The Drosophila homolog of human tumor suppressor TSC-22 promotes cellular growth, proliferation, and survival

Results

The hypothesis that human TSC22D1 inhibits cell proliferation (4, 10, 13, 32–35) makes the simple prediction that loss of the fly homolog would lead to increased cell number. bun expresses six protein isoforms that have the same DNA-binding and leucine zipper domains but can be divided into two size classes [Fig. 1A and supporting information (SI) Fig. S1]. The four small isoforms, B–E, range from 189 to 254 aa; each has a unique N-terminal domain. The two large isoforms, A (1,206 aa) and F (1,097 aa), share most of a large N-terminal domain and share small sequence motifs with mammalian large TDB proteins (36). Previous studies used transposon insertion alleles that disrupt multiple transcripts (e.g. refs. 19 and 23). We obtained lethal loss-of-function alleles that eliminate the DNA-binding and dimerization domains of only the large isoforms (SI Text and ref. 36). These alleles provided a unique opportunity to investigate the function of TDB large isoforms in the follicular epithelium, where we know the most about bun functions (23, 24). Expression of the large A isoform alone rescues null and heteroallelic mutations, suggesting that the small isoforms may be dispensable (36). Initial characterization of FCs lacking large Bun isoforms suggested that absence of large isoforms has qualitatively similar, but more severe, phenotypes during late oogenesis compared with insertional mutations (data not shown).

We tested for bun regulation of proliferation using clonal analysis (Fig. 1). Paired sister clones, one composed of homozygous bun mutant cells and one of wild-type cells, were generated by FLP/FRT-mediated mitotic recombination (37). Control experiments generated sister pairs of a wild-type GFP− clone and a wild-type GFP+ clone; each averaged nine cells per clone, a ratio of 1 (Fig. 1 B and C). If bun were a negative regulator of proliferation, then each homozygous bun clone would have more cells than its wild-type sister. For the large isoform-specific allele bun[A-Q988X], mutant and wild-type sister clones averaged 6 and 13 cells per clone, respectively, a ratio of 0.45 (Fig. 1 D and E). A lethal transposon insertion with very strong FC phenotypes, bun[6903] (Fig. S1) (23), gave similar results, means of 11 mutant and 20 wild-type cells per clone, a ratio of 0.55 (n = 7, P = 0.01; data not shown). Thus, bun, particularly the large Bun isoforms, promotes FC proliferation.

Premature exit from mitotic cell cycles can lead to fewer cells. FCs proliferate until stage 6, when Notch signaling halts mitosis and initiates endoreplication cycles (27, 28). Because bun blocks Notch signaling during late oogenesis (23), we tested for altered expression of two Notch target genes, cut and emc (38, 39), in proliferating bunA mutant FCs (Fig. S2). Expression was unaffected, indicating that large Bun isoforms do not influence cell number through regulation of these Notch targets.

Studies of bun function in the adult eye indicate that bun influences photoreceptor size (36), so we examined FC size. Immediately after proliferation, there was no difference in mean cell size for mutant FCs (n = 16) compared with sister clone wild-type FCs (data not shown). Because Drosophila cells enlarge chiefly by endoreplication (40), we also examined cell size after endoreplication. At stage 10, bunA mutant FCs were significantly smaller, 200 ± 80 pixels per cell for mutant versus 350 ± 90 for wild type (Fig. 2 A –C and Table S1). We confirmed that bunA mutant FCs undergo all three endoreplication cycles using FACS analysis for DNA content (Fig. 2 D and E). Thus, large Bun isoforms are required for normal growth of FCs.

This requirement in FC growth and proliferation was unexpected. We speculated that large isoforms might regulate growth differently from insertional mutations that additionally disrupt function of small isoforms. To test this, we examined stage-10 FCs mutant for bun[out14], a semiviable allele that disrupts the bunB transcription unit with strong effects on FC morphogenesis (Figs. S1 and S3). Both the large BunA isoform and the small BunB isoform are expressed in FCs (24). Mutant FCs again were significantly smaller than sister clone wild-type FCs, 3.6 ± 1.5 versus 4.7 ± 2.4 arbitrary units (au) respectively (Table S1 and Fig. S3A). We next tested the effects of overexpression, using clonal expression of Gal4 to generate GFP+ cells overexpressing BunA, BunB, or lacZ (Fig. 2 C and Fig. S3 B–E and Table S1). BunB-overexpressing FCs were not significantly different from nearby wild-type FCs. However, BunA-overexpressing FCs were significantly larger than their neighbors, 7.6 ± 2.3 versus 6.8 ± 2.1 au, respectively. We conclude that large Bun isoforms are both necessary and sufficient for normal FC growth.

We speculated that TDB proteins might function differently in cultured cells than in intact epithelia. Thus, we tested the requirement for large Bun isoforms in a nonadherent Drosophila cell line, using double-stranded RNA-mediated interference to deplete BunA and BunF in S2 cells. Double-stranded RNAs (dsRNA) were synthesized from two different sequences unique to the BunA and BunF mRNAs. Control cultures were treated with dsRNA synthesized from the firefly luciferase gene, which is not present in the Drosophila genome. A large polypeptide, detected with antibody against the shared sequences of Bun isoforms, was significantly reduced after 4 days of treatment with either BunA/F dsRNA compared with controls (Fig. 3B).

Consistent with in vivo results, large isoform-depleted cultures proliferated more slowly than control-treated cultures (Fig. 3A). Control cell number increased exponentially between days 3 and 5 after dsRNA treatment. In contrast, numbers of BunA/F-depleted cells increased only slightly during this period. This effect was observed with both dsRNAs, indicating that it is specific to BunA/F. Proliferation rates became markedly different on day 4, so we assayed for apoptotic cells at this time. Significantly, more TUNEL-positive cells were observed in BunA/F-depleted cultures than in controls (Fig. 3D and Fig. S4), indicating that large Bun isoforms promote cell survival in culture.

To determine whether cellular growth was altered by BunA/F depletion, we examined cell size distribution using forward light scatter in flow cytometry. At 12 h, there was no difference in the profile of cell sizes between BunA/F-depleted cultures and controls. A significant shift to smaller sizes for BunA/F-depleted cells was detected at 18 h (Figs. S4C and S5) and became pronounced by 36 h (Fig. 3C), the time at which BunA/F-depletion was detectable by Western blotting (data not shown). More BunA/F-depleted cells were smaller with each additional day, whereas control cells were unchanged. The smallest, apoptotic fraction of BunA/F-depleted cells did not increase above that of the control until day 3 (Fig. S4C). Thus, decreased cellular growth preceded the increase in apoptosis.

Normal cells balance growth with cell division by coordinating cell cycle progression with the rate of protein synthesis (reviewed in refs. 41 and 42). Reduced cell size could arise from decoupling the cell cycle from cellular growth in either of two ways: an increased rate of cell division without a proportionate increase in growth or a decreased rate of cellular growth without a proportionate decrease in proliferation (43). However, there was no consistent difference in the proportion of cells in each phase of the cell cycle between BunA/F-depleted cells and controls (Fig. S4D). If all cell cycle stages are lengthened proportionately, the cell cycle profile appears normal (e.g., ref. 44). To assess the overall cell division rate, we labeled cell contents with the fluorophore carboxyfluorescein diacetate succinimidyl ester (CFSE) (45). CFSE-labeled cytoplasmic contents are divided between daughter cells at each division, so that relative rates of cell division are detected by flow cytometry as decreases in fluorescence over time (Fig. 3E). BunA/F-depleted and control cells had the same fluorescence intensity on day 0, but fluorescence of control cells was substantially lower than BunA/F-depleted cells on day 5. Thus, BunA/F-depleted cells divide less often, indicating that smaller size is not a secondary consequence of accelerated division.

Slow proliferation of BunA/F-depleted S2 cells resulted from a combination of slower cell division and increased cell death, so we assayed for apoptotic cells in mutant FCs (Fig. 4). With good nutrition, no cell death is observed in epithelial follicle cells after stage 1 (reviewed in ref. 46), but approximately half the polar cells undergo apoptosis during stages 1–4 (47). Consistent with this, wild-type sister clones had TUNEL+ FCs only during stages 1–5. In contrast, mutant clones had TUNEL+ FCs throughout oogenesis (Fig. 4B). Furthermore, 26% of bunA mutant clones had TUNEL+ FCs (Fig. 4A) compared with only 4% of wild-type sister clones (data not shown). Control sister clones were indistinguishable (data not shown). Thus, increased apoptosis likely contributes to reduced proliferation of mutant FCs.

As a final test, we overexpressed individual Bun isoforms specifically in polar cells. Excess polar cells undergo apoptosis, leaving only two in each cluster by stage 5 (47). We overexpressed BunA, BunB, or BunC using upd-Gal4 (48). At stage 5 or later, 13% of polar cell clusters with BunA overexpression had at least three cells (Fig. 4D), whereas all control clusters had two cells. Expression of the small-isoform BunC did not alter polar cell number, and the small-isoform BunB had a minimal effect (Fig. 4D). Before stage 5, expression of BunA did not increase the mean number of cells per cluster, and ectopic mitosis was not observed in 69 clusters stained for phosphohistone H3 (data not shown). Prolonged survival was not due to a change in cell fate, because only Fas3+ cells were counted in this experiment; moreover, BunA promotes loss of Fas3 from epithelial FCs (Fig. S6). Thus, increased BunA levels promote polar cell survival, and large Bun isoforms are necessary to promote epithelial FC survival. Tests for genetic interactions with the central regulators of Drosophila apoptosis also indicate that endogenous bun antagonizes apoptosis during eye development (Fig. S7).

Discussion

These data demonstrate that large Bun isoforms promote cellular growth in culture and cell autonomously in vivo. This is a rare phenotype in S2 cells; a dsRNA interference screen found that only ≈4% of Drosophila genes had a cell size, cell cycle, or cell death phenotype (49). Reduced proliferation of large isoform-depleted S2 cells is a secondary consequence of reduced cellular growth and increased apoptosis. Similarly, FCs lacking large Bun isoforms proliferate poorly, in part from increased apoptosis. Large isoforms are required for proliferation in the wing primordium and for full size of adult photoreceptors (36). In adult primordia, slowly growing cells die when they compete with adjacent cells for survival factors (e.g., in ref. 50); it is unknown whether such competition occurs in FCs.

Large Bun isoforms are important to reach normal adult size (19), but genetic studies suggest they are not central to the insulin receptor or TOR signaling pathways (36). FCs lacking the large BunA/F isoforms maintained normal size while proliferating but not during endoreplication. Unlike other genes necessary for final FC size (27, 28, 51), BunA/F do not regulate entry into endocycles via Notch target genes. Importantly, FCs lacking BunA/F can complete all three endocycles by stage 10 of oogenesis. Oogenesis slows substantially under poor nutrient conditions (52), a developmental flexibility that might permit slowly growing FCs to reach normal size. The growth pathway regulated by bun remains to be identified.

This work does not distinguish whether only the large Bun isoforms are necessary for FC growth or whether small isoforms are also needed. Insertion mutations in the bunB transcription unit, bun[6903] and bun[out14], are associated with reduced cellular proliferation and growth, but do not directly test the requirement for bunB because they may alter splicing of other isoforms. In FCs, overexpression of BunA gave increased FC size, but overexpression of BunB had no significant effect. Notably, BunA overexpression rescues size and viability of putative bun null larvae, suggesting that loss of all small isoforms has no effect on larval cell growth (36). Overexpression of BunA prolonged polar cell survival, whereas the effect of BunB was minimal. In contrast, ectopic BunB, but not BunA, could alter expression of downstream genes during late oogenesis (24). These differences in phenotypes may result from differential expression of different isoforms, or from mechanistic differences in the functions of large and small isoforms. Detailed studies of specific phenotypes are needed to conclusively distinguish the functions of large and small isoforms.

Mammalian TDB genes, called TSC22D1, TSC22D2, TSC22D3, and TSC22D4, encode both large and small isoforms, similar to bun (16, 53). All protein isoforms encoded by these genes share the same TSC-22 DNA-binding domain and a leucine zipper motif that permits both homo- and heterodimerization (16). We expected that bun would have a function similar to TSC22D1, the most similar mammalian gene by polypeptide sequence and exon/intron structure (14). Despite this similarity, we find that ectopic BunB does not accelerate apoptosis of polar cells and that cells overexpressing BunB continue to proliferate. This was a surprise, because increased levels of TSC-22 can promote cell death in tumor cell lines (10, 32, 33) and promotes cell cycle arrest induced by TGF-β or PPAR-γ (35), the opposite of the proliferation-promoting functions of large bun isoforms in flies. This apparent paradox may indicate that a different mammalian TDB protein shares the BunB function, such as Gilz, encoded by TSC22D3 (54). Gilz, promotes lymphocyte survival (17) and has functional similarities with BunB (56).

Studies of TSC22D1 focused on only the small protein isoform, TSC-22; the large isoform has not been investigated. In a genetic study of murine TSC22D1, we find similarities to bun growth phenotypes (M. Guitard, C. E. Dohrmann, T. Soma, L. L. Dobens, J. Brissette, and L.A.R., unpublished work), suggesting that this gene also shares in vivo functions with bun. Although TSC22D1 is suspected to be a tumor-suppressor gene for some tissues (7, 8, 13, 32, 55), it is overexpressed in renal cell carcinoma (56). Reduced TSC22D1 expression may promote carcinogenesis only when accompanied by mutations that suppress apoptosis, such as p53 mutations, or in tissues where expression of another TDB gene can rescue growth. Alternatively, reduced expression might promote chemotherapeutic resistance by slowing growth and proliferation (57, 58).

An emerging concept in tumor biology is that mutations can both promote carcinogenesis in one context and decrease cellular fitness in others (59). In flies, bun has context-dependent functions that might antagonize tumor progression. In most FCs, bun is dispensable for epithelial morphology; however, bun mutant FCs are sensitized to migration-inducing signals (23) and display position-dependent loss of monolayer organization (data not shown). Perhaps TSC22D1 similarly prevents neoplasia in the presence of potentially oncogenic signals. Consistent with this model, overexpression of TSC-22 inhibits anchorage-independent growth in salivary-gland tumor cells (11). The functions of bun in the follicular epithelium provide a framework for evaluating the importance of TSC22D1 in tumor progression.

Materials and Methods

Expanded methods are in SI Text.