The sugar-responsive enteroendocrine neuropeptide F regulates lipid metabolism through glucagon-like and insulin-like hormones in Drosophila melanogaster

The enteroendocrine cell (EEC)-derived incretins play a pivotal role in regulating the secretion of glucagon and insulins in mammals. Although glucagon-like and insulin-like hormones have been found across animal phyla, incretin-like EEC-derived hormones have not yet been characterised in invertebrates. Here, we show that the midgut-derived hormone, neuropeptide F (NPF), acts as the sugar-responsive, incretin-like hormone in the fruit fly, Drosophila melanogaster. Secreted NPF is received by NPF receptor in the corpora cardiaca and in insulin-producing cells. NPF-NPFR signalling resulted in the suppression of the glucagon-like hormone production and the enhancement of the insulin-like peptide secretion, eventually promoting lipid anabolism. Similar to the loss of incretin function in mammals, loss of midgut NPF led to significant metabolic dysfunction, accompanied by lipodystrophy, hyperphagia, and hypoglycaemia. These results suggest that enteroendocrine hormones regulate sugar-dependent metabolism through glucagon-like and insulin-like hormones not only in mammals but also in insects.

This is an interesting study linking the function of enteroendocrine cells (EECs) with general energy homeostasis in the Drosophila model. The authors propose that NPF (the invertebrate NPY homolog) produced by EECs serves as an "incretin-like" hormone, controlling both the production of AKH (a glucagon-like hormone) and insulin-like peptides, therefore modulating fat and sugar homeostasis in adult flies. The study is rather complete and presents numerous experimental approaches to demonstrate these functional links. However, several major links are missing, among which the demonstration that NPF produced by the EECs is secreted and that NPF derived from EECs, and not from other sources, modulates NPFR signaling in the corpora cardiacca and the Insulin-producing cells (IPCs). Experimental evidence for these two points are mandatory to support the conclusions of the paper.
Major remarks: 1-Fig2a: if the metabolic changes in TKg>NPFi are not significant, the conclusion should be differently written (the "trend" has no significance, then), and the data should not be part of a main figure. Line 130, it is improper to write "…were also…", since it suggests that the previous data demonstrates a change in regulation, which is not the case. What is the statistical significance of the data presented in Fig2b?
2-Fig3: accumulation of NPF in the EECs upon starvation suggests a defect in secretion, but is certainly not a proof. This is an important caveat of the manuscript and the authors should measure circulating levels of NPF in starved vs control food conditions. This should also be done in TKg>NPF-RNAi conditions to ensure that NPF production by the EECs majorly contributes to its circulating levels in the hemolymph. This is particularly important since NPF is also produced by specific neurons in the brain, which could equally contribute to its circulating levels. In that respect, the authors should also assess the accumulation of NPF in brain neurons upon starvation and measure circulating levels after knock-down of NPF in these neurons in fed conditions. 3-Experiments using Sut1 knock-down suggest that this transporter is required for normal sugar/fat homeostasis. However, there is no evidence that Sut1 transports glucose into cells, as hypothesized by the authors. This should be assessed experimentally. TKg>Sut1-RNAi only partly mimics the effect of starvation on NPF accumulation. Possible explanation and their experimental testing should be provided. In fact, according to fig. 3h, the difference between sut1+/-and sut1-/-is not significant, therefore questioning the result obtained with the unique sut1-RNAi line used. ): why is the co-staining between GFP and AKH only partial if GFP expression is targeted to the CCs? Suppl. Fig.4f: the absence of direct connection between NPF+ neurons and the CCs does not prove that these neurons cannot communicate with the CCs, since they could contribute to circulating PDF. This should be modified in the text and tested experimentally (see point 2). 5-p7 line 261 "…suggesting that…". This conclusion seems wrong. Bmm expression is upregulated by Akh>NPFRi, but not dHSL expression. This suggests that dHSL function is not controlled (at least transcriptionally) by NPFR signaling in the CC. Silencing dHSL expression in fat cells could rescue NPF loss-of-function starvation sensitivity by a parallel mechanism, without implying that dHSL function relies on NPFR.
6-p12 line 289: The functional link between NPFR in the IPCs, and NPF produced from the EECs is not established. Therefore, such conclusion should not be made. Again, NPF could be secreted from the central brain and activate NPFR in the IPCs. 7-Fig7d is not convincing. It is difficult to imagine that such a mild reduction in IPC neuronal activity could account for a notable difference in Dilp secretion.
8-Fig7e lacks quantification and confirmation with another marker for insulin signaling. As such this piece of data is not significant. 9-Throughout their manuscript, the authors present a set of related phenotypes due to EEC-specific NPF knockdown consisting in: resistance to starvation, reduced lipid accumulation, increased feeding and reduced glycemia. Experimental evidence suggest that these effects are due to NPF sending metabolic signals to the CC and the IPCs, AKH and Dilps hormones acting then as relays. However, the phenotypes observed after silencing NPFR in the CC and IPCs are not consistent with this idea and it is unclear what contribution NPFR in the IPCs and the CCs brings to the NPF knockdown phenotype. Strikingly, Dilp2>NPFR-RNAi animals show reduced TAG amount, glycemic levels and increased food intake, but normal resistance to starvation (Figure 8a). This contrasts with the strong decrease in resistance to starvation observed in AKH>NPFR-RNAi animals ( Figure 5b) and indicates that this phenotype is controlled by AKT only. It also indicates that resistance to starvation is independent of decreased lipid accumulation and glycemia, and increased feeding. The authors should provide an explanation for these discrepancies and mention it in the text.
Minor remark 1-The fly food recipe used should be detailed for the reader, since many metabolic phenotypes have been shown to specifically rely on the equilibrium between food components. Reviewer #3 (Remarks to the Author): In this manuscript, Yoshinari et al. report that midgut-derived NPF may act as a sugar-responsive incretin-like hormone in Drosophila. Similar to the regulation of glucagon and insulin by incretin in mammals, the authors show that NPF is released in response to sugar and regulates the counterparts of glucagon and insulin in the fruit fly, Akh and Dilps, by suppressing the secretion of the former and promoting the secretion of the latter. NPF could be the first functional homolog of incretin identified in the invertebrate. The model proposed by the authors is very appealing and shows the deep homology in the principles of metabolic regulation between the vertebrate and invertebrate.
Although, overall, the work is interesting and significant, there are some concerns need to be addressed to make the conclusions more convincing.
Major concerns: 1, The authors showed solid data that all the phenotypes induced by knocking-out or knocking-down NPF (in Figure 1) were caused by the deficiency of NPF. However, based on the current evidence, it is not very convincing to claim that these phenotypes were caused specifically/exclusively by NPF secreted from the midgut EECs. Considering the humoral nature of NPF, it is possible that the decrease of NPF peptides in the circulating system is sufficient to lead to these phenotypes, regardless the origin of the decrease. For example, brain NPF secretion may also be tightly regulated by diet, in that there are also sugar/nutrient sensors in the brain. The following efforts may help address this concern.
First, test if knocking down NPF in other tissues, especially in the central nervous system, causes the same phenotypes.
Second, test if restoring NPF in other tissues, especially in the central nervous system, by either NPF overexpression in NPF neurons or optogenetic/themogenetic activation of NPF neurons, could rescue the phenotypes seen in NPF mutant. The neural-specific manipulation could be achieved by more restrictive GAL4 lines or otd-flp.
Third, simply discuss the alternative possibilities in the discussion.
2, The evidence for the connection between NPFR and IPCs neural activity is not very convincing.
First, the NPFR couples to a Gi signaling pathway to inhibit NPFR-expressing neurons.
Second, the quantifications in Figure 7d lack an internal control, and Figure 7e does not have quantified group data. For example, in Figure 7e, the authors could use DAPI signal as an internal control to normalize the tGPH signals. Actually, the DAPI signal in Figure 7e lower panel (NPFR-RNAi group) appears dimmer than the control groups, suggesting the change in the tGPH signals may just be an artifact due to lower background in that particular staining. In Figure 7d, the authors could also use treatments, such as starvation and re-feeding, to achieve more reliable CaLexA signals, or they could use a similar assay, the TRIC assay, which has RFP expression as an internal expression control.
Minor concerns: 1, Is there a bi-directional regulation of Akh and Dilps by NPF/NPFR? Would opposite phenotypes (in starvation resistance or TAG level) be observed if NPF/NPFR is overexpressed in a wildtype genetic background? 2, The authors may include the phenotypes of Akh>Akh-RNAi or Akh-A/Akh-KO in some of the experiments in Figure 5. It would further confirm that it is Akh in the Akh cells that contributes to the phenotypes. It would also provide clues about whether the phenotypes are bi-directionally regulated. Along the same line, the authors may test if the phenotypes of Dilp2>Dilp2/3/5-RNAi recapitulate the effect of knocking-down NPFR in IPCs to further support the model in Figure 8h. 3, The authors indicated that the behavioral and metabolic phenotypes induced by knocking-down NPFR in the IPCs is caused by attenuated insulin signaling. I think it is a bit of a stretch to attribute both phenotypes to insulin signaling. Feeding and metabolism may be regulated by different mechanisms in the IPCs. For example, most IPCs produce Dilp2/3/5 as well as DSK. It has been shown in previous studies that Dilp2>DSK-RNAi flies also show increased food intake and starvation resistance. It is not rigorous to say Dilps, regulated by NPFR, play the role in both functions without excluding the potential contribution by DSK. 4, The authors consistently observed that the expression level and the protein level of a peptide (NPF and Dilps) changed in opposite directions. It would be better if more discussion could be made about potential mechanisms for this phenomenon. Along this line, is there any reason that the authors only showed the expression level of Akh but not the protein level?
Reviewer #4 (Remarks to the Author): This manuscript describes incretin-like roles for gut-derived Neuropeptide F (NPF, homologous to mammalian NPY) in adult Drosophila. The same lab had previously shown that gut NPF promotes germline stem cell proliferation in mated females (PMID: 30248087) and they now explore its metabolic roles (in mated females too).
The authors show that: 1. Constitutive loss of gut enteroendocrine (EE) cell-derived NPF leads to starvation-sensitive flies. These flies have reduced triacylglyceride and a transcriptional/metabolic profile suggestive of a starvation-like state, despite eating more. 2. The levels of NPF within EE cells are sensitive to dietary sugars and dependent on the sugar transporter1 (sut1) gene 3. In the fed state, NPF signals through its receptor to the secretory cells that make glucagon-like Akh hormone to prevent Akh release; a compelling experiment in support of this idea is that loss of Akh can rescue the lipodystrophy resulting from NPFR receptor mutation or its specific downregulation in Akh-producing cells 4. In the fed state, NPF promotes insulin release, and the NPF effects on Akh and insulin collectively regulate peripheral FOXO levels, sugar and lipid metabolism.
These conclusions are generally supported by comprehensive experiments, and the data is of very good quality. I have three general comments, and a few specific ones: 1. The roles of NPF are practically identical to those recently described for another hormone coexpressed in the same EE cells: Bursicon alpha (Bur, PMID: 30344016). This is fine, but the claims about NPF being the first incretin should be a bit toned down. More importantly, the two manuscripts together raise a few questions about how the same EE cells use different mechanisms to detect the same nutrient (sugar), resulting in potentially differential release of two different EE peptides that end up doing the same thing. This does not make sense to me and needs to be addressed somehow. Specifically, does Sut1 regulate Burs expression? Is NPF not sensitive to Glut1, previously implicated in the sugar-stimulated secretion of Burs from EE cells? Is Sut1 really acting in EE cells? The only data in support of this idea one single EE-specific sut1 RNAi downregulation, and Sut1 expression in EE has not been shown. This is important because, even though the TKg-Gal4 driver used for this downregulation is meant to be gut-specific, it is actually expressed in a small subset of neurons, so the reported phenotypes could be due to sensory/central neuronal sugar sensing/transport.
2. The authors make many statements about NPF secretion without directly demonstrating that NPF is, in fact, secreted as an incretin. I realise that it is not trivial to measure circulating hormones in flies, but this has been achieved for Burs made by the same cells (PMID: 30344016) using Western blots. Do the levels of circulating NPF decrease when NPF is downregulated form EEs and/or in response to sugar restriction?
3. As far as I can tell, all genetic manipulations were performed constitutively throughout development (TKg-Gal4is expressed in third instar larvae possibly earlier). This makes some phenotypes difficult to interpret, such as those affecting fat body lipid and starvation sensitivity; an inability to mobilise fat stores might be expected to result in increased rather than decreased fat stores. To resolve this, the authors should test whether acute depletion of NPF in adult EE cells (by means of a tub-Gal80ts transgene for example) leads to reduced lipid stores, starvation sensitivity and hyperphagia.

Specific comments
1. Lines 66-67. Not sure it is appropriate to describe Activinβ as an enteroendocrine hormone?
3. Line 162: "we hypothesised that starvation impairs NPF secretion from EECs". Consider rephrasing: "impairs" somehow implies a failure when it is in fact the adaptive, homeostatic response. 4. Line 214: "re-introduction" may be more accurate than "overexpression" given that the manipulation is conducted on a mutant that does not express NPFR.

5.
There are a few typos in some figure headings and the summary cartoon.
6. The visceral muscle expression of the NPFR knockin line is surprising in light of its previously reported expression (PMID: 11897397). Although that study used larval rather than adult guts, expression appears to be in epithelial precursors/enteroendocrine cells. The two knockin lines used to assess NPFR expression also seem to differ in the CNS: one looks almost pan-neuronal which seems unlikely. Expression analysis of the endogenous NPFR transcript or protein would increase my confidence in the reported expression.
7. The authors may want to mention Limostatin (a fly decretin, PMID: 25651184) in their Discussion, because together, the two studies suggest that an incretin may be regulating secretion of a decretin.
8. The raw data needs to be provided for metabolomics/transcriptomics datasets. 9. It seems interesting that NPF levels (both protein and transcript) are even higher in peptone-refed than normally fed animals. Might this suggest some integration of sugar/protein amounts by EE cells? The authors might want to discuss this.
10. I believe the primary reference for the Ilp3 antibody is PMID: 18972134.
11. My understanding is that most experiments were done in mated females, but the CAFÉ assay used virgin females. Please clarify/justify.

Response to Referees
Re: MS# NCOMMS-20-29096-A 'The sugar-responsive enteroendocrine neuropeptide F regulates lipid metabolism through glucagon-like and insulin-like hormones in Drosophila melanogaster' (The original title 'The enteroendocrine neuropeptide F acts as an incretin-like hormone in Drosophila')

Dear Editor and Reviewers,
We would like to thank you and the reviewers for your comments, which have strengthened our manuscript. The reviewers' concerns are reproduced below, and our responses are presented in bold. All of these changes can be tracked in the revised text by yellow markers. We hope that this revised manuscript is now suitable for publication in Nature Communications. Sincerely,

Response to Reviewer #1
This interesting paper from Yoshinari et al presents a survey of the physiological effects and pathway function of Neuropeptide F (NPF) as regulator of sugar metabolism and fat catabolism in Drosophila. The authors present extensive data indicating that NF produced in enteroendocrine cells (EECs) in the gut in response to a sugar diet controls the secretion of AKH and DILPs from the fly's corpora cardiaca (CC) and insulin producing cells (IPCs), respectively. AKH and DILP then regulate metabolism in the Fat Body via FOXO, suppressing lipolysis and promoting sugar and fat storage. This mode of regulation makes intuitive sense, and is especially interesting because the way NF is utilized closely parallels incretin function in mammals (see Fig 9). The experimental treatment is quite comprehensive, using ligand and receptor knockouts in the various organs combined with relevant genomics and metabolic assays. Although many of the effects that are documented are rather small in magnitude, the data are nevertheless generally convincing. The genetics are first rate, utilizing elegant rescue experiments in many cases. Moreover, the paper is well written and quite clear. In the very active field of organ-organ communication and metabolic control in Drosophila, this paper looks like a significant advance. I have a few minor comments about presentation, experimental design, and data quality (below), but overall I'm quite enthusiastic about this work. 4. In several instances the paper refers to gene "expression" when discussing mRNA levels measured by RT-qPCR. Please indicate when mRNA is assayed, and when protein is assayed, in both the text and the figures. 7. Figure 3, Lines 156-160: The authors find increased NPF protein in EECs after starvation, and less NPF after a sugar feeding. They conclude from this that NPF secretion is promoted by sugar feeding. This inference is critical to the authors' model of NPF function, but their conclusion is only one of several alternatives that could be taken from the observations. This is one of the weakest links in the story, and it would be good to see more data to support the authors' conclusion about NPF secretion. However, to date, we have been unsuccessful in detecting the circulating NPF levels in Drosophila hemolymph samples. We assume that this is due to the low abundance of endogenous NPF in the hemolymph, or its high rate of degradation in the hemolymph. As the reviewer noted, this is a quite challenging issue, and we cannot further try to promptly develop an appropriate detection method in the COVID-19 pandemic in Japan. I hope that the reviewer will understand our situation. We have, however, addressed this issue in the Discussion section of the revised manuscript (P20, L467-470). In addition, we have mentioned that we 'hypothesize' that starvation suppresses NPF secretion from EECs (P9, L205 -P10, L206).
Regarding DILP level in the hemolymph, we quantified DILP2 protein abundance in the hemolymph upon NPFR RNAi in the IPCs using a method with the DILP2HF strain, as previously reported (Park et al.  Fig. 4a, b). We have also moved a portion of the metabolomic data from the supplemental material to the main figure (Fig. 2a, b) 9. In Fig 2, it would be good to show Trehalose levels, since these are the main circulating sugar in fly hemolymph.

PLOS Genetics
Response: Thank you for this important suggestion. We measured circulating trehalose levels of TKg ts >NPF RNAi animals, and confirmed that knockdown of gut NPF resulted in the reduction of not only glucose but also trehalose in the hemolymph. We have described this data in Extended Data Fig. 2b and P7, L136-137. 10. A bit more data on the expression patterns of the NPFR-Gal4 lines needs to be included. Where exactly is this receptor expressed? Please provide a complete description, and more evidence in the supplement. 11. Figure 5a is another major weak link in the story. Loss of NPFR in the CC reduces Akh mRNA there, but only a little bit and the difference is not very significant. More data should be included to support the function of NPF and NPFR as regulators of Akh, if possible. The genetic analysis on this is great (Fig 5b-k), but ideally the authors would also have measures of Akh protein, or more direct measures of its activity.

Response to Reviewer #2
This is an interesting study linking the function of enteroendocrine cells (EECs) with general energy homeostasis in the Drosophila model. The authors propose that NPF (the invertebrate NPY homolog) produced by EECs serves as an "incretin-like" hormone, controlling both the production of AKH (a glucagon-like hormone) and insulin-like peptides, therefore modulating fat and sugar homeostasis in adult flies.
The study is rather complete and presents numerous experimental approaches to demonstrate these functional links. However, several major links are missing, among which the demonstration that NPF produced by the EECs is secreted and that NPF derived from EECs, and not from other sources, modulates NPFR signaling in the corpora cardiacca and the Insulin-producing cells (IPCs). Experimental evidence for these two points are mandatory to support the conclusions of the paper.
Major remarks: 1-Fig2a: if the metabolic changes in TKg>NPFi are not significant, the conclusion should be differently written (the "trend" has no significance, then), and the data should not be part of a main figure. Line 130, it is improper to write "…were also…", since it suggests that the previous data demonstrates a change in regulation, which is not the case. What is the statistical significance of the data presented in Fig2b?
Response: We would like to thank you for your constructive comments.
As you have pointed out, the expression changes of certain genes shown in the original Fig. 2a (as well as the original Fig. 2b) were not significant.
This figure includes most of the curated genes related to carbohydrate metabolism and TCA cycle regardless of statistical significance. We agree with the reviewer's opinion, and therefore, have moved the data shown in the original Fig.2a and 2b to Extended Data Fig. 4a, b. We have also carefully revised the text describing these data on P8, L164-173. In the revised text, we have specified that statistic significant was set at p < 0.05, which is described in Extended Data Tables 1 and 2.

2-Fig3
: accumulation of NPF in the EECs upon starvation suggests a defect in secretion, but is certainly not a proof. This is an important caveat of the manuscript and the authors should measure circulating levels of NPF in starved vs control food conditions. This should also be done in TKg>NPF-RNAi conditions to ensure that NPF production by the EECs majorly contributes to its circulating levels in the hemolymph. This is particularly important since NPF is also produced by specific neurons in the brain, which could equally contribute to its circulating levels.
In that respect, the authors should also assess the accumulation of NPF in brain neurons upon starvation and measure circulating levels after knock-down of NPF in these neurons in fed conditions. In addition, we have mentioned that we 'hypothesize' that starvation suppresses NPF secretion from EECs (P9, L205 -P10, L206).
Additionally, rather than directly measuring the hemolymph NPF level, we assessed the contribution of brain NPF with a fbp-GAL4, which is active in the brain NPF + neurons, but not in gut EECs. In contrast with the knockdown of NPF in the midgut, knockdown of NPF in the brain did not have significant effects on starvation resistance, TAG level, or feeding (Extended Data Fig. 3a-g). Moreover, knockdown of brain NPF did not alter Dilp2,3-5 levels or AKH levels (Extended Data Fig. 15a-d). We also confirmed that starvation does not induce accumulation of NPF in the brain (Extended Data Fig. 15g). These data suggest that brain NPF plays a distinct role with gut NPF in regulating metabolism. These data are described on P7, L149 -P8, L161 and P19, L437-446. (2) We found that the Glu700 FRET signal was reduced in sut1knockdown EECs as compared to control EECs. In addition, 24-hour starvation significantly decreased FRET signals in control EECs (Extended Data Fig. 7b). This data suggests that sut1 regulates intracellular glucose levels in EECs, which depends on dietary nutrients. These results have been described on P11, L239-244 in the revised manuscript. The experimental procedure is described on P29, L700 -P30, L706.

(3) We established a new strain to overexpress mVenus-tagged
Sut1 protein (Sut1::mVenus) and confirmed that Sut1::mVenus was localized on cellular plasma membranes (Fig. 3e), consistent with the postulate that Sut1 is involved in the transport of its bona fide substrate(s). These results have been described on P10, L230 -P11, L232 in the revised manuscript.
It should, however, be noted that we have not directly assessed the transport activity or affinity of Sut1 for glucose. While future studies should investigate this point, we believe that this is beyond the scope of the current manuscript.
Regarding the second point that the reviewer raised, we agree that the sut1-/-mutant did not show increased NPF protein in the gut, whereas NPF transcript level was significantly decreased (Fig. 3g). This situation was also observed in sut1 knockdown using another UAS-sut1 RNAi(TRiP) line. However, TKg>Sut1 RNAi(VDRCKK) exhibited increased NPF protein in the EECs. Due to these inconsistencies in the data, we have moved the data showing NPF protein levels to the Extended Data from the main body (Extended Data Fig. 6d-e, Extended Data Fig. 8b, f). In contrast, we would like to emphasize that the reduction of NPF mRNA levels with upregulation (or trended upregulation) of NPF protein level are consistent with the results obtained from the three loss of sut1 experiments (Fig. 3f,   g, Extended Data Fig. 8a). Therefore, we surmise that loss of sut1 in the EECs resulted in suppression of NPF signalling in peripheral tissues. This hypothesis is also supported by metabolic data for TKg>sut1 RNAi animals ( Fig. 3h-j, Extended Data Fig. 8c, d).

L443-444) and removed the statement "NPF neurons do not have a direct
connection with the CC and that circulating NPF in hemolymph may be received in the CC cells". To assess the involvement of brain NPF for AKH regulation, we conducted brain-specific knockdown of NPF and found that the knockdown had no significant effects on AKH mRNA and protein levels (Extended Data Fig. 15c, d, and P19, L440-443). Therefore, we conclude that NPF from the midgut, not the brain, has a significant role in the regulation of AKH. 5-p7 line 261 "…suggesting that…". This conclusion seems wrong. Bmm expression is upregulated by Akh>NPFRi, but not dHSL expression. This suggests that dHSL function is not controlled (at least transcriptionally) by NPFR signaling in the CC.
Silencing dHSL expression in fat cells could rescue NPF loss-of-function starvation sensitivity by a parallel mechanism, without implying that dHSL function relies on NPFR.
Response: Thank you for pointing out this inaccurate expression. In the revised manuscript, we have precisely described the difference in the regulatory mechanisms between Bmm and dHSL on P15, L338-339 and L345-347.
6-p12 line 289: The functional link between NPFR in the IPCs, and NPF produced from the EECs is not established. Therefore, such conclusion should not be made.
Again, NPF could be secreted from the central brain and activate NPFR in the IPCs.
Response: Thank you for this important criticism. As suggested, we assessed whether NPF from the central brain affects DILP production/secretion (Extended Data Fig.15a,b). However, knockdown of brain NPF had no significant effect on the DILP mRNA and protein levels. Therefore, we concluded that NPF from the gut, not the brain, has a significant role in the regulation of DILPs. We have mentioned this point on P19, L440-443. 7-Fig7d is not convincing. It is difficult to imagine that such a mild reduction in IPC neuronal activity could account for a notable difference in Dilp secretion.  (Fig. 7e). Furthermore, ad libitum fed NPFR RNAi animals exhibited slight decrease in the CaLexA intensity compared with ad libitum fed control animals (Fig. 7e). Together, these results suggest that, even if the reduction of CaLexA signal intensity is mild, knockdown of NPFR in the IPCs significantly reduces circulating DILP2 level. These data are described on P17, L387-393.

8-Fig7e lacks quantification and confirmation with another marker for insulin
signaling. As such this piece of data is not significant.
Response: Thank you for this important suggestion. We conducted the ratiometric analysis of membrane/cytoplasmic tGPH signal and have presented the quantitative data in Fig. 7f. Based on these results, we confirmed that tGPH signalling at the plasma membrane of the fat body was significantly reduced in Dilp2>NPFR RNAi animals.

Moreover, in response to your suggestion, we confirmed peripheral insulin signalling activity by western blotting analysis to measure the phospho-AKT/pan-AKT ratio of control (dilp2>LacZ RNAi ) and
NPFR knockdown animals (dilp2>NPFR RNAi ). Consistent with our hypothesis, knockdown of NPFR in the IPCs reduced phospho-AKT/pan-AKT ratio (Fig.7g), thus, further supporting our hypothesis that NPFR, in IPCs, regulates insulin signalling in peripheral tissues including the fat body. These results are described on P16, L399 -P17, 402. The experimental procedure is described on P34, L806-822.
9-Throughout their manuscript, the authors present a set of related phenotypes due to EEC-specific NPF knockdown consisting in: resistance to starvation, reduced lipid accumulation, increased feeding and reduced glycemia. Experimental evidence suggest that these effects are due to NPF sending metabolic signals to the CC and the IPCs, AKH and Dilps hormones acting then as relays. However, the phenotypes observed after silencing NPFR in the CC and IPCs are not consistent with this idea and it is unclear what contribution NPFR in the IPCs and the CCs brings to the NPF knockdown phenotype. Strikingly, Dilp2>NPFR-RNAi animals show reduced TAG amount, glycemic levels and increased food intake, but normal resistance to starvation ( Figure   8a). This contrasts with the strong decrease in resistance to starvation observed in AKH>NPFR-RNAi animals ( Figure 5b) and indicates that this phenotype is controlled by AKT only. It also indicates that resistance to starvation is independent of decreased lipid accumulation and glycemia, and increased feeding. The authors should provide an explanation for these discrepancies and mention it in the text.  dilp3 and dilp5 mRNA levels (Extended Data Fig. 14d). In contrast, NPFR knockdown in the IPCs did not influence Akh mRNA expression (Extended Data Fig. 14e). These data suggest that NPFR knockdown in the CC results in not only enhancing AKH production but also suppressing DILP production. These counterregulatory functions of AKH and DILPs may explain the differences in starvation resistance. The data are described on P178, L426 -P19, L434.
Minor remark 1-The fly food recipe used should be detailed for the reader, since many metabolic phenotypes have been shown to specifically rely on the equilibrium between food components. Although, overall, the work is interesting and significant, there are some concerns need to be addressed to make the conclusions more convincing.

Response
Major concerns: 1, The authors showed solid data that all the phenotypes induced by knocking-out or knocking-down NPF (in Figure 1) were caused by the deficiency of NPF. However, based on the current evidence, it is not very convincing to claim that these phenotypes were caused specifically/exclusively by NPF secreted from the midgut EECs.
Considering the humoral nature of NPF, it is possible that the decrease of NPF peptides in the circulating system is sufficient to lead to these phenotypes, regardless the origin of the decrease. For example, brain NPF secretion may also be tightly regulated by diet, in that there are also sugar/nutrient sensors in the brain. The following efforts may help address this concern.
First, test if knocking down NPF in other tissues, especially in the central nervous system, causes the same phenotypes.
Response: Thank you for raising this critical point. In response, we conducted additional experiments to knock down NPF specifically in the brain. To achieve this, we employed the fbp-GAL4 driver, which is active in the NPF+ neurons in the brain but not in gut EECs (Extended Data Fig.   3a). We found that the effect of the brain-specific knockdown of NPF on starvation sensitivity, TAGs level, feeding amounts, DILPs level, and AKH level were not significant as compared to control animals (Extended Data Fig. 3a-e, Extended Data Fig. 15a-d). These data suggest that midgut NPF has a prominent role in suppressing lipodystrophy, which is independent from brain NPF. These results have been described on P7, L151 -P8,

L157, and P19, L440-443.
Second, test if restoring NPF in other tissues, especially in the central nervous system, by either NPF overexpression in NPF neurons or optogenetic/themogenetic activation of NPF neurons, could rescue the phenotypes seen in NPF mutant. The neural-specific manipulation could be achieved by more restrictive GAL4 lines or otd-flp.
Response: In response to the reviewer's suggestion, we conducted transgenic rescue experiments in which we examined whether NPF expression in the brain restored the phenotype of NPF genetic mutant animals. To achieve this, again, we employed the fbp-GAL4 driver. In contrast with NPF restoration in EECs, we found that NPF overexpression in the brain did not restore the hypersensitivity to starvation or the reduced TAG level in NPF mutant animals (Extended Data Fig. 3f, g).
These data support our idea that NPF from the gut regulates lipid storage levels and has been described on P8, L157-161.
Third, simply discuss the alternative possibilities in the discussion.
Response: As per the reviewer's suggestion, we have included a discussion regarding the difference in functions between midgut NPF and brain NPF in the Discussion section (P23, L544 -P24, L559).

2, The evidence for the connection between NPFR and IPCs neural activity is not very
convincing. First, the NPFR couples to a Gi signaling pathway to inhibit NPFRexpressing neurons. Second, the quantifications in Figure 7d lack an internal control, and Figure 7e does not have quantified group data. For example, in Figure 7e, the authors could use DAPI signal as an internal control to normalize the tGPH signals.
Actually, the DAPI signal in Figure 7e lower panel (NPFR-RNAi group) appears dimmer than the control groups, suggesting the change in the tGPH signals may just be an artifact due to lower background in that particular staining. In Figure 7d, the authors could also use treatments, such as starvation and re-feeding, to achieve more reliable CaLexA signals, or they could use a similar assay, the TRIC assay, which has RFP expression as an internal expression control.
Response: We appreciate the reviewer's comments. As you indicated, NPFR reportedly couples to a Gi signalling pathway. However, we would like to note that NPFR is also known to be coupled with both Gαq and Gαi subunits in heterologous expression systems, which we have described on P22, L523-525. Additionally, such dual G-protein coupling has been reported in short NPF receptor (sNPFR) signalling in the IPCs and CC (Oh et al. Nature. 2019). In this case, sNPFR positively regulates IPCs neuronal activity via Gαq signalling, whereas sNPFR negatively regulates the CC via Gαi signalling. Therefore, we speculate that such differential coupling in the different cell types would also occur in NPFR.
In response to the reviewer's criticism regarding the quantification of tGPH signals, we performed the ratiometric assay for membrane/cytoplasmic tGPH signals and presented the data in Fig. 7f.
The quantitative data supports our conclusion that tGPH signaling at the plasma membranes of the fat body was significantly reduced in Dilp2>NPFR RNAi animals.

Regarding an internal control and quantitative analysis of the
CaLexA experiment, we have added further data in the revised manuscript. Specifically, we conducted a quantitative analysis of CaLex signals (Fig. 7e). Consistent with previous studies (Meschi et al. Dev Cell. RNAi animals (Fig. 7e). We also confirmed that the IPC CaLexA level of the ad libitum fed dilp2>NPFR RNAi animals was stronger than that of ad libitum control animals (Fig. 7e).

2019.), starvation was observed to reduce neuronal activity of the IPCs in both control and Dilp2-GAL4-driven NPFR
These data suggest that knockdown of NPFR in the IPCs slightly, however, significantly, decreases its neuronal activity. We believe that this additional data further supports our postulate that NPFR positively regulates IPC neuronal activity. These results are described on P17, L387-

393.
Minor concerns: 1, Is there a bi-directional regulation of Akh and Dilps by NPF/NPFR? Would opposite phenotypes (in starvation resistance or TAG level) be observed if NPF/NPFR is overexpressed in a wildtype genetic background? Fig. 1g). Therefore, as you suggested, bi-directional regulation appears to exist. We have described this data on P6, L122-123.  Genetics 2015), we found that knockdown of Akh increased the starvation resistance and TAG abundance (Fig. 5d, f).

Response: We conducted an additional experiment to examine whether overexpression of NPF in the EECs affect TAG level and observed a slight increase in the TAG level (Extended Data
Interestingly, whereas co-suppression of NPFR and Akh expression in the CC restored starvation sensitivity of NPFR knockdown, the cosuppression phenotype was stronger than a single Akh knockdown. These results suggest that not only AKH, but also unknown factor(s) from the CC, contribute to the hypersensitive phenotype of the NPFR knockdown. We have described these points on P14, L315-318.
In response to the reviewer's second suggestion, we assessed the TAG levels in Dilp2>Dilp2/3/5 RNAi animals. Knockdown of dilp3 slightly decreased the TAG levels, while knockdown of dilp2 and dilp5 resulted in no significant differences or only a slight decreasing trend in TAG abundance, respectively (Extended Data Fig.14b) In addition, we have mentioned that we 'hypothesize' that starvation suppresses NPF secretion from EECs (P9, L205 -P10, L206). specific sut1 RNAi downregulation, and Sut1 expression in EE has not been shown. This is important because, even though the TKg-Gal4 driver used for this downregulation is meant to be gut-specific, it is actually expressed in a small subset of neurons, so the reported phenotypes could be due to sensory/central neuronal sugar sensing/transport. Therefore, in conjunction with our other data (Extended Data Fig. 1a-c), it is suggested that the region-specific sugar-sensing mechanism between NPF+ EECs and Bursα+ EECs are Sut-1 and Glut1-dependent, respectively. It will also be intriguing to investigate how gut cells sense multiple dietary nutrients, in the future. We have discussed these points in the Discussion section (P20, L472 -P21, L483).

Response
To rule out the possibility that TKg-GAL4 + neurons in the brain affect NPF level in the gut, we conducted brain specific sut1 knockdown using tub>FRT>GAL80>FRT combined with Otd-FLP and nSyb-GAL4. We found that the brain specific sut1 knockdown did not affect midgut NPF mRNA level, while the midgut NPF protein level was slightly reduced (Extended Data Fig. 9a, b). Although sut1 knockdown in the brain had a minor effect on the regulation of NPF in the EECs, TKg>sut1 RNAi had a stronger effect on NPF protein accumulation with reduced NPF mRNA level. Moreover, Sut1 KI-T2A -GAL4 was not expressed in NPF + neurons in the brain (Extended Data Fig. 9d). These results suggest that the TKg>sut1 RNAi phenotype is not a secondary effect of brain NPF. These results are described on P12, L257-261.
2. The authors make many statements about NPF secretion without directly demonstrating that NPF is, in fact, secreted as an incretin. I realise that it is not trivial to measure circulating hormones in flies, but this has been achieved for Burs made by the same cells (PMID: 30344016) using Western blots. Do the levels of circulating NPF decrease when NPF is downregulated form EEs and/or in response to sugar restriction?
Response: We appreciate your comment regarding the importance of measuring hemolymph NPF levels. However, over the last 8 months and half, since receiving the reviewers' feedback, we have tried several experiments (ELISA, Dot blotting, western blotting, and LS-MS/MS analysis) to address this point. However, to date, we have been unsuccessful in detecting the circulating NPF levels in Drosophila hemolymph samples. We assume that this is due to the low abundance of endogenous NPF in the hemolymph, or its high rate of degradation in the hemolymph. In the COVID-19 pandemic in Japan, we cannot further try to promptly develop an appropriate detection method. I hope that the reviewer will understand our situation. We have, however, addressed this issue in the Discussion section of the revised manuscript (P20, L467-470).
3. As far as I can tell, all genetic manipulations were performed constitutively throughout development (TKg-Gal4 is expressed in third instar larvae possibly earlier).
This makes some phenotypes difficult to interpret, such as those affecting fat body lipid and starvation sensitivity; an inability to mobilise fat stores might be expected to result in increased rather than decreased fat stores. To resolve this, the authors should test whether acute depletion of NPF in adult EE cells (by means of a tub-Gal80ts transgene for example) leads to reduced lipid stores, starvation sensitivity and hyperphagia.
Response: Thank you very much for providing us with this important suggestion. In response, we conducted additional experiments to achieve the adult-specific NPF knockdown with tub-GAL80 ts . Consistent with the observed phenotypes of TKg-GAL4 without tub-GAL80 ts , knockdown of NPF in the adult stage also exhibited reduced TAG levels, high starvation sensitivity, hypoglycaemia, and hyperphagia (Fig.1g-i, Extended Data Fig.   2a-c). These data rule out the possibility that loss of NPF during the larval stage affects adult metabolism. We have described these results on P6, L130 -P7, L139.

Specific comments
Response: We have fixed these typographical errors. Thank you for pointing them out.
6. The visceral muscle expression of the NPFR knockin line is surprising in light of its previously reported expression (PMID: 11897397). Although that study used larval rather than adult guts, expression appears to be in epithelial precursors/enteroendocrine cells. The two knockin lines used to assess NPFR expression also seem to differ in the CNS: one looks almost pan-neuronal which seems unlikely. Expression analysis of the endogenous NPFR transcript or protein would increase my confidence in the reported expression. Fig. 11 and on P12, L276 -P13, L282. We found that NPFR-T2A-KI -GAL4 is expressed in the brain, CC, sNPF+ enteric neurons, Malpighian tubules, ovary, and gut. As far as we observed, two NPFR-KI -GAL4 lines showed similar expression patterns in the brain, including the IPCs (Fig.7a, Extended Data Fig. 11a, 14a). We also observed expression of the NPFR KI-T2A -GAL4 line in the Malpighian tubules (Extended Data Fig. 11d) Fig. 4a, Extended Data Fig. 4b, and Fig. 2d,

respectively.
We have also supplied Extended Data Tables 3 and 4, which represent raw data for metabolic analyses corresponding to Fig. 2a-c, and Extended Data Figure. 5a. 9. It seems interesting that NPF levels (both protein and transcript) are even higher in peptone-refed than normally fed animals. Might this suggest some integration of sugar/protein amounts by EE cells? The authors might want to discuss this.
Response: Thank you for providing us with these interesting comments.
As the reviewer pointed out, peptone-refeeding increased NPF mRNA and protein levels compared to the normal food-fed condition (Fig. 3a-c).
Although these data suggest that EECs sense sugar/protein levels in the diet, we did not identify the mechanism by which midgut NPF mRNA and NPF protein levels are upregulated by peptone feeding. In the revised manuscript, we have discussed this point in the Discussion section on P21, L489-491.
10. I believe the primary reference for the Ilp3 antibody is PMID: 18972134.

Response: We have cited the reference (Ref #91) for the anti-DILP3
antibody and anti-TK antibody on P28, L678 and P29, L682.
11. My understanding is that most experiments were done in mated females, but the CAFÉ assay used virgin females. Please clarify/justify.