Differential effects of commensal bacteria on progenitor cell adhesion, division symmetry and tumorigenesis in the Drosophila intestine

To maintain an effective barrier to invasion by gut-resident microbes, multipotent intestinal stem cells (ISCs) divide at a rate that matches the loss of dead or damaged epithelial cells (Alam and Neish, 2018; Amcheslavsky et al., 2009; Biteau and Jasper, 2011; Buchon et al., 2009a; Jiang et al., 2009; Shaw et al., 2010). Commensal bacteria promote epithelial growth (Broderick et al., 2014; Buchon et al., 2009b; Cheesman et al., 2011; Li et al., 2012; Zackular et al., 2013), and disruptions to microbiota composition or diversity are associated with proliferative diseases such as colorectal cancer (Gagnière et al., 2016; Zeller et al., 2014). Thus, we consider it important to understand how the gut bacterial community influences ISC division and differentiation.

Drosophila melanogaster is a popular system in which to study microbial control of ISC division because there is an extensive toolkit for host genetic manipulation, and it has a simple, cultivable microbiome that is easy to modify (Broderick and Lemaitre, 2012; Koyle et al., 2016). Importantly, key regulators of ISC division are evolutionarily conserved between flies and vertebrates (Miguel-Aliaga et al., 2018). For example, vertebrate and fly ISCs reside in a niche that uses related growth factors to direct stem cell division and differentiation (Jiang and Edgar, 2011; Morrison and Spradling, 2008; Takashima and Hartenstein, 2012). Integrins are particularly important niche regulators of ISC division. Integrins anchor fly ISCs to a basal extracellular matrix, orienting the mitotic spindle at an angle to the basement membrane, and ensuring polarized distribution of cell fate determinants (Chen et al., 2018; Goulas et al., 2012). In asymmetric divisions, the apical daughter exits the niche and terminally differentiates as a mature epithelial cell (Martin et al., 2018; Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006, 2007), whereas the basal daughter remains within the niche, where it retains ‘stemness’. Depletion of integrins from the fly midgut diminishes asymmetric division frequency, promoting symmetric expansion of stem cell lineages and epithelial dysplasia (Goulas et al., 2012; Lin et al., 2013; Okumura et al., 2014). Notably, relationships between integrins and ISC growth are evolutionarily conserved, as integrin loss also causes hyperplasia in the mouse intestine (Jones et al., 2006).

In flies, most asymmetric divisions generate a post-mitotic enteroblast that differentiates as a large, absorptive enterocyte in response to Notch pathway signals (Bardin et al., 2010; Biteau and Jasper, 2014; Guo and Ohlstein, 2015; Ohlstein and Spradling, 2007; Zeng and Hou, 2015). Loss of Notch from stem cell/enteroblast progenitor pairs leads to rapid growth of epithelial tumors characterized by hyperplastic stem cells, absence of enterocytes, and accumulation of secretory enteroendocrine cells (Patel et al., 2015). Disruptions to Notch cause similar dysplastic phenotypes in fish and mice (Crosnier et al., 2005; Qiao and Wong, 2009), and spontaneous accumulation of mutations at the Notch locus is linked to age-dependent development of intestinal tumors in adult Drosophila (Siudeja et al., 2015).

Previous work established that pathogenic Pseudomonas aeruginosa causes tumors in flies that express a latent Ras oncogene in progenitor cells (Apidianakis et al., 2009). More recently, we used the Notch-deficient model to show that commensal bacterial are also effective agents of intestinal tumorigenesis (Petkau et al., 2017). However, it is unclear which taxa promote tumors, and how this happens. To understand how gut bacteria regulate epithelial proliferation, we determined the impacts of common fly commensals on wild-type and Notch-deficient progenitors. We found that Lactobacillus brevis stimulated stem cell proliferation and tumorigenesis, whereas a close relative, Lactobacillus plantarum, did not. Upon further analysis, we found that the L. brevis cell wall was sufficient to promote tumors in Notch-deficient intestines. Mechanistically, L. brevis decreased expression and altered the subcellular distribution of progenitor cell integrins. Consistent with essential requirements for integrins in regulating asymmetric progenitor growth, we found that association with L. brevis increased the frequency of symmetric stem cell divisions, resulting in greater numbers of ISCs than post-mitotic enteroblasts. Combined, our data implicate a common fly commensal in the symmetric expansion of stem cell lineages, promoting proliferation of stem cells that harbor tumorigenic lesions.

To study bacterial effects on intestinal tumors, we used the temperature-controlled escargot-GAL4, GAL80^ts, UAS-GFP (esg^ts) transgenic fly line to express an inducible Notch RNAi construct (UAS-N^RNAi) in ISCs and enteroblasts (collectively referred to as progenitor cells) at the restrictive temperature of 29°C. Intestines of control esg^ts/+ females contained evenly distributed GFP-positive progenitors and Prospero-positive enteroendocrine cells in a simple epithelium dominated by large, polyploid enterocytes (Fig. 1B; Fig. S1A). In line with an earlier study that described massive stem cell growth in flies with Notch-deficient progenitors (Patel et al., 2015), we found that depletion of Notch (esg^ts/N^RNAi) caused multilayered midgut tumors populated by excess progenitor and enteroendocrine cells within 5 days (Fig. 1C; Fig. S1B). As tumors were evident in female intestines, but largely absent from males (Fig. S1C), we performed all subsequent experiments on adult female posterior midguts.

In contrast to intestines with a conventional microbiome, Notch-deficient tumors rarely appeared in age-matched, germ-free (GF) flies, indicating microbial requirements for tumor growth (Fig. 1D), although we cannot exclude the possibility that tumors eventually form in GF flies with age. To identify bacterial species that promote tumors, we examined posterior midguts of adult esg^ts/N^RNAi flies that we associated exclusively with common species of Lactobacillus commensals, a dominant genus within the fly microbiome (Adair et al., 2018; Wong et al., 2013). To focus exclusively on adult tumors, we raised esg^ts/N^RNAi larvae with a conventional microbiome under conditions that prevent Notch inactivation. Upon eclosion, we fed adults an antibiotic cocktail that depleted the bacterial microbiome to below detectable levels, and re-associated flies with Lactobacillus brevis (Lb), or Lactobacillus plantarum (Lp) (Fig. 1A). We compared each mono-association to conventionally reared (CR) esg^ts/N^RNAi flies that contained a poly-microbial gut microbiota. Mono-association of esg^ts/N^RNAi flies with Lp resulted in few visible tumors (Fig. 1E). In contrast, mono-association with Lb caused multiple, large tumors throughout the posterior midgut (Fig. 1F), indicating that Lb is sufficient for tumor development.

We then quantified impacts of bacterial association on midgut tumors. First, we developed a four-point system to classify intestines, ranging from no visible defects (level I) to intestines with progenitor and enteroendocrine-rich tumors (level IV; Fig. S1D). In a blinded assay, we categorized 85% of CR esg^ts/N^RNAi intestines as level IV, whereas only 20% of GF intestines belonged to the same category (Fig. 1G), confirming bacterial effects on gut tumors. Consistent with our initial observations, GF and Lp-associated intestines had similarly mild levels of midgut tumors (Fig. 1G). In contrast, all intestines associated with Lb had level IV tumors within 5 days of Notch inactivation. To measure total tumor size per midgut, we quantified the posterior midgut area occupied by progenitor and enteroendocrine cells in the respective groups. Association with Lb significantly enhanced accumulation of progenitors and enteroendocrine cells in esg^ts/N^RNAi intestines compared with CR, GF or Lp mono-associated flies, supporting a role for Lb in promoting tumors (Fig. 1H). To determine if the larger tumor areas in Lb-associated flies are a result of increased tumor initiation or accelerated tumor growth, we quantified numbers of tumors in each intestine. Similar to our assessment of tumor size, association with Lb had a significant impact on tumor numbers, resulting in approximately three times as many tumors per gut as CR counterparts (Fig. 1I). Collectively, our data indicate that association with Lb increases the frequency of midgut tumor initiation.

To determine which factors from Lb promote tumors, we measured tumors in flies that we continuously fed heat-killed Lb or cell wall derived from Lb for 5 days. GF esg^ts/N^RNAi flies fed heat-killed Lb mixed with sterile food had similar tumor levels (Fig. 2A), and similar progenitor and enteroendocrine cell expansions as Lb mono-associated flies raised on sterile food (Fig. 2B), indicating that structural components of Lb are sufficient to promote tumors. To determine whether Lb cell wall mediated these effects, we measured tumors in GF flies fed sterile Lb extracts in PBS/sucrose on filter paper. Whereas esg^ts/N^RNAi GF flies fed sterile PBS/sucrose alone had low levels of intestinal tumors, flies fed Lb cell wall extract had significantly increased tumor levels (Fig. 2C), and accumulated progenitor and enteroendocrine cells to similar levels as GF counterparts fed dead Lb (Fig. 2D). Finally, we noticed that esg^ts/N^RNAi flies mono-associated with Lb died significantly faster than CR counterparts, whereas Lb did not shorten the lifespan of esg^ts/+ controls (Fig. S2), arguing that cell wall components from L. brevis promote initiation of Notch-deficient tumors, resulting in premature host death.

To determine how Lb affects Notch-deficient progenitors, we used RNA sequencing (RNA-Seq) to identify the transcriptional profiles of fluorescence-activated cell sorting (FACS)-purified, GFP-positive progenitors from Lb-associated esg^ts/+ and esg^ts/N^RNAi intestines (Fig. 3A). As controls, we sequenced transcriptomes of esg^ts/+ and esg^ts/N^RNAi progenitors from GF flies, or flies that we mono-associated with Lp. Principal component analysis (PCA) revealed that Notch-deficient progenitors segregate from wild-type progenitors along PC1, regardless of bacterial association (Fig. 3B). Differential gene expression analysis showed that the majority of gene expression profiles altered by Notch depletion were shared between GF, Lb-associated and Lp-associated intestines (Fig. 3C), indicating the existence of a microbe-independent core response to loss of Notch in progenitors. Gene ontology (GO) term analysis of the core Notch-deficient response revealed significant upregulation of biological processes involved in cell division (Fig. 3D), and diminished expression of Notch-responsive Enhancer of split [E(spl)] complex genes required for enteroblast differentiation (Fig. 3E). In addition to effects on growth and differentiation, we observed unexpected downregulation of immune pathway regulators in progenitors that lacked Notch (Fig. 3D,E). Decreased expression of immune regulators is not secondary to tumor development, as we saw similar changes in intestines of GF and Lp-associated flies. In each case, Notch inactivation diminished expression of essential components of the Immune Deficiency (IMD) pathway, such as imd, IKKβ and Rel, as well as prominent IMD response genes such as pirk, and multiple PGRP family members (Fig. 3E). These data suggest a genetic link between Notch signaling and immune responses in the progenitor compartment, and match previous reports that mutation or activation of the IMD pathway alters expression of Notch pathway genes in the fly intestine (Broderick et al., 2014; Petkau et al., 2017).

As tumor growth frequently involves shifts in microbiota composition, we determined effects of Notch inactivation on host association with Lb and Lp. First, we measured the intestinal bacterial load of esg^ts/+ and esg^ts/N^RNAi flies that we mono-associated with the respective strains for 5 days. In these experiments, we quantified bacterial load in the same cohort of flies that we used to measure tumors in Fig. 1, allowing us to determine whether host-microbe associations correlate with midgut tumors. For Lb and Lp, we observed significantly increased bacterial loads in Notch-deficient intestines compared with wild-type controls (Fig. 4A), suggesting effects of host genotype on bacterial association. Importantly, there were no differences between Lp or Lb loads in Notch-deficient guts. Thus, the identity of associated bacteria, not the abundance, determines tumors in the host.

As bacterial association is higher in Notch-deficient intestines than in wild-type intestines (Fig. 4A), and Notch inactivation diminishes expression of IMD pathway components, we investigated whether IMD affects host association with Lb. Consistent with a role for IMD in the control of intestinal Lb, we found that imd mutants had significantly higher Lb loads than did wild-type controls 10 days after mono-association with Lb (Fig. 4B). Conversely, constitutive activation of IMD in enterocytes (Myo1A^ts;ImdCA) reduced Lb load to approximately 4% of that found in imd mutants (Fig. 4B). These data support a role for IMD in regulation of intestinal Lb. However, it is important to note that the increased bacterial abundance in imd mutants is considerably less pronounced than increases observed upon Notch depletion (compare Fig. 4A and 4B). Thus, we believe that additional, as-yet-unidentified, IMD-independent mechanisms control bacterial numbers in esg^ts/N^RNAi intestines.

Finally, we measured the effects of Notch inactivation on host association with Lactobacillus commensals. Here, we completed a longitudinal measurement of bacterial load in intestines of esg^ts/+ and esg^ts/N^RNAi flies that we mono-associated with Lp or Lb. In general, our data match earlier reports that total numbers of intestinal bacteria in flies increase with age (Clark et al., 2015; Guo et al., 2014). In wild-type esg^ts/+ intestines, the rates of increase in host-association with Lp and Lb are nearly indistinguishable, with Lb associating to lower levels at all times tested (Fig. 4C). Initially, Lb also associated with esg^ts/N^RNAi intestines to lower levels than Lp. However, we noted substantial effects of Notch inactivation on subsequent progressions in host-microbe association. In this case, association with Lp increased at a considerably slower rate than association with Lb (Fig. 4D). Exponential regression analysis revealed that host-associated Lb loads double at similar rates in intestines of esg^ts/+ (0.87 days) and esg^ts/N^RNAi (0.78 days) flies (Fig. 4C,D). In contrast, host-associated-Lp loads double at a considerably slower rate in N-deficient intestines (2.54 days; Fig. 4D), than wild-type intestines (0.97 days; Fig. 4C), suggesting that Notch knockdown hinders host association with Lp, but has minimal effects on association with Lb.

As Lb grows effectively in Notch-deficient intestines, where it promotes tumors, we reasoned that Lb will have distinct division-enhancing effects on progenitor cells. To test this hypothesis, we looked for progenitor cell transcriptional events that were specific to association with Lb. PCA and differential expression analysis identified a transcriptional response that is unique to Lb in wild-type and Notch-deficient progenitors (Fig. 3B; Fig. S3A,B). Regardless of host genotype, association with Lb specifically increased the expression of genes required for cell division, such as those involved in DNA replication and mitotic spindle organization (Fig. 5A), as well as prominent cell cycle and growth regulators (Fig. 5B), consistent with growth-promoting effects of gut-associated bacteria in Drosophila (Broderick et al., 2014; Buchon et al., 2009b). In addition to effects on cell cycle and growth pathways, we noticed a particularly striking inhibitory effect of Lb on the expression of genes involved in cell-cell adhesion, cell-matrix adhesion and cell polarity (Fig. 5A), especially genes that encode integrin complex proteins (Fig. S3C). For example, association with Lb led to diminished expression of the α- and β-integrins scab and myospheroid (mys), the talin ortholog rhea, and the integrin extracellular matrix ligand LanA (Fig. 5B). The effects of Lb on expression of genes associated with stem cell adhesion were independent of host genotype, as we observed the same phenotypes in progenitors of Lb-associated esg^ts/+ and esg^ts/N^RNAi flies (Fig. 5B).

Given the positive effects of Lb on expression of ISC division regulators in esg^ts/+ and esg^ts/N^RNAi progenitors, we asked if Lb activates ISC division in wild-type progenitors. To answer this question, we mono-associated GF wild-type (esg^ts/+) flies with Lb and quantified phospho-histone 3-positive (PH3+) mitotic cells in adult midguts. Similar to the effects on tumors, Lb stimulated division of wild-type progenitors to significantly higher levels than those of CR, GF or Lp-mono-associated flies (Fig. 5C). Thus, our data indicate that association with Lb diminishes the expression of genes required for progenitor adhesion to the extracellular matrix, and induces expression of genes required for epithelial growth, promoting ISC division in wild-type and Notch-deficient progenitors.

We were particularly intrigued by effects of Lb on expression of integrins that anchor progenitors within the niche. Therefore, we asked what effects Lb has on progenitor cell adhesion and morphology. In a preliminary experiment, we used transmission electron microscopy to visualize posterior midguts of CR and Lb-associated wild-type flies. CR intestines contained basal progenitors in close association with the extracellular matrix (Fig. 6A,B; Fig. S4A,B for additional examples). Mono-association with Lb appeared to disrupt intestinal organization, generating round progenitors that shifted apically relative to the extracellular matrix, and lacked discernible contact with larger enterocytes or extracellular matrix (Fig. 6C,D; Fig. S4C,D for additional examples). These morphological changes appear to be specific to progenitors, as no defects were apparent in the shape, or relative position, of surrounding enterocytes. The apparent shift in progenitor localization in Lb-associated intestines prompted us to ask if Lb modifies integrin distribution. To answer this question, we determined the subcellular localization of the β-integrin Mys in sagittal sections of GF intestines, or intestines that we associated with Lb. In GF esg^ts/+ flies, we detected basolateral enrichment of Mys in GFP-positive progenitors (Fig. 6E). In contrast, and similar to our electron microscopy results, we found that Lb colonization caused progenitors to ‘round up’ and adopt a more apical position within the epithelium (Fig. 6F). Furthermore, association with Lb had visible impacts on Mys localization, characterized by discontinuous basolateral distribution, and atypical apical enrichment of Mys (Fig. 6F, arrowheads). To measure effects of Lb on subcellular distribution of integrins directly, we developed an immunofluorescence-based assay that allowed us to quantify apical:basolateral ratios of Mys in progenitors (Fig. S5). With this assay, we detected basal enrichment of Mys in GF progenitors (Fig. 6H). Association with Lb shifted the distribution of Mys, resulting in significant increases in apical Mys (Fig. 6H). To determine whether Lb-dependent effects on integrin subcellular distribution are downstream consequences of ISC division, we blocked division in progenitors of Lb-associated flies by expressing the cell cycle inhibitor dacapo (esg^ts/dap) (Fig. 6I). Notably, when we examined division-impaired midguts, we found that Lb continued to cause increases in apical Mys (Fig. 6G,H), indicating that Lb alters apicobasal integrin distribution independently of ISC divisions.

Loss of intestinal integrins results in aberrant stem cell divisions with substantial effects on organization of the progenitor compartment (Goulas et al., 2012). Therefore, we investigated the effect of Lb on midgut progenitor cells. We first stained intestines of GF flies, or flies that we mono-associated with Lb for the ISC marker Delta (Fig. 7A,B). Compared with GF intestines, Lb association significantly increased the proportion of Delta⁺ cells within the esg⁺ progenitor pool (Fig. 7C). In support of an effect of Lb on ISCs we also noted that association with Lb increased expression of genes involved in stem cell identity, maintenance and differentiation, such as Dl and E(Spl) complex genes (Fig. 7D). To better understand effects of Lb on the progenitor compartment, we quantified marker expression in midguts of esg^ts, UAS-CFP, Su(H)-GFP flies that we raised under conventional conditions, germ-free conditions, or mono-associated with Lb (Fig. 7E-G). In this line, stem cells that express the progenitor cell marker esg are visible as CFP single-positive cells. In contrast, progenitors that express the enteroblast marker Su(H), are visible as CFP/GFP double-positive cells. We found that CR flies had approximately equal numbers of Su(H)-positive and Su(H)-negative progenitors, suggesting a 1:1 distribution of stem cells and enteroblasts in midguts of conventional flies (Fig. 7H). Removal of the microbiome increased the proportion of Su(H)⁺ cells, whereas mono-association with Lb had the opposite effect (Fig. 7H). Specifically, we measured a significant decrease in the proportion of Su(H)⁺ cells within the esg⁺ population of Lb-associated flies (Fig. 7H). Combined with quantification of Dl⁺ stem cells (Fig. 7A-C), our data indicate that compared with CR or GF flies, Lb increases the proportion of stem cells relative to enteroblasts within the progenitor compartment.

We did not see elevated levels of cell death (Fig. S6A) or increased expression of apoptosis regulators (Fig. S6B) within progenitors of Lb mono-associated flies compared with CR or GF controls. Thus, we do not believe that Lb affects cell-type composition within the progenitor compartment by preferentially promoting enteroblast death (Reiff et al., 2019); however, we cannot rule out the possibility that Lb affects relative enteroblast numbers by increasing the rate of terminal differentiation. As an alternative, we tested the hypothesis that Lb increases stem cell proportions by promoting symmetric stem cell divisions. For this assay, we used twin-spot mosaic analysis with repressible cell markers (MARCM) to visualize and quantify symmetric and asymmetric divisions in intestines of CR, GF and Lb mono-associated flies. In this fly line, a stem cell division differentially labels daughter cells with heritable GFP and RFP markers (O'Brien et al., 2011; Yu et al., 2009). In an asymmetric division, the marked enteroblast then differentiates into a single marked enterocyte, whereas the stem cell undergoes rounds of division and differentiation that generate a multi-lineage clone of cells that bear the opposing marker (Fig. 7I). In contrast, symmetric divisions generate sister cells of the same identity; ultimately producing neighboring clones of approximately equal size (Fig. 7J). To measure the effects of Lb on division symmetry, we induced recombination immediately after mono-association. We then allowed marked clones to develop for 8 days prior to visualization. We scored asymmetric stem cell divisions as clones in which a single cell of one color resided next to multiple cells of another color (Fig. 7I). In contrast, symmetric stem cell divisions labeled approximately equal numbers of intestinal epithelial cells with their respective markers (Fig. 7J). We rarely spotted symmetric divisions that generated two enteroblasts. Instead, most symmetric clones contained large and small nuclei, regardless of the associated microbes, confirming that symmetric ISC divisions generated multi-lineage clones in each case (Fig. 7J). Consistent with earlier reports (Hu and Jasper, 2019; Jin et al., 2017; O'Brien et al., 2011), approximately 23% of all clones were products of symmetric divisions in CR flies (Fig. 7K). Removal of the microbiome decreased the percentage of clones with symmetric signatures (Fig. 7K), indicating that gut bacteria promote symmetric stem cell divisions in conventional flies. Importantly, mono-association with Lb significantly increased the frequency of multi-lineage symmetric clones compared with either GF or CR flies (Fig. 7K). Thus, our data indicate that association with Lb promotes symmetric stem cell divisions within the adult midgut.

Excess microbiota-responsive growth supports development of inflammatory diseases, and hyperplastic expansion of cells that bear oncogenic lesions. To understand how gut bacteria cause progenitor dysplasia, we measured growth in adult Drosophila intestines that we mono-associated with common Lactobacillus commensals. We focused on L. brevis and L. plantarum, as they have established roles in Drosophila intestinal homeostasis (Combe et al., 2014; Fast et al., 2018; Iatsenko et al., 2018; Jones et al., 2013, 2015; Lee et al., 2013; Reedy et al., 2019; Storelli et al., 2011). In general, our observations match literature that highlight distinct, context-dependent effects of Lactobacillus commensals on juvenile growth, intestinal physiology and adult longevity. We identified L. brevis as a potent stimulator of intestinal tumors. Interestingly, L. plantarum did not promote tumor growth even though it grew to similar levels in the intestine as Lb, suggesting that the strain of Lp used in this study either fails to stimulate, or actively inhibits, ISC division. We do not fully understand how the host distinguishes Lactobacillus species, although bacterial metabolites are interesting candidates. For example, Lp promotes progenitor growth in larvae by stimulating reactive oxygen species (ROS) generation in enterocytes (Jones et al., 2013, 2015; Reedy et al., 2019). Likewise, Lb-derived uracil, a noxious agent that promotes tissue damage and repair, activates epithelial generation of ROS in adults (Lee et al., 2013). In addition, cell walls of Lp contain DAP-type peptidoglycan that activates the IMD pathway with substantial effects on transcription (Broderick et al., 2014; Lesperance and Broderick, 2020). In the future, it will be of interest to test the relationships between microbe-specific immune responses, ROS production and epithelial growth. As poly-bacterial communities have distinct effects on host phenotypes (Gould et al., 2018), it will also be of interest to investigate the effects of interactions among Lactobacillus species on epithelial growth and tumorigenesis.

Quantification of host-associated bacteria suggests that physiological disruptions associated with Notch deficiency promote accumulation of Lb in the gut, which fuels continued growth of Notch-deficient tumors. A similar feed-forward loop was described recently in a BMP-deficient tumor model (Zhou and Boutros, 2020), suggesting a conserved relationship between tumor growth and microbiota expansion in flies. Our transcriptional data raise the possibility that inactivation of Notch partially increases Lb loads by suppressing expression of IMD pathway regulators. However, given the massive increase in bacterial numbers observed upon Notch knockdown compared with the more moderate increases observed associated with imd mutants, we hypothesize that additional mechanisms of Notch-dependent bacterial control exist. As Notch is required for intestinal differentiation, we suggest that a mis-differentiated epithelium is incapable of fully limiting bacterial accumulation in the intestine. For example, Notch inactivation changes the composition, and possibly also gene expression, of differentiated cells within the midgut. Furthermore, Notch controls differentiation of copper cells (Wang et al., 2014), a specialized cell type that attenuates bacterial growth by establishing a stomach-like region of low pH (Li et al., 2016). Thus, we speculate that Notch-deficient intestines are impaired in their ability to eradicate bacteria due to a multifactorial network of perturbed immunity and mis-differentiation of mature epithelial cells.

To determine how Lb affects intestinal proliferation, we characterized transcriptional responses from midgut progenitors of flies inoculated with Lb. We observed significant effects of Lb on expression and subcellular localization of integrins, crucial regulators of stem cell niche interactions (Ellis and Tanentzapf, 2010; Fernandez-Minan et al., 2007; Marthiens et al., 2010; Toyoshima and Nishida, 2007) and stem cell maintenance (Goulas et al., 2012; Lin et al., 2013; Okumura et al., 2014; You et al., 2014). Typically, integrins accumulate at basolateral margins of intestinal progenitors, and anchor interphase progenitors to the extracellular matrix. In many tissues, including the fly intestine, integrins organize the stem cell division plane, by orienting the mitotic spindle. Progenitors mainly divide at angles greater than 20° to the basement membrane, leading to asymmetric divisions (Hu and Jasper, 2019; Ohlstein and Spradling, 2007), after which basal daughter cells remain in the niche and retain stemness, whereas apical daughters exit the niche and differentiate. Approximately 20-40% of divisions in the young adult intestine under homeostasis occur symmetrically, yielding clonal lineages of stem cells or enteroblasts (Hu and Jasper, 2019; Jin et al., 2017; O'Brien et al., 2011). Over time, enteroblast clones differentiate into mature enterocytes that eventually die. However, clonal stem cell lineages retain the capacity to grow and establish regional dominance within the epithelium. In some cases, symmetric divisions facilitate adaptive responses to environmental fluctuations allowing the intestinal environment to tune proliferative needs to extrinsic factors. For instance, rapid changes in nutrient availability or ingestion of toxic doses of paraquat increase the frequency of symmetric divisions and expand the stem cell pool, allowing for a rapid regenerative response (Hu and Jasper, 2019; O'Brien et al., 2011). We discovered that Lb disrupts subcellular distribution of integrins in progenitors and promotes symmetric expansion of ISCs. Lb-mediated ISC growth likely accounts for the tumor burden noted in Notch-deficient flies inoculated with Lb. However, as individual stem cell lineages accumulate mutations with age, we consider it possible that exposure to Lb also facilitates spontaneous development of intestinal tumors in wild-type adult flies. Because our focus was on intestinal progenitors it is unclear whether Lb also alters integrins in mature epithelial cells. As integrin loss in enterocytes causes delamination and subsequent stress-induced tumor growth (Patel et al., 2015), it is important to understand whether the actions of Lb on integrins is specific to progenitors or if enterocytes are also affected.

How Lb disrupts integrins and promotes stem cell growth requires clarification. As stem cells derive cues from the surrounding epithelium to direct their growth, we consider it likely that mature epithelial cells, such as enterocytes, sense L. brevis and transduce signals to ISCs to promote growth. Our transcriptional profiling of progenitors suggests that, in addition to changes in integrin expression, Lb induces transcriptional events indicative of Hippo inhibition and JAK-STAT activation, pathways known to mediate regenerative proliferation in response to intestinal damage and the microbiome (Broderick et al., 2014; Buchon et al., 2009b; Ren et al., 2010; Shaw et al., 2010). Therefore, we speculate that immune activation and damage of mature epithelial cells by Lb induces growth signals that promote stem cell division. Along these lines, Lp and Lb induce ROS production in enterocytes, which damages the intestine leading to increased intestinal mitoses (Jones et al., 2013, 2015; Lee et al., 2013; Reedy et al., 2019). Interestingly, the cell wall of Lb is sufficient for tumor growth, suggesting a role for bacterial recognition pathways in intestinal growth. Consistent with this hypothesis, constitutive activation of the IMD pathway promotes growth of Notch-deficient tumors (Petkau et al., 2017), and chronic inflammation is a risk factor for development of colorectal cancer (Kim and Chang, 2014). Based on these observations, we speculate that Lb activates epithelial ROS and IMD, resulting in damage and subsequent regenerative responses that promote ISC division and fuel tumorigenesis. In parallel with or downstream of epithelial stress, Lb decreases integrin expression in ISCs, promoting symmetric stem cell divisions that expand the numbers of stem cells and increase the regenerative capacity of the wild-type intestine as a mechanism of adaptive growth. Because of the evolutionary conservation of intestinal homeostatic regulators, we believe Drosophila will be a fruitful model for determining how gut-resident bacteria influence intestinal progenitor function.

Drosophila stocks and crosses were set up and maintained at 18-25°C on standard corn meal food (Nutri-Fly Bloomington formulation; Genesse Scientific) with a 12:12 light:dark cycle. All experimental flies were virgin female flies except where noted. Upon eclosion, flies were kept at 18°C then shifted to the appropriate temperature once 25-30 flies per vial was obtained. GF and mono-associated flies were generated as previously described (Fast et al., 2018). To generate GF flies, freshly eclosed flies were fed autoclaved food with antibiotics (100 µg/ml ampicillin, 100 µg/ml neomycin, 50 µg/ml vancomycin and 100 µg/ml metronidazole) for 5 days at 25°C. Conventionally reared controls were fed autoclaved food without antibiotics for 5 days at 25°C. To generate mono-associated animals, flies were made germ free as above then were fed 1 ml OD₆₀₀=50 of L. brevis or L. plantarum re-suspended in sterile 5% sucrose/PBS on a cotton plug overnight at 25°C. During this overnight feeding, CR and GF controls were fed sterile 5% sucrose/PBS without bacteria. The following morning, CR, Lb and Lp conditions were transferred to fresh autoclaved food at 29°C for the remainder of the experiment, whereas GF flies were maintained on autoclaved food with antibiotics. Sterility and mono-association were confirmed in each experimental vial by plating fly homogenate or fly food on MRS. Vials found to be contaminated were discarded from the experiment. Fly lines used in this study were: w;esg-GAL4,tubGAL80^ts,UAS-GFP (esg^ts), UAS-N^RNAi (VDRC ID#100002), w;Myo1A-GAL4;tubGAL80^ts,UAS-GFP (Myo1A^ts), imd, w¹¹¹⁸, UAS-ImdCA (Petkau et al., 2017), w;esg-GAL4,UAS-CFP, Su(H)-GFP;tubGal80^ts (esg^ts,UAS-CFP,Su(H)-GFP). Twin-spot MARCM flies were generated by crossing together hsFLP;FRT40A,UAS-CD2-RFP,UAS-GFP-miRNA/cyo flies and FRT40A,UAS-CD8-GFP,UAS-CD2-miRNA/cyo;tubGAL4/TM6C flies to create hsFLP;FRT40A,UAS-CD2-RFP,UAS-GFP-miRNA/FRT40A,UAS-CD8-GFP,UAS-CD2-miRNA;tubGAL4/+. Twin-spot clones were induced by a 37°C heat shock for 1 h immediately after transferring flies to fresh food after the overnight mono-association protocol. Flies were then shifted back to 29°C for 8 days before dissecting and counting clones from 30 intestines per treatment. The imd and UAS-ImdCA lines had been backcrossed into our wild-type w¹¹¹⁸ background.

L. brevis^EF (DDBJ/EMBL/GenBank accession LPXV00000000) and L. plantarum^KP (DDBJ/EMBL/GenBank chromosome 1 CP013749 and plasmids 1-3 CP013750, CP013751 and CP013752) were both isolated from our Drosophila lab stocks and have been previously described (Petkau et al., 2016). Both bacteria were streaked out on MRS plates (BD Difco, 288210) and aerobically grown at 29°C. Single colonies were picked for growth in MRS broth (Sigma-Aldrich, 69966) at 29°C (L. brevis for 2 days and L. plantarum for 1 day). To generate dead L. brevis, liquid culture was spun down, washed twice with sterile water then re-suspended in sterile water before heating to 95°C for 30 min. After heating, the killed bacteria was spun again and re-suspended to 10 mg/ml in sterile 5% yeast, 5% sucrose in PBS.

To extract the cell wall, L. brevis was heat-killed as above, cooled on ice, then run through a French Press at 20,000 psi three times to lyse the bacterial cells. After lysis, any remaining whole cells were collected and discarded by two successive spins at 2000 g. To collect the cell wall, the supernatant was spun at 10,000 g for 30 min, washed twice with 1 M NaCl and twice with sterile water before re-suspending in sterile 5% yeast, 5% sucrose in PBS. GF flies were fed a 10 mg/ml cell wall solution on filter paper disks on top of sterile food alongside 10 mg/ml dead L. brevis and sterile 5% yeast, 5% sucrose PBS without any L. brevis extracts. Dead bacteria and cell wall were continuously fed to flies during the experiment, with fresh extracts provided every second day. Sterility of dead Lb and cell wall was confirmed by plating 100 μl of extract on MRS.

Intestines were dissected in PBS, fixed in 4% formaldehyde for 20 min then blocked overnight at 4°C in 5% normal goat serum (NGS), 1% bovine serum albumin (BSA) and 0.1% Tween-20. Washes were performed in blocking solution without NGS. Primary and secondary antibody incubations were carried out for 1 h at room temperature in blocking buffer without NGS. For Delta and PH3 stains, we used a revised protocol in which 8% formaldehyde was used to fix, washes were performed in PBS with 0.2% Triton X-100 (PBST) and blocked in PBST with 3% BSA. To prepare the intestinal sections, the posterior midgut was extracted, flash frozen on dry ice in frozen section compound (VWR, 95057-838) and sectioned at 10 μm thickness; slides were stained using the same parameters as whole guts. Primary antibodies used were: anti-Prospero [1/100; MR1A, Developmental Studies Hybridoma Bank (DSHB)], anti-Mys (1/100; CF.6G11, DSHB), anti-GFP (1/1000; G10362, Invitrogen), anti-phospho-histone3 (1/1000; 06-570, Millipore), anti-Delta (1/100; C594.9B, DSHB). Secondary antibodies used were: Alexa Fluor 568-conjugated goat anti-mouse (1/500; A11004, Invitrogen) and Alexa Fluor 488-conjugated goat anti-rabbit (1/500; A11008, Invitrogen). DNA stains used were: Hoechst 33258 (1/500; H-3569, Molecular Probes), DRAQ5 (1/500; 65-0880-96, Invitrogen). Apoptotic cells were detected in dissected guts using the TMR red In Situ Cell Death Detection Kit (Roche, 12156792910) following the standard kit staining protocol. GFP primary antibody was used for the terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) experiment and cryosectioning experiments to retain GFP signal from the esg promoter. Intestines were mounted on slides using Fluoromount (Sigma-Aldrich, F4680). For every experiment, images were obtained of the posterior midgut region (R4/5) of the intestine with a spinning disk confocal microscope (Quorum WaveFX). PH3⁺ cells were counted through the entire midgut. To determine the apical and basal Mys intensity, we drew a line of 10-pixel width from the basal side to the apical (lumen side) side across GFP⁺ progenitor cells. We defined apical and basal progenitor cell borders as 50% of the maximum GFP intensity, as this GFP intensity coincides with the basal Mys intensity peak. We determined the intensity of GFP and Mys across the progenitors using the function plot profiles, copied these values into Excel and determined the apical and basal Mys intensities. All image stacking, intensity and area calculations were performed using Fiji software (Schindelin et al., 2012).

Progenitor cell isolation by FACS was adapted from previously described protocols (Dutta et al., 2013). In brief, three biological replicates consisting of 100 fly guts per replicate with the Malphigian tubules and crop removed were dissected into DEPC PBS and placed on ice. Guts were dissociated with 1 mg/ml of elastase at 27°C with gentle shaking and periodic pipetting for 1 h. Progenitors were sorted based on GFP fluorescence and size with the BD FACSAria III sorter. All small GFP-positive cells were collected into DEPC PBS. Cells were pelleted at 1200 g for 5 min and then re-suspended in 500 μl of Trizol. Samples were stored at −80°C until all samples were collected. RNA was isolated via a standard Trizol chloroform extraction and the RNA was sent on dry ice to the Lunenfeld-Tanenbaum Research Institute (Toronto, Canada) for library construction and sequencing. The sample quality was evaluated using Agilent Bioanalyzer 2100. TaKaRa SMART-Seq v4 Ultra Low Input RNA Kit for Sequencing was used to prepare full length cDNA. The quality and quantity of the purified cDNA was measure with Bioanalyzer and Qubit 2.0. Libraries were sequenced on the Illumina HiSeq3000 platform.

On average, we obtained 30 million reads per biological replicate. FASTQC was used to evaluate the quality of raw paired-end sequencing reads (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/ version 0.11.3). Adaptors and reads of fewer than 36 base pairs in length were trimmed from the raw reads using Trimmomatic (version 0.36) (Bolger et al., 2014). Reads were aligned to the Drosophila transcriptome bdgp6 with HISAT2 (version 2.1.0) (Kim et al., 2015). The resulting BAM files were converted to SAM files using Samtools (version 1.8) (Li et al., 2009). The converted files were counted using Rsubread (version 1.24.2) (Liao and Smyth, 2019) and loaded into EdgeR (version 3.16.5) (Robinson et al., 2010). In EdgeR, genes with counts fewer than 1 count per million were filtered and libraries were normalized for size. Normalized libraries were used to call genes that were differentially expressed among treatments. Genes with P<0.01 and FDR<5% were defined as differentially expressed.

PCA was performed on normalized libraries using Factoextra (version 1.0.5). Gene Ontology Enrichment Analysis and Visualization Tool (GORILLA) was used to examine GO term enrichment (Eden et al., 2009). Specifically, differentially expressed genes (defined above) were compared in a two-list unraked comparison to all genes output from edgeR as a background set. Redundant GO terms were removed.

Five flies were randomly selected from a single vial of flies for each biological replicate and surface sterilized by washing in 10% bleach and 70% ethanol. Flies were then homogenized in MRS, serially diluted and 10 µl of each dilution was plated on MRS. Colonies were counted from 10 µl streaks that had 10-200 colonies and the CFU/fly calculated.

Intestines were dissected from virgin female flies that had been kept at 29°C for 8 days following germ-free and bacterial-association protocols. Posterior midguts were excised and fixed with 3% paraformaldehyde with 3% glutaraldehyde. Fixation, contrasting, sectioning and visualization were performed at the Faculty of Medicine and Dentistry Imaging Core at the University of Alberta. Midgut sections were visualized with Hitachi H-7650 transmission electron microscope at 60 Kv in high contrast mode.

Virgin female flies were mono-associated with L. brevis or raised with a conventional microbiome as described. After mono-association, flies distributed to sterile vials with autoclaved food at a density of ten flies/vial were shifted to 29°C for the remainder of the experiment. Dead flies were counted every 1-3 days and vials were flipped three times per week to fresh autoclaved food.

Figures were constructed using R (version 3.3.1) via R studio (version 1.1.442) with easyggplot2 (version 1.0.0.9000), with the exception of GO term figures and line plots for which ggplot2 (version 3.0.0) was used. For all boxplots, box represents first and third quartiles, center line represents the median. Lower whisker represent smallest observation greater than or equal to 25% quantile − 1.5 * inter-quartile range. Upper whisker represent largest observation less than or equal to 75% quantile + 1.5 * inter-quartile range. Longevity graphs were made in Prism software along with the stats. All other statistical analysis was performed in R. Figures were assembled in Adobe Illustrator.

We thank Dr Bruce Edgar, Dr Bruno Lemaitre and Dr Lucy O'Brien for providing fly lines. We thank Dr Sarah Hughes and Dr Kristi Baker at the University of Alberta for proofreading support. We acknowledge microscopy support from Dr Steven Ogg, Gregory Plummer and Woo Jung Cho at the Faculty of Medicine and Dentistry Imaging core; flow cytometry support from Dr Aja Rieger at the Faculty of Medicine and Dentistry Flow Cytometry Core; and cryosectioning support from Lynette Elder at the Alberta Diabetes Institute Histocore. The authors wish to thank Kin Chan at the Network Biology Collaborative Centre (https://nbcc.lunenfeld.ca/) for the RNA-Seq service. Network Biology Collaborative Centre is a facility supported by Canada Foundation for Innovation, the Ontarian Government, and Genome Canada and Ontario Genomics (OGI-139).