Relaxins (Rln) and relaxin-like factors are peptides belonging to the relaxin/insulin superfamily. All members are structurally similar and are synthesized as a preprohormone consisting of a signal peptide, B-domain, C-peptide, and A-domain (Sherwood,1994; Ivell and Einspainer,2002). The mature Rln hormone is produced by cleavage of the signal peptide and C-domain, and it consists of A and B chains linked by disulfide bonds (Marriott et al.,1992). To date, three rln genes have been identified in the human genome: the rln1 gene has only been found in human and great apes and its expression is limited to the decidua, placenta, and prostate (Hansell et al.,1991); the rln2 gene is mainly expressed in the ovary and encodes the major circulating form of the hormone, commonly designated as “relaxin” (Winslow et al.,1992; Sherwood,1994); the rln3 gene is the latest identified member of the relaxin family and is mainly expressed in the brain and testis (Bathgate et al.,2002; Burazin et al.,2002). rln1 and rln2 are thought to be duplicated genes in primate lineage since only one ortholog gene, designated as rln1, has been found in other species (Bathgate et al.,2006).
Phylogenetic analyses of the relaxin peptide family indicate a Rln3-like peptide as the ancestral relaxin, whose gene is under a very strong purifying selection, highlighting the importance of its highly conserved function from fish to mammals (Hsu,2003; Wilkinson et al.,2005; Wilkinson and Bathgate,2007).
Rln2 has a well-recognised role during pregnancy, since it acts on the female reproductive tract to prepare the birth canal for parturition (Bathgate et al.,2006; Sherwood,2004). Although originally characterized as a reproductive hormone, Rln2 has emerged as an endocrine and paracrine multifunctional factor that plays important roles on both reproductive and non-reproductive tissues. For instance, it was shown that Rln2 is involved in the angiogenesis (Unemori et al.,1999), in the remodeling of the extracellular matrix (McGuane and Parry,2005), in the regulation of nitric oxide biosynthesis (Bani et al.,2001), in the invasive potential of endometrial cancer (Kamat et al.,2006), and in other physiological functions (Ivell and Einspainer,2002; Samuel et al.,2007).
As for the rln3 gene, its expression is prevalent in the brain and testis but not in female reproductive tissues, suggesting a different physiological function of the Rln3 peptide from other family members (Bathgate et al.,2002; Liu et al.,2003). In fact, ultrastructural observations of Rln3-positive neurons showed dense-cored vesicles transported to nerve endings near the synaptic terminals, suggesting that Rln3 is released into the synaptic cleft and acts as a neurotransmitter (Tanaka et al.,2005).
Interestingly, the spatial distribution of the rln3 transcript in the adult mouse and rat brain shows that the main site of gene expression is a restricted cell group of the brainstem region, known as the nucleus incertus (NI) (Bathgate et al.,2002; Burazin et al.,2002; Tanaka et al.,2005); this neuron cluster is positioned in a behaviour control network that integrates information related to memory, attentional state, and stress response (Goto et al.,2001; Olucha-Bordonau et al.,2003). In support of the central role of NI in the brain, it has been demonstrated that NI neurons of stressed rats show an increased Fos expression, suggesting a direct role in stress response mechanisms (Tanaka et al.,2005). The function of NI in stress responses is also corroborated by a high expression level of the type1 receptor for the corticotropin-releasing factor (CRF) in NI cells, which has a key role in the body response to stressors (Potter et al.,1994; Bittencourt and Sawchenko,2000; Goto et al.,2001,2005; Tanaka et al.,2005).
A great amount of experimental evidence showed that the rln3 gene is directly involved in the physiological function of NI. In fact, rln3 is up-regulated in NI neurons of stressed rats, suggesting an involvement in stress response regulation (Tanaka et al.,2005). Furthermore, a role for the Rln3 peptide in appetite regulation was hypothesized, given that intracerebroventricular injections of human Rln3 stimulate food intake in satiated rats (McGowan et al.,2005). These data, together with the neuroanatomical analyses of Rln3 peptide distribution in the adult rat brain (Ma et al.,2007), suggest a wide modulatory effect on various behavioural mechanisms, and indicate that Rln3 could be involved in different neural processes such as metabolism, stress response, and cognition in the mammalian adult brain.
Only recently, Miyamoto et al. (2008), for the first time, described the expression pattern of the rln3 gene during brain development, specifically in rat embryos. The transcript was localized in two bilateral cell groups of the pons region, presumably corresponding to the NI of an adult brain. In addition, the authors demonstrated that the serotonergic system may influence rln3 gene expression. In fact, selective depletion of cerebral serotonin, produced in raphe neurons, leads to a significant increase of rln3 gene expression (Miyamoto et al.,2008). These data suggest that rln3-expressing neurons in NI might be functional early in vertebrate embryogenesis.
To date, the role of rln3 has only been investigated in mammalian species. Here we report the study of rln3 expression in the brain of a non-mammalian vertebrate, the zebrafish Danio rerio, which has emerged as a vertebrate model for genetic, molecular, and behavioural studies. We show that, in postembryonic zebrafish brain, rln3 expression is spatially restricted in two neuron clusters, unlike the mammalian brain in which only one cell cluster (NI) abundantly expresses this gene.
RESULTS AND DISCUSSION
During zebrafish embryogenesis, at 48 hpf the hatching period starts and secondary neurogenesis takes place. Over this time, the larval brain significantly grows and differentiates to form its finer subdivisions. The embryo at 48 hpf represents the first stage of zebrafish brain development, which can be compared to related stages of neurogenesis in other vertebrates such as the mouse and rat (Mueller and Wullimann,2005), in which rln3 gene expression territories have already been characterized. We performed whole mount in situ hybridization experiments on embryos from 48 to 96 hpf, in order to compare the rln3 expression pattern of the zebrafish brain to the mammalian brain.
At the late pharyngula period (48 hpf), two bilateral rln3-expressing cell groups were found in the developing zebrafish brain in the midbrain tegmentum region (Fig. 1A,B). The expression in this restricted cell clusters starts at 40 hpf (Fig. 1K); before this stage, a faint and diffused in situ hybridization signal was revealed in the developing brain (Fig. 1J). Later, at 72 and 96 hpf, the midbrain rln3-expressing cell clusters were still evident and showed a larger cell number (Fig. 1C, D, H, I). We performed transverse sections of hybridized embryos at 72 hpf to better characterize the spatial distribution of these rln3-expressing cells. As shown in Figure 1E, they are distributed latero-dorsally to the central midbrain tegmentum in a position compatible with the periaqueductal gray matter (PAG).
Interestingly, starting from 72 hpf, the rln3 gene expression showed a new more posterior signal detected in a smaller cell group in the tegmentum/medulla region (Fig. 1C,D,H,I). In the transverse sections of hybridized embryos, the rln3-expressing cell clusters appeared grouped in the middle of central gray (Fig. 1F) and were distributed as two bilateral columns near the fourth ventricle, as shown better in the magnification (Fig. 1G).
In order to have more evidence that the anterior rln3-expressing cell cluster is in the PAG region, we analysed the spatial gene expression of the enkephalins, which are endogenous opiods considered markers for that territory in the rat brain (Smith et al.,1994). We performed whole mount in situ hybridization experiments for both paralog genes present in the zebrafish genome, proenkephalin (penk) and proenkephalin like (penkl), on embryos at 96 hpf, when the signal for the rln3 gene in the PAG is very evident. The antisense riboprobe for penk showed a restricted signal in the rombencephalic region of the zebrafish brain, whereas only a faint signal was revealed in the midbrain region (data not shown). On the other hand, the penkl gene showed signals in different brain regions; in particular, a cell group, presumably representing the PAG, was evident in the midbrain region, as shown by a dorsal view of the zebrafish brain (Fig. 2A). Therefore, we performed double in situ hybridization experiments with rln3 and penkl riboprobes. As shown in the magnification of the dorsal view of the midbrain region (Fig. 2B), the colocalization of rln3 and penkl transcripts clearly demonstrated the rln3 gene expression in the PAG.
As for the posteriormost rln3-expressing cell clusters, we better defined their spatial localization by double in situ hybridization experiments using markers of specific territories in the brainstem region. In particular, we used a tryptophan hydroxylase enzyme gene (tphR) to mark the dorsal raphe (DR) in the midbrain region, the corticotrophin-releasing hormone (crh), and the corticotrophin-releasing hormone binding protein (crh-bp) genes to mark the locus coeruleus (LC) and the superior raphe (SR), respectively, in the pons region. The double in situ hybridization showed the posteriormost rln3-expressing cell clusters positioned caudally to the dorsal raphe nucleus (Fig. 2C,D) and regionalized in the rostral portion of the pons region, as evidenced by the spatial relationship with the locus coeruleus (Fig. 2E,F) and superior raphe (Fig. 2G,H). Since both the localization of the cell cluster and the spatial distribution of the cells along the fourth ventricle closely match the one described for rln3 expression in the developing rat brain, we hypothesized that the cell cluster, identified in the pons region of the zebrafish brain, might correspond to mammalian NI. Given that the rat NI is characterised by the highest expression levels of gene encoding the corticotrophin-releasing hormone receptor type 1 (crhr1) (Potter et al.,1994), we analysed the transcript localization of crhr1 in the zebrafish brain. The whole mount in situ hybridization at 96 hpf showed a faint signal in several brain regions where, in the hindbrain, a bilateral cell group, more probably representing NI-neurons, was evident (blue arrowhead in Fig. 2I). Although we couldn't obtain a good signal in double in situ experiments, probably due to low transcript levels of crhr1, the comparison between rln3 and crhr1 signals in the hindbrain clearly shows the correspondence of the two cell clusters (Fig. 2I). We further analysed the distribution of crhr1-expressing cells in the zebrafish hindbrain by transverse sections of the hybridized embryos. As shown in Figure 2J, the crhr1-expressing cell cluster is positioned in the middle of central gray as the posteriormost rln3-expressing cell cluster. The comparison between the magnification of the transverse section of hybridized embryos with crhr1 (Fig. 2K) and rln3 (Fig. 2L) riboprobes demonstrate even more the correspondence of the two cell clusters, corroborating our hypothesis of rln3 expression in NI of the zebrafish brain.
Our data of such a restricted expression of rln3 in the developing zebrafish brain are in accordance with phylogenetic analyses that hypothesised a conserved role during vertebrate evolution for Rln3 as a neuropeptide. In addition, in situ hybridization experiments indicated remarkable differences among zebrafish and the rat rln3 expression profile. In the developing and adult rat brain, rln3 transcripts have been reported to be predominant in the NI, whereas only scattered rln3-expressing cells are present in other brain regions such as the PAG (Tanaka et al.,2005; Ma et al.,2007; Miyamoto et al.,2008). Notably, in zebrafish larva, rln3 gene expression appears first in the PAG region, where a substantial cell group shows the signal, and later in neurons of the NI (Fig. 1). Such zebrafish rln3 gene expression in the PAG neurons is very appealing since in teleost fish, as in other vertebrates, these neurons have been correlated to the production of vocal communication, which is an important feature for social behaviour (Kittelberger et al.,2006). Finally, the identification of zebrafish NI adds an important element to neuroanatomical mapping of the behaviour control network in the brainstem region of fish organisms.
In conclusion, considering the recent hypothesis of a broad modulatory activity of Rln3 in the control of metabolism, reproduction, and cognition, our findings provide the basis to analyse the involvement of this peptide and of PAG and NI in such mechanisms in the early life stage of a vertebrate model organism.
Zebrafish embryos, obtained from wild-type fish of AB strain maintained at 28°C, were cultivated and staged as described (Westerfield,1995; Kimmel et al.,1995).
Cloning of Zebrafish cDNAs
First strand cDNA was synthesized with reverse transcriptase Superscript III (Invitrogen, Milan, Italy) from total RNA of 72-hpf embryos. Using the product of reverse-transcription reaction as a template, the cDNA for rln3 (AM161137) was amplified by PCR using the following primers: (forward) 5′-AAAGCACAGGTAGACCA- TCAGG-3′ and (reverse) 5′-TGCAGCCCCATTTGCAGCAGG-3′. The cDNA for crh and crh-bp was amplified using primers described by Alderman and Bernier (2007). The cDNA for tphR (AB125219), crhr1 (XM_691254), penk (NM_182883), and penkl (NM_200083) was amplified using the following primers: tphR, (forward) 5′-GGAGAGTGTCAGGTACTGT-3′ and (reverse) 5′-CAGAAAGCCAACTTCATTCT-3′; crhr1, (forward) 5′- CAGCTCACCATGAATCCAGA -3′ and (reverse) 5′- AGAGGCAGCAGAACCAGTGT -3′; penk, (forward) 5′-CACCGGACAGACTGCAGTAAC-3′ and (reverse) 5′-CCCTGATCAACATCCTCTGG-3′; penkl, (forward) 5′-CAGTGGACATTGCCAGAGAA-3′ and (reverse) 5′-GCCTTCTTCATAAAGCCTCCA-3′. The PCR products were cloned into pGEM®-T Easy Vector.
Whole Mount In Situ Hybridization
Whole mount in situ hybridizations were performed as reported in Thisse et al. (2004). For rln3 expression, digoxigenin (DIG)-labeled riboprobes were used, whereas the probes for the other transcripts in double in situ hybridization were marked with fluoresceine. Anti-DIG antibody and NBT/BCIP (Roche) were used to detect the rln3 probe, while anti-FLUO antibody and INT/BCIP (Roche) were used to detect the second transcript. Ultrathin sections were obtained as described in Donizetti et al. (2008).
We thank Gennaro Iamunno and Franco Iamunno for preparation of the ultrathin sections.