Shavenbaby and Yorkie mediate Hippo signaling to protect adult stem cells from apoptosis

To compensate for accumulating damages and cell death, adult homeostasis (e.g., body fluids and secretion) requires organ regeneration, operated by long-lived stem cells. How stem cells can survive throughout the animal life yet remains poorly understood. Here we show that the transcription factor Shavenbaby (Svb, OvoL in vertebrates) is expressed in renal/nephric stem cells (RNSCs) of Drosophila and required for their maintenance during adulthood. As recently shown in embryos, Svb function in adult RNSCs further needs a post-translational processing mediated by Polished rice (Pri) smORF peptides and impairing Svb function leads to RNSC apoptosis. We show that Svb interacts both genetically and physically with Yorkie (YAP/TAZ in vertebrates), a nuclear effector of the Hippo pathway, to activate the expression of the inhibitor of apoptosis DIAP1. These data therefore identify Svb as a novel nuclear effector in the Hippo pathway, critical for the survival of adult somatic stem cells.


17
To compensate for accumulating damages and cell death, adult homeostasis (e.g., Polished rice (Pri) smORF peptides and impairing Svb function leads to RNSC 25 apoptosis. We show that Svb interacts both genetically and physically with Yorkie 26 (YAP/TAZ in vertebrates), a nuclear effector of the Hippo pathway, to activate the 27 expression of the inhibitor of apoptosis DIAP1. These data therefore identify Svb as a 28 novel nuclear effector in the Hippo pathway, critical for the survival of adult somatic 29 stem cells. 30

Keywords 31
Stem cells, OVOL/Shavenbaby, smORF peptides, Hippo pathway, apoptosis, Renal 32 system, Malpighian tubules, Drosophila. 33 The family of OvoL/Ovo/Shavenbaby (Svb) transcription factors has been strongly conserved 34 across evolution 1 and is characteristic of animal species. Initially discovered in flies for a dual 35 function in the development of epidermal derivatives (Svb) and of the germline (Ovo) 2, 3 , 36 mammalian orthologs (OvoL1-3) have soon been identified 4-6 . OvoL/svb genes produce 37 several protein isoforms and the existence of three partially redundant paralogs in mammals 38 complicates their genetic analysis. There is a single gene in Drosophila, which expresses 39 germline-(ovo) and somatic-specific (svb) transcripts from different promoters. Previous Tarsal-less) peptides, which belongs to a novel family of peptides encoded from small open 52 reading frames (smORF) hidden within apparently long noncoding RNAs 16 . Svb is first 53 translated as a long-sized protein that acts as a repressor (Svb REP ) 17 . Pri smORF peptides then 54 induce a proteolytic processing of Svb REP leading to the degradation of its N-terminal region 55 and releasing a shorter activator form, Svb ACT 17 . Further work has demonstrated that pri 56 expression is directly regulated by periodic pulses of steroid hormones 18 , allowing a 57 functional connection between hard-wired genetic regulatory networks (svb expression) and 58 systemic hormonal control (mediated by pri) for a proper spatio-temporal control of 59 epidermal cell morphogenesis 16  progenitors/stem cells, from basal invertebrates 30 to mammals 31-33 . Therefore, whereas a large 73 body of evidence supports a key role for OvoL/Svb in the behavior of somatic stem cells, a 74 functional investigation of their mode of action in vivo remains to be undertaken. 75 Here we built on the knowledge and tools accumulated for the study of Svb function in flies 76 to investigate its putative contribution to the behavior of somatic stem cells in the adult. We 77 show that in the Malpighian tubules, which ensure essential renal functions in insects 34, 35 , svb 78 is specifically expressed in the adult renal and nephric stem cells (RNSCs, see Fig. 1a). We 79 further find that a main function of Svb in the kidney is to protect RNSCs from apoptosis by 80 controlling the expression of the inhibitor of apoptosis, DIAP1, in interaction with Yorkie, a 81 nuclear effector of the Hippo pathway. 82

svb is expressed in Renal Nephric Stem Cells and controls their maintenance 84
To assay whether svb might be expressed in the adult, we tested large genomic reporter 85 constructs that cover each of the seven enhancers contributing to svb expression. We found 86 that one enhancer 9 , svb E10 , drove specific expression in tiny cells of the Malpighian tubules 87 ( Supplementary Fig. 1a, b). transcription factor of the Snail/SLUG family that is also expressed in intestinal stem cells 99 (ISCs 39 ) where it acts to prevent ISC differentiation 40,41 . Co-localization with an esg-LacZ 100 reporter confirmed that the svb E10 enhancer was active in RNSCs ( Fig. 1b and Supplementary 101 Fig. 1b). To define the minimal region of svb responsible for the expression in RNSCs, we 102 assayed a collection of overlapping constructs 7 . This identified two independent elements, the 103 svb E3N and svb E6 enhancers 7, 9 , which despite having distinct activities during embryogenesis 9, 104 42 drove similar expression in adult RNSCs ( Supplementary Fig. 1c). 105 Having established that two enhancers drive specific expression of svb in the adult 106 stem cells of the renal system, we next assayed consequences of depleting svb function in 107 RNSCs. We used a well-controlled genetic system, hereafter referred to as esg ts , ensuring 108 RNAi-mediated gene depletion, specifically in the stem cells and only at the adult stage 43 . We 109 also developed an image analysis pipeline, allowing automated quantification of the whole 110 population of RNSCs (see methods). In control conditions, the number of esg-positive RNSCs 111 remains stable after adult hatching, with approx. 350 cells per tubules (Fig. 1c,d). We only 112 noticed a weak reduction of RNSCs (300 cells) after one month. In contrast, esg ts -driven 113 RNAi depletion of svb led to a dramatic loss of RNSCs, which were completely absent after 114 32 days of treatment (Fig. 1c,d). The effects of svb depletion were already strong following 8 115 days of treatment, with a two-fold reduction in the number of RNSCs. Similar results were 116 observed when using an independent driver of RNSCs (Dome-meso-gal4) to direct RNAi-117 mediated knockdown of svb ( Supplementary Fig. 1d,e). The loss of RNSCs upon svb 118 depletion was also confirmed by staining against Hindsight, another transcription factor 119 specific of RNSCs within Malpighian tubules ( Supplementary Fig. 1d,e). Finally, the key role 120 of svb in the maintenance of adult RNSCs was further demonstrated by results from genetic 121 mosaics, showing that svb-null mutant cells 44 were unable to maintain RNSCs (Fig. 1e). 122 Taken together, these data thus reveal that svb is specifically expressed in RNSCs and 123 critically required for the maintenance of the adult stem cell compartment. 124 125 Svb processing is essential for its activity in Renal Nephric Stem Cells 126 In the epidermis, Svb activity relies on a proteolytic processing that causes a switch from a 127 repressor to an activator form. This processing is gated by Pri regulatory peptides, which bind 128 to and activate the Ubr3 ubiquitin E3-ligase that, in turn, triggers a limited degradation 129 operated by the proteasome 45 (Fig. 2a). Thereby, pri mediates a systemic control of Svb 130 maturation since the expression of pri is directly regulated by the ecdysone receptor (EcR) 18 . 131 To assess whether the function of Svb in Malpighian tubules also required its proteolytic 132 maturation, we investigated a putative function of pri and ubr3 in RNSCs. We screened a 133 collection of pri reporter lines 18, 46 and identified two cis-regulatory regions driving 134 expression in RNSCs ( Fig. 2b and Supplementary Fig. 2a,b). Consistently with the expression 135 of pri in RNSCs, pri depletion almost fully eliminated RNSCs upon 8 days of RNAi 136 treatment ( Fig. 2d and Supplementary Fig. 2c). In addition, a dominant negative form of the 137 Ecdysone Receptor (EcRDN) that abolishes pri expression during both embryonic and post-138 embryonic development 18 was sufficient to reduce the number of stem cells when specifically 139 expressed in adult RNSCs (Fig. 2d). Furthermore, we found that ubr3 was also required for 140 RNSC maintenance, as deduced from results of RNAi-mediated depletion and genetic 141 nullification 45 of ubr3 activity (Fig. 2c,d). Finally, the expression of OvoA that behaves as a 142 These results provide compelling evidence that the whole regulatory machinery discovered 148 for its role in the development of epidermal cells 17, 18, 45 is also at work in adult RNSCs. We 149 therefore concluded that the post-translational maturation of the Svb transcription factor is 150 essential for the maintenance of RNSCs. 151 152

Svb protects Renal Nephric Stem Cells from apoptosis 153
The loss of RNSCs observed following the lack of svb function or maturation could 154 theoretically result from at least three different causes: i) lack of proliferation, ii) precocious 155 differentiation, or iii) increased cell death. Consistent with the quiescent behavior of RNSCs, 156 we observed a very low frequency of RNSC mitosis in controls, as deduced from staining 157 with the mitotic marker Histone3-P ( Supplementary Fig. 3a) and as previously noticed 37 . showing that svb depletion does not trigger RNSC differentiation (Fig. 3a). Finally, we tested 176 whether svb-depleted RNSCs were lost because they underwent apoptosis. Since apoptotic 177 figures are difficult to observe in the digestive track 50 including the Malpighian tubules, we 178 assayed consequences of blocking programmed cell death by expressing the viral caspase 179 inhibitor p35 51 . Although the expression of p35 had no detectable effect by itself on RNSCs, 180 it significantly rescued the number of RNSCs when svb was depleted ( Fig. 3c and see below). 181 Taken together, these data show that the loss of RNSCs observed upon svb loss of function is 182 primarily due to stem cell death, indicating that a main role of Svb is to protect adult stem 183 cells from undergoing apoptosis. 184 185

Svb acts downstream of Hippo 186
During epidermal development, svb is expressed in post-mitotic cells where it acts as a 187 terminal differentiation factor that controls cell shape remodeling 15, 52 . We noticed that 188 RNSCs lacking svb displayed a reduced size, as well as an altered morphology (Fig. 3b). One 189 could speculate that these defects in cell shape might stress RNSCs and thus induce apoptosis. Previous work has shown that the Hippo pathway is a key regulator of the Drosophila gut 197 homeostasis, controlling survival and proliferation of stem cells for tissue regeneration 57, 58 . 198 Consistently, we found that the activation of Hpo induced a strong reduction in the number of 199 RNSCs. However, co-expression of OvoB, the constitutive activator isoform of Svb, together 200 with Hpo was sufficient to rescue the loss of RNSCs (Fig. 4a). These results therefore 201 demonstrated that if Svb and Hpo interact for the homeostasis of RNSCs, the loss of RNSCs 202 observed upon svb knockdown is not due to the activation of the Hippo pathway, since Svb is 203 instead acting downstream Hpo. In contrast, overexpression of Yki (mimicking a loss of 204 Hippo signaling 59 ) induced a strong increase in esg+ renal cells, which displayed abnormal 205 tumor-like morphology when compared to wild-type RNSCs (Fig. 4a). Unexpectedly, this 206 increase in the number of renal stem cells was entirely suppressed upon svb depletion, or 207 expression of the constitutive repressor OvoA (Fig. 4a and Supplementary Fig. 3f). 208 Quantification indicated that esg+ cells overexpressing Yki were even more sensitive to svb 209 loss-of-function than otherwise normal RNSCs (Fig. 4a), a result well in line with the extra-210 resistance of intestinal stem cells to apoptosis when compared to tumorous stem-like cells 50 . 211 Hence, the function of Yki in RNSCs requires Svb, suggesting that Svb was acting either 212 downstream or in parallel with this nuclear effector of Hippo. Several lines of evidence ruled 213 out the former and validated the latter hypothesis. First, knocking down Yki also led to RNSC 214 loss ( Supplementary Fig. 3c). Expression of the Svb constitutive activator (OvoB) was 215 nevertheless not able to rescue RNSC survival in the absence of Yki (as opposed to the over 216 expression of Hpo, Fig. 4a), showing that Svb requires Yki activity for RNSC maintenance 217 ( Fig. 4a and Supplementary Fig. 3c). Second, although Yki is sufficient to induce DIAP1 218 expression 60 (and see below), Yki was not able to rescue the lack of Svb while DIAP1 alone 219 did (Fig. 4a). Indeed, we found that DIAP1 was sufficient to compensate for svb-depletion 220 ( Fig. 4a and Supplementary Fig. 3f). In sum, while both Svb and Yki are required for RNSC 221 maintenance, re-expression of Yki is not sufficient to rescue the loss of svb function. 222 Reciprocally, Svb is not sufficient to rescue a proper RNSC compartment in the absence of 223 Yki, showing that Svb and Yki act in parallel for the survival of adult stem cells. 224 We thus concluded that Svb acts downstream of Hippo cytoplasmic core components and, 225 together with Yki, both nuclear factors are required to protect RNSCs from apoptosis. 226

Svb as a novel nuclear effector of the Hippo pathway 227
Having established that Svb and Yki genetically interact, we wondered whether the two 228 proteins might interact to control the expression of common target genes, e.g., DIAP1. Yki is 229 unable to bind DNA by itself and need to associate to sequence-specific transcription 230 factors 54 . Interestingly, Yki contains two WW protein domains shown to mediate interaction 231 with partners bearing PPxY motifs (such as Wts 60 , Wbp2 61 and Mad 59 ), and we detected two 232 PPxY motifs within the Svb protein, at position 523 (PPFY) and 881 (PPSY). Co-233 immunoprecipitation assays showed that Svb bound to the wild type form of Yki, while the 234 mutation of Yki WW motifs was sufficient to abrogate the interaction with Svb (Fig. 4c). A 235 second piece of evidence emerged from the comparison of chromatin immuno-precipitation 236 (ChIP-seq) datasets between Svb 14 and Yki 62 . We found that Svb and Yki share 836 common 237 genomic binding sites (Supplementary Fig. 4a) and statistical tests established the 238 significance of this overlap (Supplementary Fig. 4b). Interestingly, co-binding of Yki was rare 239 for the direct target genes of Svb identified in the epidermis 12-14 , as illustrated by shavenoid or 240 dusky-like that both lack Yki binding ( Supplementary Fig. 4d,d'). In contrast, Svb was often 241 bound to known Yki target genes, such as bantam, fat, piwi or nanos 63 ( Supplementary Fig.  242 4c,c'). ChIP-seq also revealed that Svb and Yki bound to a same region of DIAP1, within an 243 enhancer driving specific expression in adult intestinal stem cells 64 (Fig. 4b). We therefore 244 tested if Svb might regulate DIAP1 expression. Although very weak in control conditions, 245 expression of DIAP1-LacZ was strongly enhanced upon Yki overexpression. This induction 246 was completely antagonized by OvoA (Fig. 4d). Similar results were obtained with the 247 isolated DIAP1 enhancer (DIAP1-4.3-GFP) containing the binding sites of Yki and Svb, the 248 expression of which was again enhanced by Yki overexpression and abrogated upon 249 counteracting Svb activity (Fig. 4e). These data thus strengthen the conclusion that Svb and 250 Yki functionally interact in RNSCs to prevent apoptosis, at least in part through promoting 251 DIAP1 expression. 252 One important question was whether the interaction between Svb and the Hippo pathway also 253 took place in other tissues. The function of Hippo has been initially described in imaginal 254 discs, which give rise to most adult tissues 65 and Yki overexpression promotes cell 255 proliferation in both wing and eye discs 60 . We tested Svb/Yki genetic interactions in the wing 256 using collier-Gal4 that drives expression in the medial (L3-L4) intervein region. Yki over-257 expression resulted in the expansion of this region due to tissue overgrowth (Supplementary 258 Fig. 3d). In contrast, OvoA leads to both a reduction of the L3-L4 region and the absence of 259 epidermal trichomes. As in RNSCs, OvoA was epistatic to Yki, since the wing region 260 expressing both yki and ovoA was smaller than in controls and lacked trichomes 261 ( Supplementary Fig. 3d). In the eye, overexpression of Yki using the GMR-Gal4 driver 262 promoted extra cell proliferation resulting in an increased eye size. Similar results were 263 obtained following pri overexpression, and co-expressing pri and yki resulted in a synergistic 264 eye growth (Supplementary Fig. 3e). Northern blotting of RNAs extracted from adult heads 265 revealed that DIAP1 mRNA levels were increased following pri overexpression 266 ( Supplementary Fig. 4e), while there was no effect on yki or cycE mRNA. 267 We interpret these results to imply that Svb functionally interacts with Yorkie, both in adult After clearing by centrifugation at 14,000 rpm for 10 min, immuno-precipitations were done 378 from transfected lysates in lysis buffer using anti-GFP antibody (GFP-trap, Chromotek). 379 Immuno-precipitated samples were separated by SDS-PAGE and transferred to PVDF 380 membranes, then blotted using anti-GFP (TP401, Acris Antibodies, 1:10000) and anti-HA 381 (Covance, 1:2.000) antibodies. Secondary antibodies anti-mouse or anti-rabbit IgG-HRP 382 conjugates (Jackson Laboratory, 1:10.000) were detected using ECL Clarity (Biorad).             shown at the right of each picture.