Cholinergic signaling modulates intestinal pathophysiology in a Drosophila model of cystic fibrosis

Abstract

Cystic fibrosis (CF) is a monogenic genetic disease caused by mutations in the Cystic Fibrosis Transmembrane conductance Regulator (CFTR) chloride/bicarbonate channel, which is expressed in certain epithelial cells. Current therapies focus on restoring CFTR function, but many gut-related pathologies persist, highlighting the need for complementary treatments to improve the quality of life of people with CF. In this study, we use Drosophila melanogaster as a model to investigate the gut-specific effects of Cftr loss. We demonstrate that enterocyte specific knockdown of Cftr in flies recapitulates several CF pathologies, including reduced intestinal motility, nutrient malabsorption, and decreased energy stores. Using single-nuclei RNA sequencing (snRNA-seq), we identify significant transcriptional changes in the CF model gut, including the upregulation of acetylcholine esterase (Ace, human AChE), which leads to reduced cholinergic signaling. Cholinergic signaling has been shown to affect CFTR function but this is the first time CFTR loss of function has been shown to alter cholinergic signaling. Functional assays confirm that cholinergic sensitivity is diminished in CF guts. Furthermore, restoring cholinergic signaling via Ace knockdown rescues multiple CF-associated phenotypes. Additionally, we identify the transcription factor Fork head (Fkh), the Drosophila homolog of human FOXA1/FOXA2, which is known to be a positive regulator of Cftr transcription in the intestine, as a positive regulator of Ace expression in CF guts. This study establishes the Drosophila gut as a powerful model to investigate CF pathogenesis, genetic modifiers, and identifies Ace and fkh as genetic modifiers. This work also suggests that enhancing cholinergic signaling may represent a viable therapeutic strategy for gastrointestinal manifestations of CF.

Author summary

Cystic fibrosis (CF) is a genetic disease that causes complications in multiple organ systems, including the lungs and gastrointestinal tract. While recent therapies have greatly improved life span and respiratory outcomes in people with CF (pwCF), they continue to report significant gastrointestinal symptoms that impact their quality of life. Therefore, there is a need to better understand the progression of CF in the gut and identify gut specific therapeutic targets to improve patient quality of life. In this study we use Drosophila to model gut specific complications of CF and show our model recapitulates many CF clinical presentations. Furthermore, we identify acetylcholine esterase (Ace) as a gene that is increased in our CF model guts and is important for the development of CF pathologies. We additionally perform a screen to identify a transcriptional regulator of Ace, whose genetic manipulation also regulates CF phenotypes. Our findings not only identify a potential target to alleviate GI symptoms in patients with CF, but also validate the Drosophila gut as a robust model for studying CF pathogenesis, genetic modifiers, and screening therapeutics.

Introduction

Cystic fibrosis (CF) is a genetic disease that affects approximately 1 in 2,500 newborns in the United States [1]. CF is caused by mutations in the Cystic Fibrosis Transmembrane conductance Regulator (CFTR) chloride channel, which is expressed in certain epithelial tissues, including the respiratory tract, pancreas, and the gut [2–4]. The CF phenotype is most closely associated with the accumulation of mucus in the pulmonary and gastrointestinal tracts, which can lead to inflammation, bacterial infection, and malnutrition [1,5–8]. Much of CF research has focused on the lung, as the primary cause of death in people with CF (pwCF) have been related to lung complications. However, CF causes pathologies in other CFTR-expressing tissues, including the gut, that result in various clinical phenotypes [7,9–13]. For example, many pwCF are born with intestinal blockage (Meconium Ileus) and later in life have decreased intestinal motility, dysbiosis, inflammation, and poor nutrient absorption in the gut [7,9–13]. Additionally, as life expectancy has increased following better treatments, complications in the gastrointestinal tract have become an increasing cause of morbidity [10–15]. These complications include small bowel bacterial overgrowth, bowel obstructions, and increased risk of intestinal cancers [11–13]. Recent studies have suggested that CF modulator drugs may not fully correct the inflammation and dysbiosis seen in the gut of pwCF and that pwCF on modulator drugs still report GI symptoms [14,16–19]. Therefore, additional therapies targeting the gut, not directly related to modulating CF activity may be useful for quality of life of pwCF. Recently, there have been studies highlighting how the CF gut can crosstalk with many other organ systems, further highlighting the importance of investigating how loss of CFTR function effects gut biology [20–22].

The Drosophila midgut, analogous to the human intestine, is comprised primarily of 4 main cell types: the absorptive enterocytes (ECs), the neuropeptide producing enteroendocrine (EE) cells, intestinal stem cells (ISCs), and the intermediate progenitor enteroblasts (EBs) [23]. The Drosophila midgut has been used to study a number of questions in stem cell biology, cell signaling, and physiology, and many of the mechanisms active in the intestines of flies have already been shown to apply more broadly to other organisms and may therefore be relevant for human pathologies [23–25]. The fly ortholog of human CFTR has recently been identified and used to establish an intestinal CF model [26]. This model revealed that Cftr knockdown in the Drosophila intestine disrupts osmotic homeostasis and displays CF-like phenotypes in the intestinal epithelium [26]. While this work introduced Drosophila as a CF model, the phenotypes examined were largely cellular phenotypes, such as increased chloride and sodium levels in the ECs, and not clinical manifestations of CF.

Here, we further characterize the CF model gut, demonstrating that several clinical pathologies, including reduced intestinal motility and nutrient malabsorption are preserved. Furthermore, the Drosophila CF model displays characteristics consistent with a failure to thrive phenotype. Additionally, we perform single nuclei RNA-seq (snRNA-seq) to further characterize the CF gut model and learn new biology relevant to CF. Interestingly, we found increased acetylcholine esterase (Ace) expression in CF model guts, which results in reduced cholinergic signaling potential. Recent work in Drosophila has demonstrated that cholinergic signaling is important for maintenance of the intestinal epithelial barrier and is required for the intestinal epithelium to return to homeostasis after injury [27,28]. Additionally cholinergic signaling has been implicated in other diseases of intestinal inflammation such as Inflammatory Bowel Disease (IBD) but has not yet been investigated in the context of the CF intestine [29,30]. Previous studies have demonstrated that cholinergic signaling can increase CFTR function [31–34] but this is the first work that demonstrates CFTR function can have a reciprocal effect on cholinergic signaling. We further show that the decrease in cholinergic signaling observed in the CF model guts may be clinically relevant as restoring sensitivity to cholinergic signaling rescues many CF pathologies in Drosophila. Finally, we identify Fork head (Fkh), a FOXA1/A2 homolog as a transcriptional regulator of Ace expression in the CF model gut. FOXA1/A2 is a pioneering transcription factor that positively regulates the expression of Cftr and other transmembrane proteins and ion channels important for regulating ion homeostasis in the intestinal epithelium [35,36]. Altogether, our findings help establish Drosophila as a model organism to study GI manifestations of CF. We identify Ace (human: AChE) and Fkh (human: FOXA1/A2) as genetic modifiers of CF and our findings suggests that the cholinergic signaling pathway may be a viable therapeutic target in CF gastrointestinal disease.

Results

Knockdown of Cftr in the Drosophila midgut recapitulates clinical pathologies of Cystic Fibrosis

People with CF (pwCF) display a range of multiple organ specific as well as systemic pathologies, including gastrointestinal complications [10–13,37–39]. Therefore, we examined whether Drosophila CF model guts, with enterocyte-specific knockdown of Cftr (Myo1A-Gal4 > UAS-Cftr^RNAi), recapitulated typical clinical pathologies. pwCF are at increased risk for gastrointestinal cancers and microbial dysbiosis [11,13,39,40]. These clinical observations are consistent with previous findings in the Drosophila CF model, showing gut hyperplasia and increased bacterial load [26]. Additional common CF gut pathologies such as meconium ileus, distal intestinal obstruction syndrome, and constipation are linked to impaired intestinal motility [9,38,41]. To test whether the CF model exhibits reduced gut motility, we performed an excretion assay. Cftr knockdown flies showed significantly reduced excretion compared to wild-type (WT) controls, indicating impaired intestinal transit (Figs 1A and S1A). Malabsorption of nutrients in the intestine is another hallmark of CF [42]. In line with this, CF model flies exhibited increased glucose, triacyl glycerides (TAGs), and protein levels in their excreta compared to WT flies, suggesting reduced nutrient absorption in the intestine (Figs 1B-1D and S1B). Of note increased tissue breakdown and stress responses can result in the fly kidney (Malpighian tubules) depositing lipids into the excreta which may contribute to the increased nutrients we observe in the excreta in our model [43].

Fig 1. CF model gut recapitulates hallmarks of CF.

(A) CF model guts have decreased excretion rate compared to WT guts. n = 19 vials of 10-15 females from 3 independent experiments. (B) CF model guts have increased glucose in excreta compared to WT flies. n = 16 (WT), 17 (Cftr^RNAi) vials of 15-20 females from 3 independent experiments. (C-D) CF model guts have increased TAG (C) and protein (D) in excreta compared to WT flies. n = 10 vials of 60 females from 3 independent experiments. (E) CF model guts have reduced TAG levels at 2 weeks of age compared to WT flies. n = 22 (WT), 24 (Cftr^RNAi) of 5 pooled females from 4 independent experiments. (F) CF model guts have slightly reduced whole body glucose levels at 2 weeks of age compared to WT flies. n = 22 (WT), 24 (Cftr^RNAi) of 5 pooled females from 4 independent experiments. (G) CF model gut flies have reduced whole body TAG levels at 4 weeks of age compared to WT flies. n = 22 (WT), 24 (Cftr^RNAi) of 5 pooled females from 4 independent crosses. (H) CF model gut flies have reduced whole body glucose at 4 weeks of age compared to WT flies. n = 24 (WT), 23 (Cftr^RNAi) of 5 pooled females from 4 independent crosses. (A-H) P-values were calculated using the Mann-Whitney test in GraphPad prism. Error bars are mean with 95% CI. (I) CF model gut flies have reduced lifespans compared to WT flies. n = 9 (WT), 12 (Cftr^RNAi) vials with 10-15 females from 2 independent crosses, error bars are mean + /- SEM. (A-I) WT is Myo1A > + and Cftr^RNAi is Myo1A > UAS-Cftr^RNAi.

Failure to thrive, a condition where pwCF fail to gain weight and grow at the expected rate, is a more systemic pathology with a strong intestinal component [7,42,44]. Cftr knockdown flies have significantly reduced energy stores, with lower levels of whole-body TAG and glucose at both 2 and 4 weeks of age (Figs 1E-1H and S1C-S1F). The reduction of TAG levels is robust at both 2 and 4 weeks of age while the reduction of glucose stores is more apparent at 4 weeks of age. This reduction in whole body energy stores is not due to a developmental defect as inducing Cftr knockdown in adult flies using the temperature sensitive Gal80^TS repressor also leads to reduced whole body TAGs and glucose (S1G-S1H Fig). Furthermore, the reduction in whole body energy stores is not due to reduced food intake as the CF gut model flies consume similar amounts of food to WT flies over 24 hrs and in 30 minutes after starvation (S1I-S1J Fig). This decrease in whole body energy stores is consistent with a failure to thrive phenotype. As the CF gut model only has reduced Cftr expression in the midgut, it may be useful in studying gut-specific contributions to the failure to thrive outcome.

Finally, pwCF have reduced lifespans, although they have improved dramatically in recent years due to the availability of better treatments [45]. Our model fly shows that loss of Cftr in the gut alone is sufficient to reduce the lifespan of the fly, highlighting the importance of CFTR function in the gut (Figs 1I and S1K). Overall, these results show that loss of Cftr in the enterocytes recapitulates many hallmarks of CF in the gastrointestinal system.

Single nuclei analysis of CF model guts

To further characterize the CF gut model, we performed snRNA-seq on the midgut of WT and CF model gut flies. The sn-RNA-seq analysis recovered 3,669 nuclei for WT and 4,166 for CF midguts that were clustered into 21 clusters annotated using the marker genes from the previously published cell atlas of the Drosophila gut [46,47] (Fig 2A). These clusters include multiple enterocyte (EC) clusters that map along the Drosophila digestive tract and show transcriptional similarity to mammalian enterocytes [48] (S2A Fig). The posterior EC (pEC) clusters show high transcriptional similarity to the mature distal enterocytes while the anterior ECs (aEC) cluster only shows moderate similarity to mature proximal enterocytes (S2A Fig). The middle ECs (mECs), copper cells, and large flat cells (LFC) which reside in the middle of the Drosophila midgut show moderate similarity to mammalian EC clusters (S2A Fig). The Drosophila midgut does not contain specific goblet or paneth cells but the aEC clusters are transcriptionally similar to paneth cells and multiple EC clusters express genes specific to goblet cells in the mammalian gut (S2A Fig). The Drosophila enteroendocrine cells show strong sequence similarity to the mammalian enteroendocrine cells. Finally, the Drosophila, intestinal stem cells (ISCs), enteroblasts (EBs), and adult differentiating enterocytes are similar to the mammalian stem cells, transit amplifying cells, and enterocyte progenitor cells (S2A Fig).

Fig 2. snRNA-Seq of CF model guts.

(A) Annotated cell clusters of snRNA-Seq from WT (Myo1A > +) and CF (Myo1A>Cftr^RNAi) midguts, visualized with UMAP. (B) Over and underrepresented cell clusters in CF model gut snRNA-seq data set. Size of circle represents number of nuclei in CF model gut analysis for indicated cluster. (C) Number of differentially expressed genes in each cell cluster between WT and CF model gut snRNA-seq data sets. (D) Heat map of the top 5 differentially expressed secreted proteins for each cell type. Color reflects the log2fold change in expression in CF model guts compared to control per cell cluster.

UMAP analysis of the sn-RNA-seq data does not identify any unique clusters in the CF model guts, however there are a number of clusters that are over or under-represented in the CF model (Figs 2A-2B and S2B). This includes increased ISCs, which is consistent with the hyperplasia previously described in the Drosophila CF model gut [26]. Furthermore, there are a significant number of differentially expressed genes (DEGs) between WT and CFTR deficient guts across all identified cell clusters (Fig 2C and S1 File). Together these results indicate substantial differences in the transcriptional profile and cellular composition between WT and CF model guts.

Interestingly, many secreted peptides are differentially expressed indicating that the CF gut model may be a good model to study the crosstalk between the CF gut and other organs (Figs 2D and S2C and S2 File). Among the upregulated secreted proteins in the CF model midguts are several that have also been identified as key factors secreted by Yki-activated gut tumors involved in tumorigenesis and communication with other organs (Impl2, Pvf1, Itp, and Upd3) [49,50] (S2D Fig). This overlap suggests that the CF model gut may exhibit a predisposition toward intestinal tumorigenesis similar to pwCF who are at increased risk of gastrointestinal cancers [11,13,39,40].

In addition to the expected increase of cells in the ISCs cell clusters [26] there is a redistribution of cells in anterior and posterior EC cell clusters (Figs 2A-2B and S2B). In the CF preferred anterior EC cluster, one of the top 5 differentially expressed secreted peptides is Acetylcholine esterase (Ace, human AChE) (Fig 2D). Ace expression is increased overall in CF model guts but is particularly apparent in the anterior EC clusters (Figs 3A and S3A-S3B). We confirmed increased Ace expression in our CF model guts by RT-qPCR analysis of whole guts (S3C Fig).

Fig 3. CF model guts have reduced sensitivity to cholinergic signaling.

(A) Violin plot of acetylcholine esterase (Ace) expression in anterior EC2 cell clusters indicates higher Ace expression in CF model guts. Each dot represents expression in a single-nucleus, n = 137 (WT) or 479 (Cftr^RNAi) for 05_aEC2 and n = 362 (WT) or 213 (Cftr^RNAi) for 06_aEC2. P-values were calculated using the Mann-Whitney test in ggplot2. (B-C) Cftr deficient midguts have reduced sensitivity to acetylcholine (ACh) stimulation as measured by calcium levels in enterocytes. (B) Representative images of GCaMP7c fluorescence, in WT (Mex^TS > GCaMP7c, Luc^RNAi) and Cftr deficient guts (Mex^TS > GCaMP7c, Cftr^RNAi) before (T15s) or after addition of ACh (T125). Scale bar 50 μm. (C) Graph of average relative fluorescent intensity, ΔF/F0, per frame (2.5s per frame) and genotype. n = 9 (WT) and 10 (Cftr^RNAi) from 4 independent experiments. Error bars are mean + /- SEM. P-value was calculated using the Mann-Whitney test in GraphPad prism. (D-E) Ace knockdown increases sensitivity to ACh stimulation in Cftr deficient enterocytes. (D) Representative images of GCaMP7c fluorescence, in Cftr deficient (Mex^TS > GCaMP7c, Cftr^RNAi; Luc^RNAi) midguts and Cftr deficient guts with Ace knockdown (Mex^TS > GCaMP7c, Cftr^RNAi; Ace^RNAi) before (T15s) or after addition of ACh (T125). Scale bar 50 μm. (E) Graph of average relative fluorescent intensity, ΔF/F0, per frame (2.5s per frame) and genotype. n = 3 (WT), 5 (Cftr^RNAi, Luc^RNAi), 4 (Cftr^RNAi, Ace^RNAi) from 3 independent experiment. Error bars are mean + /- SEM. P-values were calculated using 2-way ANOVA with Tukey’s multiple comparisons main column effect in GraphPad prism.

Ace hydrolyzes acetylcholine into acetate and choline thereby inhibiting cholinergic signaling. Cholinergic signaling has recently been shown to be required for recovery of the intestinal epithelium after damage and for maintaining intestinal barrier function in Drosophila [27,28]. Furthermore, cholinergic signaling has been shown to increase CFTR activity but there are no studies showing CFTR function affecting cholinergic signaling [31–34]. Therefore, we investigated the importance of increased Ace expression in our CF gut model.

Reduced Cholinergic signaling in the CF Gut Model

As Ace degrades acetylcholine (ACh), the agonist of cholinergic receptors, increased Ace expression should decrease sensitivity to ACh and limit cholinergic signaling [27]. Cholinergic signaling has been proposed to use Ca²⁺ as a secondary messenger to modulate the mammalian intestinal epithelium [51]. Therefore, to test for cholinergic sensitivity in the Cftr deficient midguts we visualized Ca²⁺ by conditionally expressing the Ca²⁺ indicator GCaMP7c[52] in ECs and performed ex vivo live imaging [27]. We found that Ca²⁺ levels in CF model guts were reduced in response to ACh, indicating that CF model guts had a dampened response to ACh stimulation compared to WT guts (Fig 3B-3C).

Cholinergic receptors include both muscarinic and nicotinic receptors, however previous sequencing of the Drosophila gut only identifies the presence of certain nicotinic receptor subunits and no muscarinic subunits [53,54]. Therefore, we treated the Drosophila midguts with nicotine, which activates nicotinic receptors without being degraded by Ace [55]. Nicotine treatment increases Ca²⁺ levels in WT and to a lesser extent in CF model guts (S3D-E Fig). This suggests that cholinergic signaling may be impaired at the receptor level in the CF model gut in addition to the increased Ace expression. In our snRNA-seq data set nicotinic receptors have low coverage and were therefore filtered out of the differential expression analysis (S3 File). However, when the nicotinic receptors are included in the analysis, we observe a trend towards reduction in overall nicotinic receptor subunit gene expression (S3F-G Fig). We confirmed this overall reduction in nicotinic receptor subunits by RT-qPCR (S3H Fig). We observe a significant decrease in beta1, beta2, alpha5 and alpha7 nicotinic receptor subunits with no change in the alpha4 and alpha5 subunits and were not able to detect the alpha6 subunit mRNA via RT-qPCR (S3H Fig). These data confirm that cholinergic signaling is reduced in the CF model by both reduced levels of nicotinic receptor subunits and by increased Ace expression.

To test the importance of Ace expression in the CF gut models sensitivity to cholinergic signaling we used the Mex-Gal4 (enterocyte specific driver) together with the Gal4 repressor Tubulin-Gal80^TS to drive expression of both Ace^RNAi and Cftr^RNAi in adult Drosophila for 2 weeks (S4 Fig). When Ace levels are reduced in the CF background the response to ACh is increased to WT levels (Fig 3D-3E). This result indicates that reducing Ace expression in CF model guts is sufficient for rescuing sensitivity to cholinergic signaling even if nicotinic receptor subunit levels are also lower in the CF model. This is likely because Ace knockdown in the CF background reduce Ace levels below WT levels (S3I Fig). Therefore, if decreased cholinergic signaling is important for CF pathophysiology reducing Ace expression should alter CF phenotypes in the Drosophila model.

Increasing cholinergic signaling in CF model gut rescues several CF phenotypes

Recent work has demonstrated that loss of cholinergic signaling after damage leads to increased proliferation of ISCs and a failure to return to homeostasis [27]. Therefore, we examined whether Ace expression affected the hyperplasia present in the CF model gut. Indeed, Ace knockdown in the CF model gut reduced proliferation in the gut indicating that decreased cholinergic signaling contributes to the increased proliferation observed in the CF intestine (Fig 4A).

Fig 4. Increasing cholinergic signaling rescues CF phenotypes.

(A) Ace knockdown rescues the increased proliferation seen in CF model guts. n = 7 (WT; Luc^RNAi), 10 (Cftr^RNAi, Luc^RNAi), 11 (Cftr^RNAi; Ace^RNAi). (B) Ace knockdown increases intestinal motility as measured by excretion rate in CF model guts. n = 16 vials with 9-13 female flies from 3 independent experiments. (C) Ace knockdown rescues whole body TAG levels at 2 weeks of age in CF model guts. n = 19 (WT; Luc^RNAi), 20 (Cftr^RNAi, Luc^RNAi), 22 (Cftr^RNAi; Ace^RNAi) of 5 pooled females from 4 independent experiments. (D) Ace knockdown reduces the amount of glucose remaining in excreta in CF model gut flies. n = 14 (WT; Luc^RNAi), 14 (Cftr^RNAi, Luc^RNAi), 15 (Cftr^RNAi; Ace^RNAi) vials of 15-20 females from 3 independent experiments. (E-F) Ace knockdown reduces the amount of TAG (E) and protein (F) remaining in excreta in CF model gut flies. n = 9 vials of 60 flies from 3 independent experiments. (G) CF model gut flies have increased intestinal permeability, measured by the presence of normally non-permeable blue dye in the hemolymph at 35 days of age, which is rescued by Ace knockdown. n = 9 (WT; Luc^RNAi), 10 (Cftr^RNAi, Luc^RNAi), 10 (Cftr^RNAi; Ace^RNAi) vials of 9-13 females from 3 independent experiments. (A-G) P-values were calculated using ordinary one-way ANOVA with Tukey’s multiple comparisons test in GraphPad prism. Error bars are mean with 95% CI. RNAi constructs were expressed in enterocytes using the enterocyte specific driver Mex^TS (Mex-Gal4 with the temperature sensitive Gal4 repressor tubulin Gal80^TS) (S4 Fig).

We also found that increasing cholinergic signaling by reducing Ace levels in the CF model gut rescued the excretion rate indicating that cholinergic signaling can regulate intestinal transit in the CF model (Fig 4B). Reducing Ace expression in the CF model gut was also sufficient for rescuing both the whole-body TAG stores and the malabsorption of nutrients observed in the CF model gut at 2 weeks of age (Fig 4C-4F).

Recently cholinergic signaling has been linked to intestinal barrier function in the Drosophila gut [28]. As there is evidence of increased intestinal permeability in pwCF [56], we tested whether the CF model gut had decreased intestinal barrier function. To test intestinal barrier function, we performed a Smurf assay, where flies are fed a blue food dye that in healthy guts is impermeable to the intestinal barrier. When intestinal barrier function is impaired the blue dye leaks into the hemolymph and the fly turns blue (Smurf) [57]. We found that CF model guts have decreased barrier function demonstrated by the increase percentage of Smurf flies at 35 days of age (Fig 4G). Importantly, intestinal barrier integrity is rescued by increasing cholinergic signaling via Ace knockdown in the gut (Fig 4G).

Altogether these results indicate that the reduction in cholinergic signaling observed in CF model guts contributes to multiple CF pathologies.

Fkh regulates Ace transcription in CF model gut

As Ace RNA levels are changed in the CF model gut, we performed a screen of transcription factors to identify how Ace expression is regulated downstream of loss of CFTR function. To be included in the screen the transcription factor had to 1) have expression in the single cell data where Ace is expressed (S5A Fig) and 2) have potential binding sites as assessed by the TF2G database (S1 Table) [58]. We identified 18 putative transcription factors (TFs) and used RNAi to test their ability to regulate Ace transcription in the Cftr deficient background (Myo^TS> Cftr^RNAi, TF^RNAi) by quantifying Ace levels (S5B Fig). The top 2 hits with the lowest Ace expression in the CF background were fork head (fkh) RNAi lines (Fig 5A). In Drosophila Fkh has been shown to regulate ISC proliferation and expression of nutrient transporters in the intestine [59,60]. Interestingly, the mammalian ortholog FOXA1/A2 regulates expression of CFTR and other transmembrane proteins important for regulating ion homeostasis in the intestinal epithelium [35,36]. Additionally, fkh expression has the highest correlation with Ace expression in the aEC2 CF cell cluster of the transcription factors we tested (S5C Fig). Furthermore, Fkh has 2 ChIP-Seq peaks identified at the Ace promoter in an embryonic ChIP-seq data set in a region identified as a cis-regulatory module (CRM) [61] for Ace (S5D-S5E Fig). In addition to the Fkh peaks at the Ace promoter there are an additional 2 Fkh ChIP-seq peaks in the intronic regions of Ace (S5D Fig). These intronic binding sites could also be important for regulation of Ace transcription especially since the Fkh mammalian ortholog, FOXA1, has been shown to regulate CFTR expression through intronic binding sites [62–64]. Together these data support regulation of Ace transcription by Fkh as Fkh is 1) expressed in cells that express Ace (S5A and S5C Fig) 2) has the ability to bind to the Ace promoter and intronic regions (S5D-S5E Fig) and 3) knockdown of Fkh reduces Ace mRNA levels (Figs 5A and S5F). Therefore, we investigated Fkh activity and function in the CF model guts.

Fig 5. Fkh regulates Ace transcription and CF pathologies in the CF model gut.

(A) Screen of transcription factors identifies Fkh as a candidate transcription factor of Ace in the CF model guts. n = 2 biological replicates of 15 pooled guts. error bars are mean + /- SD. RNAi constructs were expressed in adult flies for 2 weeks using Myo^TS (Myo1A-Gal4, tub-Gal80^TS) driver (S5B Fig). (B) Representative images of increased Fkh nuclear localization in CF model guts compared to WT. scale bars 25μm. (C) Graph of nuclear/cytoplasmic ratio from confocal imaging indicates Fkh has increased nuclear localization in Cftr deficient guts. n = 12 from 4 independent experiments. P-values were calculated using the Mann-Whitney test in GraphPad prism. Error bars are mean with 95% CI. (D) fkh knockdown rescues the hyperplasia observed in CF model guts. n = 7 (WT; Luc^RNAi), 9 (Cftr^RNAi, Luc^RNAi), 9 (Cftr^RNAi; fkh^RNAi) from 2 independent experiments. (E) fkh knockdown increases intestinal motility as measured by excretion rate in CF model guts. n = 16 vials with 9-13 female flies from 3 independent experiments. (F) fkh knockdown rescues whole body TAG levels at 2 weeks of age in CF model guts. n = 19 (WT; Luc^RNAi), 20 (Cftr^RNAi, Luc^RNAi), 22 (Cftr^RNAi; fkh^RNAi) of 5 pooled females from 4 independent experiments. (G) fkh knockdown does not reduce the amount of glucose remaining in excreta in CF model gut flies. n = 14 (WT; Luc^RNAi), 14 (Cftr^RNAi, Luc^RNAi), 15 (Cftr^RNAi; fkh^RNAi) vials of 15-20 females from 3 independent experiments. (H) fkh knockdown reduces TAG levels in the excreta of CF model gut flies. n = 9 vials of 60 flies from 3 independent experiments. (I) fkh knockdown does not reduce the level of protein in the excreta of CF model gut flies. n = 9 vials of 60 flies from 3 independent experiments. (J) Fkh depletion rescues intestinal barrier function in flies with Cftr deficient enterocytes as assessed by percentage smurf flies at 35 days of age. n = 9 (WT; Luc^RNAi), 10 (Cftr^RNAi, Luc^RNAi), 10 (Cftr^RNAi; fkh^RNAi) vials of 9-13 females from 3 independent experiments. (D-J) P-values were calculated using ordinary one-way ANOVA with Tukey’s multiple comparisons test in GraphPad prism. Error bars are mean with 95% CI. (D-J) Share values for WT; Luc^RNAi with Fig 4A-4G as Ace and fkh knockdown in CF background were performed at the same time and used the same controls. (D-J) RNAi constructs were expressed in enterocytes using the enterocyte specific driver Mex^TS (Mex-Gal4 with the temperature sensitive Gal4 repressor tubulin Gal80^TS) (S4 Fig).

As there is minimal difference in the expression of fkh between WT and CF midguts in our single cell data and that Fkh activity is known to be regulated by nuclear translocation [60,65], we examined Fkh nuclear localization in our model. We used confocal microscopy to image the localization of Fkh in the anterior midguts, the region with highest Ace expression in the Cftr deficient guts. We found increased nuclear localization of the Fkh protein in the nuclei of CF model compared to WT guts consistent with increased Fkh transcriptional activity (Figs 5B-5C and S5G-S5H). Fkh nuclear localization can be inhibited by phosphorylation by mTOR [60,65]. Consistent with increased Fkh nuclear localization there is decreased mTOR activity, measured by phospho-4EBP levels in anterior midguts, in the CF model gut compared to WT guts (S5I-S5J Fig).

If Fkh is regulating Ace transcription, we expect Fkh manipulation to also modulate CF pathologies. Indeed, reducing fkh expression lowered ISC proliferation, increased intestinal motility, increased whole body TAG stores, and increased intestinal barrier integrity in the CF model guts (Fig 5D-5F and 5J). Interestingly, fkh knockdown in the Cftr deficient background does not fully rescue the nutrient malabsorption phenotype as shown by no change in the amount of glucose or protein remaining in the excreta and only partial reduction of the TAG levels in the excreta (Fig 5G–5I). This is the one phenotype tested that does not phenocopy Ace knockdown, likely due to other transcriptional targets of Fkh, such as nutrient transporters, also being downregulated [60]. These results suggest that increased Fkh nuclear localization downstream of CFTR loss of function regulates CF pathophysiology likely through increasing Ace expression.

Discussion

We show that the Drosophila CF intestinal model recapitulates several clinical pathologies of CF, including poor intestinal motility, nutrient malabsorption, reduced whole body energy stores, and decreased intestinal barrier function. Additionally, we performed snRNA-seq on Cftr deficient and WT midguts uncovering an upregulation of acetylcholine esterase (Ace) expression in Cftr deficient enterocytes. This upregulation in Ace results in a decreased sensitivity to acetylcholine in Cftr deficient enterocytes. Importantly, we demonstrate that increasing cholinergic sensitivity in Cftr deficient enterocytes by Ace knockdown rescues several clinical pathologies of CF. Finally, we identify the FOXA1/A2 ortholog Fkh as a putative transcription factor for Ace, that has increased nuclear localization in the Cftr deficient midguts (Fig 6).

Fig 6. Model Figure.

Loss of CFTR function in enterocytes leads to increased Fkh nuclear localization, likely from reduced mTOR activity. Increased Fkh nuclear localization increases Ace transcription. Increased Ace degrades acetylcholine, reducing cholinergic signaling in CF enterocytes. This reduction of cholinergic signaling impairs the maintenance of the intestinal epithelium resulting in increased ISC proliferation, decreased intestinal motility, decreased intestinal barrier function and decreased nutrient absorption. This figure was created in BioRender. Lane, L. (2025) https://BioRender.com/r4x91pu.

Previous work has shown that cholinergic signaling can increase CFTR activity [31,32], but it has not been shown previously that CFTR can affect cholinergic signaling. Here, we demonstrate that loss of CFTR function leads to an increase in Ace expression and a reduction in the guts sensitivity to cholinergic signaling, indicating that there may be reciprocal regulation between cholinergic signaling and CFTR activity. This is interesting as it has been recently demonstrated in Drosophila that cholinergic signaling is required for a return to homeostasis after damage to the gut [27]. This decrease in cholinergic signaling potential may be preventing the gut from returning to homeostasis from the damage being done by loss of CFTR. This may be because the CF gut is constantly being damaged and it may be energetically unfavorable to be constantly returning to normal homeostasis, but further studies are required to explore this hypothesis.

Additionally, we demonstrate that increasing the sensitivity to cholinergic signaling in CF model guts rescues many pathological phenotypes. Although decreasing Ace expression can increase cholinergic signaling in ECs it is possible that some or all of the rescue of CF pathologies with Ace knockdown is not cell autonomous. Ace reduces the levels of acetylcholine in the intercellular space and can affect cholinergic signaling in any nearby cell type that expresses nicotinic receptors, such as the EEs and visceral muscle. Further work will be required to definitively show which cell types are most affected by the increase in Ace expression in the CF model ECs. However, whether decreasing Ace expression in the CF model ECs is working cell autonomously or increasing cholinergic signaling in other cell types, it still improves many CF related phenotypes. Cholinergic signaling is already being investigated as a therapeutic target for diseases of intestinal inflammation such as IBD [29,30,66] which shares many GI symptoms with CF [67,68]. This work suggests AChE inhibition may also be therapeutically beneficial for patients with CF as well.

We also show that Ace transcription downstream of Cftr loss of function is at least in part regulated by an increase in Fkh activity. We provide evidence including Fkh ChIP-seq peaks at the Ace promoter and intronic region and a transcriptional response of Ace to Fkh depletion, that supports direct regulation of Ace by Fkh. However, further work such as mutational analysis would be needed to confirm direct regulation of Ace by Fkh and to fully characterize how Fkh regulates Ace transcription.

Interestingly, the human orthologs of Fkh FOXA1/FOXA2 are known to regulate the expression of CFTR as well as other ion channels important for intestinal cell function [35,36]. Therefore, in the context of ECs with decreased CFTR activity increasing FOXA1/FOXA2 activity could be a compensatory mechanism to try and increase CFTR function. The increase in Fkh activity in the CF model is not due to an increase in fkh transcription but instead a result of increased Fkh nuclear localization. We observe evidence of decreased mTOR activity in the CF model gut which is consistent with increased Fkh nuclear localization [60,65]. However, further studies are required to determine if the decrease in mTOR activity is cell autonomous and directly related to the loss of CFTR function. However, while Fkh activity is required for increased Ace transcription in the Drosophila CF model, it is likely not sufficient to increase Ace mRNA levels. Treatment of flies with rapamycin, which has been demonstrated to increase Fkh nuclear localization [60,65], does not increase Ace mRNA levels after 10 or 20 days of treatment, although it is increased after 50 days, indicating that Fkh activity alone is not sufficient to increase Ace transcription in the midgut [69]. Therefore, other transcription factors may also be important for the increased transcription of Ace observed in the CF model. Additional studies are needed to investigate whether any of the other transcription factors that reduced Ace levels in our screen are important in the context of CF. Finally, it will be important to test if this interaction between CFTR loss of function, increased Fkh activity and increased Ace expression is conserved in mammalian systems, as FOXA1/A2 and AChE could be therapeutically relevant targets.

We performed snRNA-Seq to compare the intestines of WT and the CF Drosophila model. While no unique cell clusters were identified in the CF model, we observed an increase in ISCs and newly differentiated ECs, along with a shift in EC cluster composition. The expansion of the stem cell population aligns with the increased risk of cancer in pwCF [11,13,39,40]. We further investigated the upregulation of Ace in the CF gut, prompted by recent studies highlighting the role of cholinergic signaling in maintaining intestinal epithelial homeostasis in Drosophila [28]. In addition, we identified a substantial number of differentially expressed genes (DEGs) across all cell types, presenting numerous opportunities for future investigation. Notably, many differentially expressed secreted peptides may play a role in inter-organ communication. Together with observed changes in whole-body metabolite levels and lifespan, these findings suggest that the Drosophila CF model may provide valuable insights into how gut dysfunction influences systemic physiology.

Our study further establishes Drosophila as a useful model to study CF. Importantly, CF clinical outcomes are only partially determined by CFTR mutations and are influenced by other genetic modifiers [70–73]. Many of these modifiers have been identified by GWAS studies while others are hypothesized based on interaction with CFTR or other experimental evidence [70–73]. However, in most cases, these putative modifiers have not been validated and studied in detail [70–73]. This work highlights the potential of Drosophila for investigating potential CF genetic modifiers. For example, the fly model can be used to screen which of these potential genetic modifiers alter CF pathophysiology. Understanding how CF modifiers affect physiology could lead to new drug targets that could be beneficial to all CF patients regardless of the type of CFTR mutations. Similarly, Drosophila would be an ideal model for a first pass to screen chemical libraries for potential CF therapeutics as a relatively cheap in vivo model with clinically relevant phenotypes.

Methods

Fly lines

The following fly stocks were used in this study: Drivers from Perrimon lab stocks: Mex1-Gal4 [74] (Mex), Mex1-Gal4 Tubulin-Gal80^TS (Mex^TS), Myo31DF^NP0001-Gal4 (Myo1A-Gal4) [75], Myo1A-Gal4 Tubulin-Gal80^TS(Myo1A^TS).

UAS-RNAi lines from NIG and BDSC Drosophila Stock Centers: UAS-Cftr^RNAi- NIG Stocks: 5789R-1 & 5789R-4, UAS-Ace^RNAi- BDSC 25958, UAS-Luciferase^RNAi-BDSC 31603, UAS-fkh^RNAi- BDSC 27072, UAS-fkh^RNAi #2- BDSC 58059, UAS-Blimp-1^RNAi- BDSC57479, UAS-Ken^RNAi- BDSC-34739, UAS-CF2^RNAi- BDSC 57256, UAS-Jim^RNAi- BDSC 42662, UAS-br^RNAi- BDSC 33641, UAS-Eip93F^RNAi BDSC 57868, UAS-Xrp1^RNAi- BDSC 51054, UAS-nub^RNAi- BDSC 28338, UAS-GATAd^RNAi- BDSC 34640, UAS-Cic^RNAi- BDSC 25995, UAS-Ets21c^RNAi- BDSC 39069, UAS-br^RNAi #2- BDSC 27172, UAS- rel^RNAi- BDSC 28943, UAS-pdm2^RNAi- BDSC 29453, UAS-pdm2^RNAi #2- BDSC 50665, UAS-rel^RNAi #2- BDSC 35661, UAS-CG5953^RNAi- BDSC 57543, UAS-CG5953^RNAi #2- BDSC 57287, UAS-GATAd^RNAi-BDSC 33747, UAS-BTb-vii^RNAi- BDSC 28912, UAS-Trl^RNAi- BDSC 40940, UAS-Trl^RNAi #2- BDSC 41582, UAS-Eip74ef^RNAi- BDSC 29353.

Excretion assay

Flies were fed overnight on lab food with 2 g/100 ml FD&C Blue Dye (12–15 flies per vial for female and 15–20 flies per vial for male). After overnight feed they were moved to 5 ml culture tube without food for 2 hrs (female) or 1 hr 45 minutes (males), time points where the flies have not completely cleared the food from their midguts. Excreta was collected in 400 μl of 0.05% PBST and 100 μl was used in triplicate to measure blue absorbance at 625 nm using the SpectraMax Paradigm Multi-mode microplate reader (Molecular Devices). A standard curve was made with serial dilutions of the FD&C Blue Dye in 0.05% PBS and values were normalized to WT for each experiment.

Whole body metabolite measurements

For each replicate 5 flies (females) or 8 flies (males) were collected in eppendorph tubes containing 1.0 mm Zirconium Oxide Beads (Next Advance Lab Products—ZROB10) and frozen on dry ice and stored at -80C. Samples were moved to ice and 500 μL of ice-cold PBS with 0.05% Triton-X-100(Sigma) was added to the tubes. Homogenization was carried out using a TissueLyser II (QIAGEN) for 2–3 cycles of 30 seconds each, at an oscillation frequency of 30 Hz/s. Tubes were centrifuged for 1 minute at 3500 g to remove debris, and the homogenate was immediately used for glucose, protein, and TAG quantification. Quantifications were performed using a SpectraMax Paradigm Multi-mode microplate reader (Molecular Devices). For protein measurements, 5 μL of the homogenate was mixed with 200 μL of the BCA Protein Assay Kit in triplicate (Pierce BCA Protein Assay Kit), followed by incubation for 30 minutes at 37°C with gentle shaking in 96-well Microplates (Greiner Bio-One) and measuring absorbance at 562nm. Glucose quantification involved mixing 10 μL of homogenate with 100 μL of Infinity Glucose Hexokinase Reagent in triplicate (Thermo Fisher Scientific—TR15421), incubating for 30 minutes at 37°C in 96-well Microplates UV-Star (Greiner Bio-One – 655801) with gentle shaking and measuring at 340nm. For triglycerides, 5 μL of the homogenate was mixed with 150 μL of Triglycerides Reagent in triplicate (Thermo Fisher Scientific—TR22421) and incubated for 10 minutes at 37°C in 96-well Microplates (Greiner Bio-One) with gentle shaking and absorbance was measured at 520 nm. Values were calculated from standard curves using seven serial dilutions (1:2) of BSA (Pierce BCA Protein Assay Kit), glycerol standard solution (Sigma), or glucose standard solution (Sigma). Glucose and TAG levels were normalized to BCA protein levels and then normalized to WT levels for each independent cross.

Glucose in excreta

Flies were fed overnight on food containing 60 g/L glucose, 60 g/L yeast, 40 g/L, 40 g/L cornmeal, 10 g/L agar and 1 g/ L FD&C Blue dye with 15–20 flies per vial. Flies were transferred to 5 ml culture tubes with 200 ul of food containing 60 g/L glucose, 60 g/L yeast, 40 g/L, 40 g/L cornmeal, 10 g/L agar and 1 g/ L FD&C Blue dye for 6 hr. Excreta was collected in 200 μl of 0.05% PBST and transferred to a 96 well microplate (Greiner Bio-One – 655801) and absorbance was measured at 625 nm using the SpectraMax Paradigm Multi-mode microplate reader (Molecular Devices). A standard curve was made with serial dilutions of the FD&C Blue Dye in 0.05% PBS. Glucose quantification was performed as in whole body metabolite measurements and glucose levels were normalized to total excreta amount (625 nm) absorbance before being normalized to the WT condition in each experiment.

TAG and Protein in Excreta

Flies were fed overnight on food containing 60 g/L glucose, 60 g/L yeast, 40 g/L, 40 g/L cornmeal, 10 g/L agar and 1 g/ L FD&C Blue dye with 60 flies per bottle. Flies were transferred to 5 ml culture tubes with 200 ul of food containing 60 g/L glucose, 60 g/L yeast, 40 g/L, 40 g/L cornmeal, 10 g/L agar and 1 g/ L FD&C Blue dye for 4 hr. Excreta was collected in 200 μl of 0.05% PBST and transferred to a 96 well microplate (Greiner Bio-One – 655801) and absorbance was measured at 625 nm using the SpectraMax Paradigm Multi-mode microplate reader (Molecular Devices). A standard curve was made with serial dilutions of the FD&C Blue Dye in 0.05% PBS. TAG quantification was performed as in whole body metabolite measurements section. Protein levels were measured via BCA assay as described in whole body metabolite measurements section. TAG and protein levels were normalized to total excreta amount (625 nm) absorbance before being normalized to the WT condition in each experiment.

Feeding Assay

24hr. Feeding Assay: The amount of food consumed over a 24hr period was measured using the DIETS assay [76]. 25 flies were placed in a vial with an agar plug (as water source) and a small amount of food (100ul) was placed in an eppendorph tube cap. The cap with food was affixed with tape to the side of the fly vial. The flies were acclimatized to the system for 24hrs before being transferred to a new vial with agar plug with a weighed eppendorph tube cap with 100 μl of food. After 24 hrs of consumption the food was weighed again. Empty vials with food caps were used as evaporation controls. Food intake per fly was calculated as (change in weight- evaporation controls change in weight)/ number of flies.

30 minute blue dye feeding assay: Flies were starved overnight before being put on standard lab food containing 1g/100ml FD and C blue dye for 30 minutes. 5 flies with heads removed were homogenized in 600μl of.05% PBST and tubes were centrifuged for 1 minute at 3500 g to remove debris. 100 μl was used in triplicate per replicate to measure blue absorbance at 625 nm using the SpectraMax Paradigm Multi-mode microplate reader (Molecular Devices). A standard curve was made with serial dilutions of the FD&C Blue Dye in 0.05% PBS and values were normalized to WT for each experiment.

Lifespan assay

15 (females) or 20 (males) 2–4 day old adult flies were placed in vials on standard laboratory food. Flies were flipped every other day and vials were scored for percent survival.

Smurf assay

Smurf assay was performed as previously described [57]. Briefly 10–15 female flies were raised on standard fly food and flipped every 2 days for 33 days before being put on standard fly food supplemented with 2.5 g per 100 ml of FD & C blue dye for 2 days. At 35 days each vial was scored for percentage smurf flies under a dissection microscope.

RNA isolation, RT and quantitative PCR

For RNA isolation, 10–15 adult guts were dissected, and tissues were homogenized in TRIzol reagent (Ambion). The supernatant was collected and processed using Direct-zol RNA MicroPrep columns (Zymo Research) following the manufacturer’s protocol. Reverse transcription was performed using the iScript cDNA Synthesis Kit (Bio-Rad). Quantitative real-time PCR (qRT-PCR) was carried out on a CFX96 Real-Time System (Bio-Rad) using iQ SYBR Green Supermix (Bio-Rad). qRT-PCR reaction volume used was 10 µl (2 µl 5 µM Primer pair mix + 5 µl 2x SYBR Green+ 3 µl cDNA). Relative mRNA levels were determined using the ΔΔCt method, with mRNA levels normalized to Rpl32.

qPCR primers:

Gene	Forward	Reverse
Ace	AGGTGCATGTCTACACGGG	ACGTTGGTGTTGGGGTTCC
nAchRa3	ATGAAGTGGTTTCAAGTGACCAT	CAAATCGTCGTACAATCGTTTCG
nAchRa4	TGGGTGTGGACCTATCTGAAT	AAGTATGGTTCGTCGCAACAA
nAChRa5	CAGCAACTCACAACACTGCAA	CCGTGGTTGGCTACATCCTC
nAchRa7	ATGCTGGTCTATGGCCTGG	GAGTAGCCGCTTCTCATGGG
nAchRb1	TGGAGTCTTCCTGCAAATCCT	GCCACAAGCACCAGGATG
nAchRb3	ATGACGACGACTCCCAAGATA	AAGAAGCATCCCCATTAGCATTT
Rpl32	AGCATACAGGCCCAAGATCG	TGTTGTCGATACCCTTGGGC

Single-nuclei RNA-seq

Single-nucleus suspension and FACS: Single-nucleus suspension was conducted as previously described [47]. Briefly, 70 guts per condition were dissected in cold Schneider’s medium, flash-frozen and stored at -80°C. Prior to FACs sorting, samples were spun down and Schneider’s medium was exchanged with homogenization buffer [250mM Sucrose, 10mM Tris pH8, 25mM KCl, 5mM MgCl, 0.1% Triton-X, 0.5% RNasin Plus (Promega, N2615), 50x protease inhibitor (Promega G6521), 0.1mM DTT]. Using 1ml dounce (Wheaton 357538), nuclei were released by 20 loose pestle strokes and 40 tight pestle strokes while keeping samples on ice and avoiding foam. Next, nuclei were filtered through 5 ml cell strainer (40 μm), and using 40 μm Flowmi (BelArt, H13680-0040). Nuclei were centrifuged, resuspended in PBS/0.5%BSA with 0.5% RNase inhibitor, filtered again with 40 μm Flowmi and stained with DRAQ7 Dye (Invitrogen, D15106). Single nuclei were sorted with Sony SH800Z Cell Sorter at PCMM Flow Cytometry Facility at Harvard Medical School and 100k nuclei per sample were collected in PBS/BSA buffer.

10x genomics and sequencing: Single nuclei RNA-seq libraries were prepared using the Chromium Next GEM Single Cell 3’ Library and Gel Bead Kit v3.1 according to the 10xGenomics protocol. Approximately 16,500 nuclei were loaded on Chip G with an initial concentration of 700 cells/µl based on the ‘Cell Suspension Volume Calculator Table’. Sequencing was conducted with Illumina NovaSeq 6000 at Harvard Medical School Biopolymers Facility.

10x data processing: Raw sequencing data were aligned to the Drosophila melanogaster reference genome (BDGP6.32, Ensembl release 104) using 10x Genomics Cell Ranger v7.1.0. Low quality nuclei with less than 500 Unique Molecular Identifiers (UMI) were filtered out. The default clustering and uniform manifold approximation and projection (UMAP) generated by 10X Loupe Browser was used to cluster cells. Cell types were manually annotated using the top marker genes from each cluster. The filtered count matrix and metadata were loaded into R (v4.4.0) and further processing was performed using the default Seurat workflow (v5.2.1). Differential gene expression analysis was performed using a Mann Whitney-U test with a cut-off of p-value < 0.05 and abs(Log2FC) >.5 and percent expression > 1%.

Normalized expression violin plots and heatmaps were generated using Seurat, ggplot2 (v3.5.2), ComplexHeatmap (2.24.0) and pheatmap (v1.0.12). hdWGCNA (v0.4.05) was used to create metacells for which Pearson correlation tests were performed for candidate transcription factors and Ace expression.

Cell type comparison between Drosophila and mammalian intestinal epithelium

We used DIOPT (release 7, score>3) [77] to map mouse genes to Drosophila orthologs. The Drosophila orthologs of markers identified in various cell types from mammalian datasets were grouped respectively as cell type specific gene sets. Next, we compared the cell type specific gene sets from the mammalian study [48] and the cell type markers from Drosophila [46]. P-value of enrichment was calculated based on the hypergeometric distribution and the similarity of Drosophila markers with mammalian markers is reflected by the negative log10 of the P values.

GCaMP7c live imaging

GCaMP7c live imaging was performed as previously described [27]. Guts were dissected in fresh HL3 buffer (1.5mM Ca2 + , 20mM MgCl2, 5mM KCl, 70mM NaCl,10mM NaHCO3, 5mM HEPES, 115mM Sucrose, 5mM Trehalose) and placed in eight-well clear bottom cell culture chamber slides with HL3. Guts were stabilized with a Nylon mesh (Warner instruments, 64–0198) and paper clips cut in small identical pieces. ACh and Nicotine sensitivity assay: ACh and Nicotine sensitivity assay was done using LSM780 and LSM980 microscopes with 40x water objective lens. Each frame (~2.5 sec/frame) is the maximum projection of 5–6 z-stacks (2.96µm/z) and was acquired with 488nm excitation for GFP. 5mM Acetylcholine (Acetylcholine Chloride, Sigma A2661) or 1.33 mM Nicotine (Sigma, N3876) were added at the 10th frame (~25 sec). For Fig 3B-3C an old aliquot of Acetylcholine was used at 10mM after testing responses in WT midguts to a range of concentrations. When a new aliquot of Acetylcholine was purchased, a range of concentrations were tested in WT gutsand 5mM Acetylcholine gave a similar response in WT guts as10mM of old Acetylcholine aliquot. Therefore for Fig 3D and 3E 5mM Acetylcholine was used.  All images were taken from similar areas in the anterior midgut between R1-R2. Fiji was used for assembly and calculation of fluorescence per frame. DF/F0 = Ffr-F0/ F0. Ffr is the fluorescence per frame and F0 (baseline fluorescence) is the average fluorescence intensity of the first 9 frames (fr1-fr9).

PH3 + counts

Guts were dissected and fixed in 4% PFA for 20 minutes. They were washed 3x in 0.1% PBST and incubated for 1 hr at RT in 5% Normal Goat Serum in 0.1%PBST. Guts were incubated for overnight with rabbit anti-pH3 (Millipore #06–570; 1:3000). Guts were then washed for 30 minutes 3x in 0.1% PBST then incubated for 2hrs at RT with secondary antibody, Alexa Fluor 555-conjugated goat anti-rabbit IgG (1:200) and DAPI (1:3000). The guts were then washed 3X in 0.1%PBST for 30 minutes and mounted in VECTASHIELD antifade mounting media (Vector Laboratories). Slide genotype was blinded and PH3 + nuclei were counted with an epifluorescence microscope.

Fkh nuclear localization

Guts were dissected and fixed in 4% PFA for 20 minutes. They were washed 3x in 0.1% PBST and incubated for 1 hr at RT in 5% Normal Goat Serum in.1%PBST. Guts were incubated for 2 days at 4C with Rab-anti-FKH [60,78–80] (1:100)(gift from Dr. Chandrasekaran, St. Mary’s College of California) in 5% Normal Goat Serum in 0.1%PBST. Guts were then washed for 30 minutes 3x in 0.1% PBST then incubated overnight in secondary antibody, Alexa Fluor 555-or 647 conjugated goat anti-rabbit IgG (1:200) and DAPI (1:3000). The guts were then washed 3X in 0.1% PBST for 30 minutes and mounted in antifade mounting media. Fluorescence was imaged using W1 Yokogawa spinning disk, Nikon inverted Ti2 confocal microscope. Percent nuclear localization was calculated on maximum intensity projections in Fiji.

Phospho-4EBP1 Staining

Guts were dissected and fixed in 4% PFA for 30 minutes. They were washed 3x in 0.1% PBST and incubated for 1 hr at RT in 5% Normal Goat Serum in.1%PBST. Guts were incubated for 2 days at 4C with Rab-anti-p4EBP1 (1:800) (CST 2855S) in 5% Normal Goat Serum in 0.1%PBST. Guts were then washed for 30 minutes 3x in 0.1% PBST then incubated overnight in secondary antibody, Alexa Fluor 555-or 647 conjugated goat anti-rabbit IgG (1:200) and DAPI (1:3000). The guts were then washed 3X in 0.1% PBST for 30 minutes and mounted in antifade mounting media. Fluorescence was imaged using W1 Yokogawa spinning disk, Nikon inverted Ti2 confocal microscope. Average staining intensity was calculated on maximum intensity projections in Fiji.

Statistics and Reproducibility

Prism (https://www.graphpad.com/scientific-software/prism/) or ggplot2 was used to create graphs as well as to perform statistical analysis. Statistical tests are indicated in figure legends. No statistical analysis was conducted to determine sample sizes. Randomization was not performed. Blinding experiments were conducted during mitotic division counts. No data were excluded. Confocal images shown in figures are representative images of two to 4 independent experiments with similar result.

Supporting information

S1 Fig. Related to Fig 1.

(A) CF model guts have decreased excretion rate compared to WT guts as measured by the amount of excreta collected over a 1.75 hr time period. n = 13(WT), 15 (Cftr^RNAi) vials of 10–15 males from 3 independent crosses. (B) CF model guts have increased glucose in excreta compared to WT flies. n = 11 (WT), 12 (Cftr^RNAi) vials of 15–20 males from 2 independent crosses. (C-F) CF model guts have reduced whole body energy stores to WT flies. (C) CF model guts have reduced TAG levels at 2 weeks of age compared to WT flies. n = 24 of 8 pooled males from 4 independent crosses. (D) Male CF model gut flies have no significant difference in whole body glucose levels at 2 weeks of age compared to WT flies. n = 23 (WT), 22 (Cftr^RNAi) of 8 pooled males from 4 independent crosses. (E) CF model gut flies have reduced whole body TAG levels at 4 weeks of age compared to WT flies. n = 19 of 8 pooled males from 4 independent crosses. (F) CF model gut flies have reduced whole body glucose at 4 weeks of age compared to WT flies. n = 24 (WT), 23 (Cftr^RNAi) of 8 pooled males from 4 independent crosses. (G-H) Decreased whole body metabolites are not due to a developmental defect in CF model guts. (G) Whole body TAG levels are reduced in CF model guts compared to WT when Cftr knockdown is induced by temperature shift in 1–2 day old adults. n = 6 of 5 (female) or 8 (male) pooled flies. (H) Whole body glucose levels are reduced in CF model guts compared to WT when Cftr knockdown is induced by temperature shift in 1–2 day old adults. n = 6 of 5 (female) or 8 (male) pooled flies. (I) CF model gut flies eat a similar amount of food over a 24 hr period as WT flies. n = 9 vials with 25 flies from 2 independent experiments. (J) CF model gut flies eat a similar amount of food in the 30 minutes after starvation as WT flies. n = 8 of 5 pooled females. (A-J) P-values were calculated using the Mann-Whitney test in Graphpad prism. Error bars are mean with 95% CI. (K) CF model gut flies have reduced lifespan compared to WT flies. n = 11 vials with 10–15 males from 2 independent crosses. (A-F, I-K) WT is Myo1A > + and Cftr^RNAi is Myo1A > Cftr^RNAi. (G-H) WT is Myo^TS > + and Cftr^RNAi-TS is Myo^TS > Cftr^RNAi.

(TIF)

S2 Fig. Related to Fig 2.

(A) Comparison of fly gut cell types and mammalian intestine cell types using marker genes identified. (B) Differences in cell type composition between WT and CF guts illustrated as a stacked bar chart depicting percentage of cell belonging to each cell cluster in snRNA-seq data for WT and CF model guts. (C-D) Many secreted peptides are differentially expressed between WT and CF guts across cell clusters. (C) Heat map of the differentially expressed secreted proteins. Color reflects the log2fold change of expression in CF model guts comparing to control in each cell type. Gene names of differentially expressed genes can be found in S2 File. (D) Violin plots of expression of secreted peptides important for Yki^act gut tumor physiology (Itp, Impl2, upd3, and pvf1) in snRNA-seq data.

(TIF)

S3 Fig. Related to Fig 3.

(A-B) Ace expression is upregulated Cftr deficient midguts compared to WT midguts in snRNA-seq data. (A) Ace expression in each cell in WT and Cftr deficient guts plotted onto UMAP (B) Violin plots of Ace expression in each cell cluster identified in snRNA-Seq data set. (C) Ace expression is increased in Cftr deficient guts compared to WT guts via RT-qPCR analysis of whole guts. n = 4 of 10–15 pooled guts from 1 independent experiment. (D) Representative images of GCaMP7c fluorescence, in WT (Mex^TS > GCaMP7c, LucRNAi) and Cftr deficient guts (Mex^TS > GCaMP7c, Cftr^RNAi) before (T15s) or after addition of Nicotine (T125). Scale bars are 50 μm. (E) Graph of average relative fluorescent intensity, ΔF/F0, per frame (2.5s per frame) and genotype. n = 8 (WT) and 7 (Cftr^RNAi) from 3 independent experiments. Error bars are mean + /- SEM and pValue was calculated using the Mann-Whitney test in Graphpad prism. (F) Average normalized expression of all nicotinic receptors in WT and CF single nuclei RNA-seq data set. (G) Heat map of average expression of nicotinic receptor subunits in WT and CF midguts from snRNA-seq data set. (H) Relative expression of indicated nicotinic receptor subunit in WT and CF model guts as assessed by RT-qPCR. n = 6–7 replicates of 10 pooled guts from 2 independent experiments. pValues were calculated using ordinary one-way ANOVA with Tukey’s multiple comparisons test in GraphPad prism. Error bars are mean + /- SD (I) Relative Ace expression in guts of indicated genotype. n = 6 replicates of 10 pooled guts from 2 independent experiments. P-values were calculated using ordinary one-way ANOVA with Tukey’s multiple comparisons test in GraphPad prism. Error bars are mean with 95% CI.

(TIF)

S4 Fig. Related to Fig 3–5.

Experimental set up for Figs 3D-3E, 3I, 4, and 5D-5J. Mex^TS (Tub-Gal80^TS, Mex-Gal4) virgin females were crossed to males with the indicated UAS-RNAi constructs and raised at 18C, nonpermissive temperatures. 1–3 day old adults were moved to 29C for 2 weeks (35 days for smurf assay) and were than used in indicated assays. Drosophila image from bioicons licensed under CC0.

(TIF)

S5 Fig. Related to Fig 5.

(A) Violin plots of expression of candidate Ace transcription factors in anterior EC cell clusters from snRNA-seq data. (B) Diagram of experimental set up for Ace transcription factor screen. Myo^TS (Tub-Gal80^TS, Myo1A-Gal4); UAS-Cftr^RNAi females were crossed to male flies with UAS-RNAi against candidate transcription factor and raised at 18C, nonpermissive temperatures. 1–3 day old adults were moved to 29C for 2 weeks and guts were dissected to assay for Ace expression levels. (C) Scatter plots showing correlation of Ace and candidate transcription factor expression levels in meta-cells from the 05-aEC2 CF cell cluster (Cell cluster with overall highest Ace expression). (D) Data from the ChIP–seq database indicating enrichment of Fkh at the promoter of Ace. (E) JBrowse viewer of Ace gene with Janelia Gal4 line putative brain enhancer regions and REDfly transcriptional regulatory regions. (F) Relative Ace expression in guts of indicated genotype via RT-qPCR. n = 7 replicates of 10 pooled guts from 2 independent experiments. (G) Representative images of decreased Fkh nuclear staining in CF model guts with fkh RNAi. scale bars 25μm. (D) Quantification of decreased Fkh staining in CF model guts with fkh RNAi. n = 5 from 1 independent experiment. P-values were calculated using the Mann-Whitney test in GraphPad prism. Error bars are mean with 95% CI. (I) Representative images of decreased phospo-4EBP1 staining in CF anterior midguts compared to WT. scale bars 25μm. (J) Quantification of phospho-4EBP1 staining in WT and CF deficient guts. n = 10 from 2 independent experiments. p values were calculated using the Mann-Whitney test in GraphPad prism. Error bars are mean with 95% CI. Drosophila image in panel B is from bioicons licensed under CC0.

(TIF)

S1 Table. TF2TG.

(PDF)

S1 File. Differentially expressed genes in sn-RNA-seq.

(XLSX)

S2 File. Differentially expressed secreted peptide genes in sn-RNA-seq.

(XLSX)

S3 File. Average expression nicotinic receptors in sn-RNA-seq.

(XLSX)

S1 Data. Numerical data underlying main figure graphs.

(XLSX)

S2 Data. Numerical data underlying supplementary figure graphs.

(XLSX)

Data Availability

The scRNA-seq portal for data mining is available at https://www.flyrnai.org/scRNA/. Raw RNA-seq reads and processed data files have been deposited in the NCBI Gene Expression Omnibus (GEO) database under accession code: GSE300597.

Funding Statement

This work was funded by NIH F32 Ruth-Kirschstein Postdoctoral Fellowship (NIDDK) 5F32DK130290-03 (https://www.niddk.nih.gov/), CF Foundation Postdoctoral Research Fellowship Award LANE21F0-00816F221(https://www.cff.org/) which provided salary for E.A.L.. N.P. is an investigator of the Howard Hughes Medical Institute (https://www.hhmi.org/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.