The Notch and EGFR signaling regulate caspase inhibitor Diap1 to match supply with intestinal demand

via also enteroblasts. These data provide into how drives adult and of caspase provide new insights into the contribution of programmed cell death in adult tissue homeostasis, new avenues for future investigation of intestinal EBs nodal EGFR. epistatic data


Introduction
The Drosophila intestinal epithelium is renewed completely several times over its 40-50 day adult life in a process that takes one to three weeks under normal homeostatic self-renewal (Antonello et al., 2015;Ohlstein and Spradling, 2007). However, after an overt damage, it is renewed in just two-three days (Buchon et al., 2009;Ohlstein and Spradling, 2007). The wide range of physiological turnover time reflects the stochastic damage of absorptive enterocytes (ECs), the main cells in the intestinal epithelium , by exposure to pathogens and toxins present in food and chemical and physical stress, and the adaptive capacity of the intestine. The intestine also contains secretory enteroendocrine (EE) cells, which constitute only 10% of the intestinal population and renew themselves at a slower rate than ECs (de Navascués et al., 2012;Sallé et al., 2017;Parasram et al. 2018). Intestinal cell turnover is sustained by a small population of ISCs scattered throughout the epithelium that, as observed in other high turnover epithelia in mammals (Simons and Clevers, 2011), divide regularly and produce, with each division, one cell that differentiates and one that remains undifferentiated and will divide in the same way to produce a new cell stem and a cell committed towards the same lineage as the previous division or into the other intestinal cell type (e.g. Micchelli and Perrimon, 2006;Ohlstein and Spradling, 2006;2007). ISCs can also divide symmetrically (de Navascués et al., 2012;O'Brien et al., 2011) and directly differentiate into EE without division (Zeng and Hou, 2015). However, in the highly renewing intestine, stem cells must operate rapidly and efficiently by providing enough new cells to replenish daily tissue demand. Moreover, since multipotent ISCs have different options in terms of cell lineage, an outstanding open question is how individual stem cells rapidly and adaptively produce the distinct tissue cell types at the right number to sustain tissue homeostasis in short-and long-term.
In Drosophila, ISCs are the only mitotic cells and their immediate committed progeny, enteroblasts (EBs), terminally differentiate without further division into only two possible cell types (EC or EE) through differential Delta (Dl)-Notch (N) activation in the daughter cell that become the committed progenetitor cell (Ohlstein and Spradling, 2007;Perdigoto et al. 2011). These features of the adult fly gut simplify the analysis of the dynamics of ISC production during homeostasis. Drosophila ISCs divide slowly but continually and produce EBs that can remain incompletely differentiated for long periods in the absence of a local (Antonello et al., 2015). The existence of such a pool of EBs in homeostatic intestines was first suggested in studies of infection challenge (Buchon et al., 2009) but only formally established by multicolour tracing methods (Antonello et al., 2015). After infection, or genetic induction of enterocyte death (Buchon et al., 2009;, the ISC proliferation rate increases to cope with the increased demand for new cells. The mitotic index can increase from 3 to 5 mitosis per midgut to more than 100 mitosis per gut, however, ISC proliferation rate peaks only 24-48 hours post challenge (Buchon et al., 2009;). During the time interval between challenge and the raise of ISC proliferation, it has been shown that the 'pre-existing' EB pool serves as the intestine's first defence (Buchon et al., 2009;Antonello et al., 2015). A fundamental question is how the stem cell population performs in times of low intestinal demand since maintaining an 'unnecessary' population of immature EBs may increase work load and metabolic demand and thus lead to poorer organ performance or risk of tissue hyperplasia (Zhai et al., 2015).
Developing organs often use a strategy of overproduction followed by culling of the excess cells via programmed cell death (PCD) to ensure correct organ size and shape (Fuchs and Steller, 2011). Although counterintuitive that adult stem cells overproduce (i.e. divide superfluous), here we discovered that ISCs divide in greatly excess to physiological demand and cull excess enteroblast cells via a PCD similar to those operating during morphogenesis. We found Dl-induced N activation primes enterocyte-committed progenitor cells to death via the pro-apoptotic transcription factor Klumpfuss/WT, which also regulates fate diversification. Excecution of cell death involves activation the RUNX homologue Lozenge and is counteracted by environmental survival signals acting via the Epidermal Growth Factor Recepto (EGFR) in enteroblast cells, impinging on regulation of the caspase inhibitor DIAP1. Loss of diap1 resulted in complete loss of enteroblast and abrogation of intestinal renewal, whereas blocking caspase genes uncovered that more than half of the enterocyte-committed progenitor cells produced by the ISCs are eliminated by PCD in the physiological intestine in conditions of low demand. In addition, selective elimination of apoptosis in progenitor cells is sufficient for tumorigenesis to occur. These data provide new insights into the contribution of programmed cell death in adult tissue homeostasis, opening new avenues for future investigation of intestinal cancer.

Intestinal Stem Cells Do not Adjust their Division to Slowing Intestinal Cell Replacement
author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https: //doi.org/10.1101/493528 doi: bioRxiv preprint In order to test whether ISCs division adapt to situations of low demand, we developed a 'Low demand' protocol to minimize the need for intestinal cell replacement ( Figure 1A). The key features of our protocol to assess ISC's production that differentiate from previously studies are: first, to minimize changes of pathogens accumulation in the food, which is the leading cause of EC damage (Apidianakis and Rahme, 2010), we frequently transferred flies to fresh food vials (i.e. 3-4 days old flies were transferred to fresh food vials every 48 hour: Figure 1A). Second, we used our tracing method ReDDM ('Repressible Dual Differential Marker': Antonello et al., 2015), which allows to trace differentially stem and progenitor cells and the differentiated progeny ( Figure 1B). New ECs are detected as RFP-only cells by the ReDDM tracing method ( Figure 1B), whereas ISCs and/or EBs are detected as double GFP-RFPlabelled cells. Control flies were transferred to fresh vials every week as previously reported (Antonello et al. 2015;Jian et al., 2009) and as expected from unpredictable fluctuating demands, there was increased renewal of ECs and substantial variation among individual midguts ( Figure 1C, pattern histogram). This widely used standard culturing condition will refer to as 'Variable' demand. In 'Low' demand, few EC had renewed at day 7, 14 and even 21 after tracer induction (esg ReDDM -midgut; Figure 1C, solid histogram) compared with control gut in normal culturing condition (new food vial every week) where the intestinal epithelium had renewed completely after three weeks ( Figure 1C), as previously reported (Antonello et al., 2015;. In spite of the slow epithelial cell turnover in 'Low demand' condition, ISCs continued dividing at the same rate that midguts reared in normal culturing conditions ( Figure 1D). PH3 + counts showed that ISCs in 'Low demand' maintained a constant proliferation rate (2-5 mitosis/midgut) at day 7, 14 and 21 ( Figure 1D). Representative images of guts reared at the standard condition (refered as variable demand) and the low demand condition are in Figure 1E-G. No accumulation of EBs was observed in 'Low' midguts after three weeks of continued ISC division ( Figure   1G). This could not be attributed to terminal differentiation because the ReDDM tracing method detected very few renewed ECs (RFP-only labelled cells reflect the newly renewed ECs, Figure 1F-G). Prompted by this observation, we next investigated directly whether stem cell production might be regulated by a PCD.

Undifferentiated Progenitor Cells and Tumor Masses
author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint We first drove in age-synchronized cohorts of adult flies the expression of the baculovirus caspase inhibitor p35 in ISCs and EBs using esg ReDDM system and found a significant accumulation of esg-positive (esg+ve) cells after 7 days of p35 expression (esg ReDDM >p35: compare with control Figure 1H and I) and the occasional presence of tumour-like masses ( Figure 1J). This evidenced that ISCs produce in excess to demand uncovered a role of caspases in controlling stem and/or progenitor cell number. We next used the caspase sensor Apoliner that allows the detection of an even rarer number of apoptotic cells because it marks early steps of apoptosis while the cell still appears morphologically normal (Bardet et al., 2008).
To distinguish from apoptosis induced in aging or damaged ECs and EEs, we drove Apoliner specifically in ISC and EBs using esg-Gal4 and in this way we mapped caspase activity within the EB population (Figure S1A-C). This provides a first hint that apoptosis occurs in EB population. This notion is reinforced later by functional studies, and analysis of apoptosis using TUNEL (Terminal Deoxynucleotidyl Transferase (TdT)-mediated dUTP Nick-End Labelling; see below).

Diap1.
The pro-apoptotic proteins Reaper (Rpr), Head involution defective (Hid) and Grim regulate most apoptotic deaths by counteracting Diap1 (Hirata et al., 1995;Holley et al., 2002;Meier et al., 2000) ( Figure 2A). The H99 deficiency removes these pro-apoptotic genes (Fuchs and Chen, 2013) and the adult intestine of flies heterozygous for Df(3L)H99 had a two-fold increase in number of EB ( Figure 2B and C) as determined by counting Diap1-GFP cells ( Figure 2D), consistent with apoptosis-mediated culling of EBs being a physiological process. Importantly, in Df(3L)H99 heterozygous flies, many EBs show reduced expression levels of Diap1 compared with control guts, as suggested by the reduction of fluorescence intensity of the reporter ( Figure 2B, C and E). One interpretation of this finding is that EBs author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint with low Diap1 can survivel when gene dosage of of pro-apoptotic genes is reduced. This would suggest that survival of an enteroblast might depend on the balance between survival and apoptotic inputs.
To investigate the dynamics of PCD in regulating EB number without perturbing ISCs, we then used a Gal4 enhancer trap in klu gene, encoding the Drosophila Wilms' Tumour 1 homologue (Klein and Campos-Ortega, 1997), which we found drives expression selectively in EBs ( Figure S1E and F). Lineage tracing of klu-Gal4 (klu + ) cells using ReDDM showed all differentiated cells derived from klu + EB cells are ECs and none of them are destined to EE fate ( Figure S1F-H).
We assayed adult-and EB-selective RNA interference (RNAi)-based silencing of the caspases Debcl Diap1-RNAi, we used the TUNEL assay, which detects extensive DNA degradation (late stages of apoptosis). TUNEL method led to strong decay of fluorescent reporters and we could only identify progenitors marked by GFP using the stronger Gal4 driver, esg-Gal4 (Ohlstein and Spradling, 2006) but not using klu-Gal4 or Su(H)-Gal4.
Silencing of the caspases Debcl and Drice using esg-Gal4 were accompanied by a decrease in the number of apoptotic esg-positive (esg + -ve) cells, while silencing Diap1 or expression of Dronc were accompanied by an increased number of apoptotic esg + cells assayed by TUNEL method ( Figure 2L) and anti-activated caspase-3 labelling ( Figure S2F). These observations support the notion that homeostatic ISCs often overproduce in relation to demand and that a significant number of EBs is eliminated by apoptosis in low demand.
author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint We hypothesized that when ECs are more rapidly renewed in the intestine the occurrence of EB death diminishes as the risk of EB accumulation is also lower. This would suggest that PCD in the adult midgut EBs is a tuneable process. Because all EBs express and require Diap1 (Figure 1 and 2), we first hypothesised that cell death might be a 'default' fate of EBs that is counteracted by Diap1. We thus investigated how the EB death and life decisions might be regulated.

Enteroblast Death/Life Decision is Governed by Opposing Notch and EGFR Activity
PCD is essential during development, the removal of faulty cells during adult tissue homeostasis, the establishment of immune self-tolerance, and the killing by immune effector cells (Baehrecke, 2002;Mollereau et al., 2012;Protzer et al., 2008). Moreover, spontaneous apoptosis has also been observed in some adult tissues within the stem/precursor cell compartment (Domen and Weissman, 1999;Potten, 1992). Caspase-induced cell death in old or damaged cells often makes use of the cell death-and stress-responsive Jun N-terminal Kinase (JNK) cascade and it is typically associated with a compensatory proliferation to stimulate cell replacement (Biteau et al., 2008;Chen, 2012). In contrast, the purpose of culling excess but healthy cells is reducing their number and this type of apoptosis typically occurs without compensatory proliferation during development (Raff, 1996). We did not find evidence that apoptosis in the EBs may be regulated by JNK because EBs labelled by Diap1-GFP did not overlap with JNK activity measured using the JNK activity-sensitive reporter TRE-dsRED (Chatterjee and Bohmann, 2012) ( Figure S3A and B). Thus, EB death appeared to be triggered by signals different to those regulating death of damaged or aged intestinal cells supporting the idea of a physiological cell death aimed at culling EB number.
The N and EGFR pathways are prime candidates as they often act antagonistically to promote cell death/life decisions during development (Dominguez et al., 1998;Protzer et al., 2008). Both N (Micchelli and Perrimon, 2006;Ohlstein and Spradling, 2007;Perdigoto et al., 2011) and EGFR Zhai et al., 2015) have critical roles in stem-cell renewal, EB specification, differentiation, and lineage commitment in adult gut homeostasis and regeneration Jiang and Edgar, 2012;Kapuria et al., 2012;Perdigoto et al., 2011) but their role in EB death/survival has not been explored. We investigated the effect of cell-autonomous manipulations of N and EGFR signalling in EBs using the klu ReDDM ( Figure 3A-E). Changes in EB number were determined by quantification klu + cells using lineage-tracing ReDDM ( Figure 3F), complemented by assessment of author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint cell deaths within the progenitor cell compartment detected by GFP + TUNEL + cells using esg ReDDM ( Figure 3G). EC renewal (quantified as RFP + -only cells) and non-autonomous effects on stem cells mitosis (PH3 + cells) were also assessed in the klu ReDDM guts with altered N and EGFR activity ( Figure   S3C and D).
EB-selective overexpression of the dominant negative form of N ( Figure 3B) and of a constitutively active allele of EGFR protein ( Figure 3C) both led to EB cells accumulation (compared to control Figure   3A). Endogenous over-activated N signalling by RNAi silencing of Hairless (H) ( Figure 3D), the main antagonist of N (Bray, 2016), or by expression of a constitutively active N intracellular domain (data not shown) and expression of a dominant negative form of EGFR ( Figure 3E) decreased EB number as assessed by reduced number of klu + cells ( Figure 3F). Consistently, the loss of EBs caused by gain of N signalling or by EGFR inactivation in EBs was rescued by concomitant inhibition of apoptosis via Debcl-RNAi expression, also allowing new ECs to be formed ( Figure 3H and I). Previous work has established that activation of EGFR in ISCs by EGF-like ligands produced by different intestinal cells in response to damage acts as a paracrine non-autonomous signalling, stimulating ISC proliferation Zhai et al., 2015). In agreement with this, we observed that activation of EGFR in EBs using klu ReDDM system, non-autonomously stimulated ISCs division ( Figure 3, arrowheads indicate PH3 + , quantified in Figure S3 and extra ISCs labelled by Dl marker is shown in Figure 3C inset).
These observations indicated that the massive accumulation of klu + EB cells in the midgut with constitutive activation of the EGFR pathway arose by both the suppressed EB deaths (autonomous effect) and increased ISC divisions (non-autonomous effect). Thus, in addition to its documented role in EB for stimulating ISC proliferation Zhai et al., 2015), our data illustrate that the activity of EGFR is also crucial to suppress PCD in EBs. Consistently, we found that gain and loss of N and EGFR were accompanied by increased or reduced apoptotic esg + GFP + as assessed by TUNEL + method ( Figure 3G), and that these gains and loses can be attributed to EB death/survival as shown in Figure 2). Specific manipulation in progenitors using klu ReDDM also suggests that the increase in ISC division is likely a secondary consequence of progenitors escaping cell death (quantification of PH3 + in Figure S3D).
Dl is the only N ligand acting in adult intestinal homeostasis and regeneration (Ohlstein and Spradling, 2007). Only ISCs express Dl, suppressing stemness in the neighbour daughter cell, driving N-mediated progenitor cell commitment towards the EC lineage (e.g. Ohlstein and Spradling, 2007;; Perdigoto author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint et al., 2011). Our data suggests that Dl-mediated activation of N also 'primed' progenitor cells to PCD.
Thus N and EGFR signalling counterbalances EB death/life decision in the adult midgut along with their crucial role in stem cell self-renewal, proliferation, differentiation, and lineage commitment e.g. Ohlstein and Spradling, 2007;Perdigoto et al., 2011;Zhai et al., 2015).
This suggests that within the same cell, N and EGFR influence several cellular fates, including cell death.
We hypothesize that EBs not receiving sufficient survival signals (e.g. EGFR low ) will be driven to PCD via the pro-apoptotic input by Dl-N signalling, while EBs with low N activation or receiving high antiapoptotic input (EGFR high ) will accumulate undifferentiated or terminally differentiate upon receiving terminal differentiation input. These findings support that Diap1 gene is a nodel point of the impact of N and EGFR activity in EB death/survival decision (see below and Discussion). In addition, N and EGFR signalling are both capable to promote terminal differentiation, suggesting that death/survival/differentiation may represent a continuum rathen than discreet cell fates.

Culling, Fate Diversification, and Differentiation
To identify candidate transcription factors that may determine the outcome of N signalling, we focused on evolutionarily conserved N targets, the RUNX homologue Lozenge (Lz/RUNX) and Klu/WT1 (Protzer et al., 2008). Klu/WT is also a conserved mediator of PCD during development (Rusconi et al., 2004).
Using antibodies against lz/RUNX protein and lz-Gal4 reporter revealed that the occurrence of Lz + cells in a physiological midgut was extremely low (only 1-2 Lz + per posterior midgut; Figure S3E). Tracing the fate of lz + cells using lz ReDDM showed that they are likely EB 'committed' to die as they did not yield differentiated ECs ( Figure S3F). Consistent with this idea, blocking apoptosis via p35 expression (lz ReDDM >p35: Figure S3G) increased the number of cells derived from lz + progenitor, supporting that lzexpression may mark EB terminally fated to die. Forcing expression of lz in ISCs and EBs using esg ReDDM led to significant increase in apoptotic esg + cells comparable to those obtained by manipulation of N and EGFR (TUNEL + esg + , Figure 3G). Additionally, esg ReDDM >lz midgut also showed precocious terminal differentiation, reflected as small sized ECs ( Figure S3H and H'), suggesting that Lz/RUNX is downstream of N signalling as no enteroendocrine (Pros + ) cells appear to be traced after lz overexpression. When lz was depleted via RNAi using klu ReDDM during 7 days ( Figure 3K), we observed a significant increase in EB number and, when using esg-Gal4, decreased TUNEL + esg + cells ( Figure 3G).
Quantification of EC turnover showed that EB loss was not due to terminal differentiation ( Figure S3C) author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint and PH3 + labelling revealed that lz-RNAi in EBs also non-autonomously moderately increased ISC mitosis ( Figure S3D). This observation is consistent with prior work in the developing fly retina (Behan et al., 2002;Siddall et al., 2003;Wildonger, 2005) showing that Lz promotes apoptosis through downregulation of EGFR signalling. Consistently, lz overexpression in ISCs/EBs using esg ReDDM suppressed accumulation of EB caused by activated EGFR signalling ( Figure S3I) while depleted lz partially suppressed EB loss by EGFR DN (Figure S3J). Collectively, these findings suggest Lz is required for the irreversible commitment to death downstream of N, and possibly other factors yet to be defined in the adult Drosphila midgut. In EBs with sufficient survival signals, ectopic lz expression triggered precocious terminal differentiation. This supports that N-Lz axis also regulate terminal differentiation along with EGFR and other differentiationg signalling.
Overexpression and depletion of klu also influenced esg + cell death decision as detected by TUNEL method ( Figure 3G). Additionally, a role of Klu in sustaining EC fate commitment was also uncovered from klu-RNAi experiments using esg ReDDM and klu ReDDM (see Figure S3). EC fate determination requires the Dl-mediated activation of N signalling in EBs (Kapuria et al., 2012;Perdigoto et al., 2011). In the absence of N activation, upon stem cell division both daughter cells adopt the ISC fate or terminally differentiate as EEs, which can be marked by Pros (Jiang and Edgar, 2012). However, in the presence of a high Dl-N signalling, the daughter cell receiving N signalling commits to differentiation towards the EC lineage, which is reflected by the committed progenitor cell increasing its size over time via endoreplication (Lucchetta and Ohlstein, 2012;Perdigoto et al., 2011). This fate-determination is generally depicted as an irreversible commitment. However, we observed that depleting klu by RNAi in EBs (using klu ReDDM ) which is normal gut will only yield ECs caused a switch of their fate towards the EE lineage ( Figure S3K-N). klu RNAi midguts showed cells with large nuclei (high DNA content shown by DAPI indicate polyploid cells) labelled with Pros (red arrowheads, Figure S3K), supporting conversion of EC-'committed' EBs towards the EE lineage. Typically wild type midguts have a ratio of 10% EE (Pros + ) and 90% EC (de Navascués et al., 2012), and no EE cells are normally derived by klu + progenitors ( Figure S3M). However, klu depleted klu + EBs (klu ReDDM >klu-RNAi, 7 days) generated approximately 33% of EEs instead of 0%, indicating a continuous requirement for Klu gene expression to maintain the EC commitment fate. This experiment also uncovered an unanticipated plasticity of EBs to adopt an alternative fate. These experiments can be interpreted in two alternative models. In one model, Klu + activity is required in all EBs for their robust establishment of the EC lineage, as well the regulation author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint of their number. In the second model, ISCs generates stochastic EC and EE cell fates (~50% EC and ~50% EE) but N activation, and upstream signals such as insulin, could bias the fate towards EC lineage, resulting in ~00% EC and ~10% EE. In such scenario, Klu + would be required to robustly maintain the EC fate in a subset of the EBs, which could account for the observation that EC fate still occur in klu ReDDM >klu-RNAi guts. Regardless the mechanism, these findings are in line with recent studies of mammalian stem cell systems that suggest that progenitor cells are primed, not committed, and cell fate decisions remain tuneable by external inputs (Notta et al., 2015).

Steady Stem Cell Divisions
Both PCD and stem cell division are energetically costly (Vaux and Korsmeyer, 1999). Thus, to further explore the potential advantages of this apparently costly strategy for adult tissue renewal, we built a computational model based on these experimental observations to evaluate the performance of a stem cell system in response to injury. We compared a hypothetical system controlling production only by mitosis with a system in which stem cells produce a continuous pool of progenitor cells with further control of their numbers by PCD. The model includes a feedback mechanism by which the steady-state number of EC determines both the rate of ISC division and the probability of EB to undergo either differentiation or apoptosis ( Figure S4). The model predicts a tighter control of the number of EC under homeostatic conditions and a faster recovery from acute damage (a sudden EC loss for example through injury) ( Figure 4A and B).
A simulated intestinal turnover with single ( Figure 4A) and two successive challenges ( Figure 4B) revealed that both models cope with a loss of around 30 EC per day, which was the observed average intestinal cell loss in 'Low demand' ReDDM-tracing experiment ( Figure 1A, right graph). However, when we challenged the models with one or two successive acute damages ( Figure 4A and B, respectively), only the model with pre-existing EBs owing to continual ISC divisions predicted returns to homeostasis after one to two days consistently with the estimated recovery time determined by previous experimental studies (Amcheslavsky et al., 2008;Buchon et al., 2009;. These simulations hinted at EB death fate in the intestine being a tuneable fate decision presumably by extrinsic cues that also stimulate EB differentiation for cell replacement. Consistent with this idea, we noticed that EB express a varying level of Diap1 (assessed by GFP fluorescent intensity, Figure 4C). author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint Importantly, single manipulations of EGFR, N activity or caspases inhibition markedly changed numbers of Diap1 + cells (EBs: Figure 4D; see also quantification of terminal differentiation, new ECs in Figure   4E) along with Diap1 levels as assessed by changes in their fluorescent intensity ( Figure 4C). The baculovirus p35 antiapoptotic factor blocks apoptosis downstream of the Diap1 transcriptional repression by Rpr/Hid/Grim factors (Bergmann and Steller, 2009;Hirata et al., 1995;Holley et al., 2002;Meier et al., 2000). As anticipated, expression of the p35 caused accumulation of EBs with many of them exhibiting low Diap1-GFP levels ( Figure 4C and D). This would suggest that EBs cells displaying low Diap1 levels would normally undergo PCD but escape death by expressing the anti-apoptotic factor p35.
Importantly, this 'tuneable' Diap1-GFP is also observed when gene manipulations are done using another promoter, esg-Gal4 ( Figure 4F). Note that EBs with constitutively active EGFR had an excess of EBs but with highly variable Diap1-GFP fluorescence level intensity ( Figure 4F) as seen in quantification in Figure 4C with the klu-Gal4, supporting that Diap1 transcription is tuneable by various signals simultaneously. This is consistent with Diap1 transcription being directly regulated by several pathways intestinal stem/progenitor cell compartment (Potten et al., 2002) were attributed to a protection strategy to eliminate damaged or aged stem cells. Recently studies in mice have shown that endogenous apoptosis helps to regulate stem or progenitor cell numbers and regeneration . Our study thus provides a paradigm for how the culling process may operate during cell turnover and how this process is interwoven with proliferation and cell fate determination to ensure that the correct cell types and number are produced.

DISCUSSION
We have found that steady-state intestinal stem cells production is not solely controlled by mitosis, but also by a culling process of progenitors. Our observations support a model whereby adult ISC overproduce progenitor cells to ensure rapid intestinal cell renewal in the face of sudden and author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint unpredictable demands, thereby efficiently preserving homeostasis and intestinal barrier. Under normal physiological conditions, demand equals supply by ISCs, but in low demand (e.g. during period of fasting) the ISC's production exceeds the tissue demand and EB number is adjusted by a N-Klu-Lzmediated death via caspase-dependent programme. We have also shown that the elimination of surplus EBs is a critical tumour suppressor strategy. Thus ISCs performance both promote and limit tumorigenesis, and this findings may explain earlier observations of regeneration defects when endogenous inhibitors of apoptosis were impaired . Moreover, our study identifies nodal points for N and EGFR. including Diap1 (survival) and an effector of cell death Lz. Our epistatic data suggest that Lz may mediate cross-talk between N and EGFR to reinforce cell death commitment by dampening EGFR signalling downstream of the receptor as seen during development (Wildonger, 2005), which may explain how these pathways determine robust outcomes of N signalling.
N signalling requires the continuous interaction of N protein with its membrane-bound ligand Dl in the adjacent stem cells (Ohlstein and Sprandling, 2007;Simon and Clevers;, Liang et al., 2017. EGFR signalling can be activated in the EB in response to multiple EGF-like signals released by the niche, as well as dying ECs (Liang et al., 2017) that also stimulate ISC's proliferation, providing different scenarios for how survival signals may modulate committed progenitor-cell numbers. N and EGFR oppositely control death and life decisions in other cellular context during development (Baker and Yu, 2001;Gilboa and Lehmann, 2006;Protzer et al., 2008). However, in these other developmental contexts apoptosis is highly stereotyped with an invariant outcome and often occurs after cell fate determination. In adult tissues with high and constant demand for cell turnover, the supply of precursor cells needs to be regulated dynamically and adaptively in coordination with cell fate diversification to respond efficiently to changing environmental conditions and with sudden increase in demand. Indeed, while N and EGFR act oppositely to control death/survival of EBs, they are both positively required for EB terminal differentiation. Moreover, the fact that Dl-N signaling can directly drive expression of both rpr (Krejci et al., 2009) and Diap1 (Djiane et al. 2013) supports that N activation may 'prime' EB to death and that execution of the death programme depends on the balance of pro-apoptotic and pro-survival signals the EB cells receive, which may be reflected by Diap1 levels. These data also indicate that EBs are actually executing ternary (self-renewal, death, or differentiation) not binary choice (self-renewal or differentiation).
We postulate that a death signal emanating from the stem cells in direct contact with its committed author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint progeny allows a flexible accommodation of EB-number to ISC production. This death programme is also tuneable by survival signals produced by dying differentiated cells, from the niche, or other environmental cues, which further attunes EB-number regulation to varying physiological and pathological conditions. In the haematopoietic system stochastic cell choice provides flexibility for the maintenance of production of all blood cell lineages in the face of substantial demand for one particular lineage (Enver et al., 1998). In a speculative manner, we suggest that ISC dividing ahead of demand may similarly generate stochastic cell choice (EC and EE) with a high N-Klu biased fate towards the EC lineage. Derangement of apoptosis-mediated EB number regulation along with fate conversion may explain the previously observed tumors associated with impaired N signalling  as supported by our findings ( Figure 1J). Similarly, the same mechanism of altered apoptosis within the stem/progenitor cell compartment causes hyperplasia and tumour formation in the murine intestine . Our study provides a regulatory logic for the adjustment of progenitor numbers intertwined with both fate diversification and tissue demand.

TUNEL Assay in Stem and Progenitor Cells
Adult midgut of the indicated genotypes and age were dissected, fixed, and immunostained to detect dying GFP-labelled progenitor cells by in situ Cell Death Detection Kit (Roche Applied Science, Grenzach, Germany) according to manufacturer´s protocol followed by a DAB-reaction (Thermo Fisher, Schwerte, Germany). GFP + ve and TUNEL + ve cells were quantified using a Nikon Fluorescence microscope (Eclipse 90i). Data represent the proportion of TUNEL + ve cells relative to total GFP + ve cells.
Graphs and all statistical analyses were performed using Graphpad Prisma 6, and data were analysed using ANOVA analysis of variance with Bonferroni correction statistical test and Student´s t-tests. author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint

Image Acquisition
Confocal images were obtained with a Leica TCS SP5 inverted confocal microscope, using a 1024 × 1024 image size. Stacks were typically collected every 1 µm and the images were reconstructed using max projection. Images were evaluated and scaled using Fiji/ImageJ. In all cases, the images shown in the Figures are representative of the effect of the genetic manipulation.

Quantitative PCR
To assess the efficacy of the RNAi transgenes, mRNA was extracted from wandering third instar larvae with the corresponding RNAi transgene (hsp70-Gal4 > UAS-RNAi) or without (control, hsp70-Gal4 >) after a 1 h heat shock at 37 °C to induce transgene expression. To determine mRNA levels we used superScript First-Strand Synthesis System for RT-PCR (Invitrogen) and SYBR Green PCR Master kit (Applied Biosystems), according to the manufacturer's instructions. The cDNAs were amplified using specific primers designed using the ProbeFinder software by Roche Applied Science, and RpL32 was used as a house-keeping gene for normalization.
The following primers were used:

Quantification and Statistical Analysis, Cell Counting and Fluorescence Measurements
For progenitor cell counts, 20x images of ReDDM midguts of the different genotypes and conditions were cropped with ImageJ (Fiji 64bit) for processing and quantification with Matlab. A self-written script optimized for the ReDDM method that analyses quality (size, colour) and quantity (count events) in posterior midguts was used to measure the number of progenitor cells (double positive mCD8::GFP H2B::RFP cells) and DAPI nuclei (Antonello et al., 2015). Measurements of fluorescence intensity of author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint Diap1-GFP levels were acquired with a fixed 488nm laser intensity and images of the Diap1-GFP midguts were analysed using a Fiji-script for intensity per cell available from the authors. Representative images are shown in all panels. For initial experiments exploring the role of apoptosis, at least 20 posterior midguts were scored. For complex experiments analysing many genes conditions usually cell counting were done in at least four to 10 posterior midguts.

Enteroblasts.
author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint (A) Scheme for tracing intestinal progenitor cell number regulation in midgut with low intestinal renewal ('Low') using the lineage tracing ReDDM system (Antonello et al., 2015). The strategy relies on minimizing exposure of flies to contaminated food by transferring flies to fresh food vials every two days ('Low') as compared with normal culturing conditions in which flies are typically transferred to fresh food vial every week. This 'Low' demand strategy effectively minimized intestinal renewal. (B) Scheme of the ReDDM tracing method. This system uses two fluorescent transgenes with short-(membrane CD8::GFP, green) and long-term (nuclear H2B::RFP, red) stability and the Gal80 repressor (tub1-   (n= 9, 9, 9, 9). Box and whisker plot showing the means and values range. P values (p < 0,02) calculated by one-way ANOVA with Bonferroni correction. Scale bars, 100µm.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/493528 doi: bioRxiv preprint