Inverse regulation of two classic Hippo pathway target genes in Drosophila by the dimerization hub protein Ctp

Abstract

The LC8 family of small ~8 kD proteins are highly conserved and interact with multiple protein partners in eukaryotic cells. LC8-binding modulates target protein activity, often through induced dimerization via LC8:LC8 homodimers. Although many LC8-interactors have roles in signaling cascades, LC8’s role in developing epithelia is poorly understood. Using the Drosophila wing as a developmental model, we find that the LC8 family member Cut up (Ctp) is primarily required to promote epithelial growth, which correlates with effects on the pro-growth factor dMyc and two genes, diap1 and bantam, that are classic targets of the Hippo pathway coactivator Yorkie. Genetic tests confirm that Ctp supports Yorkie-driven tissue overgrowth and indicate that Ctp acts through Yorkie to control bantam (ban) and diap1 transcription. Quite unexpectedly however, Ctp loss has inverse effects on ban and diap1: it elevates ban expression but reduces diap1 expression. In both cases these transcriptional changes map to small segments of these promoters that recruit Yorkie. Although LC8 complexes with Yap1, a Yorkie homolog, in human cells, an orthologous interaction was not detected in Drosophila cells. Collectively these findings reveal that that Drosophila Ctp is a required regulator of Yorkie-target genes in vivo and suggest that Ctp may interact with a Hippo pathway protein(s) to exert inverse transcriptional effects on Yorkie-target genes.

Introduction

The LC8 family of cytoplasmic dynein light-chains, which includes vertebrate LC8 (aka DYNLL1/DYNLL2) and Drosophila Cut-up (Ctp), are small highly conserved proteins that are ubiquitously expressed and essential for viability^1,2,3,4. The LC8 protein is 8 kilodaltons (kD) in size and was first identified as an accessory subunit in the dynein motor complex, within which an LC8 homodimer binds to and stabilizes a pair of dynein intermediate chains (DIC)^1,5,6. However, the LC8 protein has since emerged as a general interaction hub with multiple dynein/motor-independent roles and binding partners^3,7,8. In fact the majority of LC8 protein in mammalian cells is not associated with either dynein or microtubules¹ and LC8 orthologs are encoded in the genomes of flowering plants that otherwise lack genes encoding heavy-chain dynein motors⁹.

Accumulating evidence has reinforced the idea that the primary role of LC8 in mammalian cells is to facilitate dimerization of its binding partners via LC8 self-association, a mechanism that has been termed ‘molecular velcro’⁷. LC8 can be found in association with over 40 proteins that function in diverse cellular processes, including intracellular transport, nuclear translocation, cell cycle progression, apoptosis, autophagy and gene expression^8,10. LC8 is found in both the nucleus and cytoplasm and can interact with partners in either compartment. For example the mammalian kinase Pak1 binds and phosphorylates LC8 in the cytoplasm, which in turn enhances the ability of LC8 to interact with the BH3-only protein Bim and inhibit its pro-apoptotic activity^11,12. Accordingly, overexpression of LC8 or the phosphorylation of LC8 by Pak1 enhances survival and proliferation of breast cancer cells in culture¹². LC8 also binds estrogen receptor-α (ERα) and facilitates ERα nuclear translocation, which in turn recruits LC8 to the chromatin of ERα-target genes^13,14,15. In the cytoplasm, LC8 is also found in association with the kidney and brain expressed protein (KIBRA), which is an upstream regulator of the Hippo tumor suppressor pathway¹⁶. KIBRA binding potentiates the effect of LC8 on nuclear translocation of ERα, suggesting crosstalk may occur between LC8-regulated pathways¹⁵. The KIBRA-LC8 complex also interacts with Sorting Nexin-4 (Snx4) to promote dynein-driven traffic of cargo between the early and recycling endosomal compartments¹⁷. Thus, LC8 has been linked to a variety of proteins in both the cytoplasm and nucleus that play important roles in signaling, membrane dynamics and gene expression.

Drosophila Ctp differs from vertebrate LC8/DYNLL by only four conservative amino acid substitutions across its 89 amino acid length. Similar to mammalian LC8, phenotypes produced by Ctp loss in flies imply roles in multiple developmental mechanisms. Drosophila completely lacking Ctp die during embryogenesis due to excessive and widespread apoptosis^2,18. Partial loss of Ctp function causes thinned wing bristles and morphogenetic defects in wing development, as well as ovarian disorganization and female sterility². Within salivary gland cells, Ctp promotes autophagy during pupation¹⁹, while in neuronal stem cells it localizes to centrosomes and influences mitotic spindle orientation and the symmetry of cell division²⁰. Testes mutant for ctp have motor-dependent defects in spermatagonial divisions as well as motor-independent defects in cyst cell differentiation²¹. A recent study linked ctp mRNA expression to the zinc-finger transcription factor dASCIZ and showed that knockdown of either Ctp or dASCIZ reduces wing size²². In sum, this diversity of effects produced by Ctp loss in different Drosophila cell types suggest that Ctp plays important yet context specific roles in vivo. However our knowledge of molecular pathways that require Ctp and in turn underlie these developmental phenotypes associated with Ctp loss, remain poorly characterized.

Here we use a genomic null allele of ctp and a validated ctp RNAi transgene to assess the role of the Ctp/LC8/DYNLL protein family in pathways that act within the developing Drosophila wing epithelium. We find that clones of ctp null cells are quite small relative to controls and that RNAi depletion of Ctp shrinks the size of the corresponding segment of the adult wing without clear defects in mitotic progression or tissue patterning. The effect of Ctp depletion on adult wing size is primarily associated with a reduction in cell size, rather than cell division or cell number, implying a role for Ctp in supporting mechanisms that enable developmental growth. In assessing the effect of Ctp loss on multiple pathways that control wing growth, we detect robust effects on one–the Hippo pathway. The Hippo pathway is a conserved growth suppressor pathway that acts via its core kinase Warts to inhibit nuclear translocation of the coactivator Yorkie (Yki), which otherwise enters the nucleus, complexes with the DNA-binding factor Scalloped (Sd) and activates transcription of growth and survival genes^23,24,25,26. In parallel to the effect of Ctp loss on clone and wing size, Ctp loss alters expression of the classic Yki target genes bantam and thread(th)/diap-1 in wing pouch cells. Parallel genetic tests confirm a requirement for ctp in Yki-driven tissue growth in the wing or eye. Quite unexpectedly however, Ctp loss has opposing effects on bantam and diap1 transcription in wing pouch cells: bantam transcription is strongly elevated while diap1 expression is strongly decreased in cells lacking Ctp. In each case, these effects map to small segments of DNA in the ban and diap1 promoters that recruit Yki transcriptional complexes^24,25,27. Epistasis experiments confirm that Yki is required to activate the bantam promoter in Ctp-depleted cells and that transgenic expression of Yki can overcome the block to diap1 transcription. In sum these data argue that Ctp supports physiologic Hippo signaling in wing disc epithelial cells and that Ctp likely interacts with an as yet unidentified Hippo pathway protein(s) to exert inverse transcriptional effects on Yorkie-target genes. These types of inverse effects have not previously been described within the Hippo pathway and imply that distinct subsets of genes within the Yorkie transcriptome can be simultaneously activated and repressed in developing tissues via a mechanism that involves Ctp.

Materials and Methods

Fly Strains

All crosses were maintained at 25 °C unless otherwise noted. Alleles used in these studies (Bloomington stock number indicated) are as follows: thread-lacZ (#12093), ex697 (ex-lacZ, #44248), e2f1rM729 (e2f1-lacZ, #34054), UAS-yki-V5 (#28819), yki^B5 (#36290), mirr^DE-Gal4 (#29650), EcRE-lacZ (#4517), Stat92E-10XGFP (#26198) (courtesy of R.Read), UAS-diap1 (#6657), UAS-p35 (#5073), UAS-dcr2;enGal4,UAS-GFP (#25752) obtained from the Bloomington Drosophila Stock Center. UAS-yki-IR (v104523) and UAS-ctp-IR (v43116) were obtained from the Vienna Drosophila Resource Center (VDRC). Other alleles used were ctp^ex3 (Gift of Bill Chia), enGal4/CyO, HRE-lacZ, HREmut6-lacZ, 2B2-lacZ, 2B2C-lacZ (all gift of D.J. Pan), brC12-lacZ (K. Irvine), GMR-Gal4, UAS-ykiS168A:GFP (K. Harvey), E-spl(m)ß-CD2, Su(H)-lacZ, PCNA-GFP, ban-3xGFP and UAS-yki (D.J. Pan).

Immunofluorescence Microscopy

Immunostaining and confocal microscopy performed as described previously on a Zeiss 510 inverted confocal microscope²⁸. Primary antibodies include mouse anti-β-Gal 1:1000 (Promega); rabbit anti-LC8 (1:100) (W. Sale), mouse anti-Ci (1:50), mouse-α-Notch 1:10, mouse anti-Wg (1:800), mouse anti-Cut (1:100) and mouse anti-DIAP1 (1:50) (DSHB); rabbit α-yki 1:1000 (K. Irvine); mouse α-rat CD2 1:100 (Research Diagnostics, Inc.); mouse anti-V5 (1:200, Invitrogen); mouse anti-dpERK (1:10,000, Sigma); rabbit anti-phospho histone H3 (1:100), rabbit anti-phospho-Smad1/5 (1:100), rabbit anti-DCP1 (1:150), rabbit anti-Caspase3 (1:100) (Cell Signaling); mouse anti-dMyc (monoclonal P4C4-b10); mouse anti-Cyclin E (H. Richardson). BrdU assays performed as described previously (Robinson et al. 2010) with mouse anti-BrdU (1:100; Becton Dickinson).

Wing/Eye Measurements

Eyes/wings were imaged on a Leica DFC500 CCD camera and quantified with Image J. Posterior compartment ratio (PCR) = posterior compartment size/total wing size.

Results

Ctp is required for imaginal-disc derived adult tissues to grow to normal size

To test the role of ctp in the developing wing disc, a UAS transgene encoding a ctp RNA interference (RNAi) cassette (Vienna line #43116) was expressed in the posterior compartment of the larval wing disc at 25 °C using the engrailed-Gal4 driver (enGal4). Depletion of Ctp protein in posterior cells of en > ctp-IR discs was confirmed by immunostaining with an antibody raised against the Chlamydomonas reinhardtii homolog of Ctp that also cross-reacts with metazoan LC8/Ctp proteins¹ (Figure S1). Ctp-depleted larval wing discs develop into adult wings with a fully penetrant phenotype of a shrunken posterior blade (Fig. 1A–C). This effect occurs without obvious scarring of the adult wing or major disruption in its pattern of venation, with the exception of occasional truncated cross veins (e.g. wing in Fig. 1B). Ctp-depletion with a second larval driver, mirr^DE-Gal4, which is active in the dorsal half of developing larval discs²⁹, shrinks the dorsal half of the adult eye (more darkly pigmented region above the dotted lines in Fig. 1D,E), contracts the dorsal surface of the adult head and leads to small, thin thoracic bristles (Figure S2). The Ctp-RNAi small-bristle phenotype matches a thin-bristle phenotype observed in adult flies carrying the hypomorphic ctp mutation, ctp^ins1 and further confirms the validity of the RNAi cassette². As in the wing, Ctp depletion did not obviously scar the dorsal domain of the adult eye or strongly disrupt its surface organization, suggesting that Ctp depletion impairs growth of these organs but less so the mechanisms that drive their morphological patterning.

Figure 1

ctp loss reduces compartment size and clonal growth.

Paired images of control (A,D) and Ctp-depleted (B,E) adult wings and eyes generated using the en-Gal4 (A,B) or mirr^DE-Gal4 (AKA dorsal eye (DE)-Gal4) (D,E) drivers in combination with the ctp^IR transgene. en is expressed in posterior wing cells (below dotted line in (A,B)). DE is expressed in dorsal eye cells (above dotted line in (D,E)). (C) Box-plot quantitation of the wing posterior compartment ratio (PCR = P area/P + A areas) of the genotypes in (A) (n = 22) and (B) (n = 31). Standard error of the mean (SEM) is indicated. *p < 0.0001. (F) Confocal image of a ctp mosaic L3 wing disc (ctp^ex3,FRT19A/ubi-GFP,FRT19A,hsFlp) containing GFP-negative ctp^ex3 clones (arrows) and their paired GFP-positive twinspots. (G,H) Histogram plots of the two-dimensional area of multiple ctp^ex3 clone:twinspot pairs in the hinge/notum (G) (n = 29) or pouch (H) (n = 47) of L3 wing discs. Each pair of bars on the X-axis represents a distinct clone:twinspot pair. The clones are significantly smaller than their paired twinspots, *p = 0.0002 for (G), *p < 0.0001 for (H). (I) Bar graph plot of average ratios of ctp^ex3 clone:twinspot sizes in the hinge/notum versus the pouch. SEMs are indicated. *p = 0.0034.

To better assess the effects of Ctp loss on cell proliferation and survival, ctp mutant clones (GFP-) and their wildtype twinspots (GFP+) were generated by combining a heatshock(hs)-induced Flp mitotic recombination system with the ctp^ex3 null allele². Within dissected larval wing discs, GFP-deficient ctp^ex3 clones could be found in the pouch, hinge and notum at 48 hrs post-clone induction (Fig. 1F). In the hinge and notum, these ctp null clones can reach a fairly large sizes (see arrows in Fig. 1F) but are nonetheless ~2-fold smaller than their control twinspots (Fig. 1F,G,I). This trend toward small clones is more pronounced in the pouch where ctp-null clones are occasionally missing and consistently quite small relative to their age-matched twinspots (e.g. see Fig. 1F). A quantitative measurement across multiple pouch clone pairs with extant twinspots shows an approximate 5-fold reduction in ctp mutant clone area (Fig. 1H,I). This apparently unequal growth deficit elicited by ctp loss in hinge/notum cells vs. pouch cells could nonetheless stem from perturbation of an otherwise ubiquitous cellular process, such as the mitotic spindle defects observed in ctp mutant neuronal stem cells²⁰. Alternatively, it could indicate a role for Ctp in a molecular pathway or process that is especially important in pouch cells, similar to the cell-type specific and motor-independent requirement for ctp in germline and somatic cyst cells²¹.

Ctp loss elevates apoptosis and division of wing disc cells and reduces their size

To better assess the effects ctp loss on division and survival of larval wing disc cells, larval wing cells depleted of Ctp were probed with antibodies to cleaved caspase-3 (cC3), the incorporated nucleotide base analog bromodeoxyuridine (BrdU) and the mitotic marker phospho-histone H3 (pH3) (Fig. 2A–C). cC3 is moderately elevated in the posterior compartment of en > ctp-IR wing discs (Fig. 2A,A′) and an antibody to the cleaved caspase Dcp-1 detects elevated apoptosis in ctp^ex3 mutant wing pouch cells (Figure S3). This finding in disc cells parallels the widespread apoptosis that occurs in ctp mutant embryos² and raises the possibility that death of ctp mutant cells may deprive the developing wing of a sufficient pool of cells to achieve normal size. To test this hypothesis, two transgenes encoding either the Drosophila protein Diap1, which blocks caspase cleavage, or the baculoviral protein p35, which inhibits active cleaved caspases³⁰, were co-expressed with ctp-IR from the en driver (Fig. 2D). Neither of these anti-apoptotic transgenes were able to significantly rescue small en > ctp-IR wings, indicating that cell death is unlikely to be the sole cause of wing blade undergrowth induced by chronic Ctp-depletion.

Figure 2

Effects of Ctp loss on the division, survival and size of wing cells.

Confocal images of en > ctp^IR L3 wing discs stained for (A′) cleaved Caspase-3 (cC3), (B′) the base analog BrdU, or (C′) phospho-histone H3 (PH3). Anti-Cubitis interruptus (Ci) staining marks the anterior domain (A–C). (D) Quantitative analysis of PCR in the indicated genotypes. en > diap1 (n = 13) is significantly larger than en > diap1 + ctp^IR (n = 12) (*p < 0.0001) and en > p35 (n = 10) is significantly larger than en > p35 + ctp^IR (n = 17) (*p < 0.0001). (E–E″) Box-plot quantitation and images of wing hair cell density in a fixed area in the posterior wing between the L4 and L5 wing veins in control (en > + ) (n = 10) or Ctp-depleted (en > ctp^IR) (n = 9) adult wings. SEMs are indicated. *p < 0.0001. (F) Comparative effects of Ctp-depletion (en > + vs. en > ctp^IR) on cell size (data in 2E) and PCR (data in 1C). en > + values are standardized to 100%.

The early mitotic marker phospho-histone H3 (pH3) is increased among Ctp-depleted larval wing pouch cells (Fig. 2C,C′). Taken in isolation, this higher abundance of pH3-positive ctp-IR larval wing cells could be indicative of an accelerated rate of mitotic entry or a slower rate of M-phase transit. Although disrupting LC8 function in cultured mammalian cells does not impede mitotic progression³¹, a past study in Drosophila attributed an excess-pH3 phenotype in Ctp-depleted cells to dynein-motor defects that block cells in mitosis and thus enrich for M-phase markers²². This mitotic-block model logically predicts that Ctp-depleted cells accumulate in M-phase and thus depopulate other phases of the cell cycle. However, BrdU-incorporation analysis of ctp-IR wing pouch cells does not show a depletion of S-phase cells and in fact seems to show an increase in the frequency of S-phase entry (Fig. 2B,B′). A very similar increase in cell division occurs in Drosophila germ cells lacking ctp²¹. Intriguingly ctp loss reduces cell division among neural stem cells³², suggesting that Ctp plays distinct proliferative roles in different cell types. In combination, the pH3 and BrdU data presented here argue that Ctp-depleted larval wing disc cells are able to actively transit between the mitotic and DNA synthesis phases of the cell cycle, suggesting that a cell cycle block is not the prime cause of small ctp-IR adult wings.

Genetic manipulations that reduce cell size can also shrink Drosophila adult organs. Thus, the lack of experimental evidence pointing to excess apoptosis or a proliferative deficit among Ctp-depleted cells prompted analysis of the effect of Ctp loss on cell size. Each cell in the adult wing generates a single hair, enabling a quantitative determination of cell density derived from hair counts within a fixed region of the adult wing. Applying this approach to an area of the posterior wing blade between the L4 and L5 veins reveals that en > ctp-IR wings have significantly higher hair density relative to control en > + wings, indicating that Ctp-depleted cells are smaller (Fig. 2E). The measured effect of en > ctp-IR depletion on cell size (~14% smaller; Fig. 2E) is of similar magnitude to its effect on the relative size of the posterior wing blade (~17% smaller; Fig. 1C), indicating that effects on cell size, rather than cell number, are likely be a central cause of small en > ctp-IR wings (Fig. 2F).

The smaller cell size of Ctp-depleted wing cells combined with the shorter, thinner thoracic bristles seen with mirr^DE-Gal4 driven Ctp (Figure S2) is reminiscent of phenotypes induced by mutations in genes that support metabolic process of growth, such as the diminutive gene or the Minute ribosomal RNA genes³³ (and reviewed in ref. 34). The diminutive (dm) locus encodes the Drosophila Myc protein (dMyc), a well-established pro-growth transcription factor that promotes cell and tissue growth³⁵. Notably, ctp^ex3 null clones generated in the eye disc show reduced levels of dMyc protein as detected by immunofluorescence (Fig. 3A,A′). Within the wing disc, dMyc protein is normally detected throughout the dorsal and ventral halves of the pouch, with a region of cells along the dorsoventral boundary that express less dMyc³⁶. In en > ctp-IR discs, dMyc staining is intact in the corresponding anterior regions but decreased in these areas that express the ctp-IR transgene (Figure S4). Consistent with this link between Ctp and dMyc protein levels, heterozygosity for the null allele dMyc^PL35 further shrinks en > ctp-IR wings (Fig. 3B). In sum these data suggest that reduced cell size contributes significantly to the undergrowth of en > ctp-IR adult wings and potentially link ctp to one or more of the transcriptional, translational and post-translational mechanisms that control dMyc levels in disc cells^28,37,38.

Figure 3

Effect of Ctp loss on a panel of disc cell proliferation and/or growth pathways.

(A) Confocal image of an L3 eye disc (posterior to the right) mosaic for ctp^ex3 clones marked by the absence of β-galactosidase (βgal) (A) and stained for dMyc (A′). Note the drop in anti-dMyc fluorescence in ctp^ex3 clones in (A′). (B) Quantitation of PCR in the indicated adult genotypes shows that heterozygosity for the dMyc^PL35 allele (n = 6) has a dominant enhancing effect on the en > ctp^IR phenotype (n = 31). SEMs are indicated. *p = 0.0037. (C–K, M) Confocal images of L3 wing discs with Ctp depleted in the posterior domain (right-side in all images, with Ci marking the anterior domain (C–E,H,J,K) or transgenic UAS-GFP marking the posterior domain (F,G,I,M)) and analyzed for each of the indicated factors: (C′) anti-Notch (N) to detect the C-terminal fragment of the Notch receptor; (D′) the E2F-activity reporter PCNA-GFP; (E′) the Notch-activity reporter E(spl)mβ-CD2; (F′) the E2f1-lacZ enhancer trap; (G′) the Notch-activity reporter Su(H)-lacZ; (H′) the Jak-Stat activity reporter Stat10x-GFP; (I′) anti-Wingless (Wg) to detect the ligand of the Wg/Wnt pathway; (J′) anti-phospho-Mothers against decapentaplegic (Mad) to detect signaling through the Dpp pathway; (K′) anti-Cut, a target of both Wg and Notch; (M′) the ecdysone receptor (EcR) reporter activity EcRE-lacZ. (L,N) L3 eye discs mosaic for ctp^ex3 clones marked by the absence of GFP (L,N): (L′) anti-diphospho-Erk to detect signaling downstream of the Drosophila epidermal growth factor receptor (DER); (N′) anti-CyclinE (CycE), which is rate-limiting for progression into S-phase.

Ctp is dispensable for multiple signaling pathways but genetically interacts with Yorkie

Developing imaginal discs are exposed to signals from an array of conserved developmental signaling pathways, some of which are proposed to affect dMyc levels or activity (e.g. Notch and Wg)³⁹. Hence, a panel of reagents that detect activity changes in a variety of major cell pathways known to be active in the larval wing were assayed in en > ctp-IR discs (Figs 3 and S5). These included Su(H)-lacZ and E(spl)-mβ-CD2 (Notch pathway)^40,41, PCNA-GFP and E2f1-lacZ (E2F/Rb pathway)^42,43, EcRE-lacZ (EcR pathway)⁴⁴, Stat-10xGFP (Jak-Stat pathway)⁴⁵ and antibodies to the intracellular domain of Notch (N-icd), phospho-Mad (pMad), Wingless (Wg), the Wg/Notch regulated transcription factor Cut⁴⁶, the EGFR pathway component diphospho-Erk (dpErk) and the G1/S regulator CyclinE⁴⁷. None of these markers were evidently altered in the posterior domain of en > ctp-IR discs (Fig. 3C–K,M) or ctp^ex3 clones in the eye disc (Fig. 3L,N). In parallel genetic experiments, a loss-of-function allele of the pro-growth gene yorkie (yki) was found to dominantly enhance the en > ctp-IR small-wing phenotype (Fig. 4A). The Yki protein is the main target of the Hippo pathway and acts as nuclear co-activator for Scalloped (Sd)-dependent induction of Hippo target genes, which include dMyc⁴⁸. An enlarged-wing phenotype produced by expression of a UAS-yki-V5 transgene from the en > Gal4 driver is significantly suppressed by co-depletion of Ctp (Fig. 4B). Likewise ctp knockdown can suppress eye overgrowth induced by GMR-Gal4 driven expression of Yki^S168A, a hyperactive phospho-mutant form of Yki (Fig. 4C). Together, this evidence points towards a functional interaction between ctp and yki in disc cells destined to form the wing blade and eye.

Figure 4

Genetic interactions between ctp and yki in control of wing and eye size.

Box-plot representation of the effect of (A) the yki^B5 allele (n = 13), *p < 0.0001 or (B) a UAS-yki-V5 transgene (n = 22, with GFP n = 12, with ctp^IR + GFP), on PCR in control (en > GFP, n = 12) or Ctp-depleted (en > ctp^IR + GFP, n = 18) adult wings. All *p values < 0.0001. (C) Box-plot showing the effect of GMR-Gal4 driven Ctp-depletion at 29 °C among eye cells on final adult eye area (2-dimensional en face circumference) and the ability of the ctp^IR transgene to suppress this metric in the background of the GMR-yki^S168A hyperactive allele. SEMs are indicated; (GMR, n = 24; GMR > ctp^IR, n = 22; GMR > yki^S168A, n = 41; GMR > yki^S168A + ctp^IR, n = 24), both *p values < 0.0001.

Ctp loss reduces thread/diap1 transcription

The proposed dynein-independent role of mammalian LC8/Ctp as a dimerization hub for cytoplasmic and nuclear complexes (reviewed in refs. 3,7) suggests that Ctp could be involved in modulating activity of the Hippo pathway in vivo. This hypothesis was tested by assessing expression of the thread(th)/diap1 gene, a key anti-apoptosis factor and canonical Yki transcriptional target^23,49,50 in Ctp-depleted wing pouch cells. The steady-state level of Diap1 protein is reduced but not eliminated in the posterior compartment of en > ctp-IR larval discs (Fig. 5A,B), which could explain the mild increase in cC3 signal observed in ctp knockdown discs (see Fig. 2A,A′). The th-lacZ reporter, which is a Yki-responsive ‘trap’ of the bacterial β-galactosidase gene inserted into the endogenous th/diap1 locus⁵¹, also shows reduced expression in Ctp-depleted and ctp^ex3 mutant cells in the wing pouch (Fig. 5C,D,K,L and Figure S6E,F). A series of successively smaller promoter fragments of the th/diap1 promoter driving lacZ have been used to define a minimal Hippo response element (HRE) that responds to Yki hyperactivation in larval disc cells²⁵. Expression of two of these reporters, 2b2-lacZ and 2b2c-lacZ, is strongly reduced in response to ctp knockdown (Fig. 5E–H). Baseline expression of the minimal HRE-lacZ is fairly low, but its expression is also reduced in Ctp-depleted cells (Fig. 5I,J with co-expression of dicer2 to enhance ctp knockdown and Figure S6A,B, without dicer2). Furthermore, a mutant version of the minimal HRE lacking the Sd-binding site necessary for Yki dependent transcription (HREmut6-lacZ)²⁵ has lowered expression and no longer responds to loss of ctp (Figure S6C,D). In sum, these data provide evidence that Ctp is required to support transcription of the Yki target gene thread/diap1 in larval wing pouch cells.

Figure 5

Ctp supports expression of the diap-1/thread locus.

Confocal images of control (en > + in (A,C,E,G,I)) or Ctp-depleted (en > ctp^IR in (B,D,F,H,J)) L3 wing discs stained for Ci (anterior domain) (A,B) and Diap1 proteins (A’,B’), or with anti-ßgal to detect expression of endogenous th-lacZ (C’,D’) as well as sequentially smaller Yki-responsive promoter elements 2B2-lacZ (E’,F’), 2B2C-lacZ (G’,H’) and Hippo response element (HRE)-lacZ (I’,J’). Co-expressed UAS-GFP transgene marks the posterior compartment (C–J) and the addition of dicer2 (+dcr) was used to enhance ctp knockdown in (I,J). Note decreased expression of all three Diap1 markers in cells depleted of Ctp in the posterior (right-side) compartment. (K,L) Heat-shock induced GFP-negative ctp^ex3 null clones in an L3 wing disc in the background of the th-lacZ reporter. Area in L is a magnified view of the anterior clone in K. Note the drop in th-lacZ expression in ctp^ex3 cells.

Ctp loss elevates transcription of the bantam microRNA locus

Analysis of a second well-validated Yki target, the pro-growth bantam (ban) microRNA, was carried out to determine whether the requirement for Ctp is unique to th/diap1 transcription, or can be extended to other well-validated Yki-target genes as well. Two ban transcriptional reporter transgenes were used for these studies: ban3-GFP, which contains a large proximal fragment of the ban promoter driving GFP and brC12-lacZ, a Yki-responsive 410-bp promoter fragment that lies within the ban3 region and contains two Yki-association sites^27,52. Surprisingly and in polar contrast to th/diap1, expression of both ban reporters was strongly elevated in Ctp-depleted wing pouch cells (Fig. 6A–D). The brC12 reporter is also strongly induced in ctp^ex3 null wing pouch clones (Fig. 6E,F; two independent discs are shown with multiple ctp^ex3 clones), confirming the link between endogenous Ctp and ban promoter activity. Thus, while Ctp normally supports expression of the Yki-response gene th/diap1 in wing pouch cells, it has the inverse role of repressing transcription of the Yki-responsive locus ban. A third classic Yki reporter, expanded-lacZ, did not respond to Ctp-loss in otherwise wildtype larval wing cells or in those overexpressing Yki (Figure S7).

Figure 6

Ctp represses expression of the bantam locus.

Confocal images of control (en > + in (A,C)) or Ctp-depleted (en > ctp^IR in (B,D)) L3 wing discs in the background of the bantam reporters ban3-GFP (A’,B’) or brC12-lacZ (C’,D’). Anti-Ci marks anterior cells in (A,B) while a co-expressed UAS-GFP transgene marks posterior cells in (C,D). Note increased expression of both ban3 and brC12 upon depletion of Ctp in the posterior (right-side) compartment. (E,F) Heat-shock induced GFP-negative ctp^ex3 null clones in an L3 wing disc in the background of the brC12-lacZ reporter. Two different discs are shown. Note consistently elevated brC12-lacZ expression in ctp^ex3 cells, particularly within the pouch.

Epistatic relationships between yki and ctp in control of th/diap1 and ban reporters

The opposing effects of Ctp loss on the minimally Hippo-responsive th/diap1 and ban promoter fragments imply an Yki-dependent mechanism links Ctp to expression of these genes. To test these relationships, the effect of Ctp loss on brC12-lacZ and th-lacZ were reassessed either in the presence of overexpressed Yki, or RNAi depletion of endogenous Yki. Elevated brC12-lacZ expression upon Ctp-depletion is suppressed by concurrent RNAi depletion of Yki, indicating that Yki is required for maximal induction of ban following Ctp loss (Fig. 7A). Furthermore, expression of a UAS-yki transgene does not obviously elevate induction brC12-lacZ above the level observed in ctp^ex3 null clones (Figure S8A,A′). Indeed within the center of the L3 wing pouch, brC12-lacZ is more highly expressed in ctp null cells than in those expressing a yki transgene (Figure S8B; compare centrally located ctp^ex3 clone vs. surrounding Yki-expressing cells in the posterior domain). Thus, Ctp-depletion robustly activates brC12-lacZ expression in pouch cells via a mechanism that requires Yki. Because overexpression of Ctp in wing pouch cells via UAS-ctp has no effect on wing growth or Yki readouts (data not shown), the epistatic relationship between ctp and yki on the diap-1 promoter is more challenging to assess. However, transgenic expression of UAS-yki remains capable of inducing th-lacZ expression in wing disc cells depleted of Ctp (Fig. 7I–L), suggesting that ctp acts either upstream or parallel to yki. Thus, manipulating Yki levels modifies the effects of Ctp depletion on both th/diap1 and ban reporters, which provides some evidence that ctp inversely modulates Yki activity toward each of these target genes.

Figure 7

Epistatic relationship between Yki and Ctp-regulated growth.

(A–D) Anti-βgal staining of L3 wing discs to detect brC12-lacZ expression in the background of (A’) control en > GFP; (B’) Ctp-depleted, en > GFP + ctp^IR; (C’) Yki-depleted, en > GFP + yki^IR; or (D’) simultaneous depletion of Yki and Ctp, en > GFP + ctp^IR + yki^IR. Co-expressed UAS-GFP transgene marks the posterior compartment (A–D). (E,F) Quantitation of the effect of Ctp and/or Yki depletion on brC12-lacZ expression determined by measuring anti-βgal pixel fluorescence intensity across the anterior:posterior (A:P) boundary (left-to-right) in the white boxes in (A–D). GFP-expression domain is indicated. Average fluorescence of the anterior domain was set to zero (0). (E) ■ en > GFP, Δ en > GFP + ctp^IR. (F) ○ en > GFP + yki^IR, ▼ en > GFP + yki^IR + ctp^IR. (G–J) Anti-ßgal staining of L3 wing discs to detect th-lacZ expression in the background of (G’) control en > GFP; (H’) en > GFP + ctp^IR Ctp-depleted; (I’) en > GFP + yki^WT Yki-overexpression; or (J’) simultaneous expression of Yki and depletion of Ctp, en > GFP + yki^WT + ctp^IR. Co-expressed UAS-GFP transgene marks the posterior compartment (G–J). (K,L) Quantitation of the effect of Ctp depletion and/or Yki overexpression on th-lacZ expression determined by measuring anti-ßgal pixel fluorescence intensity across the anterior:posterior (A:P) boundary (left-to-right) in the white boxes in (G–J) ⋄ en > GFP, ▲ en > GFP + ctp^IR. (L) ● en > GFP + yki^WT, □ en > GFP + yki^WT + ctp^IR.

Wing disc cells lacking the Yki-binding protein Myopic also show selective effects on Yki-target genes⁵³. Myopic tethers Yki to endosomes for eventual degradation in lysosomes and its loss causes Yki to accumulate, leading to induction of ex and ban but not thread/diap1⁵³. To test whether Ctp is required for Yki cytoplasmic trafficking in a manner similar to Myopic, endogenous Yki and a V5-epitope tagged form of Yki were visualized in Ctp-depleted pouch cells (Figure S9). Yki steady state levels and nuclear:cytoplasmic distribution are unaltered in Ctp-depleted cells, indicating that wholesale changes in Yki protein dynamics and trafficking in ctp mutant wing pouch cells are unlikely to drive downstream effects on th/diap1 and ban expression.

Discussion

Here we define a role for the Drosophila protein Ctp, a member of the LC8 protein family, in regulating expression of two Hippo target genes, th/diap1 and bantam, in larval disc cells. Ctp is a member of a highly conserved family (Ctp, Dlc1 and DYNNL1/DYNNL2) of small proteins that were first identified as components of cytoplasmic dynein motors^1,5,6 but are now recognized to also act as interaction hubs for many proteins with roles in diverse processes such as autophagy, signal transduction, cell:cell adhesion and transcription^3,7,8. Largely because of this diverse set of potential effector pathways, the role of LC8 proteins in specific cellular and developmental contexts is not particularly well defined. We find that larval wing compartments depleted of Ctp by RNAi give rise to smaller compartments in the adult wing that are populated by smaller cells. Importantly, the Ctp-depleted larval precursors of these adult wing cells express markers of both M and S-phase and thus do not appear to undergo cell cycle arrest as reported elsewhere. Parallel analysis with a ctp genomic allele confirms that Ctp supports clonal growth in the larval wing disc, particularly in the pouch and that levels of the pro-growth transcription factor dMyc are reduced in ctp mutant epithelial cells. Testing candidate growth regulatory pathways active in the wing pouch uncovers a specific role for Ctp in regulating expression of genes regulated by the Hippo pathway target Yorkie (Yki), which is can positively regulate dMyc transcription. Activity reporters of two Yki-responsive genes, thread/diap1 and bantam, each respond to ctp loss in pouch cells and these effects map to smaller regions of the thread/diap1 and bantam promoters that contain Yki-responsive elements. Reduction of Yki activity enhances the Ctp-small wing phenotype and depletion of Ctp in turn blunts eye and wing growth driven by yki transgenes. These data are consistent with Ctp normally regulating Yki activity and Yki-dependent growth in imaginal disc epithelia.

The effects of Ctp on growth and Yki-target gene transcription are complex and intriguing. While Ctp supports Yki activity on the thread/diap1 promoter, it restricts activity of the bantam promoter and has no effect on a third Yki reporter, ex-lacZ. Moreover transgenic expression of ban drives tissue overgrowth^54,55 but upregulation of ban transcription in ctp mutant cells is paradoxically associated with tissue undergrowth. This apparent ban paradox may be explained by the finding that ban promotes tissue growth through dMyc^36,56, so that Ctp-depleted cells with reduced dMyc levels may be resistant to ban-induced growth.

The opposing effects of Ctp on thread/diap1 and ban transcription differ from core Hippo components, which affect these targets in a uniform way. The mechanism through which Ctp achieves its unique effects on thread/diap1 and bantam is not known. Intriguingly human LC8 interacts with the Kibra protein, a conserved element of the Hippo pathway¹⁵. Drosophila Kibra forms a complex with Merlin and Expanded and together these proteins promote phosphorylation of Warts⁵⁷. However Warts uniformly represses expression of Yki-target genes, making it unlikely that Ctp acts via a Kibra-Warts axis to exert inverse effects on thread/diap1 and ban. The Kibra-LC8 complex interacts with Sorting Nexin-4 (Snx4) to promote dynein-driven traffic of cargo between the early and recycling endosomal compartments¹⁷. Yki associates with endosomes⁵³ and endosomal traffic modulates levels of Yki protein and the transcription of its nuclear targets^53,58. However, Ctp loss does not obviously alter steady state levels of Yki or its distribution in wing disc cells (see Fig. 7) and although vertebrate LC8 (aka DYNLL1) is as a high-confidence interactor of the Yki vertebrate homolog Yap1^59,60, an equivalent association was not detectable in cultured Drosophila S2 cells (data not shown).

In addition to cytoplasmic effects, Ctp/LC8 proteins also have nuclear roles in transcriptional control³. For example, LC8 interacts with the estrogen receptor (ER), can promotes ER activity and is found with ER on the promoters of induced genes¹⁴. A recent study confirmed that ER dimerizes when bound to DNA and can activate target gene transcription either as a monomer or dimer⁶¹. Drosophila Ctp may thus interact with Yki, or alternatively with a nuclear effector in a distinct pathway that converges on Yki, to bias Yki promoter selectivity and /or to modulate formation of higher order transcriptional complexes with promoter-specific roles in vivo. Future studies will be required to identify Ctp interacting proteins within the Hippo pathway and to resolve the precise link between Yki and the multifunctional and highly conserved Ctp/LC8 protein.