Diverse Tumor Pathology due to Distinctive Patterns of JAK/STAT Pathway Activation Caused by Different Drosophila polyhomeotic Alleles

Abstract

Drosophila polyhomeotic (ph) is one of the important polycomb group genes that is linked to human cancer. In the mosaic eye imaginal discs, while ph^del, a null allele, causes only non-autonomous overgrowth, ph⁵⁰⁵, a hypomorphic allele, causes both autonomous and non-autonomous overgrowth. These allele-specific phenotypes stem from the different sensitivities of ph mutant cells to the Upd homologs that they secrete.

POLYCOMB group (PcG) genes encode conserved epigenetic regulators that form complexes to repress target gene transcription (Schuettengruber et al. 2007). Drosophila polyhomeotic (ph) encodes a core component of polycomb repressive complex 1 (Shao et al. 1999). Ph plays vital roles in diverse biological functions such as neurodevelopment and cell cycle regulation (Narbonne et al. 2004; Wang et al. 2006; Feng et al. 2011). Deregulation of PcG genes has also been correlated with various types of human cancer (Bracken and Helin 2009). Therefore, elucidating how ph controls cell proliferation has important implications in both basic research and human health.

We have previously reported that different ph alleles cause tissue overgrowth in different ways. Although a ph null allele, ph^del, causes only non-autonomous cell overproliferation, a ph hypomorphic allele, ph⁵⁰⁵, causes both autonomous and non-autonomous cell overproliferation (Feng et al. 2011) (Figure 1, A–C). In mosaic tissues, we define overproliferation of mutant cells as autonomous and overproliferation of genotypically wild-type cells induced by mutant cells as non-autonomous. We have also elucidated the signaling pathway involved in ph^del-induced non-autonomous cell overproliferation. In summary, elevated Notch activity in ph cells up-regulates the expression of JAK/STAT pathway ligand Upd homologs, which in turn activate the JAK/STAT pathway in neighboring wild-type cells and cause their overproliferation (Feng et al. 2011). In this study, we addressed why both a ph null allele and a ph hypomorphic allele cause tumors but in such different ways.

Figure 1.

Notch and upd homologs are required for both autonomous and non-autonomous overproliferation induced by ph⁵⁰⁵. (A–C) ph^del, a ph null allele, induces only non-autonomous overproliferation, while ph⁵⁰⁵, a ph hypomorphic allele, induces both autonomous and non-autonomous overproliferation. Mosaic eye discs of wild-type allele (A), ph^del (B), and ph⁵⁰⁵ (C) were analyzed. ey-flp was used to induce mosaics, and mutant cells were positively labeled by GFP (green) using MARCM (Lee and Luo 1999). DNA was stained with DAPI (blue). (D–G) The removal of Notch or all three upd homologs from ph⁵⁰⁵ cells suppresses the enlarged eye phenotype induced by ph⁵⁰⁵. Adult eyes mosaic for wild-type allele (D), ph⁵⁰⁵ (E), ph⁵⁰⁵-Notch (F), and ph⁵⁰⁵-upd^d1-3 (G) were analyzed. ey-flp was used to induce mosaics. upd^d1-3 is a deletion that lacks all three upd homologs (Feng et al. 2011). To remove Notch or all three upd homologs specifically from ph⁵⁰⁵ cells in mosaic eyes, ph⁵⁰⁵-Notch and ph⁵⁰⁵-upd^d1-3 double-mutant lines were generated and were used to perform mosaic analyses (F and G). (H–K) Notch and upd homologs are required not only for non-autonomous but also for autonomous overproliferation induced by ph⁵⁰⁵. Eye discs mosaic for wild-type allele (H), ph⁵⁰⁵ (I), ph⁵⁰⁵-Notch (J), and ph⁵⁰⁵-upd^d1-3 (K) were stained with PH3 (red), a mitotic marker. ey-flp was used to induce mosaics, and mutant cells were positively labeled by GFP (green). DNA was stained with DAPI (blue). Note that when Notch or all three upd homologs were removed from ph⁵⁰⁵ cells, both non-autonomous and autonomous overproliferation was suppressed.

We first tested whether the same signaling pathway underlay non-autonomous overproliferation induced by both ph^del and ph⁵⁰⁵. The functions of Notch and Upd homologs in the ph⁵⁰⁵ mosaic eyes were examined with the same strategy used for ph^del (Feng et al. 2011). A ph⁵⁰⁵-Notch double-mutant line was generated, and eyes mosaic for this line were essentially of the same size as wild-type eyes (Figure 1, D vs. F). The mosaic eye discs had normal size and normal cell proliferation level, as shown by PH3 staining, which marks mitotic cells (Figure 1, H vs. J). Moreover, the size of ph⁵⁰⁵-Notch clones was significantly reduced when compared to that of ph⁵⁰⁵ clones (Figure 1, I vs. J). These results indicated that Notch was required for both autonomous and non-autonomous overproliferation induced by ph⁵⁰⁵.

We next recombined ph⁵⁰⁵ with upd^d1-3, a deficiency line that lacks all three upd homologs in the Drosophila genome (Feng et al. 2011). Mosaic analyses were then performed using this double-mutant line. ph⁵⁰⁵-upd^d1-3 mosaic eyes were significantly smaller than ph⁵⁰⁵ mosaic eyes and were comparable to wild-type eyes (Figure 1, D, E, and G), indicating that tissue overgrowth was largely suppressed. PH3 staining of the double-mutant mosaic eye discs showed that these discs had relatively normal size and cell proliferation level (Figure 1, H vs. K). Importantly, ph⁵⁰⁵-upd^d1-3 clones were also drastically reduced in size compared to ph⁵⁰⁵ clones (Figure 1, I vs. K). These results indicated that Upd homologs are required not only for non-autonomous but also for autonomous cell overproliferations induced by ph⁵⁰⁵.

It is not surprising that the same signaling pathway is responsible for non-autonomous overproliferation induced by both ph^del and ph⁵⁰⁵, and it is not completely unexpected that Notch is also required for ph⁵⁰⁵-induced autonomous overproliferation, as Notch is a transcription factor that has been shown to autonomously regulate cell proliferation (Artavanis-Tsakonas and Muskavitch 2010). However, the three Upd proteins are secreted ligands (Hombría et al. 2005; Arbouzova and Zeidler 2006) and are not expected to have any direct effect on autonomous cell proliferation. To interpret our observations, we hypothesized that ph⁵⁰⁵ cells still respond to Upd ligands secreted by themselves in an autocrine or paracrine manner and therefore overproliferate. On the other hand, ph^del cells were no longer responsive to Upd ligands.

To functionally test this hypothesis, we again applied the double-mutant strategy, taking advantage of the fact that the genes domeless [dome, encoding the only transmembrane receptor of the Drosophila JAK/STAT pathway (Brown et al. 2001)] and hopscotch [hop, encoding the only Drosophila JAK kinase (Binari and Perrimon 1994)] are also on the X chromosome as is ph. We first recombined ph⁵⁰⁵ with two dome alleles to generate ph⁵⁰⁵-dome double-mutant lines. Eye discs mosaic for these lines were still significantly larger than wild type, but the size of double-mutant clones was dramatically reduced, so that only a tiny portion of the disc was composed of mutant cells (Figure 2, A and B). PH3 staining indicated that the non-autonomous proliferation level was still high, but autonomous proliferation largely disappeared (Figure 2, A and B). We further examined the adult eyes mosaic for such double-mutant lines and found that these eyes were still much larger than wild type and similar to ph⁵⁰⁵ mosaic eyes in size, but they generally were not folded as seen in ph⁵⁰⁵ mosaic eyes (Figure 2, H, J, and K).

Figure 2.

The JAK/STAT pathway is involved in autonomous overproliferation induced by ph⁵⁰⁵. (A–C) The removal of dome or hop from ph⁵⁰⁵ cells suppresses autonomous but not non-autonomous overproliferation. Mosaic eye discs of wild-type allele (A), ph⁵⁰⁵-dome (B), and ph⁵⁰⁵-hop (C) were stained with PH3 (red), which marks mitotic cells. ey-flp was used to induce mosaics. Mutant cells were labeled by GFP (green), and DNA was stained with DAPI (blue). To remove dome or hop from ph⁵⁰⁵ cells, ph⁵⁰⁵-dome and ph⁵⁰⁵-hop double-mutant lines were generated and were used for mosaic analyses. (D–F) Eye discs mosaic for wild-type allele (D), ph^del-dome (E), and ph^del-hop (F) were stained with PH3 (red) as controls. ey-flp was used to induce mosaics. Mutant cells were labeled by GFP (green), and DNA was stained with DAPI (blue). (G–O) When the JAK/STAT pathway components dome or hop were removed from ph⁵⁰⁵ or ph^del cells in mosaic eyes, the eyes were still much larger than wild type. Adult eyes mosaic for wild-type allele (G), ph⁵⁰⁵ (H), ph^del (I), two ph⁵⁰⁵-dome double-mutant lines with different dome alleles (J and K), ph⁵⁰⁵-hop (L), two ph^del-dome double-mutant lines with different dome alleles (M and N), and ph^del-hop (O) were analyzed. ey-flp was used to induce mosaics.

Next we generated a ph⁵⁰⁵-hop double-mutant line. We found that autonomous proliferation in mosaic eye discs of this double mutant was also significantly suppressed, with mutant cells accounting for only a small portion of the whole disc. On the other hand, non-autonomous cell overproliferation was not affected, and the overall size of these discs was still significantly larger than wild type (Figure 2C). Adult eyes mosaic for this double mutant showed similar phenotypes as those of ph⁵⁰⁵-dome mosaic eyes. These eyes were still significantly larger than wild type, but they were generally not folded (Figure 2L). Therefore, the removal of either dome or hop from ph⁵⁰⁵ cells only suppressed autonomous overproliferation and did not affect non-autonomous overproliferation, making such double-mutant mosaic discs phenotypically similar to ph^del mosaic discs.

As controls, ph^del-dome and ph^del-hop double-mutant lines were also generated using the same dome and hop alleles. Mosaic analyses on eye discs showed that the removal of dome or hop from ph^del cells did not affect non-autonomous cell overproliferation. It did, however, mildly reduce the mutant clone size (Figure 2, D–F), suggesting that ph^del cells might still have a weak response to Upd ligands. Adult eyes mosaic for these double-mutant lines were phenotypically indistinguishable from ph^del mosaic eyes (Figure 2, I and M–O), consistent with the above observations in mosaic eye discs.

Finally, we asked why ph^del and ph⁵⁰⁵ cells respond differently to the Upd ligands secreted by themselves. We hypothesized that some of the JAK/STAT pathway modulators might be differentially expressed in ph^del and ph⁵⁰⁵ cells. To test this hypothesis, we chose TU-tagging, a technique that enables the purification of RNA from mutant cells without having to physically isolate such cells (Miller et al. 2009). Briefly, Drosophila is unable to synthesize uridine monophosphate from uracil due to the lack of phosphoribosyltransferase (UPRT). When exogenous UPRT is expressed in mutant cells by MARCM (Lee and Luo 1999), such cells acquire the ability to utilize uracil. If these larvae are fed with 4-thiouracil (4-TU), a uracil derivative that contains a thio group, only mutant cells would be able to use 4-TU and eventually incorporate thio-containing uridine into newly synthesized RNA. This treatment has little toxicity, and the thio-labeled RNA can be purified from total RNA using conventional biochemical methods.

We performed TU-tagging to isolate RNA from ph^del cells and ph⁵⁰⁵ cells and used quantitative RT-PCR to examine candidate gene expression (Figure 3A). The expression of the JAK/STAT pathway receptor dome was significantly higher in ph⁵⁰⁵ cells than in ph^del cells. A higher receptor expression might sensitize ph⁵⁰⁵ cells to the Upd ligands. The levels of enok and socs42a, both negative regulators of the JAK/STAT pathway (Arbouzova and Zeidler 2006; Müller et al. 2008), were also significantly higher in ph⁵⁰⁵ cells compared to ph^del cells. This might represent feedback loops that negatively regulate the pathway activity. In fact, several such negative feedback loops, in which elevated pathway activity upregulates a negative pathway regulator, have been reported in JAK/STAT pathway (Arbouzova and Zeidler 2006).

Figure 3.

Molecular mechanism underlying different responses of ph^del and ph⁵⁰⁵ cells to Upd ligands. (A) TU-tagging followed by real-time PCR revealed differential expression of JAK/STAT pathway components and major regulators. UPRT and GFP were expressed in mutant cells by MARCM with ey-flp. Mid-third instar larvae were fed with 4-TU for 10–12 hr. Total RNA was extracted from eye discs, labeled, and purified according to the published protocols (Miller et al. 2009). Real-time PCR was then performed using purified RNA (and total RNA for tubulin and GFP only). The GFP-to-tubulin ratio was three to six times higher in purified RNA than in total RNA, indicating that TU-tagging successfully enriched RNA from mutant cells. (B and C) Models of overproliferation patterns and underlying signaling pathways in ph^del (B) and ph⁵⁰⁵ (C) mosaic eye discs.

Together, we conclude that both ph^del and ph⁵⁰⁵ cause autonomous overexpression of Upd homologs in mutant cells, which represents the only driving force of cell overproliferation in ph^del and ph⁵⁰⁵ mosaic tissues and in essence acts non-autonomously to activate JAK/STAT pathway. The different phenotypes of these two types of mosaics are due to different sensitivity of mutant cells to Upd homologs. ph⁵⁰⁵ mutant cells robustly respond to Upd ligands that they secrete. Therefore, Upd ligands secreted by ph⁵⁰⁵ cells simultaneously induce overproliferation in both mutant and wild-type cells. In contrast, ph^del cells are largely insensitive to Upd ligands, so that Upd ligands secreted by ph^del cells induce only overproliferation in wild-type but not mutant cells. Furthermore, differential expression of the JAK/STAT pathway receptor dome might underlie the different sensitivity of ph^del and ph⁵⁰⁵ cells to Upd ligands. Models of cell proliferation patterns and the underlying signaling pathways in ph^del and ph⁵⁰⁵ mosaic tissues are given in Figure 3, B and C.