Differential Level of RXFP3 Expression in Dopaminergic Neurons Within the Arcuate Nucleus, Dorsomedial Hypothalamus and Ventral Tegmental Area of RXFP3-Cre/tdTomato Mice

RXFP3 (relaxin-family peptide 3 receptor) is the cognate G-protein-coupled receptor for the neuropeptide, relaxin-3. RXFP3 is expressed widely throughout the brain, including the hypothalamus, where it has been shown to modulate feeding behavior and neuroendocrine activity in rodents. In order to better characterize its potential mechanisms of action, this study determined whether RXFP3 is expressed by dopaminergic neurons within the arcuate nucleus (ARC) and dorsomedial hypothalamus (DMH), in addition to the ventral tegmental area (VTA). Neurons that express RXFP3 were visualized in coronal brain sections from RXFP3-Cre/tdTomato mice, which express the tdTomato fluorophore within RXFP3-positive cells, and dopaminergic neurons in these areas were visualized by simultaneous immunohistochemical detection of tyrosine hydroxylase-immunoreactivity (TH-IR). Approximately 20% of ARC neurons containing TH-IR coexpressed tdTomato fluorescence, suggesting that RXFP3 can influence the dopamine pathway from the ARC to the pituitary gland that controls prolactin release. The ability of prolactin to reduce leptin sensitivity and increase food consumption therefore represents a potential mechanism by which RXFP3 activation influences feeding. A similar proportion of DMH neurons containing TH-IR expressed RXFP3-related tdTomato fluorescence, consistent with a possible RXFP3-mediated regulation of stress and neuroendocrine circuits. In contrast, RXFP3 was barely detected within the VTA. TdTomato signal was absent from the ARC and DMH in sections from Rosa26-tdTomato mice, suggesting that the cells identified in RXFP3-Cre/tdTomato mice expressed authentic RXFP3-related tdTomato fluorescence. Together, these findings identify potential hypothalamic mechanisms through which RXFP3 influences neuroendocrine control of metabolism, and further highlight the therapeutic potential of targeting RXFP3 in feeding-related disorders.

RXFP3 (relaxin-family peptide 3 receptor) is the cognate G-protein-coupled receptor for the neuropeptide, relaxin-3. RXFP3 is expressed widely throughout the brain, including the hypothalamus, where it has been shown to modulate feeding behavior and neuroendocrine activity in rodents. In order to better characterize its potential mechanisms of action, this study determined whether RXFP3 is expressed by dopaminergic neurons within the arcuate nucleus (ARC) and dorsomedial hypothalamus (DMH), in addition to the ventral tegmental area (VTA). Neurons that express RXFP3 were visualized in coronal brain sections from RXFP3-Cre/tdTomato mice, which express the tdTomato fluorophore within RXFP3-positive cells, and dopaminergic neurons in these areas were visualized by simultaneous immunohistochemical detection of tyrosine hydroxylase-immunoreactivity (TH-IR). Approximately 20% of ARC neurons containing TH-IR coexpressed tdTomato fluorescence, suggesting that RXFP3 can influence the dopamine pathway from the ARC to the pituitary gland that controls prolactin release. The ability of prolactin to reduce leptin sensitivity and increase food consumption therefore represents a potential mechanism by which RXFP3 activation influences feeding. A similar proportion of DMH neurons containing TH-IR expressed RXFP3related tdTomato fluorescence, consistent with a possible RXFP3-mediated regulation of stress and neuroendocrine circuits. In contrast, RXFP3 was barely detected within the VTA. TdTomato signal was absent from the ARC and DMH in sections from Rosa26-tdTomato mice, suggesting that the cells identified in RXFP3-Cre/tdTomato mice expressed authentic RXFP3-related tdTomato fluorescence. Together, these findings identify potential hypothalamic mechanisms through which RXFP3 influences neuroendocrine control of metabolism, and further highlight the therapeutic potential of targeting RXFP3 in feeding-related disorders.
Indeed, the ARC is a major mediator of orexigenic and anorexigenic signaling within the brain (e.g., Abizaid et al., 2006;Sohn, 2015;Andermann and Lowell, 2017). The ARC is highly neurochemically heterogeneous and comprises multiple neuronal populations, which are characterized for research purposes as subpopulations that express characteristic markers such as cocaine-and amphetamine regulated transcript (CART), neuropeptide Y, agouti-related peptide (AgRP), and proopiomelanocortin (POMC) (Sohn, 2015;Andermann and Lowell, 2017;Campbell et al., 2017). In order to characterize how RXFP3 signaling in the hypothalamus can influence feeding, metabolism and related neuroendocrine events, it is crucial to identify which neuronal populations express RXFP3 within key areas such as the ARC. The ARC contains dopaminergic neurons (Brown et al., 2010) that are an essential component of the prolactin negative-feedback loop, and are involved in appetite regulation (Zhang and Van Den Pol, 2016). In addition to prolactin's classical roles in lactation, it also decreases leptin sensitivity within the hypothalamus and thus increases food consumption in response to, and in preparation for, greater metabolic demands during the perinatal period (Grattan, 2015). In male rodents, prolactin has been shown to inhibit leptin release from adipose tissue (Brandebourg et al., 2007), suggesting that prolactin's effect on leptin is not sex-specific. Dopaminergic neurons are also present within the neighboring DMH, but their specific physiological roles are not well characterized (Pirník et al., 2018). Extrahypothalamic expression of RXFP3 in feeding-related brain areas has been documented, including in the ventral tegmental area (VTA) (Smith et al., 2010), which contains a major population of dopaminergic neurons that form the mesolimbic dopaminergic pathway. This pathway drives motivation and goal-directed behavior for both natural (e.g., food) and other (e.g., drug) rewards (Volkow et al., 2017).
The aim of this study was to determine whether RXFP3 is expressed by dopaminergic neurons within the ARC, DMH and VTA, to identify or exclude potential direct RXFP3-mediated effects on neurons in these areas. As a fully validated RXFP3 antibody is not yet available, transgenic RXFP3-Cre/tdTomato mice were used to visualize RXFP3-expressing neurons (Ch'ng et al., 2019). Histochemical detection of tdTomato fluorescence was combined with immunohistochemical detection of tyrosine hydroxylase (TH), an enzyme involved in the synthesis of catecholamines (Björklund and Dunnett, 2007), including dopamine (Daubner et al., 2011). TH is a highly reliable and commonly used marker for dopaminergic neurons, and the vast majority of neurons containing TH immunoreactivity (TH-IR) within the ARC, DMH, and VTA have been confirmed as dopaminergic (Zhang et al., 2010;Zhang and Van Den Pol, 2015).

RXFP3-Cre/tdTomato and Rosa26-tdTomato Mice
All experiments were performed in accordance with the Prevention of Cruelty to Animals Act 1986 under guidelines of the National Health and Medical Research Council Code of Practice for the Care and Use of Animals for Experimental Purposes in Australia. These experiments were approved by the Florey Institute of Neuroscience and Mental Health Animal Ethics Committee (#14-035-FINMH).

Immunohistochemistry
Brain sections from RXFP3-Cre/tdTomato mice were removed from cryoprotectant and placed into mesh inserts (to reduce handling damage) in a 6-well tissue culture plate containing phosphate buffer (PB; 2.7 mM KCl, 11.2 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , pH 7.4). Sections were washed in PB containing 0.05% Triton-X for 4 × 5 min on an orbital mixer, followed by 4 × 5 min PB. Sections were then incubated in blocking solution at room temperature [10% normal donkey serum (NDS; Cat#S30-100ML, Merck, Castle Hill, NSW, Australia) in PB] for a minimum of 1 h, and left overnight in PB with 2% NDS and rabbit anti-TH primary antibody at room temperature (1:1,000 dilution; Cat#AB152, Millipore, Bayswater, VIC, Australia). The next day, sections were washed in PB for 6 × 5 min, and incubated for a minimum of 1 h in PB with 2% NDS and donkey anti-rabbit Alexa488 secondary antibody (1:500 dilution; Cat#ab150061, Abcam, Melbourne, VIC, Australia). Sections were washed in PB for 5 × 5 min, mounted on glass microscope slides and coverslipped using DAKO fluorescent mounting medium (Cat#S302380-2, Agilent, Mulgrave, VIC, Australia) before storage at 4 • C.

Microscopy, Data Collection, and Statistical Analysis
A single, representative image from each of three (3) RXFP3-Cre/tdTomato mice through each area of interest was obtained using a 40 × objective on a Leica Confocal SP5 microscope (Leica, Macquire Park Germany). Using the Allen Institute Mouse Brain Reference Atlas (Lein et al., 2007) to identify surrounding visual markers, images of coronal brain sections were obtained specifically focusing on the ARC, DMH and VTA (-2.06, -2.06, and -3.6 mm A/P from bregma, respectively). Manual cell quantification was undertaken using ImageJ software, whereby individual neurons in separate channels were counted/tagged, before images were merged to determine co-localisation counts. Graphs were generated and data were analyzed using Graphpad Prism software (V8; Graphpad, San Diego, CA, United States), and statistical comparisons between groups were performed via parametric paired t-tests through p value (p), t ratio (t), and correlation coefficient (r). A single representative image through the ARC and DMH from a Rosa26-tdTomato mouse was collected as above, which revealed an absence of strong tdTomato-expressing cells within these regions.

RESULTS
In situ hybridisation data from the Allen Brain Atlas (Lein et al., 2007) demonstrates that TH mRNA is highly expressed within the ARC (Figures 1A,B), and a similar distribution of neurons containing TH-IR was observed in RXFP3-Cre/tdTomato mice ( Figure 1C). TdTomato fluorescence was equally pronounced within the ARC (but not adjacent regions), and neurons that co-expressed both markers were easily identified (Figures 1D-F). TH-IR-and tdTomato-positive cells were present in roughly equal numbers (between 35 and 40 cells/section; p = 0.271, t = 1.5, r = -0.19; Figure 1G). Approximately 20% of these two neuronal populations expressed both markers (p = 0.146, t = 2.3, r = 0.77; Figure 1H), and when the total number of both tdTomato and TH-IR cells were summed, approximately 10% exhibited double-labeling. A high density of tdTomato-positive cells (and neuronal fibers) was also observed surrounding the median eminence in the pars tuberalis of the pituitary gland (PT; Figure 1C; data not quantified).
TH mRNA is highly expressed within the DMH (Figures 2A,B; Lein et al., 2007), and neurons containing TH-IR were clearly visible within this region, but not surrounding regions ( Figure 2C). Similar to the ARC, roughly equal numbers of cells containing TH-IR and tdTomato-fluorescence were present in the DMH (between 19 and 23 cells/section; p = 0.122, t = 2.6, r = 0.91; Figures 2D-G). Approximately 20% of both TH-IR-and tdTomato-positive neurons were double-labeled (p = 0.183, t = 2.0, r = 0.05; Figure 2H), which equated to less than 10% of the total number of cells positive for either signal displaying coexpression.
Similarly, in situ hybridisation data from the Allen Brain Atlas reveals high levels of TH mRNA within the VTA (Figures 3A,B; Lein et al., 2007). Although the two-dimensional area occupied by the VTA in coronal sections changes substantially in the anterior-posterior direction, its borders were clearly delineated in our sections by the distribution of neurons containing TH-IR. This aided the identification of VTA-containing sections at a consistent coronal level in different RXFP3-Cre/tdTomato mice, at which TH-IR neurons were similarly enriched ( Figure 3C). Very low numbers of tdTomato-positive cells were identified in this region (Figures 3D-F), in contrast to the large number of TH-IR neurons observed (p = 0.004, t = 8.1, r = 0.74; Figure 3G). Sparse double-labeled cells were identified in some of the sections analyzed, but this represented essentially 0% cells containing TH-IR co-labeled, and a small percentage of tdTomato-positive cells expressed TH-IR (p = 0.391, t = 1.0, r = 1.0; Figure 3H). There was also essentially 0% of the total number of cells positive for either signal that displayed coexpression.
In order to confirm that tdTomato expression within the ARC and DMH of double mutant RXFP3-Cre/tdTomato mice  is Cre-dependent, a single mutant Rosa26-tdTomato mouse was analyzed. Bright tdTomato-expressing cells were absent within the ARC ( Figure 4A) and DMH (Figure 4B) of these mice. Interestingly however, a high density of bright tdTomato-fluorescence was present within the PT (Figure 4A), which demonstrates that tdTomato-expression within this area does not indicate the presence of Cre or RXFP3 in RXFP3-Cre/tdTomato mice.

DISCUSSION
A key finding of this study was that a small population of dopaminergic neurons within the ARC express RXFP3. It is therefore feasible, that RXFP3 activation inhibits this small proportion of ARC dopaminergic neurons, as electrophysiological studies suggest that postsynaptic RXFP3 signaling is predominantly inhibitory in rat and mouse (Kania et al., 2017;Ch'ng et al., 2019;Kania et al., 2020). These data are in line with cell-based experiments demonstrating that RXFP3 activation reduces intracellular cAMP (see Kocan et al., 2014 for review). Although the functional significance of this modest expression of RXFP3 by ARC dopaminergic neurons is currently unknown, it is possible that modulation of this population could contribute to the effects of centrally administered RXFP3 agonists on feeding and neuroendocrine function.
Prolactin binds to prolactin receptors on ARC dopaminergic neurons (Lerant and Freeman, 1998), which normally stimulates the release of dopamine into the portal blood system causing negative feedback and reduced prolactin release. Potential RXFP3-mediated inhibition of ARC dopaminergic neurons may therefore cause more subtle/reduced dopamine release into the portal blood system, resulting in greater activation of lactotroph cells in the anterior pituitary (Mansour et al., 1990) and increased prolactin secretion. Prolactin increases feeding by decreasing leptin sensitivity, and this process involves increased expression of hypothalamic suppressor of cytokine signaling (SOCS) proteins (Buonfiglio et al., 2016). In a powerful demonstration of the cross-talk between leptin and prolactin, chronic ICV injections of prolactin countered the anorexigenic effects of acute leptin administration (Grattan, 2015;Buonfiglio et al., 2016). Interestingly, increased plasma leptin has been observed following both acute (McGowan et al., 2014) and chronic (Hida et al., 2006;McGowan et al., 2006) injection of relaxin-3 into rat brain. Further investigating the possible influence of RXFP3 on leptin pathways is therefore of interest, especially as the actions of RXFP3 likely involve changes in gene expression and other medium to long-term effects, as well as rapid neuronal inhibition (or activation) (McGowan et al., 2006;Kocan et al., 2014).
Notably however, a majority of the ARC tdTomato (RXFP3)-positive cells were negative for TH-IR. In light of the neurochemical heterogeneity of this region, with several populations characterized by their expression of CART, neuropeptide Y and other peptidergic markers (Campbell et al., 2017), TH-negative/RXFP3-positive neurons in the ARC may constitute one or more of these other populations. ARC CART neurons express leptin receptors (Kristensen et al., 1998), and trigger anorexigenic pathways and reduced food consumption (Ahmadian-Moghadam et al., 2018). It is therefore possible that RXFP3 may be expressed by ARC CART neurons, and that RXFP3-mediated inhibition of these neurons may contribute to the orexigenic actions of RXFP3. Also of note, the number and density of tdTomato-positive cells within the ARC was greater than reported in our previous study (Smith et al., 2010), which detected Rxfp3 mRNA using radioactive in situ hybridisation histochemistry. This discrepancy may be due to an increased sensitivity of target detection provided by the transgenic RXFP3-Cre/tdTomato mice, whereby cells that express Rxfp3 mRNA at low levels, still express sufficient Cre-recombinase to drive robust expression of tdTomato, which is driven by a strong promoter at the Rosa26 locus. The different strains of mice used in these studies may also contribute to this differential observation.
Another key finding of this study was that within the DMH, a relatively low level of coexpression of TH-IR and tdTomato fluorescence was observed. The functional relevance of this co-expression is currently unknown, especially since the role of DMH dopaminergic neurons remains largely unexplored. However, the DMH is considered to be a relay station for multiple feeding, neuroendocrine and stress pathways, as it receives multiple inputs from the majority of hypothalamic regions including the ARC (Pirník et al., 2018), and has complex outputs to autonomic and other pathways (Brasil et al., 2020). Notably, it contains a range of anorexigenic neuron populations that may express RXFP3, including cholinergic and proopiomelanocortin (POMC) neurons. Furthermore, prolactinreleasing peptide is also produced by a subset of DMH neurons (Jeong et al., 2017;Pirník et al., 2018). Lesions of the DMH result in decreased food intake, but the precise contributions of different neuronal subpopulations in this effect requires further investigation (Jeong et al., 2017).
In this study, tdTomato was not consistently observed in VTA dopaminergic neurons, suggesting that RXFP3 is unlikely to have a direct influence on VTA neuron activity. A previous study reported a high level of RXFP3 in the VTA using immunohistochemistry with a non-validated, (and commercially discontinued) anti-RXFP3 antibody (Paul et al., 2019). Antibodies that have high sensitivity and specificity against G-protein coupled receptors (GPCRs) such as RXFP3 are rare, as receptor antigens usually misfold once they are no longer in the lipid bilayer following injection into animals (Michel et al., 2009). Therefore, this may explain the discrepancy between this previous immunohistochemical study, and the finding of low VTA tdTomato described here. However, earlier in situ hybridisation studies also suggest moderate levels of Rxfp3 mRNA are present within the VTA (Lein et al., 2007;Smith et al., 2010). Paul and colleagues also observed that Rxfp3 mRNA was enriched within GABAergic VTA neurons, and that relatively more RXFP3-expressing neurons were present within the rostromedial regions of this structure (Paul et al., 2019), which is slightly different to the more caudal/central coronal level studied here. Further studies using other methods to detect RXFP3 in this area are therefore warranted, as transgenic mouse strains such as the one used here are sometimes prone to falsenegative detection of markers (see below).
Lastly, the absence of bright tdTomato-expressing cells within the ARC and DMH of a Rosa26-tdTomato mouse demonstrated that tdTomato expression is Cre-dependent (and therefore, represents RXFP3) within these areas in RXFP3-Cre/tdTomato mice. However, the presence of bright tdTomatoexpressing cells within the PT of a Rosa26-tdTomato mouse suggests that RXFP3 is absent from this structure, and highlights the importance of examining these mice as controls (Madisen et al., 2010). Further studies to investigate the potential expression and function of RXFP3 within other regions of the pituitary gland, are nonetheless warranted. For example, RXFP3 expression was recently detected in the LβT2 mouse pituitary gonadotroph cell line (Kanasaki et al., 2019), and data from the Genotype-Tissue Expression (GTEx) project describes RXFP3 expression within human pituitary (and adrenal) samples 1 . The potential source of relaxin-3 that might access these receptors is unknown, however. 1 www.gtexportal.org/home/gene/RXFP3 Furthermore, several rodent studies have observed that relaxin-3 can influence the hypothalamic-pituitary adrenal and gonadal axes, although some of these effects may be due to relaxin-3 binding to RXFP1, rather than RXFP3 (Zhang et al., 2017;De Ávila et al., 2018).
Investigating the identity and function of RXFP3-positive cells in other brain areas may also provide mechanistic insights into how RXFP3 modulates feeding behavior and relevant neuroendocrine systems. In this regard, a previous study observed direct RXFP3-mediated inhibition of oxytocin and vasopressin neurons within the rat paraventricular hypothalamic nucleus (PVN) (Kania et al., 2017). A recent follow-up study characterized these mechanisms further, and demonstrated that local PVN injection of an RXFP3 antagonist prevented binge eating in female rats (Kania et al., 2020). The absence of tdTomato fluorescence within the PVN of RXFP3-Cre/tdTomato mice was therefore notable, particularly as previous in situ hybridisation studies have demonstrated high levels of Rxfp3 mRNA expression within this structure (Smith et al., 2010;Ma et al., 2017). The reason for this discrepancy is unknown, but may be due to the different mouse strains used or may represent false-negative fluorophore expression, whereby the fluorophore fails to be faithfully expressed within some neuronal populations, an issue often associated with transgenic reporter mice (Yang and Gong, 2005).
In summary, this study identified that a small proportion of hypothalamic dopaminergic neurons express RXFP3. Of particular interest, the detection of a small population of ARC dopaminergic neurons that express RXFP3 highlights a novel potential mechanism by which RXFP3 activation may inhibit these neurons to promote food intake and influence neuroendocrine functions.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

ETHICS STATEMENT
The animal study was reviewed and approved by the Florey Institute of Neuroscience and Mental Health Animal Ethics Committee.

AUTHOR CONTRIBUTIONS
AG and CS conceived the project and designed the study. LV performed the experiments and completed the analysis, including preparation of the figures, with contributions from SC, CZ, and IE. LV and CS wrote the manuscript. JR, LG, LR, LH, and RW provided technical advice, project guidance, and other assistance.