Mast cells link immune sensing to antigen-avoidance behaviour

The physiological functions of mast cells remain largely an enigma. In the context of barrier damage, mast cells are integrated in type 2 immunity and, together with immunoglobulin E (IgE), promote allergic diseases. Allergic symptoms may, however, facilitate expulsion of allergens, toxins and parasites and trigger future antigen avoidance1–3. Here, we show that antigen-specific avoidance behaviour in inbred mice4,5 is critically dependent on mast cells; hence, we identify the immunological sensor cell linking antigen recognition to avoidance behaviour. Avoidance prevented antigen-driven adaptive, innate and mucosal immune activation and inflammation in the stomach and small intestine. Avoidance was IgE dependent, promoted by Th2 cytokines in the immunization phase and by IgE in the execution phase. Mucosal mast cells lining the stomach and small intestine rapidly sensed antigen ingestion. We interrogated potential signalling routes between mast cells and the brain using mutant mice, pharmacological inhibition, neural activity recordings and vagotomy. Inhibition of leukotriene synthesis impaired avoidance, but overall no single pathway interruption completely abrogated avoidance, indicating complex regulation. Collectively, the stage for antigen avoidance is set when adaptive immunity equips mast cells with IgE as a telltale of past immune responses. On subsequent antigen ingestion, mast cells signal termination of antigen intake. Prevention of immunopathology-causing, continuous and futile responses against per se innocuous antigens or of repeated ingestion of toxins through mast-cell-mediated antigen-avoidance behaviour may be an important arm of immunity.

mast cell-deficient mice. Accordingly, this paper should be cited as showing both the key role of IgE in mediating resistance to potentially lethal injections of bee venom and also for providing evidence concerning the importance of mast cells in mediating such IgE-dependent resistance. However, the current paper uses IgE responses to a commonly employed antigen (OVA) to provide evidence that such IgE-dependent sensing of a presumably innocuous food allergen (OVA) can promote aversive behavior. The evidence that this response was partially dependent on serotonin (via one type of serotonin receptor: 5-HTR3) is of interest. The use of sophisticated methods to reduce the development of spurious results, including special cages (IntelliCages) for housing the mice, is appropriate. Finally, the authors comment at the end of the Discussion on the potential involvement of GDF15 in this response (they indicate that they were in communication with R. Medzhitov during the preparation of their manuscript). In contrast to the Medzhitov manuscript, the Rodewald manuscript used OVA admixed with sucrose. Therefore, the two manuscripts reached somewhat similar conclusions using either OVA alone (Medzhitov group) or OVA plus sucrose (Rodewald group).
C. Data & methodology: validity of approach, quality of data, quality of presentation. In general, data & methodology were appropriate. The use of IntelliCages to ensure that fewer potential biases were introduced during the experiments was particularly appropriate.
D. The use of statistics and treatment of uncertainties seem appropriate. E. Conclusions: robustness, validity, reliability. The data shown appear to be quite robust. Furthermore, the use of basophil-deficient Mcpt8-Cre mice to show that basophils were not involved in the avoidance behavior was appropriate, given the fact that the Cpa3-Cre/+ mice have a 40% reduction in basophils (in addition to their striking deficiency in mast cells).
F. Suggested improvements: experiments, data for possible revision. 1) The next to last sentence needs revision. In lines 429-430, the authors write: "The present work identifies them as sensor cells of the nervous system,....". It isn't clear how mast cells, of hematopoietic origin, qualify as being "of the nervous system". The meaning of this sentence should be clarified.
Minor points: 1) Line 140-142: I suggest writing: "None of these assays distinguished mast cell-deficient mice from their wild type littermates, indicating that Cpa3 Cre/+ mice had no behavioral deficits measured by these assays that could have confounded the drink avoidance experiments." 2) Lines 163-164: I suggest simply omitting: ", which may be testimony to the close relationship of basophils and mast cells in some tissues,".
3) Lines 381-382: "In light of this overall debatable role of mast cells in immunological protection," seems to be too general a statement, particularly since refs. 43 & 44 showed that IgE (ref. 43 & 44) and mast cells (ref. 43) could confer increased survival to mice injected with potentially lethal amounts of bee venom. I suggest the following text instead: "Given the complexities in understanding the actual roles of mast cells in immunological protection, the role for mast cells observed here in antigen avoidance behavior is intriguing.". 4) Lines 427-428: It seems a bit odd to cite only a paper concerned with human mast cells (ref. 32) when this manuscript was concerned solely with mouse mast cells. I suggest also citing a reference to the distinctive position of mouse mast cells among mouse hematopoietic cells (Dwyer DF, Barrett NA, Austen KF; Immunological Genome Project Consortium. Expression profiling of constitutive mast cells reveals a unique identity within the immune system. Nat Immunol. 2016 Jul;17 (7):878-87.).
G. References: appropriate credit to previous work?
The key role of IgE in mediating acquired protection against the adverse effects, including death, of toxin exposure was reported in two of the cited papers: (ref. 43 [Immunity 39, 963-975, 2013] andref. 44 [Immunity 39, 976-985, 2013]). However, the former paper also showed, in Fig. 6F-H, that passive immunization with serum from bee venom-immunized wild type mice failed to increase the resistance against challenge with a potentially lethal dose of bee venom in two different types of mast cell-deficient mice. Accordingly, this paper should be cited as showing both the key role of IgE in mediating resistance to potentially lethal injections of bee venom and also for providing evidence concerning the importance of mast cells in mediating such IgE-dependent resistance.
H. Clarity and context: lucidity of abstract/summary, appropriateness of abstract, introduction and conclusions. The Abstract should be revised in line 36 to say: "Here we show that antigen-specific avoidance behavior in inbred mice....". The authors also reported (lines 131-132): "...that a minor component of the OVA avoidance response is immunization-dependent and mast cell-independent (Fig. 1d). This is reflective of the detailed and careful approach used by the authors in the analysis of this complex behavior.
Referee #4 (Remarks to the Author): Major 1. Mast Cell deficient mice may have secondary deficiencies; as such, to truly prove a role for mast cells, adoptive transfer of mast cells into the cpa3 MC KO strain should be performed. 2. The passive sensitization experiment revealed only modest effect, in a fraction of the mice. This raises the question of whether the mast cells were sufficiently sensitized. The experiments are missing the positive control (allergen-induced anaphylactic response) as well as assessment of mast cell degranulation, on a single cell level. This is important as the investigators need to prove the role of IgE, and their data suggests that IgE-mediated mast cell activation (which may happen in their experiment), is not sufficient for the avoidance behaviour.
3. The experiments with C57Bl/6 mice ( Figure 2), also need positive control (anaphylaxis) as per point above. 4. I have concerns about the experiments with the dorsal root ganglia. It is commendable that the authors indicate that the tracing experiments were derived from two mice; this is an insufficient number of mice to study, especially in view of the negative results. Related to this, the investigators, employed Wnt1|GCaMP3 engineered mice (plexus experiments); it remains unclear if similar negative results would be derived with other genetically engineered strains or best wild-type mice. This should be addressed. 5. The findings are very interesting, but the investigators are not providing a mechanism beyond mast cells. The link between mast cells and neurological pathways is purely speculative as they brain-axis experiments are largely negative. The single experiment with palonosetron does not prove peripheral afferent nerve involvement. Deeper experiments such as dose-response, use of additional inhibitors and/or KO mice, and ruling out off-target effects will be important. 6. The authors have not proven dependence on IgE.
Minor 1. Review of Figure 1 shows that some mice escape allergen avoidance. How is this explained? Do these mice exhibit intestinal pathology? 2. The investigators are implicating allergen activated stomach and intestinal mast cells, but not mast cells that reside in the oral cavity. They should also rule out a role for esophageal mast cells. 3. In view of the negative data in Figure 4, the authors are obliged to include additional controls. Please verify the KO at the protein level, including deficiency of histamine in the Hdc -/-mice. 4. Food allergic humans have 'ad-lib'diets, yet avoidance is not universal, such as in patients with eosinophilic esophagitis (that have abundant mastocytosis and food specific IgE). Please speculate on how this may occur. 5. The authors stated that they used the molecular signature of 200 inflammatory genes from the Libero et al. manuscript (cell systems, 2015). However, Liberzon et al. manuscript contains multiple data sets. Please explain from which dataset(s) the 200 genes were chosen and why these specific genes were chosen. 6. It is not clear why the genetic analysis of the 200 genes was performed only on the duodenum as the stomach, and small intestine demonstrated mast cell hyperplasia. Due to the differences between the organs, it is essential to perform the analysis on each organ separately. In addition, a full bulk RNA-sequencing will be more informative regarding the changes in the tissue, if any. 7. In the method-Immunization-the rationale for mixing the cytokines with their antibodies in one cocktail and injecting them into the mice is unclear. 8. Extended figure 8D-why did you stain for IgE? The lines in the text which refer to this figure are 346-347, which refer to mast cells and enteroendocrine cell 5-HTR expression, and IgE is not a marker for either of these. Please add the correct staining. 9. In extended figure 1, please add a legend for the grey and red marks. 10. Extended figure 3K-please make a readable table of the 200 genes you chose for the genetic analysis, as it is hard to read them in the heatmap. 11. Extended figure 3J-Why did only the duodenum in the representative figure as the stomach and small intestine demonstrate the mast cell hyperplasia.?

Point-by-point response to the reviewers' comments (response in italics)
We are grateful to the reviewers for their insightful and very helpful comments. We feel that the manuscript has been significantly improved as a result of the reviewers' queries which importantly led us to make additional experiments. In the submitted revision, we now include comprehensive new data as requested and as described in detail below.

Reviewer #1:
We thank the reviewer for his/her positive and helpful comments. Please find below and in the manuscript the requested additional information and the requested control data.
1-Can egg white 8% sucrose solution affect behavior and metabolism in mice model? (PMID 27126968) please explain it in text.
Humans and mice have an innate preference for sweet solutions (PMID 25815979). Therefore, the 8% sucrose affects the behavior in creating a preference of mice for the sweetened egg white solution. Indeed, this preference behavior is a central element of the experiments because it makes mice prefer the sweetened egg white solution to study avoidance. The sucrose concentration was adjusted for the inbred strains used (BALB/c and C57BL/6). Chronic consumption of sucrose over the span of several months can induce metabolic changes such as increased body weight, glucose intolerance, and adiposity (PMIDs 31255519, 15975159, 36229459, 35933635). We did not observe increased body weight in mice by the end of the IntelliCage experiments. This suggests that the short time frame of sugar consumption is insufficient to alter the metabolism of the mice.
We refer to this information in the Results (page 5, line 110): "The 8% sucrose affects the behavior in creating a preference of mice for the egg white water", and in Fig. 1  3-The authors must explain egg white is endotoxin free or not ? Please explain it.
We measured the endotoxin concentrations of 11 of our independent stock preparations of 20% egg white water that we used in our behavioral experiments from early 2020 to mid 2022 with the Limulus amebocyte lysate test. Endotoxin concentrations of the egg white solutions were on average 0.78 ± 0.17 EU/ml (see Figure 1 for reviewer below). This endotoxin level is considerably lower than the reported endotoxin concentration in standard mouse chow (approximately 20 EU/ g; PMID 18990206), or the median endotoxin concentration of commercially available pasteurized milk (102.5 EU/ml;PMID 25527628

4-The authors must explain cfos expression in their expression.
To test mast cell-specific and allergen-induced cFos-expression in the central nucleus of amygdala (CeA), paraventricular nucleus (PVN), parabrachial nucleus (PBN), and nucleus tractus solitarius (NTS) of the brain, we immunized wild type (+/+) and mast cell-deficient (Cre/+) mice according to our standard regimen (Figure 2a for reviewer below). On day 21 mice were only given access to 20% egg white water containing 8% sucrose solution in place of their normal water bottle (Figure 2a for reviewer below). After three hours of voluntary drinking, mice were sacrificed, and the brains were analyzed for cFos-expression by immunohistochemistry. Mice were not given a choice in this experiment, and we observed reduced and variable egg white water intake in wild type mice compared to mast cell-deficient mice ( Figure 2b for reviewer below) which indicates initial antigen avoidance. At this time point (3h) and under these conditions, we did not observe increased cFos-expression in wildtype mice compared to mast cell-deficient mice (Figure 2c-f for reviewer below). In fact, the number of cFos+ neurons in the CeA was increased in mast cell-deficient compared to wild type mice (see Figure 2c for reviewer below). Mast cell-deficient mice which consumed more egg-white sugar solution may have increased palatability (sugar and protein) signaling (PMID 22815514). In any case, these experiments did not further illuminate cFos signals in brain areas associated with avoidance. Fig. 2: cFos-expression in the brain after antigen consumption a, Type 2 immunization scheme, and experimental timeline for analysis of cFos induction. b, On day 21 mice were given 20% egg white water containing 8% sucrose in place of their regular drinking bottle, and total egg white water intake of OVA-alum immunized BALB/c Cpa3+/+ and Cpa3Cre/+ mice was recorded after three hours. c, d, e, f, Mice were perfused and the brains were isolated and processed for immunohistochemistry staining with an antibody specific for cFos. Shown is the quantification of cFos-positive neurons in the central nucleus of the amygdala (CeA) (d), the paraventricular nucleus (PVN) (e), the parabrachial nucleus (PBN) (f), and the nucleus tractus solitarius (NTS) (g). The bars represent the mean values, and each dot is a single mouse. Statistical analysis was performed using students T-test (b-g). The exact P values are shown.
5-It would be proper to say that vagal and trpv1 neurons are not involved in this pathway since the signal generated by mast cells can reach the brain via both nerves and secreted mediators via systemic circulation. Therefore, the comment on line 314 should be reorganized.
We apologize for lack of clarity here. We have reorganized this text to read (page 14, line 375): "The fast avoidance reaction of some immunized mice (29% of wild type mice within the first day, and some immediately after their first licks) (Extended Data Fig. 9a Fig. 10i, j). None of these experiments revealed evidence for a neuronal route transmitting the avoidance signal (Extended Data Fig. 10b-j). However, since Trpv1-expressing (RTX-sensitive) neurons represent only a subset of all dorsal root ganglion neurons, a function for dorsal root ganglion neurons, that are insensitive to RTX ablation, in signaling antigen avoidance cannot be ruled out." 6-The author must re-draw figure 4f. Please use molecules and receptors from this finding of manuscript.

, b) could be due to rapidly acting humoral factors in the blood circulation or to direct signaling from mast cells to neurons. […] Avoidance signaling via extrinsic vagal neurons was assessed by subdiaphragmatic vagotomy of immunized mice (Extended Data 10g). In addition, we tested the effect of Resiniferatoxin (RTX)-mediated depletion of extrinsic Trpv1-expressing vagal-and dorsal root ganglion sensory neurons on antigen avoidance (Extended Data
We have redrawn this figure (now Figure 5), and indicate, among other changes, IgE and 5lipoxygenase activating protein (FLAP). We also added our new findings on immune activation and inflammation under non-avoidance conditions.

7-L144 Which peptide specific is antigen specific IgE ?
The IgE-antibody used for passive sensitization is the clone E-C1 (also known as OE-1) purchased from Chondrex Inc. This antibody has been raised against full length OVA protein, and possibly recognizes a repetitive epitope (https://www.chondrex.com/documents/3006-3008_IgE_Animal_Model.pdf). To our knowledge, the epitope has not been mapped.
8-FIG2. Separate figures should be prepared for experiments with 1% and 0.25% sucrose in FIG2d. The addition of alum control would be welcome for FIGa,b, and c. In addition, the pvalue between +/+ and Cre/+ should be given in the experiment with 0.25% sucrose and it is necessary to prove that the experiment works as in the BALB/c mouse in FIG2d.
The revised data from former Fig. 2 Fig. 3a-f. We refer to these experiments in the results (page 8, line 213):"As in BALB/c mice (Fig. 1c-e), antigen avoidance was not observed in alum only immunized C57BL/6 mice (Extended Data Fig. 3a, b, d, e)."...."This treatment further enhanced the avoidance response which, however, remained mast cell-and OVA immunization-dependent (Extended Data Fig. 3c, f)." 9-FIG3c-The authors propose that Nr4a1-GFP mice have increased numbers of mast cells in the stomach 3 hours after 25% OVA exposure. However, in FIG3c the p-value is not significant for stomach delta Nr4a1-GFP. Figure 3c showed the difference (delta) in Nr4a1-GFP expression in tissue mast cells of OVA-alum or alum immunized mice challenged by OVA or BSA. We chose this comparison for normalization because we noticed differences in baseline GFP fluorescence in mast cells derived from different tissues. We now show mean fluorescence intensities for mast cells from each organ. p-value for stomach mast cells is significant (Figure 4c).

The original
10-FIG4e -Alum control should be added to the experiment setup and should be shown in the figure.
We repeated this experiment, and included alum controls (Extended Data Fig. 8g). We initially observed a partial response of OVA-alum immunized mice towards higher egg white water consumption after palonosetron treatment compared to vehicle controls. To more definitively clarify the role of 5-HTR3a in antigen avoidance behavior, we consulted with statisticians, increased the number of tested mice to n = 21 for vehicle and n = 23 for palonosetron, and included the requested alum controls in these new experiments (Extended Data Fig. 8g). The data do not confirm an increased egg white water consumption after 5-HTR3a blockade. In addition, palonosetron-treatment altered the egg white water preference of alum immunized mice (Extended Data Fig. 8g), suggesting complex involvement of 5-HTR3a in antigen avoidance.
In summary, it is unlikely that 5-HTR3a has a major role in antigen-avoidance behavior. For clarity, we have removed the mast cell-deficient groups from the figure (Extended Data Fig. 8g), and have changed the text on page 13, line 365 to read: "Hence, stomach mast cells are loaded with 5-HT which upon release may signal via its ionotropic serotonin 3 receptor (5hydroxytryptamine receptor 3; 5-HTR3), which is a key regulator of visceral malaise, nausea and emesis. We tested the role of 5-HTR3 in antigen avoidance by treatment of mice with the specific inhibitor palonosetron. Palonosetron did not significantly decrease antigen avoidance behavior (Extended Data Fig. 8g)."

11-Demonstrating the intracellular calcium increase in intestinal cells would be good in terms of showing direct activation.
We thank the reviewer for this interesting suggestion. In new experiments we attempted to measure, as proposed, the intracellular calcium increase in intestinal mast cells. The frequency of mast cells in the intestine epithelial cell fraction is much lower than in the stomach, therefore it was necessary to first enrich mast cells by Percoll gradient centrifugation. The subsequent protocol was performed as described for stomach mast cells (Figure 4d, e). However, such isolated intestinal mast cells did not increases intracellular calcium levels during antigenstimulation, albeit ionomycin-induced calcium increases were apparent (see Figure 3 for reviewer below). Thus, we cannot distinguish whether intestinal epithelial mast cells are unsensitive to ex-vivo antigen stimulation or whether the cells became refractory due to the gradient purification.

Reviewer #2
We thank the reviewer for his/her insightful comments. We addressed the key requests to demonstrate an inflammatory gut phenotype following continuous allergen intake, and to provide data on effectors downstream from mast cells. We now provide new data from comprehensive experiments on the first point ( Fig. 3; Extended Data Fig. 4; Extended Data Fig. 6, Extended Data Fig. 7), and experiments inhibiting leukotriene synthesis (Fig. 4).
The authors analyze the gut phenotypes of the mast cell-deficient animals and show that they are not different from non-sensitized control animals. They use this absence of inflammatory phenotype to argue for the protective role of avoidance behaviors. However, this conclusion is based on correlative observation of negative data. To make such a conclusion, the authors should first show that continued consumption of allergen-water (as in mast cell-deficient background) promotes gut inflammation.
We agree that this is a central aspect of this work that was only indirectly addressed in the first version. The reviewer asks whether continued consumption of allergen-water (as in mast celldeficient background) promotes gut inflammation. While we have done this (see below), this experiment is confounded by the possibility that the absence of mast cells may ameliorate the development of gut inflammation (PMID 14660743). Wild type mice, on the other hand, avoid continued consumption of allergen-water when given the choice, and drink less and variable amounts when not given the choice (see Figure 4a, b for reviewer below). A comparison of the antigen amounts for wild type mice given the choice, or treated with gavage, and mast cell deficient mice under continued consumption of allergen-water are shown in Figure 4c for reviewer below. This comparison shows that the amount of antigen given by gavage is modest, and in fact considerably lower than the amount of antigen consumed voluntarily by mast celldeficient mice. Fig. 4: Antigen doses in various experimental conditions. a, Type 2 immunization scheme, and experimental timeline for consumption of egg white water only. b, On day 21 mice were given 20% egg white water containing 8% sucrose in place of their regular drinking bottle, and total egg white water intake of immunized BALB/c Cpa3+/+ and Cpa3Cre/+ mice was recorded after three hours. Immunized wild type mice given only egg white water (allergenwater) immediately drank less than immunized mast cell-deficient mice. This resulted in a trend towards reduced OVA consumption (average = 0.5g difference) and a strongly increased standard deviation in wild type mice (0.708g) compared to mast cell-deficient mice (0.346g). c, Estimated average OVA intake of immunized BALB/c Cpa3+/+ and Cpa3Cre/+ mice per day over the course of the IntelliCage experiment (in Fig. 1). For comparison, OVA amounts given to immunized mice by gavage is shown. Avoiding wild type mice consumed between 1.5 and 36 mg OVA, and mast cell-deficient mice between 380 and 1340 mg OVA. We administered 50 mg OVA per gavage (8 times over 16 days).
Regarding the new experiments, we studied consequences of continued antigen uptake via oral gavage in wild type mice, and of continued consumption of egg white water in mast cell deficient mice (outlined in Extended Data Fig. 4a). To briefly summarize these experiments, we analyzed stomach and small intestine by flow cytometry for inflammatory cells (Extended Data Fig. 4d,e), and by RNA sequencing (Fig. 3;Extended Data Fig. 6;Extended Data Fig. 7, of tissue lysates which captured changes in gene expression in an unbiased manner, i.e. without restricting analyses to cell subsets. We also analyzed serum cytokines (Extended Data Fig. 4f, g). All of these approaches revealed broad immune activation and inflammation in immunized wild type mice. In mast cell-deficient mice that were drinking antigen voluntarily, we also found an induction of immune response genes, however, its magnitude was reduced and its kinetics altered (Extended Data Fig. 7). These data suggest that the immunological and inflammatory response that is prevented by mast cell-mediated antigen avoidance behavior is largely but not exclusively driven by mast cells. Nonetheless, the observed immune activation and the inflammatory phenotype argue in our view for the protective role of avoidance behaviors.
These experiments are described in a new paragraph "Antigen avoidance behavior prevents immune activation and inflammation" in the main text page 9, line 235. Immune activation in non-avoiding mast cell-deficient mice is presented in a new paragraph on page 11, line 301.
Finally, we note that the comparison of plain water (no antigen), free choice (only voluntary amount of antigen) and gavage (defined larger amount of antigen) was key to uncover gene expression profiles for each condition (Fig. 3). These data may also define a threshold for the amount of voluntary antigen uptake.
To support the publication of this work, this reviewer believes that the authors need to provide positive results: e.g., identification of effectors downstream of mast cells in promoting avoidance behaviors. (Fig. 4f) in antigen avoidance by pharmacologically inhibiting FLAP using the specific inhibitor MK-886. To this end, BALB/c wild type mice were alum-or OVA-alum immunized, and treated, one hour before the avoidance test, with MK-886. FLAP-inhibition significantly reduced avoidance. The results have been added to a paragraph "Antigen avoidance behavior depends on 5lipoxygenase-activating protein (FLAP)" beginning on page 13, line 336. The data are shown in Fig. 4g, h.

Reviewer #3:
We thank the reviewers for their very positive and insightful comments, the suggested text changes, and the helpful discussion on protective roles of IgE.
1) The next to last sentence needs revision. In lines 429-430, the authors write: "The present work identifies them as sensor cells of the nervous system,....". It isn't clear how mast cells, of hematopoietic origin, qualify as being "of the nervous system". The meaning of this sentence should be clarified.

We clarified this sentence to read (page 17, line 472): "The present work identifies them as sensor cells linking antigen-recognition elicited by type 2 immune responses to behavior.
Minor points: 1) Line 140-142: I suggest writing: "None of these assays distinguished mast cell-deficient mice from their wild type littermates, indicating that Cpa3 Cre/+ mice had no behavioral deficits measured by these assays that could have confounded the drink avoidance experiments." The sentence has been changed accordingly (page 6, line 150).
2) Lines 163-164: I suggest simply omitting: ", which may be testimony to the close relationship of basophils and mast cells in some tissues."

This part of the sentence has been deleted.
3) Lines 381-382: "In light of this overall debatable role of mast cells in immunological protection," seems to be too general a statement, particularly since refs. 43 & 44 showed that IgE (ref. 43 & 44) and mast cells (ref. 43) could confer increased survival to mice injected with potentially lethal amounts of bee venom. I suggest the following text instead: "Given the complexities in understanding the actual roles of mast cells in immunological protection, the role for mast cells observed here in antigen avoidance behavior is intriguing." The sentence has been changed as proposed (page16, line 427). 4) Lines 427-428: It seems a bit odd to cite only a paper concerned with human mast cells (ref. 32) when this manuscript was concerned solely with mouse mast cells. I suggest also citing a reference to the distinctive position of mouse mast cells among mouse hematopoietic cells (Dwyer DF, Barrett NA, Austen KF; Immunological Genome Project Consortium. Expression profiling of constitutive mast cells reveals a unique identity within the immune system. Nat Immunol. 2016 Jul;17 (7):878-87.).

We thank the reviewers for pointing out this shortcoming. The Dwyer et al. reference is now cited (page 17, line 471)
G. References: appropriate credit to previous work?
The key role of IgE in mediating acquired protection against the adverse effects, including death, of toxin exposure was reported in two of the cited papers: (ref. 43 [Immunity 39, 963-975, 2013] and ref. 44 [Immunity 39, 976-985, 2013]). However, the former paper also showed, in Fig. 6F-H, that passive immunization with serum from bee venom-immunized wild type mice failed to increase the resistance against challenge with a potentially lethal dose of bee venom in two different types of mast cell-deficient mice. Accordingly, this paper should be cited as showing both the key role of IgE in mediating resistance to potentially lethal injections of bee venom and also for providing evidence concerning the importance of mast cells in mediating such IgEdependent resistance.

We agree, of course, that appropriate credit is due to previous work, and thank the reviewers for referring to the important specific findings regarding protective roles of mast cells shown in ref 45 (Marichal et al Immunity 2013). We have revised this sentence (page 15, line 423), and hope that this specific findings for the role mast cells in ref 45 will satisfy the reviewers' request.
H. Clarity and context: lucidity of abstract/summary, appropriateness of abstract, introduction and conclusions. The Abstract should be revised in line 36 to say: "Here we show that antigen-specific avoidance behavior in inbred mice....".

Reviewer #4: We thank the reviewer for his/her excellent and insightful suggestion to clarify the role of IgE in antigen-avoidance, and to interrogate immunological changes in stomach RNA-expression following antigen intake. We have now made new experiments with IgE-deficient mice and
performed sequencing experiments to address these point directly as described below. Towards identifying a mechanism downstream of mast cells, we provide evidence for a role of leukotrienes, mediators known to be released from IgE-activated mast cells, in antigen avoidance behavior.
Major 1. Mast Cell deficient mice may have secondary deficiencies; as such, to truly prove a role for mast cells, adoptive transfer of mast cells into the cpa3 MC KO strain should be performed. Fig. 2a-h). (Fig. 2a-c), given that mast cells are the major effector cells for IgE in antigen-exposed tissues. (PMIDs 16127161, 8613053, and 12217411), and a lower than normal concentration of mast cells in intestines (PMID 24516385, 24416383, and 28264908). Second, the phenotypes of adoptively transferred mast cells have been reported to differs from the corresponding native mast cell populations (PMID 15771585, 25727288, and 23127755). Finally, adoptively transferred mast cells may undergo phenotypic changes that are not fully consistent with their anatomical sites (e.g. protease expression pattern) (PMID 11337367). In a recent review from Stephen J Galli, Nicolas Gaudenzio, and Mindy Tsai in Annu Rev Immunol ('Mast Cells in Inflammation and Disease: Recent Progress and Ongoing Concerns) (PMID 32340580) the authors summarized these pitfalls and raised a note of caution in the area of adoptive transfer of mast cells. We do share this conclusion, and have hence refrained from the suggested adoptive transfer of BMMC into Cpa3-Cre mice.

The reviewer requests adoptive transfer of mast cells into Cpa3-Cre mice. However, this approach suffers from several shortcomings: Adoptive transfer of in vitro-derived mast cells (BMMC) does not reconstitute a normal physiologic distribution of mast cells, resulting in a higher than normal concentration of mast cells in the stomach
2. The passive sensitization experiment revealed only modest effect, in a fraction of the mice. This raises the question of whether the mast cells were sufficiently sensitized. The experiments are missing the positive control (allergen-induced anaphylactic response) as well as assessment of mast cell degranulation, on a single cell level. This is important as the investigators need to prove the role of IgE, and their data suggests that IgE-mediated mast cell activation (which may happen in their experiment), is not sufficient for the avoidance behaviour.
The essential role of IgE has now been demonstrated (see response to point 6 below). We agree that passive transfer of IgE only partially led to gain of avoidance (Fig. 2f, g). The reviewer raises the interesting question whether mice sensitized for avoidance are also sensitized for anaphylaxis. We injected BALB/c mice with IgE monoclonal antibody E-C1 (as in Fig. 2f, g), and measured passive systemic anaphylaxis following antigen challenge by OVA gavage.
Compared to OVA-alum immunized BALB/c mice OVA gavage could not induce a temperature drop in passively IgE-sensitized mice (Fig. 2l). As requested by the reviewer, we also analyzed mast cell degranulation on a single cell level based CD63 expression (PMID 34233046). In line with the partial avoidance response, only half of the mice showed activation of stomach mast cells upon OVA contact (Fig. 2j), and none of the mice showed activation in small intestinal epithelial mast cells (Fig. 2k). We conclude that IgE transfer can partially induce avoidance behavior, however, immunization is required to attain full mast cell activation and avoidance.
We present and consider these findings in the results on page 8, line 195.
In response to this request, we immunized C57BL/6 mice (bearing the Nr4a1-GFP reporter) with OVA-alum as in the avoidance experiments, followed by anaphylaxis assay. Mice were challenged by OVA or BSA gavage, or by intravenous injection of OVA or BSA (Extended Data Fig. 3j) Fig. 3l) in mice receiving OVA-gavage compared to BSA-controls, albeit reaching statistical significance only for intestinal mast cells.

. Only intravenous injection of OVA induced a temperature drop (Extended Data Fig. 3m). Analysis of mast cell activation on the single cell level (GFP-expression) by flow cytometry revealed increased GFP expression of stomach (Extended Data Fig. 3k) and intestinal mast cells (Extended Data
We present these findings in the results on page 9, line 226. Hence, type2 immunization in C57BL/6 mice can lead to gastro-intestinal mast cell activation (Extended Data Fig. 3k, l) and avoidance (Extended Data Fig. 3b, e) without sensitization for anaphylaxis by gavage (Extended Data Fig. 3m). This separation was not evident in BALB/c mice (see point 2 above).
4. I have concerns about the experiments with the dorsal root ganglia. It is commendable that the authors indicate that the tracing experiments were derived from two mice; this is an insufficient number of mice to study, especially in view of the negative results. Related to this, the investigators, employed Wnt1|GCaMP3 engineered mice (plexus experiments); it remains unclear if similar negative results would be derived with other genetically engineered strains or best wild-type mice. This should be addressed.
We have now analyzed a total of n = 5 alum immunized Wnt1|GCaMP3 mice, and n = 5 OVAalum immunized Wnt1|GCaMP3 mice. Repetition confirmed the first data set. We also modified the way the data are displayed. Instead of snap shots of recorded ganglia, we now show representative calcium traces of individual neurons (Extended Data Fig. 10b, d). The new total number of recorded submucosal plexus neurons is n = 178 for alum, and n = 188 for OVAalum. The new number of recorded myenteric plexus neurons is n = 442 for alum, and n = 419 for OVA-alum. Information on numbers of mice and neurons are provided in the legend of Extended Data Fig.  10 (page 43, lines 1292).
Regarding the model, the Wnt1|GCaMP mouse is widely used and well established in studies of the enteric nervous system with the advantage of having broad expression of the calcium indicator in all enteric neurons throughout the GI tract. Some studies have used different promoters to drive GCaMP expression in targeted subpopulations of enteric neurons nitrergic neurons;cholinergic neurons). However, to our knowledge there are no reported differences between these genetically engineered strains (where GCaMP expression is driven by different promoters) that could impact gut physiology or potentially influence the results. We do report similar results using BALB/c wild type and mast cell-deficient mice when the calcium sensor GCaMP6 was transduced using a viral vector construct (AAV9-CaMKII-GCaMP6; Extended Data Fig. 10f). However, using this approach, the viral vector only transduced GCaMP6 expression in a fraction of enteric neurons (mostly myenteric plexus neurons). Again, this highlights the advantage of using the transgenic Wnt1|GCaMP mouse model for pan-neuronal expression of the reporter.
5. The findings are very interesting, but the investigators are not providing a mechanism beyond mast cells. The link between mast cells and neurological pathways is purely speculative as the brain-axis experiments are largely negative. The single experiment with palonosetron does not prove peripheral afferent nerve involvement. Deeper experiments such as dose-response, use of additional inhibitors and/or KO mice, and ruling out off-target effects will be important.
We thank the reviewer for his/her interest in our work, and we agree, of course, that a mechanism beyond mast cells is desirable. Towards a mechanism, we have performed several new experiments. First, we demonstrate that antigen-avoidance behavior is fully IgE-dependent (using IgE-deficient mice; see below under point 6). Regarding serotonin, we initially observed a partial response of OVA-alum immunized mice towards higher egg white water consumption after palonosetron treatment compared to vehicle controls. To more definitively clarify the role of 5-HTR3a in antigen avoidance behavior, we consulted with statisticians, increased the number of tested mice to n = 21 for vehicle and n = 23 for palonosetron, and included the requested alum controls in these new experiments (Extended Data Fig. 8g). The data do not confirm an increased egg white water consumption after 5-HTR3a blockade. In addition, palonosetron-treatment altered the egg white water preference of alum immunized mice (Extended Data Fig. 8g), suggesting complex involvement of 5-HTR3a in antigen avoidance. In summary, it is unlikely that 5-HTR3a has a major role in antigen-avoidance behavior. For clarity, we have removed the mast cell-deficient groups from the figure (Extended Data Fig. 8g), and have changed the text on page 13, line 361.
In further search for effectors, we addressed the role of lipid mediators known to be released from IgE-activated mast cells. We tested FLAP-dependent leukotrienes (Fig. 4f) in antigen avoidance by pharmacologically inhibiting FLAP using the specific inhibitor  To this end, BALB/c wild type mice were alum-or OVA-alum immunized, and treated, one hour before the avoidance test, with MK-886. FLAP-inhibition significantly reduced avoidance (Fig.  4g, h). We interpret these results cautiously, given that not all mice responded to the inhibitor, and given that the effect was mostly seen at early timepoints.
These experiments are presented in a new paragraph in the results section (page 13, line 336). We also discuss these findings (page 16, line 447) in the context of the work on GDF-15 by Florsheim et al. (PMID 36712030).
We acknowledge that our data regarding the link between mast cells and neurological pathways are negative. However, we still consider our comprehensive analyses useful. We excluded in KO mice roles for major mast cell mediators, including histamine and several proteases (including Mcpt6), which can directly stimulate sensory neurons (PMIDs 27793571;12388180;7810655). Moreover, we analyzed major branches of potential gut-brain signaling pathways (enteric neurons, vagal neurons, and RTX-sensitive neurons). Taken together, we and others may build on this negative data to identify the mast cell-and antigen-responsive neuron populations, if any.

The authors have not proven dependence on IgE.
We analyzed  mice, and found that antigen avoidance is critically dependent on IgE (new data in Fig. 2a-c). We describe these data in a new paragraph on page 7, line 178.
Minor 1. Review of Figure 1 shows that some mice escape allergen avoidance. How is this explained? Do these mice exhibit intestinal pathology?
As shown in Fig. 1c, all immunized wild type animals avoided antigen, albeit with variable kinetics. Therefore, we find no evidence for escape. Extended Data Fig. 4a;new data in Fig. 3;Extended Data Fig. 6;Extended Data Fig. 7,. We found broad immune activation and inflammation in immunized wild type mice under non-avoidance conditions (Fig. 3a,b,d,f). We also compared plain water (no antigen) to free choice (only voluntary amount of antigen), and identified mild gene expression changes that may define a threshold for the amount of voluntary antigen uptake (Fig. 3c, e). By contrast, the differences avoidance versus non-avoidance were drastic (Fig. 3a,  b, d, f).

The question regarding pathology is important. We studied immune activation and inflammation under conditions of avoidance versus non-avoidance (experiments outlined in
These experiments are described in a new paragraph "Antigen avoidance behavior prevents immune activation and inflammation" in the main text page 9, line 235.
2. The investigators are implicating allergen activated stomach and intestinal mast cells, but not mast cells that reside in the oral cavity. They should also rule out a role for esophageal mast cells.
To address this point, we prepared single cell suspensions from esophageal and gingival tissues of OVA-alum immunized mice. Tissues were analyzed by flow cytometry for CD45+ CD11b-MHCII-cells expressing Kit and Fc RI (i.e. mast cells). This revealed a small population of mast cells in the gingiva (0.51% of live CD45+ CD11b-MHCII-cells), whereas mast cells were undetectable in the esophagus of mice (see Fig. 5 for reviewer below). We now refer to esophagus on page X: "We also analyzed esophagus and colon but detected no (esophagus), or only minute numbers (colon) of mast cells which precluded their further analysis (not shown)." 3. In view of the negative data in Figure 4, the authors are obliged to include additional controls. Please verify the KO at the protein level, including deficiency of histamine in the Hdc -/-mice. Fig. 6 for reviewer below genomic PCR data from our own laboratory on all mutant mouse strains involved.
Lack of Cpa3 activity in Mc-cpa Y356L,E378A mice has been reported in Figure 3e of Schneider et al., 2007 (PMID 17923505). Western blot analysis proving lack of Cpa3 and Mcpt5 in Cpa3-/-mice has been reported in Figure 1d of Feyerabend et al., 2005 (PMID 15988029). Western blot analysis proving lack of Mcpt6 in Mcpt6-/-mice has been reported in Figure 2a of Shin et al., 2008 (PMID 18354212). Histamine deficiency in brain, skin, stomach, spleen, kidney and plasma of Hdc-/-mice is reported in table 1 of Ohtsu et al., 2001 (PMID 11478947). Lack of IgE antibodies in Igh7-/-mice is reported in Figure 2b of Oettgen et al., 1994 (PMID 8047141). Lack of basophils in Mcpt8-Cre animals has been reported in Figure 1b of Ohnmacht et al., 2010 (PMID 20817571). Lack of Mast cells in Cpa3Cre animals has been reported in Figure  1b of Feyerabend et al., 2011 (PMID 22101159).
4. Food allergic humans have 'ad-lib'diets, yet avoidance is not universal, such as in patients with eosinophilic esophagitis (that have abundant mastocytosis and food specific IgE). Please speculate on how this may occur.
In eosinophilic esophagitis (EoE) it is known that "food allergens (e.g., milk, egg, wheat, soy) are well-established disease triggers, with dietary elimination of specific food allergen triggers or elemental diet therapy resulting in disease remission in a majority of subjects" (PMID 36351516 We now refer to this analysis on page 11, line 293.
6. It is not clear why the genetic analysis of the 200 genes was performed only on the duodenum as the stomach, and small intestine demonstrated mast cell hyperplasia. Due to the differences between the organs, it is essential to perform the analysis on each organ separately. In addition, a full bulk RNA-sequencing will be more informative regarding the changes in the tissue, if any. Due to the differences between the organs, it is essential to perform the analysis on each organ separately.
We thank the reviewer for the suggestion to analyze, in addition to the small intestine, also the stomach by RNA-sequencing. This experiment yielded interesting new data (Fig.3;Extended Data Fig. 4,5,6,7;.
These experiments are described in a new paragraph "Antigen avoidance behavior prevents immune activation and inflammation" in the main text page 11, line 293.
7. In the method-Immunization-the rationale for mixing the cytokines with their antibodies in one cocktail and injecting them into the mice is unclear.
We apologize for the lack of clarity. We provide this information in the Methods (page 19, line 532): "Mixing IL-3 with the anti-IL-3 antibody MP2-8F8, and IL-4 with the anti-IL-4 antibody 11B11 generates cytokine-antibody complexes that display increased activity in vivo 1 , which we exploited here to increase the magnitude and duration of cytokine effects in vivo." 8. Extended figure 8D-why did you stain for IgE? The lines in the text which refer to this figure are 346-347, which refer to mast cells and enteroendocrine cell 5-HTR expression, and IgE is not a marker for either of these. Please add the correct staining.
As detailed under major point 5 above, we have no evidence for a role of serotonin. We have therefore omitted the previous Extended Data Figure 8D, E. 9. In extended figure 1, please add a legend for the grey and red marks.
We have added the missing legends to the graphs in Extended Data Fig. 1. 10. Extended figure 3K-please make a readable table of the 200 genes you chose for the genetic analysis, as it is hard to read them in the heatmap. Fig. 3k) has been removed. Based on our new RNA Seq data (Fig.3;Extended Data Fig. 4,5,6,7;Supplementary  11. Extended figure 3J-Why did only the duodenum in the representative figure as the stomach and small intestine demonstrate the mast cell hyperplasia.?

The former heatmap (Extended Data
We have removed the previous histological analyses in favor of the more comprehensive, sensitive and accurate tissue analysis by total lysate RNA Seq (Fig. 3;Extended Data Fig. 4,5,6,7;. In addition to the augmented mast cell gene signature (Fig. 3,  d, f, g), we analyzed mast cell hyperplasia by flow cytometry (Fig. 1h-j; Extended Data Fig. 2e,  f; Extended Data Fig. 3h, i). Figures 4C and 4D-Is there an increased level of Mast cells in the Cre/+ group in the passive sensitized group compared to the unsensitized groups?

Extended
We thank the reviewer for this question which prompted us to re-calculate intestinal mast cell numbers. We display numbers now as percentages which is the more robust analysis that bypasses poorly controllable factors such as the efficiency of tissue digestion. Of note, for the stomach this was not necessary because the release of mast cells from the epithelium is robust. Regarding the specific reviewer question, we found no significant differences in intraepithelial stomach or intestinal mast cells comparing in passive sensitized Cre/+ mice compared to the unsensitized (alum) Cre/+ mice (Cre/+ mice data are directly compared in Figure 7 for reviewer below; corresponding data in the paper are shown in Fig. 1h, i; Fig. 2h, i). 13. Figure 4C-indicate which color is what.
The graphs describing the average daily egg white water intake in the tested mouse mutants have been moved to Extended Data Fig. 8a-d, and all panels contain detailed group descriptions below the x-axis.