1. Developmental Biology
  2. Stem Cells and Regenerative Medicine
Download icon

Multiscale analysis reveals that diet-dependent midgut plasticity emerges from alterations in both stem cell niche coupling and enterocyte size

  1. Alessandro Bonfini
  2. Adam J Dobson
  3. David Duneau
  4. Jonathan Revah
  5. Xi Liu
  6. Philip Houtz
  7. Nicolas Buchon  Is a corresponding author
  1. Cornell Institute of Host-Microbe Interactions and Disease, Department of Entomology, Cornell University, United States
  2. Institute of Molecular, Cell and Systems Biology, University of Glasgow, United Kingdom
  3. Université Toulouse 3 Paul Sabatier, CNRS, UMR5174 EDB (Laboratoire Évolution & Diversité Biologique), France
  4. Instituto Gulbenkian de Ciência, Portugal
Research Article
  • Cited 0
  • Views 1,153
  • Annotations
Cite this article as: eLife 2021;10:e64125 doi: 10.7554/eLife.64125

Abstract

The gut is the primary interface between an animal and food, but how it adapts to qualitative dietary variation is poorly defined. We find that the Drosophila midgut plastically resizes following changes in dietary composition. A panel of nutrients collectively promote gut growth, which sugar opposes. Diet influences absolute and relative levels of enterocyte loss and stem cell proliferation, which together determine cell numbers. Diet also influences enterocyte size. A high sugar diet inhibits translation and uncouples intestinal stem cell proliferation from expression of niche-derived signals, but, surprisingly, rescuing these effects genetically was not sufficient to modify diet’s impact on midgut size. However, when stem cell proliferation was deficient, diet’s impact on enterocyte size was enhanced, and reducing enterocyte-autonomous TOR signaling was sufficient to attenuate diet-dependent midgut resizing. These data clarify the complex relationships between nutrition, epithelial dynamics, and cell size, and reveal a new mode of plastic, diet-dependent organ resizing.

Introduction

Nutrition is a principal determinant of animal health and fitness, affecting aging, metabolic disease, and fecundity (López-Otín et al., 2016). Understanding how diet impacts physiology has broad societal implications, especially at a time when the average body mass index of human populations is continuously increasing and weight is a critical risk factor for disease (Finkelstein, 2014; Kebede and Attie, 2014). The impact of diet on health is evolutionarily conserved: in Drosophila melanogaster and other organisms such as mice and zebrafish, dietary restriction has been extensively studied, revealing an inverse relationship between lifespan and caloric content, as well as fecundity (Piper and Bartke, 2008; Simpson and Raubenheimer, 2012). Interestingly, the relationship of lifespan and reproduction has been shown to depend not only on quantity (i.e., calories), but also on the relative proportions of certain nutrients (e.g., amino acid imbalance; Grandison et al., 2009; Solon-Biet et al., 2019; Solon-Biet et al., 2014). This suggests that organismal physiology is influenced not only by quantity of food, but also by qualitative changes to dietary composition (Piper et al., 2017).

For animals, the gut is the sole interface with ingested food, and as such is an important regulator of organismal physiology. Drosophila is no exception: most digestion and absorption occur in the endoderm-derived midgut. The Drosophila midgut is a regionalized, tubular epithelial monolayer sheathed in visceral muscles (Demerec, 1950) and is akin in function to the mammalian intestine (Apidianakis and Rahme, 2011; Liu et al., 2017). In Drosophila, five main midgut regions have been described, which can be grouped for simplicity as an anterior midgut dedicated to digestion (regions 1 and 2), an acidic middle midgut (region 3), and a posterior midgut specialized for absorption (regions 4 and 5) (Buchon et al., 2013). The midgut epithelium is mostly composed of a population of mature absorptive cells called enterocytes (ECs), which is maintained by proliferative intestinal stem cells (ISCs) differentiating through a transient enteroblast (EB) phase. ISCs can also differentiate into enteroendocrine cells (EEs) through a pre-EE progenitor phase (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006; Zeng and Hou, 2015). The ISCs are distributed throughout the epithelium and have the ability to divide either asymmetrically (giving rise to one new ISC and one differentiated cell) or symmetrically (resulting in two identical progeny; de Navascués et al., 2012; Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2007). Multiple pathways orchestrate ISC proliferation, either in response to changes in the nutritional environment or following infection-, abrasion-, or chemical-derived stresses (Bonfini et al., 2016; Buchon et al., 2009b; Buchon et al., 2009a; Buchon et al., 2010; Jiang et al., 2009). The TOR and insulin pathways regulate ISC proliferation and the growth of EBs and ECs on nutrient-rich diets (Amcheslavsky et al., 2011; Choi et al., 2011; Haller et al., 2017; Kapuria et al., 2012; H-J et al., 2015; O’Brien et al., 2011; Strilbytska et al., 2017; Wen et al., 2017). At the onset of the regenerative response, cytokines (Unpaired 2 and 3 [Upd2/3]) are released, resulting in the secretion of epithelial growth factors (e.g., Vein or Vn) by visceral muscles. Vein and other EGFs, together with EB-derived Wingless (Wg), initiate ISC proliferation via activation of the EGFR, JAK-STAT, and Wnt pathways (Biteau and Jasper, 2011; Buchon et al., 2009a; Jiang et al., 2009; Zhou et al., 2013). Under homeostatic conditions, the activity of these pro-mitotic pathways in ISCs is thought to be coupled with the expression levels of ligands secreted from the niche, ultimately determining ISC behavior (Liang et al., 2017). Both the frequency and the type (symmetrical vs. asymmetrical) of ISC mitosis are thought to underlie the dynamic response of the midgut tissue to diverse physiological conditions. To contextualize these signaling pathways, we must understand how their regulation varies in distinct physiological conditions – such as on different diets – and characterize how their variable outputs at the cellular level scale to the growth of specific regions and the whole gut.

Organ growth has been studied since the 1930s (Penzo-Méndez and Stanger, 2015; Twitty, 1930; Twitty and Schwind, 1931), mostly in the context of development. In Drosophila, the study of imaginal discs (Bryant and Levinson, 1985; Gokhale and Shingleton, 2015; Neufeld et al., 1998) revealed that both cell number and cell size contribute to the final size of the adult organ (Neufeld et al., 1998). In imaginal discs, cells divide and increase in number until the organ reaches a set size. Strikingly, in this system cell growth is dominant over, and compensates for, cell division defects (Neufeld et al., 1998). However, not all modes of growth regulation can be understood based on these developmental models. For instance, adult organs such as the intestine can still grow in adults but can also shrink in response to stimuli: the adult midgut can reshape itself, with bouts of organ shrinkage followed by regrowth in response to both damage (Buchon et al., 2010; Buchon et al., 2009b; Buchon et al., 2009a; Jiang and Edgar, 2009) and nutrient availability (O’Brien et al., 2011). Similar adaptive intestinal recovery after fasting and refeeding has been described in vertebrate models, highlighting the strong evolutionary conservation of the intestine’s functional response to nutrient availability (Tamaoki et al., 2016; ÖH et al., 2012). As the midgut epithelium faces routine shedding of epithelial cells, regulation of adult midgut size fundamentally differs from the imaginal disc because its size can be determined by not only gain but also loss of cells. Therefore, to complete our understanding of how organ size is regulated, we must integrate new information on the balance of cell gain and loss, and the Drosophila midgut is an ideal model to investigate these questions.

Epithelial dynamics in the Drosophila midgut are described by two complementary models that capture the turnover of cells. A first model assumes that, in homeostatic conditions, the constant loss of mature epithelial cells is compensated by ISC proliferation, which is regulated by a feedback loop of pro-mitotic signals from dying ECs and the ISC niche (Buchon et al., 2009b; Buchon et al., 2009a; Jiang et al., 2009; Liang et al., 2017; Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2007). In this model, cell gain is mechanistically coupled to cell loss, ensuring homeostasis. A second model allows for the midgut to respond to the presence of food. Such adaptive growth, presumably underpinned by changes in ISC activity and their mode of division (O’Brien et al., 2011), is thought to balance the costs of maintaining a midgut against the benefits of acquiring nutrients from food. This model implies that the coupling between cell loss and ISC proliferation is not maintained, thus allowing for changes in cell number and organ size. To date, evidence for each of these two models has relied on the assumption that measurements of ISC proliferation in specific regions capture epithelial dynamics throughout the whole midgut, and consequent regulation of organ size. In addition to epithelial dynamics, changes in EB and EC ploidy in response to diet or infection have been reported, but not integrated into models of midgut growth (Choi et al., 2011; Xiang et al., 2017). Since ploidy often correlates with cell size, and most epithelial mass comprises ECs, these findings suggest that EC size could be an additional factor determining midgut size. To fully characterize epithelial cell dynamics, global organ-scale measurements of both cell gain and loss, and how they vary, are needed.

In this study, we characterize the response of the midgut to dietary variation, integrating the behavior of stem cells and ECs into a unified model, and scaling from molecular and cellular effects to resizing of the whole organ. We show that qualitative dietary variation can regulate midgut size, determined by opposing effects of sugar and a panel of other nutrients. We also outline novel mechanisms that regulate midgut size, showing that a high-sugar, low-yeast diet decouples stem cell proliferation from pro-mitotic niche signals by inducing translational blockage. Organ-level quantifications of cell gain, cell loss, and cell size indicate that midgut resizing is an emergent property of these three aspects of epithelial dynamics. However, we find that the main driver of midgut size in response to nutrient quality is in fact EC size, which is regulated by autonomous TOR signaling, and can even compensate for deficiencies in stem cell proliferation. Altogether, these findings provide a new, integrative perspective on the environmental, cellular, and molecular regulation of tissue homeostasis.

Results

Diet composition affects overall size and regional allometry of the midgut

It was previously described that the Drosophila midgut requires food in order to properly develop after eclosion (Choi et al., 2011; O’Brien et al., 2011). Diet composition can also affect the size of the adult midgut (Ponton et al., 2015). As D. melanogaster feeds naturally on rotten fruits (Kohler, 1993), we manipulated sucrose and yeast, which are widely used in laboratory Drosophila media and considered representative of natural nutrient sources. To ensure we measured only responses to the adult diet and not developmental differences, larvae were raised on a common pre-experiment diet before being moved to experimental diets (Figure 1A). We chose two isocaloric diets, which differed only in the relative abundance of ingredients (Figure 1B), based on their degree of difference in midgut size despite being isocaloric (Ponton et al., 2015). A detailed description of the recipe for these diets (and all other diets used in this study) can be found in Supplementary file 1. One diet was rich in sucrose and low in yeast (high sugar [HS] diet), while the other was rich in yeast and low in sugar (high yeast [HY] diet) (Figure 1B). The midguts of wild-type mated female Canton-S (Cs) flies feeding on the HY diet were on average 31% longer (Figure 1C–E, Video 1) and 44% wider than the midguts of flies feeding on the HS diet (Figure 1—figure supplement 1A). We found that width measurements were variable along the midgut and often affected by the volume of the internal bolus and microscope slide compression, so we decided to focus on midgut length as proxy of overall size. For reference, compared to two ‘standard’ diets in the field, midgut size of flies on the HY diet resembled in size Bloomington cornmeal and Bloomington molasses (Figure 1—figure supplement 1B). We also tested whether differences in midgut length depended on sex and mating status since midgut size and stem cell behavior are affected by these parameters (Ahmed et al., 2020; Reiff et al., 2015; White et al., 2021). Indeed, we found that guts of unmated females and males were significantly less different between HS and HY diets than the ones of mated females (Figure 1—figure supplement 1C), so we continued to work with mated females in subsequent experiments. Since diet composition affects feeding rate (Carvalho et al., 2005), we asked if an increased food intake, and thus calories, could explain the observed increase in midgut length on the HY diet. Flies on HS fed more than flies on HY, suggesting that growth on HY is not due to increased ingestion (Figure 1—figure supplement 1D). Diet also has the potential to affect midgut microbes, which regulate midgut homeostasis (Broderick et al., 2014; Buchon et al., 2009a). We therefore repeated our experiments in germ-free conditions, finding that microbes were not required for the response of midgut length to diet (Figure 1—figure supplement 1E). These experiments demonstrate that dietary composition affects midgut size independently of gut microbes and the caloric content of diet.

Figure 1 with 3 supplements see all
Diet composition affects size and regional allometry of the midgut.

(A) Illustration of general dietary treatment design. Flies were reared on pre-experiment diet during development. At eclosion, flies were allocated to either high sugar (HS) or high yeast (HY) before midgut dissection at 5 days post eclosion. (B) Nutritional composition (proteins, carbohydrates, and lipids) of the two isocaloric diets used as a basis for this study as calories per liter of food: enriched in sugars (HS) or yeast (HY). (C–E) Canton-S (Cs) flies fed on HS diet (C) have shorter midguts than flies on HY (D). Quantification of midgut length for HS vs. HY at 5 days post eclosion (E). (F) Midgut length response to diet is strongly variable across the Drosophila Genetic Reference Panel (DGRP), with HY being generally longer than HS (i.e., the ratio length on HY/length on HS is between 1 and 1.4). (G) Midgut resizing is allometric between regions of the midgut. Posterior midguts of flies fed HY diet exhibit a greater increase than anterior regions. For the violin/dot plots shown in this figure, white dots represent single midgut measurements. Lozenges represent mean of repeats. Violin plots are color coded according to diets (HS = red, HY = light blue throughout the article). Numbers in parentheses at the bottom of charts indicate sample size. Additional information on the statistics can be found in Supplementary file 2. Scale bars are 500 µm for all images.

Video 1
Nano CT scan and rendering of internal organs.

Video shows flies kept on high-sugar (HS) diet on the left, and flies kept on high-yeast (HY) diet on the right. At 5 s, a rendering of the ovaries appears in blue, showing increased size on HY diet. At 10 s, a rendering of the midgut (anterior and middle in light blue, posterior in green, and hindgut in purple) appears, showing increased size of the midgut on HY diet.

In nature, most traits vary in magnitude due to genetic variation and because of genotype-by-environment variation (Albert and Kruglyak, 2015; Timpson et al., 2018). It is also known that individuals in a population vary in their physiological response to diet (Garlapow et al., 2015; Jehrke et al., 2018; Uchizono and Tanimura, 2017). We therefore wondered whether the effect of diet composition on midgut size that we saw with the Canton-S strain was variable and generalizable. We took the genetically diverse lines that constitute the Drosophila Genetic Reference Panel (DGRP, Mackay et al., 2012) and measured the impact of the HS and HY diets on midgut size, with three iterations of full experimental replication. In 184 out of 188 lines tested, flies feeding on HY had qualitatively larger midguts than flies feeding on HS (HY/HS ratio higher than 1, Figure 1F). The magnitude of response was genetically variable, with the fold-increase in size varying from 0.98 to 1.47 (median = 1.18). The response to diet was statistically significant in 132 lines, and we note that the 56 nonresponsive lines were found throughout the size distribution on either HS or HY. We found that ~15% of the variance in response to the diet was explained by genetic variation, which was lower than what was found for body size plasticity (between 33% and 52%; Lafuente et al., 2018). Collectively, these results confirmed that an elevated yeast to sucrose ratio generally promotes midgut size, but that this is a quantitatively variable trait (Figure 1F).

The digestive tract comprises multiple distinct and functionally specialized regions (Buchon et al., 2013; Dutta et al., 2015; Marianes and Spradling, 2013), as depicted in Figure 1—figure supplement 2A. We tested whether the diet-dependent change in midgut size acts uniformly on all regions of the Drosophila midgut by measuring the length of anterior, middle, and posterior midguts in the DGRP. Overall, all midgut regions were longer on the HY diet than the HS diet (Figure 1—figure supplement 2B). However, change in posterior midgut length predicted change in total length with 80% accuracy (i.e., R2 = 0.8), while anterior and middle midgut explained 60 and 40% of total midgut length change, respectively (Figure 1—figure supplement 2B). In addition, the posterior midgut was more responsive to food than the anterior midgut or the entire midgut (Figure 1—figure supplement 2B). These results suggest that different midgut regions resize to a variable extent, and that diet composition affects the allometry of midgut regions. By quantifying the relative proportions of the anterior and posterior midguts of flies feeding on the HS and HY diets, we found that the length of the anterior midgut exceeds that of the posterior on the HS diet (Figure 1G). However, on the HY diet, the lengths of both regions are close to equal (Figure 1G). This suggests that the posterior midgut is more consistently responsive to changes in diet composition than the anterior midgut and that the relative proportions of the midgut regions change with diet. Altogether, our data demonstrate that diet composition affects the allometry of midgut regions, and that resizing is subject to genetic variation.

DGRP lines are fully genome-sequenced, with publicly available data on genetic polymorphisms (Mackay et al., 2012). We therefore sought to identify the genetic determinants underlying population variation in diet-dependent midgut resizing. We tested the association of the change in midgut length on HS and HY diets with ~1.9M genetic variants. In total, we identified 638 loci as strongly associated with response to diet (p<10–9), including loci mapped in genes coding for proteins associated with cell junctions (intronic variants in CadN, Nrg, and Magi), cell division (5′ UTR variant in insc and modifier in slik), epigenetic regulation (missense variant in Su(var)2-HP2), and growth/differentiation (intronic variant in tkv) (Figure 1—figure supplement 3A). All these processes have been associated with tissue turnover, suggesting that changes in midgut length could result from altered cell dynamics in the midgut (Chen et al., 2020; Hung et al., 2020; Izumi et al., 2012; Li et al., 2013a; Li et al., 2013b; Ma et al., 2019; Tian et al., 2017). Thus, natural variation in the response of midgut size to dietary changes maps to genes involved in functions of probable relevance to organ growth.

Sugar opposes yeast-induced increase in midgut size

Our results demonstrated that diet composition influences midgut size; however, it remained unclear whether the amount (or lack) of yeast and/or sucrose is responsible for this change or whether the relative proportions of sucrose and yeast were responsible. To answer this question, we utilized a nutritional geometry approach (Simpson and Raubenheimer, 1995), which enables one to separate phenotypic impacts of relative versus total nutrient availability. We systematically varied the amounts of yeast and sucrose across 28 diets, studying five different ratios and four different caloric levels, with additional diets in points of interest in the diet space (Figure 2—figure supplement 1A). Nutritional geometry revealed yeast as a major driver of midgut length: increasing the amount of yeast in the diet increased midgut length. Interestingly, we also detected an opposite effect of sucrose: increasing levels of sucrose abrogated the growth-promoting effect of yeast (Figure 2A). To test the impact of total nutrient ingestion, we measured the amount of food ingested per diet in the same preparations of flies (Figure 2—figure supplement 1B), which allowed us to plot total midgut length over diet, normalized to food passage (Figure 2—figure supplement 1C). This revealed a large area of diet space devoid of points, precluding a meaningful surface-plot analysis, so we plotted midgut length over ingestion-normalized estimates of yeast, sugar, and yeast to sugar ratio (Figure 2B–D). This correlative approach showed that midgut length increased as a function of yeast ingested, before a plateau and then a slight decrease (Figure 2B), while increasing the amount of sucrose ingested decreased length (Figure 2C). Overall, midgut length seemed to be proportional to the yeast to sucrose ratio (Figure 2D), consistent with our conclusions that sucrose opposes yeast-induced growth.

Figure 2 with 2 supplements see all
Sugar opposes yeast-induced increase of midgut length.

(A) Midgut length is maximized at specific points in diet space. Adult flies were maintained for 5 days from eclosion on one of 28 diets based on different caloric concentration and yeast to sucrose ratios (see Figure 2—figure supplement 1A for scheme on diets used and sample size). The list of recipes can be found in Supplementary file 1. The figure shows contours of a thin-plate spline (Generalized Additive Model) of length (mm, coded by colors) as a function of yeast and sucrose in diet. Colored dots represent mean of samples in a particular diet. (B–D) Yeast and sucrose have mutually opposite impacts on midgut length. Plots show an increase in midgut length with increased amount of yeast ingested (B); a decrease in midgut length with increased amount of sucrose ingested (C); and an increase in midgut length with ratio of yeast to sucrose ingested (D). (E) Several nutrients from yeast (proteins, lipids, vitamins/minerals) are required to increase midgut length. Nutrients from yeast (proteins, amino acids, lipids, cholesterol, vitamins/minerals) were added against a base diet of only the amount of sucrose found in high yeast (HY) and devoid of yeast. Letters above violin plots represent grouping by statistical differences (post hoc Tukey on GLMM). Bars beneath the main plot describe caloric content provided by the different components. (F) Midgut size is opposed by sugar, but not other added calories. Diet with only lipids, isocaloric with high-sugar (HS) and HY diets, results in midguts of lengths comparable to those on HS diet. Substitution of sucrose from HS diet with isocaloric lipids (yeast:lipid 1:14) results in midguts as long as those on HY. Midguts of flies reared on a diet substituting sucrose in HY diet with lipids (yeast:lipid 1:0.7) are also similar in length to those of flies fed HY. Letters above violin plots represent grouping by statistical differences (post hoc Tukey on GLMM). Bottom part of the chart (bar graph) describes caloric content provided by the different components. (G) Opposition by sugar of yeast-induced growth is not specific to sucrose. Statistical comparisons were performed with HS vs. HY for each sugar. All flies for experiments in this figure were moved on the experimental diets at eclosion and dissected 5 days post eclosion. For the violin/dot plots shown in this figure, white dots represent single midgut measurements. Lozenges represent mean of repeats. Violin plots are color coded according to diets (HS = red, HY = light blue, cream for other diets). Numbers in parentheses at the bottom of charts indicate sample size. Additional information on the statistics can be found in Supplementary file 2.

Yeast, which appeared to drive midgut size, is a complex nutrient source, comprising ~45% proteins, ~40% carbohydrates, ~8% lipids, vitamins, and mineral traces. We asked whether a single nutrient class might account for the overall effect of yeast on midgut length by adding specific nutrients to a diet containing only the amount of sucrose found in HY, devoid of any yeast. These added nutrients were one of a source of proteins, amino acids, lipids, vitamins, and minerals. Surprisingly, no single nutrient recapitulated the effect of HY (Figure 2E). However, a combination of proteins, lipids, and vitamins promoted growth to a point that was statistically indistinguishable from midguts of flies fed HY (Figure 2E). We speculated that additional characteristics of the HY diet could be influencing midgut size, such as texture (Li et al., 2018). The addition of fibers to the diets (Figure 2—figure supplement 2A) did not affect midgut length, although altering density by varying agar content could affect midgut length (Figure 2—figure supplement 2B). However, none of these modifications were able to mimic the impact of yeast on midgut length. Together, these experiments suggest that a panel of yeast-derived nutrients is likely required to promote midgut growth, rather than a single nutrient, as is the case for other phenotypes, such as longevity and fecundity (Grandison et al., 2009).

To further test the opposite effect of sugar on midgut size, we analyzed the impact of diets with the caloric content of sucrose substituted with lipids. A lipids-only diet, isocaloric to HS and HY diets, resulted in short midguts (yeast:lipid 0:1, Figure 2F), in agreement with the hypothesis that yeast is required for growth. However, the addition of a small quantity of yeast (the same quantity found in the HS diet) to a lipid-based and sucrose-free diet led to midgut lengths comparable to those on HY diet (yeast:lipid 1:14, Figure 2F). These results demonstrate that in the absence of sucrose a small amount of yeast is sufficient for midgut growth, while on the HS diet (yeast:sugar 1:14), the same amount of yeast is not sufficient. This result confirms the opposing role of sucrose on yeast-induced growth.

Sucrose is one of many sugars, and itself a disaccharide of glucose and fructose. It was possible that one of these moieties alone may have been responsible for blocking the growth-promoting effect of yeast, or that this was an effect specific to disaccharides. We therefore compared the impacts of equivalent levels of sucrose, glucose, fructose, and maltose (a glucose disaccharide) on midgut size (Figure 2G). All four sugars decreased midgut length. To help distinguish between sensory and metabolic mechanisms regulating the size of the midgut, we also tested the effects of a nutritious, but not palatable sugar (sorbitol) and of a palatable, but not nutritious sugar (arabinose) on the size of the midgut (Figure 2—figure supplement 2C; Burke and Waddell, 2011). Response to sorbitol was similar to the response to sucrose, suggesting that sensory mechanisms are not key in defining the size of the midgut. On the other hand, HS diet made with arabinose was lethal (flies survived only 2–3 days), and HY diet made with arabinose reduced midgut size, suggesting some damaging effect of this sugar on either the midgut itself or organismal metabolism. Altogether, these data demonstrate that multiple nutrients from yeast collectively increase midgut size, including proteins, lipids, vitamins, and minerals. By contrast, sugars oppose the impact of yeast on midgut size.

Diet composition affects both cell number and EC size in the midgut

Organ size can originate both in changes in cell numbers and cell size (e.g., Neufeld et al., 1998). We asked whether diet composition influences cell number or cell size in the Drosophila midgut, with particular attention given to the region that responds the most to diet, the posterior midgut (region 4, Figure 1G; Buchon et al., 2013). By combining immunostaining and transgenic cell type-specific labels, we quantified the number of ISCs (esg+ Su(H)-, cells), EBs (esg+, Su(H)+ cells), EEs (Prospero+ cells), and ECs (esg-, prospero-, larger polyploid cells) in the midguts of flies feeding on HS or HY diets (Figure 3A-C). All the different cell types increased in number on the HY diet compared to the HS diet (Figure 3C), and their relative proportions did not change (Figure 3D), demonstrating that posterior midguts resize without noticeable changes in cellular composition.

Figure 3 with 1 supplement see all
Diet composition affects both cell number and enterocyte size in the midgut.

(A–D) High-yeast (HY) diet increases numbers of all midgut cell types. Representative pictures of midguts from flies kept on high-sugar (HS) (A) or HY (B) diet. Green arrows indicate intestinal stem cells (ISCs), marked only by GFP (green), red arrows indicate enteroblasts (EBs), marked by GFP and GBE Su(H)-lacZ (red), and white arrow indicate enteroendocrine (EE) cells, marked with anti-Prospero antibody (white). All nuclei are stained with DAPI (blue). Quantification of total cell numbers in the posterior midgut (R4) for HS and HY, statistical analysis is comparing HS vs. HY for each cell type (C). HS and HY diets do not affect the relative proportion of cell types in the midgut (error is standard error of the mean D). (E–G) Diet affects enterocyte size. Representative picture of midguts stained with anti-mesh antibody on HS (E) vs. HY (F) diet. Quantification of EC size of flies on HS or HY diet confirms an increase in cell size on HY diet (G). All flies for experiments in this figure were moved on the experimental diets at eclosion and dissected 5 days post eclosion. For the violin/dot plots shown in this figure, white dots represent single midgut measurements. Lozenges represent mean of repeats. Violin plots are color coded according to diets (HS = red, HY = light blue). Numbers in parentheses at the bottom of charts indicate sample size. Additional information on the statistics can be found in Supplementary file 2. Scale bars are 25 µm for all images.

ECs are the biggest and most numerous cells in the Drosophila midgut. For this reason, we hypothesized that EC size could also contribute substantially to overall midgut size, and that their resizing was more likely to regulate organ size than equivalent resizing of other cell types. We therefore focused on EC size to evaluate whether changes in cell size could also underlie changes in midgut size between HS and HY. We used density of EC nuclei as a proxy for cell density and found that the density of ECs (ECs per µm2 of midgut tissue) was lower on HY than on HS diet (Figure 3A and B, Figure 3—figure supplement 1A), suggesting a difference in cell size. Using junction markers (anti-mesh antibody), we labeled EC membranes, which allowed us to directly measure the surface of average EC cross-sections (as depicted in Figure 3—figure supplement 1B) as an indicator of EC size (Figure 3E and F). ECs were 154% larger (median size) on the HY diet than on the HS diet (Figure 3G). We also determined that EC height was increased on HY diet (Figure 3—figure supplement 1C), meaning the increased EC area was not due to lateral stretching but to an increase in total cellular volume. ECs are polyploid cells, and variation in ploidy can underlie variation in cell size (Edgar and Orr-Weaver, 2001; Orr-Weaver, 2015). Therefore, we asked if the increase in size of ECs was accompanied by an increase in ploidy. We dissociated nuclei and measured ploidy through FACS. Ploidy profiles were similar on HS and HY, suggesting that the difference in midgut size due to diet is not a result of a change in ploidy (Figure 3—figure supplement 1D–H). Our results demonstrate that diet composition influences both the size and numbers of cells that build the Drosophila midgut.

The midgut plastically resizes in response to shifts in diet composition

We next asked how the size difference between HS and HY is established, and if this is a plastic process. We measured the growth kinetics of midguts on the HS and HY diets during the first few days after emergence. The HY diet sustained continuous midgut growth in the first 5 days post eclosion, whereas length remained similar to that at eclosion on HS (Figure 4—figure supplement 1A). We next asked if, given enough time (up to 28 days), midguts on the HS diet would be able to grow to levels comparable to the HY diet. We found that midguts of flies kept on the HY diet increased in size, while the size of those kept on the HS diet decreased over the course of 28 days (Figure 4—figure supplement 1B and C). Does this change represent a developmental program triggered strictly post eclosion (Buchon et al., 2013) or a dynamic response to nutritional variation (O’Brien et al., 2011; Obniski et al., 2018)? We tested whether the midguts of flies maintained on either HS or HY can resize in response to subsequent dietary changes beyond the first 3 days of maturation (Buchon et al., 2013) using two different approaches. We first tested whether midguts of flies maintained on either HS or HY for 7, 14, or 21 days could still resize in response to a diet switch. Midguts still experienced HY-mediated increase in size after being on the HS diet for either 7, 14, or 21 days (Figure 4—figure supplement 1D), and midguts experienced HS-mediated shrinkage after being on the HY diet for 7, 14, or 21 days (Figure 4—figure supplement 1E), suggesting that a shift in diet composition can resize the midgut plastically during the fly’s entire adult healthspan. We also switched flies alternately between HS and HY diets every 7 days for 3 weeks and found that their midgut was able to resize following multiple variations in diet (Figure 4A), reminiscent of what has been already documented for cell number plasticity in similar experiments (O’Brien et al., 2011). Altogether, this revealed that shifts in diet composition can plastically resize the midgut of Drosophila.

Figure 4 with 2 supplements see all
Shifts in diet composition lead to plastic midgut resizing and changes in absolute and relative cell loss and gain.

(A) Midguts can respond plastically to changes in isocaloric diets. Midgut length increases from eclosion (day 0) on high yeast (HY) for 7 days, then decreases when switched to high sugar (HS) for additional 7 days but can reincrease size upon a further 7 days HY feeding. Letters above violin plots represent grouping by statistical differences (post hoc Tukey on GLMM). (B) Mitotically active cells visualized by phospho-histone H3 (pH3) immunostaining are more numerous on HY diet than on HS diet. pH3+ cells gradually increase over time on HY, but not HS diet. Letters above violin plots represent grouping by statistical differences (post hoc Tukey on GLMM). Flies were put on the diets at eclosion, days on x-axis reflect days from eclosion. (C, D) Shifting between diets impacts pH3+ cell number in both growth (HS to HY) and shrinkage (HY to HS) experiments. Days in chart are from eclosion. Statistical comparisons are vs. pre-shift measurement. (E, F) Clonal assay with EsgF/O system put on either HS or HY from eclosion for 5 days, illustrates increased number of marked cells on HY (F) vs. HS (E) diets in region 4 of the midgut. GFP, in green, marks all cells made since the EsgF/O system was activated. (G–L) Cell loss assay enables analysis of the impact of diet composition on replacement ratio and rate. Description of experimental design is found in Materials and methods and illustrated in Figure 4—figure supplement 1H. In brief, this assay allows us to mark enterocytes (ECs) and enteroblasts (EBs) at the start of the experiment (9 days post eclosion) and to count their numbers 14 days after shifting dietary conditions (23 days post eclosion) recapitulating growth and shrinkage of the midgut, thus estimating cell gain and cell loss in these conditions. Representative pictures for the cell loss assay in growing conditions (G, H) and shrinkage conditions (I, J). In red 5966GS>His2B-RFP, marking EB and EC. Number of ECs in the posterior midgut, both marked (red, old ECs) and unmarked (blue, new ECs) by RFP, error bars are SE from three repeats (K). Data shown as rate relative to experiment start (cell/initial EC/day) (L). Number on bar in red is the ratio of EC gained/EC lost (see Materials and methods for formula). For the violin/dot plots shown in this figure, white dots represent single midgut measurements. Lozenges represent means of replicate experiments. Violin plots are color coded according to diets (green = eclosion, HS = red, HY = light blue, HY to HS = pink, HS to HY = purple). Numbers in parentheses at the bottom of charts indicate sample sizes. Additional information on statistics can be found in Supplementary file 2. Scale bars are 50 µm for all images.

Shifts in diet composition change absolute and relative levels of cell loss and ISC proliferation

We next asked whether changes in midgut size would be accompanied by changes in epithelial turnover. Study of environmental regulation of the midgut has so far focused largely on stem cell behavior. However, ISC proliferation is only half of the equation that governs epithelial turnover in the midgut. While most studies in Drosophila have focused exclusively on ISC proliferation and the associated ‘cell gain,’ the rate of ISC proliferation can only make sense of total cell gain when examined relative to the rate of cell loss. To understand diet-dependent midgut growth, we designed an integrative analysis of stem cell proliferation in the context of overall epithelial cell dynamics. We reasoned that any net change in cell number upon a diet shift must be the result of changes in absolute levels of cell gain (ISC proliferation) and cell loss, which together lead to a relative cell replacement ratio (i.e., the number of cells gained for each cell lost). Therefore, we measured each parameter on the HY and HS diets, and their response to diet switching.

We first quantified ISC proliferation in the midguts of flies fed HS or HY diets by immunostaining against the mitotic marker phospho-histone 3 (pH3). pH3-positive cells were more abundant on the HY diet compared to HS diet throughout the first 4 weeks of adulthood (Figure 4B). We also quantified this mitotic index across the diet space in which we had previously measured midgut length (Figure 2A). Indeed, ISC proliferation peaked in the same diets that were associated with long guts, showing promotion by yeast, but opposition by sugar (Figure 4—figure supplement 1F and G ). We then asked whether ISC proliferation tracked the changes in length driven by switching between diets. Switching flies from HS to HY increased ISC proliferation (Figure 4C), and vice versa after switching from HY to HS (Figure 4D). Do these mitoses translate into more progeny (ECs and EEs)? We tested directly for an impact of diet on ISC proliferation by monitoring the number of new cells using the esgF/O lineage tracing tool, which marks daughters of proliferating cells with GFP (Jiang et al., 2009). More cells were marked by GFP on the HY diet than on the HS diet (Figure 4E and F), indicating that HY-associated mitosis increases gain of new cells. These data indicate that diet composition alters ISC proliferation rate, with higher proliferation correlated to midgut size.

We then examined EC loss on the HS and HY diets. We reasoned that EC loss in the Drosophila midgut could be quantified by labeling ECs and EBs with a transient pulse of a long-term stable label (Figure 4—figure supplement 2A). Conducting this pulse chase over 14 days, which is a period longer than that in which we had recorded diet-induced midgut resizing, would allow us to quantify cell loss accompanying size plasticity. Histone2B-RFP (His2B-RFP) labels nuclei extremely stably (Antonello et al., 2015). We confirmed that His2B-RFP was not quenched up to 14 days after transient expression in tissues that do not turnover cells in the same manner as the midgut (crop and hindgut), in both HS and HY conditions (Figure 4—figure supplement 2B). This result indicates that, in ECs, a loss of RFP+ nuclei would reflect a loss of cells, not a loss of fluorophore. We expressed His2B-RFP transiently (using the hormone-dependent 5966GS EB-EC-specific driver and 3 days of feeding the inducer, RU486) (Figure 4—figure supplement 2A), after which RFP was detectable in most ECs (Figure 4G,I). Five days after start of the chase, no cell loss was detectable at the organ level in midguts maintained constantly on HY diet (HY to HY condition, Figure 4—figure supplement 2C), so we proceeded to quantify a later timepoint (14 days post chase start, Figure 4K). By 14 days after chase start, some, but not all, ECs retained RFP fluorescence (Figure 4H and J). This system therefore allowed us to quantify the cells gained (increase in DAPI+ RFP- cells) and lost (decrease in DAPI+ RFP+ cells) on each of the HS and HY diets (Figure 4G–K, Figure 4—figure supplement 2A). This experiment showed that numerous cells are gained on the HY diet, independent of previous diet (i.e., HS to HY and HY to HY), but also that a high number of cells are lost when flies started the experiment on the HY diet, independent of the present diet (i.e., HY to HS and HY to HY, Figure 4—figure supplement 2D), suggesting that current diet and dietary history could both influence tissue turnover. We note that, as guts on the HY diet are bigger, dietary history could just reflect initial midgut size. To better characterize the coupling between cell gain and cell loss, we further calculated the rates of cell gain and loss per initial EC per day (number of cells gained/lost per initial EC per unit of time, Figure 4L) as this rate takes into account size difference. Surprisingly, the rate of EC loss (per initial EC) was higher on the HY diet (HY to HY) than on the HS diet (HS to HS). This suggested that, taken alone, the mere loss of ECs cannot fully explain the small size of guts on the HS diet.

What was the balance of relative cell gain and loss upon HY or HS feeding? We compared the results of our ISC proliferation and EC loss indices (Figure 4K and L, Figure 4—figure supplement 2D). When midguts were growing on HY, cell gain exceeded loss, generating an overall replacement ratio greater than 1 (Figure 4L). Accordingly, when midguts were shrinking on HS, the replacement ratio was lower than 1 (Figure 4L). Of note, in none of our four conditions we did detect a replacement ratio close to 1, indicating that neither HS nor HY diets supported a strict coupling between cell gain and loss. We note that, importantly, similar replacement ratios occurred with very different absolute rates of cell gain and loss. For instance, replacement ratio was similar when flies were constantly fed the HS diet, or after switching from an HY to an HS diet (replacement ratio ~0.5), despite considerably different levels of absolute cell gain and loss in these two different conditions. In addition, the replacement ratio was higher in flies switched from HS to HY diet than in flies constantly feeding on the HY diet, confirming an influence of dietary history on the rates of cell gain and loss. This indicates that neither ISC proliferation nor cell loss alone capture the nature of epithelial turnover; rather, quantification of both parameters and calculation of the replacement ratio is required. This also indicates that the diet on which flies are feeding on is not the only factor influencing the absolute and relative rates of cell gain and loss, the dietary history or possibly initial organ size are also key factors. However, additional factors, in addition to cell proliferation and cell loss, must control the size of the midgut since the cell replacement ratio does not entirely translate into the increase in midgut size (e.g., on HY diet, replacement ratio is 1.46 while midgut area posterior increase is 1.14). Altogether, our data demonstrate that diet composition independently alters the rates of cell gain and cell loss, and their relative ratio underpins the emergent property of midgut growth.

Sugar induces translational stress in the midgut, which uncouples ISC proliferation from expression of pro-mitotic niche signals

Our phenomenological investigations revealed several unexpected results: (1) yeast and sugar have opposing effects on midgut size; (2) organ size is determined by both cell number and cell size; and (3) epithelial cell gain and loss, induced by dietary variation, do not linearly follow patterns intuitively expected based on midgut size. We wondered what mechanistic processes might underlie these observations. We performed an RNA-seq analysis of dissected midguts of flies that were fed either the HS or HY diet from eclosion and examined the kinetics of gene expression from eclosion on to days 1, 2, 3, and 5. An unsupervised method of grouping samples according to gene expression (principal components analysis) revealed a very clear two-phase process (Figure 5A). An apparently programmatic series of changes in gene expression occurred during the first day. These changes were largely diet-independent and correspond to maturation of the midgut (Buchon et al., 2013). However, in the following days, transcriptomes diverged substantially between the HS and HY diets. We identified differentially expressed genes (DEGs) between the HS and HY diets at days 3 and 5. Gene ontology enrichment analysis of the genes significantly more expressed on one diet or the other revealed striking differences (Figure 5B). Genes upregulated on the HY diet included numerous genes involved in digestion (proteolysis, carbohydrate metabolism, lipid metabolism, sterol transport) and respiration or oxidative stress (examples in Figure 5C). Among the digestive enzymes identified as upregulated on the HY diet, proteases, peptidases, and amino acid metabolic processing enzymes were 10–20-fold upregulated relative to HS, corresponding to dietary protein content. Similar results were found for genes involved in sterol transport, lipid metabolism, and carbohydrate digestion. These results collectively suggest higher digestion of protein, lipids, and carbohydrates on the HY diet, even though HS contained more sugar. On the HS diet, unexpectedly, genes upregulated compared to on the HY diet were ones involved in the regulation of growth, tissue development, proliferation, response to stress, and signal transduction (Figure 5B), despite the HS diet being growth repressive. Together, these findings suggest conflicting processes of cell proliferation/growth versus arrest, in which growth-promoting mechanisms are activated by, but fail to overcome, the effect of sugar.

Figure 5 with 2 supplements see all
Sugar uncouples intestinal stem cell (ISC) proliferation from niche signal expression by inducing translational stress.

(A) Diet influences midgut transcriptomes after an initial programmed developmental transition. The plot shows a principal components analysis (PCA) of the whole transcriptome, with means per diet per day ± standard error (three repeats). Numbers on the plot represent the day of dissection from eclosion. Lines connect the datapoints sequentially (day 0 to day 1, day 1 to day 2, etc.), and show the divergent transcriptomic trajectory followed by midguts on the two different diets from eclosion. (B, C) Diet modulates expression of functionally distinct gene classes. Midguts of flies fed high-yeast (HY) diet show higher expression of genes with digestive functions, while high-sugar (HS) diet involves mainly genes attributed to stress response and growth. X-axis represents the statistical significance of the gene ontology (GO) categories (y-axis) after adjustment for multiple testing. Size of the dot is proportional to number of genes in the given GO category (B). Table of genes significantly differently expressed, between HS and HY, as a ratio of HS/HY, representing midgut response to HS and HY diets; additional information on the statistics is found in Materials and methods; asterisks denote genes significantly different for p-value but not for adjusted p-value (C). (D) Cell proliferation is possible on HS diet when genetically induced. Progenitor-specific (EsgTS) overexpression of a constitutively active form of Ras (UAS-RasV12) and of UAS-Tor-DER (EGFR Active), both known proliferative inducers, allows for increased proliferation on HS diet. Flies were 6 days old when dissected. (E) Enterocyte-specific overexpression (MyoTS) of UAS-upd3-OE and UAS-spi-SEC elicit increased proliferation, strongly only on HY diet, and weakly on HS with UAS-upd3-OE. Flies were 9 days old when dissected. (F–H) General translation is lower on HS than HY, shown by puromycin incorporation assay. Images show lower incorporation on HS (F, F′) than HY (G, G′) in region 4 of the midgut from 5-day-old Canton-S (Cs) flies that were shifted on HS or HY at eclosion. Quantification of mean pixel intensity of puromycin stain (H). (I–K) p-eIF2α stain is elevated on HS (I, I′) compared to HY (J, J′). Quantification of mean pixel intensity of p-eIF2α stain in region 4 of the midgut from 5-day-old Cs flies that were shifted on HS or HY at eclosion (K). (L) Re-enabling translation can restore mitosis in midguts shrinking after being shifted from HY to HS diet for 7 days. Initial shift was performed 12 days post eclosion. Blocking translational inhibition with ActTS>Gcn2 IR or ActTS>LK6 IR is sufficient to increase pH3+ cells in midguts of flies on HS diet. However, ActTS>PEK IR and ActTS>AMPKα-IR had no effect on the number of pH3+ cells. Statistical comparisons are vs. control. (M) Despite increased mitotic activity following repression of translational inhibition in ActTS>Gcn2 IR or ActTS>LK6 IR, midgut size was still reduced on flies kept on HS diet. The statistical comparison is comparing interaction between diet and fly lines. For the violin/dot plots and boxplots showed in this figure, white dots represent single midgut measurements. Lozenges represent mean of repeats. PCA, violin plots, box plots, and the PCA are color coded according to diets (green = eclosion, HS = red, HY = light blue). Numbers in parentheses at the bottom of charts indicate sample size. Additional information on the statistics can be found in Supplementary file 2. Scale bars are 50 µm for all images.

Figure 5—source data 1

RNA-seq count data for Figure 5A–C.

https://cdn.elifesciences.org/articles/64125/elife-64125-fig5-data1-v2.csv
Figure 5—source data 2

Selected gene ontology (GO) categories for Figure 5B.

https://cdn.elifesciences.org/articles/64125/elife-64125-fig5-data2-v2.csv
Figure 5—source data 3

All gene ontology (GO) results for Figure 5B.

https://cdn.elifesciences.org/articles/64125/elife-64125-fig5-data3-v2.csv
Figure 5—source data 4

Differentially expressed genes for high sugar (HS) vs. high yeast (HY) (days 3 and 5), from which representative genes are shown in Figure 5C.

https://cdn.elifesciences.org/articles/64125/elife-64125-fig5-data4-v2.csv
Figure 5—source data 5

Numeric data for Figure 5D.

https://cdn.elifesciences.org/articles/64125/elife-64125-fig5-data5-v2.csv
Figure 5—source data 6

Numeric data for Figure 5E.

https://cdn.elifesciences.org/articles/64125/elife-64125-fig5-data6-v2.csv
Figure 5—source data 7

Numeric data for Figure 5H.

https://cdn.elifesciences.org/articles/64125/elife-64125-fig5-data7-v2.csv
Figure 5—source data 8

Numeric data for Figure 5K.

https://cdn.elifesciences.org/articles/64125/elife-64125-fig5-data8-v2.csv
Figure 5—source data 9

Numeric data for Figure 5L.

https://cdn.elifesciences.org/articles/64125/elife-64125-fig5-data9-v2.csv
Figure 5—source data 10

Numeric data for Figure 5M.

https://cdn.elifesciences.org/articles/64125/elife-64125-fig5-data10-v2.csv

We examined the pro-proliferative genes more highly expressed on the HS diet in detail and found many genes shown in previous studies to be linked to epithelial stress and compensatory turnover (Buchon et al., 2009b). These included genes involved in SAPK signaling (JNK, P38), apoptosis, as well as most pathways controlling ISC proliferation and differentiation (EGFR, JAK-STAT Wnt, hippo, Notch pathways). Importantly, genes coding for classic pro-proliferative niche signals (vn, Krn, Spi, wntD, upd3) were also more highly expressed. To date, these signals have been shown to be upregulated in the niche in response to stress, providing a cue for ISCs to proliferate and replace damaged or lost differentiated cells (Bonfini et al., 2016). Accordingly, we detected a shorter lifespan on HS diet compared to HY diet, confirming the deleterious impact of the HS diet (Figure 5—figure supplement 1A). Upregulation of these pathways in response to the HS diet suggested that the midguts on HS diet are stressed, yet ISCs failed to respond (Figure 4). This raised the question of whether ISCs on the HS diet are unable to proliferate at all, or whether HS uncouples nonautonomous niche signaling from an appropriate proliferative response. Such uncoupling could result from the inability of ISCs to respond to niche-derived signals or from the inability of the niche to translate or secrete niche signals, leading to an indirect decrease in proliferation.

We therefore first asked whether ISCs of flies fed the HS diet had the ability to proliferate in response to a pro-mitotic cell-autonomous signal. In progenitor cells, we overexpressed the oncogene RasV12 (esgTS>UAS-RasV12), a driver of cell proliferation, and UAS-Tor-DER (an activator of EGFR pathway) and again quantified pH3-positive cells after feeding on HS or HY diet. With overexpression of both constructs, ISC proliferation was higher than in control flies and statistically indistinguishable in the same genotype between HS and HY diets (Figure 5D). These results confirm that ISCs retain proliferative capacity on the HS diet, but proliferation is uncoupled from expression of pro-mitotic genes. Having shown cell-autonomous proliferative capacity was retained, we then asked whether HS diet uncouples ISC proliferation from non-autonomous niche signals by testing if stem cell proliferation could be induced through over expression of two ligands from the niche (ECs), upd3 and spitz, which are known to be involved in the response to stress in the midgut (Buchon et al., 2010). We used MyoTS to drive overexpression of UAS-upd3-OE and UAS-spi-SEC in ISCs. Both constructs were able to significantly drive proliferation on HY diet; however, on the HS diet spitz overexpression did not lead to increase proliferation and overexpression of upd3 led to a smaller degree of increase in proliferation. This result suggests that diet uncouples niche signals from stem cell activity by acting on the niche itself or blocking ISCs from receiving said signals (Figure 5E).

We next asked how ISC proliferation could become uncoupled from the high expression of niche derived pro-proliferative signals. Our RNAseq experiment indicated that the midguts of flies feeding on the HS diet express higher levels of genes involved in translation inhibition, including PEK and the Gcn2 kinase (Figure 5B and C). Reduced translation of niche-derived signals has been previously shown to impair midgut repair after infection (Chakrabarti et al., 2012), and we hypothesized a similar effect could occur upon HS feeding. We first assessed whether global translation was lower in midguts of flies feeding on HS than on HY by measuring puromycin incorporation 3 hr after feeding. Puromycin incorporation was significantly higher on HY than HS, demonstrating that general translation was lower on HS diet (Figure 5F–H). A central regulator of global translation is eIF2α, which mediates a decrease in global translation when phosphorylated during metabolic stress. In our RNAseq, expression of two eIF2α kinases (PEK and Gcn2) was increased on the HS diet, suggesting that diet-dependent eIF2α phosphorylation could explain ISC-niche uncoupling. Immunostaining confirmed that eIF2α phosphorylation was higher on the HS diet (Figure 5I–K). p-eIF2α was detected both in progenitor cells and ECs but absent or only slightly detected in the visceral muscles (Figure 5—figure supplement 2A, D'). To directly test the role of translational inhibition in sugar-induced blockage of ISC proliferation, we knocked down expression of multiple genes involved in translation inhibition, including Gcn2, Lk6, AMPKα, and PEK, under control of an inducible and ubiquitous Actin that could act in all the midgut cell types that displayed eIF2α phosphorylation. Knockdown with RNAi of either Gcn2 (an eIF2α kinase) or Lk6 (an eIF4E1 kinase) in adult flies increased ISC proliferation on the HS diet (Figure 5L), or when shifting for a week from HY to HS diet (Figure 5—figure supplement 2E), compared to controls. However, global expression of the same RNAi constructs was not able to increase midgut length on the HS diet (Figure 5M) or to prevent midgut shrinking when transitioning from HY to HS (Figure 5—figure supplement 2F). To determine in which cell type translational inhibition had a role in blocking stem cell proliferation, we knocked down Gcn2 in progenitor cells (EsgTS) and ECs (MyoTS). We found that it was possible to increase proliferation when knocking Gcn2 in ECs, but not in progenitors (Figure 5—figure supplement 2G). This, in addition with previous results (Figure 5D and E), suggests that on the HS diet a translational inhibition in ECs involving Gcn2 uncouples ISC proliferation from increased expression of pro-proliferative niche signals.

The midgut can resize independently of stem cell proliferation

To date, ISC proliferation has been assumed to underpin midgut growth in response to nutrition (O’Brien et al., 2011). However, our finding that restoring ISC proliferation was not sufficient to increase size on HS diet or abrogate the shrinking effect of switching from HY to HS diet indicated that additional mechanisms must underlie midgut size plasticity. We first explored this using natural population variation in the DGRP. We focused on a panel of seven DGRP lines (Figure 6A) including lines that respond and lines that do not respond to diet composition, but have either a short midgut, a middle-sized midgut, and a long midgut; we measured midgut length and ISC proliferation rate (pH3 immunostaining). We first noticed, as previously published, that the proliferative response was variable across different Drosophila lines (Figure 6A; Tamamouna et al., 2020). ISC proliferation and midgut size were not positively correlated amongst these genotypes on either diet, except for one line (Figure 6A). In addition, midgut resizing did not correlate with change in proliferation between the HS and HY diets (Figure 6—figure supplement 1A), suggesting that proliferation could be dispensable for resizing.

Figure 6 with 1 supplement see all
The midgut can resize independently of stem cell proliferation.

(A) Mitotic index does not correlate with midgut length. Quantification of pH3+ cells across a selected panel of high and low responder Drosophila Genetic Reference Panel (DGRP) lines shows that midgut length does not correlate with cell proliferation, flies were moved on the experimental diets at eclosion and dissected 5 days post eclosion. (B–E) Blocking EGF signaling with UAS-Egfr-IR in progenitor cells (C) results in a change in conformation of progenitor cells, which assume a more rounded shape compared to the traditional triangular shape and seem to be less in number (not quantified) compared to control (B), consistent with lower pH3+ counts than control (D). Statistical comparison for (D) is for the interaction between diet and genotype. Data for (D) is from 19 days post eclosion. However, these EsgTS>UAS-Egfr-IR midguts are still able to reach a similar length to controls (E). In (E), high sugar (HS) was dissected at 12 days post eclosion, and HS to high yeast (HY) 7 days later. (F, G) Insulin signaling with a dominant negative construct in progenitor cells results in less proliferation (F) despite EsgTS>UAS-InR-DN resulting in the same midgut length growth as the control (G). Statistical comparison for (G) is for the interaction between diet and genotype. Data for (F) is from 19 days post eclosion. In (G), HS was dissected at 12 days post eclosion, and HS to HY 7 days later. (H–J) Increase in midgut length despite proliferation blockage is accompanied with compensatory area increase of enterocyte (EC). Representative pictures of midguts stained with membrane marker mesh (white), shifted from HS to HY at 12 days post eclosion, and kept on HY for seven additional days show bigger cells on EsgTS>UAS-Egfr-IR (I) compared to control (H). Quantification of EC cell size shows compensatory effect in ECs (J). Statistical comparisons for (E, G, J) are for the interaction between diet and genotype. For the violin/dot plots and boxplots showed in this figure, white dots represent single midgut measurements. Lozenges represent mean of repeats. Violin plots and box plots are color coded according to diets (HS = red, HY = light blue, HS to HY = purple). Black lines connecting means visualize the interaction between diets. Numbers in parentheses at the bottom of charts indicate sample sizes. Additional information on the statistics can be found in Supplementary file 2. Scale bars are 10 µm for all images.

We next directly tested whether ISCs are required for the midgut to resize. EGFR signaling is central to ISC function (Buchon et al., 2010; Jin et al., 2015). We confirmed that RNAi knockdown of the EGF receptor (esgTS>Egfr IR) greatly reduced the number of progenitor cells and resulted in a complete loss of proliferative cells (Figure 6B–D). However, despite the loss of proliferative cells, these flies were still able to grow their midgut upon switching from HS to HY diets (Figure 6E). In addition to the EGFR pathway, insulin/IGF-like signaling is a master regulator of ISC proliferation, notably in response to diet (Amcheslavsky et al., 2009; Biteau et al., 2010; Choi et al., 2011; O’Brien et al., 2011; Strilbytska et al., 2020; Veenstra et al., 2008). We therefore repressed the insulin pathway by overexpressing a dominant-negative insulin receptor in progenitor cells (esgTS>InR DN). This strongly decreased ISC proliferation (Figure 6F), but these midguts were nevertheless able to grow on the HY diet (Figure 6G). We observed similar results when we overexpressed the pro-apoptotic gene reaper in progenitor cells (esgTS>rpr OE), which resulted in complete loss of marked progenitor cells in region 4 of midguts already at 7 days post TARGET system activation on the HS diet (Figure 6—figure supplement 1C) compared to control midguts (Figure 6—figure supplement 1B), loss which was also evident after shifting flies on HY from HS for an additional week (Figure 6—figure supplement 1E). However, midguts were still able to increase in size similarly to controls (Figure 6—figure supplement 1F). Altogether, these results demonstrate that, surprisingly, proliferative capacity in ISCs can be dispensable for diet-dependent midgut resizing.

Our original hypothetical model was that midgut size would be regulated by the sum of cell gain, cell loss, and cell size, according to the morphological changes observed in Figure 3. While long-term maintenance of a stem cell pool is presumably important for survival, our present functional genetic data show that, in the short term, midgut resizing does not strictly require ISC proliferation. How? We noticed that ECs were larger when EGFR signaling was reduced in progenitor cells (Figure 6H and I). When we quantified EC sectional area in esgTS>EGFR IR flies, we found that it did indeed increase after switching from HS to HY diets, and more than in control midguts (Figure 6J). We visualized a similar morphology also in esgTS>rpr OE midguts shifted from HS to HY (Figure 6—figure supplement 1E). This suggests that EC size can compensate for the lack of increase in cell number when ISC proliferation is blocked, outlining a possible multicellular homeostatic mechanism involving a feedback between ISCs and ECs.

EC resizing is necessary for diet-dependent midgut plasticity

Our data suggested that EC size could be an essential driver of midgut resizing. We first asked if EC size covaries with midgut size, as measured in Figure 4A. Indeed, EC size corresponded to midgut size across diet manipulations (Figure 7—figure supplement 1A). We wondered whether this correspondence would generalize across genetically variable genotypes. We went back to the seven DGRP lines we previously analyzed for proliferation and length and, in the same midguts, measured EC size (Figure 7A). EC size and midgut length were generally positively correlated, sometimes very strongly, especially on HY diet. Additionally, midgut resizing correlated with change in EC area between the HS and HY diets (Figure 7—figure supplement 1B), suggesting that EC size could be a strong driver of midgut plasticity.

Figure 7 with 3 supplements see all
Enterocyte (EC) resizing is required for midgut plasticity.

(A) EC size mostly correlates with midgut length. Quantification of EC area across a selected panel of Drosophila Genetic Reference Panel (DGRP) lines comprising high and low responder shows that midgut length mostly correlates with EC cell area, especially in high-yeast (HY) diet. Lines on plot show smoothed splines. Flies were moved on the experimental diets at eclosion and dissected 5 days post eclosion. (B–H) The TOR pathway regulates cell size. Representative pictures of single-cell clones (hsFlp; Act >STOP>Gal4, UAS-GFP) suggest that compared to GFP- cells TOR downregulation (UAS-Tor-IR, GFP+, B) results in smaller cells, while TOR hyperactivity (UAS-Rheb-OE, C) increases cell size. Quantification of clone size in (D). Single-cell clones are marked with GFP (green). Dissection for clones was performed 14 days post eclosion. Knockdown of TOR with MyoTS, an EC-specific driver, leads to the increased number of small ECs (F) compared to control (E); EC-specific GFP is indeed visible in small cells. Quantification of EC area (G). Accordingly, blocking TOR pathway components in ECs (MyoTS) inhibits diet-induced midgut growth. Control showed in chart is representative of multiple experiments. Statistical analyses were performed only on appropriate repeat/experiment and comparing interaction between diet and fly line. For (E–H), flies were dissected for the first timepoint (HS) at 12 days post eclosion, and then shifted on HY for additional 7 days. (I, J) Representative picture utilizing single-cell clonal system suggests that blocking Atg2 (hsFlp; Act>STOP>Gal4, UAS-GFP>UAS-Atg2-IR, GFP+ cells) results in bigger ECs compared to control GFP- cells, quantified in (J). Dissection for clones was performed 14 days post eclosion. (K, L) Blocking autophagy reduces midgut resizing upon shrinkage (HY to HS for 7 days). Blocking Atg8a expression with RNAi in ECs (MyoTS>UAS-Atg8a-IR) results in less length shrinkage compared to control midguts. Blocking Atg2 expression with RNAi in ECs (MyoTS>UAS-Atg2-IR) results in less width shrinkage compared to control midguts. For (J, K), flies were dissected at the first timepoint (HY) at 12 days post eclosion, and then shifted on HS for additional 7 days. For the violin/dot plots and boxplots showed in this figure, white dots represent single midgut measurements. Lozenges represent mean of repeats. Violin plots and box plots are color coded according to diets for (G, H, K, L) (HS = red, HY = light blue, HY to HS = pink, HS to HY = purple) or to condition for (D, J) (GFP+ in green, GFP- in gray). Black lines connecting means visualize the interaction between diets. Numbers in parentheses at the bottom of charts indicate sample size. Additional information on the statistics can be found in Supplementary file 2. Scale bars are 25 µm for (B, C, I) and 10 µm for (E, F).

We attempted to manipulate EC size by functional genetics to test the role of EC size on midgut resizing. We focused on the TOR pathway because of its evolutionarily conserved role as a regulator of cell size (Blenis, 2017; Gonzalez and Rallis, 2017). Additionally, we found in our RNA-seq analysis that several targets of Foxo, including foxo itself, are upregulated on HS diet (chico, InR, Impl2, Thor, Figure 5C), indicating that the TOR pathway is downregulated on HS diet (Hay, 2011; Rera et al., 2012). We also separately tested a reported of Foxo activity, thor-lacZ (Karpac et al., 2011), and found that it upregulated on HS diet compared to HY (Figure 7—figure supplement 2A, C). This confirmed that physiological TOR signaling is modulated by diet. We first asked whether the TOR pathway could control EC size cell autonomously using a clone-tracing system (hsFlp; Act>STOP>Gal4,UAS-GFP), which allowed us to selectively inhibit (UAS-Tor-IR) or activate (UAS-Rheb-OE) the TOR pathway specifically in fluorescently labeled single ECs. TOR activation enlarged ECs, and knock-down decreased EC size (Figure 7B–D), confirming that TOR is a cell-autonomous regulator of EC size. We next asked whether TOR was required to modify EC size in response to diet. When TOR was knocked down in the ECs (MyoTS>UAS-TOR-IR, Figure 7E–G) of midguts switched from HS to HY, change in EC size was attenuated, confirming that the TOR pathway mediates diet-dependent EC resizing.

We next asked whether TOR pathway’s effect on diet-dependent EC resizing was sufficient to alter diet’s effect on overall midgut size. We knocked down a range of genes coding for proteins in the TOR pathway and measured midgut growth after switching from HS to HY. Each of these knockdowns qualitatively diminished growth, and this effect was statistically significant in most cases (Figure 7H). Knockdown of Tor itself (MyoTS>UAS-TOR-IR) led to a significant decrease in growth compared to control when moved on the HY diet, nearly abolishing any change in size. Furthermore, repeating this experiment using a gene-switch system (i.e., 5966GS) to drive RNAi against Tor in ECs led to similar results (Figure 7—figure supplement 2D). We then tested the impact of other components of TOR signaling. TOR is found in two macromolecular complexes, mTORC1 and 2, respectively (Wullschleger et al., 2006). The mTORC1 complex integrates Raptor, and EC-specific RNAi against Raptor was not statistically significant, but qualitatively diminished regrowth after switching from HS to HY (Figure 7H). Other classical components of the TOR pathway, including the transcription factors Myc and SREBP, as well as the kinase S6K, were all significantly required for regrowth (Figure 7H). Together, these data demonstrate that the TOR pathway regulates the growth of ECs, and that EC resizing is required for concomitant increase in organ size. mTORC1 is a negative regulator of autophagy (Dossou and Basu, 2019; Kim and Guan, 2019). In addition, our RNA-seq experiment (Figure 5) showed that genes related to autophagy were significantly upregulated on HS diet (Atg2, Atg17, Atg101). These observations outlined a hypothesis in which TOR repression on the HS diet increases autophagy, consequently decreasing cell and midgut size. Atg2 and Atg8a are core essential genes required for macro-autophagy (required for phagophore and autophagosome formation) and downstream of TOR pathway (Mulakkal et al., 2014). We first confirmed that autophagy could influence EC size cell-autonomously by knocking down Atg2 with a clone-tracing system. RNAi of Atg2 in singly labeled ECs (hsFlp; Act>STOP>Gal4, UAS-GFP>UAS-Atg2-IR) produced larger cells, consistent with our hypothesis that autophagy reduces EC size (Figure 7I and J). We then tested the requirement for Atg2 and Atg8a for dietary regulation of midgut size. We knocked down Atg8a (MyoTS>UAS-Atg8a-IR) or Atg2 (MyoTS>UAS-Atg2-IR) in ECs specifically and measured the impact on midgut shrinkage when flies transition from the HY to the HS diet (Figure 7K and L). Each Atg gene was required for complete shrinkage after switching from HY to HS diets. Performing this experiment with a gene-switch system (i.e., 5966GS) to drive RNAi led to similar results (Figure 7—figure supplement 2E). Altogether, our results demonstrate that the TOR/autophagy pathway is an important regulator of diet-dependent EC resizing, which itself is required for midgut resizing following a change in diet.

Discussion

The intestinal epithelium is the vanguard of an organism’s interaction with the external environment, which includes food, microbes, and pathogens (Miguel-Aliaga et al., 2018). Consequently, the midgut and its homeostasis are vital for host health and have been shown to determine animal lifespan (Biteau et al., 2010). Here, we demonstrate that the midgut is a plastic interface between the host and nutrients, which can dynamically resize throughout the healthy phase of early life in response to changes in the nutritional composition of the diet. To date, the dogma has been that ISC proliferation is the primary driver of midgut responses to environmental variation, but correspondence to resultant midgut size has been neglected. Our data paints a more nuanced picture, showing coincident plasticity in cell gain (ISC proliferation), cell loss, and cell resizing (Figure 7—figure supplement 2A). ISC proliferation was surprisingly dispensable for plasticity on the timescale of our experiments, whilst EC size emerges as an essential determinant of nutrition-dependent midgut resizing. Diet-dependent intestinal resizing also occurs in mammals and reptiles, and it will be interesting in future work to characterize whether equivalent mechanisms are also conserved (Mao et al., 2013; Navarrete et al., 2015; Petit et al., 2007; Thulesen et al., 1999; Xie et al., 2020).

Diet-dependent midgut resizing differentially affected specific midgut regions, producing an overall allometric response to diet. This suggests that either size control in each region depends on different gene networks, as it is the case for thorax and abdomen plasticity (Lafuente et al., 2018), or that the sensitivity of those networks is differentially primed in specific regions. Strikingly, nutrition-dependent midgut resizing varied extensively across lines in the DGRP panel, demonstrating interindividual genetic variation in the ability of the midgut to respond to diet. Altogether, we propose a model in which diet shapes organismal health not only through the nutrients it provides, but also through changes in the structure of the gastrointestinal (GI) tract that secondarily influence host physiology. Future work will allow us to determine the contribution of the plasticity of the GI tract in physiological responses to nutrition beyond the midgut.

Multiple nutrients affect midgut resizing

Work using Drosophila as a model to dissect the impact of nutrition on host physiology and lifespan demonstrated that proteins and amino acids have a central role in determining animal health. For example, single amino acids such as methionine mediate most of the tradeoff between reproduction and aging (Grandison et al., 2009). Selected amino acids also influence diverse aspects of host physiology such as survival of infection or the developmental rate of larvae (Bing et al., 2018; Zhang and Rubin, 2013). Previous studies focused on elucidating the impact of diet on midgut growth were conducted using either a starved vs. fed paradigm, or a basic diet vs. basic diet with added yeast paste, making it impossible to evaluate the contribution of different nutrients or to disentangle nutrient composition from caloric content (Choi et al., 2011; O’Brien et al., 2011). In our study, we demonstrate that qualitative variation in isocaloric diets drives midgut resizing, and we make use of nutritional geometry to demonstrate that both diet density and diet composition influence midgut size. While yeast ingestion drives midgut growth, ingested sugar results in midgut shrinkage, demonstrating again that calories per se do not limit midgut growth. Using rescue-like experiments, we investigated the nutrients in yeast that underlie midgut resizing. As expected, protein was required for diet-dependent midgut growth. However, protein alone was not sufficient to induce midgut growth: all components of yeast tested together, including lipids, vitamins, and minerals, were required for growth. These results are somewhat surprising and could be because midgut physiology is different from the physiology of other organs that respond to proteins. When considering ovaries, nutrients such as lipids have accumulated in the fat body during development (Kühnlein, 2011) and are thus readily available while proteins are limiting for the generation of new eggs. It is possible that in normal conditions midgut metabolism depends entirely on food for its maintenance while other organs use both stored reserves and nutrients from food. In such a scenario, all nutrients are effectively essential nutrients, acting as both signals and substrates for cell maintenance and production. Our study also demonstrated that changes in texture, as also shown in Li et al., 2018, affected midgut size, suggesting that multiple processes and probably a complex gene regulatory network influence midgut plasticity.

Adaptation to nutrients vs. pathophysiology of sugars

While all the nutrients tested influence midgut size, not all calories are made equal. Specifically, multiple dietary sugars in high amount resulted in midgut resizing. By contrast, an isocaloric diet with higher amount of yeast caused net midgut growth. Two hypotheses could explain the opposite effects of yeast and sugar. If the midgut size is optimized to nutrient availability (O’Brien et al., 2011), we would expect that a large midgut is required to potentiate digestion and absorb the complex panel of nutrients sourced from yeast, but not for the simpler task of sugar digestion and uptake. Intrinsic to this model is an expectation that the metabolic cost of growing the midgut is outweighed by diet’s nutritional reward, likely the reproductive benefit of yeast consumption (Skorupa et al., 2008).

However, an alternative hypothesis is also possible, namely that toxicity of high levels of sugar stunt midgut growth. This hypothesis is supported by our observations that the HS diet promoted a transcriptomic signature of stress, inhibited global translation, and decreased lifespan. This decreased lifespan was not mediated by an increase in fecundity because fecundity is lower in the absence of yeast, suggesting that sugar indeed is deleterious for health. This toxic effect of sugar on midgut tissue renewal could accelerate age-related pathologies such as barrier dysfunction, leading to early death. Of note, while age-associated microbial dysbiosis in the midgut is a key factor that determines fly lifespan, the microbiota was not required for diet-dependent midgut plasticity, suggesting that diet and microbes may additively regulate lifespan. The nature of the stress associated with HS diet remains elusive as most stress-responsive pathways ended up upregulated in midgut of flies fed on the HS diet.

Is midgut homeostasis only an apparent property?

Several studies have demonstrated that stress or damage inflicted to the midgut result in a higher expression of niche-derived pro-mitotic signals. For instance, microbial infection, ingestion of damaging chemicals, or genetic elimination of ECs all result in a compensatory increase in ISC proliferation (Amcheslavsky et al., 2009; Buchon et al., 2010; Buchon et al., 2009b; Buchon et al., 2009a; Jiang and Edgar, 2009; O’Brien et al., 2011). In addition, maturation of pro-mitotic ligands in ECs is regulated by caspase activity in dying cells (Liang et al., 2017), resulting in a direct coupling between cell loss and cell gain. These results together have yielded a model in which the midgut is a homeostatic tissue, keen to maintain a set size that the midgut would return to after challenge. Most importantly, this model suggests that ISC proliferation is coupled to EC loss in a fixed manner. One strong limitation to these models is that they rely mostly on measures of proliferation or local quantifications (e.g., clones) and lack quantification of cell loss. A perfectly homeostatic regulation of the midgut epithelium would imply that cell loss equals cell gain, which leads to a replacement ratio of 1. Using pulse chase experiments to track both the number of cells gained and lost in each midgut on both the HS and HY diet, we found that whether on HS, on HY, or during diet shifts, we never detected a cell replacement ratio approaching 1. Even in flies feeding constantly on the HY diet, more cells were generated in the midgut than lost, resulting in the midgut constantly growing up to 28 days post eclosion. We therefore propose a model (Figure 7—figure supplement 3A) in which there is not a constant coupling between cell loss and gain, but instead a flexible connection between ISC proliferation and cell loss as a function of diet, contributing to midgut size. In such a model, homeostasis could be only apparent and emerge from specific conditions where cell gain and loss are equal, which were by chance used in previous studies, rather than inherently programmed. Alternatively, the response to damaging stimuli, such as infection or chemical damage, could trigger a different type of response, less sensitive to variations in ISC niche coupling. It is interesting to note that most experiments documenting tissue repair have been performed on a rich diet, akin to our HY diet, that we find is permissive to growth. It is therefore difficult to tease apart the relative contribution of diet-dependent midgut growth and damage-dependent repair signals in the described homeostatic regrowth. To further integrate these different models, it will be interesting to revisit classical experiments that have led to the notion of epithelial homeostasis of the midgut on multiple diets.

We also observe that while the cell replacement ratio will determine the increase or decrease in midgut cell numbers, and midgut size, similar ratios can occur at very different rates of cell gain and loss. Our results demonstrate that, while diet determines the overall replacement ratio, the ‘history’ of the diet on which the organisms fed a few days prior also matters. For instance, while we detected a replacement ratio around 0.5 on both midguts staying on HS or transitioning from HY to HS, the rates of gain and loss were higher on the transition. We hypothesize this could originate in the fact that the initial size of the organ differs between experiments (short midgut at the start of HS to HS, long midgut at the start of HY to HS). Future experiments should tackle how past physiological conditions influence later epithelial dynamics. We argue that to understand and capture cell dynamics within a tissue both the replacement ratio and the rates of cell gain and cell loss need to be measured. In the current study, estimating only ISC proliferation would have led to an incomplete picture of tissue turnover.

Finally, these results were obtained by keeping flies constrained to an imposed specific diet, to tease apart the impact of different nutritional components on the midgut. However, a caveat of this conclusion on midgut homeostasis is that, in the wild, flies do not have a fixed diet and may vary their feeding among different available nutrients, possibly alternating periods of midgut growth with periods of midgut shrinkage. We do not know how this process would affect midgut homeostasis, but it may (or may not) result in an overall homeostatic replacement rate.

Contributions of EC size and cell numbers to midgut resizing

Most studies of midgut size have focused on stem cell proliferation, leading to the conclusion that an increase in midgut size is concomitant to an increase in ISC activity and symmetrical division. However, it remains unclear whether the increase in mitosis is causal or even required for resizing. During development, imaginal discs grow through a combination of proliferation and cell size growth, with growth being dominant over, and compensating for, cell division defects (Neufeld et al., 1998). We found that diet-dependent midgut growth is concomitant to both increased cell number (connected to increased ISC proliferation) and increased EC size (Figure 7—figure supplement 3A). In addition, blocking ISC proliferation did not prevent midgut growth on the HY diet, suggesting that ISC proliferation can be dispensable for midgut resizing. In such midguts, the increase in EC size was higher than in control intestines, compensating for decreased cell number. As both proliferation and size increase on the HY diet, and as EC size dominates, this raises the question of what the main driver of organ size is. One answer came from looking at lines from the DGRP panel. In these 11 wild-caught isolines, almost no positive correlation was observed between proliferation rate and midgut size on either diet (only one line made exception and only on HY diet). In contrast, a larger number of lines showed a correlation between midgut length and EC size (showing an overall strong correlation between increase in EC size and increase in midgut length when comparing HY/HS), indicating that EC size may drive midgut resizing. It remains possible that this principle of cell size would vary amongst lines, especially in lines that resize strongly in response to diet as there is a physical limitation to the size a cell can reach. It is also possible that the apparent dominance of size over proliferation is just a consequence of the size difference between progenitors and mature cells. It would require a very high number of mitoses to accumulate enough small progenitors to compensate for one EC. Accordingly, forced ISC over proliferation, for example, by RasV12 overexpression (Buchon et al., 2010), does not increase midgut size, but rather shortens the midgut due to epithelial multilayering. This suggests that proliferation and cell size need to be properly coordinated to lead to normal changes in organ size. This hypothesis agrees with a recent study demonstrating a balance between stem cell mitosis and EC nucleus growth in the midgut of Drosophila (Tamamouna et al., 2020). Additionally, we cannot discount a role for EBs in organ regulation. Regulation of EBs could be, for example, a critical step in setting up the size for new ECs, and EBs are known to be able to differentiate at different rates, or undergoing apoptosis (Reiff et al., 2019). We therefore propose that the organ reaches a set size based on diet, suggesting that the notion of a ‘size meter,’ introduced in developmental biology, could also apply to adult tissues (Shingleton, 2010). However, the nature of such a ‘size meter’ remains elusive, and its characterization will be complicated by the fact that it is nutrient dependent. Moreover, in physiological conditions midgut size may just be a function of midgut digestive capability. In this context, it is possible that while still being able to grow, a midgut expanding only through EC size is not as efficient as a normally growing midgut, or more prone to deregulation in later phases of life.

Translation control as a general regulator of ISC coupling to its niche

Our results suggest that high dietary sugar levels lead to a decrease in coupling between ISC proliferation and the expression of pro-mitotic signals (Figure 7—figure supplement 3A). ISCs seem to be still capable of proliferating on the HS diet, as expression of Rasv12 and Tor-DER led to a strong increase in proliferation, comparable to midguts on the HY diet. Uncoupling is mediated by a Gcn2-dependent stress response that decreases global translation in the midgut, leading to a decrease in production of niche signals. A caveat of this conclusion is that we relied on pH3 as the sole readout for these experiments. It is possible to speculate that, despite having increased pH3 counts, ISCs may be cycling slower or entertaining different modalities of division (symmetric vs. asymmetric), which would result in a low proliferative ability despite the increased pH3 count. Interestingly, infection with lethal dose of Pseudomonas entomophila leads to similar translation blockage, and rescuing translation alleviates pathogenicity (Chakrabarti et al., 2012). These together demonstrate that global translation levels can affect ISC coupling to its niche in multiple physiological contexts. We therefore propose a general model of ISC control where stress and nutrients alter both transcription and translation of niche-derived signals, ultimately leading to ISC proliferation.

Conclusions

In this study, we find that the midgut plastically resizes in response to shifts in diet composition. This resizing is mediated by the balance of cell gain and loss, stem cell-niche coupling, and EC size. Dietary manipulation also challenges our view of epithelial dynamics in a physiological context, demonstrating that nutrients affect the rates of both cell loss and cell gain independently. Importantly, diet-dependent midgut plasticity is a variable trait in the population, opening the possibility that interindividual differences in nutritional physiology originate in variations in midgut plasticity. Our work points to intestinal plasticity and its regulation by diet as an important, overlooked, and complex phenomenon likely to impact health and disease.

Materials and methods

Key resources table

Request a detailed protocol

See Appendix 1—key resources table.

Fly stocks

Request a detailed protocol

Drosophila stocks were maintained at room temperature (~23°C) on yeast-cornmeal medium (pre-experiment diet, Supplementary file 1) or at 18°C in a 12:12 hr light/dark cycle incubator. Canton-S (Cs) (BDSC: 64349), a wild-type inbred line, was used as a wild-type for all experiments not involving transgenic constructs or the DGRP panel. Gal4 drivers used were ‘w-;Esg-Gal4; UAS-GFP, tub-Gal80TS (EsgTS, progenitor-specific, Micchelli and Perrimon, 2006); ‘Esg-Gal4, tub-Gal80TS, UAS-mcherry-CD8” EsgTS, progenitor-specific, Nagy et al., 2018; ‘Esg-Gal4, UAS-GFP, tub-Gal80TS; Act>STOP>Gal4,UAS-flp’ (EsgF/O, progenitors + marked lineage, Jiang et al., 2011); ‘w-; Myo1A-Gal4, UAS-GFP, tub-Gal80TS; upd3-lacZ’ (MyoTS, EC-specific, Buchon et al., 2010); ‘Actin5C-Gal4/Cyo; tub-Gal80TS,UAS-GFP’ (ActTS, whole fly); ‘5966GS’ (EC-EB-specific, gene-switch RU486-dependent, Guo et al., 2013); ‘hsFlp; Act>STOP>Gal4,UAS-GFP’ (single-cell clonal system, Ito et al., 1997); ‘P{ry[+ t7.2] = hsFLP}12,y[1] w[*]; P{w[+ mC] = UAS GFP.S65T}Myo31DF[T2]; P{w[+ mC] = Act5 C(-FRT)GAL4.Switch.PR}3/TM6B, Tb[1]’ (ActGS, this is P{Act5C(FRT.y[+])GAL4.Switch.PR}3 with the y[+] FRT cassette removed by FLP recombination, BDSC 9431). UAS lines used were (with reference or Bloomington Drosophila Stock Center code in parenthesis) UAS-Histone2B-RFP (Mayer et al., 2005), UAS-Gcn2-IR (67215), UAS-LK6-IR (60003), UAS-EGFR-IR (60012), UAS-Tor-IR (34639), UAS-Myc-IR (36123), UAS-raptor-IR (34814), UAS-S6k-IR (41702), UAS-SREBP-IR (34073), UAS-RagA-B-IR (34590), UAS-RagC-D-IR (32342), UAS-Atg8a-IR (34340), UAS-Atg2-IR (34719), UAS-Ras85DV12 (64195), UAS-Tor-DER (‘yw; Pw+; (UAS torD-DER)’ II RJH430, gift from M. Freeman), UAS-upd3-OE (gift of M. Crozatier) (Brown et al., 2001), UAS-spi-SEC (58436), UAS-Rheb-OE (9690), UAS-rpr-OE (5823), UAS-InR-DN (8252, the line was backcrossed six times into an outbred population with a w-Dah background). Reporter lines used: GBE-Su(H)-lacZ (Micchelli and Perrimon, 2006), thor-lacZ (9558). DGRP: we used 188 lines; measurements can be found in Figure 1—source data 2 (Mackay et al., 2012).

Experimental design

Request a detailed protocol

Mated female flies were used for all experiments, except for Figure 1—figure supplement 1C. Flies, of which the progeny would be used for experiments, were allocated in equal numbers (~20 females and ~10 males) in tubes containing the pre-experiment diet and allowed to seed the tube for 2 days in a 12:12 hr light cycle at 25°C incubator. F1 progeny were collected every 3 hr on the day of eclosion and immediately transferred to the experiment diets to avoid confounding effects of adult feeding on the pre-experiment diet. To generate conditional knockdowns/overexpressions with the Gal4–Galt80TS system (TARGET; McGuire et al., 2003), crosses were made using ~15 female flies and 5 males and transferred during development to a 12:12 hr light cycle and 18°C incubator on the pre-experiment diet. Gal4–Galt80TS parents were always females and were crossed to male flies carrying UAS constructs. Parents were removed after 5 days to control for fly density of the F1 progeny. The F1 progeny were collected every 3 hr on the day of eclosion and transferred to experimental diets, where they were kept at 18°C for 5 days to allow for proper midgut development. F1 progeny were then moved for 7 days at 29°C to allow for transgene induction to take effect. This timepoint was considered as day 0; on this day, some flies were dissected from each group and the remaining were shifted from HS to HY, or from HY to HS, and kept on the new diet for 7 days until dissection. Gal4–Galt80TS flies crossed to Cs were used as experimental control, except for experiments involving UAS-InR-DN, where w-Dah was crossed to Gal4–Galt80TS flies and used as a control. For experiments involving overexpression of UAS-RasV12 and UAS-Tor-DER, F1 flies were kept during development at 18°C to keep the TARGET system off. At eclosion, flies were transferred on HS diet for 2 days and split afterward on HS and HY for 3 days. Flies were then shifted at 29°C to induce transgenes, and midgut were dissected after ~12–16 hr induction. For experiments involving overexpression of UAS-upd3-OE and UAS-spi-SEC, we followed the same protocol as for UAS-RasV12, but flies were dissected after 3 days at 29°C. For experiments involving genetic manipulation with 5966GS, we utilized the same timing as the experiments performed with the Gal4–Galt80TS system but supplying food with RU486 to induce UAS-driven transgenes, and flies were kept at room temperature for the whole duration of the experiments. 100 μL of a 5 mg/mL solution of RU486 (Cat# M8046, Sigma-Aldrich) in 80% ethanol was added on top of the food and dried for at least 16 hr to allow for vehicle evaporation, as previously published (Biteau et al., 2010). Flies of identical genotype were used as control for this experiment but were moved instead onto vials containing only the vehicle solution (80% ethanol), which were manipulated in the same way as the vials containing RU486.

Food production

Request a detailed protocol

Food was cooked in an Erlenmeyer flask (Corning, Glendale, AZ) on a hot plate with a magnetic stirrer (VWR, Radnor, PA). Half of the total volume of water together with agar (SKU# 41054, Mooragar, Rocklin, CA) was brought to boiling temperature while stirring. The water-agar solution was then removed from the heat and cooled while stirring. Yeast (Cat# 903312, MP Biomedicals, Irvine, CA), sugar (Walmart, Bentonville, AR), and other diet-specific ingredients were added at this point unless temperature labile. More water, up to 90% of the final volume, was then added. Food was allowed to cool to ~60°C, before adding acid mix (in the amount specified in Supplementary file 1), and water to reach the final volume. Acid mix recipe: for 1 L, 418 mL of propionic acid (CAS # 79-09-4, EMD Millipore, Darmstadt, Germany), 41.5 mL of phosphoric acid (Cat# 2796-16, Macron, Avantor, Center Valley, PA), 540.5 mL of water. Food was cooled to ~40°C before being aliquoted (8 mL per vial [Cat# 75813-162], VWR). Other ingredients used: inulin (Cat# CAAAA18425-09, VWR), cellulose (Cat# IC19149980, VWR), pectin (Cat# P9135, Sigma-Aldrich), lard (land o lakes, Walmart), casein (Cat# C5679, Sigma-Aldrich), AA mix (TD.10473 and TD.110036, Harlan Laboratories, Inc, IN; Lee and Micchelli, 2013), vitamin and mineral mix (TD.10475, Harlan Laboratories; Lee and Micchelli, 2013), cholesterol (Cat# C8667, Sigma-Aldrich), moldex (Cat# QB-A611-0572-159, Neta Scientific, Hainesport, NJ), yellow cornmeal (Aunt Jemima, Walmart), glucose (Cat# A16828, Alfa Aesar, Tewksbury, MA), maltose (Cat# M5885, Sigma-Aldrich), fructose (Cat# F0127, Sigma-Aldrich), arabinose (Cat# 80502-266, VWR), and sorbitol (Cat# 76177-308, VWR). Bloomington standard diets were made with nutri-fly mixes: BL cornmeal (Cat# 66-112, Genesee Scientific, El Cajon, CA) and BL Molasses (Cat# 66-116, Genesee Scientific). Refer to Supplementary file 1 for a detailed list of ingredients for each diet.

Immunohistochemistry and histology

Request a detailed protocol

After dissection, Drosophila midguts were fixed in 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA, Cat# 15713S) in 1× PBS (Cat# 003002, Thermo Scientific, Waltham, MA) for 45–90 min and subsequentially washed three times with 0.1% Triton X-100 (Cat# T8787, Sigma, St. Louis, MO) in PBS. Midguts to be immunostained were then incubated for an hour in blocking solution (1% bovine serum albumin [Cat# 12659, EMD Chemicals, San Diego, CA], 1% normal donkey serum [RRID:AB_2337258, Jackson Laboratories, West Grove, PA] in PBS). Overnight primary antibody staining was performed at room temperature in the blocking solution. Midguts were washed three times with 0.1% Triton X-100 in PBS, and overnight secondary antibody staining was performed in blocking solution. Primary antibodies used: rabbit anti-pH3 (1:000, Cat# 06-570, EMD Millipore-Sigma), mouse anti-pH3 (1:000, Cat# 05-806, EMD Millipore-Sigma), rabbit anti-β-galactosidase (1:1000, Cat# A11132, Invitrogen, Carlsbad, CA), mouse anti-Prospero (1:100, Cat# MR1A, DSHB), rabbit anti-mesh (1:2000, Cat# 995-1, gift from Mikio Furuse, Izumi et al., 2012), rabbit anti-p-eIF2α (1:500, Cat# 3398, Cell Signaling Technologies, Danvers, MA), and mouse anti-puromycin (1:100, Cat# PMY-2A4, DHSB). Secondary antibodies used: donkey anti-rabbit Alexa 555 (1:2000, Cat# A31572, Thermo Fisher), donkey anti-mouse Alexa 555 (1:2000, Cat# A31570, Thermo Fisher), donkey anti-rabbit Alexa 647 (1:2000, Cat# A31573, Thermo Fisher), and donkey anti-mouse Alexa 647 (1:2000, Cat# A31571, Thermo Fisher). DNA was stained with DAPI in PBS and 0.1% TritonX (1:50,000 from a stock with 10 mg/mL concentration, Cat# D9564, Sigma-Aldrich) for 30 min, and samples received a final three washes in PBS before mounting in Citifluor AF1 antifade medium (Cat #17970-100, Electron Microscopy Sciences). Imaging was performed on a Zeiss LSM 700 fluorescent/confocal inverted microscope (Zeiss, Oberkochen, Baden-Württemberg, Germany).

Midgut length and width measurements

Request a detailed protocol

To measure midgut length, width, and area, tiled images of entire midguts were acquired with fluorescence imaging with a 10× objective and assembled into single images of each midgut with Zen imaging software (Zeiss). Midgut length, width, and area were measured with FIJI (Schindelin et al., 2012) by drawing of a spline line along the midgut (length), or perpendicularly to it (width) (see Figure 1D for an example), and drawing a polygonal selection along the region of interest (area).

Nano-CT scan

Request a detailed protocol

Flies were treated as previously published (Mattei et al., 2015) for image acquisition. Images were visualized with OsiriX DICOM viewer (Rosset et al., 2004). To create 3D reconstructions, organs were visually identified and manually marked for each image composing the nano-CT scan stack. Video clips were rendered with OsiriX and assembled using Adobe Premiere Pro.

Feeding assay

Request a detailed protocol

We used a modified version of the method described in Min and Tatar, 2006 to allow survey of solid food passage. FD&C1 blue dye (Cat# 700010-048, VWR), which is a dye not affected by pH and digestive enzymes (Shimada et al., 1987; Tanimura et al., 1982), was added to solid food. The volume of food ingested is proportional to the concentration of dye found in the feces. Pools of 20 female flies and 5 males were placed for 24 hr in a 50 mL tube, the cap of which contained 2 mL of the diet of interest, dyed with 0.25% of FD&C1 blue dye. Feces deposited on the side of the falcon tube were resuspended with 2 mL of water and the optical density of the resultant solution was measured at 625 nm with a SmartSpec 3000 Spectrophotometer (Bio-Rad, Hercules, CA). The same flies were subjected to multiple (five) consecutive measurements for each repeat and showed similar measurements at each timepoint. These consecutive measurements also ensured that midgut capacity did not interfere with the amount of measured egested food since the midgut will be already filled with blue food.

Generation of germ-free flies

Request a detailed protocol

Eggs were suspended in 1× PBS and successively rinsed in 70% EtOH for 1 min and dechorionated using 10% bleach for ~10 min. Eggs were then transferred under a sterile laminar flow hood, where they were rinsed three times with sterile ddH2O. The eggs were finally transferred into sterile vials with sterilized fly food. Flies were tested for the presence of bacteria after each experiment by plating homogenates on MRS agar plates.

DGRP analysis

Request a detailed protocol

Response indices were calculated as per-line difference in mean length, divided by whole-DGRP difference in mean length. Broad sense heritability of the response was calculated as the proportion of variance of the ratio of midgut length from each diet within line (as given by the estimate of the random effect ‘DGRP ID’) over the total variance in ratio (given by variance within line summed with residual variance). To determine how many DGRP lines responded to the diet treatment, we tested for the difference in total midgut length between the two diets for each of the DGRP line with a t-test. The p-value of each of the comparisons was corrected for multiple tests using a false discovery rate method.

Genome-Wide Association Study (GWAS) analysis

Request a detailed protocol

The genetic diversity of the DGRP lines comprises about 4 million SNPs. We selected the SNPs (n = 2,174,256) for our association study based on two criteria: (1) avoid a complete collinearity (possibly confounding) between alleles and Wolbachia status (i.e., we excluded cases where one allele corresponds to Wolbachia infection and the other to an uninfected status); and (2) we had enough lines per treatment to run the model. Prior to each test, we therefore calculated a two-by-two matrix with Wolbachia status and allele identity (i.e., W+/allele1, W-/allele1, W+/allele2, W-/allele2) summarizing the sum of lines for each category. We further included in our association only the SNPs where at least three of the categories had five lines. All the analyses were performed in R. To test for the response to diet, we next estimated the significance of the difference between alleles of the difference in midgut length on each diet at each selected SNP. We used a Generalized Linear Mixed Model (GLMM, function HLfit from the R package ‘spaMM’). The model was as follows: 'DifferenceTotalLength'~ DifferenceGenotypeWeight + SNP + (1|wolbachia/DGRP_lines) + (1|block). The variable ‘Block’ accounted for group of flies dissected in the same day (it would not be possible to dissect the whole DGRP for one repeat in 1 day), and the identity of the lines was accounted for as random effect following a Gaussian distribution. We compared the log likelihood of the complete model to a model lacking the main effect SNP to calculate p-values. We performed a likelihood ratio test in R as follows: pchisq(Chi2_LRT_snp, df = 1, lower.tail = F) where Chi2_LRT_snp is 2 × (log likelihood complete model – log likelihood reduced model).

Candidate SNPs had p-values between 10–6 and 10–9. To understand the difference between alleles at a given candidate SNP, we characterized the implication of the mutation on gene function (e.g., missense mutation, point mutation in regulatory sequence, etc.). The DGRP SNP positions are provided for the version 5 of the D. melanogaster genome. We then converted the positions in the equivalent for version 6 with the convert tool from Flybase. The characterization of the mutation at each candidate SNP was then provided using the Variant Effect Predictor (VEP) from the website Ensembl (http://www.ensembl.org/info/docs/tools/vep/index.html). Full results are provided (Figure 1—figure supplement 3A,Figure 1—source data 2 and Figure 1—source data 2), and highlighted candidates were selected based on the shape of the peak in the Manhattan plot and the function provided by VEP. In order of priority, our highlighted candidates had either non-synonymous mutations, mutations in the 5′ and 3′ UTR regions, or mutations in introns (which could be located in gene enhancers).

Cell counts pH3+ cells were counted directly through the fluorescent microscope eyepiece using a 20× objective. To score numbers of each cell type, EsgTS females were crossed to GBE-Su(H)-lacZ males. Progeny carrying both constructs were dissected and cells were labeled with anti-Prospero (EE), anti-βGAL antibodies (EB), DAPI (all cells, polyploid cells counted as EC), and the Esg-specific GFP expression (progenitor cells). Stacked images of region 4 of the midgut (Buchon et al., 2013) encompassing one hemisphere of the midgut along the dorsal/ventral axis were acquired with confocal microscopy with a 20× objective using Zen imaging software (Zeiss). Stacked images were subjected to orthogonal projection, resulting in one image comprising all cells in half of a midgut along the dorsal/ventral axis (Zen/FIJI). The number of Esg+ and GBE-Su(H)+ cells was manually counted. The number of DAPI+ cells with EC comparable size and Prospero+ cells was counted through a semi-automated macro that we developed in FIJI. ISC number was determined by subtracting GBE-Su(H)+ cells from the total number of Esg+ cells. EB number was determined by scoring GBE-Su(H)+ cells. EC number was determined by counting the DAPI stained nuclei within the proper size range. EE cells were determined by scoring Prospero+ cells. The area of the midgut in the acquired image was also measured to determine cell density. Cell density was then multiplied by total area of the region to obtain the total number of cells within the said region. This number was multiplied by 2 to account for only half of the midgut having been imaged. Total area of the region was obtained as described in the section ‘Midgut length and width measurements’.

Cell size measurement

Request a detailed protocol

For EC area measurements, midguts were stained with anti-mesh antibody to highlight cell contours. Z-stacks were acquired with a 20× objective using Zen imaging software (Zeiss). Stacked images were manually measured using the polygon selection tools in FIJI. For each cell, the Z-plane with the larger area was selected for the measurements. To avoid selection bias, the measured cells were adjacent to each other, rather than picked at random throughout the midgut (staining quality permitting). Roughly 30 cells per midgut were measured. For EC height measurements, Z-stacks with a 1 µm step size were acquired from MyoTS>Canton S, which resulted in EC cytoplasm being labeled with GFP. Z-stack position at the top and bottom of single ECs was measured, and EC height inferred. Roughly 30 ECs per midgut were measured.

Ploidy measurement

Request a detailed protocol

Dissected midguts, with proventriculus cut off (25 for sample), were frozen upon dissection in a –80°C freezer in Grace’s medium (Cat# 11605094, Thermo Fisher Scientific) with added 10% DMSO (Cat# 80058/040, VWR) until dissociation. On dissociation, samples were thawed on ice. Midguts were spun down, and freezing media was removed. 2 mL of ice-cold homogenization buffer (HB) was added with the midguts to a Dounce homogenizer. Guts were subjected to 50 strokes with loose pestle. After 5 min on ice, 10 additional strokes were performed. After five additional minutes, 20 strokes with tight pestle were performed. Samples were then run through a 70 µm cell strainer and spun down for 5 min at 500 g at 4°C. Supernatant was then removed and 600 µL of Wash buffer were added, and nuclei were resuspended in it. Samples were centrifuged again for 10 min at 750 g at 4°C and filtered a second time through a 40 µm cell strained. Wash solution was removed and 300 µL of PBS with DAPI (10 µg/mL) were added to each sample. Samples were run in an Aria Fusion sorter at Cornell Institute of Biotechnology – FACS core. Samples were analyzed with FCS express 6 (De Novo software). Samples were gated to remove doublets and debris, and ploidy peaks percentages were measured. 6X HB composition: 30 mM CaCl2 (Cat# 122950, Beantown Chemical, Hudson, NH), 18 mM Mg(Ac)2 (Cat# 12225, Alfa Aesar), 60 mM Tris pH 7.8 (Cat# 0497, VWR), 0.1 mM PMSF (Cat# A0999, AppliChem, Council Bluffs, IA), and 1 mM β-mercaptoethanol (Cat# 0482, Amresco). 1X HB: 6HB to 1HB, 320 mM sucrose (Cat# 902978, MP Biomedicals), 0.1 mM EDTA (Cat# E177, VWR), 0.1% NP40 (Cat# 19628, United States Biochemical, Cleveland, OH), and protease inhibitor tablet (Cat# A32963, Thermo Fisher Scientific). Wash buffer: 10 mM Tris-HCl pH 7.4, 10 mM NaCl (Cat# 470302-512, Ward’s Science, St. Catharines, ON, Canada), 3 mM MgCl2 (Ca# 442611, EMD Chemicals), 0.1 % Tween-20 (Cat# P1379, Sigma-Aldrich).

Cell loss assay

Request a detailed protocol

5966GS>Histone2B-RFP flies were fed either HS or HY diet for 5 days from eclosion. RU486 supplemented food was produced as previously described. Flies were then fed on HS and HY supplemented with RU486 for 3 days, moved back on HS and HY without RU486 for 2 days to allow for its elimination, and dissected (day 0). Flies were subsequently shifted to HS and HY for 14 days, when they were dissected, to create the four conditions analyzed (HS to HS, HS to HY, HY to HY, HY to HS). Confocal Z-stacks of half a midgut hemisphere were acquired with a 20× objective to determine the number of RFP+ and RFP- cells. The area of the midgut in the acquired image was also measured to determine cell density. Tiled images were acquired with a 10× objective to determine regional midgut area, which was measured with FIJI as previously described (see section ‘Midgut length and width measurements’). Density was multiplied by midgut regional area and then by 2 (to account for only half midgut being imaged) to determine the total cell number per midgut region. To calculate turnover, we divided the new cells made by the number of lost cells. To find the total number of lost cells, we had to factor both RFP+ and RFP- lost cells (not 100% of polyploid cells was marked at day 0). To calculate the RFP+ lost cells, we subtracted ending RFP+ cell number from initial RFP+ cell number (initial RFP+ - ending RFP+). The number of lost cells that were initially unmarked was then calculated by multiplying the amount of initial unmarked cells (initial RFP-) by the ratio of RFP+ leaving [1 – (ending RFP+/initial RFP+)]. This resulted in the formula describing number of unmarked cells lost to be: initial RFP- * [1 – (ending RFP+/initial RFP+)]. Addition of these two numbers of cells lost (marked and unmarked) gave us the total number of lost cells: (initial RFP+ – ending RFP+) + (initial RFP- * [1 – (ending RFP+/initial RFP+)]). New cells were calculated by subtracting from the ending number of RFP- cells, the amount of initial RFP- cells – the amount of initial RFP- cells lost (as previously calculated). This results in new cells being calculated with the following formula: ending RFP- – (initial RFP- – ((initial RFP- * [1 – (ending RFP+/initial RFP+)]))). The final resulting formula is as follows: ending RFP- – (initial RFP- – ((initial RFP- * [1 – (ending RFP+/initial RFP+)])))/((initial RFP+ – ending RFP+) + (initial RFP- *[1 – (ending RFP+/initial RFP+)])).

RNA-seq generation and analysis

Request a detailed protocol

50 midguts for each condition/repeat (three repeats) were dissected at eclosion and on days 1, 2, 3, and 5 post eclosion, immediately transferred to ice-cold Trizol (Cat# 15596018, Life Technologies, Carlsbad, CA), and homogenized. Total RNA was isolated using a hybrid Trizol-Rneasy (Cat# 74106, Qiagen, Hilden, Germany) protocol, as previously published (Houtz et al., 2019; Troha et al., 2018). RNA was quantified with Qubit (Thermo Fisher) and quality checked via a fragment analyzer at Cornell genomic facility. QuantSeq 30 mRNA-Seq Library Prep Kit FWD (Cat# 015.2X96, Lexogen, Vienna, Austria) was utilized to prepare 3′ end RNA-seq libraries. Libraries were quality checked before pooling and sequencing with the Illumina Nextseq 500 platform by the Biotechnology Resource Center (BRC) Genomics Facility at the Cornell Institute of Biotechnology (http://www.biotech.cornell.edu/brc/genomics-facility). 5–6 million reads were sequenced per sample, which approximately equals a 20× coverage by conventional RNA-seq. Libraries were checked by FastQC, adaptors trimmed by CutAdapt (Martin, 2011), aligned to the fly genome using Tophat (Trapnell et al., 2009), and counted with HTSeq (Anders et al., 2015). Enumerated reads were analyzed in R (3.3.1)/BioConductor using DESeq2, testing the effect of diet with a likelihood ratio test, using day 3 and 5 data. GO analysis was performed in R with topgo. Data have been submitted to ArrayExpress (E-MTAB-10812).

Translation assay

Request a detailed protocol

General translation was measured with a puromycin incorporation assay (David et al., 2012; Deliu et al., 2017). Cs flies fed either HS or HY diet for 5 days were moved to HS or HY diet, which had 150 µL of 5 µL/mL of puromycin (Cat# P8833, Sigma-Aldrich) in water added to the surface of the food just prior to the experiment. Flies were dissected after 3 hr, following standard immunochemistry protocol as previously described.

Single-cell genetic manipulation

Request a detailed protocol

hsFlp; Act>STOP>Gal4,UAS-GFP flies have a base leakiness in their activity. This system was crossed with UAS constructs of interest to generate GFP marked single cells expressing the UAS construct of interest in an otherwise normal midgut. We acquired images of these midguts stained with anti-mesh antibody to mark cell membranes and measure cells size. For each GFP+ clone (driving the transgene of interest), we quantified ~5 GFP- control cells (not driving the transgene of interest) from the same image.

Statistical analysis

Request a detailed protocol

We provide the complete statistical formula and raw data used for each experiment in the R markdown file (Supplementary file 2; https://dduneau.github.io/Bonfini_eLife_2021/Bonfini_eLife_2021.html). In brief, we mostly used generalized linear mixed models (function fitme from the R package ‘spaMM’; Rousset and Ferdy, 2014). To compare the difference between factors, such as diet or genotype, the model was, for example, as follows: Gut_length~ Diet + (1|Repeat). To compare the difference in response, such as the midgut length on each diet for different genotypes, the model was as follows: Gut_length~ Diet + Genotype+ Diet:Genotype (1|Repeat), where ‘Diet:Genotype’ represents the interaction between the variables. The variable ‘Repeat’ describes the experimental replication and was accounted for as random effect following a Gaussian distribution. We then tested the difference between main effect (or the difference in response) by comparing the log likelihood of the complete model to a model lacking the main effect (or the interaction) to calculate p-values (displayed in the figures). We performed a likelihood ratio test in R as follows: pchisq(Chi2_LRT, df = 1, lower.tail = F), where Chi2_LRT is 2 × (log likelihood complete model – log likelihood reduced model). Normal distribution and homoscedasticity of the residuals were tested with Shapiro–Wilk normality tests and Brush–Pagan tests, respectively. In many cases, the response variable was log transformed to improve model fit. To characterize differences between several conditions, general linear hypotheses tests were applied, using a Tukey post hoc pairwise comparisons (i.e., fitting an adequate model followed by a glht function -alpha = 0.05 from the package multcomp in R; Hothorn et al., 2008). Pearson correlation tests were performed (function cor.test from the default R stat package) for Figures 6A and 7A. The p-values were corrected and adjusted by the false discovery rate correction (Benjamini and Hochberg, 1995).

Appendix 1

Appendix 1—key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
genetic reagent (D. melanogaster)Canton-SBloomington Drosophila Stock CenterBDSC: 64349FLYB: FBst0064349RRID:BDSC_64349Flybase genotype: Canton-S
genetic reagent (D. melanogaster)EsgTSMicchelli and Perrimon, 2006N/Aw-;Esg-Gal4; UAS-GFP, tub-Gal80TS
genetic reagent (D. melanogaster)EsgTSNagy et al., 2018N/Aw-;Esg-Gal4,tub-Gal80TS,UAS-mcherry-CD8
genetic reagent (D. melanogaster)EsgF/OJiang et al., 2011N/Aw-;Esg-Gal4, UAS-GFP, tub-Gal80TS; Act > STOP > Gal4,UAS-flp
genetic reagent (D. melanogaster)MyoTSBuchon et al., 2010N/Aw-; Myo1A-Gal4, UAS-GFP, tub-Gal80TS; upd3-lacZ
genetic reagent (D. melanogaster)ActTSThis publicationN/AActin5C-Gal4/Cyo; TubGal80ts,UasGFP
genetic reagent (D. melanogaster)5,966GSGuo et al., 2013FLYB: FBti0150384w-; 5,966GS/Cyo
genetic reagent (D. melanogaster)Single cell clonal systemIto et al., 1997N/AhsFlp; Act > STOP > Gal4,UAS-GFP
genetic reagent (D. melanogaster)ActGSBloomington Drosophila Stock CenterBDSC: 9431FLYB: FBst0009431RRID: BDSC_9431P{ry[+ t7.2] = hsFLP}12,y[1] w[*]; P{w[+ mC] = UAS GFP.S65T}Myo31DF[T2]; P{w[+ mC] = Act5 C(-FRT)GAL4.Switch.PR}3/TM6B, Tb[1]
genetic reagent (D. melanogaster)UAS-His-RFPMayer et al., 2005N/AUAS-Histone2B-RFP
genetic reagent (D. melanogaster)UAS-Gcn2-IRBloomington Drosophila Stock CenterBDSC: 67215FLYB: FBst0067215RRID: BDSC_67215
genetic reagent (D. melanogaster)UAS-LK6-IRBloomington Drosophila Stock CenterBDSC: 60003FLYB: FBst0060003RRID: BDSC_60003
genetic reagent (D. melanogaster)UAS-EGFR-IRBloomington Drosophila Stock CenterBDSC: 60012FLYB: FBst0060012RRID: BDSC_60012
genetic reagent (D. melanogaster)UAS-Tor-IRBloomington Drosophila Stock CenterBDSC: 34639FLYB: FBst0034639RRID: BDSC_34639
genetic reagent (D. melanogaster)UAS-Myc-IRBloomington Drosophila Stock CenterBDSC: 36123FLYB: FBst0036123RRID: BDSC_36123
genetic reagent (D. melanogaster)UAS-raptor-IRBloomington Drosophila Stock CenterBDSC: 34814FLYB: FBst0034814RRID: BDSC_34814
genetic reagent (D. melanogaster)UAS-S6k-IRBloomington Drosophila Stock CenterBDSC: 41702FLYB: FBst0041702RRID: BDSC_41702
genetic reagent (D. melanogaster)UAS-SREBP-IRBloomington Drosophila Stock CenterBDSC: 34073FLYB: FBst0034073RRID: BDSC_34073
genetic reagent (D. melanogaster)UAS-RagA-B-IRBloomington Drosophila Stock CenterBDSC: 34590FLYB: FBst0034590RRID: BDSC_34590
genetic reagent (D. melanogaster)UAS-RagC-D-IRBloomington Drosophila Stock CenterBDSC: 32342FLYB: FBst0032342RRID: BDSC_32342
genetic reagent (D. melanogaster)UAS-Atg8a-IRBloomington Drosophila Stock CenterBDSC: 34340FLYB: FBst0034340RRID: BDSC_34340
genetic reagent (D. melanogaster)UAS-Atg2-IRBloomington Drosophila Stock CenterBDSC: 34719FLYB: FBst0034719RRID: BDSC_34719
genetic reagent (D. melanogaster)UAS-Ras85DV1Bloomington Drosophila Stock CenterBDSC: 64195FLYB: FBst0064195RRID: BDSC_64195
genetic reagent (D. melanogaster)UAS-Tor-DERgift from M. FreemanN/Ayw; Pw+; (UAS torD-DER) II RJH430,
genetic reagent (D. melanogaster)UAS-upd3-OEgift of M. Crozatier Brown et al., 2001N/A
genetic reagent (D. melanogaster)UAS-spi-SECBloomington Drosophila Stock CenterBDSC: 58436FLYB: FBst0058436RRID: BDSC_58436
genetic reagent (D. melanogaster)UAS-Rheb-OEBloomington Drosophila Stock CenterBDSC: 9690FLYB: FBst009690RRID: BDSC_9690
genetic reagent (D. melanogaster)UAS-rpr-OEBloomington Drosophila Stock CenterBDSC: 5823FLYB: FBst005823RRID: BDSC_5823
genetic reagent (D. melanogaster)UAS-InR-DNBloomington Drosophila Stock CenterBDSC: 8252FLYB: FBst008252RRID: BDSC_8252The 8,252 line was backcrossed six times into an outbred population with a w-Dah background
genetic reagent (D. melanogaster)GBE-Su(H)-lacZMicchelli and Perrimon, 2006N/A
genetic reagent (D. melanogaster)thor-lacZBloomington Drosophila Stock CenterBDSC: 9558FLYB: FBst009558RRID: BDSC_9558
genetic reagent (D. melanogaster)DGRP panelMackay et al., 2012N/ALines used are reported in Figure 1—source data 2
chemical compound, drugRU486Sigma-AldrichCat# M8046;Puchem#: 24278572100 μL of a 5 mg/mL solution of RU486 per vial
chemical compound, drugFD&C1 blue dyeVWRCat# 700010–048; MDL# MFCD00012141
chemical compound, drugPuromycinSigma AldrichCat# P8833;MDL# MFCD00012691; PubChem# 24898984
chemical compound, drugTrizolLife TechnologiesCat#15596018
otherYeastMP biomedicalsCat# 903,312
otherinulinVWRCat# CAAAA18425-09
othercelluloseVWRCat# IC19149980
otherpectinSigma AldrichCat# P9135
otherlardWalmartLand o lakes
othercaseinSigma AldrichCat# C5679
otherAA mixHarlan Laboratories, Lee and Micchelli, 2013Cat# TD.10473 & TD.110036
othervitamin and mineral mixHarlan Laboratories; Lee and Micchelli, 2013Cat# TD.10475
othercholesterolSigma AldrichCat# C8667
othermoldexNeta ScientificCat# QB-A611-0572-159
otheryellow cornmealWalmartAunt Jemima
otherglucoseAlfa AesarCat# A16828
othermaltoseSigma AldricCat# M5885
otherfructoseSigma AldrichCat# F0127
otherarabinoseVWRCat# 80502–266
othersorbitolVWRCat# 76177–308
othernutri-fly BL cornmealGenesee scientificCat# 66–112
otherBL MolassesGenesee scientificCat# 66–116
otherDAPISigma-AldrichCat# D95641:50,000 of a 10 mg/mL stock solution
otherCitifluor AF1Electron Microscopy SciencesCat #17970–100
antibodyAnti-pH3 (Rabbit polyclonal)EMDMillipore - SigmaCat# 06-570RRID:AB_310177IF 1:1,000
antibodyAnti-pH3 (Mouse monoclonal)EMDMillipore - SigmaCat# 05-806RRID:AB_310016IF 1:1,000
antibodyanti-β-Galactosidase (Rabbit polyclonal)InvitrogenCat# A11132RRID:AB_221539IF 1:1,000
antibodyAnti-Prospero (Mouse monoclonal)DSHBCat# MR1ARRID:AB_528440IF 1:100
antibodyAnti-Mesh (Rabbit polyclonal)Gift from Mikio Furuse, Izumi et al., 2012995-1RRID:AB_2568117IF 1:2000
antibodyAnti-peIF2α (Rabbit monoclonal)Cell signaling technologiesCat# 3398RRID:AB_2096481IF 1:500
antibodyAnti-puromycin (mouse monoclonal)DHSBCat# PMY-2A4RRID:AB_2619605IF 1:100
antibodyAnti-mouse Alexa 555 (Donkey polyclonal)Thermo FisherCat# A31570RRID:AB_2536180IF 1:2000
antibodyAnti-rabbit Alexa 555 (Donkey polyclonal)Thermo FisherCat# A31572RRID:AB_162543IF 1:2000
antibodyAnti-mouse Alexa 647 (Donkey polyclonal)Thermo FisherCat# A31571RRID:AB_162542IF 1:2000
antibodyAnti-rabbit Alexa 647 (Donkey polyclonal)Thermo FisherCat# A31573RRID:AB_2536183IF 1:2000
commercial assay or kitQuantSeq 30 mRNA-Seq Library Prep Kit FWDLexogenCat#015.2 × 96
software, algorithmFijiSchindelin et al., 2012RRID# SCR_002285
software, algorithmOsiriX DICOM viewerRosset et al., 2004RRID# SCR_013618
software, algorithmAdobe Premiere proAdobeRRID# SCR_021315
software, algorithmAdobe PhotoshopAdobeRRID# SCR_014199
software, algorithmAdobe IllustratorAdobeRRID# SCR_010279
software, algorithmRstudioRStudio Team (2020). RStudio: Integrated Development for R. RStudio, PBC, Boston, MA URL http://www.rstudio.com/.RRID# SCR_000432
software, algorithmFCS express 6De Novo softwareRRID:SCR_016431https://denovosoftware.com/
software, algorithmFastQCBabraham BioinformaticsRRID:SCR_014583https://www.bioinformatics.babraham.ac.uk/index.html
software, algorithmCutadaptMartin, 2011RRID:SCR_011841
software, algorithmTopHatTrapnell et al., 2009RRID:SCR_013035https://ccb.jhu.edu/software/tophat/index.shtml
software, algorithmHTSeqAnders et al., 2015RRID:SCR_005514https://htseq.readthedocs.io/en/master/

Data availability

Data have been submitted with an ArrayExpress accession E-MTAB-10812.

References

  1. Book
    1. Demerec M
    (1950)
    Biology of Drosophila
    Cold Spring Harbor Laboratory Press.
    1. Ito K
    2. Awano W
    3. Suzuki K
    4. Hiromi Y
    5. Yamamoto D
    (1997)
    The Drosophila mushroom body is a quadruple structure of clonal units each of which contains a virtually identical set of neurones and glial cells
    Development 124:761–771.
    1. Navarrete J
    2. Vásquez B
    3. Del Sol M
    (2015)
    Morphoquantitative analysis of the ileum of c57bl/6 mice (Mus musculus) fed with a high-fat diet
    International Journal of Clinical and Experimental Pathology 8:14649–14657.
  2. Book
    1. Simpson SJ
    2. Raubenheimer D
    (2012)
    The Nature of Nutrition: A Unifying Framework from Animal Adaptation to Human Obesity
    Princeton University Press.

Decision letter

  1. Lucy Erin O'Brien
    Reviewing Editor; Stanford University School of Medicine, United States
  2. Utpal Banerjee
    Senior Editor; University of California, Los Angeles, United States
  3. Lucy Erin O'Brien
    Reviewer; Stanford University School of Medicine, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This tour-de-force study provides a granular, quantitative, cell-to-organ scale analysis of dietary impact on tissue homeostasis and morphological plasticity in the Drosophila adult intestine. The authors systematically and dynamically vary dietary composition and assess the relative balance of stem cell proliferation and differentiated cell loss, cell and organ morphology, and response to niche signaling. The data presented give a highly detailed view of how the intestine dynamically responds to dietary change, leading the authors to a provocative interpretation that tissue homeostasis is a metastable state.

Decision letter after peer review:

Thank you for submitting your article "Diet composition resizes the fly midgut by affecting stem cell-niche coupling and enterocyte gain, loss, and size" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by Lucy Erin O'Brien as Reviewing Editor and Reviewer #1, and Utpal Banerjee as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Summary:

The intestine's adaptive response to dietary change is conserved throughout the animal kingdom. This manuscript provides an in-depth examination of the nutritional inputs and cellular elements that mediate diet-driven adaptation of the adult Drosophila midgut; its holistic scope and rigorously quantitative approach make it one of the most comprehensive examinations to date and provide new insights that will be of wide interest.

All three reviewers recognized the overall substantial contributions of this work; at the same time, they had concerns about certain experimental conditions and particular aspects of data presentation and interpretation.

Essential revisions:

(1) Tor/autophagy experiments. Given that the myo1A driver expresses in both enterocytes and neurons, it is important to repeat at least one key Tor/autophagy experiment using a different enterocyte driver to confirm that the observed effects are due specifically to genetic manipulation of enterocytes. In addition, inclusion of Tor gain-of-function data or a readout of endogenous Tor activation in HY vs HS conditions would strengthen the authors' conclusions, if such data can be readily obtained; in light of eLife's resubmission policies during COVID, however, these gain-of-function experiments are not essential.

(2) Homeostasis. Two Reviewers had concerns that the replacement ratio experiments, which are the basis of the authors' conclusions regarding homeostasis, were performed under physiological conditions that have been shown not to be homeostatic. In addition, one Reviewer had questions regarding the robustness of cell labeling and the formulas used to calculate new and old cells. Please address these comments in the manuscript text and/or response to reviews. Additional controls for the cell labelling, if they are available, would help to validate the assay (but are not essential).

(3) Diet manipulations. Please address the following reviewer comments, either in the text/response to reviews or by providing additional data. (a) Effect of gut capacity was not taken into account for excreta-based measurements of food ingestion. (b) Effects of diet on cell-type proportions in this study appear to differ compared to Obniski Dev Cell 2018. (c) Potential alternative interpretations of "antagonistic" effect of sugar on protein.

(4) Some improvements for clarity and context. I invite the authors to consider the Reviewers' comments with regard to (a) clarifying the key takeaways for a broad audience; (b) including crucial experimental details, such as animal age and mating status, in the Results (not just the Methods); (c) improving the supplemental tables that contain the diet recipes.

Reviewer #1:

In this tour-de-force study, the authors have used an impressively thorough, deconstructed, quantitative approach to examine the resizing plasticity of the adult Drosophila midgut in response to different dietary compositions and the underlying cellular behaviors that drive these physical changes. Specifically, when comparing flies raised on a high yeast (HY) or a high sucrose (HS) diet, they found that a HY diet increased midgut length, while a HS diet appeared to antagonize this effect. This diet-dependent organ resizing involves changes in cell number and cell size, and is both reversible and repeatable throughout the animal's adult life. HS promotes growth pathway gene expression, but without increased ISC proliferation. However, ISC proliferation alone cannot explain HY-induced midgut length increase. Rather, increased EC size, through TOR signaling, is required to increase midgut length, as blocking TOR signaling prevents midgut growth upon shifting from HS to HY diets.

This unprecedented work provides a highly granular, multi-dimensional understanding of diet-driven organ size plasticity. It examines multiple aspects that are more typically considered only in isolation (e.g. nutritional geometry, GWAS analyses, whole-organ morphometry/allometry, total counts of individual cell types, RNA seq analyses, detailed time courses) and synthesizes them into a comprehensive amalgamation of resizing phenomena. Along the way, the authors arrive at a number of important and surprising conclusions that will reverberate in the midgut field and in adult stem cell and tissue biology more generally. Further, the often complex data was superbly presented in creative and clear graphics, and the text was a true pleasure to read.

It should be noted that while the calculation of a "replacement ratio" is an illuminating and original approach to quantify cell turnover, there are three important elements that might deserve consideration. Specifically, the authors' calculations would benefit from further clarifications. Furthermore, there seems to be no evidence for highly efficient and uniform His::RFP retention, two essential preconditions for their replacement assay. Finally, the authors' interpretation that homeostasis may be merely "apparent" rests on an unaddressed, and potentially incorrect, assumption that replacement ratios are be constant throughout the animal's lifetime, and does not consider the possibility that guts are likely initiating age-associated dysplasia during the chase period.

Major comments:

1. The calculation of a "replacement ratio" is an illuminating and original approach to quantify cell turnover. The authors' findings that replacement ratios were significantly different from 1 for all conditions that they tested could be paradigm-challenging--if it is correct.

While I feel that this analysis was thoughtfully approached, there are three important elements that I either did not understand or else feel deserve consideration. Details are below. In brief: (a-d) I was confused by some parts of the authors' calculations. (e) I am concerned that there was no evidence for highly efficient and uniform His::RFP retention, two essential preconditions for their replacement assay. (f) Finally, I feel that the author's interpretation that homeostasis may be merely "apparent" rests on an unaddressed, and potentially incorrect, assumption that replacement ratios are be constant throughout the animal's lifetime, and does not consider the possibility that guts are likely initiating age-associated dysplasia during the chase period.

a. If I am understanding the analyses correctly, then it seems that gut size measurements do not support the calculated replacement ratios. For instance, the HY ratio from labeling days 0-14 (days 10-24 post-eclosion) is 1.64--this would imply that the gut should double in size every ~15 days. But the HY gut size, as measured by the authors, stays nearly the same over days 7-21 (Figure 4S1C) and only increases modestly by day 28. Can the authors explain these differences?

b. I was confused by the actual formula described in the Methods. For instance, the authors state, "The number of cells lost that were initially unmarked was calculated by first calculating the percentage of RFP+ cells present after 14 days (initial RFP+ number/ final RFP+ number). Shouldn't the % of RFP+ cells present after 14 days be the final/initial RFP+ number? As another example, "The number of cells lost that were initially unmarked (initial RFP- * (initial RFP+ number/ final RFP+ number)) was then calculated." I don't understand why the product of initial RFP- and the ratio of initial/final RFP+ cells yields the numbers of RFP- cells that were lost over time. Can the authors clarify their formula?

c. More generally--since all new cells are RFP-, I am having trouble grasping how one can count the numbers of initial RFP- cells that are lost without assuming that cell addition = cell loss (which is what the authors are refuting)?

d. The authors state that "not 100% of polyploid cells was marked at day 0". What % of polyploid cells were unmarked at day 0? If RU486 is administered for a longer period, will all polyploid cells acquire labeling (i.e. have they saturated the labeling potential)?

e. To use a label-retention protocol in this highly quantitative manner, two preconditions must be true: that (1) label retention is highly efficient virtually all cells labelled with HisRFP at day 0 retain the label 14 days later, and that (2) initial labelling and label retention are the same in HS and HY conditions. The images presented by the authors in 4G-J do not validate that these preconditions are met. They show that cells are capable of retaining the label, but they do not measure how many cells retain the label or whether labeling and retention are not altered by diet. Can the authors provide evidence for diet-insensitive and efficient label retention over the course of their experiments? If not, then this major caveat should be discussed in the text and the conclusions tempered accordingly.

f. Replacement ratio are likely not constant over the chase period that was used in these experiments (10-24 days post eclosion). One might argue that when the chase period begins (day 10 post eclosion), Drosophila adults are already "past their prime" (peak egg laying is ~4-7 days post eclosion), and that during the chase period the guts are undergoing age-associated dysplasia, which is morphologically apparent by 24 days (on a "standard" fly diet). Indeed, some prior evidence (Liang 2017 Figure 1) would support the notion that replacement ratio in early adults (4-8 days post eclosion) is indeed close to 1. Have the authors examined replacement ratios for chase periods that initiate earlier (e.g. 6-7 days post eclosion) and/or are shorter in duration (e.g. 2-4 days)? If not, then the possibility that replacement ratios are not constant, and that the ratios measured in Figure 4 may be affected by age-associated dysplasia, should be discussed.

2. The interpretation of the Figure 5, S51, S52-associated experiments strikes me as overstepping some limitations of the data to discount the role of progenitors (see a-c below). I do not feel that additional experiments are needed because the study's main contribution is the organ-scale and ECs analyses and not ISCs. However, the authors may wish to consider tempering the text of the Results to allow for alternative interpretations regarding the ISC responses, similar to the more equitable treatment that ISCs receive in the Discussion.

a. The conclusions about stem cell divisions rely on one sole readout of stem cell activity in the paper: PH3 staining. This measurement is a snapshot of cells in mitosis at a single point in time. whether those cells are cycling faster or slower is unknown. In addition, PH3 staining belies richer and potentially relevant information about stem cell division behaviors, such as progeny produced per stem cell and sibling fate outcomes, which can be revealed through genetic lineage tracing.

b. The conclusion that stem cells are "uncoupled" from growth signals by inhibition of translation in stem cells under HS conditions would require multiple additional experiments to be convincingly demonstrated. As one alternative explanation is that the growth signals themselves, despite being transcribed, are not translated, secreted, or otherwise available to signal to the stem cells.

c. The authors do not consider that enteroblasts represent an additional potential "node" for organ size control--they can differentiate at different rates, or even undergo apoptosis as reported recently (Reiff, 2019).

Reviewer #2:

This is an exciting manuscript that describes and characterises differential effects of high-yeast (HY) and high-sugar (HS) diets on migdut length in adult Drosophila. Through a series of carefully controlled experiments, the authors establish that a combination of various yeast-derived nutrients promotes midgut growth, whereas sugar antagonises yeast-induced growth independent of other confounds (e.g. effects on food intake). As well as confirming previous effects of diet on intestinal stem cell (ISC) proliferation, the authors uncover an additional layer of diet-dependent ISC regulation by identifying a role for HS-induced translational inhibition in restraining ISC proliferation. They also identify enterocyte (EC) size as a better predictor of the genetic/physiological variability in midgut size, and uncover a role for EC size in mediating midgut size plasticity. Mechanistically, they suggest that the diet-dependent regulation of EC size is mediated by the TOR/autophagy pathway.

Collectively, the experiments described in this manuscript advance our understanding of the nutritional plasticity of the adult intestine at the molecular, cellular and whole-organ levels. There are only a few conclusions that require additional clarifications and/or stronger data to support them. These concern previously reported effects of lipids, the physiological role of TOR signalling, possibly dietary effects on enterocyte ploidy and stem cell division mode, and the quantifications of food intake.

1. The HS and HY diets in this study differ in lipid composition. A recent study showed that, during the initial period of postnatal gut growth, dietary cholesterol modulates the number and fraction of enteroendocrine (EE) cells via its effects on Notch signalling (PMID: 30220569). In this manuscript, cell type ratios seem unchanged in the two diets (Figure 3S1A), which seems at odds with this study. Would the authors agree? I wondered whether the partial supplementations in Figure 2B did affect EE/EC ratios, perhaps in contrast to the full HY diets?

2. All of the experiments that address contributions of TOR signalling/autophagy used the MyoIA-Gal4 driver, which is expressed in ECs, but also in quite a few neurons including enteric neurons. At least one of these experiments should be repeated with mex-Gal4 or another EC-specific driver. Also, the loss-of-function data provided so far suggests that TOR signalling can mediate the diet-dependent difference in EC size, but not necessarily that it does so. A TOR signalling gain-of-function experiment in the context of gut size, and/or demonstration that the TOR signalling status of EC normally differs between HS-and HY-fed flies, would strengthen the case for this pathway as a physiological mediator of this difference. Related to this, I am not sure that the description of the "TOR/autophagy pathway as a master regulator" in line 545 is warranted either way. What do the authors mean by master regulator in this context?

3. Also related to the previous point, the authors do not seem to have considered possible differential effects of the diets on EC ploidy and/or ISC division mode (symmetric vs asymmetric). Do the larger enterocytes have increased ploidy? Do they become larger as a result of overall cell growth/ploidy or increased cell volume/dilution?

4. The authors carefully consider possible differences in caloric intake/content between the two diets: they use isocaloric diets, and attempt to rule out differences in food intake between the two diets. However, the use of excreta as a proxy for the latter is only justified when gut capacity remains unchanged (PMID 21195352), which is unlikely to be the case in their study where differences in gut lengths are apparent. A simple way around this is to quantify ingested dye colorimetrically following ingestion of a short pulse of dye-laced food (see, for example, PMID: 33116314).

Suggestions/specific points

1. I realise that elucidating how exactly sugar antagonises the effect of yeast is beyond the scope of this particular manuscript but, alongside the various sugars tested in Figure 2D, it would have been very informative to test the effects of a palatable, but not nutritious, sugar (e.g. arabinose) vs a nutritious, but not palatable sugar (e.g. sorbitol). This would help distinguish between sensory vs metabolic mechanisms.

2. The manuscript clearly shows that the two diets differentially impact ISC proliferation, EC loss and the gut's transcriptional profiles. How are these three features related? Its seems to me that one intriguing possibility is that young and old ECs differ in their transcriptional identity and/or ability to handle different diets, and different diets result in different fractions of young vs old EC. EC physiology/transcriptional profile may also be related to EC size/ploidy. If the authors have a cellular readout for one of the genes transcription of which is diet dependent, a simple co-staining of the HisRFP or esgFO experiments may help resolve this, but otherwise I suggest that the authors at least discuss this idea, possibly at the expense of the Discussion section "Is midgut homeostasis only an apparent property?" section. I thought this was a weak discussion point because (1) most experiments in the manuscript were conducted during the postnatal gut growth period (ie non-homeostatic) and (2) both the HS and HY diets are "imposed", not particularly ecological, diets. Hence, we do not really know whether the size of the gut would be stable in truly wild-type flies with food choices.

3. The fact that mated females were used for all experiments is only mentioned in the Methods. It would make more sense to state this in the manuscript, and justify the rationale for this choice; the dietary plasticity of the midgut may well differ between males, virgin and mated females in light of other data including a recent study by the authors (PMID: 26216039, 32641829, https://doi.org/10.1073/pnas.2018112118).

4. Both the main text and figures could do with stating more clearly whether the data was acquired during the gut growth period post-eclosion and/or the "homeostatic" period that follows. I realise that the information is there but it is sometimes hard to find.

5. Line 1336-7: what do the authors mean by "midgut re-sizing is allometric"? Allometric relative to what? The statement in the main text make sense to me (e.g. diet changes the allometry of different gut regions), but it is less clear to me what this sentence means.

6. What vitamins/minerals were provided in the Harlan mix? I was unable to find the composition online based on the catalogue number provided in the Methods.

Reviewer #3:

The authors use length of gut as measurement to asses growth in response to diet composition. They also measure a large panel of fly strains. They assess the ration of sugar to yeast, and conclude that sugar has a strong antagonistic effect on growth. They also test by examining genome wide association data, and transcriptome analysis, and examine translational control, TOR pathway components.

There are also some concerns about how to define low growth versus antagonistic effects. There is clearly different growth rate with the sugar vs yeast diet, but to suggest that high sugar has antagonistic effect is not sufficiently convincing, because it is equally good to describe it as low growth rate, with lower nutrient say by dilution as one would expect. Nonetheless, the idea that sugar (sucrose, glucose etc), not just carbohydrates or polysaccharides, actually has an antagonistic effect is an attractive direction. One possible improvement will be to investigate further into the "antagonistic" effect of sucrose, first to convincingly show that it is not just a low growth condition, then to investigate possible cellular or molecular targets that antagonize the growth. The results from transcriptome data, and the involvement of TOR and translational control in cell size is not unexpected as these pathways have been shown to be involved in cell growth in ISCs and EBs before, and their function in ECs is therefore a confirmation of previous work.

Large amount of data and analyses are provided, but everything seems to contribute to something, making the conclusions ambiguous and no less confusing than the beginning. Substantial reorganization and rewriting may simplify the presentation and focus on their main idea for more convincing arguments.

line 147 – The CS flies results first mentioned should be directly integrated into other flies, because the CS results (31% longer, 44% wider) seem to be to the extreme ends when compare to all 188 lines examined (median 1.18). Perhaps better to shorten this section with focus on the population study, which is much more valuable than individual strains.

Line 168, stated that 184/188 had larger midguts, but then followed by 126 lines statistically significant. They should reconcile these statements.

Concerning HY or HS diet, the authors should give a better description of these compositions at the beginning of result section, and the justification. The exact composition is quite important as these experiments occupy a large portion of the initial experiments. The detail confuses me: P. 85, supp Experimental conditions: Table 1, sucrose for preexperiment is 40 g/L, High sugar diet is 168.1, low sugar diet is 74.1. Why does this low sugar diet still contain so high of sucrose, what is the reasoning behind this choice? Also, these tables in the supp are poorly formatted and many small prints are missing/shifted in the pdf and cannot be followed easily.

Yeast should also contain plenty of sucrose and polysaccharides, or precursors of sucrose. Has this been accounted for in their experiments comparing the amount of sucrose/yeast ratio?

Line 425, the use of Rasv12 to stimulate ISC over sugar diet perhaps is not fair. While there is some increase of ligands in their RNAseq data, there may not be high level enough to produce proliferative effect. Have they shown that there is activation of these pathways/target genes in ISCs under high sugar? At the end, it is probably better to use overexpressed ligands rather than Rasv12 to do this experiment.

Line 450, the lack of increased gut length after knockdown of translational inhibitor in sugar diet requires further investigation. It again suggested a delicate balance can affect the outcome, because knockdown of these translational inhibitors affects other cell or tissue parameters.

The combination of sucrose and yeast is widely used in flies, but isn't trehalose is the main balance of source of energy in flies and synthesized from glucose, and how that is calculated into their "isocaloric" argument of the diet used for the comparison?

https://doi.org/10.7554/eLife.64125.sa1

Author response

Essential revisions:

(1) Tor/autophagy experiments. Given that the myo1A driver expresses in both enterocytes and neurons, it is important to repeat at least one key Tor/autophagy experiment using a different enterocyte driver to confirm that the observed effects are due specifically to genetic manipulation of enterocytes. In addition, inclusion of Tor gain-of-function data or a readout of endogenous Tor activation in HY vs HS conditions would strengthen the authors' conclusions, if such data can be readily obtained; in light of eLife's resubmission policies during COVID, however, these gain-of-function experiments are not essential.

To address the first part of this revision point, we have now knocked down Tor and manipulated autophagy using another enterocyte driver, the hormone-inducible driver 5966GS, to express RNAi against Tor and Atg8a. Both experiments recapitulate our previous findings with the MyoTS driver, confirming enterocytes as the relevant cell population (Figure 7—figure supplement 2D, E and main text lines 582-583 and 605-608, figure legend lines 1960-1968, material and methods lines 859-867).

Additionally, we have added quantitative data of the clone-tracing system (hsFlp; Act>STOP>Gal4,UAS-GFP) experiment to manipulate the TOR pathway and autophagy. We compared GFP+ clones (expressing RNAi or overexpression constructs) to regular GFP- cells in the same midgut. The results corroborate our initial images. We have added these quantifications to Figure 7D for Tor-IR and Rheb-OE and Figure 7J for Atg2-IR. We have also replaced the original images to better reflect the quantification (Figure 7B-C for Tor-IR and Rheb-OE and Figure 7I for Atg2-IR). We also made changes in figure legend at lines 1671 and 1679 – 1680 and in material and methods at lines 1110 to 1113.

We have also conducted new experiments to test the physiological regulation of TOR pathway by our experimental diets, by quantifying in vivo levels of a 4ebp reporter (Thor-lacZ transcription reports Foxo activity and is inversely correlated to TOR activity). These results align with the signal from our prior RNAseq experiment, corroborating the response of TOR to experimental diets. Figure 7—figure supplement 2A-C. Changes in text can be found at lines 565-567 (main text), and lines 1959 – 1960 (sup. figure legend). We also changed TOR/autophagy is a master regulator to “an important regulator” (line 607-608).

Overall, these results confirm our previous findings and strengthen the conclusion that TOR pathway activity in enterocytes partly mediates midgut resizing by diet.

(2) Homeostasis. Two Reviewers had concerns that the replacement ratio experiments, which are the basis of the authors' conclusions regarding homeostasis, were performed under physiological conditions that have been shown not to be homeostatic. In addition, one Reviewer had questions regarding the robustness of cell labeling and the formulas used to calculate new and old cells. Please address these comments in the manuscript text and/or response to reviews. Additional controls for the cell labelling, if they are available, would help to validate the assay (but are not essential).

We have tackled these concerns both through new experiments and changes in the text.

2a: concern regarding homeostatic conditions

To address the first concern that the experiment was conducted in non-homeostatic conditions, we have performed the cell loss assay in two new conditions:

– First, we have repeated the experiment shown in Figure 4G to L, but dissected flies at an earlier timepoint (5 days after shift, for a total age of 15 days post eclosion, instead of 24 days as the experiment presented initially in the first submission of manuscript). We have added this result in Figure 4—figure supplement 2C and in main text at lines 373-377. We have also included here (Author response image 1) these new data shown in the same format as Figure 4L (showing replacement ratios and rates of cell loss and gain). For this new set of data, we had to perform corrections to the chart, since there was no detectable cell loss, and in some cases we had slightly more Histone RFP+ cells counted at 5 days after induction, as expected from a noisy biological system. For panel 1 illustration, we corrected the data to show 0 cell loss. As shown in Author response image 1, even at 5 days post chase, the ratios are reminding of what we observe at the later timepoint for the various dietary conditions.

Author response image 1
Data shown as rate relative to experiment start (cell /initial EC/ day) for Figure 4-figure supplement 2C (5 days post chase start, 15 days from eclosion).</Author response image 1 title/legend>.

In order to strengthen our message, we have modified the main text at lines 406-409 and the figure legend at lines 1849-1852 to reflect the addition of this new panel in the manuscript.

Second, we have utilized the approach suggested by reviewer #1, to perform a cell loss assay in younger adults. In order to do so, we moved flies at eclosion for 1 day on either HS or HY diet. After this one day, we moved the flies on food with added RU486 for 3 days (until 4 days post eclosion). We then dissected flies at this first time point and removed flies from the hormone to start the chase. We then dissected flies 5 days post chase start, at 9 days post eclosion. Again, we find an increase in the size of the gut in this time frame for HY to HY, coinciding with an increase in total cell number, while HS to HS flies stayed at around the same size and experienced limited cell proliferation, as inferred by the number of blue cells. We have not included this data in the manuscript, as this dataset has a different experimental design, but please find the results here in Author response image 2.

Author response image 2
Number of ECs in the posterior midgut, both marked (Red, old ECs) and unmarked (Blue, new ECs) by RFP, error bars are SE from 3 repeats, for the experiment described above.

Additionally, we have investigated whether, in our experimental conditions, the gut is in a state of dysplasia. To do so, we have surveyed the state of the midgut with a EsgTS>UAS-RFP at the timepoints presented in the original manuscript for beginning and end of chase. Flies were kept at room temperature (as the histone RFP experiments) and shifted at 29°C 3 days before dissection to activate the system. Overall, we do not see accumulation of progenitors, or progenitors with apparent abnormal shapes in any of these conditions (Author response image 3) . We therefore conclude that we could not find dysplasia apparent at these timepoints.

Author response image 3
EsgTS>UAS-RFP at the start of the chase (Day0) vs at the end of it (Day14) for region 4 of the midgut.

Scale bars are 50 µm.

Overall, while we agree with reviewer #1 that the replacement ratio is not likely to be constant throughout the lifespan of the flies, as also indicated by our proliferative data in Figure 4B and previous publications, and that it is likely than on certain diets and at certain age there may be a 1:1 replacement ratio leading to homeostasis, as shown for example in Liang 2017 (10.1038/nature23678), on the diets in our manuscript we observe a non 1:1 replacement ratio also at earlier time points, suggesting that our results are not a consequence of accelerated renewal due to dysplasia.

We would also like to answer here a comment from reviewer #2, that the “Is midgut homeostasis only an apparent property?” section may be weak because most experiments were conducted during the post-natal growth period. While many experiments in the manuscript are indeed performed in the initial post-natal growth period, most of the experiments to survey cell gain and cell loss are made at later time points (and now also at earlier timepoints). In addition, we show that the increase in size or shrinking occur plastically, both early on and later in life. We therefore believe that we capture cell dynamics over a broad age range and cover the range usually considered as perfectly homeostatic. We agree that it may confusing to keep track of the time at which each experiment was performed only using the information in material and methods, as remarked also by reviewer #1, so we have added in the figure legends the age of the flies for each experiment.

Additionally, while the focus of the comments has been on the HY diet, since it is the one more resembling previously used “standard” conditions, the HS diet also does not show a 1:1 replacement condition, suggesting that indeed there is not a constant coupling between cell loss and gain, but instead a flexible connection between ISC proliferation and cell loss as a function of diet, contributing to midgut size. Regarding this finding, reviewer #2 pointed out that both HS and HY diets are “imposed” and not particularly ecological diets, and hence we do not know whether the size of the gut would be stable in wild-type flies with food choices. We agree with the reviewer that the experiments in this manuscript are not relevant to an ecological context, but they are designed to study the effect of specific diets and ratio of nutrients. We have added in the discussion a sentence to highlight this point at lines 731-736 of the discussion.

2b: concern regarding the robustness of our cell loss assay

As our cell chasing system is indeed an important component of the manuscript, we have also added details to demonstrate the robustness of the 5966GS mediated cell labelling system. First, we would like to indicate that the same His-2BRFP that we used in our manuscript, has been previously used in at least two publications in the Drosophila midgut: PMID 26077448, where it is indicated that it persists for at least 28 days in the intestine (albeit under a “data not shown” label) and PMID 33135280. Additionally, we decided to also test the robustness of the system directly. To this purpose, we utilized ActGS>UAS- His-2BRFP to drive expression in all tissues in Drosophila, so that we could assay the dye stability in tissues that do not undergo turnover like the midgut. In such tissues, the lack of His-2BRFP signal after a chase (the cell would be DAPI+ RFP-) would not be due to the fact that new cells appear, but because the RFP is lost from these cells. We selected the crop and the hindgut for their proximity to the midgut and their lack of turnover in basal conditions. We utilized the same experimental settings as in Figure 4K. We observed high stability of the His-2BRFP in both these tissues, and in a similar manner on both diets. We have included this piece of data in Figure 4—figure supplement 2B. We have added a related description in the main text at lines 363-371, in the figure legend at lines 1840-1846 and in mat and met at lines 820-823.

Related to cell labelling, we agree with reviewers that not all cells are initially marked by His-RFP, in Figure 4K. This is in our hands the case of all enterocyte drivers: they drive in most but not all enterocytes. Importantly, in Figure 4K it is possible to see under the label HS and HY the number of cells that were initially marked (in red) or not marked (in blue). For guts on HS diet we counted an avg of 152 unmarked cells over a total of 1242 cells (12.2% of cells, so 87.8% of marked cells), meaning that the labelling coverage was high. For guts on HY diet we counted an avg of 258 cells unmarked over 1855 cells (13.9% of cells, so 86.1% of marked cells). As a corollary experiment to the new cell loss assay included for this revision, we also kept some flies for 5 days instead of 3 on RU486. We have obtained very similar percentage of cell marking, as it is shown in Author response image 4, suggesting that our experimental conditions already maximize the number of ECs labelled.

Author response image 4
Percentages of RFP+ cells (RFP) over total number of cells (Dapi) at the start of a pulse chase experiment.

2c: concern about the calculation of replacement ratios

Regarding the formula to calculate cell loss, we would like to thank reviewer 1 for their comment, as we noticed that we had reported the formula incorrectly in the mat and met. We corrected the material and method section to reflect our calculations at lines 1069-1080. Importantly, we calculate the number of cells lost from the epithelium using solely RFP+ ECs and assume that this ratio of cell loss is similar also for the few unmarked cells.

Regarding the concern that a gut with a replacement ratio of 1.46 would double in size in 15 days, we think that while the gut changes how much it is turning over (as shown also in Figure 4B for proliferation), this does not (always) translate in a change in gut size (as visible in Figure 4B [pH3] and related 4S1C [length]). This result is in agreement with our finding in Figure 6 and 7 that the proliferative state of the gut is not the only, or even the central determinant of the final size of the organ. Overall, the total amount of cells for the posterior region changes of around 1000 cells (Figure 4K), but the size of these cells is likely to be slightly smaller (or with a different organization), leading to a comparable final size, which is nevertheless still bigger than the starting midguts. We have added a related comment at lines 406-409 to clarify.

Overall, we believe that the results from these experiments provide enough information to convince of the functionality of our EC loss assay, and therefore of our conclusions.

(3) Diet manipulations. Please address the following reviewer comments, either in the text/response to reviews or by providing additional data. (a) Effect of gut capacity was not taken into account for excreta-based measurements of food ingestion. (b) Effects of diet on cell-type proportions in this study appear to differ compared to Obniski Dev Cell 2018. (c) Potential alternative interpretations of "antagonistic" effect of sugar on protein.

3a: concern about excreta based measurements

Regarding the concern about the excreta-based measurements, reviewer #2 is worried that the different sizes of guts on different diets would influence the amount of excreta due to different capacities. We believe that the design of our experiment (that lasts over multiple days) is such that we measure transit at a scale where the volume of one gut is neglectable. Precisely, we have taken our measurements over a 24-hour period, and it was previously published that the transit for the food is shorter, with most food transiting the gut in the first 2-3 hours from ingestion (https://doi.org/10.1128/mBio.01453-17). Additionally, we have taken our measurements by using the same flies consecutively for 5 days, taking a measurement every 24 hours (for a total of 3 repeats, and 5 timepoint for each repeat). This means that at each day but the first, the gut was already filled with blue food from the previous day. Since the measurements at each timepoint were extremely close to each other, we consolidated them in the manuscript. In order to convince the reviewer of this robustness, we present, in Author response image 5, an example of the time course for one diet belonging to the geometrical framework. We do see for some diets a slightly lower input in the initial day, in agreement with a possible role of gut content size. However, it becomes neglectable at day 2 and is not a concern as our assay lasts for 5 days total. We made changes in the material and methods to better describe the method used as lines 941-944.

Author response image 5
Feeding assay showing data per each day instead of consolidated for one of the diets in the nutritional geometry experiment.

3b: concern about lipids and cell composition

Regarding the comment on differences in cell type proportion (reviewer #2), we do not think that our data is in contrast with the Obniski Dev Cell. While it is true that the amount of lipids is different between the HS and HY diets, there is a critical difference. While in the Obniski Dev Cell the amount of lipids is the only difference between the 2 diets used (addition or not of lipids), this is not the only change found between the HS and HY diets: we do not know about the effects of changes in lipid amount, on top of also varying amount of sugar and yeast. Additionally, in our own experiment, the composition of lipids between diets was the same, as lipids were provided by either high or low amount of yeast, while in the Obniski Dev Cell a cocktail of different lipids was added. Overall, we think we cannot compare these diets, and thus their effect on the ratio between EEs and ECs.

3c: concern about the interaction between sugar and yeast

We agree with reviewer# 3 that the interpretation of the interactions between sugar and yeast are difficult to untangle, but we believe our manuscript makes the point for an actual antagonism. We have several arguments that we believe convergently suggest an antagonistic role of sugar on yeast triggered growth, and we believe they demonstrate that the role of sugar goes beyond simply affecting the amount of yeast indirectly:

– First, our nutritional geometry approach in Figure 2 allows to map the contribution of sugar quantity, yeast quantity, and their relative amounts. Using this approach, we found that the more sugar was ingested, the smaller the gut would be, across different yeast to sugar ratios, and across different caloric contents. This means that for a given quantity of yeast ingested, the more sugar is ingested, the shorter the gut will be. This is evidence that the amount of yeast is not the only driver of gut size.

– Additionally, the experiment performed in Figure 2F shows that a low amount of yeast in the diet is enough to provide for gut growth if the rest of the calories is not provided by sugar, but by lipids. This demonstrates that sugar opposes the growth of the gut directly. Considering that yeast is indeed the main driver of gut growth, we believe this clearly shows the antagonistic role of sugar toward yeast induced midgut growth.

– Additionally, our transcriptomic study shows that the high sugar diet elicits a stress response in the gut. Regardless of the specific consequences of this stress, we think this “toxic effect of sugar” is clearly different from the effect of yeast on host transcriptome.

– Our mechanistic insights demonstrate that sugar directly uncouples ISC from their niche, thus decreasing cell proliferation, while yeast promotes proliferation. We have now expanded this conclusion to demonstrate that sugar probably affects the translation of proliferative signals in the niche, thus altering ISC-niche coupling. Specifically, to further investigate mechanisms through which HS diet blocks ISC proliferation via translation, we investigated in which cell type GCN2 mediated translational stress is important. We performed knock-down of Gcn2 via RNAi in progenitor cells and in ECs respectively. We found that it was possible to increase proliferation only when knocking Gcn2 in ECs, but not in progenitors (Figure 5—figure supplement 2G), suggesting that sugar induces stress in ECs, thus decreasing their ability to send pro-proliferative signals. This result also reinforces the idea that sugar does directly uncouple pro-proliferative signals and proliferation is in the niche. Changes in main text can be found at lines 498-502 and at lines 1889-1892 in the figure legend.

(4) Some improvements for clarity and context. I invite the authors to consider the Reviewers' comments with regard to (a) clarifying the key takeaways for a broad audience; (b) including crucial experimental details, such as animal age and mating status, in the Results (not just the Methods); (c) improving the supplemental tables that contain the diet recipes.

We have implemented suggested changes from reviewers, both by adding new results and through changes in text/organization. In this section of the response, we would also like to tackle comments that were not directly related to any of the other 3 main points but implemented in our revision as we believe they improve the manuscript.

4a Comments on proliferation and pH3

– Regarding the comments from reviewer #1 about the limitations of our study of proliferation and uncoupling. We agree with the reviewer that pH3 only reveals one facet of proliferation. In the descriptive part of our work, we demonstrate in detail the differences in proliferation between the HS and HY diet, using pH3, flip out tracing, and now our new cell gain and loss assay (5966GS His-RFP). So, we believe we make a strong case that all these indicators reflect a lack of proliferation on the HS diet. Nevertheless, we agree that when focusing on uncoupling between ISCs and their niche, we work with pH3 counts as a main phenotype, as we have established that this varies with diet, and as this is independent of Gal4 manipulation. Sadly, additional ways to monitor proliferation would be very complicated to combine with Gal4 manipulation of the translation machinery. Nevertheless, we have modified the text at lines 784-788 to highlight the possibilities suggested by the reviewer. In addition, we now have additional experiments that help clarify the role of progenitors in the response to diet:

– First, we demonstrate that transgene-based expression of pro-mitotic signals in ECs is sensitive to diet. In HS, overexpression of either upd3 or spitz leads to less proliferation on HS than on HY (Figure 5E). In addition, when translation blockage is alleviated from ECs, but not progenitors, proliferation is rescued as measured by pH3 (Figure 5—figure supplement 5G). Altogether, these data suggest that the “uncoupling” we propose occurs really at the level of the ECs. We now clarify this in the text and agree with the interpretation of the reviewer about the uncoupling. We do think that what the reviewer described is the uncoupling, which acts through a problem in the ability of ECs to perform translation. We indeed think that these are the mechanisms through which uncoupling may happen between the niche and stem cells, hence the focus on translation in Figure 5. We have clarified this concept at lines 455-457.

– We have also reinforced our conclusion that over the period of time we focus on, progenitors are dispensable for gut growth. Ablation of progenitors using the pro apoptotic gene reaper (Figure 6—figure supplement B-F) shows that the gut is still able to grow in absence of progenitors (see later for more detailed comment on this experiment). However, we agree with the reviewer that enteroblasts represent a potential “node” for organ size control. Possibly, the ability for EC size to compensate for the lack of proliferation is not infinite and progenitors would reveal a more central role in longer studies. Since we did not include any experimental result about this point, we had left it out. We have now included it in the discussion at lines 766-769.

4b Comments on our diets in comparison to other diets

Reviewer #1 asked how our diets compared to standard diets used in the field of Drosophila. We have surveyed the effect of “standard” fly food diets on the length of the midgut. We have not included these diets directly in the nutritional geometry, as they also contain cornmeal and therefore would add additional complexity, but we have included this result in Figure1—figure supplement 1B and lines 152-154 in main text, and lines 1698-1699 in figure legend. This experiment showed that our HY diet has similar size to both Bloomington diets.

4c Diverse comments on minor points and confusing language and descriptions

– Regarding the comment on antagonism not being the best descriptor of the relationship between yeast and sugar: we changed antagonism to opposite throughout the manuscript.

– Regarding the comment that Lipid HS is a confusing name, we agree with the reviewer. Accordingly, to precise the diet composition and clarify our message, we changed Lipid HS to Yeast:Lipid 1:14 and Lipid HY to Yeast: Lipids 1:0.7. We did also change HS to Yeast: Sugar 1:14 (HS) and HY to Yeast:Sugar 1:0.7 (HY), and Lipids only to Yeast:Lipid 0:1. We hope this will make it more clear to understand which diet is what.

– As requested, we moved Figure2—figure supplement 1 D-F to the main figure 2 (now B, C, D) and re-arranged the figure and figure legend accordingly.

– As requested, we moved Figure3—figure supplement 1 A to the Figure 3 (now D) and re-arranged the figure and figure legend accordingly.

– As requested, we removed “continuously” from line 317. Guts do become significantly bigger at day 28 on HY and smaller on HS compared to the start of the experiment, so we kept the rest of the description.

– Regarding the comment asking about the practice (and rationale) of drawing lines on a PCA plot (Figure 5A), the lines describe the trajectory from eclosion of each diet. For each diet, they connect Day 0 to 1, then 1 to 2, 3 to 5. We briefly updated the figure legend at lines 1591 – 1593 to make this clearer. We also added the number of replicates (3) for the RNA-seq in the figure legend, as requested. For the part of the comment asking for single datapoints: for this visualization we prefer a single point for each sample, as we are not trying to show how much the samples from different repeats are close to each other as a mean of “quality control”, but to give a biological message on the changes and differences between eclosion and the 2 different diets. Additionally, a crossbar is present on the plot representing the standard error estimated from the three replicates.

– Regarding the comment on RasV12 overexpression (5D) having similar levels of proliferation as Canton-S in Figure 4B, we have included in the panel (5D) the control and an additional mean of eliciting proliferation, a UAS-Tor-DER construct. It is possible to see that the base level of pH3 stain in these flies is lower than in Canton-S, and that over-expression of both constructs with EsgTS results in a strong increase in pH3 cells. We made few changes in the main text at lines 460-463 to implement this change. We also precised the protocol used for the induction of pH3 with this method at line 853-857.

– Regarding the comment asking how midgut width was measured: we did measure width in the wider part of the anterior, middle and posterior midgut, as the reviewer mentioned similar to the 3 yellow lines perpendicular to the gut in Figure 1 C, D. We used the sum of these 3 widths for the plot. We updated description of midgut width calculation in Figure 1—figure supplement 1A at lines 1695-1698.

– We added a note about the plasticity experiment in O’Brien 2011 at lines 330-331 as requested.

– Regarding the comment asking for the differences between 5L, M vs figure supplement 5E, F. The experiment in 5L, M shows the data for guts kept on HS diet for 1 week after Gal4 activation. The data in Figure5—figure supplement 2E shows that data for guts that were first kept for 1 week on HY diet and the shifted-on HS diet. We have corrected an error in the palette in Figure5—figure supplement 2E, which was the one used for HS instead of HY to HS.

– Regarding the comment for the figure legend of Figure 6, that in the experiment where we drive EGFR-IR, depleted was too strong of a word, we changed description of effect of Egfr on ISCs in the figure legend at lines 1630-1632 to be more in phase with what was written in the results. We also performed an additional experiment, found in figure 6 —figure supplement 1 B-F, where we over-express reaper in progenitor cells with EsgTS. Over-expression of reaper result in loss of marked progenitor cells after 7 days at 29C (12 days from eclosion) in region 4 of the midgut (flies were on HS diet). Upon shifting these flies on HY, we observe a change in the morphology of the midgut similar to EsgTS>Egfr-IR (increased space between nuclei and lack of small cells). We also observe a change in the size of the midgut similar to the control, as observed with EsgTS>Egfr-IR, reinforcing our conclusion on the ability of the gut to change its size over a 7 day time period without the need for stem cells. Changes in main text can be found at lines 529-535 and 545-547, and at lines 1920-1931 in the figure legend.

– We added n numbers in Figure 6A and 7A as requested. Additionally, regarding the comment on why 7/14 sample being statistically correlated is a “generally positive correlation”, the generally positive correlation is more visible for HY (added a note for this in the figure legend), since HS diet do not vary in size as much and it is harder to observe any phenotypes there. This effect is also enhanced when comparing this result with the results in Figure 6A, which show a marked lack of correlation.

– We double checked all the references in the main text to figures to make sure they properly coincide, thanks for noticing. We also double checked the bibliography for mistakes.

– Regarding the comment on the title not doing justice to the manuscript, we renamed it to: Multiscale analysis reveals that diet-dependent gut plasticity emerges from alterations in both stem cell niche coupling and enterocyte size.

– Regarding the comment from reviewer#2, that it would be interesting to elucidate the role of sugar in antagonizing the effect of yeast, in particular to distinguish sensory vs metabolic mechanisms, we do agree and thank the reviewer for their comment. We followed the reviewer recommendations, and we have surveyed the effects of a palatable, but not nutritional sugar (arabinose) and of a non-palatable, but nutritious sugar (sorbitol). We found that using sorbitol, we have a similar response as sucrose, indicating that taste per se it is not required for the effect of sugar on the midgut. On the other hand, arabinose was lethal to the flies in just 3 days on the HS diet and reduced the overall size of the midgut on HY after 5 days. Considering the effect on HY diet, it is possible that arabinose is having some sort of stressful effects on the fly midgut or on organismal metabolism, thus leading to a shortened gut and a quick demise of the flies, rather than just being the lack of nutrition. We have added these data to Figure 2—figure supplement 2C, in the main text at lines 267-275 and in the figure legend at lines 1770-1774.

– Regarding the comment from reviewer#2 enquiring about possible changes in ploidy, we have now surveyed the impact of our HS and HY diet on cellular ploidy via FACS. Overall, we do not see an effect on ploidy due to HS and HY diet. We have added this result in Figure 3—figure supplement 1D-H, in the main text at lines 302-307, in material and method at lines 1028-1052, and in figure legend at lines 1803-1807.

– Regarding the comment of reviewer #2 on the possibility that young and old ECs differ in their transcriptional identity and capacity to handle diets/ challenges, we agree that it is indeed intriguing. Indeed, we have already two members of our lab working on this question! However, we believe this is beyond the scope of this publication.

– Regarding the comments on the importance of the sex/mating status of the flies on midgut growth, as also recently shown by us in a recent PNAS study, we agree with the reviewer that this is an important point to check. We have now added a panel in Figure 1—figure supplement 1 (new C) where we investigate the impact of HS and HY diets on also on un-mated females and males. Indeed, we found that mated females are the most responsive to diet compared to the other two conditions. We have made changes to the main text at lines 154-159, and in the sup Figure legend at lines 1699-1702.

– Regarding the question from reviewer#2 on “what do the authors mean by “midgut re-sizing is allometric?”, we changed the “midgut re-sizing is allometric” to midgut re-sizing is allometric between regions to clarify this point at line 1501.

– About the request for the composition of the vitamins/mineral mix, it is possible to find the composition in PMID: 23844001, File S1: Summary of food recipes used in this study (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3699577/bin/pone.0067308.s008.xlsx), CDF 100-500K Recipe tab.

– Regarding the comment from reviewer#3, that the part on Canton-S flies in Figure 1 should be shortened to focus on the DGRP, we would prefer to keep as it is, since we use Canton-S for many experiments in the manuscript, and we would prefer to have a clear representation of the phenotype in this fly line.

– Regarding the comment from reviewer#3 that “Line 168, stated that 184/188 had larger midguts, but then followed by 126 lines statistically significant. They should reconcile these statements”, we made changes when describing the DGRP to better convey our message. The first statement at lines 177-178 reflects that midguts on HY were quantitatively larger than midguts of flies on HS (ratio HY/HS was higher than 1 in 184/188 cases). The second statement at lines 179-181 is instead checking if the midguts were statistically different between HS and HY. We have also changed our statistical analysis to and because we noted an error in our tabulation of lines showing a statistically significant response to diet, which was previously read from a model summary (in error). We have remedied this error using post-hoc tests, see also changes at lines 958-961.

– Regarding the comment from reviewer#3 on “the rationale of the choice of HS and HY”, we used these diets based on a previous study (PMID: 25520356, now added to the manuscript ref.) for their diverging sizes despite being isocaloric.

– Regarding the comment from reviewer#3 on why the HY diet contains so much sugar, the idea was to study the effects of varying ratios of nutrients in an isocaloric context, not to have a sugar free or sugar rich comparison. In this context, the HY diet is not supposed to just be a low sugar diet, it is an isocaloric diet with a different ratio of Yeast:Sugar. We have also updated the results with this information at lines 144-146.

– Regarding the comment from reviewer#3 on the supplemental table being poorly formatted, we thank the reviewer. Formatting was lost when uploading the table as a csv file. We have now moved the tables from the uploaded csv file to a word document, for better formatting (now Supplementary File 1).

– Regarding the comment from reviewer#3 “if the amount of nutrients has been accounted when comparing”, this has been taken in account when calculating the amount of nutrients (see Figure 1B), but for nutritional geometry the ratios are based on grams of yeast and sugar used to cook the diets.

– Regarding the comment from reviewer#3 on RasV12 to stimulate ISC being “not fair”, we want to clarify that we used this construct to assay the proliferation capabilities of ISCs on HS condition, we were not trying to imply that RASv12 was the factor responsible for proliferation. Additionally, we checked the ability of stem cells to proliferate when expressing UAS-upd3-OE and UAS-spi-SEC from enterocytes via MyoTS, as suggested by the reviewer. We find that while both constructs can induce proliferation on HY, they are not able to do so on HS, suggesting again a role for the niche in the uncoupling between pro-growth signals and stem cells (Figure 5E). We update description in main text at lines 465-474, in figure legend at lines 1605-1607 and in Materials and methods at lines 857-859.

– Regarding the comment from reviewer#3 that trehalose is the main circulating sugar, and if this has been taken in account when calculating the isocaloric status of the diets: trehalose is indeed one of 2 circulating sugars in Drosophila, but the type of circulating sugar does not change the amount of calories ingested by the flies by eating sucrose in the HS and HY diets, even if nutrients are then converted to different components during digestion and metabolic processes, so we do not think this is a problem for the calculation of the isocaloric ratio of the diets.

– We have also added statistical brackets where applicable to figures, to ease recognition of what samples are being compared.

https://doi.org/10.7554/eLife.64125.sa2

Article and author information

Author details

  1. Alessandro Bonfini

    Cornell Institute of Host-Microbe Interactions and Disease, Department of Entomology, Cornell University, Ithaca, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing - original draft, Writing - review and editing, conceived the study, designed, and performed most of the experiments, contributed to dissections for DGRP, performed RNA-seq analysis, performed image analysis and organs rendering for NanoCT scan, wrote original manuscript, contributed to dissections for DGRP, contributed to dissections for DGRP, contributed to dissections for DGRP
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6642-8665
  2. Adam J Dobson

    Institute of Molecular, Cell and Systems Biology, University of Glasgow, Glasgow, United Kingdom
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing - original draft, Writing - review and editing, conceived the study, contributed to dissections for DGRP, performed GWAS analysis, performed RNA-seq analysis, performed experiments for NanoCT scan and image acquisition, contributed to dissections for DGRP, contributed to dissections for DGRP, contributed to dissections for DGRP
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1541-927X
  3. David Duneau

    1. Université Toulouse 3 Paul Sabatier, CNRS, UMR5174 EDB (Laboratoire Évolution & Diversité Biologique), Toulouse, France
    2. Instituto Gulbenkian de Ciência, Oeiras, Portugal
    Contribution
    Data curation, Formal analysis, Validation, Writing - review and editing, performed GWAS analysis, performed statistical analyses and R-code improvements
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8323-1511
  4. Jonathan Revah

    Cornell Institute of Host-Microbe Interactions and Disease, Department of Entomology, Cornell University, Ithaca, United States
    Contribution
    Investigation, Writing - review and editing, contributed to dissections for DGRP, contributed to dissections for DGRP, contributed to dissections for DGRP
    Competing interests
    No competing interests declared
  5. Xi Liu

    Cornell Institute of Host-Microbe Interactions and Disease, Department of Entomology, Cornell University, Ithaca, United States
    Contribution
    Investigation, Writing - review and editing, contributed to dissections for DGRP, contributed to dissections for DGRP, contributed to dissections for DGRP
    Competing interests
    No competing interests declared
  6. Philip Houtz

    Cornell Institute of Host-Microbe Interactions and Disease, Department of Entomology, Cornell University, Ithaca, United States
    Contribution
    Investigation, Writing - review and editing, contributed to dissections for DGRP, contributed to dissections for DGRP, contributed to dissections for DGRP
    Competing interests
    No competing interests declared
  7. Nicolas Buchon

    Cornell Institute of Host-Microbe Interactions and Disease, Department of Entomology, Cornell University, Ithaca, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing - original draft, Writing - review and editing, conceived the study, performed RNA-seq analysis, contributed to dissections for DGRP, wrote original manuscript, supervised the project and funding., contributed to dissections for DGRP, contributed to dissections for DGRP, contributed to dissections for DGRP
    For correspondence
    nicolas.buchon@cornell.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3636-8387

Funding

National Institutes of Health (1R21AG065733-01 1R01AI148541-01A1)

  • Alessandro Bonfini
  • Jonathan Revah
  • Xi Liu
  • Philip Houtz
  • Nicolas Buchon

National Science Foundation (IOS-1656118 IOS-1653021)

  • Alessandro Bonfini
  • Jonathan Revah
  • Xi Liu
  • Philip Houtz
  • Nicolas Buchon

UK Research and Innovation (MR/S033939/1)

  • Adam J Dobson

Agence Nationale de la Recherche (ANR-10-LABX-41 ANR-11-IDEX-0002-02)

  • David Duneau

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank M Furuse for reagents. We thank Rachel L Fay and Yunan Nie for help with project-related experiments not included in the manuscript. We thank the Cornell Institute of Biotechnology imaging (NanoCT), genomic facilities (RNA-seq), and FACS core for help in setting up protocols and general assistance. We thank the Lazzaro lab for access to equipment and reagents.

Senior Editor

  1. Utpal Banerjee, University of California, Los Angeles, United States

Reviewing Editor

  1. Lucy Erin O'Brien, Stanford University School of Medicine, United States

Reviewer

  1. Lucy Erin O'Brien, Stanford University School of Medicine, United States

Publication history

  1. Received: December 16, 2020
  2. Accepted: September 22, 2021
  3. Accepted Manuscript published: September 23, 2021 (version 1)
  4. Version of Record published: October 19, 2021 (version 2)

Copyright

© 2021, Bonfini et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,153
    Page views
  • 228
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Developmental Biology
    Yanyan Ding et al.
    Research Article

    Although thrombosis has been extensively studied using various animal models, our understanding of the underlying mechanism remains elusive. Here, using zebrafish model, we demonstrated that smarca5-deficient red blood cells (RBCs) formed blood clots in the caudal vein plexus. We further used the anti-thrombosis drugs to treat smarca5zko1049a embryos and found that a thrombin inhibitor, argatroban, partially prevented blood clot formation in smarca5zko1049a. To explore the regulatory mechanism of smarca5 in RBC homeostasis, we profiled the chromatin accessibility landscape and transcriptome features in RBCs from smarca5zko1049a and their siblings and found that both the chromatin accessibility at the keap1a promoter and expression of keap1a were decreased. Keap1 is a suppressor protein of Nrf2, which is a major regulator of oxidative responses. We further identified that the expression of hmox1a, a downstream target of Keap1-Nrf2 signaling pathway, was markedly increased upon smarca5 deletion. Importantly, overexpression of keap1a or knockdown of hmox1a partially rescued the blood clot formation, suggesting that the disrupted Keap1-Nrf2 signaling is responsible for the RBC aggregation in smarca5 mutants. Together, our study using zebrafish smarca5 mutants characterizes a novel role for smarca5 in RBC aggregation, which may provide a new venous thrombosis animal model to support drug screening and pre-clinical therapeutic assessments to treat thrombosis.

    1. Cell Biology
    2. Developmental Biology
    Maia Kinnebrew et al.
    Research Advance

    A long-standing mystery in vertebrate Hedgehog signaling is how Patched 1 (PTCH1), the receptor for Hedgehog ligands, inhibits the activity of Smoothened, the protein that transmits the signal across the membrane. We previously proposed (Kinnebrew et al., 2019) that PTCH1 inhibits Smoothened by depleting accessible cholesterol from the ciliary membrane. To directly test the effect of PTCH1 on accessible cholesterol, we measured the transport activity of PTCH1 using an imaging-based assay to follow the kinetics of cholesterol extraction from the plasma membrane of live cells by methyl-β-cyclodextrin. PTCH1 depletes accessible cholesterol in the outer leaflet of the membrane in a manner regulated by its ligand Sonic Hedgehog and the transmembrane potassium gradient. We propose that PTCH1 moves cholesterol from the outer to the inner leaflet of the membrane in exchange for potassium ion export. Our results show that proteins can change accessible cholesterol levels in membranes to regulate signaling reaction.