PAF-Myc-Controlled Cell Stemness Is Required for Intestinal Regeneration and Tumorigenesis

SUMMARY

The underlying mechanisms how self-renewing cells are controlled in regenerating tissues and cancer remain ambiguous. PAF (PCNA-associated factor) modulates DNA repair via PCNA. Also, PAF hyperactivates Wnt/β-catenin signaling independently of PCNA interaction. We found that PAF is expressed in intestinal stem and progenitor cells (ISCs and IPCs) and markedly upregulated during intestinal regeneration and tumorigenesis. Whereas PAF is dispensable for intestinal homeostasis, upon radiation injury, genetic ablation of PAF impairs intestinal regeneration along with the severe loss of ISCs and Myc expression. Mechanistically, PAF conditionally occupies and transactivates c-Myc promoter, which induces the expansion of ISCs/IPCs during intestinal regeneration. In mouse models, PAF knockout inhibits Apc inactivation-driven intestinal tumorigenesis with reduced tumor cell stemness and suppressed Wnt/β-catenin signaling activity, supported by transcriptome profiling. Collectively, our results unveil that PAF-Myc signaling axis is indispensable for intestinal regeneration and tumorigenesis by positively regulating self-renewing cells.

eTOC Blurb

Controlling the proliferation of self-renewing cells is crucial for tissue homeostasis and regeneration. Kim et al. show that PAF (PCNA-associated factor), dispensable for intestinal homeostasis, is required to positively regulate the proliferation of self-renewing cells via Myc transactivation during intestinal regeneration and tumorigenesis.

INTRODUCTION

Stem cells play key roles in tissue homeostasis and regeneration by self-renewing and repopulating progenitor cells (Fuchs et al., 2004; Morrison and Spradling, 2008). In the small intestine, two major ISCs co-exist. Crypt base columnar cells (CBC) ISCs marked by the high Lgr5 expression are highly proliferative and essential for the intestinal homeostasis (Barker et al., 2007). The other ISCs located at position 4 (+4) and labeled by Hopx, Lrig1, Bmi1, and Tert are quiescent during intestinal homeostasis, whereas conditionally activated upon tissue damage. (Montgomery et al., 2011; Powell et al., 2012; Sangiorgi and Capecchi, 2008; Takeda et al., 2011). Accumulating evidence suggests that +4 ISCs function as a reservoir of ISCs during regeneration (Buczacki et al., 2013; Tian et al., 2011). Additionally, the committed progenitor cells (Dll1+, Alpi+, and Krt19+) located above the position +4 cells also dedifferentiate into ISCs for intestinal regeneration (Asfaha et al., 2015; Tetteh et al., 2016; Van Es et al., 2012), implying the involvement of the cell plasticity in rebuilding intestinal epithelium. Although the extensive lineage tracing studies have been used to identify ISCs or reservoir ISCs/IPCs populations, still the underlying mechanisms how these ISCs/IPCs cells are activated and expanded during regeneration remain elusive.

It was proposed that cancer stem cells (CSCs) are a subpopulation of tumor cells, which drives tumor growth by self-renewing and giving rise to the daughter cells (Nguyen et al., 2012). The identities of CSCs are still controversial, however, it is plausible that CSCs might be related to therapeutic resistance and tumor recurrence (Dean et al., 2005; Kreso and Dick, 2014). Expression of CD44, CD133, and Lgr5 have been suggested as a maker for stemness of colorectal cancer (CRC) cells (O’Brien et al., 2007; Ricci-Vitiani et al., 2007; Schepers et al., 2012; Zeilstra et al., 2008; Zhu et al., 2009). Nonetheless, how CSCs are maintained and expanded were not fully understood.

PAF (also known as p15/KIAA0101/NS5ATP9/OEACT-1) was initially identified as a proliferating cell nuclear antigen (PCNA)-interacting protein (Yu et al., 2001). PAF is implicated in both DNA repair and cell proliferation (Emanuele et al., 2011; Povlsen et al., 2012). PAF binds to PCNA sliding clamp and regulates DNA replication and repair (De Biasio et al., 2015). Importantly, PAF expression is significantly upregulated in many human cancers (Cheng et al., 2013; Hosokawa et al., 2007; Jain et al., 2011; Jung et al., 2013; Kais et al., 2011; Kato et al., 2012; Mizutani et al., 2005; Wang et al., 2016; Yu et al., 2001; Yuan et al., 2007). In pancreatic cancer cells, PAF overexpression is necessary for pancreatic cancer cell proliferation (Hosokawa et al., 2007). Additionally, PAF is associated with MAPK hyperactivation via transcriptional activation of the late endosomal/lysosomal adaptor, MAPK and mTOR activator 3 (LAMTOR3), which is involved in the initiation of pancreatic intraepithelial neoplasia (Jun et al., 2013). Moreover, PAF hyperactivates Wnt/β-catenin signaling in CRC cells as a co-factor of β-catenin/EZH2 transcriptional complex, resulting in the development of intestinal adenoma (Jung et al., 2013).

In this study, we sought to interrogate how stem cells are activated by radiation injury. Our unbiased gene expression screening identified that, among DNA repair-related genes, PAF expression is associated with controlling ISCs/IPCs. Further comprehensive and genetic approaches revealed that PAF-Myc signaling axis is indispensable for intestinal regeneration and tumorigenesis by positively controlling the expansion of stem cells.

RESULTS

Upregulation of PAF Expression upon Radiation Injury

To identify essential genes associated with DNA repair during tissue regeneration, we conducted qRT-PCR array for DNA repair gene collections from irradiation (IR) treated mouse small intestine (1 day post-injury [1 dpi], 10 Gy) (Figures 1A and 1B). Among differentially expressed 79 genes, PAF expression was highly upregulated by IR (7^th ranked). Additionally, IR upregulated PAF expression in the mouse small intestine in a dose- and time-dependent manner (Figures 1C and 1D), which led us to hypothesize that PAF plays crucial roles in intestinal regeneration.

Figure 1. Upregulation of PAF Expression upon Radiation Injury.

(A, B) Gene expression profiling of DNA repair genes upon radiation injury in mouse small intestine. After treatment of 10 Gy irradiation (1 day post-injury [1 dpi]), the whole small intestine samples were analyzed by qRT-PCR (N=3). PCNA is the fourth and PAF are the seventh upregulated genes among the 79 genes related to DNA repair.

(C) Time-dependent upregulation of PAF expression upon IR injury. At 0, 1, 2, 4, and 7 dpi, the small intestine samples were collected and analyzed by qRT-PCR (N=3).

(D) Dose-dependent upregulation of PAF expression upon injury. 6 and 10 Gy irradiation were used (1 dpi). Student’s t-test; error bars = S.E.M; asterisks: P< 0.05.

PAF Expression in Replenishing and Regenerating Intestinal Crypts

To elucidate the in vivo roles of PAF in intestinal regeneration, we generated PAF KO mouse model using clustered regularly interspaced short palindromic repeat (CRISPR)/Cas9 gene targeting system (Wang et al., 2013; Yang et al., 2013)(Figure S1). Of note, PAF KO mice are viable without any discernible phenotypes. To examine PAF expression in the crypt, we conducted immunohistochemistry (IHC) using two different PAF monoclonal antibodies (Figures 2A and S2B). We confirmed the specificity of PAF antibody using PAF KO mice as a negative control (Figures 2A and S2B). We located 5~10 cells of PAF positive (PAF+) cells existed in the one section of crypt (Figures 2A and 2B). PAF was expressed in 1~2 cells of CBC ISCs and 3~8 cells of transit-amplifying (TA) cells (Figures 2A and 2B), but not in the villi (Figure S2A). Most PAF+ cells belong to Ki67+ cells (17.71% of Ki67+ cells), a marker for cell proliferation (Figures 2C and 2E). Despite the additional role of PAF in DNA repair as a PCNA-interacting protein (Povlsen et al., 2012), only 23.01% of PAF+ cells were PCNA+ cells (Figures 2D and 2E). Furthermore, PAF KO mice did not display differences in the expression of PCNA in the crypts, compared to PAF wild-type (WT) mice (Figure 2F), implying the existence of PCNA-independent functions of PAF in the intestine.

Figure 2. PAF Expression in Normal and Regenerating Intestinal Crypts.

(A) PAF expression in the small intestine. PAF WT and KO mice were analyzed for immunofluorescent (IF) staining of PAF (arrows). Asterisks mark non-specific staining signals.

(B) Quantification of PAF+ cells in the small intestinal crypts.

(C) Co-immunostaining of mouse small intestine (PAF WT and KO) for PAF and Ki67. White arrows: PAF+:Ki67+ cells; yellow arrows: PAF+:Ki67+ CBC cells. Asterisks mark non-specific staining signals.

(D) Co-immunostaining of mouse small intestine (PAF WT and KO) for PAF and PCNA. White arrows: PAF+:PCNA+ cells; yellow arrows: PAF+:PCNA+ CBC cells; Asterisks mark non-specific staining signals.

(E) Quantification of PAF+ cells in K67+ or PCNA+ cell population. Of note, PAF+:Ki67- cells were rarely found (approximately 1/50 crypts).

(F) No effects of PAF KO on PCNA expression pattern. IF staining of the mouse small intestine for PCNA.

(G) No expression of PAF in the Paneth cells. Co-immunostaining of the small intestine for PAF and Lysozyme. White arrows: PAF+ cells; yellow arrows: PAF+ CBC cell.

(H) Co-expression of PAF and Lgr5 in the small intestine. Co-immunostaining of the small intestine of Lgr5-EGFP-CreERT2 mouse strain. White arrows: PAF+:Lgr5+ cells; yellow arrows: PAF+:Lgr5+ CBC cells. Asterisks indicate non-specific staining signals.

(I) PAF expression in Lgr5+ cells. FACS-isolated Lgr5+ cells were analyzed for qRT-PCR. Asterisk=P<0.05.

(J, K) PAF expression in TERT+ and Bmi1+ cells. Co-immunostaining of TERT-Tdtomato-CreERT2 or Bmi1-EGFP knock-in mouse intestine samples for PAF. Arrowheads: PAF+:TERT+ (J) and PAF+:Bmi1+ (K) cells; arrows: PAF+ cells. Asterisks mark non-specific staining signals.

(L, M) The increase of PAF+ cells upon radiation injury. Immunostaining of mouse intestine (0, 1, 2, 4, and 7 dpi; 10 Gy) for PAF (L). White arrows: PAF+ cells; yellow arrows: PAF+ CBC cells. Quantification of PAF+ cells in the regenerating crypts (M).

The representative images were shown from at least three independent experiments. Scale bars=20μm. See also Figures S1 and S2.

To determine whether PAF is expressed in ISCs, we used a Lgr5-EGFP-CreERT knock-in mouse. PAF+ cells were correlated with Lgr5+ CBC cells and Lgr5+ progenitor cells localized at transit amplifying zone (TA zone) (Figure 2H), confirmed by qRT-PCR of fluorescence-activated cell sorting (FACS)-isolated Lgr5+ cells (Figure 2I). The Paneth cells between CBC ISCs did not express PAF (Figure 2G). We further asked whether PAF is expressed in the quiescent ISCs at position 4, using TERT-tdTomato-CreERT2 (Jun et al., 2016) and Bmi1-EGFP (Hosen et al., 2007) knock-in mice. Some population of TERT+ (52.94%) and Bmi1+ cells (63.16%) expressed PAF (Figures 2J, 2K, S2D, and S2E). These results indicate that PAF+ cells include some population of CBC (Lgr5^high) ISCs, position 4+ ISCs (TERT+ and Bmi1+), and TA progenitor (Lgr5^low) cells.

Given the upregulation of PAF expression by IR (Figure 1), we monitored PAF+ cells in the regenerating intestinal crypts. Interestingly, upon IR exposure (10 Gy), the number of PAF+ cell was increased until 4 dpi and then decreased at 7 dpi (Figures 2L and 2M), which is consistent with PAF mRNA upregulation (Figures 1C and 1D). At 1~2 dpi, the remaining ISCs started to divide. At 4 dpi, the regenerating crypts were enlarged, and newly generating CBC cells and TA cells reappeared. At 7 dpi, the regeneration process was mostly completed (Jun et al., 2016; Suh et al., 2017). Intriguingly, PAF+ cells remained at position 4-6 and 8-9 at 1~2 dpi, whereas PAF+ CBC cells were lost by cell death at 1 dpi (Figures 2L and 2M). The newly generated CBC cells and TA cells highly expressed PAF at 4 dpi, and PAF+ cells were partially restored as CBC and TA cells (7 dpi) (Figures 2L and 2M). Given the high enrichment of PAF expression in the remaining and active ISCs/IPCs in the regenerating crypts, these results imply that PAF might be involved in controlling ISCs/IPCs during regeneration.

Impaired Intestinal Regeneration by PAF KO

To directly test whether PAF is engaged in intestinal regeneration, we utilized PAF KO mouse model. Prior to the experiments, we examined the intestinal morphology and several differentiated intestinal epithelial cell (IEC) markers in PAF KO mice. PAF KO mice displayed no differences in intestinal morphology and IEC differentiation (Figures S3A and S3B). To further address whether PAF KO affects intestinal homeostasis, we performed BrdU incorporation assays (Figures S3C and S3D). The number of BrdU incorporated cells in the crypts of PAF KO mice was the same as that in WT after 2h BrdU induction (Figure S3C). Furthermore, PAF KO mice showed a similar migration rate during 3 days of tracing of BrdU+ cells (Figure S3D). These data indicate that genetic ablation of PAF does not affect the proliferation and migration of IECs. Additionally, we analyzed the expression of various genes related to intestinal homeostasis by qRT-PCR (Figure S3E). Despite the slight decrease of c-Myc expression in PAF KO intestine (~36%), the expression of most genes was not altered in the crypts of WT and PAF KO mice. It is noteworthy that conditional deletion of c-Myc in the intestine has no impact on intestinal homeostasis (Bettess et al., 2005). These results suggest that PAF is dispensable for intestinal homeostasis.

Next, we asked whether PAF is required for tissue regeneration. We found that upon IR injury, PAF KO mice showed severe defects in the intestinal regeneration (Figures 3A-3D). At 4 dpi, the number of viable crypts was markedly reduced by 10 Gy (50.5% decrease) and 12 Gy (100% decrease) IR treatment (Figures 3B and 3E). The crypt structure, represented by lysozyme staining, was also severely disintegrated in PAF KO mice (Figures 3D and 3F) compared to that in WT mice. These results suggest that PAF is indispensable for intestinal regeneration.

Figure 3. Impaired Intestinal Regeneration by PAF KO.

(A-D) Impaired intestinal regeneration after irradiation (4 dpi; 10 and 12 Gy). Hematoxylin and eosin staining of the mouse small intestine samples of PAF^+/+ (WT) or PAF^-/- (KO) mice (A); IHC of the small intestine samples for Ki67 (B); IHC of the small intestine samples for E-cadherin, a marker for epithelial cell (C); IHC of the small intestine samples for Lysozyme, a marker for Paneth cell (D); Yellow arrows: crypts.

(E) Quantification of the viable crypts. Crypts that showed five successive Ki67+ cells were counted as a viable crypt.

(D) Quantification of the intact localization of Lysozyme. Crypts possessing at least three Lyso+ cells localized at the crypt bottom were counted as intact localization.

The representative images were shown from at least three animals for each condition. Scale bars=100μm. See also Figures S3 and S4.

PAF Is Required for ISCs/IPCs Expansion in Regenerating Crypts

To understand the underlying mechanism of intestinal regeneration defects by PAF KO, we first examined DNA repair process in PAF KO mice. Interestingly, no significant increase of cell death (Apoptotic body and Active-Caspase 3 staining) was detected in PAF KO mice during 48h after IR (Figures S4A-S4D). Initial recognition of DNA damages, checked by ATM pS1981(1h) and phospho-γH2AX (6h) (Figures S4E, S4F, and S4G), and downstream targets of DNA damage responses were not changed in PAF KO mice (Figures S4H-S4K). DNA double-strand breaks repair (6h~48h after IR) represented by γ-H2AX was also similar between PAF WT and KO mouse intestine samples (Figures S4E and S4F). These data suggest that regeneration defects in PAF KO mice are not due to impaired DNA repair processes.

Next, given the specific expression of PAF in ISCs/IPCs (Figures 2H-2K), we asked whether PAF KO-induced defects in intestinal regeneration is due to the dysfunction of ISCs/IPCs. To test this, we performed crypt organoid culture assays. Interestingly, the crypt organoids derived from PAF KO mice showed the low efficiency of organoid formation (Figures 4A and 4B) as well as decreased growth and budding efficiency (Figures 4A, 4C, and 4D). These results imply that PAF might be required for intestinal regeneration possibly by controlling ISCs/IPCs in a cell-autonomous manner.

Figure 4. PAF Is Required for ISCs/IPCs Expansion in Regenerating Crypts.

(A-D) Reduced organoid development by PAF KO. The crypt organoid growth assays from the small intestine samples of PAF WT and KO mice (A); quantification of organoid efficiency (N≥2000 from the three independent experiments) (B); size (5 days after seeding; N≥50) (C); budding efficiency (N≥50) (D). Asterisks=P< 0.01. Scale bars=100μm.

(E, F) Impaired ISCs/IPCs expansion by PAF KO. Visualization of Lgr5+ EGFP cells by GFP immunostaining of the small intestines from PAF WT;Lgr5-EGFP and PAF KO;Lgr5-EGFP mice at 0 and 7 dpi (E). Scale bars=50μm; Quantification Lgr5+ (EGFP+) existing crypts per field (F). At least 30 fields of view were counted; Arrows (EGFP+ crypts); asterisks (EGFP-regenerating crypts).

(G, H) Analysis of Lgr5 expression by FISH in PAF WT and PAF KO regeneration crypts. The representative images were shown. (G). Scale bars=20μm; % of Lgr5+ (EGFP+) crypts in the field (H). At least 10 fields of view were counted.

(I-K) Reduced single cell organoid development by PAF KO. Representative images of single cell (Lgr5+) organoids (day 5) derived from PAF WT;Lgr5-EGFP and PAF KO;Lgr5-EGFP mice (I). Scale bars=50μm; quantification of organoid development efficiency (J) (N≥5000 cells were analyzed for from three independent experiments); size (K) (N≥30 single cell organoids were analyzed). (L) Gene expression analysis of FACS-isolated Lgr5^low (GFP+) cells in the IR-treated intestine from PAF WT;Lgr5-EGFP and PAF KO;Lgr5-EGFP mice (2 dpi, 10 Gy). qRT-PCR analysis. Asterisks=P< 0.05. See also Figure S5.

To further assess underlying mechanisms of dysregulated ISCs/IPCs in PAF KO mice during intestinal regeneration, we generated PAF KO;Lgr5-EGFP-CreERT mice and analyzed the Lgr5+ ISCs/IPCs population after radiation injury. After IR treatment, Lgr5^high ISCs cells are depleted by DNA damage-induced apoptosis. However, a small population of Lgr5^low cells (including quiescent ISCs and TA progenitor cells) survives and contributes to intestinal regeneration (Buczacki et al., 2013; Metcalfe et al., 2014; Muñoz et al., 2012). In the normal intestine of PAF KO mice, the number of Lgr5+ ISCs/IPCs was not changed (Figures S5A and S5B), which is consistent with the BrdU incorporation assay results (Figures S3C and S3D). Similarly, PAF KO did not affect the number of Lgr5^low cells population upon IR (Figures S5C and S5D), suggesting PAF KO has no effects on Lgr5+ ISCs/IPCs on the overall cell viability in the early stage of regeneration (1 dpi, 10 Gy). However, PAF KO mice exhibited significantly delayed restoring the Lgr5+ cells in regenerating crypts at the late time point (7 dpi) (Figures 4E-4H). Of note, in the treatment of 10 Gy IR, PAF KO mice displayed about 50% of viable crypts (4 dpi) (see Figures 3A-3E). Although the regenerating crypts restored their morphology similar to WT in PAF KO mice (7 dpi), the number of Lgr5+ cells were not restored. To complement the in vivo result of PAF KO-decreased Lgr5+ ISCs/IPCs, we also employed single cell organoid cultures of FACS-isolated Lgr5+ cells from the mouse intestine. Strikingly, single cell organoids from PAF KO Lgr5+ cells showed three times lower efficiency in the organoid formation and displayed severe growth defects compared to PAF WT Lgr5+ organoids (Figures 4I-4K). PAF KO Lgr5+ cells grew markedly slowly during the 2 weeks in single cell organoid culture. These in vivo and in vitro results suggest that PAF is required for the expansion of ISCs/IPCs during intestinal regeneration.

PAF Transactivated c-Myc Is Required for ISCs/IPCs Expansion upon Radiation Injury

Next, we sought to determine the molecular mechanism of PAF-induced ISCs/IPCs expansion. We collected remaining Lgr5+ cells from IR-treated intestine samples of Lgr5-EGFP-CreERT and PAF KO mice and examined the expression of target genes of the Wnt, Notch, Hippo-YAP, Hedgehog, and TGF-β/BMP pathways (Figure 4L and S5E). In the regenerating crypts (1 and 2 dpi), remaining Lgr5^low cells from PAF KO mice exhibited a marked decreased expression of Wnt target genes (c-Myc, Cyclin D1, and Lgr5) (Figure 4L and S5E). The well-established Wnt target gene, Axin2, was also decreased in the remaining Lgr5^low cells at 2 dpi. Downregulation of Wnt target genes was also observed in whole crypts fraction at the late time point (4 dpi) (Figure S5F). IHC confirmed that c-Myc and Cyclin D1 expression was significantly downregulated in 4 dpi PAF KO mouse intestine (Figures 5A, 5B, and S5G-S5I). Additionally, we found notable PAF+:Myc+ cells in position 4-6 at 1 dpi, when ISCs are activated for repopulation (Figure 5C). Whereas c-Myc expression was upregulated during regeneration, PAF KO showed a decrease in c-Myc expression (Figures S5G-S5I). These results suggest that PAF might be required for c-Myc upregulation during intestinal regeneration.

Figure 5. Requirement of PAF-Myc Axis for ISCs/IPCs Expansion.

(A, B) Downregulation of c-Myc and Cyclin D1 in PAF KO crypts (4 dpi, 10 Gy). IHC for c-Myc or Cyclin D1. Hematoxylin or DAPI for nuclear counterstaining (blue).

(C) Co-expression of PAF and c-Myc in surviving cells in the regenerating crypts. Arrows: PAF+:Myc+ cells; asterisks: disappeared CBC cells.

(D) Conditional recruitment of PAF and β-catenin to TCF-binding elements (TBEs) in c-Myc and CCND1 (Cyclin D1) proximal promoter. Chromatin immunoprecipitation (ChIP) assays of the mouse small intestine (PAF WT and KO; 1 and 2 dpi; 10 Gy). IgG ChIP and PAF KO small intestine samples served as negative control. RNA Pol II ChIP (positive control for gene transactivation). GAPDH promoter served as negative control.

(E-G) Rescue of PAF KO-induced organoid growth failure by c-Myc expression. The Lgr5+ cells isolated from (PAF WT and KO) were transduced with Retroviruses encoding c-Myc and RFP (red fluorescent protein) and cultured for organoid development. Arrows indicate the budding. The representative images (E); quantification of organoid efficiency (F) and size (G). Asterisks: P<0.05 (N≥30). N.S. (not significant; P≥0.05).

(H, I) Rescue of PAF KO-induced organoid growth failure by ectopic expression of wild-type PAF and PIP mutant PAF (mutPIP-PAF). The Lgr5+ cells isolated from PAF KO;Lgr5-EGFP were transduced with retroviruses encoding wild-type PAF and mutPIP-PAF and cultured for organoid development. The representative images (H); Quantification of organoid size (I) Asterisks: P<0.05 (N≥20).

(J) Illustration of the working model. Upon irradiation injury, the highly proliferative cells (Lgr5^high ISCs and some of Lgr5^low TA cells) undergo apoptosis. PAF and β-catenin transactivate c-Myc in the surviving ISCs/IPCs (Lgr5^low), which leads to the expansion of ISCs/IPCs and the subsequent rebuilding of the intestinal epithelium. Scale bars=50μm (A, B, E, H) and 20μm (C); the representative images were shown from at least three independent experiments. See also Figure S5

As a cofactor of β-catenin transcription complex, PAF upregulates c-Myc expression in CRC (Jung et al., 2013), which led us to determine whether PAF transactivates c-Myc in the regenerating intestine. To test this, we performed a chromatin immunoprecipitation (ChIP) assay of the mouse small intestine using PAF antibody. The small intestine samples from PAF KO mice served as a negative control (Figure 5D; Lane 4). We found that endogenous PAF occupied the TCF-binding element (TBE)-containing proximal promoter of c-Myc and CCND1 (Cyclin D1) in the normal intestine (Figure 5D; Lane 1, 0 dpi). It is noteworthy that despite the massive death of PAF+ cells upon IR, some PAF+ cells survived at 1 dpi (Figures 2L and 2M). Nonetheless, we found that IR conditionally induced the recruitment of PAF and β-catenin to the c-Myc promoter (Figure 5D; lane 3, 2 dpi). These results suggest that PAF is conditionally associated with and transactivates the c-Myc promoter during intestinal regeneration, similar to PAF-induced c-Myc transactivation in CRC cells (Jung et al., 2013).

Next, we asked whether c-Myc mediates PAF-controlled ISCs/IPCs expansion by rescue assay of the Lgr5+ single cell organoid culture. We found that ectopic expression of c-Myc using retrovirus rescued the reduction of organoid-forming efficiency and proliferation in PAF KO Lgr5+ organoids (Figures 5E-5G). These results suggest that PAF-transactivated c-Myc is required for ISCs/IPCs expansion during intestinal regeneration.

Additionally, we performed rescue experiments using wild-type PAF and PIP mutant PAF (mutPIP-PAF) (Jung et al., 2013). It has been suggested that the function of PAF in DNA repair is mediated by the interaction of PAF with PCNA via PIP, a PCNA-interacting protein motif (Emanuele et al., 2011; Povlsen et al., 2012). However, mutPIP-PAF expression was sufficient to recover PAF KO single cell organoids (Figures 5H and 5I), suggesting that PAF-mediated DNA repair through PCNA binding is not involved in growth defect of PAF KO organoids.

Attenuation of Intestinal Tumorigenesis by PAF KO

PAF is significantly upregulated in many human cancers (Cheng et al., 2013; Hosokawa et al., 2007; Jain et al., 2011; Jun et al., 2013; Jung et al., 2013; Kais et al., 2011; Kato et al., 2012; Mizutani et al., 2005; Wang et al., 2016; Yu et al., 2001; Yuan et al., 2007), indicating the potential roles of PAF in promoting tumorigenesis. Additionally, given the pivotal roles of PAF and c-Myc in controlling ISCs/IPCs activation (Figures 5 and S5G-S5I) and initiating intestinal tumorigenesis (Jung et al., 2013; Sansom et al., 2007), we next asked whether PAF contributes to intestinal tumorigenesis by positively modulating the cancer cell stemness. We assessed the expression of PAF in intestinal adenomas driven by Apc mutation using Apc^Min/+ mouse model (Moser et al., 1990). IHC results showed that PAF was markedly upregulated in intestinal adenomas of Apc^Min/+ mice (Figure 6A). sqRT-PCR of individually isolated intestinal adenomas and adjacent normal intestine samples of Apc^Min/+ mice (age of 14 and 16 weeks) also showed the marked upregulation of PAF in Apc mutation-driven adenomas (Figure 6B), which confirms the upregulation of PAF in CRC (Jung et al., 2013). Of note, PAF+ cells were a subpopulation of Ki67+ or PCNA+ cells in adenomas (27.68% and 26.49%, respectively)(Figures S6A-S6D).

Figure 6. Attenuation of Intestinal Tumorigenesis by PAF KO.

(A, B) Expression of PAF in Apc^Min adenomas. Immunostaining of Apc^Min/+ intestinal adenoma (16 wk old) (A), scale bar=50μm; semi-QT-RT-PCR (B). Apc WT: Wild-type (Apc^+/+) intestine sample; N1-4: normal adjacent intestine samples; T1-16: intestinal adenomas.

(C) The extended life span of Apc^Min/+ mice by PAF KO. Kaplan-Meier survival curve of Apc^Min/+ (N=16), PAF Het;Apc^Min/+ (N=15), and PAF KO;Apc^Min/+ (N=17).

(D) H&E staining of the small intestines from Apc^Min/+ and PAF KO;Apc^Min/+ (age of 16 weeks). Asterisks indicate intestinal adenomas. Scale bars=2mm.

(E, F) Decreased tumor burden of Apc^Min/+ mice by PAF KO. The number of tumors (≥1.5mm) (E) and tumor volumes (mm³) (F) were quantified; 16 wk old; N=7 for each group.

(G, H) Reduced cell proliferation of Apc^Min tumors by PAF KO. Ki67 staining of small intestine adenomas from Apc^Min/+ or PAF KO;Apc^Min/+ (16 wk old) (G); quantification (H). Asterisk: P<0.001.

(I, J) Decreased differentiation of the Paneth cells of Apc^Min tumors by PAF KO. Lysozyme staining of small intestine adenomas from Apc^Min/+ or PAF KO;Apc^Min/+ (16 wk old) (I); quantification (J). Asterisk: P<0.001.

(K, L) No change in β-catenin level and activity of Apc^Min tumors by PAF KO. Immunostaining of small intestine adenomas from Apc^Min/+ or PAF KO;Apc^Min/+ (16wk old) for total β-catenin (K) and active (unphosphorylated) β-catenin (ABC) (L). Scale bars=100μm.

(M) Downregulation of c-Myc of Apc^Min tumors by PAF KO. c-Myc IHC; 16 wk old. Scale bars=100μm.

(N) Co-expression of c-Myc and PAF in Apc^Min tumors. Arrows: PAF+:Myc+ cells. Scale bars=20μm.

(O) Downregulation of Cyclin D1 of Apc^Min tumors by PAF KO. Cyclin D1 IHC; 16 wk old. Scale bars=100μm.

The representative images (N≥3) were shown. See also Figure S6.

To test whether genetic ablation of PAF suppresses tumorigenesis in a mouse model, we established Apc^Min/+ (control group) and PAF KO;Apc^Min/+ compound strains (experimental group) and. The median survival of Apc^Min/+ mice was 117 days (N=16). Surprisingly, PAF KO;Apc^Min/+ mice displayed marked extended survival (median 184 days, N=17; P=0.0001) (Figure 6C). Moreover, PAF KO;Apc^Min/+ mice showed a decreased number and size of adenomas, compared to those in Apc^Min/+ mice (Figures 6D, 6F, and S6E). Of note, PAF heterozygous KO (PAF Het;Apc^Min/+) did not affect mouse survival and the number and size of adenomas compared to PAF KO;Apc^Min/+ (Figures 6C-6F). Ki67 IHC showed reduced cell proliferation in adenomas developed in PAF KO;Apc^Min/+ mice compared to that in Apc^Min/+ control mice (Figures 6G and 6H). The number of Paneth cells was also diminished in intestinal adenomas of PAF KO;Apc^Min/+ mice (Figures 6I and 6J). Given that the Paneth cell differentiation is driven by Wnt/β-catenin signaling (van Es et al., 2005), it is possible that PAF KO might induce the overall downregulation of Wnt/β-catenin target genes in tumor cells. Based on our previous finding that PAF hyperactivates the Wnt signaling as a cofactor of the β-catenin transcriptional complex (Jung et al., 2013), we tested whether PAF KO suppresses Wnt/β-catenin signaling in intestinal adenomas. While PAF KO did not affect the level and activity of the β-catenin protein in adenomas (Figures 6K and 6L), PAF KO notably downregulated the expression of Wnt/β-catenin target genes, c-Myc and Cyclin D1, in intestinal adenomas (Figures 6M-6O and S6G). Furthermore, we found that PAF was co-expressed with c-Myc in intestinal adenomas of Apc^Min/+ mice, similar to their co-expression in the regenerating crypts (Figure 6N and S6F). These results suggest that PAF KO attenuates intestinal tumorigenesis by downregulation of β-catenin target genes including c-Myc.

Reduced CRC cell stemness by PAF KO

Previously, we found that PAF-Wnt signaling axis is required for the maintenance of breast cancer cell stemness (Wang et al., 2016). Moreover, having determined that PAF is indispensable for ISCs/IPCs expansion in regeneration (Figures 4), we next examined the role of PAF in controlling the stemness of CRC cells. We analyzed the expression of several known CRC stemness makers (CD44, CD133, and Lgr5) (O’Brien et al., 2007; Ricci-Vitiani et al., 2007; Schepers et al., 2012; Zeilstra et al., 2008; Zhu et al., 2009). Adenomas from PAF KO;Apc^Min/+ mice showed marked decreased expression of CRC stemness markers (CD44, CD133, and Lgr5)(Figures 7A-7D, S7A, and S7B). To better understand the role of PAF in CRC cell stemness, we cultured tumor organoids from Apc^Min/+ and PAF KO;Apc^Min/+ crypts. Owing to hyperactivation of Wnt signaling, Apc^Min/+ organoids develop into the sphere shape (cystic) without budding (Sato et al., 2011a). Surprisingly, the organoids derived from PAF KO;Apc^Min/+ mouse intestine showed severe defects in growth (Figures 7E and 7F). Although the initial organoid forming efficiency was similar (Figure 7G), PAF KO; Apc^Min/+ organoids did not grow until 21 days, compared to the organoids from PAF WT;Apc^Min/+ (Figures 7E and 7F). One copy deletion of PAF (PAF Het;Apc^Min/+) did not suppress the cystic organoid growth (Figure S7C), consistent with in vivo results from PAF Het;Apc^Min/+ mice (see Figures 6C-6F). We further assessed the effects of PAF knockdown on cancer cell stemness using human CRC cells. Colorectal CSCs exhibit the enrichment of CD44 and CD133 expression (Kemper et al., 2010; O’Brien et al., 2007; Ricci-Vitiani et al., 2007). We found the significantly decreased population of CD44+:CD133+ cells in PAF-depleted HT29 cells (Figure S7D). Furthermore, PAF knockdown inhibited colonosphere formation of HT29 (Figure S7E and S7F). These results suggest that PAF is required for the maintenance of CRC cell stemness.

Figure 7. Decreased CRC Cell Stemness by PAF KO.

(A) Gene expression analysis of CD44, CD133, Lgr5, and c-Myc in the normal intestine or adenomas from Apc^Min/+ and PAF KO;Apc^Min/+ (16 wk old). At least three individual samples from WT and normal region of Apc^Min/+ and PAF KO;Apc^Min/+ were used as the control.

(B, C) Downregulation of CD44 and CD133 expression in Apc^Min tumors by PAF KO. IF staining for CD44 and CD133; 16 wk old. Scale bars=100μm.

(D) Downregulation of Lgr5 in Apc^Min tumors by PAF KO. IF for GFP (adenomas of Apc^Min/+;Lgr5-EGFP and PAF KO;Apc^Min/+;Lgr5-EGFP [16 wk old]). Scale bars=100 μm.

(E) Reduced cystic organoid development by PAF KO. Representative images of organoids (day 5) derived from Apc^Min and PAF KO;Apc^Min adenomas

(F, G) Quantification of organoid size (F) (N≥30 cystic organoids were analyzed); efficiency (G) (N≥5000 cells were analyzed for from three independent experiments). Scale bars=100μm. Asterisk: P<0.001.

(H) Heatmap gene expression profile generated by significant differential expression (P<0.05) of Apc^Min/+ and PAF KO;Apc^Min adenomas by RNA sequencing (N=2 per group).

(I) Significantly PAF upregulated signaling pathways identified by GSEA analysis (KEGG) using RNA-seq result. P<0.05.

(J) Gene set enrichment analysis (GSEA) for Wnt signaling pathway. NES: normalized enrichment score; FDR: False detection rate; P-value: Nominal P-value.

(K) Illustration of the working model. During tumorigenesis, Wnt/β-catenin signaling is hyperactivated and PAF is upregulated. PAF and β-catenin transactivate c-Myc and CRC stemness-related genes (CD44, CD133, and Lgr5), which leads to the increase of CRC stemness.

See also Figure S7, Tables S1 and S2.

Next, for an unbiased assessment of PAF-controlled transcriptome in intestinal tumors, we also performed RNA-seq of PAF WT and KO Apc^Min mouse tumors (Figure 7H). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis showed that Wnt signaling is markedly downregulated in PAF KO tumor cells (Figures 7I, S7G, and Table S1), which was also confirmed by Gene Set Enrichment Analysis (GSEA) (Figure 7J and Table S2). These results strongly suggest that PAF positively modulates Wnt/β-catenin signaling in CRC, similar to that during intestinal regeneration.

DISCUSSION

Herein, our comprehensive approaches revealed that the PAF-Myc axis is required for ISCs/IPCs expansion during intestinal regeneration and tumorigenesis.

c-Myc is not required for normal intestine homeostasis but is indispensable for intestinal regeneration (Ashton et al., 2010; Bettess et al., 2005). However, it was unknown how c-Myc contributes to intestinal regeneration. In our experimental setting, ectopic expression of c-Myc rescued PAF KO-induced defects in Lgr5+ single cell organoid growth (see Figures 5E-5G), which strongly suggests that c-Myc mediates PAF-controlled ISCs/IPCs expansion during intestinal regeneration. Interestingly, PAF KO did not affect intestinal homeostasis, which requires constitutively active Wnt signaling in the crypts. Nonetheless, there were no changes in the expression of Wnt/β-catenin target genes (Cyclin D1, CD44, and Lgr5) in PAF KO intestine (without IR injury)(see Figure S3E). However, under specific physiologic or pathologic conditions such as regeneration or tumorigenesis when enhanced Wnt signaling is required (Cadigan and Waterman, 2012; Clevers and Nusse, 2012; Polakis, 2012), the highly upregulated PAF hyperactivates Wnt/β-catenin transcriptional complex and transactivates c-Myc, which subsequently triggers the activation of self-renewing cells (Figures 5J and 7K). This is also supported by the marked upregulation of PAF expression in the regenerating crypts (see Figures 1A-1C, and 2L) and CRC (Jung et al., 2013).

Accumulating evidence suggests the pivotal roles of c-Myc in regulating stem cells. c-Myc is necessary for efficient cellular reprogramming of induced pluripotent stem cells (iPSCs) (Araki et al., 2011; Takahashi et al., 2007). In ESCs, c-Myc modulates cell stemness and differentiation by amplifying protein biosynthesis (Nie et al., 2012; Varlakhanova et al., 2010). c-Myc also regulates the balance between self-renewing and differentiated/committed hematopoietic stem cells by controlling the interaction with their niche (Laurenti et al., 2008; Wilson et al., 2004). Given the specific and dynamic expression of PAF in ISCs/IPCs cells (see Figures 2) and PAF-transactivated c-Myc (see Figure 5), it is probable that PAF may play pivotal roles in governing various stem cells via c-Myc. Indeed, in the small intestine, PAF KO decreased the expression of stem cell marker, Lgr5, in the regenerating crypts accompanied with the downregulation of c-Myc. Furthermore, c-Myc rescued PAF KO-induced failure of organoid growth (see Figures 5E-5G), indicating that PAF functions as an upstream molecule of c-Myc in controlling the stem cells.

70% of human CRC displays a significant upregulation of c-Myc (Erisman et al., 1985; Sikora et al., 1987). In mouse models, genetic ablation of c-Myc suppresses intestinal tumorigenesis driven by Apc mutations (Sansom et al., 2007). Similarly, PAF KO inhibits intestinal adenoma development in Apc^Min/+ mice with reduced c-Myc expression (see Figures 6M and S6G). Moreover, the depletion of PAF decreased the expression of CRC cell stemness markers, (CD44, CD133, and Lgr5) in Apc^Min adenomas (see Figures 7A-7D) with downregulation of c-Myc (see Figures 6M and 7A). Although the effects of PAF KO on CRC cell stemness should be further elaborated with cell ablation, lineage-tracing, and serial transplantation assays, our results are somewhat similar to our previous study that PAF is required for the maintenance of breast cancer cell stemness (Wang et al., 2016). Although Myc depletion rescued phenotypes of tumorigenesis driven by APC mutation in the mouse model (Sansom et al., 2007), the ectopic expression of Myc failed to rescue the organoid growth in PAF KO;Apc^Min/+ condition. These results imply that additional factors/pathways might be involved in PAF-controlled CRC cell stemness. Transcriptomic analysis of Apc^Min/+ and PAF KO;Apc^Min/+ adenomas revealed that in addition to Wnt signaling, other oncogenic pathways including Hedgehog and TGF-G signaling were modulated by PAF (Figures 7H-7J and S7H). Given the oncogenic function of Hedgehog and TGF-G signaling in CRC (Munoz et al., 2006; Takaku et al., 1998; Varnat et al., 2009), it is possible that PAF-controlled colorectal cancer stemness is mediated by several oncogenic pathways including Wnt-Myc, Hedgehog, and TGF-β signaling. Although we here limited our scope to PAF-activated Wnt-Myc axis, it is necessary to further examine how PAF is associated with various oncogenic signalings beyond Wnt/β-catenin pathway.

PAF directly binds to PCNA (De Biasio et al., 2015). Nonetheless, it is highly likely that PAF-controlled stem cells might be independent of PCNA interaction, which is supported by the following results: (a) Without IR treatment, PAF KO is sufficient to suppress organoid growth (Figures 4A-4D and 4I-4K), which rules out the potential involvement of PAF-mediated DNA repair in tissue regeneration; (b) PAF transactivates β-catenin target genes including c-Myc in a PCNA-independent manner (Jung et al., 2013); (c) Instead of DNA repair genes, c-Myc is sufficient to rescue the PAF KO phenotypes in the single cell organoid growth (see Figures 5E-5G); (d) Ectopic expression of PIP-mutated PAF fully rescues the defect of PAF KO organoid growth (see Figures 5H and 5I); (e) Only some of PCNA+ cells are PAF+ cells in the normal intestine (26.49%) and adenomas (23.01%) (see Figures 2D, 2E, S6C, and S6D), indicating the potential roles PAF in a PCNA-independent manner; (f) PAF KO mice showed no abnormality in PCNA expression and IECs growth in the crypts (see Figure 2F); (g) PAF KO did not affect the DNA double-strand breaks and DNA damage foci formation in the intestine (see Figures S4E and S4F), indicating that PAF is dispensable for genomic stability or DNA repair. Thus, these results strongly support that PAF-mediated intestinal regeneration is independent of PCNA and DNA repair pathway.

Although PAF+ cells partially mark ISCs and TA cells in the normal intestine, PAF KO mice did not display defects in tissue homeostasis (see Figures S3). Tissue injury upregulates PAF expression, which enhances c-Myc transcription for the subsequent expansion of ISCs/IPCs. This is also supported by co-expression of PAF and c-Myc in the surviving cells (not the Paneth cells) of regenerating crypts (see Figure 5C). Employing PAF reporter and lineage tracing mice will provide further insights into how PAF+ cells contribute to tissue regeneration and cancer.

Despite the crucial roles of PAF in regulating self-renewing cells in intestinal regeneration and tumorigenesis, it is still unclear how PAF is upregulated in the regenerating crypts and CRC. Due to the high expression of PAF in Apc^Min/+ tumors (see Figures 6A and 6B), it is reasonable that Wnt/β-catenin signaling might directly transactivate PAF. Although the proximal promoter of PAF contains multiple TBEs (Wang et al., 2016), we found that manipulation of Wnt/β-catenin signaling did not affect the transcription of PAF in both IECs and CRC cells (Figures S7I and S7J). Given our previous finding that Oct4, Nanog, and Sox2 transactivate PAF in breast cancer cells and mammary epithelial cells (Wang et al., 2016), the involvement of such iPSC-inducing factors in PAF regulation should be addressed in future studies.

PAF is significantly upregulated in CRC and contributes to tumorigenesis, whereas dispensable for tissue homeostasis. Therefore, PAF might be a viable molecular target for cancer treatment with minimal damage to the normal cells. Conversely, tweaking PAF might provide new ways to manipulate tissue regeneration. Collectively, our findings unveil the essential role of the PAF-Myc signaling axis in controlling stem cell activation in regeneration and cancer.

STAR METHODS

CONTACT FOR REAGENT ANDRESOU RCESHARING

Additional information and requests for reagents should be directed to and will be fulfilled by the Lead Contact, Jae-Il Park at jaeil@mdanderson.org

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mouse Strains

We generated PAF KO strain using CRISPR/Cas9 gene targeting system. Apc^Min/+ (JAX: 002020) and Lgr5-EGFP-IRES-CreERT2 (JAX: 008875) mice were obtained from the Jackson Laboratory. All mice were maintained on a C57BL/6 background except Bmi1^EGFP mice (BKa.Cg-Ptprcb Bmi1^tm1IlwThy1a/J). Both male and female of PAF KO, C57BL/6J (WT control), or Lgr5-EGFP compound strains were used for regeneration experiments in 8-10 weeks of age. For intestinal adenoma quantification and Kaplan-Meier survival curve, only male mice (Apc^Min/+ and PAF KO;Apc^Min/+ compound strains) were used. All mouse experiments were performed under MD Anderson guidelines and Association for Assessment and Accreditation of Laboratory Animal Care International standards.

METHOD DETAILS

Generation of PAF KO Mice

To establish PAF KO mice, two guide RNAs (gRNAs) targeting exon 1 of PAF gene and Cas9 mRNA (Sigma) were injected into the mouse zygotes (The Genetically Engineered Mouse Facility, MD Anderson). gRNA sequences were as follows: #1: 5’-GTTCCCGCCACCGTTTAAATGGG-3’, #2: 5’-ACCAAAGCAAACTACGTTCCAGG-3’. Four strains harboring PAF mutation (gene targeting efficiency: 4/7=57.14%) were identified. Two heterozygote strains (strain 4 and 7) carrying 118bp and 125bp deletion of PAF (PAF null) between the two gRNAs targeted region were selected as founders. Using the strain 7 as a founder, at least 5 times subsequent backcross with C57BL/6 was conducted to minimize the off-target effects. For PCR genotyping of PAF KO following primer pairs and cycling conditions were used: primers: #F: 5’ - AGAATCGAGGTTCTCAAGCG-3’; #R: 5’-CCTTCTAGCTGCTCAATGGG-3’, PCR conditions: 10 min at 95°C, followed by 40 cycles o f 95°C for 15 sec, 65°C for 15 sec, 72°C for 30 sec and post-elongation at 72°C for 5 m in. PAF WT makes 280 bp, and PAF KO makes 155bp (125bp deletion) of PCR product.

Radiation Injury

For intestinal damage, 8-10 weeks old mice were treated with γ-irradiation (10 or 12 Gy) using irradiator (Nasatron). The intestinal regeneration was assessed at 4 and 7 dpi.

Gene Expression Analysis

For small-scale screening of DNA repair gene expression in mice, mRNAs from the irradiated mouse small intestines (1 dpi, 10 Gy) were analyzed with RT² Profiler™ PCR Array Mouse DNA Repair (Qiagen). Control and experimental mice were used for analysis (N=3, each). RNA was extracted from the mouse small intestine and organoids using the TRIzol (Invitrogen) as the manufacturer’s instructions. iScript cDNA synthesis kit (Biorad) with 1 μg of RNA was used for cDNA synthesis. For gene expression analysis of Lgr5+ cells, sorted 100-500 Lgr5+ cells were collected from Lgr5^EGFP or PAF KO;Lgr5-EGFP mice in normal and IR-treated conditions. cDNA was synthesized using REPLI-g WTA Single Cell Kit (Qiagen #150063). The quantitative real-time PCR (qRT-PCR) was performed using a 7500 real-time PCR machine (Applied Biosystems) with specific primers listed in Table S3. Target gene expression was normalized by Hypoxanthine phosphoribosyltransferase 1 (HPRT1) or 18s rRNA. Comparative 2^-ΔΔCt methods were used for quantification of qRT-PCR results.

Crypt Organoid Culture

Crypt organoid culture and Apc-mutated organoid culture were performed based on the previous studies (Sato et al., 2011b; Sato et al., 2009). Briefly, the small intestine samples were opened longitudinally and washed with PBS several times. For extracting the crypt, the small intestine was incubated with 5mM EDTA/PBS for 30 min and the crypt-rich supernatant was collected after vigorous shaking. After filtering through the 100 strainers, the same number (300~500) of crypts were seeded in 50 μl Matrigel (BD). 500 μl of ENR medium (Advanced DMEM/F12 media supplement with EGF (20 ng/ml, Peprotech), Noggin (100 ng/ml, Peprotech) and R-spondin1 (500 ng/ml, Peprotech) were added every two days. For single cell organoid culture, crypts fractions collected through the 70 μm (BD) cell strainer was incubated with Accumax (Stem cell technology 07921) and DNAse (0.8 mg/ml, Sigma) for 30 min. Passaging though the 40 μm strainer, ~5000 Lgr5^high and Cytox Blue (Life technologies) negative cells were collected by cell sorting (MoFlo, Beckman Coulter) and seeded in Matrigel. ENR medium with Jagged-1 peptide (1 μM, AnaSpec) was supplied every 2 days. Organoid efficiency was calculated by counting viable crypts or Lgr5+ cells (at day 3) in total crypts (300~500 per well, crypt organoid) or total seeded cells (N=~2000 per well, single cell organoid). For Apc^Min/+ organoids, adenomas were collected and suspended with Advanced DMEM/F12 media supplement with EGF (20 ng/ml) and Noggin (100 ng/ml). Fresh media were supplemented every 3 days until the next passage (10-14 days). Cystic organoid efficiency was calculated by counting viable cells (at day 3) in total suspended cells (Single cell organoid). The size of organoids was analyzed by measuring the area of the middle section of organoids under the microscope using AxioVision software (Zeiss) (at least 30 organoids per group for all experiments).

Organoid Retrovirus Infection

For organoid gene transduction, we utilized the modified method based on the previous reference (Onuma et al., 2013). FACS sorted Lgr5^high single cells (~5000) from PAF WT and PAF KO mice were incubated with media containing retroviruses expressing c-Myc with RFP (MSCV-c-Myc-IRES-RFP, Addgene [#35395]), or wild-type PAF and mutPIP-PAF (Jung et al., 2013) for 6h at 37 °C with polybr ene (7 μg/ml) and Jagged-1 peptide (1 μM, AnaSpec). Then, infected cells were seeded on 50 μL Matrigel/well in a 12-well plate with conventional ENR media. RFP+ cells were able to observe 2 days after infection. Only RFP+ cells were considered as transfected cells. For selecting nt-PAF or mutPIP-PAF infected organoids, blasticidin (10 μg/ml) was treated.

Lgr5 mRNA Fluorescence In Situ Hybridization (FISH)

WT or PAF KO intestinal tissue sections (0 and 7 dpi) were processed for Lgr5 mRNA FISH according to the manufacturer’s protocol (Invitrogen, FISH TagTM RNA Green Kit, with Alexa Fluor® 488 dye). Probe was designed as 530 base pair length targeting the coding sequence of Lgr5. A probe for sense strand was used as a negative control.

Mammalian Cell Culture and Sphere Formation Assay

CRC Cell line (HT29) was maintained in DMEM media containing 10% fetal bovine serum. For gene depletion, lentiviruses encoding short hairpins against PAF mRNA (MISSION shRNA, Sigma) were stably transduced into target cells and selected using puromycin (2 μg/ml). For counting the CSC population, trypsinized each cell line was counted and incubated with antibodies: CD44v6-APC (1:100, BD-Pharmingen [G44-26]) and CD133-FITC (1: 200, Miltenyi Biotec). Dead cells were excluded by Cytox Blue staining. FACS analysis was performed using FacsJazz Cell Sorter (BD). For sphere formation assay, the limited number of HT-29 cells (5000 cells per ml) were plated in triplicate in the ultra-low attachment plates and grown for six days in serum-free stem cell medium (SCM) supplemented with B27 (Invitrogen), EGF (20 ng/ml, Invitrogen), and bFGF (10 ng/ml, Invitrogen). The number and size of spheres were quantified using AxioVision software (Zeiss).

Chromatin Immunoprecipitation Assay

The mouse small intestines (the duodenum) were minced and cross-linked with 1% formaldehyde for 15 min at room temperature. After quenching by glycine, samples were incubated with lysis buffer (0.5% NP40, HEPES 25 mM, KCl 150 mM, MgCl₂ 1.5 mM, 10% glycerol and KOH pH 7.5) containing proteinase inhibitors. Then the nuclear fractions were collected after centrifugation. Cell lysates were subjected to sonication (10 times, 30s on and 30s off, Bioruptor 300 [Diagenode]) with ChIP-lysis buffer (Tris 50 mM pH 8.0, NaCl 150 mM, 0.1% SDS, 0.5% deoxycholate, 1% NP40 and EDTA 1 mM). Supernatant from lysates was used for immunoprecipitation with the primary antibodies. The following antibodies were used for ChIP: RNA Polymerase II (1μg/ml, EMD Millipore [CTD4H8]), mouse-anti PAF (1 μg/ml, Abcam [ab56773]), and normal mouse IgG (1 μg/ml, Invitrogen). ChIP amplicons were detected by ChIP-PCR using the primers listed in STAR Methods.

RNA-sequencing

The total RNA from Apc^Min/+ and PAF KO;Apc^Min/+ adenomas (two biological replicas) were used for RNA-seq. The transcriptome sequencing was performed by BGI (www.bgi.com/global/) with BGISEQ-500. Reads are mapped by Tophat2 and Differential genes are defined by Cuffdiff with P<0.05. KEGG analysis was performed by DAVID functional annotation analysis. GSEA analysis is performed with normalized FPKM of all genes with default parameters (Number of permutations =1000, collapse dataset=true, permutation type=phenotype).

QUANTIFICATION AND STATISTICAL DETAILS

The Student t-test was applied for comparison of two samples. P-values below the 0.05 were considered significantly different. At least three biological and experimental replicas were used for statistical analyses, otherwise described in Figure legends. Error bars represent standard error (S.E.M).

DATA AND SOFTWARE AVAILABILITY

The accession number for RNA-seq data reported in this paper is GSE109209. All data is available upon request.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse anti-PAF (KIAA0101)	Santa Cruz Biotechnology	Cat# sc-390515
Mouse anti-PAF (KIAA0101)	Abcam	Cat# ab56773
		RRID:AB_943922
Rabbit anti-PCNA	Cell Signaling Technology	Cat# 13110
Rabbit anti-Bmi1	Cell Signaling Technology	Cat# 6964S
		RRID:AB_10828713
Rabbit anti-RFP	Rockland	Cat# 600-401-379S
		RRID:AB_11182807
Rat anti-BrdU	Abcam	Cat# ab6326
		RRID:AB_305426
Rabbit anti-Ki67	Abcam	Cat# ab16667
		RRID:AB_302459
Mouse anti-E-Cadherin	BD	Cat# 610182
	Biosciences	RRID:AB_397581
Rabbit anti-Lysozyme	Abcam	Cat# ab108508
		RRID:AB_10861277
Rabbit anti-Villin	Thermo Fisher Scientific	Cat# PA5-22072
		RRID:AB_11155190
Rabbit anti-Chromogranin A	Abcam	Cat# ab15160
		RRID:AB_301704
Rabbit anti-Phospho-Histone H2A.X (Ser139) (20E3)	Cell Signaling Technology	Cat# 9718S
		RRID:AB_2118009
Rabbit anti-Cleaved Caspase-3 (Asp175) (5A1E)	Cell Signaling Technology	Cat# 9664
		RRID:AB_2070042
Rabbit anti-β-Catenin (D10A8)	Cell Signaling Technology	Cat# 9587S
		RRID:AB_10695312
Rabbit anti-Non-phospho (Active) β-Catenin (D13A1)	Cell Signaling Technology	Cat# 8814s
		RRID:AB_11127203
Rabbit anti-c-Myc (N262)	Santa Cruz Biotechnology	Cat# sc-764
		RRID:AB_631276
Rabbit anti-Cyclin D1 (92G2)	Cell Signaling Technology	Cat# 2978P
		RRID:AB_10839128
Rat anti-CD44	BD	Cat# 550538
	Biosciences	RRID:AB_393732
Rat anti-CD133	eBioscience	Cat# 14-1331-82
		RRID:AB_467471
Mouse anti-P53 (Ab-1)	Thermo Fisher Scientific	Cat# MS-104-P0
		RRID:AB_64407
Rabbit anti-P21 (M-19)	Santa Cruz Biotechnology	Cat#sc-471
		RRID:AB_632123
Rabbit anti-Phospho-Chk1 (Ser345)	Cell Signaling Technology	Cat#2341
		RRID:AB_330023
Mouse anti-ATM pS1981	Rockland	Cat# 200-301-500
		RRID:AB_828098
Chemicals, Peptides, and Recombinant Proteins
5-Bromo-2’-deoxyuridine	Sigma	Cat# B5002
Y-27632 dihydrochloride	Sigma	Cat# Y0503
Recombinant Mouse R-Spondin 1	R&D Systems	Cat# 3474-RS-050
Recombinant Murine Noggin	Peprotech	Cat# 250-38
Recombinant Murine EGF	Peprotech	Cat# 315-09
Jagged-1 peptide	AnaSpec	AS-61298
Critical Commercial Assays
REPLI-g WTA Single Cell Kit	Qiagen	Cat# 150063
Experimental Models: Cell Lines
FHC	ATCC	CRL-1831
		RRID:CVCL_3688
HCT116	ATCC	CCL-247
		RRID:CVCL_0291
SW620	ATCC	CCL-227
		RRID:CVCL_0547
HT29	ATCC	HTB-38
		RRID:CVCL_0320
Experimental Models: Organisms/Strains
Mouse: Lgr5-EGFP-IRES-creERT2 (B6.129P2-Lgr5tm1(cre/ERT2)Cle/J)	The Jackson Laboratory	JAX: 008875
		RRID:IMSR_JAX:008875
Mouse: ApcMin/+ (C57BL/6J-ApcMin/J)	The Jackson Laboratory	JAX: 002020
		RRID:IMSR_JAX:002020
Mouse: TERTTCE/+	Jun et al., 2016	Jae-il Park, jaeil@mdanderson.org
Mouse: Bmi1EGFP (BKa.Cg-Ptprcb Bmi1tm1Ilw Thy1a/J)	The Jackson Laboratory	JAX: 017351
		RRID:IMSR_JAX:017351
Mouse: C57BL/6J	The Jackson Laboratory	JAX: 000664
		RRID:IMSR_JAX:000664
Recombinant DNA
MSCV-c-Myc-IRES-RFP	Kawauchi et al., 2012	Addgene #35395
Sequence-Based Reagents
Primers for qPCR and ChIP, see Table S3	This paper	N/A
Deposited Data
RNA-seq data set	This paper	GEO: GSE109209

HIGHLIGHTS.

• PAF is expressed in intestinal stem cells and upregulated in regenerating crypts

• PAF transactivates c-Myc in intestinal stem cells of the regenerating crypts

• PAF-Myc axis is required for intestinal regeneration

• PAF-Myc axis is required for stemness of colorectal cancer cells