hUT (pEC 50 = 7.99 ± 0.15 and 7.52 ± 0.11 respectively). Nevertheless, they were respectively 28 and 83 times less potent than hUII (pEC 50 = 9.44 ± 0.12) to mobilize intracellular calcium. mURP, which is also the natural ligand of hUT, exhibited an intermediate potency (pEC 50 = 8.61 ± 0.16) statistically distinct from the others tested peptides (Fig. 11B).

Discussion
The aim of the present work was to compare the distributions and in vitro activities of the urotensin II-related peptides URP1 and URP2 in zebrafish. Previous studies had shown that urp2 gene expression occurs in cells contacting the central canal in the spinal cord [26], while urp1 mRNA was confined in the caudal neurosecretory system [25]. Our data do not confirm the presence of urp1 in the caudal spinal cord reported in the eel [25]. Instead urp1 + cells were found, as urp2 + cells, close to the central canal of the spinal cord. However, in contrast to urp2 mRNA which exhibits alternate zones of strong and weak staining [26], urp2 mRNA was present approximately at the same intensity along the entire length of the spinal cord. Worth mentioning, all urp1 + cells were urp2 + , while the opposite was not necessarily true, at least in adults.
On the basis of a former study reporting the occurrence of UII-like immunoreactive material in CSF-cNs [47], we suggested that urp2 + cells might be CSF-cNs [26]. CSF-cNs are neurons, described in all vertebrate groups [48][49][50], located on the edge of the neural tube lumen, socalled because they are in direct contact with the CSF via their dendritic pole. In Xenopus as in zebrafish, CSF-cNs have been shown to project an ascending axon ventrally in the spinal cord [39,[51][52][53]. Moreover, analysis of the UII-like immunoreactive-containing system from various fish species showed that CSF-cNs send fibers to several brain regions, including medulla oblongata, thalamus, hypothalamus and telencephalon, but also in the spinal cord ventrolateral surface [47,54].
Although their functions remain largely unknown, CSF-cNs are classically considered as sensory neurons, but the exact nature of their stimuli is not clearly established. Recent studies have shown that they are GABAergic (see [42] for review). They also express PKD2L1, a transient receptor potential channel that could play a role as a sensory receptor [42,43], in good agreement with studies describing these neurons as chemoreceptors or mechanoreceptors sensing the CSF chemical composition and/or movements [42,50]. In addition, using the powerful optogenetic approach, Wyart et al. [53] have shown that CSF-cNs might be involved in motor control as part of the neuronal network that controls spontaneous swimming movement of zebrafish larvae. Thus, our data demonstrate that urp1 + and urp2 + cells both express GABA and pkd2l1 indicating that they actually are CSF-cNs and that URP1 and URP2 might have a role in the neuronal network that controls spontaneous swimming in zebrafish, at least in larvae.
It is likely that the presence of URP1 and/or URP2 in CSF-cNs is an ancestral feature of gnathostomes since the occurrence of CSF-cNs containing UII-like immunoreactivity has also been reported in a chondrichthyan species, Hydolagus collieri [54]. Beside URP1 and URP2, several other neuropeptides have been detected in CSF-cNs, such as vasoactive intestinal peptide (VIP) and SS [53,[55][56][57]. In this respect, it is noteworthy that in coho salmon Oncorhynchus kisutch, SS-and UII-immunoreactive material do not localize in the same CSF-cNs [58] suggesting the occurrence of distinct CSF-cN types. Indeed, in zebrafish embryos, in which GABAergic CSF-cNs were named Kolmer-Agduhr (KA) cells to distinguish them from ciliated ependymal cells [51], KA cells have been divided into two subpopulations, namely KA' and KA" cells, on the basis of their developmental origin and location. Whereas the dorsal KA' cells are derived from olig2 + motoneuron precursors, the more ventral KA" cells develop from the lateral floor plate [34,42,59]. Our results showing that urp1 and urp2 gene expression was restricted to the KA" ventral subpopulation, reinforce the idea of the CSF-cNs diversity and reveal urp1 and urp2 as new markers for KA" cells in zebrafish. In adults, the fact that urp1 + and urp2 + cells also occur in a ventral position indicates that KA" cells conserve their location relative to the central canal during development. Whether all these different neuropeptides could help define additional types of CSF-cNs remains to be determined.
At the brain level, urp1 + cells were restricted to the rhombencephalon. In embryos, they were observed from 48 hpf, i.e. more than 24 h later than those detected in the spinal cord, in a small bilateral area also expressing isl1 and ss1 genes. In agreement with previous studies [33][34][35][36][37][38][39][40][41][42][43][44], these cells likely correspond to motoneurons in the vagal motor nucleus. Supporting this idea, most of the urp1 + cells in adults were identified as cholinergic neurons located in the same nucleus [38]. These data indicate that urp1 gene expression in the vagal motor nucleus persists throughout the entire development period until adulthood. It is noteworthy that UII and URP are known to be expressed mainly in motoneurons in tetrapods [11,[21][22][23][24]60]. While urp1 and urp2 gene expression largely overlap in the spinal cord, their respective patterns differ completely in the brain since urp2 + cells appear to form an extension of the spinal urp2 CSF-cNs system below the fourth ventricle [26].
In trout (O. mykiss) and eel (A. japonica), central injection of UII evokes an increase in arterial blood pressure and heart rate [25,[62][63][64][65]. Similar effects have been reported with URP1 [25]. It has been suggested that UII and URP1 may act directly at the central level, even though their site of action are still uncertain [25]. The present data show that the glossopharyngealvagal motor nerve nuclei are a putative source of URP1 indicating that URP1, rather than UII absent here [17,19,20,25], may act as a central regulator of the cardiovascular activity. Likewise, central pharmacological effects of UII on motor activity [61] could be achieved by URP1 expressed in the intermediate reticular formation, one type of zebrafish hindbrain nuclei projecting to the spinal cord [65].
To date, the precise mechanism of the URP1 action is unkown. Calcium mobilization assay showed here that synthetic URP1 and URP2 are active peptides toward human urotensin II receptor. This is consistent with the presence of the cyclic hexapeptide (CFWKYC) which is the minimal core involved for the activity of these peptides [31] and which is also present in UII and URP [27,28]. Previous docking [66,67] and photolabelling studies [68] showed that the side chains of Phe 6 and Lys 9 of hUII interact respectively with the Met 185 residue located in the fourth transmembrane domain (TMD) and the Asn 130 residue in the third TMD of the hUII receptor. The lower potencies of URP1 and URP2 toward hUT compared to hUII and mURP, the natural cognate peptides of this receptor, might be attributed to their longer C-terminus end. Indeed, steric hindrance and/or chemical nature of these three residues (VTN and SQN for URP1 and URP2 respectively) could interfere with the optimal positioning of the peptides within their putative hydrophobic binding pocket or their interaction with the second and third extracellular loop of the human receptor as shown for UII and URP [69].
UT has long been considered to be the only high affinity receptor for the UII family peptides, at least in mammals [7]. However, in a recent study, Tostivint et al. [28] provided evidence that the vertebrate ancestor likely possessed five distinct UT subtype genes, called UTS2R1-5 and that most of them have been preserved in teleosts. In zebrafish for example, four UT-like sequences have been identified that correspond to Uts2r1, the UT homologue, Uts2r2, uts2r3 and Uts2r4, while in stickleback, Uts2r2 seems to have been lost but is replaced by Uts2r5 [28]. In contrast, only the UTS2R1 subtype is still present in mammals. Considering the large number of UT receptor subtypes in teleosts, it is highly probable that URP1 and URP2 are able to bind to more than one receptor subtype. In support with this view, it is noteworthy that the Met 185 and Asn 130 residues mentioned above are conserved in the four putative UII receptors identified in zebrafish (data not shown). It is now evident that further studies will be needed to determine which receptor subtype the different UII-related peptides can preferentially bind to. In this regard, it will be interesting to determine whether the Uts2r1 subtype, the hUT counterpart studied in the present work, is the most efficient target of URP1 and URP2 or not.
In conclusion, we show here that, in the spinal cord, urp1 and urp2 are colocalized mainly in CSF-cNs, while, in the hindbrain, urp1 but not urp2 is confined to motoneurons in the glossopharyngeal-vagal motor nerve nuclei. We also demonstrate that URP1 and URP2 are active peptides toward human urotensin II receptor with similar potencies. Taken together, the functional redundancy of URP1 and URP2 as well as the colocalization of their mRNAs in the spinal cord suggest the robustness of this peptidic system and its physiological importance in zebrafish. These results provide the basis for further studies to improve our understanding of the physiological functions of URP1 and URP2 by using zebrafish as an experimental model.