Zrf1 is necessary for ISC proliferation during homeostasis and regeneration

To identify intrinsic factors essential for ISC proliferation, we performed an RNAi screen of candidate genes biased towards DNA-binding proteins enriched in intestinal progenitors [3,12,13]. Utilizing the ISC-specific gene expression driver Escargot (esg)-Gal4 combined with the TARGET system (tub-Gal80^ts) [14] and UAS-GFP, hereafter referred to as egt^ts, we selectively drove UAS-RNAi in adult ISC/EBs. Progenitors were visualized using UAS-GFP, and anti-phospho-histone 3 (pH3) staining served as a marker for ISC mitosis. Among ~120 RNAi lines targeting ~73 genes, three independent RNAi transgenes targeting CG10565 (hereafter referred to as Zrf1) significantly reduced ISC proliferation during homeostasis and in response to damage-inducing diets (Figs S1Aand 1A-1C). Zrf1 knockdown also resulted in fewer Delta-LacZ-positive and GFP-positive cells (Fig 1D, 1E, 1J and 1K). Lineage tracing using the egt^tsF/O system [15] revealed that progenitors failed to divide following Zrf1 RNAi knockdown (Figs 1F, 1G, 1J, S1B and S1C). CRISPR/Cas9-mediated knockout (egt^ts > Cas9.P2) of Zrf1 in ISC/EBs produced a similar phenotype (Fig 1H-1J). Importantly, the reduction in ISC proliferation was not a result of cell death since blocking apoptosis, by overexpressing either DIAP or p35 in progenitors, could not rescue this defect (Fig 1C). The proliferation in response to damage in ISCs was unaffected when Zrf1 was downregulated in adult enteroblasts (Su(H)^ts) or in EEs (pros^ts) (Fig 1L). Regeneration of ISCs was only impaired when Zrf1 was knocked down specifically in ISCs (egt^ts + Su(H)-Gal80 or Dl^ts) (Figs 1L and S1E). Altogether, these results indicate that Zrf1 is cell-autonomously required for the proliferation of ISCs during homeostasis and regeneration.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Genetic screen to identify Zrf1 as a regulator of ISC proliferation during homeostasis and in response to damage.

(A) Box plot showing ISC mitoses per midgut in egt^ts > GFP flies after 10 days of RNAi induction targeting candidate genes enriched in intestinal progenitors. Hits from the screen are listed with their human orthologs and DIOPT scores. Zrf1 RNAi lines (HMS00624, KK105682) led to reduced ISC mitosis compared to the control (Luciferase, Luc). Numbers below each box plot indicate the number of midguts analyzed. See also S1 Data Table A. (B) Quantification of Zrf1 transcript levels in midguts after 10 days of Zrf1 knockdown in midgut progenitors, showing reduced expression relative to the housekeeping gene RpL32 compared to controls. See also S1 Data Table B. (C) Box plot showing ISC mitoses per midgut after Zrf1 knockdown during homeostasis and following treatment with damage-inducing agents (bleomycin, paraquat, Ecc15). ISC mitoses are also shown following co-expression of anti-apoptotic factors (DIAP or p35). Numbers below each plot indicate the number of midguts analyzed. See also S1 Data Table C. (D–H) Confocal images of midguts from egt^ts flies treated with bleomycin. (D) Control (Luciferase, Luc). (E) Zrf1 knockdown showing reduced ISC proliferation. (F–G) Lineage tracing using egt^tsF/O system: (F) Control (Luc) and (G) Zrf1 knockdown. (H) CRISPR/Cas9-mediated knockout of Zrf1 (egt^ts > Cas9.P2) showing similar reduction in ISC division. GFP marks progenitor cells; DAPI marks nuclei. Scale bar: 50 µm. See also S1 Data Table D. (J) Box plot quantifying the percentage of GFP+ cells from experiments after Zrf1 knockdown and control (Luc) in midgut progenitors. (K) Quantification of Dl-LacZ positive cells (ISC) after midgut-progenitor-specific Zrf1 knockdown in comparison to control. See also S1 Data Table E. (L) Box plots showing ISC mitoses per midgut after Zrf1 downregulation in specific gut cell types: ISCs (egt^ts + Su(H)-Gal80), enteroblasts (Su(H)^ts) and enteroendocrine cells (pros^ts). Midguts were treated with bleomycin. Numbers below each box plot indicate the number of midguts analyzed. See also S1 Data Table F. Box plots show median, quartiles, and range. Statistical significance determined by Student’s t-test; ns = not significant, *p < 0.05, **p < 0.01, ***p < 0.001 ****p < 0.0001.

https://doi.org/10.1371/journal.pgen.1011910.g001

Zrf1 promotes ISC proliferation in response to damage

To determine whether Zrf1 is sufficient to promote ISC proliferation, we tested transgenes for effective Zrf1 overexpression. We independently drove overexpression of four transgenes in ISC/EBs. These included two EP-like insertions, P{EP}G4964 and EP{EPgy2}Y14386, which both consist of a UAS sequence inserted into the 5’ region upstream of the endogenous Zrf1 gene as well as two UAS-FlyORF lines (Zrf1^ORF.3xHA and Zrf1^ORF.CC). Overexpression of all Zrf1 constructs resulted in a 3-to-5-fold increase in Zrf1 levels, with the exception of P{EP}G4964 (Fig 2A). Interestingly, upregulating Zrf1 in progenitors only led to a moderate increase in ISC proliferation in response to damage but not during homeostasis (Fig 2B). Furthermore, CRISPR/Cas9-mediated transcriptional activation (egt^ts > Cas9.VPR) of Zrf1 also led to a damage-specific increase in ISC proliferation (Fig 2C and 2D). These results suggest that Zrf1 is sufficient to increase ISC proliferation only in response to damage.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Zrf1 promotes ISC proliferation in response to damage.

(A) Relative Zrf1 expression levels in midguts from egt^ts > GFP flies after 10 days of overexpression using different transgenes. Expression was measured relative to the housekeeping gene RpL32. Overexpression of Zrf1 (EPgy2Y14386, ORF.3xHA, and ORF.CC) resulted in increased Zrf1 transcript levels compared to the control (wild type, w¹¹¹⁸). See also S1 Data Table I. (B) ISC mitoses per midgut in egt^ts > GFP flies overexpressing Zrf1 transgenes (EPgy2Y14386, ORF.3xHA, ORF.CC) during homeostasis and after bleomycin treatment (bleo). Number of midguts analyzed is shown below each box plot. See also S1 Data Table J. (C) Quantification of ISC mitoses in whole midguts of egt^ts > GFP + dCas9-VPR flies, comparing sg-Zrf1 to controls (attP40, sg-intergenic) under homeostasis and bleomycin treatment conditions. See also S1 Data Table K. (D) Quantification of GFP+ progenitor cells in the anterior midgut (AMG) of egt^ts > GFP + dCas9-VPR flies, showing the effect of sg-Zrf1 compared to sg-intergenic controls. See also S1 Data Table L. (E–H) Confocal images of ISC/EB progenitors in egt^ts > GFP flies overexpressing fly Ras1A or human gain-of-function Raf (hRaf^gof) alone or in combination with Zrf1 knockdown. Knockdown of Zrf1 rescued ISC overproliferation induced by activated Ras1 and hRaf. GFP marks progenitor cells; DAPI marks nuclei. Scale bar: 50 µm. (I) ISC mitoses per midgut after 5 days of ectopic activation of the EGFR/Ras/Raf signaling components (Egfr^TOP4.2, Ras1A, Raf^F149, hRaf^gof) in egt^ts > GFP flies with and without Zrf1 knockdown (Zrf1HMS00624). See also S1 Data Table M. (J) ISC mitoses per midgut after 8 days of Notch knockdown in egt^ts > GFP flies with and without Zrf1 knockdown. See also S1 Data Table N. (K) ISC mitoses per midgut after 5 days of overexpressing wildtype Yorkie (Yki^WT) or constitutively active Yorkie (Yki^3SA) in egt^ts > GFP flies with and without Zrf1 knockdown. See also S1 Data Table O. (L) Growth curves of S2R+ cells over a 6-day period showing cell proliferation after dsRNA mediated knockdown of Myc (dsMyc), Zrf1 (dsZrf1), and LacZ (dsLacZ, control). See also S1 Data Table P. (M) Relative expression of Myc and Zrf1 in S2R+ cells following knockdown of Myc, showing that Myc knockdown reduces Zrf1 transcript levels but not vice versa. See also S1 Data Table Q. (N) Relative expression of Myc and Zrf1 in S2R+ cells following knockdown of Zrf1, indicating that Zrf1 knockdown does not affect Myc expression. See also S1 Data Table R. (O) Relative expression of Myc and Zrf1 following Myc overexpression (pAWGFP-Myc) in S2R+ cells, showing that Myc overexpression increases Zrf1 transcript levels. See also S1 Data Table S. (P) Relative expression of Myc and Zrf1 in S2R+ cells following Zrf1 overexpression (pAWGFP-Zrf1), showing that Zrf1 overexpression does not affect Myc transcript levels. See also S1 Data Table T. Box plots show median, quartiles, and range. Statistical significance determined by Student’s t-test; ns = not significant, *p < 0.05, **p < 0.01, ***p < 0.001 ****p < 0.0001.

https://doi.org/10.1371/journal.pgen.1011910.g002

Zrf1 interacts with multiple signaling pathways to regulate ISC proliferation

Bleomycin-induced damage in the gut is known to activate EGFR signaling [15]. Since the loss-of-function and overexpression of Zrf1 resulted in the most pronounced defects in response to damage, we sought to test if it genetically interacts with the EGFR/Ras/Raf pathway. Ectopic expression of activated EGFR (TOP4.2), Ras (Ras1^A), Raf (Raf^F149) or the human Raf (hRaf^gof) in progenitors resulted in ISC overproliferation (Fig 2E, 2G and 2I), which could be rescued by Zrf1 knockdown (Fig 2F, 2H and 2I), which places it downstream of the EGFR/Ras/Raf pathway. However, Zrf1 is not restricted to EGFR signaling, since interaction experiments place it downstream of both Notch and Yorkie pathways as well (Fig 2J and 2K). These findings imply a broad function for Zrf1, reminiscent of the role of Myc in the midgut [11]. A previous study showed that knockdown of Myc in ISCs caused stunted proliferation in response to damage, and like Zrf1, downregulating Myc rescued overproliferation defects caused by various signaling pathways [11]. Thus, we suspected Zrf1 to be in the same pathway as Myc and found compelling evidence to support this when examining the literature. ChIP-seq of Myc in Kc cells showed that Myc binds to Zrf1 [16]. In addition, microarrays of larval tissue indicate that Zrf1 is upregulated following Myc overexpression [17]. Finally, co-expression analysis of scRNA-seq data show Zrf1 as a putative member of the Myc regulon [18].

To test whether Zrf1 is downstream of Myc, we knocked down Myc in S2R+ cells which resulted in a decrease in Zrf1 transcript levels (Fig 2M). Knockdown of Zrf1 stunted S2R+ proliferation but did not affect the levels of Myc (Fig 2L and 2N). Accordingly, Myc overexpression increased the levels of Zrf1, whereas upregulating Zrf1 did not affect Myc levels (Fig 2O and 2P). In conclusion, our results suggest that Zrf1 function downstream of Myc, integrating multiple signaling pathways to regulate ISC proliferation.

Zrf1 Protein-Protein interaction network identifies association with RISC

To gain insight into the function of Zrf1, we performed a search for interacting proteins, which we reasoned would identify regulators and localization partners. The Zrf1 protein complex was immunoprecipitated under mild detergent conditions (NP40) using GFP-trap beads from S2R+ cells expressing N-terminally GFP-tagged Zrf1 and subjected to mass spectrometry (IP-MS) (Figs 3A, 2B and S2A-S2C). Because IP-MS captures both direct binders and indirect co-complex members, we applied the AFM-LIS (AlphaFold-Multimer-Local Interaction Score) analysis pipeline in three iterative rounds to predict high-confidence direct interactors (Fig 3C) [19]. In Round 1, Zrf1 was modeled against each MS candidate, and only models exceeding a Local Interaction Score (LIS) of ≥ 0.203 and a Local Interaction Area (LIA) of ≥ 3432 were retained as primary interactors. In Round 2, these primary hits were screened against the subset of MS candidates that had failed in Round 1 to uncover bridging (“secondary”) interactions. In Round 3, the remaining MS candidates were modeled against the consolidated core set to reveal tertiary contacts. Ultimately, a Zrf1 PPI network comprising approximately 40 direct PPIs was built. Details for whole prediction results and predicted aligned error maps for all positive PPI predictions are shown in S3 Fig. Network analysis revealed associations with ribosomal proteins and components of the RNA-induced silencing complex (RISC) among other proteins (Fig 3C). Consistent with this, Zrf1 IP-MS showed >2-fold spectral counts for Rm62, Fmr1, Vig, Vig2, AGO2 and Tdrd3 in comparison to controls. To test whether these proteins interact with Zrf1, we conducted co-immunoprecipitation (Co-IP) experiments with S2R+ cells transfected with C-terminally 3xHA-tagged Zrf1 and N-terminal-GFP-tagged binding partners (Fig 3D). In total, Zrf1-3xHA immunoprecipitated with 17 out of 19 GFP-tagged binding partners. This included CG7182, CG8635, Hsc70–4, mod, penguin, rin, Rm62, Vig, two isoforms of Vig2, FDY (a Y-chromosome duplication of vig2), CG8545, 128up and Tdrd3. Reciprocal Co-IP found that C-terminal-3xHA-tagged Rm62, Fmr1, Vig, Vig2a, Vig2b, AGO2, FDY and Tdrd3 precipitated with GFP-Zrf1 (Figs 3E and S4A-S4D). Antibodies against these proteins showed that endogenous levels also immunoprecipitate with Zrf1, indicating a robust physical association between Zrf1 and RISC components (Figs S4E and 3F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. IP-MS-AFM reveals Zrf1 Protein-Protein Interactions (PPI) with components of the RISC components in Drosophila.

(A) Schematic of the experimental workflow for immunoprecipitation-mass spectrometry (IP-MS). S2R+ cells were transfected with constructs expressing either GFP or GFP-Zrf1. Following cell lysis, GFP-trap agarose beads were used to immunoprecipitate complexes, which were then analyzed via mass spectrometry after Coomassie staining. (B) Coomassie-stained gel showing proteins immunoprecipitated with GFP-Zrf1 compared to GFP alone. (C) Sequential AlphaFold-Multimer (AFM) screen scheme and protein interaction network generated from the Zrf1 IP-MS interactome analysis. The screening begins with the 1st AFM, predicting interactions between Zrf1 (central node, green) and the entire interactome to identify direct interactors with Zrf1 (Group 1, beige nodes). The 2nd AFM further explores interactions within Group 1 and between Group 1 and the indirect interactome, identifying new interactors (Group 2, orange nodes) and connections within Group 1. Finally, the 3rd AFM involves predictions between Group 2 proteins and additional indirect interactors, forming Group 3 (light blue nodes). Edge thickness represents the Local Interaction Score (LIS), where thicker lines indicate stronger predicted interactions. (D) Validation of Zrf1 interactions with candidate proteins identified in IP-MS through co-immunoprecipitation (Co-IP). S2R+ cells were transfected with GFP-tagged candidate proteins and HA-tagged Zrf1. Immunoprecipitation with anti-GFP was followed by immunoblotting with anti-HA to detect Zrf1. Representative interactions with various proteins, including Rm62, Vig, and others, are shown. (E) Reciprocal Co-IP with GFP-tagged Zrf1 and HA-tagged components of RISC, such as Fmr1, Vig, Vig2a, and others. (F) Immunoblots validating the association of Zrf1 with endogenous RISC components. Proteins including Fmr1, Vig, Vig2, AGO2, Rm62, and Tdrd3 were detected, confirming the interactions of Zrf1 with the RISC complex.

https://doi.org/10.1371/journal.pgen.1011910.g003

To determine if the physical interactions between Zrf1 and RISC was conserved in humans, we conducted IP-MS on N- and C- terminally epitope-tagged ZRF1 in HEK293T cells (under mild detergent conditions) (S5A-S5C Fig). The MS spectrum counts of both tagged versions of ZRF1 were enriched for SERBP1 (ortholog of Vig, Vig2 & FDY), human orthologs of Rm62 (DDX17, DDX21 and DDX5) or Fmr1 (FXR1, FMRP and FXR2), but not for TDRD3, HABP4 or members of the AGO-family (S5D Fig). Nonetheless, we conducted Co-IP experiments to determine if the PPIs of Zrf1 are conserved. C-terminal-epitope-tagged SERBP1, HABP4, DDX17, DDX21, or DDX5 all precipitated with Flag-ZRF1 (S5F Fig top). ZRF1-GFP precipitated with N-terminal-Flag-tagged AGO1 (S5F Fig bottom). Reciprocal Co-IP confirmed the ZRF1-SERBP1 and ZRF1-HABP4 interactions. Additionally, IP with a ZRF1 antibody successfully detected endogenous FXR1, FMRP and FXR2 (S5E Fig).

Components of RISC genetically interact with Zrf1 to regulate ISC proliferation

We hypothesized that perturbing components of RISC would mimic the effects of Zrf1 knockdown. Utilizing CRISPR, we generated mutants for vig and vig2, and assessed ISC proliferation during gut damage (Fig 4A). Homozygous and transheterozygous mutants of vig, vig2, and AGO2 exhibited reduced ISC proliferation in response to bleomycin, while FDY mutants showed no defects (Fig 4B). Additionally, double heterozygous animals — Zrf1²⁴²/+ with either fmr1(fmr1^d113M/Zrf1²⁴²), vig2 (vig2^Gd59-75/Zrf1²⁴²), or AGO2 (AGO2/ + ;Zrf1²⁴²/+) —displayed a synergistic decrease in ISC proliferation compared to individual heterozygotes (Fig 4C). These results collectively suggest a genetic interaction between Zrf1 and the RISC components (Fmr1, vig2, and AGO2) pivotal for regulating ISC proliferation during damage recovery.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Genetic interaction between Zrf1 and RISC components regulates ISC proliferation during midgut damage.

(A) Western blot analysis of adult fly lysates showing protein levels of Vig and Vig2 in various genetic backgrounds. Samples include wild-type controls (Canton-S, w¹¹¹⁸), and mutant alleles for vig (vig^d65-72, vig^d63-73, vig^C274), vig2 (vig2^Gd59-75, vig2^PL00470), and a double mutant (vig2^PL00470/vig2^Gd59-75). Tubulin (tub) was used as a loading control. The blots confirm the effective reduction of Vig and Vig2 in the mutant backgrounds. (B) ISC mitoses per midgut after 3 days of bleomycin (bleo) treatment in various female and male genotypes. Mutants for vig and vig2 exhibit a reduction in ISC proliferation compared to controls (w¹¹¹⁸). Mutants for AGO2 (AGO2^414/414 and AGO2^414/454) also show decreased proliferation. See also S1 Data Table U. (C) ISC mitoses per midgut after 3 days of bleomycin treatment in double heterozygous animals. Females carrying Zrf1²⁴²/ + combined with heterozygous mutant alleles for RISC components (Fmr1^d113M, vig^d65-72, vig^d63-73, vig2^Gd59-75, and AGO2⁴⁵⁴) are compared to individual heterozygotes. Significant reductions in ISC proliferation are observed in certain double heterozygotes, suggesting a synergistic interaction between Zrf1 and these RISC components. Error bars represent the interquartile range; n = number of midguts analyzed for each genotype. See also S1 Data Table V. Box plots show median, quartiles, and range. Statistical significance determined by Student’s t-test; ns = not significant, **p < 0.01, ****p < 0.0001.

https://doi.org/10.1371/journal.pgen.1011910.g004

Zrf1 is not involved in RNAi-mediated post-transcriptional silencing

In light of the established roles of AGO2 and Rm62 in RNAi-mediated post-transcriptional silencing [20,21] and the physical association between Zrf1 and RISC, we posited that Zrf1 might play a role in the RNAi pathway. Thus, we assessed the knockdown efficiency of dsGFP in S2R+ cells expressing GFP following pre-treatment with various dsRNA. Initially, cells were exposed to either dsLacZ (control), dsAGO2, or dsZrf1. After 3–5 days, the same cells were subjected to GFP dsRNA treatment or left untreated. The effectiveness of RNAi-mediated silencing was assayed by measuring GFP levels normalized to tubulin. As anticipated, RNAi-mediated GFP knockdown was compromised in cells pre-treated with dsAGO2 compared to the control. Intriguingly, knockdown of Zrf1 using two independent dsRNAs did not impact GFP knockdown (Fig 5A and 5B). Our findings indicate that Zrf1 does not participate in RNAi-mediated post-transcriptional silencing.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Zrf1 mutants exhibit deregulation of chromatin-associated defects.

(A) Western blot analysis of S2R+ cells pre-treated for 5 days with various dsRNAs (dsLacZ, dsZrf1, dsAGO2). Cells were subsequently treated with dsGFP to assess RNAi-mediated silencing. Immunoblots show levels of AGO2, GFP, and tubulin (loading control). (B) Quantification of GFP levels from (A), normalized to tubulin. Knockdown of Zrf1 did not significantly change GFP silencing efficiency, supporting that Zrf1 is not involved in RNAi-mediated post-transcriptional silencing. See also S1 Data Table W. (C) Representative images of fly eyes of w^m4h, ac¹/ Y males crossed with various RISC and Zrf1 mutants. (D) Quantitative analysis of relative mean pigmentation intensity from (C). Box plots compare pigmentation across RISC and Zrf1 mutant backgrounds, normalized to control (w^m4h, ac¹/Y). Reduced pigmentation corresponds to enhanced suppression of PEV, while increased pigmentation indicates loss of PEV suppression. See also S1 Data Table X. (E) Extra sex comb phenotypes in Pc¹ and double heterozygous Pc¹; Zrf1 mutant males. Images show T2 and T3 legs of control (Pc¹/+) and double heterozygous (Pc¹/Zrf1¹⁴⁸, Pc¹/Zrf1¹²⁷) flies. Blue arrows point to the presence of ectopic sex combs, indicating synergistic enhancement of Pc¹ phenotypes by Zrf1 mutants. (F) Percentage distribution of extra sex combs in Pc¹ and double heterozygous Pc¹; Zrf1 mutant males. Bar graph displays the proportion of males with 0, 1, 2, 3, or 4 extra sex combs per male across genotypes, with Zrf1 mutants showing an increased frequency of ectopic sex combs. See also S1 Data Table Y. Box plots display median, quartiles, and range; statistical significance determined by Student’s t-test; ns = not significant, **p < 0.01, ***p < 0.001, ****p < 0.0001.

https://doi.org/10.1371/journal.pgen.1011910.g005

Zrf1 mutants dominantly suppress position-effect variegation and enhance polycomb defects

Previous studies have revealed that RISC components have non canonical roles in modifying heterochromatin gene silencing [22,23]. To investigate whether Zrf1 shares similar functions, we used the positive-effect variegation (PEV) reporter In(1)w^m4h, where the white (w) gene is transposed close to pericentromeric heterochromatin on the X-chromosome by a large chromosome inversion [24]. PEV results in suppressed w⁺ expression in w^m4h/Y males, leading to mottled eye pigmentation (Fig 5C). Conventionally, enhancers and suppressors of PEV have been associated with genes that regulate chromatin condensing and de-condensing, respectively. Confirming earlier studies [25,26], AGO2 (AGO2⁴¹⁴/+) and Rm62 (Rm62⁰¹⁰⁸⁶/+) mutants suppressed w^m4h PEV, while Fmr1 (Fmr1^d50M/ + & Fmr1^d113M/+) mutations did not. In addition, vig alleles (vig^d65-72/ + & vig^d63-73/+ but not vig^C274/+) moderately suppressed PEV, whereas vig2 alleles (vig2^PL00470 & vig2^Gd59-75) exhibited strong suppression. Finally, FDY mutations (FDY^MUT_L3 & FDY^MUT_L64) also dominantly suppressed PEV when compared to the control (FDY^WT_E5) (Fig 5C and 5D).

Similar to the components of RISC, heterozygous alleles of Zrf1 dominantly suppressed w^m4h PEV. This encompassed a deficiency that covers Zrf1 (BSC553), a P-element insertion with a weak mini-white (Zrf1^G4964) and five indel mutants (Zrf1⁰¹⁶, Zrf1⁰¹⁸, Zrf1¹²⁷, Zrf1¹⁴⁸ & Zrf1²⁴²) generated using CRISPR/Cas9 (Figs 5C, 5D and S6A). Collectively, our findings unveil Zrf1 and RISC components (except Fmr1) as dominant suppressors of PEV, suggesting a broad role in chromatin condensation.

Notably, Polycomb group proteins are also known to regulate transcriptional repression of target genes through chromatin modifications. A previous study highlighted the ability of ZRF1 to bind to H2Aub1 and displace PRC1, activating previously repressed genes [10]. Another study also showed that plant ZRF1 possesses both PRC1-related and PRC1-unrelated functions [27]. Given this context, we hypothesized that Zrf1 might interact with Polycomb (Pc), influencing chromatin structure and gene expression. Using Pc¹ animals, a weak amorphic allele of Pc, we tested for genetic interactions between Zrf1 and Pc. In wildtype flies, sex combs are only found at the distal end of the male foreleg (T1) (Figs 5E and S6B). However, heterozygous Pc¹ males have varying degrees of extra sex combs in the second (T2) and third (T3) legs. Zrf1 heterozygous mutant males (Zrf1¹⁴⁸/ + , Zrf1²⁴²/+) are indistinguishable from control animals. Interestingly, double heterozygous animals of Zrf1 and Pc alleles (Zrf1¹⁴⁸/Pc¹, Zrf1²⁴²/Pc¹) synergistically enhanced the sex comb defect observed in Pc¹ heterozygous animals alone (Figs 5E, 5F and S6B), suggesting that Zrf1 enhances the de-repressed genes regulated by Pc.

Zrf1 as a key regulator of ISC proliferation

Although Zrf1 orthologs share some conserved roles across species, their functions have diverged considerably. In mammals, ZRF1 displaces Polycomb complexes to activate developmental genes critical for cell fate transition [10]. Similarly, in plants, Zrf1 orthologs promote the activation of genes associated with flowering and growth [28,27]. By contrast, the yeast ortholog Zuotin operates as a molecular chaperone within the ribosome-associated complex, assisting nascent polypeptide folding and maintaining proteostasis [29]. The functional differences highlight the unique value of Drosophila as a model to uncover aspects of Zrf1 biology that may be obscured in other organisms.

Our results establish that Zrf1 is essential for ISC proliferation during both homeostasis and regeneration. Our RNAi screen identified Zrf1 as a critical factor, with its knockdown leading to a significant reduction in ISC proliferation. This phenotype was not rescued by blocking apoptosis, indicating a specific requirement for Zrf1 in ISC proliferation rather than cell survival. Interestingly, Zrf1 overexpression promoted ISC proliferation only in response to damage, suggesting that its role is tightly regulated by cellular context and environmental cues.

The integration of Zrf1 into multiple signaling pathways, including the EGFR/Ras/Raf, Notch, and Yorkie pathways, underscores its broad regulatory capacity. Our interaction experiments position Zrf1 downstream of these pathways, akin to the role of Myc in the midgut. This finding aligns with previous studies showing that Myc and Zrf1 share common regulatory networks [11], further supported by ChIP-seq and microarray data [17,16]. Altogether, our results suggest that Myc likely regulates the direct transcription of Zrf1.

Zrf1 protein-protein interaction network and association with RISC

The construction of a Zrf1 protein-protein interaction network using AlphaFold-Multimer and Local Interaction Score (LIS) analysis revealed significant associations with components of the RNA-induced silencing complex (RISC) [19]. Co-IP experiments confirmed these interactions, highlighting the potential involvement of Zrf1 in RNA processing and silencing mechanisms. However, our functional assays demonstrated that Zrf1 does not participate in RNAi-mediated post-transcriptional silencing, redirecting our focus toward its chromatin-related functions.

Zrf1 as an epigenetic factor

Our findings that Zrf1 mutants dominantly suppress position-effect variegation (PEV) and enhance Polycomb (Pc) defects provide compelling evidence for a role of Zrf1 in epigenetic regulation. The synergistic interaction between Zrf1 and Pc¹ alleles, resulting in enhanced sex comb phenotypes, suggests that Zrf1 modulates chromatin structure and gene expression in concert with Pc group proteins. This is consistent with previous reports indicating the ability of Zrf1 to displace Pc complexes from chromatin to activate gene expression [10]. Interestingly, other chromatin regulators such as Piwi have also been implicated in maintaining heterochromatin organization and repressing transposable elements in ISCs [30,31]. Although we did not detect a direct interaction between Zrf1 and Piwi in our study, both factors may act through parallel pathways to safeguard chromatin integrity in ISCs.

Implications for cancer and therapeutic potential

The role of Zrf1 in regulating ISC proliferation and chromatin dynamics suggests potential implications for cancer biology. Given that Zrf1 can function as both a tumor suppressor and an oncogene depending on the context, understanding its regulatory networks could provide new targets for cancer therapy [32,33]. For example, in the context of gastric carcinoma, where Zrf1 is overexpressed [34], targeting its interaction with Pc and RISC components could offer novel therapeutic strategies.

Future directions

Future research should aim to elucidate the precise molecular mechanisms underlying Zrf1’s interactions with Pc and RISC components. Additionally, investigating Zrf1’s role in other tissues and cancer types will be crucial for translating these findings into clinical applications. Understanding how Zrf1 integrates multiple signaling pathways to regulate stem cell behavior and chromatin dynamics will provide valuable insights into stem cell biology and open new avenues for therapeutic development.

In conclusion, our study highlights the critical role of Zrf1 in ISC proliferation and chromatin regulation, with significant implications for stem cell biology and cancer therapy. By unraveling the complex networks involving Zrf1, this study paves the way for future research and potential clinical applications targeting Zrf1 and its associated pathways.

RNAi screening

We used esg-Gal4, UAS-GFP, tub-Gal80^ts (egt^ts) to knockdown genes in adult ISCs/EBs and quantified ISC proliferation by counting the number of mitotic (pH3-positive) cells per midgut. UAS-RNAi lines were obtained from Bloomington (BDSC), Vienna (VDRC) and NIG/Kyoto Drosophila stock centers. UAS-RNAi stocks that are homozygous lethal on second chromosomes were placed over CyO, twi-Gal4, UAS-2EGFP balancers. Virgin egt^ts females were crossed to UAS-RNAi males and reared at 18°C (restrictive temperature). After eclosion, egt^ts + UAS-RNAi females were kept at 18°C for 3 days and shifted to 29°C (permissive temperature) for 10 days before their whole midguts were dissected. Flies were transferred to fresh vials every 2–3 days. Mated females were used in all experiments and reared with male flies at the ratio of 2:1. Whole midguts were analyzed for pH3 positive counts.

Infection and compound feeding

The OD₆₀₀ of an overnight suspension of Ecc15 was determined by measuring two diluted (1:10 & 1:100) aliquots of the culture. After centrifuging the culture at 3000g for 20min, the pellet was resuspended in 5% sucrose with a volume to a final OD₆₀₀ of 100. Paper towels cut out to the shape of the vial base were soaked with 500uL of the concentrated Ecc15 solution. For oral infection, flies were starved for 2h at 29°C in an empty vial before transferring to vials containing Ecc15.

Immunohistochemistry and imaging

Whole midguts were dissected in PBS and fixed in 4% PFA in PBS at room temperature for 30 minutes. Fixed guts were washed in 0.1% Triton X-100 in PBS (PBST) once then blocked with a blocking buffer (0.1% Triton, 5%NDS in PBS) for 30 minutes at RT. Primary antibodies were incubated overnight at 4°C in the blocking buffer. Guts were washed 3x in a blocking buffer and incubated with secondary antibodies overnight at 4°C along with DAPI (1:1000 of 1mg/ml stock). After antibody staining, guts were washed 3x in PBST and mounted in Vectashield antifade mounting medium (Vector Laboratories H-1200). Tape was used as a spacer to prevent coverslips from crushing the guts. Antibody dilutions used were as follows: rabbit anti-pH3 (1:1000, Millipore 06–570), mouse anti-Dachshund (1:20, DSHB AB_579773), chicken anti-GFP (1:1000, Abcam ab13970), donkey anti-rabbit 565 (1:2000, Molecular Probes A31572), goat anti-mouse 633 (1:2000, Thermo Scientific A-21240) and goat anti-chicken 488 (1:2000, Thermo Fisher Scientific A-11039). Guts were imaged on a spinning-disk confocal system, consisting of a Nikon Ti2 inverted microscope equipped with a W1 spinning disk head (50um pinhole, Yokogawa Corporation of America) and a Zyla 4.2 Plus sCMOS monochrome camera (Andor).

RNA extraction and quantitative real-time PCR

10-15 midguts were dissected in 1xPBS and homogenized in 300uL of TRIzol (ThermoFischer, 15596–026) using RNase-free pestles. RNA was extracted using Zymo Direct-zol RNA MicroPrep kit (R2050) and subsequently DNase-treated using Turbo DNA free (AM1907). 400–450ng of the resulting RNA was reverse transcribed using Bio-Rad iScript Select cDNA synthesis kit (708896) and Sybr green (1708880) based qRT-PCR was performed to determine the levels of gene expression. qRT-PCR primers were designed using FlyPrimerBank.³⁰ Primers that were not from the literature were tested for their efficiency by running qRT-PCR on serial dilutions of pooled cDNA. Only primers with efficiencies of greater than 85% were selected for further use. The primer pairs tested are as follows: Zrf1 (PP12004: 81.2%, PP24706: 80.3%, PP36474: 96.0%), RpL32 (PD41810: 99.6%), myc (PP17271: 75.9%, PP29594: 102.1%, PP4159: 94.9%).

Cell culture and RNA interference

Drosophila S2R+ cells were cultured at 25°C in Schneider’s medium with 10% fetal bovine serum (FBS) and streptomycin. HEK293T cells were cultured at 37°C in DMEM (Invitrogen 10-017-CV) supplemented with 10% FBS and streptomycin. For RNAi experiments, DNA templates for dsRNA were obtained from the Drosophila RNAi screening center (DRSC). Amplicons used in this study were: Zrf1(#11620), Myc (#23360, #37533), Fmr1 (#26781), vig (#03631), vig2 (#14402), AGO2 (#10847), Rm62 (#25109), Tdrd3 (#10016) and LacZ dsRNA. A second dsRNA Zrf1 template was generated by amplifying Zrf1 cDNA with primers containing overhanging T7 sequences (JL278 and JL279). DNA templates were amplified with T7 primers using Phusion polymerase. PCR products were in vitro transcribed (IVT) using MEGAscript T7 Transcription Kit (ThermoFisher, AMB1334-5). S2R+ were treated with dsRNAs by the “bathing” method (https://fgr.hms.harvard.edu/drsc-cellrnai).

JL278: TAATACGACTCACTATAGGGTCTGGAGCAATGAGAACGTG

JL279: TAATACGACTCACTATAGGGGGTTGACTGATTGTTCGCCT

Cell viability

1x10⁵ cells were plated into 96-well plates. Treated with 500ng of dsRNAs in a total volume of 80ul. 10ul was used and counted for cell number with a haemocytometer.

Plasmids. Molecular cloning – Drosophila constructs

pEntr-GFP-Stop: The pEntr-dTOPO backbone was amplified from pEntr-Akt (Gift from P. Saavedra) with the primers pEntr_F & pEntr_R. GFP was amplified from the plasmid pAW-vhh05-GFP (gift from J. Xu) with JL201 and JL202. The two fragments were joined by Gibson assembly (NEB, E2611S). pEntr-GFP-Zrf1-Stop: The backbone of pEntr-GFP-stop was PCR amplified with the primers pEntr_F & JL203. The Zrf1 coding sequence was amplified from the vector LD23875 with JL204 & JL205. The linker (GGGSGSG) was introduced between the GFP and Zrf1 by overhanging sequences on the primers. Amplicons were joined by Gibson assembly. pEntr-Zrf1-NoStop: The pEntr-dTOPO backbone was amplified from pEntr-GFP-stop with the pEntr_F & pEntr_R. The Zrf1coding sequence was amplified from the plasmid LD23875 with JL171 & JL172 (primers that omitted the stop codon). Amplicons were joined by Gibson assembly. pEntr-AGO2-RB-NoStop: The AGO2 coding sequence was amplified from the plasmid pAFW-AGO2 (Addgene #50554) with the primers JL267 & JL268. The fragment was inserted into the pEntr backbone via Gibson assembly. pEntr-FDY-NoStop: Two gBlock (IDT) dsDNA fragments (gJL1 & gJL2) encoding the entire FDY coding region (Accession KR781487) were joined with the pEntr backbone using Gibson assembly. pEntr-Tdrd3-NoStop: The Tdrd3 coding sequence was amplified from the plasmid RE1471 with JL293 & 294. Final vectors: pEntr223 vectors containing full-length cDNAs of dMyc, Hspa14 (CG7182), Zc3h15 (CG8635), Hsc70–4, mod, peng, Fkbp39, Rm62, rin, Fmr1, vig, vig2, Ref1, RpS3, 128up, CG8584 and yps were from the FlyBi ORFeome Collection. The pEntr-mCherry-NoStop was a gift from B. Xia. All fly pEntrvectors containing cDNAs were transferred into the Drosophila gateway vectors pAW, pAWH, pAWM and pAGW (Carnegie Science/Murphy lab) using LR clonase II enzyme mix (Invitrogen, 11791–020) (see table for full list of vectors).

JL171: CCGCGGCCGCCCCCTTCACCATGACGAGCGGTACGGTAGC

JL172: GGGTCGGCGCGCCCACCCTTTTTGACCGCCGCCTGTG

JL201: CCGCGGCCGCCCCCTTCACCATGGTGAGCAAGGGCGAG

JL202: GGGTCGGCGCGCCCACCCTTTCACTTGTACAGCTCGTCCATGCC

JL203: GCCGGAGCCGCTGCCACCTCCCTTGTACAGCTCGTCCATGC

JL204: GGAGGTGGCAGCGGCTCCGGCATGACGAGCGGTACGGTAGC

JL205: GGGTCGGCGCGCCCACCCTTTCATTTGACCGCCGCCTG

JL267: CCGCGGCCGCCCCCTTCACCATGGGAAAAAAAGATAAGAACAAGAAAGGAG

JL268: GGGTCGGCGCGCCCACCCTTGACAAAGTACATGGGGTTTTTCTTCA

JL293: CCGCGGCCGCCCCCTTCACCATGGAATTAGGCAAGAAACTACG

JL294: GGGTCGGCGCGCCCACCCTTATGCTCTCGTTTGTGGG

Molecular cloning – Human constructs

pEntr-ZRF1-NoStop: The human ZRF1 coding sequence was amplified from the plasmid pcDNA3.1-ZRF1 (GenScript, OHu13274) with JL186 & JL188. pEntr-SERBP1-NoStop: The human SERBP1 coding sequence was amplified from the plasmid UAS-SERBP1 (DGRC, hGUHO03824) with JL313 & JL314. Final vectors: pDonr221-HABP4-NoStop (DNASU HsCD00745639), pEntr-mCherry-NoStop and other pEntr vectors containing cDNAs were transferred into the mammalian gateway vectors, pHAGE-N-FLAG-HA (pHFH; gift from R. Viswanatha) or pCSF107mT-GATEWAY-3’LAPtag (pCSF-GFP, Addgene #67618, gift from S. Entwisle) using LR clonase II enzyme mix. Other plasmids containing epitope-tagged cDNAs were used and are listed as follows: pDEST-myc-DDX17 (Addgene #19876).

JL186: CCGCGGCCGCCCCCTTCACCATGCTGCTTCTGCCAAGCG

JL188: GGGTCGGCGCGCCCACCCTTTCATTTCTTGGCTCTACTTGCATTCAGCA

JL313: CGCGGCCGCCCCCTTCACCATGCCTGGGCACTTACAGGAAG

JL314: GGGTCGGCGCGCCCACCCTTAGCCAGAGCTGGGAATGCC

Immunoprecipitation and immunoblotting

Immunoprecipitation was performed as previously described.³¹ S2R+ or HEK293T cells were split 2 days before transfection. 2ug of DNA was transfected into 20 x 10⁶ of cells in a 100mm dish with Effectene transfection reagent (Qiagen 301427) following the manufacturer’s protocol. After 3 days, cells were lysed with IP-lysis buffer (Pierce 87787) containing 2X protease inhibitor cocktail (ThermoFisher, 78430) and 2X PhosSTOP (Sigma-Aldrich 4906837001). Lysate was incubated with 25ul of Chromotek-GFP-trap beads (Bulldog Biotechnology gta-20) or anti-Flag agarose (Sigma A2220) and subject to end-over-end agitation for 2hrs at 4°Cs. Beads were washed 3–4 times with 500ul of lysis buffer. Beads were eluted in 25ul of sample buffer (Pierce 39001 with 10% 2-Mercaptoethanol), boiled for 5min and sedimented by centrifugation at 2500x g for 5min.

For immunoblotting, samples were run on 4%-20% polyacrylamide gel (Bio-Rad 4561096) and transferred to an Immobilon-P polyvinylidene fluoride (PVDF) 0.45um membrane (Millipore IPVH00010). The membrane was blocked with 5% skim milk in TBST (TBS with 0.05% Tween-20) at room temperature for 1h and probed with primary antibody in 5% skim milk overnight at 4°C. Membrane was probed with HRP-conjugated secondary antibody at room temperature for 1h and signal detected by enhanced chemiluminescence (ECL;Amersham RPN2209; Pierce 34095). Concentrations used for primary antibodies were as follows: chicken anti-GFP (1:10000, aveslabs GFP-1020), use anti-HA.11 (1:1000, BioLegend 901514), mouse anti-Fmr1 (1:100, DSHB 5A11), rabbit anti-Vig (1:30000, gift from N. Riddle) & rabbit-Vig2 (1:10000, gift from N. Riddle), mouse anti-dAGO2 (1:400, gift from H. Siomi), rat anti-Rm62 (1:2000, gift from E. Lei), rabbit anti-Tdrd3 (1:2000, gift from W. Wang), mouse anti-lost (1:100, DSHB), mouse anti-Tubulin (1:1000, Sigma T5168).

Measuring eye pigmentation

w^m4h, y¹, ac¹ females were crossed with different alleles and reared at 25°C. In the progeny, the left eye of 2-day-old adult male flies were imaged and used for measuring eye pigmentation. Images were separated into the RBG channels, and the mean grey intensity of the eye was measured under the red channel. This value was subtracted from the background mean grey intensity and compared between genotypes as a fold-change of the w^m4h, y¹, ac¹ control.

Generating fly mutants with CRISPR

vig mutants: y; P{TKO.GS04822}attP40 (BL81492) was crossed with y,w; nos-Cas9/CyO. y,w; nos-Cas9/ P{TKO.GS04822}attP40 males were crossed with y,w; Gla/CyO females. Resulting offsprings were outcrossed individually to establish stocks. A ~ 500 bp region covering the seed sequence was amplified with JL317 & JL318 and subcloned into pJET1.2/blunt vector for sequencing. The selected mutants, vig^d65-72 and vig^d63-73 have predicted premature stop codons that result in respectively truncated proteins of 59aa and 58aa.

Zrf1 mutants: y,w;P{HD_CFD0021}attP40/CyO-GFP (v341280) was crossed with y,w; nos-Cas9/CyO. Male y,w; nos-Cas9/attP40 was crossed with y,w; IF/CyO. Resulting offspring were outcrossed individually to establish stocks. Established stocks were crossed with a deficiency stock. Stocks resulting in non-complementation were deemed mutants.

JL317: ATCGAATCTCACAGGGTGTGAGAG

JL318: GTTTGTTGTCGGCCGATGGA

vig2 mutants: y; P{TKO.GS04324}attP40 (BL84243) was crossed with y,w; nos-Cas9/CyO. y,w; nos-Cas9/ P{TKO.GS04324}attP40 males were crossed with y,w; TM3/TM6b females. Resulting offsprings were outcrossed individually to establish stocks. A ~ 600 bp region covering the seed sequence was amplified with JL315 & JL316 and subcloned into pJET1.2/blunt vector for sequencing. The selected mutants, vig2^d53-72, vig2^d57-66 and vig2^G.d59-75, have predicted premature stop codons that result in respectively truncated proteins of 61aa, 26aa and 25aa.

JL315: GCCATTGGGACCAAAGTAACTCTCAAA

JL316: TTACAGGGCAGCATTTTCCACG

Sampling and statistics

Experimental data was always collected alongside a control group. Most if not all experiments were conducted once with at least a sample size of three. All statistics were collected in the traditional frequentist fashion. The Shapiro-Wilk test was used to test normality. For experiments comparing only two groups, two-tailed unpaired t-test were used. Ordinary one-way ANOVA for parametric multiple comparison and the non-parametric Kruskal-Wallis test was used for comparing two or more groups. Subsequent Dunn’s multiple comparison test was used to compare each group with the control.