Hemoglobin in the blood acts as a chemosensory signal via the mouse vomeronasal system

The vomeronasal system plays an essential role in sensing various environmental chemical cues. Here we show that mice exposed to blood and, consequently, hemoglobin results in the activation of vomeronasal sensory neurons expressing a specific vomeronasal G protein-coupled receptor, Vmn2r88, which is mediated by the interaction site, Gly17, on hemoglobin. The hemoglobin signal reaches the medial amygdala (MeA) in both male and female mice. However, it activates the dorsal part of ventromedial hypothalamus (VMHd) only in lactating female mice. As a result, in lactating mothers, hemoglobin enhances digging and rearing behavior. Manipulation of steroidogenic factor 1 (SF1)-expressing neurons in the VMHd is sufficient to induce the hemoglobin-mediated behaviors. Our results suggest that the oxygen-carrier hemoglobin plays a role as a chemosensory signal, eliciting behavioral responses in mice in a state-dependent fashion.

that is sensitive to blood. The authors further demonstrate that the amino acid Gly17 in the hemoglobin protein is crucial for the interaction between blood and the receptor Vmn2r88. Furthermore, the authors show that hemoglobin acts as a chemosensory signal to the brain that has an important behavioral function. They do so by showing that hemoglobin exposure results in "exploratory" behaviors in lactating mothers. Even though the G protein-coupled receptor Vmn2r88 was recently identified as important for hemoglobin sensing in mice (Isogai et al., Cell 2018), the authors make additional important contributions that make the first part of the paper particularly impressive. Specifically, the authors show that Gly17 on hemoglobin is a crucial site for receptor interactions. This finding might also provide an evolutionary perspective on blood sensing and is crucial for understanding the interaction between hemoglobin and the Vmn2r88 receptor. Another central addition to the Isogai et al., paper is the second part of this study, which identifies a behavioral output following the olfactory sensing of hemoglobin, and identifies potential downstream neuronal circuits that are important for executing the behavioral outputs. This is an important new discovery as in Isogai et al., the authors could not identify a major behavioral affect when mutating the Vmn2r88 receptor. Taking these new discoveries into consideration, this paper can to be suitable for publication in Nature Communications if some further work would be added to strengthen the second part of this study (see below). Major concerns: 1. Mothers and non-parental mice. While most of the major new findings in the second part of the paper are performed in lactating females, the optogenetics and experiments of Figs. 1-3 are performed in naïve males. At the very least, the optogenetic experiments should be performed in lactating females as well. Furthermore, it would be helpful to know if the Hb fos responses are comparable in the VNO \ AOB of mothers vs. naïve females. Such data could add mechanistic insights to the changes in responses to hemoglobin observed between naïve females and mothers in downstream brain regions. Other work on the accessory olfactory system seem to support this line of inquiry (Dey et  2. Behavioral responses to hemoglobin. In order to identify how hemoglobin affects behaviors through neuronal activity, the authors show: 1. That exposure to hemoglobin results in elevation of cfos expression in VMHdm of mothers. 2. That hemoglobin exposure results in digging and rearing behaviors. 3. That artificial activation of SF1+ neurons in VMHdm can result in behaviors such as digging and rearing in naïve males. However, this does not clarify the question of whether or not hemoglobin actually elicits digging\rearing behaviors through SF1+ neurons in VMHdm in mothers. In order to connect these observations, the authors should use a loss of function approach. Ideally, the authors should reversibly silence SF1+ neuronal activity in VMHdm and test whether the behavioral responses to hemoglobin are affected. This can potentially show that SF1+ neurons in VMHdm are necessary for executing rearing\digging behaviors upon sensing hemoglobin. 3. "Risk assessment" vs. "exploratory" behavior. The authors use the term "exploratory behaviors" when they describe the behavioral outputs observed in response to hemoglobin (in mothers), based on negative state (e.g., cort levels). However, it can be argued that the authors could not detect anxiety markers because the anxiety levels are too low to produce detectable increases. Furthermore, the behavioral data that are presented in the paper may be more accurately described as risk assessment, a mild form of defensive behavior or what Fanselow and Zhuravka described as an "early defensive mode" (Fanselow and Zhuravka, Behavioral Processes 2019). Importantly, it has been demonstrated earlier that the VMHdm and specifically SF1+ neurons in VMHdm are important in processing defensive behaviors including rearing (Kunwar et al., eLIFE 2015, Wang et al., Neuron 2015. In addition, digging can be considered as a form of a risk assessment (for example Gozzi et al., Neuron 2010). The authors should therefore consider switching\adding the term "risk assessment" in order to be more consistent with the literature (Gross and Canteras, NRN 2012, Ribeiro-Barbosa et al., Neuroscience & Biobehavioral Reviews 2005). Interestingly, the authors show using an innate fear inducing volatile odor (2MT, fig.5f), that mice transit from an immobility\freezing behavior elicited by high doses of 2MT (freezing due to immediate danger) to digging behavior ("reduce the likelihood of encountering danger" Fanselow and Zhuravka, Behavioral Processes 2019). Since both low doses of 2MT and hemoglobin elicit similar risk assessment\exploratory behaviors, it would be informative to examine whether a lower dose 2-MT activates the same SF1+ VMHdm neurons as Hb in mothers (e.g., by fos-catFISH or single-cell imaging). This is especially important since TMT, a compound related to 2-MT, has previously been shown not to activate VMHdm (Perez-Gomez et al., 2015), in contrast to other predator odors.
Minor concerns: -It has been demonstrated that outputs from VMHdm to AHN are important for avoidance behaviors, and that VMHdm to dPAG is important for immobility (Wang et al., Neuron 2015). The authors could address these findings made by previous works through examining cFOS expression in the AHN of mothers (and if needed, change figure 7f accordingly).
-In several places, the authors measure cfos activity "Per region". This is not clear. Did the authors collect the whole region? Did the authors randomly apply a single brain slice from each region? Since many of the observations described here rely on this analysis, this should be explained in more detail and include exact coordinates.
-Why are there responses in the VNO of naïve males and females if they have no behavioral relevance? The authors should clarify this in the text.
-What is the role for the main olfactory bulb in hemoglobin sensing in mothers? This is important especially since it has been shown that there is functional plasticity to behaviorally relevant odors in the main olfactory bulb of mothers (Vinograd et al., Cell Reports 2017). - Figure 6 b and 6 c. What is the difference between the control groups in figure 6b (rearing time) and figure 6c? Similarly, what is the difference between the hb condition in figure 6b and hb +/+ condition in figure 6c? If these are different animals for the same conditions the authors should group them together to increase the statistical power.
- Figure 1, individual mice. While in all figures the variance can be estimated by explicitly showing the data for individual subjects on top of the bar graphs, it will be informative to add this in figure 1 as well (and not just error bars).
-In supplemental figure 7, mice spent similar time in a 2-chamber assay where hemoglobin was compared to the control. However, several concerns arise from this interpretation made by the authors. Most importantly, it seems that the mice tested in the assay were male mice and not mothers. This is a surprising choice for this assay, considering the fact that males did not show a significant behavioral difference in other assays, and other works imply that hemoglobin might be attractive to males (Isogai et al., Cell 2018).
In summary this is a strong submission and our comments should be taken as constructive remarks to improve the paper and eliminate confusion and ambiguity for the readers. We congratulate the authors on their beautiful and impressive work.
Reviewer #3: Remarks to the Author: In this paper, Osakada and colleagues -stimulated by initial observations that blood activates the VNO -identify the specific blood ligand responsible for this activation (beta-globin) and its corresponding VNO receptor (Vmn2r88). Using Vmn2r88-knockout mice, they demonstrate that smelling blood causes exploratory behavior (rearing and digging) in a receptor-dependent manner, and that this response is elicited in lactating mothers, but not virgin females or males. The behavioral response is attributed to the specific activation of dorsal PAG, MeA and dorsal VMH (dVMH) in lactating mothers, and was recapitulated by optogenetic activation of SF1 neurons in dVMH. On the basis of these data, the authors claim that the smell of blood activates a specific behavioral response that is dependent on an AOB-MeApv-VMHd-PAGd pathway and unique to lactating mothers.
Overall, the experiments identifying beta-globin and Vmn2r88 (Figs 1 and 3) in this paper are elegant and convincing, and consistent with a recent paper (Isogai, 2018) that also found betaglobin and Vmn2r88 to mediate the sensing of blood VNO. The subsequent neural and behavioral analysis, however, could be improved to rule out alternative explanations for the authors claims. In particular, I am concerned that the data may reflect a general response to VNO activation that could include a much larger category of ligands than just beta-globin. For example, the authors show that 2MT also elicits digging behavior when presented at very low concentration. In addition to this overall feedback, I have several concerns and questions that will clarify this manuscript: 1. To clarify the specific effects of hemoglobin, it would be helpful to see the same behavioral experiments repeated for other ligands, including some that would likely covary with hemoglobin (e.g. other infant-associated ligands), and others that are independent. 2. Fig 7 shows the result of optogenetic stimulation of SF1 neurons in dVMH. SF1 is neurons are allegedly involved in fear responses and comprise the majority of dVMH cells. Interpreting this result requires knowing whether blood activates these SF1 neurons. 3. Figure 2d,e is used to show that the H78N mutation in beta-globin does not inhibit VNO activation. However, data are only shown for two mutant proteins and a no-odor control, but not for the WT protein. Thus it cannot be determined from the figure whether H78N leads to a reduction of VNO activation compared to the WT protein. 4. Is the ability of VNO neurons to detect hemoglobin state dependent (like the fos activity reported in the brain)? 5. If blood is placed on a pup, how does the mother behaviorally react? 1 We greatly thank all reviewers for the enthusiasm about our study, and for their constructive comments. We have performed additional experiments and modifications in response to the comments, which have resulted in significant improvement of the paper. Below we first summarize the additional experiments in this revision, and then provide point-by-point responses to the reviewer's comments.

Additional experiments performed:
-Adding new conditions to open field, digging behavior, and pup retrieval assays with lactating mothers. In the manuscript from Osakada et al., the authors define molecular mechanisms for mouse detection of blood. They nicely, identify a single receptor expressed in the vomeronasal organ (VNO) that detects a particular amino acid residue of hemoglobin. Further they show that detection of hemoglobin triggers apparent exploratory behaviors specifically in lactating females. Blood as a sensory cue is somewhat novel and it is not entirely clear what survival advantage is afforded by this sensory pathway in mice, but it is interesting that it is unique to lactating females. This part of the manuscript is solid and very well done. They move on to look at downstream brain areas involved in the behavioral response to hemoglobin, by examining hemoglobin-evoked c-fos activity, and they find activity in typical parts of the VNO pathway. They then, activate neurons in the ventromedial hypothalamus (VMHd) and find that weak activation produces behaviors similar to those elicited by hemoglobin. The implication of the neural circuit involved is a quite weak. Nonetheless, the finding of the receptor and ligand that play a role in a novel behavior, is well done and interesting. I offer the following, mostly minor concerns and suggestions.

2
We thank the reviewer for careful read of the manuscript and important suggestions. 1. In Figure 2c. Why do the authors not show the amino acid sequence from frog and zebra fish to be consistent with Figure 2b? Presumably, they also lack the G17 amino acid. Although this is really minor, it would be nice to see and it would be more continuous with the previous graph.
We thank the reviewer for this suggestion. We added the sequences of frog and zebrafish β-globin (new
We sincerely apologize that our previous description was not precise. We used 2MT (2-Methyl-2-thiazoline, Sigma-Aldrich, M83406), diluted in mineral oil from 100-fold to 50,000-fold. We have changed the description of the x-axis and figure legends in new Figure 5e and 5f. 5. In Figure 6, the authors use an open field assay to ask if hemoglobin triggers anxiety like behaviors. They do not observe any clear signs of anxiety suggesting that pre-exposure to hemoglobin does not produce anxiety. However, without a positive control it is unclear if the assay is suitable to detect pre-exposure evoked anxiety. The authors should show, for example, that pre-exposure to 2MT causes some measure of anxiety in this assay. In addition, it would be useful to look at hemoglobin exposure during the assay. What if hemoglobin causes a transient state of anxiety that is reset by introduction to the assay?
We thank the reviewer for these important comments. We performed open field assays with pre-exposure of 2MT but there was no decrease in the total center time (Revise figure 2 below, new Figure 6 in the manuscript), suggesting that as this reviewer pointed out, we cannot mention anxiety in this open field assay.
On the other hand, rearing enhancement was observed specifically upon stimulation with hemoglobin that was diminished in the hemoglobin receptor-deficient mice, suggesting that a type of exploratory behavior was induced. It should be noted that the rearing enhancement was observed when the assay was performed without bedding, whereas digging but not rearing behavior was seen in the presence of bedding (  Lastly, our model in Figure 7f appears to be supported by these new results, but the circuit is still loosely defined as the reviewer mentioned. Thus, we decided to move the model to Supplementary figure 10h to advocate potential neural circuits responsible for hemoglobin information and its outputs in lactating females.

Reviewer #2 (Remarks to the Author):
This elegant study identifies a specific receptor in the vomeronasal organ (VNO) of mice, Vmn2r88, that is sensitive to blood. The authors further demonstrate that the amino acid Gly17 in the hemoglobin protein is crucial for the interaction between blood and the receptor Vmn2r88. Furthermore, the authors show that hemoglobin acts as a chemosensory signal to the brain that has an important behavioral function. They do so by showing that hemoglobin exposure results in "exploratory" behaviors in lactating mothers.
Even though the G protein-coupled receptor Vmn2r88 was recently identified as important for hemoglobin sensing in mice (Isogai et al., Cell 2018), the authors make additional important contributions that make the first part of the paper particularly impressive. Specifically, the authors show that Gly17 on hemoglobin is a crucial site for receptor interactions. This finding might also provide an evolutionary perspective on blood sensing and is crucial for understanding the interaction between hemoglobin and the Vmn2r88 receptor. Another central addition to the Isogai et al., paper is the second part of this study, which identifies a behavioral output following the olfactory sensing of hemoglobin, and identifies potential downstream neuronal circuits that are important for executing the behavioral outputs. This is an important new discovery as in Isogai et al., the authors could not identify a major behavioral affect when mutating the Vmn2r88 receptor. Taking these new discoveries into 8 consideration, this paper can to be suitable for publication in Nature Communications if some further work would be added to strengthen the second part of this study (see below).
We appreciate the reviewer's careful read and all the valuable and constructive comments.

Revise figure 7
Hemoglobin activates Vmn2r88 expressing cells in the VNO and neurons in the AOB in both lactating and virgin females. 11 a Vmn2r88 ISH and pS6 immunostaining of VNO sections from mother and virgin female mice exposed to hemoglobin (Hb) or distilled water. Scale bar, 50 µm. b Quantification of visualized neurons in the VNO. The Wilcoxon signed-rank test. 3. "Risk assessment" vs. "exploratory" behavior. The authors use the term "exploratory behaviors" when they describe the behavioral outputs observed in response to hemoglobin (in mothers), based on negative state (e.g., cort levels). However, it can be argued that the authors could not detect anxiety markers because the anxiety levels are too low to produce detectable increases. Furthermore, the behavioral data that are presented in the paper may be more  fig.5f), that mice transit from an immobility¥freezing behavior elicited by high doses of 2MT (freezing due to immediate danger) to digging behavior ("reduce the likelihood of encountering danger" Fanselow and Zhuravka, Behavioral Processes 2019). Since both low doses of 2MT and hemoglobin elicit similar risk assessment¥exploratory behaviors, it would be informative to examine whether a lower dose 2-MT activates the same SF1+ VMHdm neurons as Hb in mothers (e.g., by fos-catFISH or single-cell imaging). This is especially important since TMT, a compound related to 2-MT, has previously been shown not to activate VMHdm (Perez-Gomez et al., 2015), in contrast to other predator odors.
We agree with the reviewer's comment. We changed the relevant descriptions in the manuscript to 'exploratory and/or risk assessment behavior'. Regarding the VMHd activation, we did not see a significant -In several places, the authors measure cfos activity "Per region". This is not clear. Did the authors collect the whole region? Did the authors randomly apply a single brain slice from each region? Since many of the observations described here rely on this analysis, this should be explained in more detail and include exact coordinates.
We examined hemoglobin-induced c-Fos expression in the anterior hypothalamic nucleus (AHN) of lactating females (Revise figure 9 below). There was no significant increase in c-Fos expression in the AHN (the number of c-Fos+ cells, control; 37.8 ± 7.8 (n = 4), hemoglobin; 60.3 ± 14.6 (n = 5), p=0.05 by the Wilcoxon signed-rank test.
We apologize that our previous explanation about the quantification of staining results was not sufficient. In our staining experiments, we collected multiple sections from each region to cover entire populations (eg; in the VMHd we counted activated cells in the 5 coronal sections from anterior to posterior). As the reviewer suggested, we added the number of sections we counted per region in the legends of Figure 4 and Our results from c-Fos analysis and manipulation assays suggest that a responsible circuit of hemoglobin signal in lactating females seems to be VNO-AOB-MeA-VMHd (SF1-positive cells)-PAGd. As it was pointed out, both male mice and virgin females showed VNO activity towards hemoglobin (Figure 3 and Revise figure 7a-b above), but there was no clear induction of c-Fos expression in higher brain regions such as the VMHd and PAG (Figure 4) and no behavioral outputs in male and virgin females (Figure 5d). It is possible that this difference was caused at downstream regions that are affected by the internal state such as lactating.
The results are somewhat consistent with those in the recent paper about non-volatile protein pheromone, darcin, suggesting that the circuit from the AOB to MeA can be responsible for the integration of pheromonal information, whereas the transmission process to downstream brain regions is affected by other factors such as the internal state (Demir et al. 11 ). We added discussion in the revised manuscript (page 9 line 20-27).
-What is the role for the main olfactory bulb in hemoglobin sensing in mothers? This is important especially since it has been shown that there is functional plasticity to behaviorally relevant odors in the main olfactory bulb of mothers (Vinograd et al., Cell Reports 2017).
As the reviewer mentioned, plasticity in the main olfactory system is relevant to odor-driven behaviors in mothers. However, as for specific behaviors observed in hemoglobin-stimulated mothers, the genetic deletion of Vmn2r88 impaired these outputs (Figure 5c and 6c), suggesting that only the vomeronasal olfactory system is involved in evoking hemoglobin-dependent behaviors and the main olfactory system seems to have a minor role in this context in lactating females. Nonetheless, a possible involvement of volatile odor signal through the main olfactory system remains to be elucidated.
- Figure 6 b and 6 c. What is the difference between the control groups in figure 6b (rearing time) and figure 6c? Similarly, what is the difference between the hb condition in figure 6b and hb +/+ condition in figure 6c? If these are different animals for the same conditions the authors should group them together to increase the statistical power.
In figure 6b, we used pure C57BL/6 mice purchased from Japan SLC or Japan CLEA to perform the open field assay. And then the result showing hemoglobin-dependent rearing enhancement led us to perform the additional assay with Vmn2r88 mutant mice to confirm the output could be controlled by hemoglobin receptor; Vmn2r88. Vmn2r88 knock-out mice are also C57BL/6 background but its origin was different from C57BL/6 mice from the vendors, so we made the control group for the assay using Vmn2r88-/-. Siblings of Vmn2r88 knock-out females (Vmn2r88+/+) were used in figure 6c. Thanks for pointing this out. We added the data of individual subjects in Figure 1a, b, c, d, e, h, and Supplementary figure 2a.
-In supplemental figure 7, mice spent similar time in a 2-chamber assay where hemoglobin was compared to the control. However, several concerns arise from this interpretation made by the authors. Most importantly, it seems that the mice tested in the assay were male mice and not mothers. This is a surprising choice for this assay, considering the fact that males did not show a significant behavioral difference in other assays, and other works imply that hemoglobin might be attractive to males (Isogai et al., Cell 2018).
In our 2-chamber assay (now in new Supplementary figure 8 in the revised manuscript), we used lactating females as test mice. We apologize that our explanation in the figure and text was not clear. We modified the figure legends and text to clearly mention that lactating mothers were used.
In summary this is a strong submission and our comments should be taken as constructive remarks to improve the paper and eliminate confusion and ambiguity for the readers. We congratulate the authors on their beautiful and impressive work.
Thank you very much for all the constructive comments and suggestions. We greatly appreciated all comments that made this study more convincing.

Reviewer #3 (Remarks to the Author):
In this paper, Osakada and colleagues -stimulated by initial observations that blood activates the VNO -identify the specific blood ligand responsible for this activation (beta-globin) and its corresponding VNO receptor (Vmn2r88). Using Vmn2r88-knockout mice, they demonstrate that smelling blood causes exploratory behavior (rearing and digging) in a receptor-dependent manner, and that this response is elicited in lactating mothers, but not virgin females or males.
The behavioral response is attributed to the specific activation of dorsal PAG, MeA and dorsal VMH (dVMH) in lactating mothers, and was recapitulated by optogenetic activation of SF1 neurons in dVMH. On the basis of these data, the authors claim that the smell of blood activates a specific behavioral response that is dependent on an AOB-MeApv-VMHd-PAGd pathway and unique to lactating mothers.
Overall, the experiments identifying beta-globin and Vmn2r88 (Figs 1 and 3) in this paper are elegant and convincing, and consistent with a recent paper (Isogai, 2018) that also found beta-globin and Vmn2r88 to mediate the sensing of blood VNO. The subsequent neural and behavioral analysis, however, could be improved to rule out alternative explanations for the authors claims. In particular, I am concerned that the data may reflect a general response to VNO activation that could include a much larger category of ligands than just beta-globin.
For example, the authors show that 2MT also elicits digging behavior when presented at very low concentration. In addition to this overall feedback, I have several concerns and questions that will clarify this manuscript: 1. To clarify the specific effects of hemoglobin, it would be helpful to see the same behavioral experiments repeated for other ligands, including some that would likely covary with hemoglobin (e.g. other infant-associated ligands), and others that are independent.
Thank you very much for careful read and important suggestions. To show the specificity of hemoglobin itself as a vomeronasal ligand for digging behavior, we added some control groups. Here, we used ESP1, one of the peptide pheromones received by another specific vomeronasal receptor (Vmn2r116, also known as V2Rp5) as a negative control (Haga et al. 12 ). ESP1 did not enhance digging behavior (Revise figure 10b below) nor rearing behavior (Revise figure 2b) in lactating mothers. We also added a fresh blood-stimulated group as a positive control into experiments as well (digging assay: Revise figure 10 below, pup retrieval assay: Revise figure 12 below). These results support that there is a specificity of hemoglobin's effect on digging behavior in lactating females. In addition, we performed the open field assay with ESP1 and 2MT stimulations (Revise figure 2). The results that both ESP1 and 2MT stimulation did not increase rearing behavior suggested that enhancement of rearing behavior in the open field assay is specifically derived by hemoglobin.

Revise figure 10
Hemoglobin and fresh blood but not ESP1 enhanced digging behavior of lactating female mice a Timeline for digging assay using lactating female mice and with cotton exposure swabbed with either ESP1 (20 µg), hemoglobin (Hb, 300 µg), fresh blood (30 µL), or control buffer. Pups were removed from the mother's cage 60 min prior to cotton exposure and behavior recording. b Quantification of the total digging time duration (sec) of lactating mothers with prior cotton exposure. n = 6-15. Error bars, S.E.M. *p<0.05 by the Wilcoxon signed-rank test.

Revise figure 2
Hemoglobin but not ESP1 and 2MT enhances rearing, a type of exploratory behavior, in lactating mothers.
a Schematic illustration of the timeline of the open field assay with cotton pre-exposure. Cotton exposure was performed in their home cage (with their pups). b Quantification of total distance, total center time, moving speed, and rearing time duration of lactating mothers, pre-stimulated with control buffer-, ESP1-, 2MT-(low 2MT: 10000-fold dilution, high 2MT: 10-fold dilution) or Hb-cotton swabs in the open field assay for 10 minutes. n = 8 for control, n =7 for Hb, and n =5 for ESP1, low 2MT, and high 2MT. Error bars, S.E.M. a Images from dual-color ISH staining of sections including the VMH from hemoglobin (Hb) or distilled water (control) -stimulated C57BL/6 mother mice labeled with the SF1 cRNA probe (green) and c-Fos cRNA probe (magenta). Scale bar, 100 µm. b Quantification of c-Fos-positive neurons overlapping with or without SF1 in the VMHd. The number of sections counted to determine the number of c-Fos-positive neurons in each brain area was 6. Error bars, S.E.M. n = 4. *p<0.05 by unpaired two-sided Student's t-test. 3. Figure 2d,e is used to show that the H78N mutation in beta-globin does not inhibit VNO activation. However, data are only shown for two mutant proteins and a no-odor control, but a Vmn2r88 ISH and pS6 immunostaining of VNO sections from mother and virgin female mice exposed to hemoglobin (Hb) or distilled water (control). Scale bar, 50 µm. b Quantification of visualized neurons in the VNO. 9 sections from each of 3 animals were quantified. The x-axis shows that the number of pS6-(green), Vmn2r88-(magenta), and double-(yellow) positive cells per VNO section. 5. If blood is placed on a pup, how does the mother behaviorally react?
We conducted a new pup retrieval assay with a slight modification. We painted fresh blood, hemoglobin, and control vehicle directly onto the back of pups just before the retrieval assay. The result showed that not only hemoglobin but fresh blood painting elicited delay in pup retrieval and digging enhancement of lactating females (Revise figure 12 below, new Supplementary figure 7 in the revised manuscript).

Revise figure 12
Pup retrieval assay with C57BL/6 lactating female mice. a Timeline for pup retrieval assay using lactating female mice and pups painted either hemoglobin (300 µg Hb), fresh blood (30 µL), or control buffer. Pups were removed from the female's cage 30 min prior to behavior recording. b Combined percentage of pups with Hb, fresh blood, or control buffer-painted on their back (out of three) retrieved by an animal group as a function of time. c Quantification of the digging time duration of lactating mothers in pup retrieval assays. n = 5 for control, n =7 for Hb, and n =6 for fresh blood.

Responses to reviewers' comments
We greatly thank all reviewers for the enthusiasm about our study, and for their constructive comments. We have performed one additional experiment and some modifications in response to the important comments from the reviewers, which have resulted in significant improvement of the paper. The revised manuscript is much improved with new data and clarification. However, I have one further concern about the new data. The loss of function experiment implicating the SF1 neurons in hemoglobin-evoked behavior is great (Figure 7c), but it is missing an important control. The authors express the DREADD, hM4D-mCherry, in these neurons and show the administration of CNO eliminates behaviors evoked by hemoglobin. However, a control group of mice expressing mCherry needs to be included. What if CNO alone causes the loss of the exploratory behaviors? This seems particularly important given the increasing reports of off target effects of CNO.

RESPONSE:
We thank the reviewer for the comment. We performed additional pharmacogenetic loss-of-function assays in the VMHd of wild type lactating females to make up for the missing control. Here we used the same experimental conditions originally mentioned in our method section, not using SF1-Cre transgenic female mice but wild type C57BL/6 female mice. In the VMHd of virus injected wild type female mice, there was no viral gene expression (no mCherry expression that represents the expression of DREADD-Gi encoded in the virus) in the VMHd (Revise figure 1d) and there was no significant decrease in total digging duration after CNO administration comparing with that of saline (Revise figure 1e). This data supports our original results that showed a significant decrease in digging behavior after CNO administration in SF1-Cre mothers infected with DREADD-Gi encoding virus (Revise figure 1b and 1c). Taken together, this behavioral change was not because of administration of CNO itself but of silencing the activity of specific neurons in lactating female mice. We put this new result in Supplementary Figure 10 of our updated manuscript.