Numerous vital behaviors such as the finding of food sources, avoiding predators, and recognizing conspecifics are triggered by olfactory cues. In mammalian species, the relevant odorous molecules are detected by specialized olfactory sensory neurons (OSNs) which are organized in structurally and functionally distinct chemosensory subsystems [for reviews see (Breer et al., 2006; Ma, 2007; Munger et al., 2009)]. One special subsystem is located within the main olfactory system (MOS) and is formed by the so-called OR37 subfamily of odorant receptors (ORs) (Strotmann et al., 1992). This type of OR is structurally unique due to an insertion of six amino acids in its third extracellular loop (Kubick et al., 1997); furthermore, it exists exclusively in mammalian species (Hoppe et al., 2006), and against the general trend for ORs, the various OR37 subtypes are highly conserved among each other and across species borders from opossum to human (Hoppe et al., 2006). These features have led to the speculation that the OR37 subsystem may be tuned to detect special chemical compounds, which might be particularly relevant for mammals. In fact, we have recently demonstrated that the OR37 receptors are specifically activated by long-chain fatty aldehydes (Bautze et al., 2012); these compounds are extremely hydrophobic and were never described as active ligands for any other ORs in previous studies.

These aspects raise the question whether the information received via the OR37 sensory cells may also be processed in the brain in a way that differs from that of general OSNs. Indeed, the visualization of axonal projections from OSNs expressing defined OR37 receptors into the main olfactory bulb (MOB) has revealed that they target only a single glomerulus (Strotmann et al., 2000), a pattern which is clearly different from that of most other OSN populations which typically send their axons to two glomeruli [for recent reviews see (Imai et al., 2010; Sakano, 2010; Mori and Sakano, 2011; Murthy, 2011)]. Interestingly, the glomeruli for the different OR37 subtypes were found to be grouped together in the ventral domain of the MOB, a region which is believed to play a role in the processing of socially relevant olfactory cues (Schaefer et al., 2001, 2002; Lin et al., 2007). Using DiI and a genetic tracing approach we have recently provided first data showing that also the connectivity from a defined OR37 glomerulus into higher brain centers is organized in a special manner (Bader et al., 2012). In contrast to the prevailing principle that projection neurons from the MOB, the mitral/tufted (M/T) cells, send their axons to the piriform cortex, the anterior olfactory nucleus or the cortical amygdala which together form the olfactory cortex, a connectivity between an OR37 glomerulus and the medial amygdala (Me) was discovered. This finding is particularly notable since the Me receives input primarily from a different chemosensory subsystem, the accessory olfactory system (Kevetter and Winans, 1981) which is involved in the processing of pheromonal information (Halpern and Martinez-Marcos, 2003). The connectivity from the OR37 glomerulus was found to be confined to even a particular subregion of the Me, namely the posterior-dorsal subnucleus (MePD). These data thus provided first evidence that the projection from the OR37 subsystem into higher brain centers is in fact different from the general design of the MOS.

Our data have indicated that in addition to the special relay into the amygdaloid complex, a connection between the OR37 glomerulus and distinct hypothalamic nuclei, the supraoptic nucleus (SO), and the paraventricular nucleus (PVN) seems to exist. It remained unclear in these cases, however, whether the connectivity into these regions is made up directly by the axons of projection neurons originating from an OR37 glomerulus, or indirectly via for example the amygdala. Furthermore, toward an insight into the functional relevance of this connection, the identification of the cell type(s) to which it is formed, is an essential step. It was therefore the aim of the present study to further characterize the type of connectivity between the OR37 system and these particular hypothalamic nuclei.

To address the question whether a direct cellular connectivity from an OR37 glomerulus into the PVN of the hypothalamus may exist, we exploited the fact that the lipophilic dye molecule DiI, when inserted into the plasma membrane of a given cell, spreads by lateral diffusion within the processes of this particular cell; such an approach has been used in previous studies to unravel the direct connectivity between defined brain regions (Godement et al., 1987; Honig and Hume, 1989). The fluorescently labeled OR37 glomerulus in our OR37-ITGFP transgenic mouse lines (Figure 1A) (Strotmann et al., 2000) served as a landmark for the deposition of a dye crystal (Figure 1B), anticipating that the dye would get into contact with the dendrites of projection neurons which are associated with OR37 glomeruli. After an appropriate incubation time of the tissue, serial sections through the hypothalamus were analyzed. The PVN can be recognized due to its position in close vicinity to the third ventricle (Figure 1C); however, for an unambiguous identification we used the fact that it contains, among many others, arginine-vasopressin (AVP)-producing neurons (Kadar et al., 2010; Biag et al., 2012). Immunohistochemical analyses of serial sections through the hypothalamus in fact revealed AVP-positive cells within a well circumscribed region close to the third ventricle (Figure 1D), thereby delineating the nucleus. Every other section was examined for the presence of DiI-fluorescence. Close inspections revealed DiI-labeled filamentous structures within the region of the AVP-positive cells (Figure 1E); they were sparsely dispersed throughout the corresponding region. After a longer incubation time, a picture as representatively shown in Figure 1F emerged. At high magnification, ramifications of the fiber-like structures with small terminal-like endings were visible; many of them were covered with varicose-like elements, indicating a possible en passant synaptic arrangement with cells within the PVN. The application of DiI onto the OR37 glomerulus in the MOB thus clearly labeled fibers in this particular hypothalamic nucleus (n = 6); a close examination of the nucleus revealed no staining of cell bodies. Together these results support the concept that a direct connectivity between the OR37 region of MOB and this hypothalamic nucleus exists which is formed by the axonal processes of projections neurons which are situated at the site of dye application at the OR37 glomeruli.

To examine whether OR37 M/T-cells also project directly to the SO, we examined this region after placement of the DiI-crystal onto the fluorescent OR37 glomerulus. The SO can be found at the ventral base of the brain in immediate vicinity to the optic tract (see Figure 2A). Also in this case, we made use of the fact that it contains AVP-producing neurons to unequivocally identify it (Ludwig et al., 2005). By using an immunohistochemical approach, a densely packed group of AVP-positive cell bodies could be visualized in neighborhood to the optic tract (Figure 2B). Consecutive sections were examined for the presence of the dye. In contrast to the rather sparse staining in the PVN, a thicker bundle of DiI-positive fibers was visible extending to the portion of the SO where the corresponding cell somata are located (Figure 2C) (n = 6). These data thus provided evidence that also the SO of the hypothalamus receives direct input from M/T-cells originating at the OR37 region of the MOB.

Regarding the functional role of the relays between the OR37 region of the MOB and these hypothalamic nuclei, the knowledge about the cell type(s) to which this connection is formed, is an essential step. To address the question we used our previously generated transgenic mouse lines in which OR37C expressing sensory neurons coexpress the transsynaptic marker WGA (Bader et al., 2012); in these transgenic mice the cells in the hypothalamus to which this marker is transferred can be identified by their WGA-immunoreactivity, whereas in wild type animals no such labeling is visible (data not shown). The PVN is one of the most complex and heterogeneously organized nuclei of the brain; its major function is the secretion of diverse neuropeptides which are produced in distinct cell populations (Kadar et al., 2010; Biag et al., 2012). To reduce the experimental expense necessary for a molecular phenotyping of the WGA-immunoreactive cells, we first exploited the fact, that the PVN is not homogenously packed with neurons and those cells producing the different neuropeptides are not evenly distributed, but arranged in characteristic patterns (Biag et al., 2012). In a first approach the distribution of WGA-immunoreactive cells within the PVN was systematically analyzed (n = 2). By combining the anti-WGA immunohistochemistry with Nissl staining, it became evident that no WGA-immunoreactive cells were found within the periventricular part of the PVN, where the cell density is typically low (as shown on the representative sections in Figure 3). Since this part of the nucleus is occupied by the somatostatin-producing cells (Biag et al., 2012), these were considered as unlikely candidates. Cells with WGA-immunoreactivity were restricted to the remaining major portion of the PVN, where the cell density is high (Figures 3A–D). A detailed mapping revealed that at the most rostral level, the WGA-immunoreactive cells tended to cluster in the ventral part of the densely packed major portion of the PVN (Figure 3A). At the mid rostro-caudal extent they were located predominantly in the dorsal half (Figure 3B) and at the most caudal level in the dorsolateral part (Figures 3C and D). This particular pattern within the PVN is most reminiscent to the previously described distribution of cells producing either AVP or oxytocin (Oxy) (Biag et al., 2012). We therefore concluded that WGA-positive cells in the PVN may represent vasopressinergic, oxytonergic, or a mixture of both cell types. To scrutinize this hypothesis, double labeling experiments using the anti-WGA antibody in combination with antibodies against the two peptides were performed. Distinct populations of oxytonergic (Figure 4A) or vasopressinergic (Figure 4D) cells could be visualized by this approach. On the same sections multiple WGA-positive cells could be observed (Figures 4B and E). An overlay of these stainings demonstrates that with the anti-oxytocin antibody hardly any of these cells was double labeled (Figure 4C). In contrast, most (95.4% ± 1.4%; n = 3; mean ± standard deviation) of the AVP-producing cells were double labeled using the anti-WGA antibody (Figure 4F). These results showed that cells in the PVN which are labeled by the transsynaptic marker are vasopressinergic, but not oxytonergic. Previous studies have shown, that also the SO of the hypothalamus is composed of AVP- and Oxy-producing cell populations (Sabatier et al., 2003; Ludwig et al., 2005), raising the question, to which cell type in this nucleus the connectivity from the OR37C glomerulus is formed. The staining with specific antibodies against Oxy or AVP, respectively, visualized distinct populations in the SO (Figures 5A and D). On the same sections numerous WGA-positive cells could be visualized (Figures 5B and E). An overlay of the anti-WGA and the anti-oxytocin immunoreactivity (Figure 5C) revealed hardly any double labeled cells. In contrast, the AVP- and WGA-immunoreactivity almost completely overlapped (Figure 5F). These results thus showed that also in the SO, the transsynaptic marker labeled vasopressinergic cells, whereas oxytonergic cells were not stained. Thus, the connections from the OR37C glomerulus are formed to cells producing AVP in both nuclei.

Previous studies have shown that vasopressinergic cells in the SO represent a homogenous group of large, so-called magnocellular neurons (Castel and Morris, 1988), which project to the posterior pituitary (Swanson and Sawchenko, 1983). In contrast, the AVP-producing cell population in the PVN is comprised not only of magnocellular neurons, which also project to the posterior pituitary, but also of parvocellular neurons with a smaller diameter which project to multiple extra-hypothalamic sites (Sawchenko and Swanson, 1982; Antoni, 1993; Hallbeck and Blomqvist, 1999). To address the question, to which of these vasopressinergic cell types in the PVN the connectivity is formed, a comparative morphological examination of the cells was performed. To determine how the morphology of the two cell types appears in our preparations, cells from the SO which is comprised exclusively of magnocellular neurons and the suprachiasmatic nucleus (SCh) which contains only parvocellular neurons (Sofroniew and Weindl, 1978) were examined. On Nissl stained sections through the SO, the magnocellular neurons appeared as large and spindle-shaped cells, (Figure 6A) and the parvocellular neurons in the SCh were smaller and mostly round-shaped (Figure 6F). On the Nissl stained sections through the PVN, neurons with a magnocellular (Figure 6B) and with a parvocellular (Figure 6G) appearance could be identified. Based on these features, also the AVP-positive neurons in the PVN could be categorized into larger magnocellular (Figure 6C) and smaller parvocellular (Figure 6H) cells. Double labeling experiments together with the anti-WGA antibody (Figures 6D and I) revealed that cells labeled by the transsynaptic marker (Figures 6E and J) are magnocellular and parvocellular vasopressinergic neurons.

In the present study we have demonstrated that an application of the tracer molecule DiI onto an OR37 glomerulus on the ventral side of the MOB results in the labeling of fibers within distinct hypothalamic nuclei, the PVN and the SO. When the lipophilic dye molecule becomes inserted into the plasma membrane of a given cell, it spreads within the processes of this particular cell (Godement et al., 1987; Honig and Hume, 1989); this strongly suggests that a direct cellular connection exists between the ventral domain of the MOB and these hypothalamic nuclei. We anticipate that after the application of the dye onto an OR37 glomerulus it comes into contact with the dendritic processes of the bulbar projection neurons which are positioned at that site (Shepherd, 1972) and distributes along their axonal connections into the labeled brain areas. Based on our observations that the transsynaptic tracer WGA, which is specifically produced in OR37C expressing OSNs, arrives within the same brain areas, we conclude that projection neurons originating from this particular glomerulus send their axons into the identified hypothalamic nuclei and that this connectivity is in fact monosynaptic. Since DiI would not only label the processes of cells which project toward those brain areas, but also those of cells which project to the application site, it is principally possible that this approach also stained centrifugal neurons that support efferent input to the olfactory bulb (Davis et al., 1978; De Olmos et al., 1978). However, the fact that we did not observe DiI-labeled cell bodies within these particular brain nuclei and, moreover, no centrifugal input from there into the MOB has been observed so far, argues against the idea that centrifugal neurons were visualized in this case.

In contrast to the DiI-tracing approach which revealed labeled fibers at the PVN and the SO, the WGA-based technique stained almost exclusively cell bodies in the same brain regions. This difference is most likely due to the different properties of the labels. Whereas, DiI is lipophilic and integrates into the cell membrane of the projection neurons and thus labels their axons (Godement et al., 1987; Honig and Hume, 1989), WGA is transferred into the synaptically coupled cells, thereby labeling also their cell body (Yoshihara et al., 1999). Unfortunately, we could not visualize the fibers of the projection neurons by the WGA approach, although WGA must have been transported along their axons to reach that site in the brain. The reason for this is currently not known; it seems conceivable that most of the protein is transferred to the postsynaptic cells, leaving too little material in the incoming axons to be detectable by the applied methods.

However, both approaches together strongly support the concept that projection neurons from OR37 glomeruli send their axons directly to the PVN and the SO of the hypothalamus. Interestingly, in the rat a direct projection of neurons from the MOB to the SO has already been described (Smithson et al., 1989). In these studies this neuronal connection was visible after injection of the anterograde tracer Phaseolus vulgaris leucoagglutinin into the MOB and confirmed by the retrograde transport of rhodamine-labeled latex microspheres or the tracer Fluoro-Gold into the MOB after injection into the SO. Electrophysiological experiments finally supported the concept of a monosynaptic input to the SO from the MOB (Yang et al., 1995). However, at that time the labeling could not be attributed to a particular OR population. No evidence for a direct connection from the MOB to the PVN was found in previous experiments (Smithson et al., 1989). The difference to our results in the mouse could either be due to technical reasons, for example differences of the tracers that were employed. Alternatively, they could be due to species differences (rat vs. mouse); it is well-established by now that although these two species are quite related, substantial differences exist in the organization of their nervous systems (Dorn et al., 1981; Hirst et al., 2003; Biag et al., 2012).

It is well-known that as an important central regulator of the endocrine system the hypothalamus receives chemosensory information which modulates for example distinct endocrine responses related to reproduction (Schally, 1980; Kalra and Kalra, 1983). In fact, the different chemosensory organs in the nose transmit odor information to the hypothalamus, however, the currently described neuronal pathways are all multisynaptic. They include input from the accessory olfactory bulb, from where projection neurons send their axons first to the Me, which in turn projects to distinct hypothalamic nuclei (Krettek and Price, 1978; Kevetter and Winans, 1981; von Campenhausen and Mori, 2000; Salazar and Brennan, 2001; Choi et al., 2005). Also the MOS sends information via distinct regions of the olfactory cortex, such as the anterior olfactory nucleus, the piriform cortex, and the olfactory tubercle (Price et al., 1991). There is even a convergence of accessory and main olfactory pathways at the level of the Me which is further relayed to the hypothalamus (Pro-Sistiaga et al., 2007; Kang et al., 2009). In contrast to these known pathways, we now provide evidence for the existence of projection neurons from the OR37 glomeruli in the MOB which establish a direct connection to the hypothalamus. It seems conceivable that signals carried by such a monosynaptic pathway may be less subject to modification and would thereby enable these projection neurons to gain unmodified access to the hypothalamus.

Our molecular phenotyping approaches of the WGA-labeled cells in the hypothalamus revealed a rather defined neuroendocrine cell population; interestingly, in both hypothalamic nuclei, vasopressinergic cells were stained by the transsynaptic marker. These results suggest that OR37 related odor information may affect the activity of these particular cell populations. The consequence of this influence is currently unknown, however, meanwhile it is well established that vasopressinergic cells in these hypothalamic nuclei not only secrete the peptide at the posterior pituitary for osmoregulation, but also release it within the central nervous system, either at axon terminals of the vasopressinergic cells that project to distinct extra-hypothalamic sites (Sofroniew, 1980; Sawchenko and Swanson, 1982; Castel and Morris, 1988; Rood and De Vries, 2011), or at their somata and dendrites (Landgraf et al., 1995; Ludwig et al., 2005), from where it diffuses into other brain regions. Interestingly, many studies have demonstrated that centrally released vasopressin is crucially involved in the regulation of distinct social behaviors (Bielsky and Young, 2004; Caldwell et al., 2008; Albers, 2012). Consistent with these findings, the vasopressin receptor types Avpr1a and Avpr1b are found in relevant brain areas (Ostrowski et al., 1994; Vaccari et al., 1998) and it was shown that both receptors mediate effects of vasopressin on social behavior (Bielsky et al., 2004; Stevenson and Caldwell, 2012). Studies performed with Avpr1b-knock-out mice showed for example that male and female animals were less aggressive in a specific social context and exhibit deficits in their olfactory-based motivation to interact with social stimuli, although they displayed no alterations in their ability to discriminate distinct odors (Stevenson and Caldwell, 2012). In rodents, social behavior is highly dependent on olfactory cues; therefore it is plausible that at the level of the central nervous system, vasopressin is involved in the regulation of odor-induced social behaviors. Our finding of a connectivity from OR37 expressing sensory neurons in the nose to vasopressinergic cells in the hypothalamus raise the possibility that odor information detected by these receptors is involved in the regulation of social behaviors. This idea is supported by our previous findings that OR37 projection neurons also terminate in the Me (Bader et al., 2012), which receives chemosensory information from the vomeronasal organ, which has long been known to be involved in the detection of social odor cues [reviewed in Halpern and Martinez-Marcos (2003)]. The convergence of OR37 projection neurons with the accessory pathway in the medial amygdala thus supports the concept of the OR37 receptors being involved in the regulation of social behavior. Interestingly, there is also some evidence that peripheral vasopressin could play a role in mediating complex social behaviors. In a very recent study it was shown that in cichlid fish dominant males have lower amounts of the fish-homolog vasotocin in their pituitary and therefore produce more urine compared to subordinate males; it was proposed that vasotocin plays a key role in orchestrating different social phenotypes (Almeida et al., 2012). Like in fish, also in mice urine is an important mediator of social odor cues, therefore it seems possible that vasopressin couples social odor information at the level of the central nervous system to related physiological changes in the periphery.