Microbial colonization induces histone acetylation critical for inherited gut-germline-neural signaling

The gut-neural axis plays a critical role in the control of several physiological processes, including the communication of signals from the microbiome to the nervous system, which affects learning, memory, and behavior. However, the pathways involved in gut-neural signaling of gut-governed behaviors remain unclear. We found that the intestinal distension caused by the bacterium Pseudomonas aeruginosa induces histone H4 Lys8 acetylation (H4K8ac) in the germline of Caenorhabditis elegans, which is required for both a bacterial aversion behavior and its transmission to the next generation. We show that induction of H4K8ac in the germline is essential for bacterial aversion and that a 14-3-3 chaperone protein family member, PAR-5, is required for H4K8ac. Our findings highlight a role for H4K8ac in the germline not only in the intergenerational transmission of pathogen avoidance but also in the transmission of pathogenic cues that travel through the gut-neural axis to control the aversive behavior.

The ChIP-MS assay suggests that during P. aeruginosa infection, PAR-5 is directly bound to histone. To further substantiate this result, we validated the binding using coimmunoprecipitation (New S3A Fig) and in vivo binding (New 3A, 3B, 3C, and 3D). Whole animal immunofluorescence IF staining also reveals that induction of H4K8ac by P. aeruginosa infection is dependent on par-5 RNAi (New Fig 3F). Furthermore, new additional studies suggested by the reviewer (comment 2 below) strengthened the idea that PAR-5 controls the levels of H4K8 acetylation. More specifically, germline specific RNAi of par-5 revealed not only disruption in pathogen avoidance but also H4K8ac (New S5 and New S6).
COMMENT: 2) Using glp-1 completely gets rid of germline so there could be confounding factors that affect pathogen avoidance, which are known. Why not use germline restricted RNAi or other germ cell manipulations (or gonad loss mutants) to strengthen this section? Although glp-1 mutants can indeed reduce the number of germ cells, they do not abolish all germ cells (it is incomplete).

RESPONSE:
It is standard in the field to use glp-1 animals. Indeed, the work by Dr. Murphy that the reviewer mentioned linked the germline and pathogen avoidance using glp-1 animals [Kaletsky, et al. "C. elegans interprets bacterial non-coding RNAs to learn pathogenic avoidance." Nature vol. 586,7829 (2020): 445-451. doi:10.1038/s41586-020-2699-5]. Our data indicate that glp-1 animals, even though they have some residual germ cells, fail to avoid Pseudomonas. This indicates that the germ cells lacking in glp-1 are required for pathogen avoidance. The manuscript reads: "To study a potential role of the germline in gut-neural signaling, we used glp-1 animals, which lack most germline cells due to defects in mitotic and meiotic division [27,28]." As requested by the reviewer, we have used germline restricted RNAi to confirm that PAR-5 functions in the germline to control pathogen avoidance (Fig. 3H). We have also included new additional experiments using tissue-specific RNAi strains, whole animal IF staining, and western blot assays to further validate the role of PAR-5 mediated acetylation of H48K in the germline (New S5 and New S6). The revised manuscript reads: "Consistent with this idea, par-5 RNAi in the germline, but not in the intestine or in neurons, significantly suppressed the P. aeruginosa-induced H4K8ac (S5 Fig). Whole animal fluorescent immunohistochemistry confirmed that P. aeruginosa-induced H4K8ac is inhibited by par-5 RNAi in the germline (S6 Fig)." COMMENT: Intriguingly, glp-1 mutants appear to have a low-level of H4K8ac ( Figure 2A) with E. coli that does not increase with PA14?

RESPONSE:
We are not surprised to see H4K8ac in glp-1 animals as it should not be restricted to the germline because histone acetylation is required for several physiological processes in different tissues. fer-1 animals also have some level of basal H4K8ac that are not seen in Fig. 2A because of the differences in exposure and total amount of proteins. The quantification of the multiple experiments that takes into account the loading control (S2A Fig) shows this more clearly.
As noted by the reviewer, disruption of proper development of germline in glp-1 animals resulted in lack of P. aeruginosa-induced H4K8ac, which is indeed the core of this study.

COMMENT:
3) The identification of PAR-5 is not clear. It was identified by ChIP-ms? More details are needed here as I assume this was done with the H4K8ac antibody, but it isn't clear why this would directly pull-down PAR-

RESPONSE:
We hope the revised manuscript makes it clearer that PAR-5 was identified by ChIP-ms: "To identify potential interacting partners that may affect H4K8ac in response to P. aeruginosa colonization, we performed chromatin immunoprecipitation-mass spectrometry (ChIP-MS). A total of 25 H4K8 acetylated-interacting candidate proteins that were upregulated more than 3-fold in infected animals were identified (S1 Table). We decided to further study PAR-5 because out of all the H4K8 acetylated-interacting candidate proteins that are upregulated more than 3-fold by P. aeruginosa infection, it is the only one that is expressed in the germline and in neurons, from where it could also be involved in the control of pathogen avoidance. Another reason why we focused on PAR-5 is that it belongs to a 14-3-3 family of chaperones [31,32] that, through interactions with different proteins, can regulate PTM such as H4K8ac." The revised Method section mentions the anti-H4K8ac (ab15823, Abcam) antibody that was used to co-precipitate proteins that are bound to the specific histone. The details of cross-linking and precipitation are described in Methods.
As requested by the reviewer, we have performed additional co-immunoprecipitations that confirm the physical binding of PAR-5 with H4K8 (New S3A Fig). We also performed additional experiments to study potential nuclear interaction in vivo. We used bimolecular fluorescence complementation (BiFC) that is used to determine the physical interactions of proteins in living cells. The new study shows that there is in vivo physical interaction between PAR-5 and HIS-67, which is an ortholog of human H4 (New Fig 3A-D). Knockdown of par-5 by RNAi resulted in the significant reduction of fluorescence (New Fig 3A-C), further confirming that the presence of the two proteins is required for the GFP reconstitution. The revised manuscript reads: "We confirmed the direct binding of .
We also confirmed the protein-protein interaction in vivo using bimolecular fluorescence complementation (BiFC), which allows for the determination of physical interactions of proteins in living cells through direct visualization [33]. The BiFC constructs were engineered to individually express, under the control of the heat shock promoter , GFP protein fragments translationally fused with PAR-5 and H4, which is a C. elegans ortholog of human H4. The interaction between the two proteins would bring the non-fluorescent fragments into close proximity for reconstitution and fluorescence. Twelve hours after heat shock, we observed fluorescence, indicating a physical interaction between PAR-5 and H4 in vivo ( Fig 3A). Animals carrying BiFC constructs without H4 did not exhibit fluorescence. Knockdown of par-5 by RNAi resulted in a significant reduction of fluorescence ( Fig 3B and 3C), further confirming that the presence of the two proteins is required for the GFP reconstitution. As shown in Fig 3A and 3D, the protein interaction occurs in the nuclei of hsp-16.41-expressing cells." We do not agree with the reviewer's statement that PAR-5 is a cytoplasmic chaperone. While the first line of par-5 Overview in Wormbase (https://wormbase.org/species/c_elegans/gene/WBGene00003920#01e6-9gc4af3bd-10) reads "par-5 (abnormal embryonic PARtitioning of cytoplasm)," the same description reads "Localizes to cell cortex and nucleus." More importantly, published studies have shown PAR-5 nuclear localization (figure on the left). We agree that PAR-5 might affect H4K8ac indirectly, which should not detract from the many findings of this study (1-P. aeruginosa infection induces H4K8ac, 2-the germline is required for the Fig 1E from Berdichevsky et al. Cell, 125, 6, P1165-1177, JUNE 13, 2006. To see whether 14-3-3 proteins are present in C. elegans nuclei, we performed cell fractionation experiments…14-3-3 proteins were present in both nuclear and cytosolic fractions ( Figure 1E). These observations are consistent with the hypothesis that the interaction between SIR-2.1 and 14-3-3 proteins occurs in the nucleus. pathogen avoidance induced by intestinal bloating, and 3-intact germline and PAR-5 are required for the transmission of pathogen avoidance to the next generation). The new studies we performed to address the reviewer's critique shows PAR-5 nuclear localization and histone binding (New 3A-D, New S3A Fig). Additional studies provide further support to the role of PAR-5 on H4K8ac (3F, S5, and S6). Thus, as stated in the revised manuscript, we think that "Even though PAR-5 is required for development and its inhibition may have wide effects on the germline that might indirectly affect H4 acetylation, our results indicate that PAR-5 directly interacts with histone." COMMENT: 4) Based on the IF results, there appears to be some somatic cells with H4K8ac marks? Could the authors look at the levels of H4K8ac marks either through WB or IF for the tissue specific RNAi.

RESPONSE:
We performed the suggested experiments using both WB and IF. As shown in New S5 Fig and S6 Fig, only par-5 RNAi in the germline significantly reduced H4K8ac. The revised manuscript reads: "As shown in Fig 3H,

par-5 RNAi in the germline significantly reduced pathogen avoidance, which is consistent with our previous results and suggests that H4K8ac occurs in the germline in response to infection. Consistent with this idea, par-5 RNAi in the germline, but not in the intestine or in neurons, significantly suppressed the P. aeruginosa-induced H4K8ac (S5 Fig). Whole animal fluorescent immunohistochemistry confirmed that P. aeruginosa-induced H4K8ac is inhibited by par-5 RNAi in the germline (S6 Fig)."
COMMENT: 5) The connection to pathogen avoidance is the weakest as the germline has been shown to impact this response. The authors could look at H4K8ac levels in ftt-2 knockdown worms too since they share so much homology with PAR-5 but don't affect avoidance.

RESPONSE:
As the reviewer noted, the germline has been shown to impact pathogen avoidance. We do not have reason to believe that ftt-2 knockdown will affect H4K8ac. Our studies show that par-5 RNAi does not reduce the levels of FTT-2 (S4B and S4C Fig). The results of ftt-2 knockdown would be difficult to interpret without studying its potential effect on par-5 and the study of ftt-2 is outside the scope of this work. The several aforementioned experiments we performed to address the previous criticism provide further support to the role of the germline in pathogen avoidance and H4K8ac (New Fig 3H, New S5 Fig, and

RESPONSE:
We thank the reviewer for this suggestion. We have included the temperature and time conditions in each figure's legend.

COMMENT: 8) The authors include a whole section where they introduce PAR-5 and how 14-3-3 chaperones
are involved with PTMs such as H4K8ac without citations.

RESPONSE:
We apologize for the oversight and we thank the reviewer for pointing out these insufficiencies. We have included relevant citation/s in the revised manuscript.
COMMENT: 9) Similarly, the transgenerational marks for H4K8ac are cool, but they should reference work that has already shown transgenerational inheritance of pathogen avoidance and specifically Coleen Murphy's work as possible future targets of study or H4K8ac inheritance.

RESPONSE:
We want to clarify that unlike colonization of the gut by live replicating P. aeruginosa, the exposure to RNA extracted from the pathogen does not induce intestinal bloating. This is not surprising as bloating requires intact, live replicating bacteria. Moreover, the RNA-induced avoidance accounts for only a small portion of the P. aeruginosa-elicited avoidance (25%, left figure). We had an extensive discussion with Dr. Murphy, who agrees that our works are not sufficiently related and stated "as opposed to our work on small RNA-induced learning and transgenerational inheritance of behavior, which seems orthogonal." Dr. Murphy also stated the following: "The focus of our paper is on the small RNA-driven part of the response" (which as shown in the work by Murphy and the figure on the left, only accounts for a fraction of the total avoidance) "I am also not sure what we would say about your study, since we also see no evidence of intestinal bloating in our experiments." Indeed, Dr. Murphy herself did not address our extensive work on bloating [Dev Cell. 2019 andElife. 2019] in her study.
In deference of the reviewer's comment, we have discussed the work by the Murphy lab in the new Conclusions section: "The inheritance of avoidance elicited by small RNAs from P. aeruginosa requires the germline [35,36]. We do not know whether H4K8ac plays a role in the avoidance mediated by small RNAs, which accounts for a fraction of the avoidance elicited by P. aeruginosa." Reviewer #2 COMMENT: Although many bacterial infections are restricted to the intestine, there is increasing evidence that infection causes signaling between different tissues. C. elegans is a powerful system to study bacterial infections, pathogen avoidance, and effects that are passed down between generations. In this manuscript the authors investigate the connection between the intestine and germline, and how this affects animal behavior. The authors find that infection with Pseudomonas aeruginosa induces H4Kac methylation in the germline. This methylation also occurs by knocking down several genes that cause intestinal distension. These RNAi conditions also induce pathogen avoidance which is dependent on the germline. The authors also identify PAR-5 as interacting with Occupancy index of N2 animals after 24-hour incubation on P. aeruginosa PA14 lawns or E. coli OP50 lawns supplemented with isolated total RNA from E. coli (E.c. RNA), P. aeruginosa (P.a. RNA), or E. faecalis (E.f. RNA).

RESPONSE:
We thank the reviewer for a very good summary and support to our study.
COMMENT: 1) The authors claim that "results demonstrate that enhanced H4K8ac in the germline is required for the transgenerational pathogen avoidance induced by bloating caused by bacterial colonization of the intestine" and claim in 4c that PAR-5 is required for this transgenerational effect. The authors do not show that animals that lack H4K8ac in the germline generate progeny that are defective for bacterial avoidance. They also don't show that PAR-5 is necessary for this transgenerational effect. Although this experiment is not possible with par-5 RNAi, the authors should either conditionally deplete PAR-5 (such as with auxin inducible degradation) in the P0s, or reword the text and figure to remove these claims.

RESPONSE:
The reviewer raised a good point about the limitation of our study. As suggested, we have removed the claims and used the term 'intergenerational' instead of 'transgenerational.'

COMMENT:
2) The convention in the field is to only use "transgenerational" to refer to effects that are passed down at least three generations (Perez and Lehner nature cell biology 2019). Effects that are only shown to be passed down a single generation are referred to as "intergenerational". Although the effect shown in this manuscript may indeed be transgenerational, the authors have not shown this. The authors should either test how many generations this effect lasts or change the text to clarify that it may be either intergenerational or transgenerational.

RESPONSE:
Thank you again for highlighting this issue. We have used the term "intergenerational" in the revised manuscript.
COMMENT: 3) Insert in the following sentence "not" after "did": "As shown in Fig 2B, inhibition of aex-5 and eat-2 did elicit pathogen avoidance in glp-1 animals."

RESPONSE:
We apologize for this error that has been corrected. Figure 1C and 3A, outlines of the germline are necessary as it is hard to know where the germline is in the current fluorescent images.

RESPONSE:
To address the reviewer's critique, we performed additional experiments and took better quality images. Also, we included close-up images of the germline region and included outlines (new Fig 1C). We have included differential interference contrast (DIC) representative images (new Fig 1C) and also new images that depict the entire body of the animal (new Fig  1D). We replaced the images in Fig 3A with better quality images from the new experiments (new Fig 3F). We decided to apply outlines only to the close-up images because we believe the outlines on other figures reduce the overall quality (below).