Inhibition of Notch signaling by the p105 and p180 subunits of Drosophila chromatin assembly factor 1 is required for follicle cell proliferation Free

Nucleosomes undergo disassembly and assembly processes during DNA replication and DNA repair. Histone chaperones are critical factors mediating these processes and act by guiding the trafficking of histones and depositing them onto DNA during replication-coupled chromatin assembly (De Koning et al., 2007; Ransom et al., 2010). Chromatin assembly factor 1 (CAF1) is one of these histone chaperones and mediates the deposition of histone H3/H4 onto newly synthesized DNA (Smith and Stillman, 1991; Kaufman et al., 1995; Gaillard et al., 1996; Verreault et al., 1996). Drosophila CAF1 is composed of three subunits, CAF1-p180, CAF1-p105 and CAF1-p55 (also known as CAF1-180, CAF1-105 and CAF1-55, respectively), which correspond to human p150, p60 and p48 (also known as CHAF1A, CHAF1B and RBBP4, respectively). Drosophila CAF1-p55 and human CAF1-p48 are present not only in the CAF1 complex but also in a multitude of chromatin-modulating complexes, suggesting that CAF1 has multiple functional roles, and is not restricted to acting as a histone chaperone (Kaufman et al., 1995).

Emerging evidence has shown that CAF1 plays crucial roles in the development of multicellular organisms, including Drosophila, by regulating heterochromatin formation, signal transduction and transcription (Yu et al., 2015). These studies suggest that CAF1 not only functions as a histone chaperone promoting replication-coupled cell cycle progression, but also serves as a protein complex adaptor that integrates epigenetic regulation by interacting with heterochromatic proteins and/or transcriptional factors. This epigenetic role enables CAF1 to orchestrate cell proliferation and cell differentiation in specific cellular contexts (Yu et al., 2015). Recently, we determined that CAF1 activates Notch signaling, and that this is essential for the development of wing imaginal discs (Yu et al., 2013). Notch signaling also regulates another Drosophila developmental process, oogenesis, although the role of CAF1 here is so far unknown and is the focus of this study.

Drosophila oogenesis is a developmental process that involves highly regulated differentiation of germline and somatic follicle cells (Deng and Bownes, 1998; Klusza and Deng, 2011). From the germarium to stage 6 of oogenesis, Drosophila follicle cells undergo multiple rounds of mitosis with archetypal cell cycle phases (G1, S, G2 and M phases) to increase their number to 650 cells, which form a monolayer to cover 16 germline cells (Deng and Bownes, 1998; Deng et al., 2001; Shcherbata et al., 2004). Starting at stage 7 and ending at stage 10A, follicle epithelial cells undergo three rounds of endocycle (also called endoreplication; it duplicates genomic DNA without cell division in each cycle) to generate 16 copies of genomic DNA in each follicle nucleus (Edgar and Orr-Weaver, 2001; Shu et al., 2018). The transition from the mitotic cycle to endocycle is characterized by the sudden loss of mitotic cyclins (e.g. Cyclin A and Cyclin B) and markers [e.g. phospho-histone 3 (PH3)] (Bradbury, 1992; Deng et al., 2001; Hendzel et al., 1997), increased expression of S-phase-specific cyclins (e.g. Cyclin E) (Follette et al., 1998), and decreased expression of immature cell-fate markers (e.g. Eyes absent; Eya) (Lopez-Schier and St Johnston, 2001; Sun and Deng, 2005). At stage 10b, main-body follicle cells stop undergoing endocycle and proceed to amplify some specific gene areas (e.g. the chorion gene region; this stage is therefore called the gene amplification stage) (Calvi et al., 1998; Cayirlioglu et al., 2001; Sun et al., 2008). The differentiation processes for germline and somatic follicle cells during oogenesis are highly regulated in a temporal and spatial manner through their complicated cell–cell communication and subsequent signal transduction as well as transcriptomic reprogramming (Deng and Bownes, 1998; Klusza and Deng, 2011). Therefore, the developmental process of oogenesis must involve dynamic changes in nucleosomal conformation and accessibility to transcriptional machinery, which play critical roles in transcriptomic reprogramming. Although CAF1 is a crucial regulator in controlling the dynamics of nucleosomal conformation and determining transcriptomic reprogramming, its role in Drosophila oogenesis remains elusive.

The switch from the mitotic cycle to endocycle (M/E) during follicle cell differentiation is known to be elicited by Notch signaling (Deng et al., 2001; Lopez-Schier and St Johnston, 2001; Bray, 2006). At stage 6 of oogenesis, expression of the ligand Delta (Dl) is upregulated in germline cells, subsequently activating Notch signaling in follicle cells through Presenilin (Psn)-mediated proteolysis of Notch and the release of the Notch intracellular domain (N^ICD) (Struhl and Greenwald, 1999). The N^ICD translocates to the nucleus and causes Suppressor of Hairless [Su(H)] to shift from being a transcriptional repressor to an activator, which in turn activates Notch target gene expression (Struhl and Adachi, 1998). Disruption of Notch signaling in follicle cells by mutation, or siRNA-mediated knockdown of signaling components, blocks the M/E switch and results in continued cell proliferation as well as impaired cell differentiation (Deng et al., 2001; Lopez-Schier and St Johnston, 2001; Palmer et al., 2014).

Notch signaling is known to modulate the expression of cell-cycle regulatory genes to elicit the M/E switch of follicle cells. Notch signaling has been shown to negatively regulate the expression of String, the Drosophila homolog of the Cdc25 phosphatase, to promote the M/E switch by blocking follicle cell mitosis (Schaeffer et al., 2004; Shcherbata et al., 2004). In contrast, at the M/E switch Notch signaling upregulates the expression of Fizzy-related (Fzr, also known as Cdh1) (Schaeffer et al., 2004; Shcherbata et al., 2004). Fzr, a WD40-repeat protein, functions to regulate the anaphase-promoting complex/cyclosome (APC/C) and to promote G1 progression through proteolysis of mitotic regulators (e.g. Aurora-A kinase and Cyclins A, B and B3) in an APC/C-dependent manner (Sigrist and Lehner, 1997; Castro et al., 2002). Therefore, Fzr upregulation by Notch signaling promotes endoreplication in follicle cells through the inhibition of mitotic regulators (Schaeffer et al., 2004; Shcherbata et al., 2004).

We have previously shown that during the M/E switch, the activated Notch pathway downregulates the expression of cut, a homeodomain gene normally expressed in the mitotic follicle cells (Sun and Deng, 2005). Suppression of Cut is required for Notch-mediated upregulation of Fzr expression, resulting in the initiation and promotion of endocycle (Sun and Deng, 2005). Given that Cut is also required for maintaining the undifferentiated status of follicle cells during the mitotic stage in addition to its essential role in cell proliferation, Notch-mediated downregulation of Cut also promotes follicle cell differentiation during the endocycle stage (Sun and Deng, 2005). Moreover, we have identified a novel Notch target called Hindsight (Hnt, also known as Pebbled), a zinc-finger transcription factor, which is required and sufficient to block follicle cell proliferation and induce the M/E switch (Sun and Deng, 2007). Mechanistically, during the M/E switch Hnt downregulates the expression of Cut and String and inhibits Hedgehog signaling. These Hnt-mediated effects inhibit follicle cell proliferation and promote the occurrence of endocycle as well as follicle cell differentiation (Sun and Deng, 2007).

We recently determined that CAF1 is functionally required for in vivo Notch signaling activity during the development of wing imaginal discs (Yu et al., 2013). Drosophila CAF1 activates Notch signaling by associating with Su(H) and N^ICD to establish an active local chromatin state and activate the expression of Notch target genes, respectively (Yu et al., 2013). Here, we studied the role of CAF1 in Drosophila ovarian follicle cell differentiation. Unexpectedly, we found that CAF1-p105 and CAF1-p180 suppress Notch signaling activity to maintain follicle cell proliferation during the mitotic stage. We also revealed the regulatory relationship between CAF1-p105, CAF1-p180, Notch and Notch-regulated genes (Hnt and Cut) in ovarian follicle cells. This study reveals a novel paradigm wherein CAF1 functions to sustain cell proliferation by serving as a dual factor that can positively and negatively regulate Notch signaling in a tissue-context-dependent manner.

To examine the role of CAF1-p105 in follicle cells, homozygous clones of CAF1-p105³⁶, a null allele (Yu et al., 2013), were generated by crossing FRT42D,CAF1-p105³⁶/T(2;3) with hsFLP;FRT42D,M,GFP/CyO and subsequently using the FLP/FRT-induced mitotic recombination technique (see Fig. S1). After mitotic recombination, follicle cell clones with homozygous CAF1-p105³⁶ were GFP negative, which could be distinguished from GFP-positive follicle cell clones with wild-type CAF1-p105 alleles or heterozygous CAF1-p105³⁶. Owing to the critical role of the homeobox transcription factor Cut in maintaining mitosis of follicle cells (Sun and Deng, 2005) and the essential role of CAF1 in cell proliferation (Ramirez-Parra and Gutierrez, 2007; Yu et al., 2015), we first examined the expression pattern of Cut in mosaic egg chambers with homozygous CAF1-p105³⁶ follicle cell clones. Cut has been shown to be expressed in all follicle cells during mitotic stages 1–6 of oogenesis, then is absent from the main-body cells during endocycle stages (stages 7–10a) and reappears afterward (Fig. 1A). Immunostaining with a monoclonal anti-Cut antibody revealed that Cut expression was dramatically downregulated in homozygous CAF1-p105³⁶ follicle cell clones at the mitotic stage (stage 4 and 5 egg chambers are shown in Fig. 1B). The penetrance for this phenotype was low (14.8%, analyzed number of clones=27) when egg chambers were harvested 2 days after heat shock. However, the penetrance dramatically increased to 83% (analyzed number of clones=47) when egg chambers were harvested 3 days after heat shock. In contrast, Cut expression was not affected in control homozygous FRT42D follicle cell clones during these stages (Fig. 1C), demonstrating that Cut downregulation in early-stage follicle cells was caused by loss of CAF1-p105 function. To determine whether two other CAF1 components, CAF1-p180 and CAF1-p55, have comparable effects on Cut expression, we examined their respective mutant alleles, CAF1-p180²⁷⁰ (Song et al., 2007b) and CAF1-p55^7-150 (Wu et al., 2012), and also the UAS-RNAi line for CAF1-p180 knockdown (Yu et al., 2013). Homozygous CAF1-p180²⁷⁰ and CAF1-p55^7-150 follicle cell clones (indicated by GFP-negative cells) were generated by crossing their respective mutant lines with the hsFLP; FRT42D,M,GFP/CyO line and subsequently inducing FLP/FRT-induced mitotic recombination as described above. CAF1-p180 RNAi-expressing follicle cell clones were generated by crossing the UAS-CAF1-p180 RNAi line with the heat shock-inducible GAL4 driver line (hsFLP;act<y+<Gal4,UAS-GFP/Cyo) with the UAS-GFP transgene. After heat shock, generated RNAi-expressing follicle cell clones were GFP-positive as heat shock-induced expression of the GAL4 driver drove expression of both RNAi and GFP through binding of GAL4 to the UAS promoters of the CAF1-p180 RNAi and the GFP transgene (Fig. S2). The knockdown effect of CAF1-p180 RNAi was confirmed previously (Yu et al., 2013). Through these studies, we found that Cut expression was also lost in homozygous CAF1-p180²⁷⁰ follicle cell clones during the mitotic stage (84.6%, n=13) (Fig. 1D). Consistent with this, CAF1-p180-RNAi-expressing follicle cell clones induced by a Flip-out GAL4 also showed significant downregulation of Cut expression before the M/E switch (79.4%, n=63) (Fig. 1E). In contrast, Cut expression was not affected in homozygous CAF1-p55^7-150 mutant follicle cell clones during these early stages (n=34) (Fig. 1F). These findings, taken together, indicate that CAF1-p105 and CAF1-p180, but not CAF1-p55, are required for maintaining Cut expression in mitotic follicle cells.

Cut plays a pivotal role in maintaining mitosis and restricting follicle-cell differentiation before the M/E switch (Sun and Deng, 2005). To determine whether CAF1-p105 and CAF1-p180 are also required for sustaining the mitotic cycle in follicle cells, we immunostained phosphorylated histone H3 (PH3), a well-known mitotic marker in Drosophila (Hendzel et al., 1997; Wang et al., 2001; Karam et al., 2010), in mosaic egg chambers. We found that inactivation of CAF1-p105 resulted in suppression of histone H3 phosphorylation (Fig. 2A,B) in follicle cells before stage 6 (100 follicle cells were analyzed per experiment, n=3, P<0.01). This indicates that loss of CAF1-p105 results in suppressing mitosis of follicle cells. We were aware how follicle cells deficient for CAF1-p105 or CAF1-p180 (hereafter denoted CAF1-p105/p180) could have grown into cell clones when Cut expression was downregulated. As mentioned above, high penetrance of Cut downregulation only occurred 3 days after heat shock, indicating that this 2-day time period with Cut expression allowed CAF1-p105/p180-deficient mitotic follicle cells to grow into cell clones. We next examined whether both CAF1-p105 and CAF1-p180 are required for maintaining expression of Eya, a key protein essential for sustaining the immature follicle cell fate during mitotic stages (Fig. 1A) (Bai and Montell, 2002). The homozygous CAF1-p105³⁶ follicle cell clones (indicated by GFP-positive cells) were created by crossing FRT42D,CAF1-p105³⁶/T(2;3) with FRT42D MARCM and subsequently using FLP/FRT-induced mitotic recombination (see Fig. S3). We found Eya expression was downregulated in the CAF1-p105/p180 loss-of-function cells (79.4% for homozygous CAF1-p105³⁶ follicle cell clones, n=34; 81.5% CAF1-p180-RNAi-expressing follicle cell clones, n=54) (Fig. 2C,D). To determine whether the Eya downregulation seen upon loss of CAF1-p105/p180 was attributable to Cut downregulation, we performed the immunostaining analysis of Eya in Cut-knockdown follicle cells. As shown in Fig. 2E, knockdown of Cut led to downregulation of Eya in follicle cells at the mitotic stage (100%, n=47), demonstrating Cut is required for Eya expression in mitotic follicle cells. These findings together demonstrate that CAF1-p105 and CAF1-p180 are functionally required for maintaining proliferation and the immature cell fate of follicle cells at the mitotic stage.

To unravel the mechanism underlying Cut downregulation in CAF1-p105/p180-deficient follicle cells at the mitotic stage, we hypothesized that this phenotype might have resulted from premature activation of Notch signaling. To test this possibility, we analyzed the expression of Hnt (a Notch target gene expressed in follicle cells at endocycle stages; see Fig. 1A) (Sun and Deng, 2007) in homozygous CAF1-p105³⁶ and CAF1-p180 RNAi-expressing follicle cell clones at the mitotic stage. The homozygous CAF1-p105³⁶ follicle cell clones were created by crossing FRT42D,CAF1-p105³⁶/T(2;3) with either hsFLP;FRT42D,M,GFP/CyO (Fig. 3A) or FRT42D MARCM (Fig. 3B) and subsequently using FLP/FRT-induced mitotic recombination as previously described. The immunostaining results showed that upregulated Hnt expression was detected in homozygous CAF1-p105³⁶ follicle cell clones before the M/E switch (93.5%, n=31 for Fig. 3A; 92.9%, n=42 for Fig. 3B). Owing to the low frequency of homozygous CAF1-p180²⁷⁰ clone generation, we created CAF1-p180 RNAi-expressing follicle cell clones instead to examine the impact of CAF1-p180 deficiency on Hnt expression. This experiment showed that CAF1-p180 knockdown also resulted in early Hnt upregulation in follicle cells during the mitotic stage (83.8%, n=37) (Fig. 3C). To determine whether upregulation of Hnt is reflective of overall Notch signaling activation in CAF1-p105³⁶ mutant follicle cell clones, the expression of the Notch activity reporter E(spl)m7-LacZ (Song et al., 2007a) was examined in CAF1-p105³⁶ follicle cell clones. Expression of LacZ was found in the mutant follicle cell clones at the mitotic stage (86.8%, n=53) whereas the wild-type counterparts displayed no expression of LacZ (except in polar cells, which are known to have active Notch signaling; Chen et al., 2011) (Fig. 3D). CD2 expression from another Notch reporter, E(spl)mβ-CD2 (Song et al., 2007a), was also detected in CAF1-p180-knockdown follicle cell clones at the mitotic stage. This result was reproducible in knockdown experiments using two different CAF1-p180 RNAi fly lines (89.6%, n=48 for RNAi#1; 85.3%, n=34 for RNAi#2) (Fig. 3E,F). These data together indicate that CAF1-p105 and CAF1-p180 are involved in preventing premature Notch target gene expression during the mitotic stage of follicle cell development.

The negative effect of CAF1-p105 and CAF1-p180 on Notch target expression in early follicle cells that we observed above differs from the role of CAF1 in Notch signaling in the wing imaginal disc development, where CAF1 positively regulates the expression of Notch target genes (Yu et al., 2013). To determine whether CAF1 regulates Notch target gene expression differently in endoreplicating follicle cells, where Notch signaling has been activated, we examined the expression of Hnt in homozygous CAF1-p105³⁶ mutant and CAF1-p180-RNAi-expressing follicle cell clones, and the expression of the Notch reporter E(spl)m7-LacZ in homozygous CAF1-p105³⁶ follicle cell clones. The results showed that the expression of Hnt (n=32) and E(spl)m7-LacZ (n=40) was not affected in homozygous CAF1-p105³⁶ mutant follicle cell clones at the endocycle stage (Fig. 4A,B). Consistent with this, Hnt expression was not altered in CAF1-p180 RNAi-expressing follicle cell clones (n=44) at the endocycle stage (Fig. 4C). These findings suggest that CAF1-p105 and CAF1-p180 are dispensable for Notch target gene expression in follicle cells after the M/E switch. Therefore, results here and above suggest that the regulatory relationship between CAF1 and Notch signaling is dependent on the context.

Our previous studies have shown that Hnt is responsible for Cut downregulation during the M/E transition (Sun and Deng, 2007). Given that the inhibition of CAF1-p105/p180 elicits Hnt induction, this finding raises a possibility that the observed Cut downregulation is attributable to Hnt upregulation. To test this possibility, we conducted an epistasis study. Hnt RNAi caused a successful knockdown of Hnt in follicle cells at the endocycle stage (Fig. 5A). Co-knockdown of CAF1-p180 and Hnt restored Cut expression in follicle cells at the mitotic stage (100% rescue, n=16) (Fig. 5B), demonstrating that Hnt upregulation in CAF1-p105/p180-deficient follicle cells is responsible for Cut downregulation. This finding suggests that CAF1-p105 and CAF1-p180 maintain Cut expression in mitotic follicle cells via inhibiting Hnt expression. Consistent with our previous finding (Sun and Deng, 2007), Hnt knockdown led to significant Cut upregulation in follicle cells at the endocycle stage (100%, n=46) (Fig. 5B). The Hnt-knockdown-induced upregulation of Cut expression in endoreplicating follicle cells was not affected by co-knockdown of CAF1-p180 (Fig. 5B), which is expected as CAF1-p105/p180 regulates Cut expression in an indirect manner via Hnt.

Given that Cut functions as a transcriptional regulator, Cut may play a role in CAF1-p105/p180-dependent suppression of Notch target gene expression in mitotic follicle cells. To test this possibility, we performed a Cut knockdown study. Immunostaining of Cut confirmed the knockdown effect of Cut RNAi (Fig. 6A). The immunostaining data showed that Hnt expression was upregulated in Cut-knockdown follicle cell clones at the mitotic stage (100%, n=47) (Fig. 6B). This result was confirmed with another Cut RNAi (data not shown). To test whether Cut also negatively regulates other Notch target genes, we examined the expression of E(spl)m7-LacZ in Cut-knockdown follicle cell clones at the mitotic stage. We found that expression of E(spl)m7-LacZ was not upregulated in Cut-knockdown follicle cell clones at the mitotic stage (n=38) (Fig. 6C), suggesting that Cut is not involved in regulating expression of the Notch target gene E(spl)m7. To demonstrate that Cut downregulation in CAF1-p180-deficient follicle cells at the mitotic stage is the cause of Hnt upregulation, we performed an epistasis study to examine whether ectopic Cut overexpression is able to rescue the Hnt upregulation phenotype observed in CAF1-p180 RNAi-expressing follicle cell clones at the mitotic stage. Immunostaining of Cut confirmed that UAS-Cut drove the ectopic expression of Cut in follicle cells at the endocycle stage (Fig. 6D). We found that ectopic expression of Cut abolished the Hnt upregulation induced by CAF1-p180 knockdown (100%, n=35) (Fig. 6E). These results demonstrate that CAF1-p180 negatively regulates Hnt expression via Cut in mitotic follicle cells.

In CAF1-p105/p180 loss-of-function follicle cells prior to the M/E switch, Cut is downregulated, whereas Hnt and Notch target gene reporters are upregulated. Because these events are associated with activation of Notch signaling, we next asked whether Notch is responsible for these events observed in CAF1-p105/p80-deficient follicle cells.

To unravel whether Notch is required for the premature activation of Notch target genes in CAF1-p105/p180-deficient follicle cells at the mitotic stage, epistasis studies were performed. The double knockdown of both Notch and CAF1-p180 in mitotic follicle cells by the Flip-out method (Fig. S2) rescued the CAF1-p180-deficient phenotypes (Cut downregulation, induction of the Notch reporter E(spl)mβ-CD2, and Eya downregulation) observed in CAF1-p180 RNAi-expressing follicle cell clones at the mitotic stage (100% rescue; n=21 for Cut staining; n=18 for CD2 staining; n=15 for Eya staining) (Fig. 7A–C). This result is consistent with another rescue study showing that expression of the Notch RNAi in homozygous CAF1-p105³⁶ MARCM mutant follicle cell clones at the mitotic stage also rescued the CAF1-p105-deficient phenotypes, including Cut downregulation (100% rescue, n=22) and Hnt upregulation (100% rescue, n=27) (Fig. 7D,E). Immunostaining of the Notch intracellular domain (N^ICD) confirmed the Notch knockdown effect of the RNAi in homozygous CAF1-p105³⁶ MARCM mutant cell clones (Fig. 7F). These findings, taken together, demonstrate that Notch is mandatory for premature activation of Notch target gene expression and subsequent Cut downregulation in CAF1-p105/p180-deficient follicle cells at the mitotic stage.

In this study, our novel findings have revealed that CAF1-p105 and CAF1-p180 are essential for the progression of the mitotic stage during Drosophila ovarian follicle cell differentiation because they prevent premature activation of Notch signaling (Fig. 8). Under CAF1-p105/p180 deficiency, premature activation of Notch signaling is induced to suppress follicle cell proliferation via Hnt-mediated downregulation of Cut expression (Fig. 8), a transcription factor functionally required for follicle cell mitosis (Sun and Deng, 2005). Moreover, we for the first time found that maintenance of Cut expression by CAF1-p105/p180 in mitotic follicle cells is crucial for negative regulation of Hnt expression. According to these findings, we hypothesize that Cut is an intrinsic repressor of Hnt expression, and activation of Notch signaling by CAF1-p105/p180 deficiency, or by Dl, overcomes the inhibitory effect of Cut and activates Hnt expression, which subsequently downregulates Cut expression in a negative-feedback manner. Therefore, our studies reveal a novel negative-feedback loop in the regulatory relationship between Hnt and Cut (Fig. 8). Our current findings and previous discovery that Hnt is required for downregulating Cut to promote the M/E switch and endocycle of follicle cells (Sun and Deng, 2007) indicate the pivotal role of this negative-feedback regulatory loop between Cut and Hnt in the progression of the mitotic stage, M/E switch and the endocycle stage during follicle cell differentiation.

The differential roles of CAF1-p105 and CAF1-p180 in wing imaginal disc development (Yu et al., 2013) and follicle cell differentiation suggest that CAF1-p105 and CAF1-p180 play a dual role in regulating the Notch signaling pathway depending on the tissue context (Fig. 8). Interestingly, loss of CAF1-p55 failed to activate Notch signaling in mitotic follicle cells, indicating that CAF1-p55 is involved in activating Notch signaling in wing imaginal discs (Yu et al., 2013), but not in repressing Notch signaling in mitotic follicle cells. In wing imaginal disc development, activation of Notch signaling is required for promoting cell proliferation via induction of Cut expression (Yu et al., 2013). In contrast, Notch is not activated during the mitotic stage of ovarian follicle cell differentiation until the transition from the mitotic stage to the endocycle stage (M/E switch). After the M/E switch, activated Notch signaling inhibits follicle cell mitosis by downregulating Cut expression and drives endoreplication in differentiated follicle cells. By comparing these two scenarios, we found a common phenomenon wherein CAF1 is necessary for proliferation of wing imaginal disc and ovarian follicle cells via activating Cut expression, which is achieved by activating and inhibiting Notch signaling activity, respectively.

Our studies also reveal that CAF1-p105/p180 is required for maintaining the immature cell fate of ovarian follicle cells through Cut-dependent activation of Eya expression (Fig. 8). These observed events are consistent with the physiological role of CAF1 in promoting DNA replication as well as cell proliferation and maintaining immature cell fate (Ramirez-Parra and Gutierrez, 2007; Yu et al., 2015). Therefore, we hypothesize that there are tissue-specific mechanisms determining the regulatory role of CAF1-p105 and CAF1-p180 in the Notch signaling pathway. Moreover, we also observed that CAF1 is functionally linked to Cut in both wing imaginal disc and follicle cell studies. This suggests that the functional connection between CAF1 and Cut is physiologically important for cell proliferation during the development of organisms. Our findings are in line with studies of Arabidopsis plants showing that CAF1 is essential for pollen cell mitosis as loss of CAF1 results in delay and arrest of the cell cycle during pollen development (Chen et al., 2008).

It is unclear how CAF1-p105 and CAF1-p180 suppress Notch target gene expression in mitotic follicle cells. However, in our previous studies of Drosophila embryos and an insect cell model (S2 cells), we found that CAF1 protein subunits associate with Su(H) (Yu et al., 2013). It is known that Su(H) is involved in maintaining the repressive chromatin state for inhibiting Notch target gene expression when it is complexed with other repressor subunits (e.g. Hairless, Gro and CtBP). Therefore, it is possible that CAF1-p105/p180 can associate with the Su(H) repressor complex to participate in inhibiting Notch target gene expression. This hypothesis is supported by our preliminary finding that CtBP knockdown in follicle cells results in phenotypes (e.g. Cut downregulation, Hnt upregulation, E(spl)m7-LacZ upregulation, Eya downregulation) highly similar to those caused by CAF1 deficiency (data not shown). This functional analogy between CAF1 and CtBP suggests that CAF1 is functionally linked to the Su(H) repressor complex in mitotic follicle cells. However, further studies are needed to address whether inactivation of other repressor components can lead to similar phenotypes and whether CAF1-p105/p180 can associate with the Su(H) repressor complex in mitotic follicle cells. Another puzzle is that Notch is required for premature induction of Notch target gene expression in CAF1-p105/p180-deficient follicle cells at the mitotic stage. This finding raises questions with regard to how Notch is activated when CAF1-p105/p180 is deficient and whether this Notch activation is dependent or independent of ligand (Palmer and Deng, 2015). Further investigations are necessary to address these questions.

In conclusion, our findings have revealed that CAF1-p105 and CAF1-p180 are physiologically required for proliferation of follicle cells during the mitotic stage through their intrinsically inhibitory effect on Notch signaling. Therefore, our study shows that CAF1 has functions beyond its role in DNA replication and plays an essential role in proliferating cells through multiple mechanisms involving its dual role in positively and negatively regulating the Notch pathway activity in a tissue-context-dependent manner.

The following fly stocks were used in this study: FRT42D, CAF1-p105³⁶/T (2;3) (Yu et al., 2013); CAF1-p180²⁷⁰, FRT101/FM7c (Song et al., 2007b); FRT82B, CAF1-p55^7-150/TM6B, Tb (Wu et al., 2012); UAS-CAF1-p180-RNAi/TM3, Sb (Bloomington, BL28918, TRiP ID# HM05129); UAS-CAF1-p180-RNAi (BL32478, TRiP ID# HMS00480); UAS-Cut RNAi (BL29625, TRiP ID# JF03304; and BL33967, TRiP ID# HMS00924); UAS-N RNAi (BL27988, TRiP ID# JF02959); UAS-Hnt-RNAi/CyO (VDRC, V101325); UAS-Cut/CyO (Sun and Deng, 2005); E(spl)m7-lacZ (Assa-Kunik et al., 2007); E(Spl)mβ-CD2 (de Celis et al., 1998). The following fly strains were used to generate mosaic mutant or RNAi-expressing clones: yw, FLP¹²²; FRT42D, M, GFP/CyO; FRT42D MARCM (Tamori and Deng, 2013); yw, FLP¹²², FRT101, ubi-GFP; hsFLP;; FRT82B, ubiGFP/TM3,Sb; yw, hsFLP; act<y+<Gal4, UAS-GFP/Cyo. The following generated fly strains were used in the study: FRT42D,CAF1-p105³⁶/CyO;E(spl)m7-LacZ/TM6B; E(spl)mβ-CD2/CyO;UAS-CAF1-p180 RNAi (BL28918 or BL32478)/TM6B; FRT42D, CAF1-p105³⁶/CyO;UAS-N RNAi/TM6B; hsFLP;act<y+<Gal4, UAS-GFP/Cyo;UAS-N RNAi/TM6B; UAS-Hnt-RNAi/CyO;UAS-CAF1-p180-RNAi/TM6B; UAS-Cut/CyO;UAS-CAF1-p180-RNAi/TM6B.

Flies were raised under standard conditions at 25°C. Mitotic clones were generated by FLP/FRT-mediated recombination. Adult female flies were heat shocked twice daily for 1 h at 37°C and incubated at 25°C for 3–4 days for generation of loss-of-function follicle cell clones. MARCM mutant clones were generated by heat shock of adult flies twice daily for 1 h at 37°C. For generation of flip-out clones, adult females were heat shocked for 45 min at 37°C and incubated at 25°C for 3–4 days before dissection.

Immunofluorescence staining and image acquisition were carried out as described previously (Deng et al., 2001; Sun and Deng, 2005). The following antibodies were used: mouse anti-Cut, 1:15 (2B10); mouse anti-Eya, 1:10 (eya10H6); mouse anti-Hnt (1G9), 1:15 (1G9); mouse anti-N^ICD, 1:50 (C17.9C6) (all from the Developmental Studies Hybridoma Bank; DSHB); rabbit anti-PH3, 1:200 (catalog no. 06-570, Millipore); rabbit anti-β-galactosidase, 1:5000 (catalog no. 08559761, MP Biomedicals) and mouse anti-CD2, 1:50 (catalog no. MCA154G, AbD Serotec). Secondary antibodies were used at the designated dilutions: Alexa Fluor 546 goat anti-mouse IgG (1:500) and Alexa Fluor 546 goat anti-rabbit IgG (1:500) (Molecular Probes). Nuclear DNA was stained with DAPI (Invitrogen). Images were captured on a Zeiss LSM 510 confocal microscope.

Data are presented as mean±s.d. A Student's t-test was used to analyze variances between two experimental datasets. P<0.05 was considered significant. Data were analyzed using the GraphPad Prism software (version 6.0; GraphPad Software, Inc).

We thank Jen Kennedy for critical reading of the manuscript. We thank the Developmental Studies Hybridoma Bank (DSHB), Vienna Drosophila RNAi Center (VDRC) and Bloomington Drosophila Stock Center (BDSC), the TRiP at Harvard Medical School for providing antibodies and fly stocks. We thank the Biological Science Imaging Core facility at Florida State University for technical support.